You are on page 1of 196

ABSTRACT This study began with an interesting empirical puzzle concerning a pressing public health issue in the developing

world. More than half a million women die related to pregnancy and childbirth each year. One in four maternal deaths worldwide occur in India. This study asks, given two Indian states with comparable socio-cultural contexts and economic indicators, what explains differing progress for maternal mortality reduction? Comparative analysis of the policy processes in each state provides insights to the research question and suggests the need for more holistic frameworks for analyzing policy processes than currently exist. Comparative case studies draw on more than 140 interviews with health policy experts, managers and service delivery personnel in the two Indian states, as well as with representatives of domestic and international NGOs and donor agencies over three months in 2007; numerous national and state government reports; policy and program documents; and reports and documents from international donor and nongovernmental organizations to triangulate data relevant to answering the research question. The findings suggest that differing social historical influences in each state; worldviews, priorities and degrees of power among major political parties; strength and ideas of key policy actors; and capacity and norms in the concerned public health bureaucracies shaped variation in availability and access to publicly provided safe motherhood services in the two states, especially for more vulnerable groups (e.g. women with lower economic, educational and social status). The results suggest a need for more holistic frameworks for analyzing policy processes frameworks that highlight the influence of such social historical influences as

major social movements and other feats of social organization (e.g. colonialism) as they shape conditions of the political environment, the power of policy actors and ideas, and organizational structures that shape subsequent policy processes. In doing so, the study identifies a neglected set of variables in historical social organization, refines our understanding of how the political environment matters, and presents politicalbureaucratic actors and ideas as a category of factors that bridges policy and management to better reflect the overall set of causal relationships that influence policy outcomes.

UMI Number: 3381604 Copyright 2009 by Smith, Stephanie Lynette All rights reserved

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

______________________________________________________________
UMI Microform 3381604 Copyright 2009 by ProQuest LLC All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

_______________________________________________________________
ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106-1346

Copyright 2009 Stephanie Lynette Smith All rights reserved

vi TABLE OF CONTENTS Chapter 1 Policy processes & safe motherhood in South India Introduction Theorizing the policy process Conclusion Chapter 2 Safe motherhood: profile & public policy Introduction Safe motherhood: scope, causes & interventions Health systems The scope of Indias maternal mortality crisis Public policy & maternal mortality in India Karnataka Tamil Nadu Conclusion Chapter 3 Methods Methods of exploration Case selection Data Primary data Secondary data Validity Chapter 4 Socio-cultural dynamics & safe motherhood Introduction 1 2 6 15 19 20 21 27 30 32 37 40 43 45 46 47 50 50 56 56 58 59

vii Economic status Education & literacy Geography Caste, class, religion & gender Conclusion Chapter 5 Politics & safe motherhood Introduction Politics & health policy in Tamil Nadu Politics & health policy in Karnataka Local & regional political dynamics Conclusion Chapter 6 The public health bureaucracy & safe motherhood Introduction The public health bureaucracy in Tamil Nadu The public health bureaucracy in Karnataka Conclusion Chapter 7 60 65 70 75 82 84 85 86 98 112 117 120 121 122 139 147

Toward a more holistic framework for analyzing policy processes 149 Introduction Summary of findings Implications of findings for analyzing the policy process Linking policy processes & outcomes At the intersection of policy & management Political environments & policy change 150 151 152 153 156 158

viii Governance structures & policy processes Social historical factors in policy processes Study limitations Conclusion 160 163 165 167

Appendix A References

Comparing frameworks for analyzing policy processes

172 174

ix TABLES 2.1 2.2 3.1 4.1 Global maternal mortality ratios with confidence intervals Summary data on India, Karnataka and Tamil Nadu Key safe motherhood indicators for India and major states 23 36 49

Economic status and safe motherhood process indicators in Tamil Nadu and 62

Karnataka, 1998-9 4.2

Economic status and safe motherhood process indicators in Tamil Nadu and 63 67 68

Karnataka, 2005-6 4.3 4.4 4.5 Education levels and safe motherhood process indicators, 1998-9 Education levels and safe motherhood process indicators, 2005-6

Rural and urban institutional delivery rates in Tamil Nadu and Karnataka, 1992-3, 73

1998-9 and 2005-6 4.6 2005-6 5.1 6.1 7.1 A.1 Share of health in revenue budget of select major states (%) Translation of political priorities in the health bureaucracy Proposed holistic framework for analyzing policy processes Comparing frameworks for analyzing policy processes

Rural and urban antenatal and postnatal care rates in Tamil Nadu and Karnataka, 73 97 148 171 172

FIGURES 6.1 Maternal death audit form, Karnataka excerpt 146

CHAPTER 1

POLICY PROCESSES & SAFE MOTHERHOOD IN SOUTH INDIA

2 INTRODUCTION Few would argue that the staggering maternal mortality burden born by women and families in the developing world more than half a million women die related to pregnancy and childbirth every year (WHO 2007) is unaffected by material socioeconomic factors, such as poverty, illiteracy and gender inequality. The magnitude of these problems and their influence on access to maternal health care, especially for vulnerable groups, would seem to explain much in developing country settings. But this study presents an empirical puzzle in which material socioeconomic factors fail to fully explain why women in one Indian state access safe motherhood services at a greater rate than women in a state with historically comparable socioeconomic indicators. This study examines how public policy processes contribute to that variation. Looking more closely at the relationships between socioeconomic factors and safe motherhood outcomes in the two comparative cases, a distinct difference emerges between the South Indian states of Karnataka and Tamil Nadu. In Karnataka, the gap in rates of access to maternal health services between more resourced and more vulnerable groups is wide, dragging down state averages. In Tamil Nadu, the gap is narrower, with women from more vulnerable groups accessing safe motherhood services at significantly higher rates than their counterparts in Karnataka. That is, women in Tamil Nadu access safe motherhood services at greater rates than women in Karnataka across income groups, levels of education and other socio-cultural categories. Why? Another distinct difference between the states is the rate at which women access safe motherhood services in the public sector. Women from more vulnerable groups reported accessing safe motherhood services in the public sector at greater rates than the

3 private sector in both states in a 1998-9 national health survey (IIPS 2001a and b). In Tamil Nadu, they accessed maternal health services in the public sector at nearly twice the rate of their counterparts in Karnataka, making up a substantial portion of the gap in rates of access between the states. What explains this difference? In Tamil Nadu, strong leadership and a policy community emerged surrounding efforts to reduce maternal mortality in the mid-1990s. These policy actors effectively integrated safe motherhood services among other service delivery priorities. They developed new agency procedures, systems for training, monitoring and evaluation to increase the availability and accessibility of safe motherhood services in the public sector, with special attention to reaching groups with less access. Norms and expectations that supported public health and social welfare policy, as well as the relatively robust capacity of the health bureaucracy in terms of human and financial resources facilitated their course of action. A significant social movement in the past century was an important factor shaping operative norms and expectations for social welfare policy benefiting vulnerable groups in the state. Tamil Nadus system of competitive party politics profoundly shaped by this social movement reinforced these as priorities for the states public health bureaucracy and provided opportunities for policy communities to promote safer motherhood services. These forces worked over time to systematically shape policy and action in ways that increased the availability and accessibility of safe motherhood services in Tamil Nadus public sector. Karnataka, like neighboring Tamil Nadu, is a relatively wealthy state benefiting from comparatively competent and stable governance in the broader Indian context, affecting better than national averages for access to safe motherhood services in both

4 states. Karnatakas performance lags next to Tamil Nadu, however. The state has had no comparable program of action to increase the availability and accessibility of publicly provided safe motherhood services and the most vulnerable groups are particularly neglected. No strong policy community has formed surrounding the issue and leadership for it is conspicuously absent. Institutional norms were less pro-poor and more oriented to high-level hospital care than to the primary health approach that serves as a critical link between women from more vulnerable groups and safe motherhood services. The states political system serves to maintain the status quo distribution of resources for public health services. Recent overtures from national leadership have not altered the equation for priority attention to expanding access to rural health care, including maternal health services, due to conflicts between parties at the state and national levels. Comparative analysis of these cases has much to contribute to understanding our empirical puzzle, as well as important factors and relationships in policy processes. The same types of factors were at work in the policy processes in both states. Differing social historical influences in each state; worldviews, priorities and degrees of power among major political parties; strength and ideas of key policy actors; and capacity and norms in the concerned public health bureaucracies shaped variation in availability and access to publicly provided safe motherhood services in the two states. The findings of this research depart from some of the more prominent frameworks for analysis of policy processes in an important way. The findings emphasize connections between policy and implementation or management, dimensions of policy processes that tend to be treated as somehow distinct and amenable to separate analysis. The research question in this study could not be sufficiently answered by focusing the analysis on

5 agenda setting or public management factors and dynamics alone, however, suggesting the need for a more holistic framework for analyzing policy processes. This is not to suggest that scholars do not recognize the integral nature of various stages, phases or dimensions of policy processes Sabatier and Jenkins-Smiths (1999) advocacy coalition framework is highly developed in this regard. But existing frameworks of analysis tend to be more geared toward understanding agenda setting or policy change or organizational performance toward policy ends than toward understanding the factors that connect them in a big picture sense. The results of this study informed development of a more holistic framework of analysis that aims to address this limitation of existing policy literature. The results also suggest important refinements to our understanding of certain key variables. Socio-economic and cultural variables are important to consider, but social history referring to the historical organization of society and including such feats of social organization as social movements, state formation and colonization is a neglected factor that importantly shapes the more contemporary policy processes that are our most common objects of study. Changes in governing coalitions or political leadership affect policy agendas, but how is dependent on the worldviews, priorities and power of political parties as revealed in this study. Lastly, policy actors need not be defined strictly in terms of a particular phase of the policy process, but are more fruitfully conceived in terms of multiple roles that may span the political-bureaucratic labyrinth in policy processes. This understanding is crucial to overcoming analytic barriers between policy and implementation in our frameworks of analysis.

6 These points form some of the central contributions of this research to our understanding of policy processes. The next section of this chapter discusses some of the key ideas it draws on and responds to in public policy literatures. THEORIZING THE POLICY PROCESS Public policy processes are complex, involving a multitude of individual and organizational actors, conflicting ideas, values and beliefs about how to define problems and what to do about them, and various institutional, legal and societal norms, rules and structures. This study is concerned with understanding how these factors and possibly others affected the policy processes that led to differences in public sector performance for safe motherhood in two Indian states. Existing approaches to policy analysis point to a number of important factors and relationships some of which are well suited to explaining our empirical puzzle and some of which this study suggests need refinement. This promises to contribute to the development of better frameworks of analysis and to further our understanding of their applications in different settings. To begin, Baumgartner and Jones (1993) punctuated equilibrium model of instability in American politics attempts to explain much of the policy process. It argues that policy processes are characterized by lengthy periods of relative stability, of incremental policymaking, that are occasionally disrupted by brief periods of wideranging policy change. They explain that policy monopolies, stable sets of policy actors supported by powerful structural arrangements and political understandings of an issue, tend to exert significant influence on incremental policy processes within particular issue areas or domains. These arrangements supporting policy ideas are generally connected to core political values. The best are such things as progress, participation,

7 patriotism, independence from foreign domination, fairness, economic growth things no one taken seriously in the political system can contest (Baumgartner & Jones 1993, p. 7). The main focus of their approach is to explain infrequent, short bursts of major policy change that are characterized by the public attention-grabbing dynamics of agenda setting that consumes officials time and resources. Competing ideas about the nature of a policys contribution or impact or changes in central political values can create openings for major change (Baumgartner & Jones 1993; Kingdon 1984; Sabatier & Jenkins-Smith 1999). According to Baumgartner and Jones (1993) model, changes to the structure of political institutions and the way issues are understood by those institutions can precipitate substantial policy change. For example, significant social movements, changes in political leadership or shifting understanding of an issue from one of personal to government responsibility could affect rapid policy change. As the authors suggest, Public policymaking responds to the great cleavages of society, traditionally organized by political parties (Baumgartner & Jones 1993, p. 21). The approach to policy analysis presented by Baumgartner and Jones (1993) in Agendas and Instability in American Politics draws attention to policy influences exerted in the broader realm of macro politics, involving partisan political conflict. Major policy change is set in motion by influences in the macro political environment and tends to involve higher profile venues for authoritative decisions, such as those in the realm of legislative or executive authority. Agenda setting in these arenas determines whether and the extent to which certain issues receive attention and resources, as well as the operating rules within which policy actors especially those representing the administrative state,

8 but including technical experts and others with public and private policy interests function within issue areas or policy domains, such as public health (Baumgartner & Jones 1993; Redford 1969). Baumgartner and Jones approach suggests that policy analyses need to account for these larger cleavages, significant shifts in thinking about how to define and solve public problems that are represented by political upheaval and changes of leadership, in order to understand the ebb and flow of policy attention. It also suggests that these have wide-ranging effects that can help to explain factors affecting decisions and actions in policy domains. Their points are well taken, but require further inquiry to understand their applicability to the Indian context. The focus on agenda setting is also limiting, downplaying the power of policy entrepreneurs and communities that design and carry out governmental programs of action that determine the availability and accessibility of public services. The advocacy coalition framework developed by Sabatier and Jenkins-Smith (1999) suggests that major policy change is a function of power and conflict between central groups of actors within policy subsystems. Drawing on Sabatier and JenkinsSmith (1999), the policy subsystem is an important concept used in this analysis to refer to a subset of a larger socio-political system in which a fairly stable set of actors influences government decisions and actions on issues within a policy domain in a given jurisdiction, such as health policy in Tamil Nadu, over time. Like Baumgartner and Jones (1993), the authors approach is sensitive to macro political influences, such as changes in systemic governing coalitions (e.g. political leadership) and public opinion. These can alter the terms of subsystem policymaking by shifting political support to different types

9 of problems and solutions. Changes in socio-economic conditions, impacts from other policy subsystems and other more stable factors such as basic constitutional and sociocultural structures, shape opportunities and constraints for advocacy coalitions to influence authoritative government decisions. The advocacy coalition framework features two central hypotheses concerning policy change. It suggests that the central value- and belief-based commitments of a governmental program of action are unlikely to be changed as long as the advocacy coalition that instituted it retains power, unless change is imposed by a hierarchically superior power or significant perturbations external to the subsystem (Sabatier & JenkinsSmith 1999). The latter are a necessary but not sufficient cause of change. While external factors are important catalysts to change, the advocacy coalition framework is most centrally concerned with attributes of and learning between groups of central actors in policy subsystems as evidenced in its remaining seven hypotheses. The advocacy coalition framework contains both a broader framework that identifies several types of variables that should be considered in analyses of policy processes and a theory of policy change and changes to core policy beliefs of actors within subsystems. The framework and theory were developed based on the U.S. system of policy and politics, containing implied assumptions about well-organized interest groups, mission-oriented agencies, weak political parties, multiple decisionmaking venues, and the need for supermajorities to enact and implement major policy change (Sabatier & Weible 2007, p. 199). Sabatier (1998) and Sabatier and Weible (2007) responded to criticisms of the models shortcomings in explaining policy processes in European corporatist regimes and less democratic societies. They adapted the model to

10 consider the degree of consensus needed for major policy change and the accessibility and number of venues actors must navigate to reach a decision point on a policy proposal. The framework, perhaps best suited to analysis of more open pluralist regimes, is quite adaptable to analysis of a range of regime types including those requiring greater and lesser degrees of consensus and more and less open political systems (Sabatier & Weible 2007). Indias political system certainly falls within these ranges. The advocacy coalition framework is useful in terms of identifying variables and suggesting causal relationships that might be important to this analysis. However, this study is less concerned with the dependent variable major policy change change that is broad in topic and scope (Sabatier 1998) and more concerned with the aspects of public service provision that affect achievement of policy goals (e.g. equitable access to maternal health services). The advocacy coalition framework implies that policy change is the primary objective of policy communities, that their struggle is waged at a political level that draws public attention and incites negotiation between political leaders in prominent forums. The struggle to influence higher-level political agendas and policy change has important implications for government decisions, actions and results, but this study suggests that it does not adequately capture the roles of policy actors in connecting policy agendas to public performance outcomes. Kingdons streams model (1984, 1995) is another important theory of the policy process. Its primary relevance here is to recognize the significance and roles of actors in policy processes. According to Kingdons approach, policy entrepreneurs and other members of policy communities frame ideas about problems and solutions, looking for windows of opportunity to advance issues on policy agendas so that they receive serious

11 attention from authoritative government actors. Kingdon emphasizes legislative and executive actors in his agenda setting model, but suggests policy entrepreneurs and communities may include those inside or outside government, in elected or appointed positions, interest group representatives, academics, consultants, and representatives of civil society organizations. Their power derives from different bases of influence, such as expertise, a position of leadership or political connections, and collective action surrounding shared concern about a particular policy issue or arena (Kingdon 1984, 1995). Kingdon is primarily concerned with major legislative policy change and suggests that administrative actors individuals and agencies are more instrumental in specifying policy alternatives or choices than agenda setting, though this formulation downplays administrative discretion as an important source of policymaking power (Appleby 1949; Meier 1979; Rourke 1984; Walt 1994). As Rourke (1984) suggests in Bureaucracy, Politics and Public Policy, The scope of administrative discretion is vast in all societies both in the everyday routine decisions of government agencies and the major innovative or trend-setting decisions of public policy (p. 37). Agents of the administrative state have the authority to establish and pursue their own priorities within broader mandates determined by constitutional structures and macro political institutions and should not be overlooked (Meier 1979). Indeed, in their advocacy coalition approach to policy analysis, Sabatier and Jenkins-Smith (1999) propose that actors engaged in policy subsystems be defined inclusively, with implementation actors as influential shapers of policy and action. Advocacy coalitions, equivalent to policy communities, serve important functions in

12 defining problems, generating and vetting policy solutions, and helping issues rise on decision agendas. They can also play important roles in shaping plans, programs and implementation as they often include the experts that inform and authorities that carry out policy points cogently made by such scholars of policy implementation as Elmore (1979), Lester and Goggin (1998), Lipsky (1980), Maynard-Moody (2000), and Vinzant and Crothers (1998), and supported by the findings of this research. Theories of the policy process also accord an important role to such factors as core values, beliefs and ideas that contribute to the identification of problems, solutions, and their importance relative to other societal issues. They motivate actors and define the structures of political institutions and government programs (Baumgartner & Jones 1993; Kingdon 1984, 1995; Sabatier & Jenkins-Smith 1999). Actors form collective action efforts to advance their beliefs and ideas to points of authoritative decision, action and institutionalization (Sabatier & Jenkins-Smith 1999), suggesting their interaction as another defining aspect of policy processes. Stones (2002) concept of causal stories is useful for understanding the importance of framing ideas in policymaking processes. Actors use causal stories to define problems in ways that assign responsibility, point to solutions and inspire collective action on their behalf (Snow et al. 1986; Stone 2002). Stone (2002) offers two useful and well known examples in which causal framing had these effects: in Silent Spring Rachel Carson (1978) persuasively reframed environmental degradation as a consequence of human activity rather than a natural occurrence, inspiring environmental activism; and in Unsafe at Any Speed Ralph Nader (1966) reframed injuries and deaths from motor vehicle crashes not as a result of accidental factors but manufacturers

13 inattention to safety features, inspiring organized consumer protection efforts. When responsibility for problems can be assigned, especially through a short and clear causal chain, policy communities are more likely to unite, to elevate issues on policy agendas and to alter institutional structures to support them (Keck & Sikkink 1998; Snow et al. 1986; Stone 2002). Ideas are a medium of exchange and a mode of influence even more powerful than money and votes and guns. Shared meanings motivate people to action and meld individual striving into collective action. Ideas are at the center of all political conflict. Policy making, in turn, is a constant struggle over the criteria for classification, the boundaries of categories, and the definition of ideals that guide the way people behave (Stone 2002, p. 11). Indeed, the advocacy coalition framework and policy streams models suggest that though policy communities are centered around interest in a given issue, that does not imply agreement about how to define a problem or what should be done about it (Kingdon 1995; Sabatier & Jenkins-Smith 1999). Conflicting ideas contribute to fragmentation and reduced power of policy communities (Kingdon 1995; Shiffman & Smith 2007). By the same token, the more ideas resonate and have salience within existing political agendas and institutions, the more influence policy actors are likely to have on decisions and action (Berman 2001; Keck & Sikkink 1998; Shiffman & Smith 2007; Stone 2002). The institutional analysis approach also makes a relevant contribution to our understanding of policy processes, suggesting that rules, norms and strategies structure the incentives that shape the behavior of policy actors (Ostrom 2007). The term institution, Ostrom (2007) explains, refers not to organizational units but to shared

14 understandings of prescriptions for behavior that are subject to more and less formal monitoring and sanction as organizations or society dictate (see also Crawford & Ostrom 2005). In other words, the shoulds and should nots contained in organizational rules and societal norms shape the behavior of policy actors, and in turn the programs of action they design and carry out. This analysis uses the term institution in the same sense, drawing on the concept to contribute to our understanding of factors that shape decisions and actions of actors in relevant policy subsystems and agencies. Drawing on broader social theories of culture, the organizational culture perspective speaks to the power of institutions and ideas. It suggests that shared values, beliefs, assumptions and understandings shape the rules, policies, strategies, goals and practices of organizations and their members (Martin 2002; Ott 1989; Schein 1992; Smircich 1983; Wilson 1989). And, by extension, the rules and practices of societies and their members are shaped by shared values, beliefs and understandings, as evidenced by the influence of class, caste and gender dynamics and flowing over into the formal institutions of the state. The organizational culture perspective assumes that many organizational behaviors and decisions are almost predetermined by the patterns of basic assumptions existing in the organization. They become the underlying, unquestioned but virtually forgotten reasons for the way we do things here, even when the ways are no longer appropriate (Ott 1989, pp. 2-3). Organizational culture is a lens through which new ideas are filtered in the administrative state to determine whether they fit with existing goals and practices. The more ideas resonate with the institutions operating in central state agencies and political organizations, the more likely they are to be integrated into the

15 sense of mission that guides their behavior. Wilson (1989) suggests how high the stakes can be on this point, stating, Tasks that are not defined as central to the mission are often performed poorly or starved for resources (p. 110). Systems of beliefs, values and ideas are represented in the structures of political and administrative institutions the stated goals, policies and standard operating procedures that guide the ways individuals and groups process information and act on it. As Sikkink suggests, "New ideas do not enter an ideological vacuum. They are inserted into a political space already occupied by historically formed ideologies. Whether or not consolidation occurs often depends on the degree to which the new model fits with existing ideologies (Sikkink quoted in Berman 2001, p. 236). Ideas gain power through their institutionalization in social, political and administrative structures (Finnemore & Sikkink 2001; Martin 2002; Ott 1989; Schein 1992). Ideas and structures are shaped by the power of actors to reinforce them or influence change over time (Keck & Sikkink 1998; Snow et al. 1986; Stone 2002). This is the stuff of policymaking from the setting of agendas to authoritative decisions and goal pursuit. CONCLUSION This dissertation draws upon these theories of the policy process in order to examine its empirical puzzle more and less equitable rates of access to safe motherhood services among women in two South Indian states and the policy process contribution to that variation. This study departs from existing frameworks of analysis in two important ways. First, it is less concerned with major policy change as the dependent variable. It is difficult to form a direct link between major policy change as conceptualized in the

16 policy streams, punctuated equilibrium and advocacy coalition frameworks of analysis and changes in safe motherhood policies, programs and outcomes in our cases. There is very little evidence to support a model in which policy communities waged high profile, conflict-ridden campaigns targeting legislative or executive venues to affect major policy change and even less to indicate that major policy change directly resulted in differences in availability of and access to safe motherhood services in either case. Rather, differences in the capacity of the health systems, the institutional norms, and the actors and ideas operating in the health policy arena in each case shaped varying degrees of availability and access to safe motherhood services, the latter particularly in recent years. Importantly, the policy community that affected change in Tamil Nadu in the past twelve to fifteen years was rather insulated. Its primary support came from bureaucratic leaders. These leaders championed the cause, networking with other members of the policy community, framing ideas about maternal mortality reduction in ways that appealed to political principals and exercising their discretion to integrate priority for the cause into the states vast public health service delivery network. This finding suggests that organizational structures, as well as political-bureaucratic actors and ideas, importantly shape the relationship between policy processes and outcomes a central concern of this study. But differences in the political environments and social history of the two states also made important contributions to shaping policy processes and outcomes in the cases. Existing frameworks of analysis point to change in governing coalitions as an important factor affecting policy agendas, but in Tamil Nadu this did not make a difference on its own. It was consistencies in the nature of the worldview, priorities and strength of Tamil

17 Nadus governing coalitions and their opposition that provided continuous opportunities for the safe motherhood policy community to promote their cause. In Karnataka, these conditions had the opposite effect, providing few opportunities for strong advocacy to emerge and take hold. There was also variable consistency in Karnatakas political priorities as weak governing coalitions fell from and ascended to power. Differences in political environments, although both featured competitive multi-party politics, affected systematic differences in policy processes, performance and outcomes in the states. In addition, differences in the states social histories Tamil Nadus widespread social movement and Karnatakas formation of disparate regions lacking a cohesive political identity were instrumental in shaping these varying political environments and subsequent policy processes. This factor is neglected in existing frameworks of analysis. To summarize, this study is interested in more than what it takes for items to rise on policy agendas, to come to points of decision or to achieve successful implementation it is interested in understanding holistically which types of factors and what types of relationships in policy processes affect varying policy outcomes. This is a tall order and one that comes with costs in terms of specificity and certainty. But it is an order that comes with the rewards of identifying neglected factors, refining under-specified variables and pushing the boundaries of our frameworks of analysis so that we can better understand the range of factors and relationships that affect policy outcomes on pressing social issues. These are a few of the contributions this study makes to our understanding of policy processes. To conclude, this chapter has set out the theoretical questions examined in this dissertation and discussed relevant conceptual and theoretical contributions from

18 public policy literatures. Chapter 2 discusses the policy and country contexts for the study, providing background information on the global safe motherhood crisis, linking safe motherhood with broader health systems policy issues, and profiling the Indian cases that inform this study. Chapter 3 provides further discussion of the study design, data and measures taken to support validity. Chapters 4 through 6 present the case evidence in comparative perspective. The first empirical chapter examines the relationship between social conditions and safe motherhood outcomes; the second, political influences; and the third, bureaucratic influences on the programs of action that shape access to safe motherhood services in Tamil Nadu and Karnataka. The concluding chapter summarizes the results and draws implications for public policy scholarship.

19

CHAPTER 2

SAFE MOTHERHOOD: PROFILE & PUBLIC POLICY

20 INTRODUCTION Safe motherhood first gained international attention as a significant problem, one primarily affecting the developing world, two decades ago when the World Health Organization released the first estimates of the global maternal mortality burden. Since then, global advocates have mobilized research efforts to improve understanding of the scope and nature of the problem, organized technical conferences to promote particular interventions and raised global awareness of the problem at international meetings and events. Published work by such scholars as AbouZahr (2001, 2003), Campbell (2001) and Shiffman and Smith (2007) chronicle policy developments pertaining to the global safe motherhood initiative. Shiffmans work contributes to our understanding of how safe motherhood rose on policy agendas in Guatemala, Honduras, India, Indonesia and Nigeria (Shiffman 2007; Shiffman & Garces del Valle 2006; Shiffman & Ved 2007). But there is very little understanding of how policy processes affect outcomes on this issue in high burden countries beyond this work. Knowledge of safe motherhood tends to be developed based on a more narrow technical perspective rather than lenses of analysis that facilitate understanding of the political tides that affect investment in broader health and social policies, and maternal health in turn. The first section of this chapter draws on that more technically oriented literature to describe the scope of the global maternal mortality crisis that takes the lives of more than half a million women annually (WHO 2007). It elucidates key safe motherhood indicators, causes of death and interventions recommended by the global policy community concerned with the issue. This is intended to provide readers with background on the state of knowledge and current thinking on the issue. It also describes

21 the context for safe motherhood policy and implementation in the developing world. The tone of this section reflects the technical tone and narrow focus of much scholarship on the matter up until it was more recently linked with the influence of the strength of broader health systems. Relevant health systems literature is briefly discussed because it enhances our understanding of the linkages between the technical issues surrounding safe motherhood and related health policy and system issues that play into policy processes. The latter part of this chapter turns to the Indian context and cases examined in this study. India accounts for a quarter of the global maternal mortality burden (WHO 2007). Safe motherhood recently gained agenda status at the national level in India (Shiffman & Ved 2007), but that does not explain historical policy developments and outcomes or guarantee impacts at the sub-national level. Largely decentralized authority for health policy and implementation to the state level and varying outcomes at the subnational level beg the question of how policy processes make a difference for access to maternal health care at this level. This study responds to that need through comparative case studies that provide insights to the empirical puzzle and further our knowledge of important factors and relationships in policy processes. SAFE MOTHERHOOD: SCOPE, CAUSES & INTERVENTIONS A maternal death is defined as the death of a woman while pregnant or within 42 days of termination of pregnancy (WHO 2004, p. 3). More than half a million women die related to pregnancy and childbirth every year worldwide, a figure that has held steady for twenty years (WHO 2007). Further, ninety-nine percent of maternal deaths occur in the developing world with Sub-Saharan Africa and South Asia accounting for 86

22 percent of global maternal deaths (WHO 2007). One in four maternal deaths occur in India. Globally, the maternal mortality ratio (MMR) is the most widely used indicator to denote the extent of the problem. The maternal mortality ratio the number of maternal deaths per 100,000 live births within a specified time period indicates the probability of a woman dying once she is pregnant. The ratio controls for fertility rates.1 World MMR was estimated at 400 for 2005 (WHO 2007). To give a sense of variation in maternal mortality ratios globally in 2005, Sierra Leone featured the highest MMR in the world at 2,100; Nigerias MMR was 1,100; Indias 450; and Irelands, the lowest MMR in the world, 1 (WHO 2007). Table 2.1 shows select maternal mortality ratios estimated for 2005, along with lower and upper estimates as reported by the World Health Organization (2007). Some of Indias south Asian neighbors are included to give a sense of the scope of the problem in that region. Sri Lanka is a successful anomaly among South Asian nations, benefiting from government investment in the health system and strategic efforts to expand access to maternal health care (Pathmanathan et al. 2003). Although a tremendous gap in this important maternal health indicator is readily observed between developed and developing nations, caution should be exercised in using individual country estimates to compare across countries or time periods (Stanton et al. 2000; WHO 2007). There are differences in data sources across countries, data reliability is a problem and different methods have been used to develop estimates over

The maternal mortality rate, based a ratio of the number of maternal deaths to the number of women of reproductive age during a given time period, is another indicator sometimes used. It does not control for fertility rates and is not reported on further in this study. For deeper analysis of safe motherhood indicators and measurement methods, see Graham et al. 1989, Stanton et al. 2000 and WHO 2007.

23 time. In addition, confidence intervals that are used to indicate the statistical reliability of maternal mortality estimates are large, due to the limited numbers of maternal deaths in short time periods and sample sizes of surveys used to produce estimates (Graham et al. 1989; Stanton et al. 2000). Table 2.1 Global maternal mortality ratios with confidence intervals MMR World Sierra Leone Nigeria Bangladesh India Pakistan Sri Lanka Ireland Source: WHO 2007 The issue of data limitations is an important one. In their study assessing maternal mortality indicators in 13 countries based on Demographic and Health Survey data commonly used by developing country governments and international donor agencies, Stanton, Abderrahim and Hill (2000) found severe limits to data quality on time of death recorded relative to pregnancy, childbirth, and the postpartum period and suggested maternal mortality is under-reported though likely to a greater extent in the more distant than recent past (p. 120). Graham and colleagues (1989) and Ronsmans and colleagues (1997) issued similar cautions regarding the reliability and use of estimates. Despite 400 2100 1100 570 450 320 58 1 Lower Estimate 220 880 440 380 300 99 39 1 Upper Estimate 650 3700 2000 760 600 810 77 2

24 measurement difficulties, the maternal mortality ratio offers an important window onto the problem and has become a key indicator for assessing the state of maternal health globally and nationally. What causes maternal mortality? Leading direct causes of maternal death include hemorrhage, hypertensive diseases, sepsis/infection, obstructed labor and complications of abortion (Khan et al. 2006; Ronsmans & Graham 2006). Scholars suggest the means of averting maternal deaths have long been available (Campbell 2001; Maine & Rosenfield 1999, p. 480). But access to and availability of life-saving interventions are not necessarily so straightforward. Thaddeus and Maines three delays framework is useful for understanding this (1994). Through a review of maternal health research, Thaddeus and Maine identified the time that elapses after the development of an obstetric complication as crucial to determining obstetric outcomes. Three types of delays emerged: delays in the decision to seek care, delays in reaching a health facility and delays in provision of adequate care at health facilities. It is important to understand that it is common practice for women in the developing world to deliver in the home contrasting with Western bio-medical childbirth norms of the modern era. The three delays are based in recognition that home deliveries in developing countries are commonly attended by unskilled family members or lay birth attendants that are not trained to recognize or manage obstetric complications; that women and their families often lack access to financial resources and transportation, resulting in delays in reaching health facilities; and that provision of adequate care is not guaranteed in health systems that lack capacity in terms of human, technical and financial resources. This framework facilitates understanding of macro-level health systems

25 limitations, as well as social barriers that play out at the micro level, such as gender and cultural dynamics that influence family decisions about maternal health care. Maternal mortality reduction is about more than a set of direct causes and technical interventions; it is about constraints and opportunities to addressing these barriers within relevant policy subsystems, and in macro political and social spheres of influence thus helping to inform the research questions of interest to this dissertation. The safe motherhood policy community has developed a number of maternal mortality reduction strategies taking social conditions and the state of health care systems in developing countries into account over the past twenty years. Some of the more prominent have been increasing training of traditional birth attendants, provision of antenatal care and risk screening, attendance of skilled health professionals at births and availability of emergency obstetric care (AbouZahr 2001, 2003; Campbell 2001; Hussein et al. 2005; Maine & Rosenfield 1999; Miller et al. 2003; Paxton et al. 2005). Skilled birth attendance, institutional delivery rates and antenatal care use are key safe motherhood indicators in addition to the maternal mortality ratio. It should be noted that the three delays model, safe motherhood indicators and maternal mortality reduction strategies and recommendations discussed in the paragraphs above and below are tailored to developing country contexts. Emergency obstetric care in the event of life-threatening complications tends to not be readily available outside urban areas and accessing it is cost prohibitive for many families. There are also significant gaps in identification of high-risk pregnancies that should be referred for delivery with the assistance of medical professionals in developing country settings. There is a body of research demonstrating that planned home deliveries for low risk pregnancies result in

26 similar outcomes, often with fewer medical interventions, as hospital deliveries attended by medical professionals in industrialized North American and European settings. See (OConnor 1993) for background on the home birth movement in the United States and large-N studies by Duran (1992), Olsen (1997) and Johnson and Daviss (2005) for additional information on this research. The global safe motherhood policy community focuses on issues affecting women in the developing world an entirely different set of circumstances from those women face in industrialized nations. They recommend skilled care for women during pregnancy and childbirth and access to emergency care. These criteria are difficult to meet outside of institutional settings and in many cases not even then. Recognizing that institutional delivery rates are an imperfect indicator of maternal health outcomes, this study nonetheless uses it as a key indicator of the state of maternal health care in conformance with the global safe motherhood policy communitys recommendations for developing country contexts. In October 2007, a special issue of The Lancet dedicated to safe motherhood reported on three key recommendations for maternal mortality reduction that were agreed upon in the global safe motherhood policy community: First, comprehensive reproductive health care, including family planning and safe abortion, or where necessary, postabortion care. Second, skilled care for all pregnant women by a qualified midwife, nurse, or doctor during pregnancy and especially during childbirth. Third, emergency care for all women and infants with lifethreatening complications (Starrs 2007). In the same issue, Freedman (2007) discussed the strength of health systems as integral to the effectiveness of these strategies (Starrs 2007), an approach consistent with the World

27 Health Organizations World Health Report 2000 and a recent report from the United Nations Millennium Project Task Force (2005). This understanding of safe motherhood problems, solutions and outcomes specific to the issue as importantly shaped by broader health systems and the social, political and administrative dynamics that shape them is important to the model of the policy process developed in this thesis. This helps us to understand the influence of other priorities in decision venues and how their impacts on programs of action might facilitate or hinder progress for safe motherhood. It also reveals several practical health system constraints and challenges to improving access to safe motherhood services. Health systems Recent scholarship on health systems defined as all the activities whose primary purpose is to promote, restore or maintain health by the World Health Organization (2000, p. 5) draws attention to the significance of the overall organization, resources and responsiveness required to provide for the health needs of populations. This is in contrast to programmatic and policy emphasis on narrowly defined health issues, such as HIV/AIDS, malaria, family planning or safe motherhood, that tend to be the foci around which policy communities form in the global health arena and which subsystem policy analysis tends to emphasize. According to the World Health Organization and others, health systems encompass the institutions and individuals engaged in formal health services, such as professional medical care, traditional healers, home care, health promotion and prevention of ill-health, as well as mechanisms for its provision (e.g. insurance) and regulation (Reich et al. 2008; UN Millennium Project Task Force 2005; WHO 2000).

28 In its The World Health Report 2000 Health Systems: Improving Performance, The World Health Organization stated: Combating disease epidemics, striving to reduce infant mortality, and fighting for safer pregnancy are all WHO priorities. But the Organization will have very little impact in these and other battlegrounds unless it is equally concerned to strengthen the health systems through which the ammunition of life-saving and life-enhancing interventions are delivered to the front line (2000, p. xii). After years of channeling attention and resources to vertical disease-specific programming designed for cost-efficiency and to bypass the shortcomings of health systems, global health policy communities are increasingly recognizing the importance of broader systems of health care delivery for shaping outcomes. This represents a significant shift in thinking about how to alleviate health problems, particularly in lowand middle-income countries a shift reminiscent of the primary health care movement linked with the Alma Ata Health for All declaration of 1978 (Campbell 2001). To this point, Gilson (2003) has importantly observed that health systems are complex socio-political institutions and are part of the social fabric of every country (p. 1461). Freedman (2005) shares this view, observing, Societal values and norms are signaled and enforced not only through interpersonal relationships, but also in the very structure of a health system (p. 21). Like Gwatkin and colleagues (2004), they call attention to how health systems tend to reflect and produce social inequities such as those defined by caste, class and gender relations in societies, linking institutional rules and norms in social, political and administrative spheres of influence to more and less equitable health outcomes a phenomena of interest to the present research.

29 Central health systems constraints and strategies to overcome them have received a good deal of attention from health policy scholars. In a 2004 article in The Lancet, Travis and colleagues stated, there is growing consensus that a primary bottleneck to achieving the MDGs [Millennium Development Goals2 which include maternal mortality reduction] in low-income countries is health systems that are too fragile and fragmented to deliver the volume and quality of services to those in need (p. 900). Major health systems constraints include shortfalls in health human resources, financing,3 information systems, equipment and drug supply, infrastructure, weak management and poor regulation (Hanson et al. 2003; Murray & Frenk 2000; Oliveira et al. 2003; Reich et al. 2008; Travis et al. 2004; WHO 2000). Drawing on the World Health Organization report of the Commission on Macroeconomics and Health (2001), Hanson and colleagues (2003) suggested infusion of additional financial resources could reduce constraints to taking up priority interventions that would strengthen health systems and promote specific health goals: for example, at the community level lack of demand and barriers to use; and at the health services level, shortages and distribution of staff, weak supervision and technical guidance, inadequate drug, medical and equipment supply, and infrastructure. However, health sector policy and strategic management constraints affecting these and other weaknesses would depend

The Millennium Development Goals set a target date of 2015 to alleviate a number of pressing social problems, including poverty, education and health goals, that United Nations member states agreed to at the turn of the century. Goal five aims to improve maternal health by reducing the global maternal mortality ratio by 75 percent from 1990 levels. 3 For a comprehensive assessment of health systems finance needs and issues, see the World Health Organization report of the Commission on Macroeconomics and Health (2001). Hongoro and McPake (2004) provide an overview of global health human resources constraints.

30 more upon support from political actors, the context for reform (e.g. corruption), and inter-sectoral policies such as civil service rules and budgeting and planning frameworks (Hanson et al. 2003). The authors suggest political factors and certain types of policies are important to strengthening health systems, but the dynamics of policy processes and how they matter remains neglected in this literature. This study addresses this gap. THE SCOPE OF INDIAS MATERNAL MORTALITY CRISIS This section presents an outline of the problem nationally using key indicators in order to provide context for the subsequent discussion of public policy and safe motherhood in India, as well as case selection and profiles. The scope of Indias safe motherhood problem is substantial with the country accounting for far more maternal deaths than any other nation. This is partly a function of the size of Indias population. The status of Indias maternal mortality ratio was on par with the average for other developing nations in 2005 (WHO 2007). As reported by the World Health Organization (2007), Indias maternal mortality ratio was fifty times the average for developed regions of the world (450 compared to 9 maternal deaths per 100,000 live births). Not comparable with the estimates developed by the World Health Organization, UNICEF and UNFPA reported above due to differing methodologies, the Registrar General of India estimated the nations maternal mortality ratio at 301 (95% Confidence Interval 285-317) for 2001-3, a reduction of nearly 25 percent from 1997-8 (Registrar General of India 2006). This survey provides the most reliable source of data on maternal mortality ratios at the sub-national level in India. The Registrar General also reported that less than thirty percent of all births in India took place in a public or private health care

31 institution in 2003.4 Leading causes of death included haemorrhage (38%), sepsis (11%), and unsafe abortion (8%) (Registrar General of India 2006). As noted in the previous chapter, safe motherhood indicators tend to vary among Indian states along the lines of socio-cultural and development indicators. Women in the Empowered Action Group states of Bihar and Jharkand, Orissa, Madhya Pradesh and Chattisgarh, Rajasthan, Uttar Pradesh and Uttaranchal, and Assam suffered maternal mortality at rates disproportionate to their representation in the female population overall, while women in the southern states of Andhra Pradesh, Karnataka, Kerala and Tamil Nadu were under-represented in this figure (Registrar General of India 2006). The Empowered Action Group is a set of states in northern India thus designated for their relatively low socio-economic indicators among Indian states. In contrast, South Indian states are known for their relatively advanced socio-economic indicators and, in the Indian context, somewhat more advanced ideas about gender equity. The maternal mortality ratio for the Empowered Action Group states in 2001-3 was 438 (95% Confidence Interval 410-467) compared to 173 (95% Confidence Interval 144-202) for the southern states. Other major states, including Gujarat, Haryana, Maharashtra, Punjab and West Bengal, featured an MMR of 199 (95% Confidence Interval 178-220) during the same time period. We do not have reliable, comparable estimates of MMR at the state level for historical comparison. As discussed earlier in the
4

It should be noted that what constitutes an institutional delivery varies substantially. Deliveries are counted as institutional regardless of the level of health facility or training of staff in attendance. Quality of facilities and equipment also varies significantly. The empirical chapters of this thesis elaborate upon some of the variation in public sector institutional delivery services. Limited though this measure is, it is one of the best available indicators of the state of maternal health services and access in India. Quality of care related to institutional delivery and other safe motherhood services is an important and neglected issue.

32 chapter, the figures reported on this indicator should be approached with caution due to large confidence intervals that make inter-state comparisons difficult. Other key safe motherhood indicators should be consulted to gain a better understanding of relative progress among states. Institutional delivery rates reflected similar disparities among these groups at 16 percent in the Empowered Action Group states and Assam, 64 percent in the southern states, and 34 percent in the other major states (Registrar General of India SRS 2006). The Registrar Generals report (2006), Maternal Mortality in India: 1997-2003, concluded that attendance of skilled health professionals at births and institutional deliveries needed to be expanded to reduce maternal mortality in India further, especially in the Empowered Action Group states and Assam. This is consistent with global safe motherhood recommendations cited previously (Starrs 2007). PUBLIC POLICY & MATERNAL MORTALITY IN INDIA Indias safe motherhood crisis is an interesting and important context in which to study policy processes. To begin, though Indias safe motherhood problem is significant and general goals for maternal mortality reduction are documented in national health policies since the early 1980s, the issue received little meaningful attention or resources from Indias national government until the 2004 national election (Shiffman & Ved 2007). A new government that promised attention to social equity goals rose to power in 2004 with a mandate to expand access to health services in rural areas. Attention from prominent national figures and unprecedented resources dedicated to maternal mortality reduction goals followed in the Government of Indias sweeping National Rural Health Mission policy introduced in 2006. As promising as this increase in attention and

33 resources was, it remains unclear whether national priority for maternal mortality reduction forms a sufficient basis for alleviation of the problem in India. It certainly did not play a substantial role in Tamil Nadus early policy attention to the issue. Shiffman and Ved (2006) noted the generation of state level commitment for the issue as a key challenge to achieving progress. The structural devolution of health policy to the sub-national level in India is an important consideration in designing and conducting this analysis. Constitutionally, health policy is on Indias concurrent list a responsibility shared by the national and state governments. The federal government has some influence as it sets national health policy goals, develops programs and funds a portion of states health activities. In practice, Indias states are largely responsible for health policy and implementation, including a significant role in service delivery. States fund the greater proportion of public health services and activities (Government of India 2005), giving local political priorities substantial influence over the sector. The National Rural Health Mission strengthens the influence of priorities determined through state-level political processes with its emphasis on devolution of this responsibility to the local level. The structure of responsibility for health policy and implementation in India suggests states are an important level to analyze policy processes affecting safe motherhood. Further, scholars of public policy processes suggest that diverse outcomes are common at the sub-national level, especially in less centralized systems (Baumgartner & Jones 1993; Walt 1994), suggesting this as an important level of analysis. The cases selected for analysis in this study feature varying safe motherhood outcomes and relatively comparable socio-cultural and economic indicators. This latter

34 feature helps to control to the extent possible for social influences on safe motherhood outcomes so that the affects of and relationships between macro political and subsystem policy dynamics may be more readily observed. Case studies of the south Indian states of Tamil Nadu and Karnataka, states historically comparable on the bases of similar sociocultural and economic indicators, raise important questions about what explains varying outcomes. Their comparable social structures and location within the same federal structure of government positions the cases well to help answer our empirical question about how policy processes have contributed to varying rates of access to safe motherhood services in the two states, as well as to reveal important factors and relationships in policy processes. Table 2.2 below summarizes key data on both states and India to examine further the comparability of Karnataka and Tamil Nadu on select social and economic indicators and show their relationship to key health indicators. It should be noted that both states compare favorably to Indias averages and the states historical figures are crucial to establishing the bases for their comparison. The Government of India (2002) reported that the per capita net domestic product, incidence of poverty and level of human development in Karnataka compared favorably to Tamil Nadu dating to the early 1980s. The poverty rate in Karnataka was, in fact, quite a bit lower than in Tamil Nadu (38 versus 52 percent). On another important indicator, female literacy, Tamil Nadu led Karnataka by seven to eight percentage points in the early 1990s on into the early 2000s. Importantly, these social and economic indicators were associated with closely comparable fertility and infant mortality rates, key public health indicators, between the states in the early and late 1990s as reported in the National Family Health Surveys (IIPS

35 2007). Total fertility rates continued to be comparable in the 2005-6 surveys, while infant mortality rates declined to a greater extent in Tamil Nadu than in Karnataka (IIPS 2007). Significantly, institutional delivery rates (the best historically comparable maternal health data we have) show disparities between the states dating to the first National Family Health Survey in the early 1990s while other key health indicators remained closely comparable. In other words, Karnataka and Tamil Nadu are historically comparable on key economic, social and health indicators with the exception of maternal health and this presents an interesting puzzle for investigation. If social and economic indicators do not fully explain the disparities, then what can help us to explain this variation? This chapter continues with a brief introduction to the states to familiarize the reader with their general geographic, socio-economic and health profiles. Chapter 4 compares relationships between key socio-economic indicators and safe motherhood outcomes in the two states in much greater detail and further discusses their relevance to answering the research questions.

36 Table: 2.2 Summary data on India, Karnataka and Tamil Nadu


India Past Per capita net domestic product, 1981-2/1997-8 Below poverty line 44 (%), 1983/1999-2000 Human development index (HDI), 1981/1991 Female literacy rate, 39 1991/2001 Total fertility rate, 3.4 1992-3/2005-6 Infant mortality rate, 79 1992-3/2005-6 Institutional delivery 26 rate, 1992-3/2005-6 41 39 67 64 90 57 65 43 68 31 2.7 2.9 2.1 2.5 1.8 54 44 57 51 65 0.302 0.381 0.346 0.412 0.343 0.466 26 38 20 52 21 1,671 2,840 1,584 2,866 1,570 3,141 Recent Karnataka Past Recent Tamil Nadu Past Recent

Sources: India National Human Development Report 2001(Government of India 2002) for population figures, per capita net domestic product, poverty line, human development, and female literacy rate; IIPS (2007) National Family Health Survey data for fertility, infant mortality and institutional delivery rates

37 Note: The infant mortality rate indicates the probability of a child dying before her first birthday; the fertility rate indicates the average number of births per woman Karnataka Karnataka is located in south India. It borders Kerala and Tamil Nadu to the south, Andhra Pradesh to the east, Maharashtra to the north and Goa tucked in at the northern tip of the states western coastline along the Arabian Sea. In 1956, Karnataka state (called Mysore until 1973) was formed along linguistic lines of Kannada-speaking areas of five territories, including Bombay, Hyderabad, Madras, the former princely state of Mysore and the independent state of Coorg (Government of Karnataka 2006). The states 191,791 square kilometers encompass mountains, plateaus and coastal areas (Government of Karnataka 2006). Karnatakas population numbered 52.7 million in 2001, sixty-six percent residing in rural areas (Government of India 2001). In 1999-2000, per capita net state domestic product stood at 16,343 Rupees (about $413 US), leading Indias average of 15,562 Rupees (about $390 US) (Government of India 2002). Agriculture made up a third of net state domestic product in 2000 while industry (16 percent) and services (54 percent) accounted for the remainder (Government of India 2001). Fifty-six percent of workforce labor was agricultural (Government of India 2001). The proportion of the population living below the poverty line decreased from 38 percent in 1983 to 33 percent in 1993-4 to 20 percent in 1999-2000 (Government of India 2002). Among major religions, Hindus made up 84 percent of the population, Muslims 12 percent and Christians 2 percent in 2000 (Government of India 2001). Karnatakas standard human development indicators have improved over time and compare favorably with Indias, just leading nationwide averages on several

38 measures. In 2001, the state led the all-India average on the Human Development Index5 (0.478 to 0.302) and Gender Development Index (0.637 to 0.609) (Government of Karnataka 2006). Literacy rates were at 67 percent compared to 65 percent and reflected similar male-female gaps: male literacy rates stood at 76 percent in both settings and female literacy at 57 percent in Karnataka and 54 percent in India overall (Government of India 2001). On the sex ratio, an important indicator of the status of women in society, Karnataka led the national average at 964 compared to 933 females per 1,000 males still reflecting a significant gender gap, though not as wide as nationally (Government of India 2001). Health indicators in Karnataka have improved steadily over time and lead Indias averages. Life expectancy in Karnataka is more that two years longer than the countrys average at 63.3 years (Government of India 2002). Fertility declined from nearly three children per woman in the early 1990s to 2.1 in 2005-6 while the national average stood at 2.7 (IIPS 2007). Infant mortality rates experienced a notable decline of approximately one third over the same period with 43 deaths per 1,000 live births in Karnataka in 20056 compared to 57 nationally (IIPS 2007). Karnataka has also made progress on safe motherhood indicators. Between the 1992-3 and 2005-6 National Family Health Surveys, overall institutional delivery rates increased from 39 to 67 percent (IIPS 2007). According to National Family Health Survey figures, both rural and urban institutional delivery rates increased substantially. The United Nations Development Programme has used the Human Development Index since 1990 as a broad measure of well being based on life expectancy, educational and economic indicators. India was ranked 126th of 177 countries in the 2006 report (UNDP 2006). The Gender Development Index, introduced in 1995, is an adjusted measure of human development that accounts for gender disparities. Scores on the indexes range from 0 to 1, with higher scores indicating relatively higher levels of human development.
5

39 Rural rates increased from 27 to 57 percent while urban rates increased from 68 to 85 percent (IIPS 2007). It should be noted, however, that the respected Sample Registration System (Registrar General of India 2006), with its larger sample size and more rigorous methods of data collection, reported overall institutional delivery figures more conservatively at 41 percent in 1991 and 50 percent in 2003. Both reports show the state leading Indias averages by more than one and a half times, suggesting the state of maternal health is somewhat more advanced in the state than nationally. Assistance of a skilled health professional (doctor, nurse or other) at deliveries, another important safe motherhood indicator, increased from 47 to 71 percent between the 1992-93 and 2005-06 surveys (IIPS 2007). That said state-level figures and rural-urban comparisons do not capture regional differences in institutional delivery rates. The gap between the lowest performing districts, Koppal and Raichur at 21 percent, and the highest performing districts, Udupi, Dakshina Kannada and Bangalore Urban all in the ninetieth percentile, is wide (Government of India 2004). All eleven of the states northern districts6 track at or below the median (61 percent in Belgaum) among twenty-seven districts, suggesting regional dynamics may be at work in determining safe motherhood outcomes in the state. There is a precipitous drop in institutional delivery rates between the top performers in the ninetieth percentile and the next tier of performers in the seventieth percentile (five districts). So while the northern districts draw down state-level figures, this forms an incomplete explanation for lagging safe motherhood performance in the state.

According to the Karnataka Human Development Report 2005, the northern districts include Bidar, Gulbarga, Raichur, Koppal, Bellary, Bijapur, Bagalkot, Belgaum, Gadag, Dharwad and Haveri.

40 Lastly, Karnatakas maternal mortality ratio compared favorably to Indias overall ratio of 301 [95% Confidence Interval 285-317] in Sample Registration System reporting for 2001-2003. The states maternal mortality ratios showed a decline from 266 [95% Confidence Interval 202-331] to 228 [95% Confidence Interval 169-287] deaths per 100,000 live births between the 1999-2001 and 2001-2003 Sample Registration System surveys (Registrar General of India 2006). Trends are not possible to assess with certainty, however, because of the large overlap between confidence intervals and longterm MMR data are not yet available at the state level. As noted previously, maternal mortality ratios are best considered in relationship to other safe motherhood indicators. Tamil Nadu Tamil Nadu is Indias southernmost state occupying its eastern coastline. It borders Kerala is to its west, Karnataka to the northwest and Andhra Pradesh to the north. The states territory covers 130,000 square kilometers and includes coastal regions, plains, valleys and mountainous areas. The modern era of the past 60 years saw Tamil Nadu formed of the pre-independence territory of the Madras Presidency, but Tamil language and history dates back nearly 6,000 years (Government of Tamil Nadu 2003). With a population of over 62 million, Tamil Nadu is one of Indias most populous states. Fifty-six percent of its population resides in rural areas (Government of India 2001). It is also one of the most prosperous of Indias states with per capita net state domestic product of 19,141 Rupees (about $470 US) in 1999-2000 compared to Indias average of 15,562 Rupees (about $382 US) (Government of India 2002). The high tech boom has had an important on the economies of Tamil Nadu and Karnataka, especially in their capital cities. Agriculture remains a significant sector in Tamil Nadu, accounting for 65

41 percent of workforce labor and nearly 20 percent of net state domestic product (Government of Tamil Nadu 2003). Services (58 percent) and industry (24 percent) make up the remainder. The proportion of the population living below the poverty line declined from 52 percent in 1983 to 35 percent in 1993-4 and 21 percent in 1999-2000 (Government of India 2002). The vast majority of the population is Hindu (88 percent), but the state also features significant Christian (6 percent) and Muslim (5.6 percent) populations (Government of India 2001). Standard human development indicators depict Tamil Nadu as a fairly progressive Indian state. Tamil Nadu leads the all-India average on the Human Development Index (0.531 to 0.302 in 2001) and Gender Development Index (0.813 to 0.676 in 1991) (Government of India 2002; Government of Tamil Nadu 2003). The state enjoys relatively high literacy rates, 73 percent compared to Indias overall rate of 65 percent (Government of India 2001). However, the gender gap in literacy nationally is also reflected in Tamil Nadu. Female literacy is 65 percent compared to male literacy at 82 percent in the state while Indias rates are 54 percent and 76 percent respectively (Government of India 2001). Although the status of women in Tamil Nadu may be considered somewhat advanced relative to other parts of India (Basu 1990), the state is not spared the phenomenon of Indias missing women (Sen 1992) as indicated by a sex ratio of 986 females per 1,000 males (Government of India 2001). It is more equitable than the all-India average of 933 females per 1,000 males, however. Lastly, it is important to note that Tamil Nadus development indicators have also shown steady improvement over time.

42 Tamil Nadus health indicators are also relatively favorable and have improved markedly in the past fifteen years. To begin, life expectancy, the infant mortality rate and the fertility rate in Tamil Nadu are advanced compared to India overall. At 64.1 years, life expectancy in the state is three years greater than the countrys average (Government of India 2002). Fertility declined from 2.5 children per woman in the early 1990s to below the replacement rate, 1.8 children in 2005-2006 in Tamil Nadu, while it hovers at 2.7 nationally (IIPS 2007). Tamil Nadu has also decreased its infant mortality rate to 31 deaths per 1,000 live births, more than halving its rate of fifteen years ago (IIPS 2007). Indias overall infant mortality rate lags at 57 deaths per 1,000 live births, though down from 79 in the early 1990s. Turning to safe motherhood indicators, the state has made a good deal of progress. According to Indias most recent National Family Health Survey, 90 percent of deliveries took place in health care institutions in 2005-06 (IIPS 2007), reflecting an increase from 64 percent during the 1992-93 survey. The most significant change is in the rural institutional delivery rate. Rural institutional deliveries increased from 50 to 87 percent in the past fifteen years while urban rates increased from 91 to 95 percent (IIPS 2007). Indias respected Sample Registration System reports more conservative figures with total institutional deliveries increasing from 57 to 65 percent in Tamil Nadu between 1991 and 2003 (Registrar General of India 2006). Either scenario puts Tamil Nadus institutional delivery rate at more than double the countrys average. Another important maternal health indicator, assistance of a skilled health professional (doctor, nurse or other) at deliveries, increased from 69 to 93 percent between the 1992-93 and 2005-06 surveys (IIPS 2007). On the maternal mortality ratio, Indias Sample Registration System

43 (Registrar General of India 2006) shows a decrease from 167 (95% Confidence Interval 111-224) to 134 (95% Confidence Interval 83-185) deaths per 100,000 live births between the 1999-2001 and 2001-2003 surveys, but large confidence intervals and the short timeframe make it difficult to assess any real trends. State health officials suggested MMR in Tamil Nadu was closer to 100 in 2007. CONCLUSION This chapter begins by drawing on existing research on safe motherhood to provide relevant background information on the scope of the global safe motherhood crisis, technical causes of maternal mortality and recommended interventions to alleviate the problem. This discussion makes two central contributions to this thesis: it describes the basis for selection of the dependent variable, access to maternal health care, used in this study and it identifies several conditions that affect it. Regarding the dependent variable, rates of access to safe motherhood services are the best available indicators to compare the state of maternal health care across states. Maternal mortality ratios are another important indicator, especially for describing the scope of the problem globally and nationally, but large confidence intervals render them unreliable indicators to compare across states or to assess trends. This review of the literature also suggests that it is important to consider several health systems indicators that affect the availability and accessibility of safe motherhood services, such as human resources, safe drug supply, equipment, facilities and quality of management. The empirical chapters of this thesis discuss how these play into the policy process and affect access to safe motherhood services in the states.

44 This chapter also identifies potentially significant causal factors related to the structure of responsibility for health policy in India. The devolution of health policy to the state level with the national level retaining a role in setting broad goals and providing some funding to the states suggests that political factors at the state level and between levels of government might be important. Indeed, the nature of party politics at the state level varies between the states and between the state and national levels, facilitating the policy process in Tamil Nadu and erecting barriers in Karnataka. These factors are suggested particularly by dynamics surrounding the recently introduced National Rural Health Mission policy by the national coalition government. Lastly, the case profiles identify an anomaly in the Karnataka case that deserves some attention as a variable potentially affecting the policy process and the outcomes of interest in that state. Districts in the northeast region of the state feature very low safe motherhood and other socio-economic development indicators. The logical assumption is that low development in that region affects low demand for safe motherhood services. Evidence presented in the empirical chapters provides an alternative explanation, suggesting that there is unmet demand due to political and administrative factors affecting inadequate service provision in the region.

45

CHAPTER 3

METHODS

46 METHODS OF EXPLORATION In-depth case studies are employed in this research to enable collection of evidence that may lead to new theoretical insights and hypotheses with respect to questions of policy processes, as well as to inform an under-explored line of inquiry in the literature on public health policy in developing countries. Following Yin (2003), exploratory case studies may appropriately be employed to answer what, how and why varieties of research questions. Here, how do policy processes shape differing rates of access to safe motherhood services in two South Indian states? Why do women from more vulnerable groups access publicly provided safe motherhood services at significantly higher rates in one state than the other? What can this tell us about important factors and relationships in policy processes? Case studies are used to capture historical and contemporary evidence that may contribute to answering the research questions and to generating hypotheses (Yin 2003). Case studies employ multiple sources of data for purposes of triangulation (Miles & Huberman 1994; Yin 2003). Case studies enable incorporation of documented historical evidence that is important to this study from such sources as government and donor reports and documents, policy directives, media reports and other archival data. They also facilitate collection of contemporary evidence that provides crucial insights to the research questions through interviews with key informants and observation of implementation sites that are key to public health policy and implementation. Qualitative methods, and case studies in particular, allow for investigation of phenomena within their real-life context and recognize that context as integral to informing the question at hand (Miles & Huberman 1994; Yin 2003).

47 Lastly, this research follows a logic of replication by examining two cases, allowing for analytic insights to be drawn from their comparison and improving the external validity of the work (Yin 2003). Although the exploratory nature of this research does not lend itself to broad generalizations, the comparative case study design may help to establish scope conditions for applicability of propositions derived from the study (George & Bennett 2004). CASE SELECTION Several factors played into the selection of cases for this research. India, accounting for a quarter of global maternal deaths, appealed based on the scope of the problem but that was neither the only nor the determining factor in selecting the country as the context for this work. The federal system in which health policy is a shared responsibility between the national and state governments and the wide variation in safe motherhood indicators between Indias states were important considerations. Briefly, health policy tends to be led from the national level while states are responsible for implementation. That safe motherhood outcomes vary amongst the states while national policy direction is uniform begs the question of what differs at the sub-national level to influence these disparities. This juxtaposition provided an important opportunity to explore the research questions, particularly at a sub-national level of analysis where contextual influences might be more readily observed. From the sample of possible cases all twenty-nine of Indias states I selected cases that were relatively comparable along the lines of social and economic development, yet varied on safe motherhood indicators. It was important to select cases in this way to control to the extent possible for the potentially powerful alternative

48 explanations that social and economic development could provide for variation in maternal health outcomes in the comparison. To put this plainly, this study takes as a premise that social and economic development shapes overall safe motherhood outcomes. It is only by holding these as constant as possible that other explanations, such as policy processes, may be discovered. The influence of policy processes on maternal health outcomes would not be so readily observable in a comparison of cases that varied widely on social and economic development. Table 3.1 below gives a sense of the variation in safe motherhood outcomes between Indias major (large population) states. A study that asked a more general question about what factors explain why some states do better than others on safe motherhood performance might select cases with the greatest variation in maternal health outcomes, such as Uttar Pradesh and Tamil Nadu. But extreme poverty, low female literacy and low human development in Uttar Pradesh are difficult to sort out from poor governance, political instability and other factors in policy processes in explaining maternal health outcomes in the state. States that are relatively comparable along these lines, that are known for relatively good governance and social progressivism form a better basis for addressing the research question at hand. Two states that are more appropriate for comparison and that were selected for study in this research are Tamil Nadu and Karnataka, two neighboring states in South India that have reputations for being relatively progressive and economically advanced. I was initially interested in Tamil Nadu because of the attention it has attracted in national safe motherhood policy circles for its leadership in this policy arena. In the end, both cases were selected for their potential to reveal important factors influencing progress on

49 safe motherhood goals. Because the cases were profiled extensively and the bases for their comparison discussed in the first two chapters, they are not discussed further here in these respects. Table 3.1 Key safe motherhood indicators for India and major states India & Major States Maternal Mortality Ratio 2001-3 All-India Uttar Pradesh / Uttaranchal Rajasthan Karnataka West Bengal Maharashtra Tamil Nadu Kerala 445 228 194 149 134 110 12 39 32 45 64 89 32 67 43 90 90 97 33 87 78 85 96 98 76 91 93 99 99 100 301 517 1992-3 26 15* 2005-6 41 22 1992-3 65 35* 2005-6 77 67 Institutional Delivery (%) Any Antenatal Care (%)

Sources: Maternal mortality ratio as reported by the Sample Registration System report for 2001-3 (Registrar General of India 2006); Institutional delivery and antenatal care rates as reported in National Family Health Survey fact sheets for 2005-6 (IIPS 2007).

50 DATA This research is informed by a number of data sources for purposes of triangulation, as well as to understand the phenomena in historical and contemporary perspective. The safe motherhood picture is not easily captured by any one data source or evidence from any one point in time. It is an evolving picture couched in an everchanging set of social, political and administrative circumstances. Recognition of the issue as a significant problem throughout the developing world first emerged in the mid1980s among international actors, such as the UN agencies. This facilitated revitalization of national policy attention to the issue of maternal mortality in India in the 1990s and into the 2000s; hence, data focus on the contemporary period of the past fifteen years. Interviews with actors close to maternal health policy and implementation are crucial to understanding influences on safe motherhood outcomes during this period. Primary data sources, including interviews with key informants and observations at policy and implementation sites, are discussed first in this section. Secondary data sources are discussed next. They importantly inform our understanding of how safe motherhood outcomes have changed over time, what policy measures have been taken to address the problem, and roles of political, administrative and social influences in facilitating or hindering progress in the states. Primary data Interviews with key informants and observations made on site visits to rural health facilities and agencies overseeing public health management form the backbone of this research. Over 14 weeks in the fall of 2006 and spring of 2007 I interviewed 138 individuals in two states and visited 54 health facilities in eight districts. I interviewed 31

51 state and district health officials, nearly evenly split between the two states, in order to gain the insights of policymakers and managers overseeing maternal health policy and implementation. Key informants included a former health secretary, a health commissioner, directors of public health (Tamil Nadu) and health and family welfare (Karnataka), reproductive and child health directors and program officers, and district health officers and other management staff. Many of these interviews took place at the offices of the primary state agencies responsible for public health in the states capital cities with individuals currently occupying relevant positions the Department of Public Health in Chennai, Tamil Nadu, and the Department of Health and Family Welfare in Bangalore, Karnataka. Other key informants in this category provided historical insights to dynamics in these agencies and the broader political and social settings based on their experience occupying public health offices during the period of interest. Twenty-two informants representing donor agencies, nongovernmental organizations and academic institutions provided similar historical and contemporary insights to the research questions, each based on her or his area of expertise. These informants represented such agencies as UNICEF, Danida and the World Bank, as well as local and international nongovernmental organizations and academic institutions. The expertise of these informants often spanned the two states and national-level public health policy dynamics. Some had backgrounds working for or in partnership with state health agencies in the area of reproductive and child health and they contributed significant insights to the historical and current limitations and potential for public health agencies in the states to effectively address safe motherhood issues.

52 An important contribution of this research is that it takes into account perspectives of multiple actors from top leaders to frontline service delivery personnel and from inside public agencies and out. I interviewed 72 frontline service delivery workers, nearly all on site visits to rural health facilities in four districts in each state. Nearly two-thirds of these interviews were with service delivery personnel in Tamil Nadu. The differing proportions of interviews with service delivery personnel in the two states to some extent reflect greater interest in and support for my investigation from the Department of Public Health in Tamil Nadu. In Tamil Nadu, I selected the three main districts I wanted to visit Vellore, Dharmapuri and Theni. The director of public health notified district officials of my visit and asked them to assist with arrangements to visit rural health facilities. In most cases, I then accompanied supervisory staff making regular rounds. My three focal districts were rural districts, all selected because they had seen their safe motherhood indicators improve substantially in recent years and they provided a good opportunity to collect evidence to explain why. I also visited a semi-rural district on the outskirts of the capital, Poonamalee, a health district that had relatively good safe motherhood indicators. I spent only one day in Poonamalee, but two to three full days in each of the other districts. I also conducted brief interviews with 13 women at rural health facilities who had current or past experience with maternal health services in the districts. In Karnataka, my site visits to four rural districts in the state received less coordinated support from the state level. I was able to visit the four districts I selected in the state and go on rounds to rural health facilities with district staff members, but visited fewer facilities in a shorter period of time and met with fewer doctors and nurses in the field than in Tamil Nadu. Absenteeism at health facilities in Karnataka contributed to

53 lack of access to informants in rural health facilities. The data collected in Karnataka suffers little, for this. Across districts, informants identified the same factors at work and I made similar observations at implementation sites. Additionally, the lower level of support for my investigation and the types of observations I made in the field there, including finding staff absent or vacant from posts, are important pieces of evidence themselves. The four districts I visited in Karnataka included Bangalore Rural, Mysore, Kodagu and Bellary. I selected Kodagu as a district with unexpectedly good safe motherhood indicators despite its rugged and remote terrain, Mysore and Bangalore Rural as middling districts, and Bellary as one of the lowest performing districts in the state. I spent two to three days in each of these districts meeting with district health officials and visiting rural health facilities. In both states, my visits to rural health facilities were dominated by visits to Primary Health Centers (PHCs) and Health Sub-centers7 (HSCs). Health Sub-centers are located at the bottom of the health service delivery hierarchy, at the village level, serving a population of approximately 5,000 (Government of India 2006). They are staffed by Village Health Nurses8 (VHNs) whose primary responsibilities include maternal health services, immunization programs and treating minor ailments. Primary Health Centers are rural health units that typically cover the area of five to six Health Sub-centers, a
7

These are called Sub-centers in Karnataka and Health Sub-centers in Tamil Nadu. They serve the same function and I use the Karnataka designation throughout for consistency and clarity, although direct quotes may reflect this small difference. 8 I try to use the term Village Health Nurse to refer to the nurses posted to Health Subcenters, as is the designation in Tamil Nadu. The title for this position has changed periodically in Karnataka, but Village Health Nurse is still understood to mean nurses posted to Health Sub-centers in the state. Some data sources do not differentiate between village health nurses and auxiliary nurse midwives and the latter referent is used. It should be assumed that auxiliary nurse midwife includes personnel functioning as village health nurses, and certainly performing maternal health service duties.

54 population of approximately 30,000 (Government of India 2006).9 At these facilities, doctors and nurses that conduct antenatal, postnatal and delivery services made up the pool of key informants. I conducted 35 individual interviews with service delivery personnel in Tamil Nadu, most during visits to 31 health facilities in the state. I also conducted six focus groups with 23 participants, most of whom were Village Health Nurses and a few nursing supervisors. In Karnataka, I conducted 14 individual interviews with service delivery personnel and observed 23 health facilities. A handful of interviews and site visits were conducted at higher-level public health facilities, including district hospitals, where I spoke with obstetrician-gynecologists and nurses. Interviews with service delivery personnel typically lasted between 15 and 45 minutes, although a few took place over several hours when medical officers accompanied me on site visits in districts. These key informants were crucial to learning what and how maternal health policies were being implemented in the states and how conditions for safe motherhood had changed over time. Interviews with state and district officials and representatives of donor and nongovernmental agencies lasted between 30 minutes and two hours, averaging a little over an hour. Some key informants in each state shared their time and insights on two or more occasions, enabling me to ask follow-up questions and delve more deeply into key issues. A purposive sample of key informants was selected through review of donor and government documents, as well as referrals from key informants to identify individuals

This population service level is the standard set at the national level for coverage in rural plains areas. Another category exists for less accessible areas and more difficult to reach populations hilly/tribal/difficult area. The standard for Primary Health Centers in these areas is 20,000 population coverage and for Health Sub-centers 3,000 population (Government of India 2006).

55 that were centrally involved in safe motherhood policy and implementation in the states. Interviews were semi-structured, focusing on the common question of key factors influencing safe motherhood outcomes in each state, but allowing for flexibility to gain the insights of each individual informant. Interview questions shifted according to the expertise of informants. For example, service delivery personnel were asked to talk about what maternal health services they provided, to give concrete examples of how they handled the last maternal health complication they encountered, and whether and how conditions for maternal health had changed over time. I used open-ended questions to draw out the unique knowledge of each informant. Interviews were extremely important because the data they provided was not available in the written record. I recorded extensive notes during all interviews and followed up interviews by recording my observations of the settings, tones, attitudes and other relevant pieces of data as soon after each interview concluded as possible. Often, this was within an hour or two. Sometimes it was at the end of the day. I also created contact summary sheets (Miles & Huberman 1994) within 24 to 48 hours after each interview with a public official or representative of a donor or nongovernmental agency. Contact summary sheets covered main issues and themes, summary data on key questions, new data or points that seemed particularly salient, issues to follow-up on and a brief discussion of how that contact contributed to answering the research question. I used these summary sheets to identify important themes and develop a coding system to identify patterns and later map out key variables and their possible relationships with safe motherhood outcomes.

56 Secondary data In addition to interviews, I carefully reviewed and cross-checked multiple documents and reports to verify interview and observation data, as well as to understand historical developments relevant to safe motherhood more completely. Documents included government reports and policy documents; health reports; reports from bilateral and multilateral donors; national, state and local government health plans; demographic, health and other surveys; published research on reproductive health and maternal mortality; and national and state media reports on health and safe motherhood. I obtained these documents from donor, NGO and university libraries, by asking key informants, and through Web-based searches. I reviewed these documents and used them to inform interview questions, check facts and construct a timeline for policy and implementation activity. They also helped to document progress on safe motherhood and other developmental indicators. I compared interview notes and written documents to identify major factors facilitating or hindering progress on safe motherhood. Documents and reports were important to providing verification of factual data. They also proved to be of assistance in cases in which key informants were unable to recall the details or timing of significant developments accurately. VALIDITY I used a number of strategies to address issues of validity in this research. To begin, I used multiple sources of evidence for purposes of triangulation an important step in case study research for construct validity (Yin 2003). Using multiple sources of evidence helped to address issues of response bias and reflexivity, as well. My expertise

57 on safe motherhood and Indian administrative systems also helped me to identify anomalies in reporting and instances when it was likely I was being told what I wanted to hear rather than facts or genuine perceptions. I was able to lessen the effects of these sources of bias by asking pointed follow-up questions, consulting other informants and referring to other sources of data. I used the research proposal as a case study protocol that included an overview of the project, procedures to be followed during field research, questions to guide the case studies, and an outline of the research report as suggested by Yin (2003) to ensure reliability and address issues of validity. The constructs I developed to operationalize key concepts in the research proposal served to guide data collection and measurement, contributing to the construct validity of the research. I also created a research database that includes interview summary sheets, a document database, interview and observation notes, and memos to improve reliability of the studies (Miles & Huberman 1994; Yin 2003). Lastly, as previously noted, I used a logic of replication in designing the research. The Tamil Nadu and Karnataka cases were conducted separately, using the same case study protocol, to contribute independent evidence and findings suitable for comparison and some degree of generalizability to questions about policy processes and their relationships to outcomes.

58

CHAPTER 4

SOCIO-CULTURAL DYNAMICS & SAFE MOTHERHOOD

59 INTRODUCTION Until the past century, in South India and around the world, giving childbirth in the home with experienced family members or lay midwives in attendance was the norm. People relied on multiple systems of medical knowledge, social customs and spiritual beliefs to guide health decisions and care and this remains the case in communities in South India and many regions of the world today. Local knowledge and communities contribute to the strength of health systems in important ways, yet a sound body of research tells us that access to quality maternal health and emergency obstetric care in the biomedical tradition is crucial to preventing maternal mortality. And, indeed, evidence suggests that South Indian women are increasingly accessing or attempting to access modern maternal health care. Between 1992-3 and 2005-6, institutional deliveries increased from 64 to 90 percent in Tamil Nadu and 39 to 67 percent in Karnataka (IIPS 2007). But not all groups have access to safe motherhood services on an equitable basis. Women from lower economic groups, with less education, from Scheduled Castes and Tribes and those residing in rural areas access safe motherhood services at lower rates than their more resourced counterparts. This study was designed to control as much as possible for the effects of these types of macro level factors by selecting states that were relatively comparable on social, cultural and economic indicators. An interesting finding emerges in this chapter though. One of the key differences between the cases is that women from more vulnerable groups accessed safe motherhood services at greater rates in Tamil Nadu than in Karnataka. This helps to explain differing outcomes between the states and opens an interesting question what explains Tamil

60 Nadus relatively more equitable safe motherhood outcomes? The answer has to do with political and administrative dynamics that have shaped health systems that provide more equitable access to safe motherhood services in Tamil Nadu than in Karnataka. These are examined in the two chapters that follow this one. This chapter presents evidence concerning the relationships between select social, cultural and economic factors and safe motherhood outcomes in each state and compares these relationships across the states. The evidence suggests such factors as economic status, education level and geographic location play important roles, but constitute partial explanations for the disparities in safe motherhood outcomes between the states. It suggests policy processes affecting the governmental programs of action to promote safe motherhood in each state plays an important role in making up this difference and requires further investigation. ECONOMIC STATUS To begin, a recent survey highlighted the import of economic factors in determining access to maternal health care in India. Both women and men identified economic barriers as important factors in their decisions against delivery in a health institution (IIPS 2007). Twenty-six percent of women and 24 percent of men reported cost and 11 and 7 percent respectively reported the related matters of distance and lack of transport as reasons for choosing home over institutional delivery (IIPS 2007). In the same survey, the third National Family Health Survey, women in the lowest income quintile reported five times the number of problems accessing medical advice or treatment for themselves as their counterparts in the highest income quintile (IIPS 2007). They identified such issues as getting money for treatment, distance to health facility,

61 having to take transport, and concern that no provider or drugs would be available at higher rates than women in any other income quintile (IIPS 2007). It is no surprise then that womens economic status in Tamil Nadu and Karnataka was associated with outcomes on various safe motherhood indicators. The 1998-9 and 2005-6 National Family Health Survey reports for Tamil Nadu and Karnataka (IIPS 2001a and b, 2008a and b) showed higher rates of antenatal care, institutional deliveries and safe deliveries amongst women with progressively higher economic status. The surveys used a common standard of living index across states based on such indicators as household amenities (e.g. drinking water, availability of toilet facilities, etc.) and ownership of durable goods (e.g. vehicle, television, etc.) to indicate relative access to resources among lower, middle and higher groups. This was the best measure available to show the relationship between economic status and safe motherhood access in the states. The data are comparable across states within the same survey year, but not across survey periods because the earlier survey distinguished three and the later survey five levels of the wealth index. The 1992-3 survey did not report wealth data. In 1998-9, women in the highest economic group had a home delivery rate of only 3 percent, while women in the medium and lower groups had home deliveries at rates of 16 and 31 percent respectively in Tamil Nadu (IIPS 2001). In Karnataka, women in the highest to lowest groups experienced home deliveries at rates of 21, 45 and 68 percent in 1998-9 (IIPS 2001). The finer grain analyses of the 2005-6 surveys show the same trends as depicted in Table 4.2 below. In addition, the surveys reported higher rates of tetanus vaccination and distribution of iron supplements, key antenatal care components, amongst women of higher economic status in both states (IIPS 2001a and b, 2008 a and

62 b). They showed the same trend in post-partum care check-ups within two months of delivery in 1998-9 (IIPS 2001a and b) and within two days of birth in 2005-6 (IIPS 2008a and b). Tables 4.1 and 4.2 display these results and highlight important differences between the states. Table 4.1 Economic status and safe motherhood process indicators in Tamil Nadu and Karnataka, 1998-9
Two or more Home delivery (%) Any antenatal care tetanus (%) vaccinations (%) Economic Karnataka group Low Medium High 68 45 21 Nadu 31 16 3 75 90 98 Tamil Karnataka Nadu 98 99 99 61 78 92 Tamil Karnataka Nadu 94 96 99 Tamil

Source: NFHS 2 Surveys, Tamil Nadu and Karnataka (IIPS 2001a and b) The 1998-9 and 2005-6 National Family Health Surveys indicate that across economic groups women in Tamil Nadu had substantially lower rates of home deliveries than women in Karnataka. Importantly, the disparity between institutional delivery rates of women in the highest and lowest wealth groups was substantially smaller in Tamil Nadu than Karnataka, as well. Tamil Nadus rates showed more equitable access to safe motherhood services across economic groups than Karnatakas in both survey years. Antenatal care evidence shows this clearly with rates of 92 percent or more for women in all economic groups in Tamil Nadu and 94 percent or more receiving two or more tetanus vaccines. The gaps between economic groups in Karnataka were much wider on these

63 indicators. These figures suggest that it is necessary to look beyond economic indicators to understand why women in Tamil Nadu particularly those in lower income groups have better access to safe motherhood services than women in Karnataka. Table 4.2 Economic status and safe motherhood process indicators in Tamil Nadu and Karnataka, 2005-6
Three or more Home delivery (%) antenatal care visits (%) Economic Karnataka group Lowest Second Middle Fourth Highest 61 53 41 17 4 Nadu 32 16 14 5 2 56 65 81 90 97 Tamil Karnataka Nadu 92 94 94 98 100 61 65 78 88 93 Tamil Karnataka Nadu 94 96 95 96 98 Two or more tetanus vaccinations (%) Tamil

Source: NFHS 3 Surveys, Tamil Nadu and Karnataka (IIPS 2008a and b) It is also relevant to look at rates of safe motherhood care in public versus private institutions. The 2005-6 survey did not report on this indicator, but in 1998-9, women with a lower standard of living accessed public delivery care at greater rates than women from higher economic groups in both states (IIPS 2001a and b). In Tamil Nadu, women from lower economic groups accessed public delivery services at a two to one ratio over private services. In Karnataka, the ratio was five to one, suggesting public sector health services were vitally important for safe motherhood for women in lower income groups in both states. It should be noted that although lower income women in Karnataka relied

64 more heavily on public than private services in some ways, 45 percent of deliveries for low-income women occurred in Tamil Nadus public sector compared with 26 percent in Karnataka. This was partly because overall institutional delivery rates were higher in Tamil Nadu, but also raises the question of why Tamil Nadus public health system served such a greater proportion of safe motherhood needs between the states and that in advance of specific and coordinated attention to the issue from policy actors. On a related economic point, it is important to observe that the incidence of poverty decreased significantly in both states in recent years. Since 1983, the proportion of Karnatakas population living below the poverty line decreased by nearly half, from 38 percent to 20 percent in 1999-2000 (Government of India 2002). In Tamil Nadu, it decreased from 52 to 21 percent in the same period (Government of India 2002). Urban and rural poverty rates were nearly even in Tamil Nadu, while Karnatakas urban population suffered a greater poverty rate (25 percent) than its rural population (17 percent) in 2000-2001 (Government of Karnataka 2006; Government of Tamil Nadu 2003). Poverty lines in the two states hovered around Rupees 325 to 350 in rural areas and Rupees 550 to 600 in 2004-5 in the two states around twenty-five cents per person per day in rural areas and less than fifty cents per person per day in urban areas (Government of India 2007). These figures give a sense of the financial barriers lowincome families faced to accessing health care, as well as what it might mean for families to break out of poverty. Navaneetham and Dharmalingams (2002) quantitative analysis of 1992-3 National Family Health Survey data for South India confirmed the significance of economic status for antenatal and delivery care outcomes in Karnataka and Tamil Nadu. Broader trends in economic growth and the decreasing incidence of poverty in the

65 states have likely had an impact on safe motherhood outcomes; however, consistently better rates of access to maternal health services in Tamil Nadu when higher poverty rates prevailed and in their more recent equivalency to Karnatakas suggest that economic factors constitute only a partial explanation for variation between the states. EDUCATION & LITERACY Educational indicators are also associated with barriers to womens access to health care and safe motherhood. In the most recent National Family Health Survey, across six levels of education ranging from none to twelve or more years completed, Indian women with successively lower levels of education reported greater problems in accessing medical advice or treatment for their own needs across all categories (IIPS 2007). Categories included: getting permission to go for treatment, getting money for treatment, distance to health facility, having to take transport, not wanting to go alone, concern that no provider would be available, and concern that no drugs would be available. Nearly 60 percent of women with no education reported at least one problem in accessing health care. This amounts to a significant problem considering that female literacy in India was only 56 percent, in Karnataka 57 percent and in Tamil Nadu 65 percent in 2001 (Government of India 2002). Navaneetham and Dharmalingam (2002) found education levels to be a significant factor in determining antenatal and institutional delivery rates in the two South Indian states based on 1992-3 National Family Health Survey data. According to National Family Health Survey data for 1992-3, 1998-9 and 2005-6, higher levels of education were associated with higher rates of institutional delivery, safe delivery, antenatal care and two of its key components (e.g. tetanus vaccine and provision

66 of iron supplements), and postnatal check-ups in both states (IIPS 2001a, 2001b; IIPS 2008a and b; Population Research Centre 1995). There was some decline in home deliveries in all categories except the highest (high school complete and above) in Karnataka between the 1992-3 and 1998-9 surveys. Data from the 2005-6 survey are not strictly comparable with previous survey data because the educational categories changed. Among illiterate women in Karnataka, home deliveries decreased from 77 to 67 percent; among literate women having education up to middle school from 47 to 39 percent; among those completing middle school from 36 to 31 percent; and among those completing high school, home deliveries were up from 14 to 18 percent between 1992-3 and 1998-9. In Tamil Nadu, home deliveries among illiterate women decreased by more than half from 53 to 26 percent. At the second level, home deliveries decreased from 27 to 16 percent; at the middle school complete level from 16 to 10 percent; and at the high school complete level from 10 to 4 percent over the same period. These results are summarized in Table 4.3. Results from the most recent survey are summarized in Table 4.4.

67 Table 4.3: Education levels and safe motherhood process indicators, 1998-9
Any antenatal care Home delivery (%) (%) Tamil Karnataka Nadu Illiterate Literate, < middle 39 school complete Middle school complete High school complete and above 18 4 99 100 95 99 31 10 93 100 82 96 16 96 100 84 96 67 26 77 Karnataka Nadu 98 62 Tamil Karnataka Nadu 98 vaccinations (%) Tamil Two or more tetanus

Source: NFHS 2 Surveys, Tamil Nadu and Karnataka (IIPS 2001a and b)

68 Table 4.4: Education levels and safe motherhood process indicators, 2005-6
Two or more 3 or more antenatal Home delivery (%) care visits (%) vaccinations (%) Tamil Karnataka Nadu No 64 education <5 years 32 complete 5-9 years 25 complete 10 or more 10 years complete 4 94 100 90 96 12 89 96 84 96 15 78 96 77 97 25 57 89 63 94 Karnataka Nadu Tamil Karnataka Nadu Tamil tetanus

Source: NFHS 3 Surveys, Tamil Nadu and Karnataka (IIPS 2008a and b) Although women with lower levels of literacy and education did not access safe motherhood services at the same rate as their more educated counterparts, their rates of access increased during the 1990s. The fifty percent decrease in home deliveries among illiterate women in Tamil Nadu is remarkable and begs explanation beyond the typically snailish pace of social change processes. As on economic indicators, the gap between lowest and highest educational groups in Tamil Nadu was much smaller than between the same groups in Karnataka, a trend reflected in all three sets of survey results. This suggests that something other than literacy or education level played an important role in

69 determining rates of access to safe motherhood services. Similar patterns on antenatal care indicators showing high coverage across groups in Tamil Nadu and gaps between lower and higher educational groups in Karnataka back up this suggestion. Additionally, mirroring the patterns of lower economic groups, women with lower education levels in Tamil Nadu delivered in public health institutions at a rate of 40 percent compared to 24 percent in Karnataka, suggesting the public sector component of Tamil Nadus health system functioned more equitably for safe motherhood across social groups than did Karnatakas. This analysis examines why and how in subsequent chapters. It is also important to observe trends toward increasing female literacy in both states. In Karnataka, overall female literacy increased from 17 to 25 to 33 to 44 to 57 percent each decade between 1961 and 2001 (Registrar General of India, Census of India, various years cited in Government of Karnataka 2006). In Tamil Nadu, female literacy increased from 51 to 65 percent between 1991 and 2001 (Government of India 2002). Increasing female literacy in the states likely played a role in improving womens access to safe motherhood services, but it does not tell the whole story. In Tamil Nadu, at the district level, the relationship between female literacy rates and institutional delivery rates did not hold consistently. Two of the districts with the lowest female literacy rates, Erode and Namakkal at 55 and 57 percent respectively, featured institutional delivery rates in the ninetieth percentile in the top third of the states districts (Census of India in Government of Tamil Nadu 2003; Government of India 2004). Even with one third of Tamil Nadus districts having female literacy rates under 60 percent, only one of Tamil Nadus districts featured an institutional delivery rate under 70 percent and fewer than a handful were below 80 percent out of thirty districts. This suggests that literacy and

70 education levels in Tamil Nadu are important, but not enough to fully explain the states relatively advanced safe motherhood indicators. In Karnataka, female literacy rates were more closely linked with institutional delivery rates at the district level. The three districts with the lowest female literacy (under 40 percent in Koppal, Raichur and Gulbarga) featured among the lowest institutional delivery rates in the state (21 to 31 percent) figures on par with the lowest performing states in the country and far lower than any district in Tamil Nadu (Government of Karnataka 2006; Government of India 2004). Correspondingly, Karnatakas three districts where female literacy was over 75 percent (Bangalore Urban, Dakshina Kannada and Udupi) featured the only institutional delivery rates that were over 90 percent in the state (Government of Karnataka 2006; Government of India 2004). These figures are spread among twenty-seven districts. The presence of this pattern in Karnataka but not in Tamil Nadu suggests intervening factors might be at work to affect more equitable access to care in Tamil Nadus health system. GEOGRAPHY The preceding observations at the district level raise the issue of geographical factors. Even within the context of South India, a geographic region often contrasted socio-culturally and developmentally with North India, geographic variations in safe motherhood indicators can be observed between rural and urban areas, as well as among district-level groupings. In Karnataka, regional disparities associated with the geopolitical origins of its northern and southern districts are readily apparent. The districts of Hyderabad-Karnataka (including Koppal, Raichur, Bellary, Gulbarga and Bidar) and Bombay-Karnataka (including Haveri, Gadag, Bagalkot, Bijapur, Dharward and

71 Belgaum), names that reflect their associations prior to the states formation in 1956, make up Karnatakas northern region. They are concentrated amongst the districts with the lowest safe motherhood and other development indicators in the state (Government of Karnataka 2006; Government of India 2004). These areas featured lower levels of human development, including health, education and economic indicators, at the states formation than the states southern and coastal districts and continue to do so (Government of Karnataka 2006). Some informants to this research suggested that the status of women is relatively lower in some districts of the northern region, restricting their access to health services (interview nos. 14, 84). A study by George and colleagues (2005) discussed further in Chapter 6 suggests that women and their families in this region do attempt to access safe motherhood services, especially once complications arise, but services are frequently unavailable or inaccessible. It is logical to conclude that this would affect lower safe motherhood indicators in the northern regions, but two notes of caution are in order. First, a 2006 World Bank report found political and administrative neglect to be a significant factor affecting slow progress for human development in Karnatakas northern regions, offering another explanation for this phenomenon and one that is examined further in subsequent chapters. The second is that if the geographical legacy of disparities in human development were so intractable as it appears in Karnataka institutional deliveries in its lowest performing districts actually decreased slightly between 1998-9 and 2002-4 from 22 to 21 percent (Government of India 2004) then they should resist change in other settings. But this was not the case in Tamil Nadu. Between 1998-9 and 2002-4, institutional deliveries increased from 55 percent in the former South Arcot district (lowest indicators in the

72 state) to 74 percent in now Viluppuram and 82 percent in present day Cuddalore (Government of India 2004). Other historically under-developed areas also showed marked improvement. This suggests that some factor other than a geographic legacy of relatively low human development might be at work in determining how health systems function for safe motherhood. Another important socio-geographic dynamic is observed in rural-urban disparities in safe motherhood indicators. Returning to recent National Family Health Survey results, Indian women residing in rural areas reported at ratios of two and three to one over their urban counterparts facing significant problems accessing medical advice or treatment for their own needs (IIPS 2007). Women and men reported opting for home delivery because it cost too much or was too far away or transport unavailable to go to a health facility for delivery at greater rates than their urban counterparts (IIPS 2007). Karnataka and Tamil Nadu both exhibited substantial gaps in institutional delivery rates between rural and urban areas in the 1990s. That gap is closing at a rapid clip in Tamil Nadu. In 2005-6, rural and urban institutional delivery rates in Karnataka stood at 57 and 85 percent respectively and in Tamil Nadu 87 and 95 percent (IIPS 2007). See Table 4.5 for a comparison of the states institutional delivery rates between 1992-3 and 2005-6. The pattern of rural-urban inequities holds across antenatal and postpartum care indicators (IIPS 2001a and b), though it is narrower in Tamil Nadu than in Karnataka (IIPS 2008a and b) as depicted in Table 4.6. As with economic and educational indicators, rural women in Tamil Nadu had better safe motherhood outcomes than their peer group in Karnataka. And, in recent years, the gap between rural and urban groups in

73 Tamil Nadu shrank to a difference of 12 percent versus 30 percent in Karnataka, making for more equitable safe motherhood outcomes in the former state. Table 4.5: Rural and urban institutional delivery rates in Tamil Nadu and Karnataka, 1992-3, 1998-9 and 2005-6
Karnataka 1992-3 Urban Rural 68 27 1998-9 79 39 2005-6 85 57 1992-3 91 50 Tamil Nadu 1998-9* 93 73 2005-6 95 87

Note: *Chennai, the urban capitol of Tamil Nadu, was reported separately in 1998-9 reporting. Its institutional delivery rate was 99 percent. Source: NFHS1 and 2 reports for Tamil Nadu and Karnataka, NFHS 3 India report (IIPS 2001a and b, 2007; Population Research Centre 1994, 1995) Table 4.6: Rural and urban antenatal and postnatal care rates in Tamil Nadu and Karnataka, 2005-6
3 or more antenatal care visits (%) Karnataka Urban Rural 82 55 Tamil Nadu 98 94 Postnatal check-up within 2 days (%) Karnataka 74 63 Tamil Nadu 94 89

Source: NFHS 3 surveys for Tamil Nadu and Karnataka (IIPS 2008a and b) Note: Data for Chennai, the urban capital of Tamil Nadu is reported separately. The corresponding antenatal care rate in Chennai was 97.7 percent and for postnatal check-up 98 percent.

74 It is also important to observe that 1998-9 National Family Health Survey data indicated consistent disparities between the antenatal care components and advice women received in rural and urban areas across both states (IIPS 2001a and b). This is important because it is one of the best quality of care indicators available and it is not reported on across other social indicators. Antenatal care components included weight measured, height measured, blood pressure checked, blood tested, urine tested, abdomen examined, internal examination, X-ray, sonogram or ultrasound and amniocentesis. Antenatal advice included diet, danger signs of pregnancy, delivery care, newborn care and family planning. Urban women received all of these at greater rates than rural women in Karnataka. The same was true in Tamil Nadu with the exception of amniocentesis. Women in Tamil Nadu, across rural and urban dimensions, reported receiving antenatal care components and advice at greater rates than women in Karnataka with the exception of advice on danger signs (urban and rural) and receiving a blood test (urban). The gap between rural and urban women on these measures was also smaller in Tamil Nadu than in Karnataka with the exception of receiving advice on danger signs during pregnancy. These suggest that rural-urban dynamics are important determinants of access to safe motherhood services, as well as the quality of safe motherhood services. That rural-urban gaps were less pronounced in Tamil Nadu suggests other factors might be at work in alleviating these health system disparities. Aggregate quality of care indicators from the 2005-6 National Family Health Survey showed that Tamil Nadu closed the gap between the states on advice about danger signs of pregnancy. Forty-eight percent of women in Karnataka and 84 percent of women in Tamil Nadu reported receiving information about specific pregnancy complications,

75 including vaginal bleeding, convulsions, prolonged labor, or where to go if complications arose (IIPS 2008). Overall, Karnataka held fairly steady on this indicator from 1998-9, but Tamil Nadus health system showed substantial improvement, nearly doubling the percentage of women receiving this kind of advice from 1998-9 (IIPS 2001; IIPS 2008). These later developments are consistent with evidence of stronger subsystem influences on behalf of safe motherhood in Tamil Nadu than Karnataka between the 1998-9 and 2005-6 surveys, phenomena discussed at length in the chapter on public administration and safe motherhood. CASTE, CLASS, RELIGION & GENDER Religion and social class or castes are other important factors to consider in the Indian context. Returning to recent health survey data, women identifying with Scheduled Tribe and Scheduled Caste groups reported more problems accessing medical advice or treatment for themselves than women from other social groups, with tribal women reporting the most problems (IIPS 2007). Navaneetham and Dharmalingam (2002) found Scheduled Tribe and Caste group to be a significant factor affecting womens use of institutional delivery services in the states in the early 1990s. In Karnataka, 58 percent of tribal women reported home deliveries in 2005-6, down from nearly 70 percent in 1998-9 and 73 percent in 1992-3 reporting (IIPS 2001; IIPS 2008; Population Research Centre 1995). Their rates of receiving any antenatal care declined from 80 percent to 72 percent and two or more tetanus vaccinations declined from 59 to 56 percent between the first two surveys. In 2005-6, 62 percent of tribal women reported

76 three or more antenatal care visits10 and there was a 60 percent tetanus vaccination rate (IIPS 2008). Karnatakas Scheduled Tribes comprised 6.6 percent of its population in 2001 (Census of India 2001). They made up 10 to 18 percent of the population in eight districts with overall institutional delivery rates ranging from 21 to 72 percent. These groups tend to live in more remote areas, have lower educational indicators, and have less access to economic resources and health services (Government of Karnataka 2006). Tribal groups made up only one percent of Tamil Nadus population in 2001 and safe motherhood indicators were not reported for this group in the survey, but my observation was that they faced the same social challenges as tribal groups in Karnataka and were also likely to be relatively less well served by the health system. Low rates of access to maternal health services among tribal groups likely factor in Karnatakas lagging performance to a greater extent that in Tamil Nadu because of their greater representation in the population. Karnatakas tribal population is still relatively small, however, and considering the greater proportion of Scheduled Castes in Tamil Nadus population this factor fails to form a sufficient explanation for the relatively greater disparities observed in overall safe motherhood outcomes between the states. Scheduled Caste groups made up 16 percent of Karnatakas population in 2001 (Government of Karnataka 2006) and 19 percent of Tamil Nadus population in 1991 (Government of Tamil Nadu 2003). These groups also showed lower safe motherhood indicators than other groups (non-Scheduled Caste or Tribe) in both states, although with some greater decrease in home deliveries in the 1990s than the Scheduled Tribe groups in The 2005-6 antenatal care indicator is different from the earlier surveys, making these incomparable over time. The first two surveys reported on any number of antenatal care visits while the most recent survey reports only on a higher standard of 3 or more antenatal care visits.
10

77 Karnataka. Home deliveries among Scheduled Caste groups decreased from 78 to 60 to 46 percent in Karnataka and 54 to 30 to 20 percent in Tamil Nadu over the course of the three surveys (IIPS 2001a and b; IIPS 2008a and b; Population Research Centre 1994, 1995). Between the first two surveys, antenatal care rates stayed about the same in Karnataka, although tetanus vaccination increased from 63 to 69 percent. In Tamil Nadu antenatal care increased from 92 to 99 percent between 1992-3 and 1998-9. In 2005-6 reporting, tetanus vaccination rates increased to 74 percent in Karnataka and decreased slightly to 96 percent in Tamil Nadu (IIPS 2008a and b). Ninety-four percent of Scheduled Caste women in Tamil Nadu received three or more antenatal care visits while the rate of access was 68 percent in Karnataka. Again, we see this more vulnerable group exhibiting better safe motherhood outcomes in Tamil Nadu than in Karnataka. We also see that the public sector disproportionately served women from Scheduled Tribe and Caste groups for institutional deliveries over the private sector. Twenty-five percent of Scheduled Tribe and 26 percent of Scheduled Caste deliveries took place in public health institutions in Karnataka in 1998-9 compared to 6 and 11 percent respectively in the private sector. In Tamil Nadu, 43 percent of Scheduled Caste deliveries took place in public health institutions compared with 25 percent in the private sector. This again suggests that the public sector plays a significant role in maternal health care provision in both states, and begs the question of why the health system functions relatively more equitably for safe motherhood in Tamil Nadu than in Karnataka. Turning to religion, National Family Health Survey reports suggest that among major religions there was less of a difference for womens access to safe motherhood

78 than some of the other social factors addressed here. The Census of India 2001 reported Karnatakas population was 84 percent Hindu, 12 percent Muslim and 2 percent Christian and Tamil Nadus 88 percent Hindu, 6 percent Christian and 5.6 percent Muslim (Government of India 2001). Navaneetham and Dharmalingam (2002) did not find religion to be a significant determinant of antenatal and institutional delivery rates in the two states in the early 1990s. In 1998-9, Hindu and Muslim women in Karnataka delivered in health institutions at about the same rate, 49 and 50 percent respectively. Institutional delivery rates were higher in Tamil Nadu among Hindus and Muslims. Rates increased among Muslim women by some 20 percent and Hindu women by 14 percent in 2005-6 reporting to 71 and 63 percent respectively (IIPS 2008). In Tamil Nadu, Muslim women had slightly higher institutional delivery rates than Hindu women, 84 and 79 percent respectively in 1998-9 and 97 and 87 percent in 2005-6. In 1998-9, Christian women in Tamil Nadu reported delivery in health institutions at a rate of 97 percent. This figure declined to 89 percent in 2005-6 reporting. Institutional delivery rates were little changed (one percent) among Muslim women in Tamil Nadu from 1992-3, but increased by 16 percent among Hindu women and 24 percent among Christian women in the state (IIPS 2001; Population Research Centre 1994). Hindu women had slightly higher rates of antenatal care and tetanus vaccination in both states in 1998-9 (IIPS 2001a and b), but this had reversed by the 2005-6 survey (IIPS 2008a and b). Christian women were not reported on in Karnataka in the earlier surveys and results were not significant in 2005-6. Relationships between religion and safe motherhood indicators in the states appear to reflect overall disparities between the states rather than to play a role in explaining them.

79 Jejeebhoy and Sathars (2001) study of womens autonomy and region in South Asia offers some insights into why religion probably was not a significant factor as suggested by the relatively small variations observed in safe motherhood indicators between major religions. They found, differences are more likely to reflect social systems as marked by region than by nationality or religion. South Indian women appear to enjoy greater levels of autonomy than north Indian women (p. 697). This highlights two important points. First, it serves as a reminder that the selection of two South Indian states for this study serves to control for influence of region, social systems and culture to some degree. As a result, we expect to find these factors playing less significant roles in determining differences in safe motherhood outcomes between these states than we would between states in south and north India, for example. Second, Jejeebhoy and Sathars study suggests gender dynamics are important in the South Asian context and, again, that these are controlled for to some extent by selecting states in a particular region. It bears noting Jejeebhoy and Sathars finding that women's autonomy-in terms of decisionmaking, mobility, freedom from threatening relations with husband, and access to and control over economic resources-is highly constrained in Pakistan and in north and south India (2001, p. 707) in all three of these South Asian regions. Within that context, they suggest that women in South India experience a greater degree of autonomy. The authors do not suggest that women in South India enjoy the same degree of autonomy as their Western counterparts. Rather, their argument that women in South India enjoy relatively more autonomy than women in North India draws on research conducted in India that demonstrates the dominant influence of behavior and norms

80 imprinted by regionally prescribed social systems, and points out that the social systems that characterize the southern region provide women more exposure to the outside world, more voice in family life, and more freedom of movement than do the social systems of the north [Dyson and Moore 1983; Basu 1992; Jejeebhoy 2000] (2001, p. 687). This kind of broad characterization is useful for our current purposes, but should not be overstated as cultural beliefs and social practices vary across households and communities (Malhotra et al. 1995; Van Hollen 2003). Basus (1990) study of cultural influences on health care use by women from Tamil Nadu and Uttar Pradesh forms the critical link between womens autonomy and safe motherhood. In Basus study, the two states represent South and North India respectively. Basu found greater female autonomy to be linked with greater confidence and willingness to seek and interact with health care providers for delivery care (1990). She noted that her findings were consistent with Dyson and Moores (1983) landmark paper On Kinship Structure, Female Autonomy, and Demographic Behavior in India. Importantly, Basu (1990) found no differences in modern medical care usage based on culture (represented by a northern and a southern state) for a number of ailments, but childbirth was an exception. She cited fears of an unfamiliar and hostile environment, as well as coercive family planning practices, as reasons to avoid institutional delivery. She concluded that Tamil women, because they had a relatively greater degree of experience with and exposure to the outside world and modern health services, were more willing to avail themselves of biomedical maternal health services. That might be so in a relative sense, but other scholars have documented womens hesitation to expose themselves to discrimination and mistreatment by the health systems in both Tamil Nadu

81 and Karnataka (Matthews et al. 2005; Van Hollen 2003). Data are not available to assess the comparability of Karnataka and Tamil Nadu on this factor or its implications for safe motherhood outcomes directly, but other scholars work suggests the states should be more comparable on this factor than would be southern and northern states. Thus, womens autonomy might play some role in shaping safe motherhood outcomes, but is another factor that is largely controlled for by selection of states in the same region. Further, although scholars agree that patriarchal social norms can act to suppress the importance of womens health at the household level (George 2007; Hutter 1997; Nichter 1989; Van Hollen 2003), the impact of womens autonomy on safe motherhood outcomes is not so straightforward. To begin, the most recent National Family Health Survey reports that the extent to which Indian women participate in decision making, including about their own health care, visits to family or relatives and making major and daily household purchases, makes little difference in the rates at which they access antenatal, institutional delivery and postpartum care (IIPS 2007). Second, womens participation in decision making about their own health care varies along the lines of economics, education, location and to a lesser extent caste or tribe lower economic and educational indicators and rural residence are associated with lower participation in health decision making (IIPS 2007). This is not to say that womens autonomy is not important, but that womens abilities and decisions to access modern maternal health care are conditioned by myriad factors. Van Hollen (2003) observed, the decision-making process is never a matter of the free will of rational, value-maximizing individuals, but, rather, it is always enacted in political-economic contexts and shaped by socio-cultural factors such as gender, class, caste, and age, in reference to how poor women and their

82 families make decisions about matters related to pregnancy and childbirth (p. 7). And so we can observe that on measures of womens autonomy and status, Tamil Nadu might be somewhat more advanced than Karnataka, but we must recognize that the relationship between these factors and these states somewhat more advanced safe motherhood indicators is mediated by a number of factors, such as economic status, education and familial decision processes. CONCLUSION Safe motherhood outcomes improved in Tamil Nadu and Karnataka over the course of the 1990s and into the 2000s. In both states, greater access to safe motherhood services was likely facilitated by reduced incidence of poverty and illiteracy. There is some variation in female literacy and Scheduled Caste composition between the states, but these and other social and cultural factors discussed in this chapter do not constitute a complete explanation for such disparate safe motherhood outcomes. The data point to two key differences between the states. First, Tamil Nadus had more equitable safe motherhood outcomes across economic, educational, geographic and caste groups that are comparable with those in Karnataka. Second, the public sector played a more prominent role in provision of safe motherhood services to women from more vulnerable groups in Tamil Nadu than in Karnataka. Women in lower income and education groups in Tamil Nadu accessed safe motherhood services in the public sector at nearly twice the rate as directly comparable groups did in Karnataka in 1998-9. These findings lay the groundwork for the analysis presented in the next two chapters. Differing social, cultural and economic conditions between the states do not provide a full explanation for Tamil Nadus superior safe motherhood outcomes. The

83 important role the public health system appears to play in providing safe motherhood services to women from more vulnerable groups in Tamil Nadu provides an important clue to more fully answering the question. The following two chapters examine how the politics of health and its administration in the two states factor into explaining this variance.

84

CHAPTER 5

POLITICS & SAFE MOTHERHOOD

85 INTRODUCTION Pregnancy and childbirth though very much experienced on an individual level are embedded in broader socio-cultural contexts that impact how women experience health systems and access safe motherhood. So too are these experiences embedded in political contexts that impact the extent to which health systems provide for safe motherhood, especially for those that face the greatest social barriers to access. In the Indian context, the central government through its political leadership plays an important role in setting national health policy goals and funding programs that support them. Central government health funding and programs make up only about 25 percent of states total health budgets, however (Government of India 2005). States, albeit with some influence from national health agendas, determine which health policies to take up and the extent to which to support them. This has a direct impact on the strength of the public health system to deliver safe motherhood services. It also affects the potential for leadership to emerge and work to integrate it among other priorities. As discussed in the previous chapter, women from more resourced groups have a record of accessing safe motherhood at greater rates whether in the public or private sector. This analysis focuses on the public sector because it plays an important role in providing maternal health services, especially to less-resourced groups that lag behind in access to safe motherhood. The public sector is positioned to play a key role in ensuring equitable access to safe motherhood through health systems development. In Tamil Nadu and Karnataka, low income, illiterate and scheduled caste women access safe motherhood services from the public sector at greater rates than the private sector (IIPS 2001a and b). Tamil Nadus public sector provided a greater proportion of safe motherhood services

86 than Karnatakas in the 1990s and its margin increased to nearly a twenty-five percent difference in 2002-4 reporting 53 percent of Tamil Nadus institutional deliveries were in the public sector compared with 29 percent of Karnatakas (RCH2 2004). This chapter takes a few steps back in the causal chain to examine political factors, particularly party politics, in each case that affect the central decision venue for health policymaking and shape the strength of health systems to provide for safe motherhood more broadly. The following chapter examines the links then from the political factors discussed here to the characteristics of the bureaucratic venue and the governmental programs of action pursued by policy actors. These ultimately affect the availability and accessibility of safe motherhood services in each state and help us to understand the variation in rates of access between them. POLITICS & HEALTH POLICY IN TAMIL NADU Public provision of safe motherhood services in Tamil Nadu must be understood from a historical perspective on the politics of social welfare in the state. Political attention to health in Tamil Nadu is rooted in an early twentieth century social movement founded by E.V. Ramaswami Periyar that promoted equality for women and a casteless society. Periyars Dravidian social movement inspired the two major political parties in the state today, the Dravida Munnetra Kazhagam (DMK) and All India Anna Dravida Munnetra Kazhagam (AIADMK), which have between them been in power in the state since 1967. The legacy of Periyars movement, including a responsive, populist political system and an active civil society that engage each other to promote social welfare in the state, is widely commented upon by students of Tamil politics (Athreya & Chunkath 2000; Geetha & Rajadurai 1998; Katzenstein 2004; Mehrotra 2006; Subramanian 1999;

87 World Bank 2006). The India Human Development Report 2001 observed the social consciousness inspired by leaders such as Ramasami Naicker Periyar, stating, The State has, historically, been a hot bed of social reform movements, often precipitating political action in the desired direction (p. 79). Scholars of Indian and Tamil politics including Kohli (1990), Sinha (2005), Subramanian (1999) and Washbrook (in Frankel & Rao 1989) have noted Tamil Nadus leading parties pursuit of populist, welfarist policies that advance such issues as health in the state. A recent World Bank report stated, Ideas emphasi[z]ing social welfarism and empowerment permeated through the DMK and the AIADMK. These ideas also made good political sense (2006, p. 57). Drawing on Venkatasubramanian (2005) to provide a historical example, the World Bank report described how the Congress party lost power in Tamil Nadu primarily because of the way it handled a major food crisis in the late 1960s. Already inclined to support social welfare policies, leaders of the DMK and AIADMK learned from the experience and developed a common commitment to making food staples affordable to all, the report stated. The report also linked political initiative for the states Nutritious Midday Meal program, family planning and improved child and maternal nutrition programs to this nexus between common political values for social welfare and the political incentives to promote these values (World Bank 2006). Scholars of Indian politics, including Drze (2004) and Subramanian (1999) have commented that the politicization of health in Tamil Nadu has driven its policy initiative and resource allocation to this area. Public health and policy experts in the state were attuned to this factor. One explained ongoing political support for maternal health policies thus, Its not a particular person or political official. Its party and they both

88 want to show work in this area. Every government for the past 30 years is taking some action, initiative in this area (interview no. 29a). A longtime public health expert in the state suggested development of Tamil Nadus model essential drug distribution system as a relevant example (interview no. 5a). He described how, in the mid-1980s, the AIADMK introduced access to essential drugs as a political issue, suggesting the public had rights to affordable and reliable access to a safe drug supply. He explained that Danida, the bilateral development agency of The Netherlands, helped develop the system, and with the opposition party monitoring progress, a corruption-proof system of essential drug distribution was created in the form of the Tamil Nadu Medical Supply Corporation. The World Bank report noted previously helps to explain why monitoring from the opposition (DMK party in this case) would help to prevent corruption. Discussing why monitoring of development programs in Tamil Nadu is effective it stated, politicians are unlikely to tolerate abuses in programmes that they consider important for ideological and political reasons (2006, p. 58). Support for the Tamil Nadu Medical Supply Corporation was driven by the idea that the government had a responsibility to provide for the social welfare of its citizens (Snow et al. 1986; Stone 2002) and this came in the form of a safe and effective drug supply in this case, ideas reinforced through the structure of political competition in the state. The Medical Supply Corporation example has direct relevance to safe motherhood insofar as the program helps ensure availability and access to essential medicines and because it shows how political incentives, such as electoral incentives, work to reinforce political priority for health policies and their effective implementation. It is also representative of the relatively strong health system Tamil Nadu political dynamics facilitated and later safe motherhood policies built upon.

89 Although I was not able to document the political roots of specific safe motherhood policies in Tamil Nadu, the state has a history of issuing government orders supporting maternal health services that spans leadership from both major parties since the mid-1990s. For example, under Chief Minister M. Karunanidhi (DMK party), when officials in the public health bureaucracy started researching causes of maternal mortality and learned that women were reluctant to seek delivery services at primary health centers because they were not staffed outside typical 9 to 5 business hours, the health secretary issued Government Order No. 396 to increase the availability of 24-hour delivery services. The 1999 Order provided for 24-hour staffing of nurses at 90 primary health centers in seven districts with the most need. Under Chief Minister J. Jayalalithaa (AIADMK party), Government Order No. 25 extended 24-hour delivery services to an additional 90 primary health centers in ten more districts in 2005 based on increased use of delivery services in 24-hour facilities and the importance of providing basic emergency obstetric and newborn care in these facilities. In 2006, under newly (again) elected M. Karunanidhi leadership (DMK party), the government moved to fill critical health staffing vacancies (The Hindu October 24, 2006; The Hindu April 16, 2007), reversing a hiring freeze of the previous government (interview nos. 29b, 32b). The new government also distributed mobile cell phones to all village health nurses in the state, a step that facilitated communication between families, nurses and actors at higher levels of the health system (interview no. 35). Other government orders, all signed by a senior health bureaucrat (notably Chunkath), have addressed human resource vacancy problems, provision of basic emergency obstetric care, strengthening of outreach systems, provision of emergency transport services, establishment of blood banks and donation programs,

90 and improvements in the maternal death audit system (Government of Tamil Nadu Government Order 1999, 2004, 2006A-G). Official reports and documents do not typically link specific initiatives in the health arena with specific political parties and dynamics in Tamil Nadu. The states initiative in the health policy arena is frequently referenced, however. The state initiated the Nutritious Midday Meal program under M.G. Ramachandrans leadership (AIADMK) in 1982. The programs success in promoting nutrition for children, pregnant and lactating women, and pensioners made it a model program that inspired introduction of Indias National Midday Meal Programme in 1995 (Government of India 2002). Karnataka introduced its statewide noon meal program in 2002, after Indias Supreme Court ordered all states adopt it in 2001 twenty years after Tamil Nadu took initiative in this area (World Bank 2006). The National Commission on Macroeconomics and Health, Ministry of Health and Family Welfare, considered the states program for reliable and affordable essential drug provision another model program for the nation (Government of India 2005). Tamil Nadu has also been recognized for its initiative in the family planning arena. The India Human Development Report 2001 noted, Tamil Nadu was among the first States, in independent India, to launch family planning programmes (Government of India 2002, p. 79). It was also one of the first states to drop official family planning targets, known for the coercive tactics used to achieve them, although family planning remains a priority. Shortly after the launch of the central government-sponsored Child Survival and Safe Motherhood (CSSM) program in 1992, the State Population Council with Tamil Nadus then chief minister as its chair announced its support for promotion

91 of institutional deliveries, an important component of maternal mortality reduction efforts (Padmanaban & Desikachari 2004). The nationally sponsored CSSM program was reported to emphasize family planning over safe motherhood (Visaria et al. 1999), but Tamil Nadus step to take ownership of safe motherhood, especially in advance of the landmark International Conference on Population and Development, signaled early initiative on the issue. More recently, and also relevant to safe motherhood, the Reproductive and Child Health 2 Programme Joint Review Committee recognized Tamil Nadu for being among the first states to certify facilities with comprehensive emergency obstetric and newborn care services (Government of India 2007), again showing the states leadership role in promoting safe motherhood in the country. Although political dynamics driving health and safe motherhood initiatives in Tamil Nadu are not widely recorded in official reports and documents, public health and policy experts frequently recognized them as key drivers. Conditions surrounding political competition in the state created an environment that fostered initiative and provided support through the health bureaucracy for issues, such as social welfare and health, which resonated with political priorities across parties. The following quotes represent informants insights to the issue: A representative of an international development agency with close knowledge of Tamil Nadus initiative for safe motherhood understood it to be based in, Political and bureaucratic will. The government is based in a social movement about women and freedom, women and childrens issues [have] long [been] a priority of the state. She also commented on the nexus between political incentives and values: Opposition parties also are monitoring and go to the press.

92 [They say] the previous government did x and now we have to show what weve done. Its part of maintaining mass appeal and also part of doing good, important work (interview no. 29a). A senior health administrator observed, In Tamil Nadu health is always perceived as a sensitive issue by politicians because so many people depend on the public health system. He said that improvements in health infrastructure, human resources and free health services happened as a result of the Periyar movements attention to family planning promotion and improving the status of women because both parties attach a lot of importance to health. Each government introduces something new in health (interview no. 32a). Public health and policy experts commented frequently on the consistent commitment to health shared by Tamil Nadus two main political parties and their leadership. They also shared the perspective that the states political, bureaucratic and technical leadership shared this commitment and served complementary functions to enhance public health services as a result. The following quotes provide examples of how informants linked political and administrative commitment to health with initiative for safe motherhood: A retired public health official recalled from his tenure in the 1990s, Tamil Nadus government is one of the progressive states in the country. There was a lot of political will for health and other development issues at the time education. Second was bureaucrats there at the time. The mission, vision of government was there, but an enthusiastic implementation by bureaucrats successive health secretaries who were dynamic, had much rural health consciousness, wanted to support. And the health secretariat also (interview no. 94).

93 A current public health official explained, Political support is good. Politicians want to develop the health department and systems because whichever party is in power will get rewards for it [presumably rewards at the ballot box]. Administrative, technocrats, political are all in sync (interview no. 39a). The government orders signed by the health secretaries under successive DMK and AIADMK leadership are one source of evidence to support these contentions. The states online archive of budget speeches dating to 2002 is another important source of data that shows successive governments committing to pursue health and safe motherhood goals building on what was done before and adding something new. Under the leadership of Chief Minister J. Jayalalithaa and the AIADMK, maternal mortality reduction and particular plans to achieve this goal were addressed in budget speeches spanning each of the fiscal years 2002-3 to 2005-6 (Government of Tamil Nadu 2002-2008). Plans included establishing 24-hour delivery services at Primary Health Centers and expanding and upgrading centers to provide comprehensive emergency obstetric and newborn care (CEmONC) services throughout the state. The theme continued when M. Karunanidhi and the DMK party returned to power. The 2007-8 and 2008-9 budget speeches recognized the contributions of expanded 24-hour delivery and CEmONC services to increasing rates of institutional delivery, and added plans to provide emergency transportation services at the block level (an administrative unit within districts). Supported with funds from the Tamil Nadu Health Systems Project, ambulances were provided to 187 blocks with plans to provide them to the remaining 198 blocks, as reported in the 2007-8 budget speech. These are important indicators that both parties were committed to improving the health system. More 24-hour Primary Health

94 Centers, increased staffing, improvements in the availability of emergency obstetric services and emergency transport have contributed to improved maternal health outcomes in recent years. Shifting gears, resource allocation to health, and particularly maternal health, appears to have been consistent with the level of political commitment attributed to the state. Indias budget systems do present difficulties in assessing this, however. The 2005 Report of the National Commission on Macroeconomics and Health addressed this issue: The budgeting system in India is based on the artificial classification of plan [new items] and non-plan [items in past budgets]. This makes it impossible to track fund flows. Since the annual planning process only considers the plan or new activities, the maintenance of assets never gets the required attention under the non-plan budget. Secondly, the aggregation of budget heads keeps changing making any trend analysis difficult. Thirdly, there is no uniformity in budget lines in the country (Government of India, p. 80). For these reasons, it is difficult to obtain and report on health budget data, especially data disaggregated to something so specific as maternal health, in a meaningful way or to make comparisons across states on particular line items. We can look at more macrolevel data, such as the share of health in states revenue budgets, over time and make some cross-state comparisons to get a relative sense of whether Tamil Nadus health budgeting is consistent with an argument for relative political priority in the state. To begin, Indias Tenth Five Year Plan: 2002-2007 indicated that Tamil Nadu was the third highest spending state per capita on health, behind only Kerala and Punjab based on a 2000 National Council of Applied Economic Research report (Government of

95 India 2002). Tamil Nadus total spending per capita was recorded at approximately Rupees 125, with about 80 percent from the state and 20 percent from the central government. Karnataka fell in at just over Rupees 100 in health spending per capita, with about 75 percent from the state and 25 percent from the central government. A comparison with states that spend less on health services per capita and have lower health indicators helps to put these figures in perspective in the Indian context. States like Bihar (Rupees 60), Madhya Pradesh (Rupees 65) and Uttar Pradesh (Rupees 70) spent much less per capita on public health services some 50 to 60 percent of what Tamil Nadu spent and central government spending makes up a greater proportion of the total. These figures make both Tamil Nadu and Karnataka look good in comparison. Citing similar figures for Tamil Nadu, a 2004 World Bank health project appraisal criticized the state for its low level of health spending, at around $3 per capita, and noted that spending on public health services had fallen from 7.5% of the state budget in the mid-1980s to 5.8% in 2001 (World Bank 2004). Again, that $3 per capita was relatively advanced in comparison to other states. And, the state budget figures noted here do not tell the whole story. The Tamil Nadu Human Development Report (Government of Tamil Nadu 2003) reported that nominal budget outlays to the states health sector increased between 1991 and 2000 from Rupees 4.11 billion to Rupees 10.51 billion. The report further stated, Except in 1995-6 and 1999-2000, the increase in nominal outlay over the previous year has exceeded 10 per cent with the highest increase of 19.8 per cent being recorded between 1990-1 and 1991-2 (2003, p. 65). The development report concurred with the World Bank assessment about the decline in health expenditure as a share of the

96 total budget over the course of the 1990s though, despite increases in public health expenditure in absolute terms. Data on Tamil Nadu, Karnataka and other major states shares of health in revenue budgets between 1985-6 and 2004-5 offers further bases for comparison. See Table 5.1 below. In Tamil Nadu, the share of health in the revenue budget decreased between 1985-6 and 1991-2, rebounded somewhat in 1995-6 and then trended downward. Karnataka and Maharashtra showed more consistent downward trends. It is unclear why Bihars (a state that spends relatively less per capita and has lower health indicators) figures went up when others went down. The National Commission on Macroeconomics and Health reported a general downward trend in health spending by the states due to fiscal pressures in the late-1980s and then again in the early- and mid-1990s with the introduction of reforms by the central government (Government of India 2005). These decreases are reflected in the figures in the table below. The Commission cautioned reading too much into these figures, stating, the actual access to services depends on other factors such as the efficiency with which the system is functioning. In other words, if the health system is inefficient or poorly managed, mere increase of financial resources may have little consequence (2005, p. 72). It is impossible to draw conclusions about the relationship between Tamil Nadus political dynamics and allocation of resources to health from these figures, but the states relatively high levels of health allocations and spending compared to other states are consistent with the proposition.

97 Table 5.1: Share of health in revenue budget of select major states (%)
1985-6 Bihar Karnataka Maharashtra Punjab Tamil Nadu 5.68 6.55 6.05 7.19 7.47 1991-2 5.65 5.94 5.25 4.32 4.82 1995-6 7.8 5.85 5.18 4.56 6.4 1999-2000 6.3 5.7 4.59 5.34 5.51 2003-4 4.84 4.85 4.39 4.27 5.26

Source: Government of India, National Commission on Macroeconomics and Health 2005, p. 72 International attention in the 1990s, including the 1994 International Conference on Population and Development (ICPD), may also have precipitated political support and initiative for the issue in Tamil Nadu. A public health expert in the state with deep historical knowledge of safe motherhood in Tamil Nadu located the origins of the states initiative thus: It started with the 1990s global discussions about the safe motherhood initiative the ICPD 1994 global advocacy conference rousing and channeling policymakers. Nobody should die from pregnancy. The Tamil Nadu government stated subsequent to ICPD that this needs action. It took on to train skilled birth attendants, dais, doctors so mothers could be in safe hands. There was political will. ICPD raised awareness. Top bureaucrats attended and came back to convince politicians. It resulted in increased will. Its spurred a chain of activities over time since then. It spins off in a chain reaction. Donors also make a commitment with the international conference. It brings all parties to synchronization of commitment (5b).

98 Indeed, following the conference, one of the key bureaucrats that attended, Sheela Rani Chunkath, championed the cause in Tamil Nadu. Her role is discussed in more detail in the next chapter, but suffice to say here that she was instrumental in aligning political priority for health in Tamil Nadu with the ICPD agenda in a way that resulted in a series of government orders, key partnerships for safe motherhood with donor agencies such as Danida and UNICEF, and a public health system mobilized on behalf of the cause in the years that followed the conference (interview nos. 5b, 32b). The ICPD agenda dovetailed nicely with the agendas of the state-level political parties in Tamil Nadu that valued health policy and had incentives to support it. The new model fit with existing ideas about government responsibility for social welfare. As Sikkink suggests possible, this provided opportunities for safe motherhood to rise on political and administrative agendas and laid the groundwork for subsystem actors to positively affect policy and implementation (in Berman 2001). POLITICS & HEALTH POLICY IN KARNATAKA Karnatakas political environment and its relationship to health initiative and spending are not so straightforward. The states safe motherhood activities and funding are more closely aligned with central government-sponsored health programs, such as the Reproductive and Child Health programs and National Rural Health Mission, than any specific initiative originating within the state. There is little evidence that health has cache as a political issue amongst parties at the state-level, or of any particular historically or socially based political value for health policy. There have been opportunities since the mid-1990s for the state to take greater initiative in the health and safe motherhood arenas. However, evidence suggests political incentives and values were

99 not aligned to elevate these on the states policy agenda in a way that would significantly impact the extent to which the health system supported safe motherhood services in the state. The state had a key opportunity to commit to improving health and safe motherhood when, in 1999 Karnatakas then chief minister, S M Krishna, formed the Karnataka Health Task Force to assess the state of the public health system and make recommendations for reform. He called on a group of independent public health experts in the nongovernmental and private sectors, academics, members of professional medical associations and bureaucrats to take up the issue. The Task Forces 2001 report identified eleven major issues of concern. Chief among them were neglect of public health, corruption, distortions in primary health care, lack of attention to health inequities, and gaps between policy and implementation. The report contained an extensive list of recommended reforms. The following were highlighted as some of the priority issues: Develop policies and data systems to promote and monitor equity in health service delivery that focus on regional disparities, gender inequalities; class and caste / ethnic inequalities; the geographical (rural / urban) divide (Karnataka Health Task Force, Issues of concern and an agenda for action, 2001, p. 5) Strengthen the skills of public health policymakers and managers from the level of medical officers managing Primary Health Centers to the state directorate level Commit to eliminating corruption in the health system, including unsanctioned fees for services; appointments, promotion and transfers; and medical education

100 Strengthen human resources management, including numbers and skills of health service personnel, as well as top-level leadership in technocratic and bureaucratic positions Develop a robust health management information system The Task Force also identified womens health generally and reproductive health more specifically among the key points of concern and issues for action in its report. The report recognized improvements in the health status of women in the state over time, but pointed to disparities between rural and urban areas, regional inequities (e.g. the lowperforming districts of northern Karnataka), and lagging progress compared to other states in South India. Explaining the status of womens health in Karnataka, the report stated, The efforts taken to address womens issues have been inadequate, distorted, vertical, top-down and have rarely emerged out of womens priority concerns (Karnataka Health Task Force 2001, Ch. 8). Regarding Reproductive and Child Health programs, the Task Force recommended that the state: Make First Referral Units fully functional, including 24-hour delivery services, emergency obstetric care, and transportation Improve the quality of service delivery, including ensuring the availability of trained staff and increasing the number of female medical officers, especially in the northern districts Combat gender discrimination against female patients and health workers

The report also criticized policies addressing womens health issues for being limited to those with an emphasis on reproductive health to the neglect of womens other health needs. The report set an agenda for action on items that would directly affect how the

101 health bureaucracy responded to womens maternal health needs, such as providing access to emergency obstetric care, but there is little evidence that any concerted effort materialized to achieve these goals into mid-2007. As of mid-2008, one member of the task force estimated forty percent of the Task Forces recommendations remained unimplemented, leaving substantial room for improvement. The Task Force reported, There is a lack of political will in that health is not high on the agenda of governance. The commitment and capacity to get plans off the ground and reach those who need to be reached is lacking, (Karnataka Health Task Force 2001, p. 5). One member of the Task Force elaborated on this point, explaining that although the chief minister and his principal secretary at the time were committed to reform, political support was not widespread enough to sustain the effort (interview no. 25). He said that the formation of the Karnataka Health Task Force in 1999 was a good time, a different time, when there was will from the chief minister, then S M Krishna of the Congress party (interview no. 25). It was under S M Krishnas leadership that the Karnataka Health Policy 2004, formulated by the Task Force, was passed. His account of the rise and fall of the health reform effort implied that political will for health varied with fluctuations in the states political leadership. If the chief minister supported health reform and only S M Krishna stands out in recent years then it would receive attention and resources. This was not a very reliable system, however, as leadership shifted frequently in the state. Another health policy expert addressed the influence of shifts in political power in the state, explaining that support for social policy, including health, was not institutionalized as a priority in the states competitive multi-party political system.

102 Neither party will bother. The opposition party is not important, she said (interview no. 26). Her point was that it was difficult to get health policy on the agenda and it could easily fall from the agenda because the parties had no pressing reasons to take up the issue or incentives to sustain priority for it in the state. There were not sufficient electoral rewards to be had for it. Karnatakas budget speeches, setting out commitments of parties in power in the state government over three administrations between the 2002-3 and 2007-8 fiscal years, reflect these observations. An initial observation, in 2002-3 and 2004-5 health policy did not even warrant a separate heading packaged as it was with education and, in one instance, civic amenities. In the first three speeches examined, prominent health issues such as polio, tuberculosis and HIV/AIDS, more obscure issues such as telemedicine and cardiology referral centers, and general proposals to increase health spending were emphasized (Government of Karnataka Budget Speeches 2002, 2003, 2004). Safe motherhood received its first mention in the 2005-6 speech with a proposal to provide 24hour delivery services through nurses posted to Primary Health Centers in select backward areas to reduce maternal mortality and infant mortality and to provide immunization services. I found no evidence this proposal had been implemented. Safe motherhood appeared one other time in the speeches surveyed, in 2007-8 when the government proposed to provide kits for new mothers containing such items as mosquito nets, soap and lotion to encourage institutional deliveries and facilitate hygiene maintenance none of which address major causes of maternal mortality. This lack of follow through and advance of ineffective strategy suggests there was little real priority

103 for safe motherhood from political principals in the state despite its appearance in recent budget speeches. The health proposals presented in Karnatakas budget speeches were comparatively vague, inconsistent and light on attention to safe motherhood next to Tamil Nadus over the same period. Tamil Nadu budget speeches set targets for maternal mortality reduction and specified goals and funding for increasing availability and access to comprehensive emergency obstetric and newborn care (CEmONC) services, addressing these issues directly in six of seven budget speeches between 2002-3 and 2008-9 and showing strategic direction for maternal mortality reduction. Karnataka budget speeches proposed substantially more restrained goals for 24-hour Primary Health Center delivery services. These centers do not provide emergency, including surgical, obstetric services that address the major causes of maternal deaths. Only two of six of the states budget speeches addressed maternal health needs between 2002-3 and 2007-8. Comparing budget speeches in the two states since 2002, Karnatakas suggest relatively less strategic direction in the health policy arena generally and relatively lower priority for safe motherhood specifically. This translated to relatively less political attention and allocation of resources to maternal mortality reduction in Karnataka than in Tamil Nadu at a minimum between 2002 and 2007. Following on this, as of late-2007 the National Rural Health Mission, a nationwide Government of India health policy that went into effect in 2006, represented another missed opportunity for action on safe motherhood in the state. The Reproductive and Child Health-2 program, Karnatakas primary vehicle for safe motherhood initiatives and funding, comes under the Missions umbrella. Government and media reports

104 criticized Karnataka for its slow implementation in 2007 and 2008. A report of the Reproductive and Child Health-2 Joint Review Mission in March 2007 suggested, Karnataka could perform much better (Government of India 2007; p. 143). It noted a lack of policy impact for the states northern districts and made several recommendations for improvement. Recommendations specifically pertaining to maternal health included: to establish honorariums to health personnel for attending any delivery as opposed to only night deliveries; to examine whether the states medical colleges could manage First Referral Units; and to give cash payments to auxiliary nurse midwives for implementing the Janani Suraksha Yojana (JSY) program a program that provides cash incentives to poor families for using government antenatal, delivery and postnatal care services (p. 143). The report also recommended that the state fill 560 auxiliary nurse midwife vacancies, crucial personnel in maternal health service provision. Six months later, in September 2007, The Hindu reported that Karnatakas slow progress had caused the Mission to ask the state specifically to fill vacancies of auxiliary nurse midwives in Primary Health Centers (Prabhu 2007). Meanwhile, The Hindu (2008) reported that Union Health Minister Anbumani Ramadoss praised Tamil Nadus implementation of the Mission and recognized the state as the best performer on a visit in early 2008. In June of 2007, at a review meeting of the National Rural Health Mission in New Delhi, Karnataka was identified as one of the eight lowest performing states in the country (The Deccan Herald 2007). The Deccan Herald (2007) reported that the poor performance of these states prompted Indias prime minister to call for a special meeting of the National Development Council at which he was to instruct the states chief ministers to increase priority for health care. According to one health policy expert in

105 tune with the politics of health in the state, Karnatakas then chief minister H D Kumaraswamy of the Janata Dal (secular) party, a figure whose leadership of a weak coalition government was under challenge lacked the power to increase attention and resources for health even if he had some desire to do so (interview no. 25). The chief ministers weakness became more evident when the states ruling coalition government (Bharatiya Janata Party and Janata Dal [secular]) dissolved in the Fall of 2007 and leadership shifted to the Bharatiya Janata Party after some months of Presidents Rule. Scholars have presented varying theories and evidence that help to explain why social welfare policies, extending to health, have not risen higher on political agendas in the state. To begin, in his study of Devraj Urs tenure as chief minister of Karnataka between 1972 and 1980, Kohli (1989) found limitations to the regimes purported pursuit of anti-poverty programs, finding that An attempt to institutionalize lower-class power within the state ... was neither intended nor achieved. The resulting reforms therefore tended to be piecemeal and superficial rather than deep and systematic (p. 146). Gould (2003), Harriss (1999) and Manor (1989) similarly found that ideological commitments as to social welfare and reform did not have much policy impact in Karnataka as perceived personal and more immediate local interests proved more powerful influences. Political patronage and power at the state level interplayed with these factors at the local level, affected by traditional social structures that drew influence away from the statelevel political arena and tended to favor maintenance rather than restructuring of the social order (Kohli 1989; Manor 1989). Decentralization is an important and related theme that has a greater impact on politics in Karnataka than in Tamil Nadu. Manor (1989) explained:

106 most of the time in most villages, politics still means power relations among persons and groups within the village or circle of villages. The modern state has been kept very largely at arms length. The power which Indias electoral system gives to ordinary people to reject unwanted state-level parties and leaders is now well understood by ordinary Karnataka villagers. But that understanding has not generated a rejection of traditional relationships and modes of conduct within the village. Nor have reformist policies and the attempt to channel a sizable share of the resources of the state to poorer villagers in the years between 1972 and 1980, and since 1983 worked substantial changes in village power relations (p. 332). A decade later, Crook and Manor (1998) suggested that although villagers perceived government to be more responsive to local priorities when power was formally decentralized to the local level, they found that democratic decentralization did little to benefit vulnerable groups poor people, Scheduled Castes (ex-untouchables) and women, including in social sectors such as health and education (pp. 76-7). This does not portend well for the impact of the National Rural Health Mission on maternal health in Karnataka with its emphasis on decentralized planning and implementation. Further, and in comparative perspective, Harriss (1999) drew on Churchs (1984) classification of types of political regimes in Indias states to suggest that states policies and performance were importantly influenced by the extent to which political parties were organized and managed well and the extent to which the interests of lower castes and classes were represented. His findings included an assessment that lower castes and classes were less well represented by political parties in Karnataka than in Tamil Nadu and that the former states political parties were less organized than the latters. He found

107 that Tamil Nadu had relatively strong parties with coherent ideologies based in their social movement roots, factors affecting progress in the social sectors. With respect to Karnataka, he referred to Crook and Manors work on decentralization to suggest increasing fragmentation in the states political system has negatively affected the states progress for poverty alleviation and other benefits to vulnerable groups. Goulds (2003) historical analysis of electoral politics in Karnataka leading up to the 1999 elections reinforced the message that weak political parties further challenged by the instability of coalition politics and lacking a clear mandate proved significant challenges to progress for social change in the state. It is in this context that health policy experts and public health officials expressed concern that political and administrative commitment to health was weak in Karnataka and contributed to slow progress for safe motherhood in the state. Representative statements include the following: One policy expert in the state observed, Health is a state subject. The state is not committed in Karnataka (interview no. 26). Another expert on health policy in Karnataka and at the national level described a lack of political priority for health in the state by stating, political will is if the chief minister takes more interest in health he doesnt even review the health department, what kind of secretaries they appoint (interview no. 25). A high-level public health official said, Commitment is not therefrom anyone at any level to develop solutions to management problems, including corruption, reliable drug supply, human resources vacancies and mismatches of skills in postings (interview no. 8).

108 A bureaucrat with knowledge of the health system and maternal health in the state said, Pressure is not coming from higher levels of government to improve health sector performance (interview no. 27). These informants were suggesting that safe motherhood suffered because political and administrative leaders failed to make sure resources were allocated, effective managers were appointed, skilled health personnel were in place, equipment and drug supply were assured in the health bureaucracy policymakers failed to ensure the components of the health system that directly affected whether women could receive life-saving maternal health care were in place and functional. This has important implications for the impact of national priority for maternal mortality reduction on policymaking, as well as outcomes, at the state level safe motherhood policy is unlikely to have much impact if the health bureaucracy does not effectively allocate resources and attention to the issue. The state-level context must be considered in relationship to safe motherhoods position on the national-level policy agenda. The Congress party and its coalition government came to power at the national level in 2004, laying out the National Common Minimum Programme11 that contained the groundwork for safe motherhood to be prioritized through the National Rural Health Mission (NRHM). A representative of a leading safe motherhood advocacy organization in India explained that as part of the National Common Minimum Programme derived from an alliance of political parties, The NRHM is very political (interview no. 3). The National Rural Health Mission signaled a boost in political priority for health, and particularly maternal health, at the

The National Common Minimum Programme is a public declaration of the priorities of the ruling party or coalition of parties represented by Indias prime minister. The current statement can be found on the prime ministers website at http://pmindia.nic.in/cmp.htm.

11

109 national level (Shiffman & Ved 2007), but ascendance of health policy at the national level did not line up well with Karnatakas political leadership. Karnatakas coalition government made up of the Bharatiya Janata Party and Janata Dal (secular) was not a part of the Congress Partys United Progressive Alliance coalition that championed the Mission. A health policy and planning expert with knowledge of the Missions implementation progress in Karnataka noted this misalignment between political leadership at the national and state levels and suggested that it restricted the public health bureaucracys ability to take action to increase institutional deliveries a core national strategy to improve safe motherhood. She cited human resources shortages as an example of a key problem and suggested it was not being resolved because it was political Karnatakas government was an opposing power to the ruling national coalition (interview no. 9c). She described barriers on both sides of the political aisle, for example: on one side, she said that the national government delayed the release of allocated funding to hinder the state governments performance an explanation given for underutilized Mission funds in 2005-06. On the other side, the state government withheld full support for the national coalition governments schemes, such as neglecting to fill human resource vacancies. This is only the perception of one health policy expert, but other experts also attributed the states lagging priority for health policy to political parties ties to national-level political interests (interview nos. 5c, 84). These dynamics only served to compound state-level political conditions affecting a lack of attention and resources to safe motherhood.

110 Shifting directions somewhat, it is important to consider whether resource allocation to health, and particularly maternal health, are consistent with the relatively low level of political commitment attributed to the state. As discussed in the section on health resources in Tamil Nadu, this is a difficult matter to untangle partly because the state is more readily comparable to Tamil Nadu than some of the lower-performing states in the country but there are some data sources we can turn to get a sense of priority for health in the state. Recalling the comparison between the states in the previous section, Karnatakas share of health in its revenue budget declined steadily from 6.55 to 4.85 percent between 1985-6 and 2003-4 (Government of India 2005). Tamil Nadus figures yo-yoed, but ultimately decreased from 7.47 to 5.26 percent over the same period (Government of India 2005). Taking another indicator of resource allocations for health, the Karnataka State Health Policy 2004, adopted under the leadership of the Congress party, suggested the state would attempt to meet the standards of seven percent of state budget allocations to health by 2005 and eight percent by 2010 based on the Government of Indias National Health Plan 2002 recommendations. The National Health Plan 2002 was formulated while the National Democratic Alliance, led by the Bharatiya Janata Party and not aligned with the Congress party, was in power. According to budget allocation plans posted on Karnatakas state finance department website, budget allocations to health ranged from one percent to 3.47 percent of the state budget between fiscal years 2002-03 and 2007-08. In the year following the National Health Plan recommendation, under Congress party leadership, the proportion of the state budget allocated to health decreased from 2.24 percent to one percent (Government of Karnataka 2002-2007; Government of

111 Karnataka 2004) a precipitous step in the opposite direction of publicly stated national and state goals. We can also look at Karnatakas resource commitments in relationship to the National Rural Health Mission. According to an article in The Hindu in September 2007, the National Rural Health Mission asked Karnataka to work toward the Missions goal of increasing state spending on health by ten percent annually (Prabhu 2007). The states Medium Term Fiscal Plan for Karnataka 2008-12 shows plans for health outlays to be increased by more than ten percent for each of the fiscal years between 2005-06 and 2008-09 increasing from Rupees 11.47 billion to Rupees 22.06 billion with those plans to be supplemented by the National Rural Health Mission (Karnataka Finance Department 2008). Government and media reports indicate that Mission funds, encompassing those funds for maternal health, have been thus far under-utilized. The Reproductive and Child Health-2 Joint Review Mission (Government of India 2007) reported that in fiscal year 2005-06, Karnataka expended only 45 percent of Rupees 288 million released (Rupees 630 million were originally allocated). Tamil Nadu used 86 percent of the Rupees 613.8 million released to it in 2005-6. The report showed that Karnataka had used 11 percent of Rupees 883.7 million allocated in 2006-07 halfway into that fiscal year, with no funds released as of that time. Further media reports indicated that Karnataka returned a substantial sum of Mission funding to the national government unspent from fiscal year 2006-07 allocations (Charan 2008). Tamil Nadu had used nine percent of the Rupees 1.07 billion allocated to it and 14 percent of the Rupees 710.3 million that had been released midway into the 2006-7 fiscal year, according to the Joint Review Report (Government of India 2007).

112 With respect to safe motherhood, although it is difficult to disentangle maternal health from budget data, there is some evidence that Karnataka has devoted relatively few resources to the issue. A budget study by a local watchdog and health advocacy group found zero expenditures for maternal and child health in the fiscal years 1999-2000 and 2000-2001 by Karnataka state (Centre for Budget and Policy Studies 2004). A senior health official in Karnataka drew a direct comparison between the two states investments in maternal health, explaining that in Tamil Nadu the state is giving it, not just the RCH centre-sponsored programs. Karnataka state doesnt give more than the centrally sponsored Reproductive and Child Health funds (interview no. 12) to address maternal health issues. The official suggested that it was Tamil Nadus investment in safe motherhood that pushed its progress beyond that of Karnataka (interview no. 12). A district health official in Karnataka, linking funding for maternal health programs with outcomes, stated, Without RCH 2 and NRHM [funding], infant mortality and maternal mortality would be much higher. Other sources of support wouldnt be there (interview no. 22). Evidence on the states dedication of resources to health and safe motherhood, though not conclusive, appears consistent with the suggestion that political commitment for health in Karnataka is relatively lower than that in Tamil Nadu. Local & regional political dynamics In addition to state-level politics, the Karnataka case raises important questions about the impact of local and regional political dynamics on health policy and implementation. Decentralization of authority for health policy and oversight emerged as an important issue in the case. A bureaucrat with knowledge of maternal health issues in Karnataka said, The state cannot necessarily control [health] programs. With decentralization,

113 local interests are overriding other issues, such as Reproductive and Child Health. Priorities are multiple infrastructure, roads, water, retail (interview no. 27). Two informants with extensive experience as public health administrators in senior positions described conflicts between the interests of the state and local political bodies as hindering effective management and health service delivery (interview nos. 8b, 17). Other informants perceived decentralization of authority over public health matters as a barrier to improving maternal health outcomes in terms of corruption and direct interference in health service delivery by local political representatives and bodies. Health policy experts and high-level health officials made general statements about corruption, for example, Corruption is top to bottom, politicians at all levels (interview no. 25) and Decentralization equals decentralized corruption (interview no. 8a) the latter referring to problems with Panchayat Raj interference. District level administrators and service delivery personnel put their perceptions of the situation in starker terms, as illustrated by their words below. All held positions in the same district. A government doctor with nearly 30 years of experience in public health and administration said, Decentralization to the grassroots level [Panchayats] has come with a lot of interference in performance of facilities that has demoralized the system. [Politicians] think they are the masters, expect doctors and staff to be under their control (interview no. 17). A district-level health administrator said, ANM and Gram Panchayat [local political representative] are joint accounts he may expect some monetary benefit and then she cant take up activities. Both have to sign. Most Gram Panchayat leaders are difficult to manage and difficult for women ANMs and health

114 workers (interview no. 22). The administrator added that the problem was not uniform, but estimated 5-10 percent of local political leaders caused problems. A female medical officer at a Primary Health Center said, Politicians try to push us to do things, try to control. Its hard to get staff here because they get harassed by politicians. Not a single person gives support (interview no. 21). The head staff nurse in one hospital said, Ministers misappropriate drugs. People complain and then ministers come here, complain, say to do this and that not by doctors orders (interview no. 18). A public health official in another, relatively low-performing district described similar challenges, noting that harassment of medical officers and other staff by Panchayat Raj members contributed to difficulties attracting and retaining service delivery personnel (interview no. 96b). He also noted corruption involving local politics and administration as a barrier to effective health management in that district. The case of the formerly independent state of Coorg, now Kodagu district, stood in contrast to the other three rural districts in which I collected data. Kodagu is one of the better-performing districts in Karnataka with an institutional delivery rate of 75.8 percent (RCH 2004). District health officials and service delivery personnel perceived panchayat representatives differently there. None uttered a negative word about them. A senior district health official said local political interest in health in Kodagu had continued from the days when the district was an independent state (interview no. 86a). He suggested that there was an ongoing interest in health issues there quite independent from state politics and that had benefited the jurisdictions health system and outcomes over time (interview no. 86a). During my site visits in Kodagu, the district health officer arranged

115 for me to attend a training session on implementation of the National Rural Health Mission. There, a panchayat representative responsible for health matters was giving information on the Reproductive and Child Health 2 program, including components on maternal and child health, to other panchayat representatives. He told me that his goal was to work with other panchayat representatives to help them better understand their roles and responsibilities with regard to health in the district (interview no. 92). Preliminary evidence suggested safe motherhood received greater political support in this district than in the others I observed. Institutional delivery rates were relatively high compared to other districts in the state. Kodagus story is consistent with evidence that regional political dynamics may also have played a role in shaping the extent to which the health bureaucracy promoted safe motherhood. While most informants discussed regional inequities in maternal health outcomes as a legacy of differences in socio-cultural factors and human development levels inherited when the state was formed in 1956, there is some evidence to suggest political and administrative neglect of the northern region has also played a role. A recent World Bank report on public health reform stated that Karnataka neglected its more backward northern districts where low female literacy rates, poor infrastructure for health and school education, and higher infant and child mortality rates remain a cause for concern (World Bank 2006, pp. 58-9). The Karnataka State Health Policy 2004 identified a need for attention to regional disparities in resource allocation in the state. The Karnataka Human Development Report 2005 went into some depth, highlighting issues of regional disparities in human development, gender, education, health promotion and economic development. The

116 report suggested the states budget procedures perpetuated regional imbalances by systematically allocating insufficient funds to historically disadvantaged districts, causing particular challenges for the districts of the northeast region. The report concluded that regional variations in human development and the skewed distribution of historically given expenditures, in favour of districts with higher human development indicators, makes it necessary to introduce strategic changes in resource allocation (Government of Karnataka 2006, p. 36). Many of the reports findings were informed by the 2002 Report of the High Power Committee for Redressal of Regional Imbalances in Karnataka. In September 2006, The Hindu reported: Noted economist and visiting professor in the Institute for Social and Economic Change (ISEC) Abdul Aziz said that even four years after the recommendations of the High-Power Committee for Redressal of Regional Imbalances (HPCRRI) was submitted to the State Government, no serious effort had been made to implement the recommendations of the committee to find a permanent solution to the problem of inequality in development. Many taluks in north Karnataka had been deprived of their share in development (The Hindu 2006). There is some evidence that regional, in addition to state, political dynamics may help to explain the extent to which the health system promoted safe motherhood across the state. The evidence presented here is not comprehensive and the hypothesis would require more investigation, but health policy and management experts consistently raised the issue of regional disparities in safe motherhood indicators and some linked these disparities with the politics of particular regions. For example, a high-level administrator in the Department of Health and Family Welfare said, Commitment and ownership by

117 politicians is not there in north Karnataka. In south Karnataka there is more political demand. It is closer to the capital (interview no. 14). An expert on health policy and implementation in one of the northern districts said, The baby that cries gets the milk people in the north dont even know what they can ask for.12 It is very underdeveloped in the north and politicians have ignored it totally (interview no. 84). This suggests different political dynamics might be at work in the northern and southern regions of the state and that these might influence different degrees of health system functioning on behalf of safe motherhood in the state. CONCLUSION Although both of these states feature competitive political environments with multiple parties, their records of initiative for health and safe motherhood are quite different. Tamil Nadus long record of initiative for health is grounded in a social movement that fostered a long-term pattern of political commitment between two major parties to values for social equity and politicization of health in the state. These created conditions in the health bureaucracy that promoted interest and investment in health system development and programs of action that showed priority for social welfare and public health. As a result, Tamil Nadu has been at the fore for promoting nutrition, family planning and safe motherhood, in addition to such health system-strengthening measures as reliable and affordable drug supply. Ongoing political attention to these issues is facilitated by shared priority and intense competition in state-level politics.

George et al. (2005) and Matthews et al. (2005) document unmet demand for services in the northern region and remote rural areas. These studies are discussed more thoroughly in Chapters 4 and 6.

12

118 In Karnataka, by contrast, there is little evidence of political drivers functioning to promote health system development and safe motherhood to the same degree as in Tamil Nadu. The Karnataka Health Task Force was a notable reform effort, but the initiative failed to gain broad political support. Evidence suggests health reform fell from the agenda with a change in government, demonstrating inconsistent political support for health in a state with relatively weak political parties that feature no clear mandate for social reform. This has affected the health bureaucracy, as well. Conditions do not favor re-allocation of priorities to elevate the status of safe motherhood, or perhaps more importantly to correct for regional imbalances in health and other human development indicators. Political support at the state level also failed to materialize for recent national health initiatives that could impact maternal health, possibly attributable in part to misaligned political incentives between national and state level party leadership. The main source of health initiative in the state came through nationally sponsored programs which, in the absence of state-level political support, received less vigorous support at the implementation level than needed to strengthen health system functioning for safe motherhood. Additionally, regional and local political dynamics might have served to exacerbate health system dysfunction, thus further detracting from provision of safe motherhood services. These cases suggest that competitive, multi-party politics can affect health system promotion of safe motherhood in different ways given different political incentives and other bases for priorities. In Tamil Nadu, parties were founded upon the bases of value for social equity and political incentives developed to reinforce priority for health and other social welfare programs, shaping a health bureaucracy attuned to these issues and a

119 health system with greater capacity to address them. Karnatakas political parties had no such strong bases for orienting its health bureaucracy and strengthening its health system to support these issues, instead caught up in a web of coalition politics and decentralization that has left parties powerless to affect broad social change. This suggests that political systems close to the implementation level play significant roles in affecting how health systems function.

120

CHAPTER 6

THE PUBLIC HEALTH BUREAUCRACY & SAFE MOTHERHOOD

121 INTRODUCTION The public health bureaucracy plays an important role in determining how well health systems function for safe motherhood in the developing world especially for women from lower income, education and social class groups, and for those residing in rural areas. These women and their families do seek out safe motherhood services, as part of regular care during pregnancy and when complications arise. The public health system often makes the difference between their having access to services that can reduce the risk of death in pregnancy or childbirth and not having access to these services. Policy processes surrounding the public health bureaucracy help determine the extent to which public health service providers and facilities are available and equipped to facilitate access to quality safe motherhood services. This chapter examines the linkages between the institutional rules, norms and priorities in health bureaucracies that in turn shape the programs of action that policy actors pursue. These programs of action affect the availability and accessibility of safe motherhood services in the two states. The chapter suggests that conditions in the health bureaucracy facilitated emergence of leadership for the issue, greater strategic planning and integration of safe motherhood as a priority among other programs in Tamil Nadu, leading to greater access and availability of services in that state. Conditions in Karnatakas health bureaucracy led to formal, rule-based implementation, less strategic planning and program implementation for the issue and imbalance among other priorities, affecting lesser availability and access to safe motherhood services in the state.

122 THE PUBLIC HEALTH BUREAUCRACY IN TAMIL NADU In the mid-1990s, a Tamil Nadu bureaucrat with a self-described passion for womens health attended the International Conference on Population and Development (ICPD) the 1994 ICPD was a pivotal event that turned global attention toward a reproductive health agenda after decades of priority for population control policies. Sheela Rani Chunkath brought that agenda back to Tamil Nadu. She returned to the state with a call to action one she was well positioned to pursue as Reproductive and Child Health Director for the state and that fit well with political and health bureaucracy priorities. Chunkath took a team problem-solving approach from the beginning, fostering development of a policy community including government officials and representatives of international donor agencies that would affect safe motherhood over the next decade. A senior health official gave an example, The entire team thought through, motivated. Staff are not always encouraged to speak up. They had to learn, be encouraged. We held workshops to brainstorm how to increase institutional deliveries. Groups across all levels [village nurses, doctors, etc.] made suggestions. One doctor said, Why dont you do case studies of maternal deaths? We found that questions were asked if deaths occurred in an institution, but not on the road (interview no. 35). As a result, Chunkath introduced the states first maternal death audits in the mid-1990s. It was an action that proved a turning point for safe motherhood in Tamil Nadu. Record keeping and reporting on maternal deaths were irregular at best before the audit system was developed, but the new system helped to build an evidence base. It was important to learn more about the circumstances that contributed to maternal deaths, as

123 well as where and how many were occurring, to inform policymaking and mobilize support and resources. Significantly, Chunkath and the team of public health and other allies she worked with understood the problem to be bigger than the epidemiological causes of maternal deaths that were typically recorded in health facilities. They believed social causes of death played as least as significant a role as biomedical causes and social causes could extend to structures of the public health system. They designed the audits to reflect this, beginning with a series of case studies carefully documenting perspectives of medical personnel and families. They learned that nearly a third of women died in transit to hospitals (interview nos. 29a, 35; Government of Tamil Nadu 2003). They also recognized that a preference for delivering at home delayed recognition and treatment of emergencies. By the time women and their families decided to go to a hospital, transportation and money for fees had to be arranged, then distances traveled. Additionally, primary health facilities often referred women to other facilities, commonly multiple other health facilities, due to poor health infrastructure and shortages in staffing that restricted the availability of safe motherhood services. The audits began to capture dynamics of the problem from social, medical and health system management perspectives. Chunkath and her team set about using evidence from the audits to help political and administrative actors understand and persuade them to take closer interest in the issue in order to motivate support for safe motherhood initiatives (interview nos. 5b, 29a, 32a, 35, 39a). One informant close to this process explained that they needed to take maternal death from a private tragedy to a public scandal, to make government responsible (interview no. 35). They used evidence from the audits to convince people there was a

124 problem and framed maternal death as a disgrace so that public officials would feel shame and be motivated to take responsibility for alleviating the problem (interview no. 32a). One public health official recalled Chunkath asking questions like, If it were your wife, what would you do? of doctors and politicians (interview no. 39a). It was important to convince actors throughout the system to identify personally with the cause and commit to it. The audits became a powerful instrument for getting other stakeholders on board. Several health policy experts and officials spoke about the role of bureaucrats and technocrats in bringing important health issues to the attention of political representatives and taking a leadership role in their promotion in political and administrative circles something they connected with initiative for safe motherhood in the state. The following quotes represent some of their insights: A senior public health official involved in promoting safe motherhood in the state said, There is political priority. Both governments [parties] are very willing to take it on. We educated the leadership, stated the problem, called for maternal health rallies (interview no. 35). One public health expert with an international development agency explained, Bureaucratic will, technical people and administrators have power to communicate and convince others and other levels (of government) to do. They have an opportunity to make a proposal and try to convince (interview no. 29a). The health bureaucracy that these actors represent, the Department of Public Health, is a powerful forum itself, giving officials voice with the health secretary and ministry to attempt to gain additional support for their pet projects.

125 A retired senior health official that was involved in the 1990s noted the complementary functions of political priority and bureaucratic leadership at the time: Tamil Nadus government is one of the progressive states in the country. There was a lot of political will for health and other development issues at the time education. Second was bureaucrats there at the time. The mission/vision of government was there, but an enthusiastic implementation by bureaucrats successive health secretaries who were dynamic, had much rural health consciousness, wanted to support, and the health secretariat also (interview no. 94). He named Chunkath as a key figure and suggested leadership continuity across political and bureaucratic actors played an important role in sustaining particular health policy efforts. Others also commented on the role of Chunkath, in particular: The focus on maternal health started in 1996 when Sheela Rani Chunkath was MCH [Maternal and Child Health] director. She started introducing womens issues and then turned to maternal health. She started data collection, spurred maternal death audits, a senior public health official explained (interview no. 32a). Another senior public health official said, Sheela Rani was an inspiring force for me. She uses force, the power of position, listens to others in forming policy and then insists the decision be implemented once made. She gave momentum and amplification to safe motherhood efforts. So many technical people are committed, but we needed her leadership, support. And she provided

126 momentum. Her name is so important in the history of MCH in Tamil Nadu (interview no. 39a). Importantly, Chunkath and other technical and bureaucratic champions of the cause used their positions and the maternal death audits to shape policymaking (interview nos. 5b, 29a, 50). The audits helped identify shortcomings in transportation needs, blood supply, and human resources. They informed a series of projects, programs and government orders to address these problems over the years, many with Chunkaths signature as health secretary on them. This signaled political priority and administrative support for maternal health, developed necessary health infrastructure and institutionalized systems of monitoring and support to guide service delivery (interview no. 5b). The previous chapter detailed a series of government orders that contributed to these efforts between 1999 and 2006. Government policy notes, a tool that lays out recent achievements and plans during a given fiscal year, are another source of information on government support for safe motherhood policy. Maternal health featured prominently in policy notes reviewed for the years 2002-3 (earliest available) through 2008-9. The following summarizes select items pertaining to promotion of safe motherhood through health system development in Tamil Nadu as highlighted in Policy Notes between 2002-3 and 2008-9 (Government of Tamil Nadu 2002-2008): Policy Note 2002-3: 105 Health Sub-centers provided water and electricity; Village Health Nurses (VHN) received accessible loans for mopeds and training to improve mobility; the Danida-Assisted Tamil Nadu Area Health Care Project supported expansion of the drug warehouse system used by the Tamil Nadu Medical Supply Corporation to provide affordable and reliable drugs to the public

127 Policy Note 2003-4: Upgrading of Health Sub-centers and Primary Health Centers in focus districts under the Danida-Assisted Tamil Nadu Area Health Care Project was nearly complete; upgraded infrastructure and health management information systems were implemented in Madurai and Theni Policy Note 2004-5: Mobility and communication training provided for VHNS; two First Referral Units in each district targeted to become Comprehensive Emergency and Newborn Care facilities; hiring of anesthetists and gynecologists for First Referral Units and Primary Health Centers planned; development of emergency transport system with nongovernmental organizations planned Policy Note 2005-6: Ambulances provided to block Primary Health Centers and protocols developed for handling emergencies, including obstetric and newborn, under World Bank-sponsored Tamil Nadu Health System Project; 260 Primary Health Centers upgraded to provide 24-hour essential obstetric and newborn care; blood donation camps planned to improve blood bank supplies Policy Note 2006-7: Following success of 24-hour delivery services in Primary Health Centers piloted in 90 PHCs in 1999 and an additional 90 in 2005, plans to extend 24-hour delivery and emergency services to 600 more PHCs; Reddy Maternity Benefit Scheme to provide Rupees 6,000 at the rate of Rupees 1,000 per month to pregnant women below the poverty line; reported 66 Comprehensive Emergency Obstetric and Newborn Care (CEmONC) Centers established and observed increases in deliveries in government institutions, cesarean operations, and sonograms; additional 32 CEmONC centers planned for 2006-7

128 Policy Note 2007-8: reported that the government approved recruitment of 400 specialist doctors and 400 staff nurses for CEmONC services and posting of these health staff had already commenced Policy Note 2008-9: reported that Primary Health Centers conducted 86.5 percent more deliveries in 2007-8 than in the previous year with 1,000 Primary Health Centers upgraded to 24-hour delivery services staffed by 3,000 staff nurses The Policy Notes summarized here demonstrate a record of attention and action by the Government of Tamil Nadu to develop the public health system in ways that would promote safe motherhood in the state. The Policy Notes emphasized investments in critical infrastructure and human resources, but bureaucratic and technocratic leaders provided the mechanisms to improve how the system functioned for safe motherhood. Health officials understood that it was not enough to have facilities and staff available, but these inputs would have a marginal impact without addressing how the health system worked. Chunkath and other champions of safe motherhood in Tamil Nadu understood that certain social norms embedded in health system structures could work to promote or discourage safe motherhood. They appealed to those institutions to benefit the cause. Some of their strategies appealed to established hierarchical rules, such as systems for monitoring, reporting and review by senior officials. Other strategies challenged established power relationships. For example, through symbolic expressions of value and respect for female health workers, such as improving village health facilities (Subcenters) with their input, improving their mobility and improving their ability to communicate with superiors and community members. Continued leadership was critical

129 to instituting these changes, but change to institutional rules and norms meant broader and more sustainable health system improvements to promote safe motherhood. To begin, in such a rigidly hierarchical society and health system, attention from senior officials, such as the director of public health, health secretary and district collectors sent a message that safe motherhood was important and should be acted upon. Health officials conveyed this message by requiring that each maternal death be reported up the chain of command, giving attention from actors in positions of power (interview nos. 5a, 29a, 35, 39a). One informant recalled maternal deaths being elevated to the status of any infectious disease high priority with notification by telegram to senior health officials (interview no. 5a). Medical officers and district health officials gave telegrams [reporting] on maternal deaths to the district collector, health secretary, RCH [Reproductive and Child Health] commissioner, director of public health we simply made a fuss, so it comes into the consciousness of people (interview no. 35). This helped to increase attention and establish value for the issue at higher levels of the public health system that would trickle down to the service delivery level. Part of making the hierarchy work for safe motherhood involved establishing a monitoring role for district collectors. This proved a powerful force. As the senior public official in each district, collectors were responsible for dealing with such health system issues as doctor absences, sub-par facilities, infrastructure development and program implementation, with maternal and child health being significant priorities. They took their responsibilities for making sure these programs were reaching people seriously (interview nos. 2, 31, 35, 67). As supervisors of the maternal death audit process they served a monitoring function, responding to public complaints about health services and

130 making the health system more responsive to the publics needs (interview nos. 32a, 35). They held monthly maternal death audit reviews with a committee of local health experts and families. Monitoring by district collectors reinforced the message that the issue was important and required attention and action at all levels. For service delivery personnel, monitoring by collectors made everyone aware of [the issues] importance (interview no. 79). One nurse supervisor described the mortal fear created among everybody by the prospect of collectors finding problems in the system (interview no. 67). Fear of sanction was one form of ensuring compliance. Others suggested that strong monitoring contributed to learning how to avoid repeating the same mistakes (interview nos. 2, 32a, 34, 82). Monitoring by district collectors channeled the power inherent in the hierarchical health governance system to provide a feedback mechanism that informed future policy and service delivery. Circumstances surrounding maternal deaths were also reviewed by the senior district health official, the deputy director of health services, together with all health service personnel involved. I was invited to observe a meeting in progress in one district. Two medical officers, four village health nurses and a nurse supervisor attended the meeting with the deputy director. He solicited insights to the circumstances surrounding the maternal death under review from each health worker. The director learned that the woman had been what they call a visiting case. Customarily, a woman returns to her parents home to deliver her first child after spending most of the pregnancy at the home of her husband and his family. It is difficult for public health officials to track and offer services in these cases. Village health nurses may not learn about the pregnancy until a very late stage, if at all, resulting in less thorough attention to prenatal care and less time

131 to establish relationships and encourage institutional or skilled attendance at delivery. In this case, the woman went to her natal home without documentation of prenatal care. The director found that the village health nurse did not properly assess the potential for risk in the pregnancy as indicated by the womans low weight gain during pregnancy and young age (18). The womans family took her to four different health facilities before she died. The director emphasized the need for the supervising medical officer to review village health nurses reports on pregnancies and suggested the nurse needed additional training to update her skills and knowledge to better assess pregnancy risks. It was unclear to me why the woman went to four different health facilities to receive care, but it was evident that the director approached the meeting as an opportunity to learn how to prevent deaths in the future at the individual and local system levels. After the meeting the deputy director explained the philosophy behind the review meetings: A meeting approach, discussion is the best approach to preventing [maternal deaths] and building reform. The next time they will pay attention and not make the same mistakes. Every month we are reviewing, having grievance meetings, so all hospitals are looked after to improve quality. Most issues are sorted out quickly. We are eliminating the social causes of maternal death so only medical causes remain (interview no. 82). The audit and review systems leveraged the power of institutional rules (hierarchy) to learn from individual cases. Policy actors used these lessons to develop and implement solutions to make the health system function better for safe motherhood in the future. Although established tools of hierarchy were powerful, champions of the cause in Tamil Nadu understood that these were limited and could serve to reinforce social

132 inequities that prevented the system from functioning better for safe motherhood. Leaders like Chunkath and Dr. Padmanaban, the director of public health at the time of this study, asked important questions about how gender inequality, poverty and social class, as well as shortcomings in health service delivery contributed to the problem. They understood that the socio-cultural influences that restricted womens access to safe motherhood services at the family and community levels were embedded in the health system and undertook to overcome some of these biases through institutional changes to the system. They recruited a core group of individuals and organizations that shared their understanding of the problem and solutions and worked to re-direct the system to function on behalf of safe motherhood (interview no. 32a). At the service delivery level, the frontline health workers that were the point of contact and care for rural communities were at the bottom of the hierarchy yet everyone from primary health physicians to senior officials recognized village health nurses as the backbone of health care (interview no. 31), providing the closest link between rural communities and the health system (interview nos. 51, 68; Narayanan 2003). They were and are responsible for educating women and their families about potential health risks related to pregnancy and the benefits of receiving prenatal, postnatal and institutional delivery care. Providing this care was in their job description, yet as little as ten years ago it was common for village health nurses to lack knowledge, skills and resources to carry out these duties effectively. It was not only a matter of commitment gender bias and low social status contributed to neglect of support for the needs of village health nurses to effectively carry out their maternal and child health care duties (interview no. 35;

133 Narayanan 2003). They had little voice in the health system, few resources to support them, and faced problems being accepted by communities. The Danida-Assisted Tamil Nadu Area Health Care Project helped to address these issues in high-need districts between 1996 and 2001, an effort Chunkath coordinated with a group of individuals that were sensitive to gender dynamics in the health and social systems. Chunkath and the team she worked with invited village health nurses to express their needs to senior officials with promise of having them met. Chunkath helped break down the power distance, communicating directly with the nurses and helping them attain voice and a modicum of respect in the health system (interview nos. 5c, 35, 67). To this end, the nurses formulated plans for new and upgraded health sub-centers; their health and safety concerns were addressed with community leaders; the government arranged to provide them with mopeds to improve their outreach abilities; and they participated in communication and technical skills training (Narayanan 2003). The Danida project invested in village health nurses needs for health, safety, technical skills, and support and helped them to achieve a different status in communities so that they could help make the health system function better for maternal mortality reduction. It reflected a change in bureaucratic priorities, as well as institutional rules and norms. Dharmapuri was a focal district for the Danida-sponsored project. The district had some of the lowest safe motherhood indicators in the state when the project commenced. I visited the district and spoke with health officials and frontline workers in April of 2007. The following quotes represent their perspectives on the influence of the project and changes in health system structures for safe motherhood:

134 One district health official said, the construction of HSCs [Sub-centers]. They were very tiny before. VHNs status was apparent. Why should they have larger and more comfortable facilities? Madam Sheela Rani [Chunkath] called together VHNs to plan the size, needs. VHNs should feel comfortable to stay there. Now 75 percent of VHNs stay there. Before it was 20 percent. HSCs are now functioning well. VHNs are providing good service. Its a chain link. New HSCs made convenient, mopeds, VHNs able to communicate and express ideas. In turn, people came for deliveries and VHNs could spend more time on health services. Thereby all health indicators improved (interview no. 67). Asked what differences she observed between when she started as a VHN twelve years ago and now, one VHN said, literacy improved, education increased, VHNs are going to the community to counsel and educate. Two-wheeler helps to go to field visits its easy to go (interview no. 53). We are motivating VHNs at review meetings to encourage institutional deliveries. At the DMCHO [District Maternal and Child Health Officer] monthly review meeting we talk about the kind of challenges they are facing, how to improve, who has a complicated case. It helps plan how to deal with cases that are high-risk, a sector health nurse said (interview no. 52). Village health nurses described how their improved skills, mobility and status helped them to assist women in accessing maternal health care at Sub-centers and at higher-level facilities. Between the 1998-9 and 2003-4 Reproductive and Child Health Surveys, institutional delivery rates in the district increased from 63 to 75 percent and safe deliveries attended by a doctor or nurse increased from 69 to 80 percent. Local and state

135 health officials and others involved in the Danida project attributed part of this improvement and better health system functioning for safe motherhood to village health nursess elevated status in Dharmapuri, along with other health system strengthening initiatives and policies. Officials and frontline workers gave credit to similar investments in Theni district, as well. Thenis institutional delivery rates increased from 69 to 90 percent between 1998-9 and 2003-4, under a special project of the first World Banksponsored Reproductive and Child Health program that featured health system strengthening, measures to improve gender equity in health, and improvements in maternal and child health (Government of India 2004; Government of Tamil Nadu 2003). It is important to recognize that even as safe motherhood gained traction, the public health system remained oriented to other priority areas priority areas determined in broader national and international political arenas, as well as at the state level. At the national level, family planning and, perhaps to a lesser extent, immunization as part of the child survival agenda, took precedent over safe motherhood historically. Indias national population control policies date to the early 1950s with strong family planning targets guiding vigorous implementation into the 1990s (Academy for Nursing Studies 2005; Visaria et al. 1999). Of Indias first nationwide program to address safe motherhood, the Child Survival and Safe Motherhood program launched in 1992, Visaria and colleagues wrote, The Family Welfare Program, however, continued to emphasize family planning service, and the child survival components of the new program especially the expansion of child immunization services were implemented earlier than the safe motherhood components (1999, p. S44). In 2005, a national study of public health nursing stated, The roles and responsibilities of ANMs have undergone changes

136 according to changing national priorities and programmes. During the [19]80s and 90s, the stress on family planning and immunization alienated the ANM from maternal and child health (Academy for Nursing Studies 2005, p. 26). Indeed, Van Hollen (2003) documented ongoing priority for family planning in Tamil Nadus public health system through the mid-1990s and I also observed this on my site visits to districts. But Tamil Nadus bureaucrats and technocrats had integrated priority for maternal mortality reduction into the health system alongside other goals, such as family planning, by the time I studied the issue. This was evidenced, for example, in the states approach to evaluating district-level progress on National Rural Health Mission goals. Health officials in two districts enthusiastically told me about how much their districts improved in newly released rankings based on institutional delivery rates, family planning and outreach performance from 2006 to 2007 and that they were going to do better than x district the following year. A long-time observer of Tamil Nadus public health system explained that there is a competitive spirit between district officials to improve and outperform other districts from one year to the next (interview no. 5c), so this was an effective mechanism to ensure attention for safe motherhood at the district level. To give another example, on a site visit to Dharmapuri district, officials proudly introduced me to Ms. C. Mahalakshmi, a village health nurse and recent winner of the Florence Nightingale National Award for service. A few weeks later The Hindu, the major English-language daily newspaper in Tamil Nadu, featured her story in an article (Prasad 2007). She received the award based on improved prenatal and community outreach services in addition to increasing the institutional delivery rate in her remote and

137 rural area to 97 percent. She was awarded a cash prize of Rupees 50,000 (about $1,200 US) and a certificate of recognition. Informants explained that her story would set a high standard and serve as an inspiration to other village health nurses to improve maternal health performance (interview nos. 5c, 51). The award served to set new standards and expectations for safe motherhood performance. Significantly, senior health officials took pains to demonstrate the kind of priority they had and wanted the health system to have for safe motherhood to change performance expectations and practices at the service delivery level. For example, one nurse supervisor explained, Madam [Chunkath] was a role model. She used to come every week, go into dirty houses without hesitation. We saw [how] to draw closer to the community (interview no. 67). This helped nurses better understand how to help the health system function better by forming relationships with women and communities, regardless of status. It became more important to ensure the system promoted maternal health than to let status barriers get in the way. In another example, a former deputy director of public health at the district level told me about a situation in which a medical officer at a rural health center reported that he referred a woman that was bleeding before delivery to a higher-level hospital with a blood bank. The woman never arrived. The director instructed the medical officer to follow up with her. He reported back that she and her husband had gone in search of a moneylender because they did not have money on hand for the unofficial fees that would be required at the higher-level, technically free, government institution. Guided by goal rather than rule, the director told the medical officer to give the family Rupees 500 to go to the hospital and that he would reimburse him from his own pocket. Reflecting on

138 the incident he explained, It was a near miss case. It shows referral [from one health facility to another] doesnt work, but close accompaniment and doing whats necessary does. That official instituted a policy that all maternity patients needing to go to higher level public health facilities in his district be accompanied by a nurse or doctor to prevent unnecessary delays in receiving appropriate care, one of the most important factors in maternal deaths. The officials actions on behalf of the individual case and at the policy level demonstrated how structural changes in Tamil Nadus public health system were used to promote safe motherhood. That policy was later adopted more widely and funds were officially allocated to cover transportation costs in cases requiring emergency care, demonstrating broader commitment to the issue at the state level. It is important to recognize that Tamil Nadu acted to orient its public health system to promote safe motherhood prior to any meaningful commitment at the national level. The National Rural Health Mission only came onto the scene in 2006 and Tamil Nadus safe motherhood policies helped to inform the reproductive and child health programs that fall under the Missions umbrella (interview nos. 2, 3). A representative of an international donor agency described how Tamil Nadu took leadership, diverged from the Centre. They said to the Centre, Okay, but were going to do it this way (interview no. 2) and that made a difference for how safe motherhood was integrated amongst other health priorities in the state. Tamil Nadu did not follow the Government of Indias health policy prescriptions as closely as many other states, suggesting a nuanced understanding of the macro political environment is important to understanding the policy process.

139 THE PUBLIC HEALTH BUREAUCRACY IN KARNATAKA Karnatakas experience with safe motherhood followed health priorities determined at the national level more closely. In the absence of state-level initiative, national programs had an important impact on how Karnatakas health bureaucracy provided for safe motherhood. In 2007, national health priorities were still oriented to family planning, although safe motherhood and child survival recently rose on the national health agenda. To illustrate, the national implementation plan for the Reproductive and Child Health Phase 2 under the National Rural Health Mission identified decreases in maternal mortality, infant mortality and fertility rates as its primary aims (Government of India undated, p. 8), but these were identified as short-term goals for the program. Medium- and long-term goals were confined to reducing the total fertility rate to achieve population stabilization. These goals demonstrated strong priority for family planning even as safe motherhood gained status on the national health agenda. The Janani Suraksha Yojana (JSY) program, part of the Reproductive and Child Health 2 program under the National Rural Health Mission, integrated family planning and safe motherhood goals. Under the original JSY guidelines, women living below the poverty line were to receive compensation for participating in safe motherhood activities, such as antenatal care and institutional delivery, but only up to two live births. Women would not receive benefits for children born after the first two in states with better safe motherhood indicators, such as Karnataka and Tamil Nadu. Under a separate but linked program, cash incentives were to be offered to women who accepted permanent sterilization methods after an institutional delivery. The Government of India later removed the two-child limit from the JSY program, but its inclusion demonstrated that

140 priority for family planning remained strong even with increasing programmatic attention to safe motherhood. Both Karnataka and Tamil Nadu participated in this program Tamil Nadu had an additional, separate program (the Reddy Maternity Benefit Scheme) under which the state provided a substantially larger sum to women than the national program did. Unlike in Tamil Nadu, Karnataka used the JSY program as its main strategy to improve safe motherhood at the time of this study. The Reproductive and Child Health Officer in one district reported that immunizations and monitoring JSY participation and benefits were his top priorities for field visits to Primary Health Centers This is the process indicator, he said (interview no. 96a). This and other observations suggested that Karnatakas health bureaucracy encouraged tracking of JSY benefits, but not necessarily important indicators of quality or survival outcomes. At the state level, Karnatakas budget analysis for the Reproductive and Child Health Policy Program (Government of Karnataka 2005) revealed significant resource allocation to family planning, safe motherhood and immunization. Over the five-year implementation period, the family planning services heading (including JSY compensation, tubectomy incentives, a pilot project for injectible contraceptives, and incentives for non-scalpel vasectomy) totaled Rupees 1.6 billion. Breaking down the category, JSY compensation totaled Rupees 9.9 million and family planning incentives totaled Rupees 6.2 million 6.1 million of this was for tubectomy incentives. Family planning operations were projected to outnumber JSY incentives by a 4:3 ratio: an estimated forty million tubectomy incentives would be given annually compared to thirty million JSY incentives. Support for immunizations, including supply of vaccines and

141 cold chain maintenance to preserve them, was budgeted at Rupees 1.59 billion nearly equaling the total budget for family planning and JSY incentives. To provide some historical perspective on how national health programs influenced bureaucratic priorities for family planning, the Karnataka Human Development Report 1999 discussed the impact of the Child Survival and Safe Motherhood program between 1992 and 1997 in Karnataka: The program provides for training dais and Auxiliary Nurse-cum-Midwives [ANM] and setting up first referral units to ensure safe deliveries. As elsewhere in the country, pressure on the ANM to achieve family welfare targets has resulted in neglect of maternal and child health work. Incentives are given only for family welfare, not for immunisation, birth spacing, ante and postnatal checkups or the distribution of nutritional supplements. (1999, p. 124). As noted above, other studies documented structural mechanisms for promoting family planning during this time period and extending into the 2000s (Academy for Nursing Studies 2005; Visaria et al. 1999). George and colleagues (2005) documented top priorities of frontline health workers in Karnataka and found that immunizations and family planning led maternal health. The authors surveyed ANMs in Koppal district between 2001 and 2003 to learn what they perceived as most pressing for their work in rural villages. They found that ANMs listed immunization as being the 1st and 2nd most important activities that they face pressure to implement, followed by family planning, with communicable diseases coming in last. Very few ANMs mentioned ANC or reproductive and child health (RCH) and only one mentioned post natal care in particular (2005, p. 16). The public health system was set up to provide for family

142 planning and immunization needs with little indication that maternal health was to receive the same level of attention. Health system priorities to promote family planning and immunization over safe motherhood extended to the Primary Health Center level. The Karnataka Human Development Report 1999 reported, PHCs [Primary Health Centers] have been seen more as institutions providing family planning services. Full time nursing care being absent in most PHCs, it is uncommon for them to even attend to normal deliveries (p. 36). George and colleagues findings based on case studies of five maternal deaths in Koppal are on point. They found: There is a lack of attention given to women seeking assistance during delivery when other competing health tasks are at hand. PHC doctors are more accountable for routine outpatient care, pulse polio campaigns, routine immunization work, tubectomy camps or even administrative meetings, than attending to women with potentially life-threatening obstetric complications. There is no back up system that ensures follow up so that women receive prompt and appropriate medical care attention during these other routine tasks or scheduled government holidays (2005, p. 29). My observations at the primary health care level in the state were consistent with George and colleagues findings. I found little evidence that the health bureaucracy featured any mechanisms to ensure responsiveness to womens needs for safe delivery or emergency obstetric care services. Checking off antenatal care indicators and administering vaccines seemed important at the service delivery level, but there was less evidence that the health bureaucracy systematically ensured that frontline workers were provided with the skills,

143 resources or support needed to help women access safe motherhood services when needed. George and colleagues study of maternal health care in Koppal district of Karnataka provided further insights to how health system structures in the state detracted from safe motherhood goals. Their study revealed that crucial health infrastructure required to provide emergency obstetric care was underprovided in the states health system. Koppal, a northern district, was tied with Raichur for lowest institutional delivery rate in the state in 2002-4 the two districts formed one larger district until recently. Just over twenty percent of deliveries in the districts occurred in health institutions in 2002-4 (Government of India 2004). The authors reported, In Koppal, investment in infrastructure has not translated into comprehensive emergency obstetric care as none of the higher-level government facilities have all the required specialists or critical supplies (George et al. 2005, p. 12). The authors found that there were no Basic Emergency Obstetric Care and no Comprehensive Emergency Obstetric Care facilities in Koppal that were fully functional according to United Nations guidelines. Missing components of Basic Emergency Obstetric Care included availability of anti-convulsant drugs and removal of retained products of conception (Safe abortion/Manual Vacuum Aspiration) and for Comprehensive EOC, blood transfusions and surgery (p. 13). Allow me to repeat, no basic or comprehensive emergency obstetric services were available in Koppal district at the time of George and colleagues study between 2002 and 2004. That meant that the states health bureaucracy failed to provide even the most basic structural elements necessary for promoting safe motherhood for emergency cases in that district regardless of whether any antenatal, skilled delivery or postpartum care was involved.

144 According to the Government of India, First Referral Units are equipped to provide round-the-clock services for Emergency Obstetric and New-born Care, in addition to all emergencies that any hospital is required to provide (Government of India 2004). According to the guidelines, such services as surgery, including Cesarean sections, and blood storage facilities must be available. Karnataka reported that all Community Health Centers in the state were First Referral Units in 2005 (Government of India 2006). The evidence from Koppal has been presented. In the course of my research, district officials and service delivery personnel in Mysore and Kodagu, two better performing districts, reported that some of their health facilities failed to meet these standards Speaking to the problem of maternal mortality, a Mysore district official said, There is a shortage of anesthetists and gynecologists in government hospitals. We dont have operating equipment in many taluks, so we refer cases to major institutions, but it takes time and complications arise (interview no. 22). It seems likely that was the case in other districts, as well. Tamil Nadu faced similar problems, but took action to make First Referral Units fully functional on a strategic basis to fill this gap in the system as discussed earlier in this chapter. Karnatakas systems for reporting and data collection also showed relatively little priority to safe motherhood in the states health bureaucracy. In 2001, The Karnataka Health Task Force reported that there were four statistics officers at the district level one each for family planning, immunization, tuberculosis control and leprosy. Reproductive and child health, including safe motherhood, were monitored by agencies outside Karnatakas public health system. The Task Force reported that certain data collection activities for Reproductive and Child Health and Primary Health Center

145 facilities status (e.g. drugs and equipment availability) were carried out by independent agencies with minimal coordination with the DHS. Reports are not regularly received by the DHS and there would appear to be little feedback into the health management system of the conclusions of such surveys (2001, Ch. 14). By way of contrast, Tamil Nadu instituted its system of maternal death audits and reviews by senior officials in the mid-1990s and also developed a health management information system to improve data collection at the primary health level (Government of Tamil Nadu 2003). As a result, the state could analyze maternal health data at the district level something most states did not have (interview no. 35). George and colleagues identified other data-related problems that showed relative inattention to safe motherhoods ultimate outcome survival in Karnatakas health system (2005). The authors reported on the emphasis of data collection on outputs, such as the numbers of women visited for antenatal care or numbers of immunizations or iron tablets given, to the neglect of health outcomes. They concluded that for ANMs and Medical Officers, The actual health outcome of the women they see, whether she survived or not, becomes incidental to their work because the outcome is not closely monitored (George et al. 2005, p. 28). George and colleagues reported, There is a government order requiring the medical officer and higher-level officers to review each maternal death, but this is rarely done. It is not in their interests to report problems (2005, p. 30). Officials in two of the districts I visited confirmed that maternal deaths were not regularly reviewed. One said, Not all [maternal] deaths are getting verification from district and taluk level. Competing priorities, meetings, are keeping them from it. Some of the verification is not right (interview no. 87).

146 Karnatakas maternal death audit form further revealed how little attention the health bureaucracy required be given to safe motherhood outcomes (George et al. 2005). The two-page form included brief demographic information; antenatal care check-offs (e.g. tetanus immunization and use of iron tablets); and a perfunctory timeline of the labor process, any complications that arose and any referrals to other facilities. There was space to document the date, time and place of death and the date and time the information was relayed to the District Health Officer, the senior district health official in Karnataka. Five very short lines were dedicated to the probable cause of death. The table is reproduced in exact proportions below: Figure 6.1 Maternal death audit form, Karnataka excerpt 1 2 Medical causes APH/PPH/Anaemia Avoidable factors / causes in the hospitals 3 4 5 Economic causes Social causes Brief summary of findings about the causes of death Source: Unpublished document author obtained from district health officials in Karnataka, original copy available from author There was also a line underneath the table that stated other information could be attached. I read more than twenty of these forms in one district and collected four representative samples. There were no attachments. The data on these forms was scant. None offered any but a biomedical explanation for the cause of death one did not offer that. It is very

147 difficult if not impossible to ascertain why these women died and what the health system could have done to prevent it. None of my samples indicated that the District Health Officer ever received it. The data collected on these forms was insufficient to gain a deep understanding of the problems contributing to maternal mortality and were likely underutilized for effective problem solving at the district level. The Karnataka Health Task Force came to much the same conclusion at the state level, suggesting that aggregate data was underutilized for meaningful monitoring, planning and action at the micro level or for strategic planning and management of the Health system (Karnataka Health Task Force 2001, Ch. 14). These translated to little meaningful attention to safe motherhood in Karnataka. CONCLUSION This chapter is about what policy actors did, what influenced their actions and what impacts it had for availability and access to safe motherhood services in the respective states. In Tamil Nadu, a health bureaucracy with rules, norms and priorities oriented to public health and social welfare goals facilitated emergence of effective leadership for the issue, initiative, strategic planning and integration among other priority issues. Policy actors framed the issue as one that resonated with ixisting ideas about what the health bureaucracy should do, gathered evidence and took strategic action. Importantly, as representatives of the health bureaucracy, key policy actors were well positioned to make authoritative decisions. This affected the governments program of action and helped to elevate the status of the issue on political and bureaucratic agendas. Next to Tamil Nadu, leadership for safe motherhood in Karnataka was conspicuously absent. Any bureaucratic attention to safe motherhood was channeled

148 through Government of India-sponsored programs this because the states political agenda did not influence priorities in the health bureaucracy toward primary health and social equity issues like safe motherhood. Recent priority for safe motherhood at the national level primarily trickled down to Karnataka through the JSY program. The JSY program promoted community demand for maternal health services, but failed to provide mechanisms to make the health system responsive to this demand. The JSY program also fed into the Karnataka health systems orientation to outputs checking off antenatal care visits, immunizations and institutional deliveries without checks for quality of maternal health care or survival outcomes. The program did little to functionally integrate safe motherhood into a health system historically oriented to family planning and immunization goals set by political principals. Table 6.1 below summarizes the ways in which varying political priorities translated through the health bureaucracy to programs of action on behalf of safe motherhood in the two states. The findings suggest that the rules, norms, and allocation of attention and resources in the health bureaucracy have important implications for the emergency of leadership, coordinated policy action, strategic planning and integration of new priorities among those pre-existing. Table 6.1 Translation of political priorities in the health bureaucracy Tamil Nadu Initiative Emergence of coordinated advocacy Strategic action and planning Integration among other priorities Karnataka Formal implementation No coordinated policy advocacy Status quo Lack of integration among priority issues

149

CHAPTER 7

TOWARD A MORE HOLISTIC FRAMEWORK FOR ANALYZING POLICY PROCESSES

150 INTRODUCTION This study began with an interesting empirical puzzle. What explains differing safe motherhood outcomes in two Indian states with comparable socio-economic indicators and competitive multi-party political environments? Significantly, one state provided more equitable access to publicly provided safe motherhood services than did the other, improving overall rates of access to a greater extent. The investigation undertaken to explain these empirical puzzles worked backward from that point, finding that a complex set of policymaking factors affected differing results in the two states. Existing frameworks of policy analysis provide important clues to identifying explanatory variables and relationships, but do not fully explain the findings of this study. The findings of this study suggest that social-historical factors are under-represented and the crucial link formed by bureaucratic leaders between policy agendas and agency practice is under-specified in existing frameworks of analysis. Some other variables are in need of refinement to become more meaningful to our analyses, such as the characteristics of major political parties. In order to link policy processes and outcomes, this study requires a holistic framework of analysis that encompasses multiple categories of factors. Such a framework must account for public service delivery outcomes, links between policy and management, the political environment and social history. This study contributes to our understanding of the relationships between these parts and their relationships to each other through a holistic approach to analyzing policy processes. Following a brief summary of the case findings, this chapter discusses the implications of the findings for analyzing policy processes, limitations of the study, and conclusions.

151 SUMMARY OF FINDINGS Tamil Nadu Since the mid-1990s, bureaucratic leaders and other policy community members in Tamil Nadu deliberately pursued government policies and programs that were more responsive to problems surrounding the availability and accessibility of safe motherhood services than they did in Karnataka. Institutional norms supported bureaucratic leaders in Tamil Nadus Department of Public Health and the health ministry in taking initiative, organizing surrounding important issues, and designing and implementing maternal mortality reduction strategies. Their ideas about safe motherhood resonated with political priorities for social welfare promotion, particularly among more vulnerable groups. The priorities of political parties were reinforced by the system of partisan competition and powerful governing coalitions in the state factors significantly influenced by a broad twentieth century social movement promoting social equity along the lines of caste, class and gender such that there was continuous priority for social welfare, including health, across successive governing coalitions over several decades. The systematic, long-term nature of these influences on health policy processes in Tamil Nadu affected a relatively strong public health system that made possible early increases in rates of access to safe motherhood services in the state, in addition to facilitating later advocacy and action more directly on its behalf. Karnataka Policy actors in Karnataka remained weak and fragmented throughout the study period, failing to coalesce surrounding problems of availability and accessibility of safe motherhood services in the state. Their ideas failed to resonate with key government

152 officials in positions of bureaucratic or political leadership. In the absence of leadership for the cause, institutional norms affected a status quo approach to safe motherhood within the Department of Health and Family Welfare and the health ministry. That meant continuing attention to high-profile infrastructure development projects and formal implementation of reproductive health programs delegated by the national government. These were not enough to alleviate disparities in the availability and accessibility of safe motherhood services for more vulnerable groups, particularly those in the northeast region. Party politics in the state did little to foster or support strategic action for safe motherhood and factored into neglect of other aspects of the health system, such as rural health human resources, equipment and facilities, that facilitate womens access to safe motherhood services. In addition, recent partisan conflict between opposing coalition governments at the state and national levels may have erected further barriers to the allocation of attention and resources that could improve access to safe motherhood services in Karnataka. IMPLICATIONS OF FINDINGS FOR ANALYZING THE POLICY PROCESS Why do women in Tamil Nadu have more equitable access to safe motherhood services than do women in Karnataka? Why do women from more vulnerable groups in Tamil Nadu have such greater access to safe motherhood services, particularly in the public sector, than do women in Karnataka? An analytic approach that emphasizes agenda setting uncovers only part of the story and an approach that emphasizes implementation likewise reveals only some of the factors at work. Connections between policy agendas, decisions, practice and outcomes the intersections between policy and management are under-treated in existing frameworks for analyzing policy processes.

153 By using backward mapping from the point of service delivery to management, decisionmaking and agenda ascendance with reference to this study, this discussion develops these connections and contributes to a more holistic understanding of policy processes. Linking policy processes and outcomes The evidence in this study suggests that publicly provided safe motherhood services are more widely accessible and of higher quality in Tamil Nadu than in Karnataka. What explains this? Consistent with Ingraham, Joyce and Kneedler Donahues (2003) framework for analyzing public management capacity, this study finds that the strength of key agencies in terms of leadership, human and financial resources, and information and capital management represents an important set of factors affecting government performance (e.g. accessibility of maternal health services). While the network of public health institutions is vast in both cases, one is clearly ahead of the other on this count and this has important implications for the supply of and demand for safe motherhood services among more vulnerable populations. While the management capacity of key agencies to implement policy is an important consideration in policy analysis, it does not typically enter the analytic calculus of frameworks more concerned with how issues rise on agendas. The table in Appendix A shows how Sabatier and Weible (2007) deal with these by including them in a list of resources that policy communities use to advance their causes within subsystems in comparison to the framework of analysis developed in this study. This study proposes a more holistic approach, suggesting that these resources are the province of organizations and that the management capacity of agencies importantly influences outcomes of policy processes.

154 Institutional norms that communicate to street level bureaucrats in government agencies, including those nurses, doctors and health education workers, and their managers how they should and should not carry out their duties, how they should and should not understand problems, and how they should and should not address particular issues, are also important factors. Street level bureaucrats in Tamil Nadu have a much more sophisticated understanding of problems related to maternal mortality and the actions they can and are responsible to take to prevent it than they do in Karnataka. This affects how they carry out their duties from routine check-ups with expectant mothers to the lengths they go to see that they are treated for risk conditions and life-threatening complications. Organizational and professional norms shape the behavior of street level bureaucrats, as well as that of their managers and policymakers. They shape the character and quality of service delivery the availability and accessibility of safe motherhood services in the public sector a factor disproportionately affecting outcomes for more vulnerable groups. Ostroms (2007) work on institutional analysis and proponents of the organizational culture perspective (Martin 2002; Ott 1989; Schein 1992; Smircich 1983) find institutional norms to be important shapers of behavior and, therefore, important shapers of organizational outcomes. The identification of norms as organizational influences that connect policy with outcomes marks an important difference between the contributions of this study to policy analysis and existing frameworks (see Appendix A for a table comparing approaches). In existing scholarship, policy beliefs, images or ideas are commonly found to be key variables motivating policy communities and policy change. Policy scholars such as Baumgartner and Jones (1993), Kingdon (1984, 1995) and Sabatier and Jenkins-Smith

155 (1999) developed their frameworks of analysis to understand forces driving agendas and policy change, finding that ideas about problems and solutions shaped policy community behavior. They also found that ideas helped problems rise on public policy agendas and solutions reach decision points among authoritative actors. This study supports those findings and suggests that an extension of the concept as embedded in institutional norms furthers our ability to connect policies with outcomes through the organizations that are tapped to pursue policy goals, thus contributing to a more holistic framework for policy analysis. The institutional norms thus far discussed and the ideas that motivate policy communities and shape agendas are related but differ somewhat. Importantly, the latter tend to have a quality of newness while the former tend to be more established assumptions the sort of things everybody knows so there is no need to name them publicly. When new or different ideas confront the established order of assumptions, varying responses are possible. As suggested by Berman (2001), Keck and Sikkink (1998), Shiffman and Smith (2007) and Stone (2002), this study finds that new ideas are more likely to be adopted the more they resonate with existing institutional structures (political, organizational and professional norms). For example, in Tamil Nadu ideas about maternal mortality reduction found resonance with professional and organizational norms promoting public health agency performance on gender, class and rural-urban social equity issues. As such, Tamil Nadus policy community and entrepreneurial leaders found a more receptive audience that was more ready to bring safe motherhood goals and practices into the fold within the states public health system than was the case in Karnataka where such norms were weaker and where the political environment was

156 less supportive. The interplay between these factors is important to understanding relationships between policy, management and outcomes in a more holistic approach to analyzing policy processes. At the intersection of policy & management Policy communities, groups of individuals that interact to some degree surrounding concern for issues in a given policy area, and their leaders form a crucial link between the setting of policy agendas, decision-making and the pursuit of policy goals by government agencies and their partners. The cases examined in this study suggest that bureaucratic leaders who championed the cause with allies in the policy community and political principals, as well as managed the integration of priority for the issue and interventions to address it into the service delivery network were crucial to changes in Tamil Nadu over the past twelve to fifteen years. These policy entrepreneurs negotiated the political environment, framing ideas about maternal mortality reduction in ways that appealed to political leadership, to bring it onto the policy agenda and to draw attention and resources for address of the problem. With support from political principals, they used their decision authority to sign Government Orders, change agency rules and procedures and set new expectations among service delivery providers for performance of the public health system. In this case, bureaucratic leaders served the function of policy entrepreneurs that networked with other policy community members, promoted their ideas, gained the endorsement of political principals, and provided managerial guidance within an extensive, hierarchical public health bureaucracy. They did not serve any one function. Any one function would have been insufficient to affect change in policy and practice.

157 This finding suggests that the more adept bureaucratic leaders are at navigating this trio of competencies, the more likely they are to affect changes in policy and practice. This finding contributes to our understanding of the nature and importance of bureaucratic leadership in policy processes. A similar approach to understanding bureaucratic leadership is being explored by public management and governance scholars. OToole, Meier and Nicholson-Crotty (2005) draw on Moores (1995) concept of managing upward to political principals, downward to agency personnel, and outward to network participants to explain public performance outcomes in public school systems in Texas. As in this research, these scholars find a political-bureaucratic role for public leaders that is under-developed theoretically in public management and policy scholarship (Meier, OToole & Nicholson-Crotty 2004). This suggests a re-thinking of the roles of policy entrepreneurs and bureaucratic leaders in our frameworks of analysis. Bureaucratic leaders are not only formulators of alternative solutions (Kingdon 1984, 1995), potential resources to advocacy coalitions (Sabatier & Weible 2007), or managers of policy implementation they are potentially all of these things, as well as authoritative government actors with their own agendas and decision powers (Meier 1979; Rourke 1984). This is not to say that they are not subject to democratic accountability mechanisms leaders in Tamil Nadu were clear that they could not have acted without the support of political principals but to highlight the roles they play in bridging the gap in our frameworks of analysis between what we tend to think of as distinctively policy and distinctively management. Identifying such a variable is crucial to development of a more holistic framework for analyzing policy processes.

158 Political environments & policy change Tamil Nadus relatively early trend toward increased access to safe motherhood services, notably among more vulnerable groups, was affected by more than recent bureaucratic leadership and organizational factors specific to the cause. In both cases the strength, priorities and worldview of governing coalitions and competitors affected the capacity of the public health system to provide these services, the understanding of what the health system and its representatives should do with respect to maternal mortality reduction, and the opportunities for policy communities to frame safe motherhood as an issue that resonated with political and bureaucratic leaders. Differences in these factors across the cases systematically affected different trajectories for equitable and improved access to safe motherhood services in each state. The influences of these three factors strength, priorities and worldview of governing coalitions and their competitors are intricately intertwined. In Tamil Nadu, the two major political parties (the DMK and AIADMK) rose to power on the basis of their shared worldview that social equity goals in terms of class, caste and gender are central public policy aims. This is highly politicized. In an atmosphere of intense competition between the parties, each attempts to maintain or regain power at least in part by demonstrating priority for social equity goals through policies, programs, resource allocations and public criticism of their rivals. These arrangements have supported regular investment in building the capacity of the public health system and directing its resources to activities that have promoted safer motherhood over several decades. In addition, they have provided systematic opportunities for policy communities and leaders to advance new ideas about health and other social policy initiatives in ways that benefit

159 more vulnerable populations. This has benefited safe motherhood in indirect and direct ways, ranging from nutrition for lactating mothers being included in child nutrition programs more than thirty years ago to expansion of emergency obstetric care services more recently. Karnatakas political parties have not shown systematic inclination toward a worldview that strongly embraced social equity or pro-poor policy goals. Health reform in the early 2000s, for example, was a pet project of a particular chief minister, but political structures were not designed to sustain that priority. Complicating matters, political parties in the state have grown relatively weak, causing governing coalitions to become just that weak coalitions of political parties that have difficulty advancing policy agendas. Because there are no strong opposition parties and none that derive power from a deliberate social welfare agenda, the states political system lacks systematic incentives for state-level parties to advance policies, programs and resource allocations in ways that regularly benefit the public health system, and safe motherhood by extension. The structure of party politics in Karnataka has affected weak leadership in the public health system and provided fewer opportunities for the also relatively weak and fragmented policy community there to advance their ideas about maternal mortality reduction on policy agendas and in practice. The factors that emerge from analysis of the cases studied here with respect to the political environment differ somewhat from how they are presented in the advocacy coalition framework, the most directly comparable holistic framework of analysis to the one developed in this study. Changes in systemic governing coalition appears in early and more recent versions of the framework as the central variable concerning political

160 party power (Sabatier & Jenkins-Smith 1999; Sabatier & Weible 2007). Like Baumgartner and Jones (1993) and Kingdons (1984, 1995) work, the advocacy coalition framework suggests that changes in political leadership can affect shifts in the ways in which problems are understood, the types of solutions that are advanced and the allocation of resources and attention to particular issues. In a two-party system, such as the U.S. system on which these scholars base their research, change in systemic governing coalition may be enough to capture attendant changes in worldview and priorities that affect agency priorities and the opportunities for policy communities to advance issues on agendas. However, the cases examined in this research suggest that these should be named and that the strength of political parties should not be assumed (see the table in Appendix A comparing frameworks on this point). The refinement proposed in this study helps to expand the application of our frameworks of analysis to other political contexts and to enhance their explanatory power. Governance structures & policy processes In both cases informing this research, governance structures, including constitutional structures, authoritative levels of government and degree of centralization at the authoritative level of government, mattered in that they provided bases for case comparison and combined with the political variables in each state to affect differing trajectories for safe motherhood policy. The assumption that constitutional structures matter is consistent with existing policy scholarship (Baumgartner & Jones 1993; Kingdon 1984, 1995; Sabatier & Jenkins-Smith 1999; Sabatier & Weible 2007). The constitutional provision of shared responsibility for health policy and implementation between states and the national government in India formed part of the basis for

161 comparing the two cases. The national government plays a prominent role in setting general health policy goals and providing some financial resources, but state governments are primarily responsible for managing their public health systems (the national government exercises little oversight) and state governments also have substantial autonomy in formulating health policy and selecting priorities among national health goals. Therefore, health policy processes in India are fundamentally shaped by the intergovernmental nature of shared responsibilities between levels of government and analysis at the state level is crucial to linking policy processes to outcomes. Delegation of authority for health policy to the state level had an important impact in each case. In Tamil Nadu, it facilitated state-level initiative for public health and safe motherhood preceding the rise of the issue on the national policy agenda. Further, it enabled Tamil Nadus political leaders to allocate resources above and beyond national recommendations. In effect, the autonomy afforded the state in this regard, in addition to its relative insulation from conflicts between national level political parties, allowed state-level political dynamics to run their course with little external interference. And, because governance authority remained relatively concentrated with the governing coalition at the state level, the effects of their decisions on health policy and programs were more likely to spread widely throughout the state. Karnataka has a different story. Delegation of authority for health policy to the state meant that weak state-level parties lacking overt priority for social policy constrained systematic advances for the public health system in ways that supported safer motherhood for more vulnerable groups. Two additional factors played important roles in the case. First, intergovernmental relations between Karnataka and the national level

162 were more contentious than they were in Tamil Nadu. Karnatakas major political parties had closer ties to national level political parties. Conflicts between opposing parties sometimes manifest in passive resistance to uptake of national health goals and programs. Second, Karnataka historically has had a more decentralized approach to governance authority. In other words, local political power and priorities tend to trump state-level political power and priorities. This has translated to uneven political and administrative support for the public health system and various policy goals across the states twentyseven districts. This analysis is fairly consistent with existing scholarship, particularly Baumgartner and Jones (1993) assertions about how the decentralized nature of federalism affects agenda setting in policy processes. In federal systems, variation in policies and outcomes at the sub-national level is to be expected. This analysis reveals further, however, how important governance factors interact with differing political structures to produce varying results in a non-U.S. context. The structure of stronger and weaker parties engaged in political competition in India, with variation at the subnational level that affects how decentralized policymaking plays out, is an important relationship to recognize in our frameworks of analysis. The way in which differing worldviews and priorities are systematically channeled to influence more or less widespread political support for any policy initiative at the sub-national (or national) level is also important to consider. There is much room for further specification of these relationships and for our frameworks to better incorporate the dynamics of intergovernmental relationships in policy processes.

163 Social historical factors in policy processes Following existing scholarship, an assumption was made at the outset of this research that socio-economic conditions and socio-cultural structures were important factors to consider in analyzing the relationship between policy processes and outcomes (Sabatier & Jenkins-Smith 1999). As such, an attempt was made to hold these as constant as possible between the two cases to form a basis for comparison. Evidence suggests that the relatively small differences between Karnataka and Tamil Nadu on economic, educational, caste and geographic variables form an incomplete explanation for their more widely varying safe motherhood outcomes. They also form an insufficient explanation for the respective governments varying performance for provision of equitable access to safe motherhood services. This study suggests that an important social variable is missing in this equation. A key difference between these cases is the historic collective action effort that transformed society and politics in Tamil Nadu over the course of the twentieth century. The Periyarled social movement promoting social equity along the lines of class, caste and gender in Tamil Nadu had widespread effects on the political environment in the state. The movement gave rise to two powerful political parties that share worldview and priorities for social equity policies. It also spurred an active civil society that attempts to hold its political leaders to their promises on these counts. Karnataka, on the other hand, features a different social history. The state is made up of disparate regions. Some hail from more socially progressive roots, including the former princely state of Mysore, coastal regions of the former Madras Presidency and independent Coorg regions where access to safe motherhood services is advanced for

164 the state. Other regions, particularly the northeast territory formerly aligned with Hyderabad, joined the state under less prosperous conditions. These areas suffer political and administrative neglect and the gap in access to safe motherhood services between more resourced and more vulnerable populations is wide. Karnatakas regions have lacked a cohesive identity, bases for social organization remain elusive, and collective action to pursue social equity policies remains relatively fragmented and weak. What this suggests is that historical social movements, collective action achievements, and other feats of social organization, such as colonialism and statereorganization, have fundamental impacts on other factors in policy processes. Social history affects governance structures, political environments, organizational capacity, the relationship between people and their government. Policy processes are about negotiating what governments should do and how they do it. Historical social forces that impact contemporary policy processes are highly relevant and underrepresented in existing frameworks of analysis (see the table in Appendix A). This study suggests we exclude them at the risk of missing an important variable. To summarize, this study suggests that in order to understand why outcomes related to important policy goals vary over time, we need more holistic frameworks for analyzing policy processes. We need frameworks that address the relationships between social history, the political environment, key actors (e.g. advocates, decision makers and bureaucratic leaders), organizations and outcomes. Each of these categories is complex and deserving of individual analysis. But understanding the place of each category in a more holistic framework of analysis can help us to better grasp their significance and

165 meaning in relationship to each other as they influence such outcomes as equitable access to healthcare. This is also an important aim of policy scholarship. STUDY LIMITATIONS It is important to recognize some of the limitations of this study. Some are related to the case studies themselves and others to the generalizability of results. The comparative case study design, extensive data collection (including more than 140 interviews) and triangulation of data support the validity of the results. The design reduces to the extent possible the likelihood that socio-cultural and economic factors could better explain divergent outcomes in the two states, particularly with regard to the rates of access to safe motherhood services between women representing directly comparable vulnerable groups (e.g. lowest economic group, illiterate, lower caste, rural). A potential limitation is the extent to which the study focuses on public over private sector influences on safe motherhood policy and outcomes. Though the evidence suggests that the public sector plays a greater role in providing maternal health services to more vulnerable populations, both sectors play important roles in providing services to the population at large and the private sectors influence on divergent outcomes may be underestimated. Roles of professional associations and other private actors have influenced health policy more generally on issues that affect safe motherhood, such as personnel contracting policies for example, but are not treated extensively in this analysis. Their direct relationship to safe motherhood policy and outcomes was difficult to assess and data were limited. Data were also limited regarding the role media might have played in creating demand for services. This is an important issue for future research.

166 Further, this research examines a particular policy issue in a particular social and political setting, leaving open questions about the generalizability of results to other policy issues and settings. The study design could have been strengthened through comparative analysis of policy processes across health issue areas or across multiple policy arenas, such as health and education. Resource constraints prevented such an extensive undertaking for this study, but comparative analysis of this nature is on the authors future research agenda. With respect to the setting, the frameworks of analysis informing the theoretical approach to the study are largely derived from study of the U.S. context. Their general applicability (with some refinement) supports the suggestion that the framework developed in this study might be appropriately applied to other national and sub-national settings. The social historical category of factors might be instrumental in this regard as we see countries, such as Sri Lanka, featuring historic collective action efforts to extend basic health and human services to more vulnerable populations, subsequent activity surrounding safe motherhood policy and some of the most advanced safe motherhood indicators in the developing world (Pathmanathan et al. 2003; WHO 2007). Generalizability is always limited in exploratory studies, but such research is undertaken with this in mind with an eye toward future research and hypothesis testing. This study offers new insights to the roles of policy actors, party politics and social history in policy processes that require further investigation in different policy arenas and settings.

167 CONCLUSION This study began with an interesting empirical puzzle and a quest to understand it in a theoretically relevant way. It began with an understanding that there were differing outcomes on an important health issue between two seemingly similar Indian states. Why was Tamil Nadu leading Karnataka on this? Upon closer examination, more equitable access to safe motherhood services through Tamil Nadus public health system made a significant difference. But why and what could it contribute to our understanding of public policy and management? It became apparent early in the course of data collection that differences in bureaucratic leadership in the states made an important difference for availability and access to safe motherhood services, especially for more vulnerable groups. These actors operated at the intersection of management and policy, shaping the resources and institutional norms of their agencies and service delivery personnel, influencing the access and power of policy communities to promote ideas about maternal health care, and negotiating their political environments. Bureaucratic leaders in Tamil Nadu had more to work with greater organizational capacity and institutional norms that supported performance to address the needs of more vulnerable groups than they did in Karnataka. The political environment they worked in provided more opportunities to champion the cause, to frame it to resonate with the worldviews and priorities of major political parties. This was importantly shaped by the states social history, particularly a widespread social movement to promote social equity on the basis of class, caste and gender.

168 Other approaches to policy analysis fall short of fully accounting for what is observed in the cases examined in this study. Existing approaches would have us focus our lens of analysis on the efforts of policy communities to convince authoritative decision makers (typically legislative) to pay attention and dedicate resources to alleviation of a specific problem within a particular issue area. This is important in the cases at hand, but incomplete. Authoritative bureaucratic leadership and a cohesive policy community made a significant difference for safe motherhood policy in Tamil Nadu, but they built upon something more foundational and their reach extended beyond policy decisions to agency performance. Absent Tamil Nadus historical collective action efforts to promote social equity goals, the political environment would likely have been different broader policy priorities supporting access to health and education for more vulnerable groups would likely not be as strong. In other words, it is less likely that the ideas the policy community tried to promote about maternal mortality reduction would have resonated to the extent that they did with the political environment and it is less likely that spillover effects from other policies improving access to safe motherhood would have been as strong as they were absent this historical social influence. This leads to our first proposition: Proposition 1: Social historical factors, such as collective action efforts, that deeply affect the political environment (including the strength, worldview and priorities of governing coalitions and their opposition), shape the opportunities for policy communities to exert influence in subsequent policy processes.

169 It is not only a change in governing coalition that shapes policy processes, but the character of governing coalitions, their power and the strength of their opposition. These do not evolve in a vacuum. They are influenced by social historical conditions that alter the terms of political power and debate. Following Baumgartner and Jones (1993), this is what the great cleavages of society do (p. 21). They affect the broader set of conditions in which policy subsystems and policy communities are embedded, thereby influencing their opportunities to develop and exert influence. Our second proposition concerns the intersection of policy and management: Proposition 2: Bureaucratic leaders form a crucial link between the policy communities, political principals and agency management that in turn influence agenda setting, authoritative decisions and agency performance in pursuit of policy goals. The ways in which bureaucratic leaders manage these relationships profoundly influences policy processes. What is different about the approach to policy analysis presented here is the identification of this important link between what are often considered distinct parts of the policy process. Of course, we know that policy processes are messy, that stages overlap, that bureaucratic actors inform and make authoritative decisions as well as manage agencies in the pursuit of policy goals. We know this and yet our frameworks of analysis poorly reflect these overlapping roles and responsibilities. This proposition is presented in an attempt to overcome conceptual barriers that direct our analyses to emphasize distinct parts of policy processes rather than the connections that form the whole. It is also presented to advise policy communities to pay attention to the

170 dimensions of influence these important actors exert to greater and lesser degrees, of course, and in greater and lesser support of any given set of policy goals. So what? Existing frameworks of analysis recognize that policy processes are embedded in complex social settings and yet we tend to examine policy processes in deceptively discrete segments, stages or phases. This greatly contributes to our understanding of portions of policy processes and yet limits our ability to respond to questions requiring more holistic answers. The big picture is important and existing frameworks of analysis even the more comprehensive fall short in providing us the conceptual tools to examine it. Based on the findings of this study and informed by existing scholarship, Table 7.1 below summarizes categories and variables for consideration toward building a more holistic understanding of policy processes. This framework has important implications for theory and practice. For theory, it identifies an important and neglected set of variables in historical social organization. It refines our understanding of how the political environment matters. And, it presents political-bureaucratic actors and ideas as a category of factors that bridges policy and management to better reflect the overall set of causal relationships that influence policy outcomes. The findings of this study and the framework it presents also have implications for transnational and domestic policy actors as they seek to understand important factors in policy processes, identify effective strategies to further their ideas, and improve public performance for those who lack voice in the process.

171 Table 7.1 Proposed holistic framework for analyzing policy processes Category Organizational structures Factors Capacity of lead agencies Financial & human resources Information & capital management Institutional norms Political Professional Organizational Politicalbureaucratic actors & ideas Elected officials Policy communities Bureaucratic leadership Upward, downward and outward Policy ideas Political environment Governance structures Strength, priorities and worldview of governing coalitions & competitors Basic constitutional structures Level of government Degree to which power is centralized Social history Socio-economic conditions Socio-cultural values & structures Social organization

172 APPENDIX A Table A.1 Comparing frameworks for analyzing policy processes Categories of factors Organizational structures Sabatier & Jenkins-Smith / Weible Baumgartner & Jones Smith Capacity of lead agencies Financial & human resources Information & capital management Institutional norms Political Professional Organizational Societal Politicalbureaucratic actors & ideas Elected officials Policy communities Bureaucratic leadership Upward, downward and outward Policy ideas Policy subsystem Advocacy coalitions Policy beliefs Resources Authority Finances Leadership Information Continued on next page Policy communities Policy images

173 APPENDIX A CONTINUED Categories of factors Political environment Sabatier & Baumgartner & Jenkins-Smith / Jones Weible Change in governing Change in governing coalition coalition Smith Strength, priorities and worldview of governing coalition & competitors Basic constitutional structures Level of government Degree of centralization Socio-economic conditions Degree to which power is centralized Socio-economic conditions Socio-cultural values & social structures Social organization Source: Baumgartner & Jones (1993); Sabatier & Jenkins-Smith (1999); Sabatier & Weible (2007)

Governance structures

Basic constitutional structures Degree of consensus needed Openness of political system

Basic constitutional structures Level of government

Social history

Socio-economic conditions

Socio-cultural values Socio-cultural values & social structures & social structures

174 REFERENCES AbouZahr, C. (2001). "Cautious Champions: International Agency Efforts to Get Safe Motherhood Onto the Agenda." Studies in HSO&P 17: 384-411. AbouZahr, C. (2003). "Safe motherhood: A brief history of the global movement 19472002." British Medical Bulletin 67: 13-25. Academy for Nursing Studies (2005). Situational analysis of public health nursing personnel in India. Hyderabad, Academy for Nursing Studies. Appleby, P. H. (1949). Policy and administration. University, Alabama, University of Alabama Press. Athreya, V. and S. R. Chunkath (2000). 'Gendering' health policy. March 5, 2000. The Hindu. Chennai. Basu, A. M. (1990). "Cultural influences on health care use: Two regional groups in India." Studies in Family Planning 21(5): 275-86. Baumgartner, F. R. and B. D. Jones (1993). Agendas and instability in American politics. Chicago, The University of Chicago Press. Berman, S. (2001). "Review: Ideas, norms, and culture in political analysis." Comparative Politics 33(2): 231-50. Campbell, O., W. Graham, et al. (2006). "Strategies for reducing maternal mortality: Getting on with what works." Lancet 368: 1284-99. Campbell, O. M. (2001). "What are maternal health policies in developing countries and who drives them? A review of the last half-century." Safe motherhood strategies: A review of the evidence. V. DeBrouwere and W. Van Lerberghe. Antwerp, Belgium, ITG Press. Centre for Budget and Policy Studies (2004). Maternal health in Karnataka - As seen from budget data. Bangalore. Charan, S. (2008). Rural health mission yet to take off in state. January 1, 2008. The Hindu. Bangalore. Cohen, M. D., J. G. March, et al. (1972). "A garbage can model of organizational choice." Administrative Science Quarterly 17(1): 1-25. Commission on Macroeconomics and Health (2001). Macroeconomics and health: Investing in health for economic development. Report of the Commission on Macroeconomics and Health. Geneva, World Health Organization.

175 Costello, A., K. Azad, et al. (2006). "An alternative strategy to reduce maternal mortality." Lancet 368: 1477-79. Crawford, S. S. E. and E. Ostrom (2005). A grammar of institutions. Understanding institutional diversity. E. Ostrom. Princeton, Princeton University Press: 582-600. Crook, R. C. and J. Manor (1998). Democracy and Decentralisation in South Asia and West Africa: Participation, Accountability and Performance. Cambridge, UK, Cambridge University Press. Deccan Herald (2007). Rural health: PM unhappy with states performance. June 6, 2007. Deccan Herald. Bangalore. Retrieved May 3, 2008, http://www.deccanherald.com/Content/Jun62007/national200706055786.asp. Dreze, J. (2004). Health checkup. March 12, 2004. The Hindu. Chennai. Retrieved April 3, 2008, http://www.hinduonnet.com/2004/03/12/stories/2004031201851000.htm. Dreze, J. and A. Sen (1997). Indian Development: Selected Regional Perspectives. Bombay, Oxford University Press. Duran, A. M. (1992). "The safety of home birth: the farm study." American Journal of Public Health 82(3): 450-3. Dyson, T. and M. Moore (1983). "On kinship structure, female autonomy, and demographic behavior in India." Population and Development Review 9(1): 35-60. Elmore, R. F. (1979). "Backward mapping: Implementation research and policy decisions." Political Science Quarterly 94(4): 601-16. Finnemore, M. and K. Sikkink (2001). "Taking stock: The constructivist research program in international relations and comparative politics." Annual Review of Political Science 4: 391-416. Frankel, F. R. and M. S. A. Rao, Eds. (1989). Dominance and state poewr in modern India: Decline of a social order. New York, Oxford University Press. Freedman, L. P. (2005). "Achieving the MDGs: Health systems as core social institutions." Development 48(1): 19-24. Freedman, L. P., W. Graham, et al. (2007). "Practical lessons from global safe motherhood initiatives: Time for a new focus on implementation." The Lancet 370: 138391. Geetha, V. and S. V. Rajadurai (1998). Towards a Non-Brahmin Millennium: From Iyothee Thass to Periyar. Calcutta, Samya.

176 George, A. (2007). "Persistence of high maternal mortality in Koppal District, Karnataka, India: Observed service delivery constraints." Reproductive Health Matters 15(30): 91102. George, A., A. Iyer, et al. (2005). Gendered health systems biased against maternal survival: Preliminary findings from Koppal, Karnataka, India, IDS Working Paper 253, Institute of Development Studies. George, A. L. and A. Bennett (2004). Case Studies and Theory Development in the Social Sciences. Cambridge, MA, MIT Press. Gilson, L. (2003). "Trust and the development of health care as a social institution." Social Science and Medicine 56: 1453-68. Glasier, A., A. M. Gulmezoglu, et al. (2006). "Sexual and Reproductive Health: A Matter of Life and Death." Lancet 368: 1595-607. Government of India (2001). Census of India 2001. Registrar General and Census Commissioner. Government of India (2002). National human development report 2001. New Delhi, Planning Commission. Government of India (2004). India reproductive and child health: District level household survey 2002-04. New Delhi, IIPS and Ministry of Health & Family Welfare, Government of India. Government of India (2004). "Guidelines for operationalising first referral units." Retrieved May 9, 2008, http://www.mohfw.nic.in/dofw%20website/FRU&_nbsp_Guidelines_2004.pdf. Government of India (2005). Report of the national commission on macroeconomics and health. New Delhi, Ministry of Health and Family Welfare. Government of India (2006). Bulletin on rural health statistics in India. Ministry of Health & Family Welfare, Government of India. Government of India (2007). Poverty estimates for 2004-05. Press Information Bureau. Retrieved May 20, 2008, http://planningcommission.nic.in/news/prmar07.pdf. Government of India (2007). Reproductive & child health programme phase II: 3rd joint review mission. Ministry of Health & Family Welfare Donor Coordination Division. Retrieved May 9, 2007, http://health.nic.in/NRHM/Documents/Final_Aide_Memoire_JRM3.pdf.

177 Government of India (undated). National program implementation plan: RCH phase II program document. Ministry of Health & Family Welfare, Government of India. Retrieved, June 10, 2008, http://mohfw.nic.in/NRHM/RCH/guidelines/NPIP_Rev_III.pdf. Government of India, IIPS, et al. (2004). Karnataka reproductive and child health: District level household survey. Ministry of Health & Family Welfare, Government of India. Government of Karnataka (1999). Human Development in Karnataka 1999. Bangalore, Planning Department, Government of Karnataka. Government of Karnataka (2002-2007). "Budget allocations." Retrieved May 9, 2008, from http://www.kar.nic.in/finance/budget-m.htm. Government of Karnataka (2002-2008). "Budget speeches." Retrieved May 9, 2008, 2008, from http://www.kar.nic.in/finance/budget-m.htm. Government of Karnataka (2004). Proceedings of the Government of Karnataka: The Karnataka state integrated health policy-reg. Government of Karnataka Proceedings of the State Cabinet, Health & Family Welfare Department, Bangalore. Retrieved October 12, 2007, http://karhfw.gov.in/documents/PDF/STATE%20HEALTH%20POLICY.pdf. Government of Karnataka (2005). Reproductive and child health project: Programme implementation plan (abridged version) 2005-2010. Department of Health and Family Welfare Services, Government of Karnataka. Government of Karnataka (2006). Karnataka human development report 2005. Bangalore, Planning and Statistics Department, Government of Karnataka. IIPS (2007). National Family Health Survey (NFHS-3), 2005-06: Key indicators, International Institute for Population Sciences. Government of Tamil Nadu (1999). Government Order No. 396. Health & Family Welfare Department. Government of Tamil Nadu (2002). Government Order (Rt.) No. 2143. Health & Family Welfare Department. Government of Tamil Nadu (2002-2008). "Policy notes." Retrieved June 20, 2008, 2008, from http://www.tn.gov.in/policynotes/default.htm. Government of Tamil Nadu (2002-2008). "Budget speeches." Retrieved June 15, 2008, 2008, from http://www.tn.gov.in/budget/budgetspeech.htm. Government of Tamil Nadu (2003). Human Development Report. Delhi, Government of Tamil Nadu in association with Social Science Press.

178 Government of Tamil Nadu (2003). Tamil Nadu human development report. New Delhi, Government of Tamil Nadu in association with Social Science Press. Government of Tamil Nadu (2004). Government Order (Ms.) No. 211. Health & Family Welfare Department. Government of Tamil Nadu (2004). Government Order (Ms.) No. 223. Health & Family Welfare Department. Government of Tamil Nadu (2005). Government Order 2D No. 25. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (Ms.) No. 33. H. a. F. W. Department. Government of Tamil Nadu (2006). Government Order No. 152. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (2D.) No. 21. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (2D.) No. 22. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (D) No. 164. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (2D.) No. 18. Health & Family Welfare Department. Government of Tamil Nadu (2006). Government Order (2D.) No. 19. Health & Family Welfare Department. Gould, H. A. (2003). "Political self-destruction in Karnataka, 1999." India's 1999 elections and 20th century politics. P. Wallace and R. Roy, eds. New Delhi, SAGE Publications: 94-140. Graham, W., W. Brass, et al. (1989). "Estimating maternal mortality: The sisterhood method." Studies in Family Planning 20(3): 125-35. Gwatkin, D. R., A. Bhuiya, et al. (2004). "Making health systems more equitable." The Lancet 364: 1273-80. Haas, P. M. (1992). "Introduction: Epistemic Communities and International Policy Coordination." International Organization 46(1): 1-35.

179 Hanson, K., M. K. Ranson, et al. (2003). "Expanding access to priority health interventions: A framework for understanding the constraints to scaling-up." Journal of International Development 15: 1-14. Harriss, J. (1999). "Comparing political regimes across Indian states: A preliminary essay." Economic and Political Weekly: 3367-77. Hussein, J. and S. Clapham (2005). "Message in a bottle: Sinking in a sea of safe motherhood concepts." Health Policy 73: 294-302. Hutter, I. (1997). Nutrition and reproduction: The socio-cultural context of food behaviour in rural South India. Cultural perspectives on reproductive health. C. M. Obermeyer, ed. Oxford, Oxford University Press. IIPS. (2007). "2005-2006 national family health survey (NFHS-3): national fact sheets." Retrieved May 28, 2007, from http://www.nfhsindia.org/factsheet.html. IIPS and Macro International (2008). National Family Health Survey (NFHS-3), India, 2005-6: Karnataka. Mumbai, International Institute for Population Sciences. IIPS and Macro International (2008). National Family Health Survey (NFHS-3), India, 2005-6: Tamil Nadu. Mumbai, International Institute for Population Sciences. IIPS and ORC Macro (2000). National Family Health Survey (NFHS-2), 1998-99: India. Mumbai, International Institute for Population Sciences. IIPS and ORC Macro (2001a). National family health survey (NFHS-2), India, 1998-99: Karnataka. Mumbai, International Institute for Population Sciences (IIPS). IIPS and ORC Macro (2001b). National family health survey (NFHS-2), India, 1998-99: Tamil Nadu. Mumbai, International Institute for Population Sciences (IIPS). Ingraham, P., P. G. Joyce & A.K. Donahue (2003). Government performance: Why management matters. Baltimore, Johns Hopkins University Press. Jejeebhoy, S. J. and Z. A. Sathar (2001). "Women's autonomy in India and Pakistan: The influence of religion and region." Population and Development Review 27(4): 687-712. Jenkins, R., Ed. (2004). Regional Reflections: Comparing Politics Across India's States. New Delhi, Oxford University Press. Johnson, K. C. and B.-A. Daviss (2005). "Outcomes of planned home births with certified professional midwives: large prospective study in North America." British Medical Journal 330(7505): 1416.

180 Justice, J. (1986). Policies, Plans, and People: Foreign Aid and Health Development. Berkeley, University of California Press. Karnataka Health Task Force (2001). Karnataka health task force report. Bangalore, Government of Karnataka. Keck, M. E. and K. Sikkink (1998). Activists Beyond Borders: Advocacy Networks in International Politics. Ithaca, Cornell University Press. Khan, K., D. Wojdyla, et al. (2006). "WHO analysis of causes of maternal death: A systematic review." The Lancet 367: 1066-74. Kingdon, J. W. (1984, 1995). Agendas, Alternatives, and Public Policies. Menlo Park, CA, Longman. Kohli, A., Ed. (1988). India's Democracy: An Analysis of Changing State-Society Relations. Princeton, Princeton University Press. Kohli, A. (1989). The state and poverty in India: The politics of reform. New York, Cambridge University Press. Kohli, A. (1990). Democracy and Discontent: India's Growing Crisis of Governability. Cambridge, Cambridge University Press. Kohli, A., Ed. (2001). The Success of India's Democracy. Contemporary South Asia. Cambridge, Cambridge University Press. Koontz, T. M. (2002). Federalism in the forest: National versus state natural resource policy. Washington, D.C., Georgetown University Press. Lester, J. P. and M. L. Goggin (1998). "Back to the future: The rediscovery of implementation studies." Policy Currents 8(3): 1-9. Lipsky, M. (1980). Street-level bureaucracy: Dilemmas of the individual in public services. New York, Russell Sage Foundation. Maine, D. and A. Rosenfield (1999). "The safe motherhood initiative: Why has it stalled?" American Journal of Public Health 89: 480-502. Malhotra, A., R. Vanneman, et al. (1995). "Fertility, dimensions of patriarchy, and development in India." Population and Development Review 21(2): 281-305. Manor, J. (1977). Political Change in Mysore 1917-1955. New Delhi, Manohar.

181 Manor, J. (1989). "Karnataka: Caste, class, dominance and politics in a cohesive society." Dominance and state power in modern India: Decline of a social order. F. R. Frankel and M. S. A. Rao, eds. Delhi, Oxford University Press. 1: 322-361. Martin, J. (2002). Organizational culture: Mapping the terrain. Thousand Oaks, Sage Publications. Matthews, Z., J. Ramakrishna, et al. (2005). "Birth rights and rituals in rural South India: Care seeking in the intrapartum period." Journal of Biosocial Science 37: 385-411. Maynard-Moody, S. (2000). "State agent or citizen agent: Two narratives of discretion." Journal of Public Administration Research and Theory 10(2): 329-58. Mehrotra, S. (2006). "Well-being and caste in Uttar Pradesh: Why UP is not like Tamil Nadu." Economic and Political Weekly: 4261-71. Meier, K. J. (1979). Politics and the bureaucracy: Policymaking in the fourth branch of government. North Scituate, MA, Duxbury Press. Meier, K.J., L.J. OToole & S. Nicholson-Crotty (2004). Multilevel governance and organizational performance: Investigating the political-bureaucratic labyrinth. Journal of Policy Analysis and Management 23(1): 31-47. Miles, M. B. and A. M. Huberman (1994). Qualitative Data Analysis: An Expanded Sourcebook. Thousand Oaks, Sage Publications. Miller, S., N. L. Sloan, et al. (2003). "Where is the "E" in MCH? The need for an evidence-based approach in safe motherhood." Journal of Midwifety & Women's Health 48(1): 10-18. Moore, M.H. (1995). Creating public value: strategic management in government. Cambridge, MA, Harvard University Press. Murray, C. J. L. and J. Frenk (2000). "A framework for assessing the performance of health systems." Bulletin of the World Health Organization 78(6): 717-31. Nakamura, R. T. and F. Smallwood (1980). The Politics of Policy Implementation. New York, St. Martin's. Narayanan, G. (2003). Best practices: Danida-assisted Tamil Nadu area health care project phase 3. Chennai, Danida Tamil Nadu Area Health Care Project. Navaneetham, K. and A. Dharmalingam (2002). "Utilization of maternal health care services in Southern India." Social Science and Medicine 55: 1849-69.

182 Nichter, M. (1989). Anthropology and international health: South Asian case studies. Boston, Kluwer Academic. O'Connor, B. B. (1993). "The home birth movement in the United States." Journal of Medical Philosophy 18(2): 147-74. Oliveira-Cruz, V., K. Hanson, et al. (2003). "Approaches to overcoming constraints to effective health service delivery: A review of the evidence." Journal of International Development 15: 41-65. Olsen, O. (1997). "Meta-analysis of the safety of home birth." Birth 24(1): 4-13. Ostrom, E. (2007). Institutional rational choice: an assessment of the institutional analysis and development framework. Theories of the policy process. P. A. Sabatier. Boulder, Westview Press: 21-64. OToole, L., K.J. Meier & S. Nicholson-Crotty (2005). Managing upward, downward and outward: Networks, hierarchical relationships and performance. Public Management Review 7(1): 45-68. Ott, J. S. (1989). The Organizational Culture Perspective. Pacific Grove, CA, Brooks/Cole Publishing Company. Padmanaban, P. and B. R. Desikachari (2004). Averting maternal deaths and disabilities: Rights based approach towards reduction of maternal mortality ratio (MMR) in Tamil Nadu. Pathmanathan, I., J. Liljestrand, et al. (2003). Investing in maternal health: learning from Malaysia and Sri Lanka. Health, nutrition and population series. Washington, DC, The World Bank. Paxton, A., D. Maine, et al. (2005). "The evidence for emergency obstetric care." International Journal of Gynecology and Obstetrics 88: 181-93. Population Resource Centre (1994). National family health survey (MCH and family planning): Tamil Nadu 1992. Bombay, Population Research Centre, The Gandhigram Institute, and International Institute for Population Sciences (IIPS). Population Resource Centre (1995). National family health survey (MCH and family planning), Karnataka 1992-3. Bombay, Institute for Social and Economic Change and International Institute for Population Sciences (IIPS). Prabhu, N. (2007). State asked to meet health spending commitment (increase 10%). September 9, 2007. The Hindu. Bangalore. Retrieved May 2, 2008, http://www.hinduonnet.com/2007/09/28/stories/2007092862800500.htm.

183 Prasad, S. (2007). Florence Nightingale award for Dharmapuri nurse. May 9, 2007. The Hindu. Chennai. Retrieved May 12, 2007, http://www.hindu.com/2007/05/09/stories/2007050903930400.htm. Redford, E. S. (1969). Democracy in the administrative state. New York, Oxford University Press. Reich, M. R., K. Takemi, et al. (2008). "Global action on health systems: A proposal for the Toyako G8 summit." The Lancet 371: 865-69. Registrar General of India (2006). Sample Registration System, Maternal Mortality in India: 1997-2003 Trends, Causes and Risk Factors. New Delhi, Registrar General of India. Ronsmans, C. and W. Graham (2006). "Maternal mortality: Who, when, where, and why." The Lancet 368: 1189-200. Ronsmans, C., A. M. Vanneste, et al. (1997). "Decline in maternal mortality in Mablab, Bangladesh: a cautionary tale." The Lancet 350: 1810-14. Rourke, F. E. (1984). Bureaucracy, politics and public policy. Boston, Little, Brown and Company. Roy, R. and P. Wallace, Eds. (2007). India's 2004 elections: Grass-roots and national perspectives. New Delhi, SAGE Publications. Sabatier, P. A., Ed. (1999). Theories of the Policy Process. Boulder, CO, Westview Press. Sabatier, P. A. (1998). "The advocacy coalition framework: revisions and relevance for Europe." Journal of European Public Policy 5(1): 98-130. Sabatier, P. A. and H. C. Jenkins-Smith (1999). The advocacy coalition framework: An assessment. Theories of the policy process. P. A. Sabatier. Boulder, Westview Press: 117166. Sabatier, P. A. and C. M. Weible (2007). The advocacy coalition framework: innovations and clarifications. Theories of the policy process. P. A. Sabatier. Boulder, Westview Press: 189-220. Schein, E. H. (1992). Organizational culture and leadership. San Francisco, Jossey-Bass Publishers. Shiffman, J. (2007). "Generating political priority for maternal mortality reduction in 5 developing countries." American Journal of Public Health 97: 796-803.

184 Shiffman, J. and A. L. Garces del Valle (2006). "Political history and disparities in safe motherhood between Guatemala and Honduras." Population and Development Review 32(1): 53-80. Shiffman, J. and S. Smith (2007). "Generation of political priority for global health initiatives: A framework and case study of maternal mortality." Lancet 370: 1370-79. Shiffman, J. and R. Ved (2007). "The state of political priority for safe motherhood in India." British Journal of Obstetrics and Gynaecology 114: 785-90. Sinha, A. (2005). The Regional Roots of Developmental Politics in India: A Divided Leviathan. Bloomington, IN, Indiana University Press. Smircich, L. (1983). "Concepts of culture and organizational analysis." Administrative Science Quarterly 28(3): 339-58. Snow, D. A., J. E. Burke Rochford, et al. (1986). "Frame Alignment Processes, Micromobilization, and Movement Participation." American Sociological Review 51(4): 464-481. Stanton, C., N. Abderrahim, et al. (2000). "An assessment of DHS maternal mortality indicators." Studies in Family Planning 31(2): 111-123. Starrs, A. M. (2006). "Safe motherhood initiative: 20 years and counting." Lancet 368: 1130-32. Starrs, A. M. (2007). "Delivering for women." The Lancet 370: 1285-7. Stoker, R. P. (1991). Reluctant partners: Implementing federal policy. Pittsburgh, University of Pittsburgh Press. Stone, D. (2002). Policy paradox: The art of political decision making. New York, W.W. Norton & Company. Subramanian, N. (1999). Ethnicity and Populist Mobilization: Political Parties, Citizens and Democracy in South India. New Delhi, Oxford University Press. Subramanian, N. (2003). Beyond ethnicity and populism? Changes and continuities in Tamil Nadu's electoral map. India's 1999 elections and 20th century politics. P. Wallace and R. Roy, eds. New Delhi, SAGE Publications: 50-93. Thaddeus, S. and D. Maine (1994). "Too far to walk: Maternal mortality in context." Social Science and Medicine 38(8): 1091-110.

185 The Hindu (2006). Minister initiates process of filling VHN posts. October 24, 2006. The Hindu. Chennai. Retrieved May 3, 2008, http://www.hindu.com/2006/10/24/stories/2006102417060400.htm. The Hindu (2006). Government must address issue of regional imbalance: expert. September 25, 2006. The Hindu. Gulbarga. Retrieved May 3, 2008, http://www.hinduonnet.com/2006/09/25/stories/2006092504820300.htm. The Hindu (2007). 6,017 vacancies in hospitals to be filled. April 16, 2007. The Hindu. Madurai. Retrieved May 8, 2008. http://www.hindu.com/2007/04/16/stories/2007041602230300.htm. The Hindu (2008). Rural health mission: Tamil Nadu the best performer, says Anbumani. March 2, 2008. The Hindu. Chennai. Retrieved May 8, 2008, http://www.hinduonnet.com/2008/03/02/stories/2008030251720400.htm. Travis, P., S. Bennett, et al. (2004). "Overcoming health-systems constraints to achieve the Millennium Development Goals." The Lancet 364: 900-06. United Nations Millennium Project Task Force on Maternal and Child Health (2005). Who's got the power? Transforming health systems for women and children. London. Retrieved April 2, 2008, http://www.unmillenniumproject.org/reports/tf_health.htm. Van Hollen, C. (2003). Birth on the Threshold: Childbirth and Modernity in South India. Berkeley, University of California Press. Vinzant, J. C. and L. Crothers (1998). Street-level leadership: Discretion and legitimacy in front-line public service. Washington, D.C., Georgetown University Press. Visaria, L., S. J. Jejeebhoy, et al. (1999). "From family planning to reproductive health: Challenges facing India." International Family Planning Perspectives 25(Supplement): S44-49. Visaria, L. and P. Visaria (1998). Reproductive health in policy and practice: India. Washington, DC, Population Reference Bureau. Wallace, P. (2003). "Introduction: The new national party system and state politics." India's 1999 elections and 20th century politics. P. Wallace and R. Roy, eds. New Delhi, SAGE Publications: 1-23. Walt, G. (1994). Health policy: An introduction to process and power. London, Zed Books. WHO (2000). The world health report 2000: Health systems: Improving performance. Geneva, The World Health Organization.

186 WHO (2004). Maternal Mortality in 2000: Estimates Developed by WHO, UNICEF and UNFPA. Geneva, Department of Reproductive Health and Research, World Health Organization. WHO (2007). Maternal mortality in 2005: Estimates developed by WHO, UNICEF, UNFPA, and The World Bank. Geneva, The World Health Organization. Wilson, J. Q. (1989). Bureaucracy: What government agencies do and why they do it, Basic Books. World Bank (2006). Reforming Public Services in India: Drawing Lessons from Success, Poverty Reduction and Economic Management Sector Unit South Asia, World Bank. Yin, R. K. (2003). Case Study Research: Design and Methods. Thousand Oaks, Sage Publications.

You might also like