You are on page 1of 7

of Hospital Staffing and Organization of Patient in their current hospital for 12 years (SD = 7.

6),
Outcomes24 conducted in 5 countries (Canada, primarily on medical/surgical units (64%).
USA, England, Scotland, and Germany). The study
Procedures
was designed to explore relationships between
hospital work environment characteristics, nurse Nurses received questionnaires through regular
staffing, and nurse and patient outcomes. In mail in the fall of 1998. Participation was anony-
mous with instructions to clarify informed consent.
Canada, nurses working in 292 acute care hospi- tals
in 3 provinces were surveyed. In Ontario and British The Dillman25 technique was used to maximize re-
Columbia, stratified random samples of nurses turn rates.
were randomly selected from the registry lists of the
provincial licensing bodies. In Alberta, the entire Instruments
population of acute care nurses was surveyed. A total
Practice Environment Scale of the Nursing Work
of 17,965 nurses returned useable questionnaires
Index
(response rate, 59%).
The results reported in this article relate to a In this analysis, we used items on the survey
subset of the Ontario and Alberta data (n = 4,606 questionnaire included in Lake’s5 modification of
and n = 3,991, respectively) who provided valid the NWI-R, the Practice Environment Scale of
responses on all variables in the analysis (N = the Nursing Work Index (NWI-PES). Items captur-
8,597). Consistent with the demographic profile of ing each of Lake’s subscales reflect 5 aspects of
nurses in Canada, nurses’ average age was 44 years professional nursing worklife environments. Re-
with 19 years of experience in nursing (see Table 1). spondents rated positively worded statements as
Most were female, diploma prepared, and worked Strongly Disagree (1), Disagree (2), Agree (3), and
full time. The majority held permanent positions Strongly Agree (4). The Canadian survey did not
(85%), whereas others had temporary positions or include 3 items included in Lake’s5 analysis of
casual positions. Of those in casual positions, most USA data (career ladder in place, use of nursing
preferred this position (61%). Nurses had worked diagnosis, and supervisors use mistakes as learning
opportunities). The nurse participation in hospital
affairs subscale (Participation) consisted of 9 items;
the nursing foundations for quality of care subscale
(Nursing Model), 8 items; nurse manager ability/
Table 1. Demographics support of nurses subscale (Leadership), 4 items;
the staff and resource adequacy subscale (Staffing),
x̄ SD 4 items; and the collegial nurse/physician relation-
Average age 44 9.3 ships subscale (Nurse/Physician Relationship), 3
Years experience 19 9.2 items. Lake5 established evidence for the construct
Years worked in current hospital 12 7.6 validity and internal consistency reliability for the
% NWI-PES.
Sex Maslach Burnout InventoryVHuman Service Scale
Female 98
Male 2 The Maslach Burnout InventoryVHuman Service
Highest educational credentials
Diploma 48 Scale (MBI-HSS) is the original version of this
Baccalaureate 28 measure, which is the most widely used measure
Masters 2 of job burnout.26 The 22-item measure comprises
Employment status
Full-time 59 3 subscales: emotional exhaustion (9 items), deper-
Part-time 40 sonalization (5 items), and personal accomplish-
Employment type
Permanent 85 ment (8 items). The items are framed as statements
Temporary 3 of job-related feelings (eg, ‘‘I feel burned out from
Casual 13
Primary specialty areas my work,’’ ‘‘I feel confident that I am effective at
Medical/surgical units 64 getting things done’’), and are rated on a 7-point
Intensive care unit 12
Obstetrics 10 frequency scale (ranging from ‘‘never’’ to ‘‘daily’’).
Operating/recovery room 6 Burnout is reflected in higher scores on emotional
Pediatrics 4 exhaustion and depersonalization and lower scores
Psychiatry 4
on personal accomplishment. A factor analysis of
the data in this study for the MBI-HSS items rep-
licated the established MBI-HSS factor structure.

262 JONA Vol. 36, No. 5 May 2006


A considerable body of research has confirmed the
validity and reliability of this measure.27,28

Depersonalization

.34
Adverse Events
Adverse events24 were measured by nurses’ reports
of the frequency of occurrence of 4 types of
negative patient incidents on their shifts over the

Accomplishment
past year: falls, nosocomial infections, medication
errors, and patient complaints. Nurses were asked

Personal

j.35
j.22
‘‘Over the past year, how often would you say each
of the following incidents has occurred involving
you or your patients.’’ Response options ranged
from 1 (never) to 4 (frequently).

Exhaustion
Emotional

.71
.30
j.28
Results
Table 2 displays the means, SDs, Cronbach !
reliability estimates, and correlations for the variables

Nursing
Model

j.39

j.27
j.25
.25
in the study. The scores on the MBI subscales are
close to the usual level for health service profes-
sionals.26 Emotional exhaustion and depersonali-

Involvement
zation are highly correlated (r = 0.71), and both are

Policy

.82
j.39

.22

j.28
j.23
moderately correlated with personal accomplish- Means, SDs, Cronbach "’s, and Correlations for Major Study Variables
ment (r = j0.28 and r = j0.35, respectively). The
strongest correlations with adverse events are with
staffing (r = j0.30), emotional exhaustion (r =
Leadership

0.30), and depersonalization (r = 0.34). All ! levels

.89
.73

.21
j.41

j.29
j.23
are in the acceptable range above .70. Regarding
the patient safety items, the most frequent were
patient complaints (M = 2.36, SD = 0.91) followed
Physician/

by nosocomial infections (M = 2.06, SD = 0.87),


Nurse

.48
.47
.51

.13
j.22

j.16
j.14
patient falls (M = 1.96, SD = 0.89), and medication
errors (M = 1.89, SD = 0.76).
Staffing

.37
.67
.64
.63
j.61

.24

j.43
j.30
Data Analysis
Through structural equation modeling, the analysis
assessed Lake’s5 factor structure for the NWI-PES,
Cronbach

the factor structure for the MBI-HSS by Maslach


.78
.83
.84
.79
.72
.91

.80

.78
.75
!

et al,26 and the measure of adverse events. The


structural equation modeling analysis also exam-
N = 8,560. All correlations significant at P G .01.

ined the fit between the hypothesized model and


11.20
0.69
0.65
0.78
0.54
0.49

7.14

5.67
0.65
SD

the data and the magnitude of the direct and


indirect effects within the model (Figure 1).
2.32*
2.82*
2.46*
2.38*
2.71*

2.06*
Mean

22.34y

y
y

6.30
37.38

Model Testing
The hypothesized model was tested with EQua-
tionS,29 a structural equation modeling statistical
Policy involvement

Sums of item ratings.


accomplishment
Depersonalization
Nurse/Physician

package. The first phase of the analysis examined


Nursing model

Adverse events
exhaustion

the measurement models of the NWI-PES, the


*Item means.
Leadership

Emotional
Table 2.

MBI-HSS, and the patient safety items. Based on


Personal
Staffing

Hoyle and Panter’s30 recommendations, several


criteria were used to evaluate fit of the models.
y

These included omnibus fit indexes such as the

JONA Vol. 36, No. 5 May 2006 263


chi-square (# 2),31 incremental fit indexes, such as errors between pairs of items within the NWI-PES
the Comparative Fit Index (CFI)32 and the Incre- factors. A confirmatory factor analysis also con-
mental Fit Index (IFI),33 and the Root Mean firmed a single factor structure of the 4 patient
Square Error of Approximation (RMSEA) advo- safety items with no correlated errors.
cated by Browne and Cudeck.34 Next, the structural relationships among the
The # 2 test is interpreted as the test of the latent variables in the model were examined. A
difference between the hypothesized model and the structural equation modeling analysis using maxi-
just identified version of the model. Low, non- mum likelihood estimation identified a good fit
significant values are desired.35 However, the # 2 of the data to the hypothesized model (# 2 =
test is very sensitive to sample size; thus, in a model 16,557.35, df = 1,346, CFI = .907, IFI = .907,
with a relatively large sample size, the null hy- RMSEA = .037). This model met the criterion for
pothesis will almost always be rejected. Because of incremental fit indexes (CFI/IFI greater than .90).
this limitation, the # 2 test was used only to evaluate All structural coefficients were statistically signifi-
the relative differences in fit among competing cant. The relationships among worklife factors and
models. Incremental fit indexes indicate the pro- burnout were consistent with those of our previous
portion of improvement of the hypothesized model research, and the posited relationships to adverse
relative to a null model, typically one assuming no events were supported by these data. However, the
correlation among observed variables. The gener- modification indexes indicated that adding 2 direct
ally agreed upon critical value for the CFI and IFI is paths to adverse events would further enhance the
.90 or higher.32,33 The RMSEA is the standardized fit of the model. When paths from staffing
summary of the average covariance residuals and is adequacy and from nursing model to adverse
thus a measure of the lack of fit between the data events were added, the # 2 improved significantly
and the model. Low values (between 0 and .06) (# 2Diff = 119.19, df = 2, P = .001), producing a
indicate a good fitting model.36 good overall fit (# 2 = 16,438.19, df = 1,344, CFI =
The confirmatory factor analysis supported the .908, IFI = .908, RMSEA = .037). In this signifi-
measurement models for Lake’s 5-factor solution cantly enhanced model fit, all coefficients, except
for the NWI-PES items and the 3-factor solution the path from exhaustion to adverse events, were
for the MBI-HSS by Maslach et al. The analysis significant (see Figure 2). This suggests that burn-
identified 10 correlated errors between pairs of out only partially mediated the relationship be-
items within the MBI-HSS factors and 7 correlated tween worklife factors and adverse events.

Figure 2. Final model. Note: Numbers in circles are error terms for the endogenous latent variables. Numbers by the
arrows are path coefficients.

264 JONA Vol. 36, No. 5 May 2006


events made a substantial improvement in # 2 .
Limitations
Both of the added path coefficients were more
We acknowledge that the findings of this study substantial than the paths from burnout to adverse
must be viewed with caution given the cross- events, with the path from exhaustion losing
sectional nature of the design, which precludes statistical significance in the context of the added
strong statements on causality. Longitudinal ana- paths.
lyses would allow us to examine the dynamic na- This pattern suggests that nurses’ psychologi-
ture of work by measuring changes in perceptions cal relationship with work is related to adverse
of working conditions over time and the impact of events in the context of their direct relationships
these conditions on nurse and patient outcomes. with workplace qualities. Both resource issues
Replication of the study in other samples of staff (adequate staffing) and values issues (use of a
nurses is needed to validate the current findings. nursing model of care) are directly relevant to the
incidence of adverse events. These same qualities
are directly related to nurses’ experience along the
Discussion continuum of burnout to engagement with work.
The results are consistent with the notion that The link between adequate staffing and
patient safety outcomes are associated with the adverse events corroborates the findings by Aiken
quality of the nursing practice work environment et al19 that linked nurse/patient staffing ratios to
and that the burnout/engagement process plays an inpatient mortality and other studies linking nurse
important mediating role. The results suggest that staffing to adverse events.15,16 In our model,
when nurses perceive that their work environment staffing adequacy was a consequence of effective
supports professional practice, they are more likely nursing leadership in the unit, which resulted in
to be engaged in their work, thereby ensuring safe collaborative relationships with physicians and
patient care. The results also support the key role greater involvement of nurses in unit governance.
of strong nursing leadership in creating conditions Both of these conditions, in turn, were associated
for work engagement and, ultimately, safe, high- with emphasis on a nursing model of care (vs
quality patient care. medical), which subsequently had both direct and
The results extend those of our previous re- indirect effects on patient safety outcomes in our
search that found support for a structural model model.
linking Lake’s5 professional practice work envi- When the hospital supported a nursing model
ronment characteristics5 to nurse burnout.4 That of care, nurses felt a greater sense of personal
model defined a fundamental role for nursing lead- accomplishment in their work, which in turn
ership in relation to the quality of worklife through translated into more positive nurse-sensitive
links with staff nurse policy involvement, staffing patient outcomes. These findings support Aiken
levels, support for a nursing model of care, and and Lake’s contentions that professional work
nurse/physician relationships. environments affect patient outcomes, as well as
Our current analysis took the conceptual model Leiter’s argument for the mediating role of burnout
a step further by examining adverse events with in this process. The results provide further support
implications for patient safety. The analysis also for Leiter and Laschinger’s4 model of nursing
provided ample support for a model in which the worklife and extend it to include patient safety
3 components of burnout mediated the relationship outcomes.
of workplace factors with adverse events. The The severe downsizing of the nursing work
hypothesized Nursing Worklife Model provided an force because of hospital restructuring in the 1990s
adequate fit with the data, consistent with the has had a major impact on nursing work environ-
notion that workplace qualities affect adverse events ments. Although nurses have responded positively
to the extent that they influence nurses’ exhaustion, to the challenges created by these conditions,
depersonalization, and personal accomplishment. their coping resources are being severely strained.
The analysis suggested that burnout’s media- Burnout results from accumulated exposure to
tion function was less than complete. In fact, stressful working conditions. Research is beginning
modification indexes suggested that both of the to document high levels of nurse burnout levels
workplace qualities with direct paths to burnoutV after a decade of restructuring.
staffing adequacy and use of a nursing model of In 2 recent Canadian studies carried out con-
careVwould further enhance the prediction of ad- currently,37,38 nurses reported severe levels of burn-
verse events. The revised model with direct paths out according to Maslach and Leiter’s norms. In
from each of these workplace qualities to adverse the study of new graduate nurses in Ontario by

JONA Vol. 36, No. 5 May 2006 265


Kim et al,37 64% of the sample reported severe visibility to staff and availability for mentoring
levels of burnout. This is particularly distressing and support.41-43 Our results suggest that this
given the current severe nursing shortage and the situation must change to prevent nurse burnout
drop in enrollment in nursing education programs. and reduce the likelihood of adverse patient events.
In the study by Greco et al,38 58% of a sample of In conclusion, the results of this study suggest
nurses of all ages who worked in acute care settings that characteristics of professional nursing work
across Ontario also reported severe levels of burn- environments described in the Magnet hospital
out. In both studies, burnout level was strongly research play an important role in the quality of
related to the degree of fit between personal ex- nurses’ worklife and patient safety outcomes.
pectations and existing worklife conditions de- Burnout seems to be a key mediating process
scribed by Leiter and Maslach.39 through which work environments affect patient
Given the manifestations of advanced stages of outcomes. The results suggest that nurse adminis-
the burnout process, it is reasonable to expect that trators must develop strategies to create work
nurses experiencing burnout would be challenged environments that allow nurses to practice accord-
to provide high quality of care. Our findings lend ing to professional standards, thereby increasing
support to this hypothesis by linking characteristics work satisfaction, preventing burnout, and assur-
of nursing professional practice environments to ing that patients are provided with safe effective
adverse patient outcomes through the mediating high-quality care.
mechanisms of burnout.
Finally, the key role played by nursing leader-
ship in this research highlights the importance of Acknowledgments
developing effective staff nurse leaders to ensure This survey was part of an international project to
that nurses feel confident and satisfied with their assess the Outcomes of Hospital Staffing, funded
work and that patients receive the quality of care by the National Institutes of Health (NRO4513),
they deserve. Nursing leadership plays a key role in in the United States, with Dr Linda Aiken as
providing the direction and infrastructure to ensure Principal Investigator. The project includes 3 prov-
that nurses are empowered to practice profession- inces of Canada (Ontario, British Columbia, and
ally, and thus, deliver high-quality care.40 Reduc- Alberta), as well as England, Scotland, Germany,
tions in management staff because of restructuring and the United States. The data used for the cur-
initiatives over the past decade, however, have rent analysis were collected in Ontario (Dr Judith
hindered nurse leaders’ ability to lead. Significantly Shamian, Principal Investigator) and Alberta (Dr
expanded spans of control have reduced their Phyllis Giovanetti, Principal Investigator).

References
1. Institute of Medicine. Keeping Patients Safe: Transforming 8. Blegen MA, Goode CJ, Reed L. Nurse staffing and patient
the Work Environment of Nurses. Washington, DC: Insti- outcomes. Nurs Res. 1998;47(1):43-50.
tute of Medicine; 2004. 9. Kovner C, Gergen P. Nurse staffing levels and adverse
2. Aiken LH, Clarke SP, Cheung RB, Sloane DM, Silber JH. events following surgery in U.S. hospitals. Image. 1998;30:
Educational levels of hospital nurses and surgical patient 315-321.
mortality. JAMA. 2003;290(12):1617-1623. 10. Needleman J, Buerhaus PI, Mattke S, Stewart M, Zelevinsky
3. Tourangeau AE, Giovanetti P, Tu JV, Wood M. Nursing K. Nurse-staffing levels and the quality of care in hospitals.
related determinants of 30-day mortality for hospitalized N Engl J Med. 2002;346(22):1715-1722.
patients. Can J Nurs Res. 2002;33(4):71-88. 11. McGillis Hall L, Doran D, Baker G, et al. Nurse staffing
4. Leiter MP, Laschinger HS. Demands and values: implica- models as predictors of patient outcomes. Med Care.
tions for nurses’ occupational health. Presented at: the First 2003;41(9):1069-1109.
Canadian Conference for Research on Mental Health in the 12. Lang TA, Hodge M, Olson V, Romano PS, Kravitz RL.
Workplace; 2005; Montreal, QB. Nurse-patient ratios. J Nurs Adm. 2004;34(7/8):326-337.
5. Lake ET. Development of the Practice Environment Scale of the 13. Aiken LH, Smith HL, Lake ET. Lower medicare mortality
Nursing Work Index. ResNursHealth. 2002;25(3):176-188. among a set of hospitals known for good nursing care. Med
6. Leiter M, Laschinger HKS. Relationships of work and Care. 1994;32(8):771-787.
practice environment to professional burnout: testing a 14. Kazanjian A, Green C, Wong J, Reid R. Effect of the
causal model. Nurs Res. 2006;55(2):137-146. hospital nursing environment on patient mortality: a
7. Institute of Medicine. To Err is Human: Building a Safer systematic review. J Health Serv Res Policy. 2005;10(2):
Health System. Washington, DC: Institute of Medicine; 1999. 111-117.

266 JONA Vol. 36, No. 5 May 2006


15. Kramer M, Schmalenberg CE. Magnet hospital nurses 31. Jöreskog KG, Sörbom D. LISREL 7: User’s Reference
describe control over nursing practice. West J Nurs Res. Guide. Mooresville, Ind: Scientific Software Inc; 1989.
2003;25(4):434-452. 32. Bentler PM. Causal modeling via structural equation system.
16. Sovie MD, Jawad AF. Hospital restructuring and its impact In: Nesselroade JR, Cattell RB, eds. Handbook of Multivariate
on outcome s: nursing staff regulations are premature. Experimental Psychology. Perspectives on Individual Differ-
J Nurs Adm. 2001;31:588-600. e nces. 2nd ed. New York, NY: Plenum; 1988:317-335.
17. Whitman GR, Kim Y, Davidson LJ, Wolf GA, Wang SL. 33. Bollen KA. Structural Equations With Latent Variables.
The impact of staffing on patient outcomes across specialty New York: Wiley; 1989.
units. J Nurs Adm. 2002;32:633-639. 34. Browne MW, Cudeck R. Single sample cross-validation
18. Maslach C, Leiter MP. The Truth About Burnout. 3rd ed. indices for covariance structures. Multivariate Behav Res.
San Francisco, Calif: Jossey-Bass; 1997. 1989;24:445-455.
19. Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. 35. Kline RB. Principles and Practice of Structural Equation
Hospital nurse staffing and patient mortality, nurse burn- Modeling. New York: Guildford Press; 1998.
out, and job dissatisfaction. JAMA. 2002;288:1987-1993. 36. Hu L, Bentler PM. Cutoff criteria for fit indexes in

20. C larke H, Laschinger HKS, Giovanetti P, Shamian J, covariance structure analysis: conventional criteria versus
Thomson D, Tourangeau A. Nursing shortages: workplace new alternatives. Struct Equ Modeling. 1999;6(1):1-55.
environments are essential to the solution. Hosp Q. 2001; 37. Kim J, Laschinger HKS, Wong C. Workplace empowerment,

4(4):50-57. work engagement and organizational commitment of new


21. Robinson SE, Roth SL, Keim J, Levenson M, Flentje JR, graduate nurses. Paper presented at: Iota Omicron Chapter/
Basher K. Nurse burnout: work related and demographic School of Nursing 18th Annual Research Conference; 2005;
factors as culprits. Res Nurs Health. 1991;14:223-228. London, Ontario.
22. Sims J. Focus on the future. J Nurs Manag. 1997;4(4):20-21. 38. Greco P, Laschinger HKS, Wong CA. Impact of leader
23. Bakker AB, Killmer CH, Siegrist J, Schaufeli WB. Effort- empowering behaviours on staff nurse empowerment and
reward imbalance and burnout among nurses. J Adv Nurs. work engagement. Paper presented at: Iota Omicron
2000;31(4):884-891. Chapter/School of Nursing 18th Annual Research Confer-
24. Aiken LH, Clarke SP, Sloane DM, et al. Nurses’ reports on ence; 2005; London, Ontario.
hospital care in five countries. Health Aff. 2001;20:43-53. 39. Leiter MP, Maslach C. Areas of worklife: a structured
25. Dillman D. Mail and Telephone Surveys: The Total Design approach to organizational predictors of job burnout. In:
Method. New York: Wiley; 1978. Cooper C, ed. Handbook of Stress Medicine and Health.
26. Maslach C, Jackson SE, Leiter MP. The Maslach Burnout 2nd ed. London: CRC Press; 2004:173-192.
Inventory. 3rd ed. Palo Alto, Calif: Consulting Psychologists 40. Clifford JC. Restructuring: The Impact of Hospital Organi-
Press; 1996. zation on Nursing Leadership. San Francisco, Calif: Jossey-
27. Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Bass Publishing; 1998.
Rev Psychol. 2001;52:397-422. 41. Canadian Nursing Advisory Committee. Our Health, Our
28. Schaufeli WB, Enzmann D. The Burnout Companion Study Future: Creating Quality Workplaces for Canadian Nurses.
and Practice: A Critical Analysis. London, England: Taylor Ottawa, Ontario: Advisory Committee on Health Human

Francis; 1998. Resources; 2002.

29. Bentler PM, Chou CP. Practical issues in structural model- 42. Norrish BR, Rundall TG. Hospital restructuring and the
ing. Sociol Methods Res. 1987;16(1):78-117. work of registered nurses. Milbank Q. 2001;79(1):55-79.

30. Hoyle RH, Panter AT. Writing about structural equation 43. alston SL, Burns LR, Kimberly JR. Does reengineering

model. In: Hoyle RH, ed. Structural Equation Modeling. really work? An examination of the context and outcomes
Concepts, Issues, and Applications. Thousand Oaks, Calif: of hospital reengineering initiatives. Health Serv Res. 2000;
Sage; 1989:158-176. 34(6):1363-1388.

Copyr ight © Lippincott Williams & Wilkins. Unauthor iz ed reproduction of this article is prohibited.
JONA Vol. 36, No. 5 May 2006 267
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

Copyr ight © Lippincott Williams & Wilkins. Unauthor iz ed reproduction of this article is prohibited.

You might also like