Articles

5 year ecacy of a bivalent killed whole-cell oral cholera vaccine in Kolkata, India: a cluster-randomised, double-blind, placebo-controlled trial
Sujit K Bhattacharya, Dipika Sur, Mohammad Ali, Suman Kanungo, Young Ae You, Byomkesh Manna, Binod Sah, Swapan K Niyogi, Jin Kyung Park, Banwarilal Sarkar, Mahesh K Puri, Deok Ryun Kim, Jacqueline L Deen, Jan Holmgren, Rodney Carbis, Mandeep Singh Dhingra, Allan Donner, G Balakrish Nair, Anna Lena Lopez, Thomas F Wierzba, John D Clemens

Summary
Background Ecacy and safety of a two-dose regimen of bivalent killed whole-cell oral cholera vaccine (Shantha Biotechnics, Hyderabad, India) to 3 years is established, but long-term ecacy is not. We aimed to assess protective ecacy up to 5 years in a slum area of Kolkata, India. Methods In our double-blind, cluster-randomised, placebo-controlled trial, we assessed incidence of cholera in nonpregnant individuals older than 1 year residing in 3933 dwellings (clusters) in Kolkata, India. We randomly allocated participants, by dwelling, to receive two oral doses of modied killed bivalent whole-cell cholera vaccine or heat-killed Escherichia coli K12 placebo, 14 days apart. Randomisation was done by use of a computer-generated sequence in blocks of four. The primary endpoint was prevention of episodes of culture-conrmed Vibrio cholerae O1 diarrhoea severe enough for patients to seek treatment in a health-care facility. We identied culture-conrmed cholera cases among participants seeking treatment for diarrhoea at a study clinic or government hospital between 14 days and 1825 days after receipt of the second dose. We assessed vaccine protection in a per-protocol population of participants who had completely ingested two doses of assigned study treatment. Findings 69 of 31 932 recipients of vaccine and 219 of 34 968 recipients of placebo developed cholera during 5 year follow-up (incidence 22 per 1000 in the vaccine group and 63 per 1000 in the placebo group). Cumulative protective ecacy of the vaccine at 5 years was 65% (95% CI 5274; p<00001), and point estimates by year of follow-up suggested no evidence of decline in protective ecacy. Interpretation Sustained protection for 5 years at the level we reported has not been noted previously with other oral cholera vaccines. Established long-term ecacy of this vaccine could assist policy makers formulate rational vaccination strategies to reduce overall cholera burden in endemic settings. Funding Bill & Melinda Gates Foundation.
Published Online October 18, 2013 http://dx.doi.org/10.1016/ S1473-3099(13)70273-1 See Online/Comment http://dx.doi.org/10.1016/ S1473-3099(13)70296-2 National Institute of Cholera and Enteric Diseases, Kolkata, India (S K Bhattacharya MD, D Sur MD, S Kanungo DIH, B Manna PhD, S K Niyogi MD, B Sarkar PhD, G B Nair PhD); Indian Council of Medical Research, New Delhi, India (S K Bhattacharya); International Vaccine Institute, Seoul, South Korea (M Ali PhD, Y A You MS, B Sah MBBS, J K Park PhD, M K Puri MSc, D R Kim MS, R Carbis BSc, A L Lopez MD, T F Wierzba PhD, J D Clemens MD); Menzies School of Health Research, Casuarina, NT, Australia (J L Deen MD); University of Gothenburg, Gothenburg, Sweden (J Holmgren PhD); Shantha Biotechnics, Hyderabad, India (M S Dhingra MD); University of Western Ontario, London, Ontario, Canada (A Donner PhD); University of the Philippines Manila, National Institutes of Health, Manila, Philippines (A L Lopez); UCLA School of Public Health, University of California, Los Angeles, CA, USA (J D Clemens); and icddr,b, Dhaka, Bangladesh (J D Clemens) Correspondence to: Dr Mohammad Ali, International Vaccine Institute, SNU Research Park, San 4-8 Nakseongdaedong, Gwanak-gu, Seoul 151-919, South Korea mali@ivi.int

Introduction
Cholera is a serious global public health problem because clean drinking water and sanitation are not universally available, and appropriate case management is not accessible to many patients. In endemic countries alone, about 14 billion people are at risk of cholera and an estimated 28 million cases and 91 000 deaths occur every year. More than half these cases and deaths occur in cholera-endemic countries in Asia and in Africa.1 Furthermore, recent outbreaks in Cuba, Haiti, and Zimbabwe show the ability of this disease to spread rapidly to new areas and to produce outbreaks with substantial morbidity and mortality. Because of their capacity to spread rapidly, cholera outbreaks can overwhelm existing public health infrastructures and require substantial resources. The economic burden of cholera in African countries alone in 200507 ranged from US$39 million to $156 million per year, dependent on the estimate of average life expectancy used.2 Additional preventive interventions are needed.

To complement improvements in access to water and sanitation and rehydration therapies, much attention has been given to development of a cholera vaccine. After several studies in the 1960s showed that injectable whole-cell cholera vaccines conferred only modest protection of short duration, often with signicant sideeects, researchers focused on oral vaccines that could eciently stimulate local immunity in the gut.3 The rst oral cholera vaccine to be prequalied by WHO for purchase by UN agencies contains a mixture of killed Vibrio cholerae O1 bacteria and the non-toxic B subunit of cholera toxin, and is marketed under the trade name Dukoral (Crucell, Netherlands). This vaccine was licensed largely on the basis of studies done more than 20 years ago in Bangladesh4 and Peru5 that showed 85% protection for the rst 46 months and 60% protection for 2 years after a primary regimen of two or three doses. The protection declined substantially in the third year and was evident against V cholerae O1 El Tor only in the rst year for individuals younger than 5 years.6 The vaccine is used primarily by people travelling from

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

developed countries to cholera-endemic areas and has not been used routinely by these countries in their public health systems. A second killed whole-cell oral cholera vaccine (Shanchol, Shantha Biotechnics, Hyderabad, India), which contained the same V cholerae O1 whole-cell strains as Dukoral, at dierent doses, and killed V cholerae O139 bacteria but not the B subunit component, was developed and licensed in India in 2009. Licensure was based in part on results of a placebo-controlled, cluster-randomised trial in which the bivalent killed whole-cell oral cholera vaccine was shown to be safe and conferred 67% ecacy (cumulative) against cholera severe enough to require treatment in a health facility at 2 years and 66% ecacy at 3 years of follow-up.7,8 The vaccine was prequalied by the WHO in 2011.

However, use of oral cholera vaccines can be hindered by the perceived short duration of protection that the vaccine confers (compared with recommended vaccines for other diseases). To assess the duration of protection of the vaccine, the results of which also have implications for the requirements for a boosting dose or reimmunisation, we extended the follow-up period to 5 years.

Methods
Study design
Our cluster-randomised, double-blind, placebo-controlled trial was done in a cholera-endemic area in the urban slums of Kolkata, India, with a population of about 109 000 individuals. Before the trial, a census was done to enumerate the population according to their regular or legal residence, map the households residing in the area, assign unique study identication numbers to each individual, and establish household socioeconomic, water-use, sanitation and hygienic characteristics. Details on the study site, study agents, study procedures, and assembly of participants for this trial have been reported previously.7,8 Residents who were at least 1 year old and not pregnant were eligible to participate in the study. The study protocol was approved by the Drugs Controller General of India, the ethics committee of the National Institute of Cholera and Enteric Diseases, the Health Ministry Screening Committee of India, and the International Vaccine Institute institutional review board. We obtained written informed consent from residents older than 18 years and from the guardians of residents aged 117 years. We also obtained written assent from residents aged 1217 years. An independent, international, data and safety monitoring board reviewed the study protocol, assessed serious adverse events, and approved freezing of data and the analytical plan before analysis was started.

3933 clusters assessed for eligibility (108 777 individuals) 1003 individuals excluded (aged <10 year) 3933 clusters randomly allocated (107 774 individuals)

Vaccine group 1727 clusters received allocated vaccine (cluster size 1302 residents, median size 10) with 33 127 individuals 239 clusters did not receive allocated vaccine* (cluster size 154 residents, median size 5) with 19 085 individuals 215 individuals medically ineligible (pregnant) 18 870 individuals did not participate

Placebo group 1768 clusters received allocated placebo (cluster size 11530 residents, median size 9) with 36 201 individuals 199 clusters did not receive allocated placebo* (cluster size 127 residents, median size 5) with 19 360 individuals 203 individuals medically ineligible (pregnant) 1 individual given wrong investigational product 19 157 individuals did not participate

0 clusters lost to follow-up between dose 1 and dose 2* 32 individuals lost to follow-up between dose 1 and dose 2 6 deaths 26 migrated out of study area 6 clusters discontinued (cluster size 13 residents, median size 2) 1128 individuals discontinued 53 incompletely dosed 7 given wrong investigational product 13 not given dose 2 because medically ineligible 1055 not given dose 2 because non-participants

2 clusters lost to follow-up between dose 1 and dose 2* (cluster size 11 residents, median size 1) 37 individuals lost to follow-up between dose 1 and dose 2 9 deaths 28 migrated out of study area 9 clusters discontinued (cluster size 13 residents, median size 1) 1175 individuals discontinued 43 incompletely dosed 16 not given dose 2 because medically ineligible 1116 not given dose 2 because non-participants

Randomisation and masking

We used stratied randomisation to preassign all eligible participants in each dwelling to one of the four letter codes that served to identify the vaccine and placebo. The assigned codes were printed in vaccination books for use during dosing. Assignment of each dwelling to one of the codes used six strata, dened by administrative ward and dwelling size (20 residents or >20 residents). An external statistician, who was masked to the identities of the codes, randomly assigned dwellings to the four codes in a 1:1:1:1 ratio within each of the strata.

1721 clusters analysed in per-protocol population* (cluster size 1287 residents, median size 10) with 31 932 individuals 0 clusters excluded from analysis 35 individuals excluded from analysis 34 incompletely dosed at dose 2 1 given wrong investigational product at dose 2

1757 clusters analysed in per-protocol population* (cluster size 11448 residents, median size 9) with 34 968 individuals 0 clusters excluded from analysis 21 individuals excluded from analysis 19 incompletely dosed at dose 2 2 given wrong investigational product at dose 2

Procedures
Each dose of the modied killed bivalent whole-cell vaccine (Shanchol) contained about 15 10 inactivated V cholerae O1 and 5 10 V cholerae O139 bacteria dispensed based upon their lipopolysaccharide contents: 600 ELISA units of lipopolysaccharide of formalin-killed

Figure 1: Trial prole *Entire cluster did not receive allocated vaccine or placebo, was lost to follow-up, or discontinued. Present in the study area during the vaccination campaign, but did not attend the vaccination outpost.

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

V cholerae O1 El Tor Inaba (strain Phil 6973); 300 ELISA units of lipopolysaccharide each of heat-killed V cholerae O1 classical Ogawa (strain Cairo 50), formalin-killed V cholerae O1 classical Ogawa (strain Cairo 50), and heatkilled V cholerae O1 classical Ogawa (strain Cairo 48); and 600 ELISA units of lipopolysaccharide of formalin-killed V cholerae O139 (strain 4260B). We used vials containing identical-appearing heat-killed Escherichia coli K12 cells as placebo. Single-dose vials were labelled with one of four letter codes, two for vaccine and two for placebo. Project sta and study individuals were unaware of the identities of the codes. Two doses of the assigned agents were given in two rounds: from July 27 to Aug 13, 2006, and from Aug 27 to Sept 10, 2006. Before vaccination, another census of the study population was done to update the census used for allocation. Surveillance was done in nine community clinics established for the trial and in two hospitals serving the study population. Study physicians completed structured forms to obtain pertinent clinical information, and obtained faecal specimens that were tested for V cholera, including identication of O1 and O139 serogroups and Inaba and Ogawa serotypes. Biotype was ascertained for all O1 isolates, and the biotype of the cholera toxin genetically encoded was identied as previously described.8,9 A diarrhoeal visit was dened as having, in the 24 h before presentation, at least three loose stools or (if one, two, or an indeterminate number of loose stools were reported) the patient must have shown evidence of some or severe dehydration according to WHO criteria.10 The onset of a diarrhoeal episode was the day on which the patient rst reported loose or liquid stools. Diarrhoeal visits for which the date of onset was less than or equal to 7 days from the date of discharge for the previous visit were grouped into the same diarrhoeal episode. The primary endpoint (a cholera episode) was dened as a diarrhoeal episode in which the diarrhoea was non-bloody, a faecal specimen yielded V cholerae O1 or O139, and a domiciliary check conrmed that the individual had visited the treatment centre for diarrhoea on the recorded date of presentation. Demographic surveillance for migrations and deaths among the study population was maintained during the 5 years of followup, and verbal autopsies were done for identication of the cause of deaths.

amount swallowed. Episodes of cholera were included in the intent-to-vaccinate analysis if they had onset between 1 day and 1825 days after the rst dose. In both analyses, we analysed only the rst episodes of cholera. Before analysis, data were frozen, and the analytic plan was approved by the data and safety monitoring board. In our study, the zero time (day 0) for per-protocol and intent to vaccinate analyses was the date of second dose and rst dose, respectively. Individuals who were present (ie, number at risk) at day 0 were assessed in the year 1 analysis. We did survival analyses to calculate vaccine protective ecacy with measurements of the time to the rst episode of cholera, censoring the follow-up of individuals who died or migrated out.7 In descriptive analyses, we tted Kaplan-Meier curves. We tted unadjusted and adjusted Cox proportional hazards regression models, after verifying that the proportionality assumptions were fullled for all independent variables,1113 and estimated the hazard ratios by exponentiation of the coecient for the vaccine variable in these models. We calculated percentage vaccine protective ecacy as (1hazard ratio) 100. We
Overall population Vaccine recipients Placebo recipients Non-participants

30 25 Incidence per 1000 per year 20 15 10 05 0

Year 1

Year 2

Year 3

Year 4

Year 5

Figure 2: Incidence of laboratory-conrmed cholera cases by study year in Kolkata, India

Vaccine n Per protocol Cases

Placebo n Cases

Crude protective ecacy (95% CI)

Adjusted protective ecacy (95% CI)

Statistical analysis
We analysed ecacy in a per-protocol population and according to intention to vaccinate. In the per-protocol analysis, we investigated vaccine protection in participants who had completely ingested two doses of assigned study treatment and assessed cholera episodes that occurred between 14 days and 1825 days after receipt of the second dose. In the intent-to-vaccinate analysis, we included all individuals who received at least one dose of study treatment, and analysed the data on the basis of the assigned treatment irrespective of the

31 932 69 82

34 968 219 36 201 234

66% (5375; p<00001) 65% (5274; p<00001)* 62% (4972; p<00001) 60% (4671; p<00001)

Intent to vaccinate 33 127

*Estimated after adjustment for variables used to stratify for randomisation (cluster size and ward of residence) and individual-level variables associated with risk of cholera (age group, individuals living in their own house, living in a household with important economic contributor with a stable occupation, living in a household with monthly household expenditure per head higher than median expenditure, and longer than median distance from the household to the nearest water body). Estimated after adjustment for variables used to stratify for randomisation (cluster size and ward of residence) and individual-level variables associated with risk of cholera (age group, individuals living in a household with literate household head, living in a household with important economic contributor having stable occupation, living in a household owning at least one luxury item, and distance from the household to the nearest health clinic).

Table 1: Overall occurrence of cholera episodes and cumulative protective ecacy of the oral cholera vaccine

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

used robust sandwich variance estimates to account for the design eect of cluster randomisation, allowing inferences for vaccine ecacy at the individual level.14 To calculate vaccine protective ecacy adjusting for putative confounders, we included stratication variables (cluster size and ward of residence) in the model irrespective of the level of statistical signicance. In addition, baseline variables that were signicantly associated with time to event at p<010 in bivariate analyses were candidates as independent variables in the model. To avoid overtting the models, we used a backward elimination algorithm to select independent model variables at p<010. The list of baseline variables is described elsewhere.8 We assessed vaccine ecacy in predened subgroups. We assessed heterogeneity of vaccine protection by levels of baseline variables in these subgroups by analysing two-way interaction terms in the models. To test whether vaccine protection declined in the 5 years of follow-up, we tested the proportional hazards assumption for the vaccine variable. To assess cases prevented from vaccination by age groups, we calculated
A
1000 Cumulative proportion without cholera 0998 0996 0994 0992 0990 0 0 Number at risk Vaccine 31 932 Placebo 34 968 365 30 532 33 466 730 28 976 31 677 1095 27 642 30 179 1460 25 964 28 555 1825 Placebo Vaccine

the prevented number of cases for every 1000 vaccinated individuals (Pr) as: Cp Cv Pr = 1000 Np Nv Where Cv is the number of cases in the vaccine group, Cp is the number of cases in the placebo group, Nv is number of individuals in the vaccine group, and Np is the number of individuals in the placebo group. Although our study protocol called for one-tailed analyses of protective ecacy, we present p values and 95% CIs for protective ecacy in a more conservative two-tailed fashion to assist interpretation of our data. The study extended the 3 year analysis by including 5 years of follow-up, and for this secondary analysis the threshold of signicance was p<005 with corresponding two-sided 95% CI. All statistical analyses were done with SAS version 9.3. This trial was registered at ClinicalTrials.gov number, NCT00289224.

Role of the funding source

The funding agencies of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had nal responsibility for the decision to submit for publication.

Results
We included 31 932 participants in 1721 clusters in the vaccine group and 34 968 participants in 1757 clusters in the placebo group in the per-protocol population (gure 1). In the intent-to-vaccinate analysis, we included 1727 clusters with 33 127 recipients of vaccine and 1768 clusters with 36 201 recipients of placebo. Individuallevel and cluster-level baseline characteristics were comparable for recipients of the vaccine and placebo, as reported previously.7 We noted no substantive imbalances in baseline variables in participants who were excluded or lost to follow-up in each group. In the 5 years of follow-up of the per-protocol population, there were 69 episodes of cholera in the vaccine group and 219 in the placebo group. For the intent-to-vaccinate population, there were 82 episodes of cholera in the vaccine group and 234 episodes of cholera in the placebo group. All detected cholera episodes were attributable to V cholerae O1 El Tor biotype, and 95% of the isolates were Ogawa serotype. One death in the vaccine group and two deaths in the placebo group were attributed to diarrhoea or gastroenteritis of presumed infectious origin (international classication of diseases 10 code A09). None of these three deaths or any other deaths could be attributed to V cholerae infection or to administration of the vaccine. Incidence of laboratoryconrmed cholera among all residents, whether or not they participated in the trial, varied by year after the

B
1000 Cumulative proportion without cholera 0998 0996 0994 0992 0990 0 0 Number at risk Vaccine 33 127 Placebo 36 201 365 31 714 34 703 730 1095 Days of follow-up 30 065 32 784 28 664 31 238 1460 27 148 29 567 1825

Figure 3: Kaplan-Meier estimates of the cumulative risk of not having cholera (A) Per-protocol population. (B) Intent-to-vaccinate population

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

Vaccine n Per protocol 1050 years 50150 years 150 years Intent to vaccinate 1050 years 50150 years 150 years 2166 7237 23 724 31 21 30 2082 7023 22 827 27 18 24 Cases

Placebo n Cases

Crude protective ecacy (95% CI)

Adjusted protective ecacy* (95% CI)

Prevented cases per 1000 vaccinations

p value

007 2263 7698 25 007 2376 7893 25 932 53 62 104 59 65 110 45% (1166) 68% (4183) 75% (6084) 43% (1063) 65% (3880) 70% (5481) 42% (564) 68% (4282) 74% (5884) 39% (461) 65% (3980) 69% (5380) 105 55 31 009 105 53 30

*Estimated after adjustment for the variables used to stratify for randomisation, and other covariates associated with the risk of cholera (listed in the appendix). Estimated for the overall interaction term between vaccine and age group variables in the model.

Table 2: Overall occurrence of cholera episodes and cumulative protective ecacy of the oral cholera vaccine by age at vaccination

vaccination period (gure 2). However, vaccine protection was evident in the 5 years of follow-up, as shown by the decreased incidence in recipients of vaccine compared with recipients of placebo. In the per-protocol analysis, the adjusted cumulative 5 year protective ecacy of this vaccine for individuals aged 1 year or older (at the time of receipt of the rst dose of vaccine) was 65% (95% CI 5274%, p<00001) and in the intent-to-vaccinate analysis the protective ecacy was 60% (4671%, p<00001; table 1). In the per-protocol analysis, Inaba serotype was detected in only four episodes in the vaccine group and ten episodes in the placebo group during follow-up (unadjusted protective ecacy 56%, 95% CI 37 to 86; p=016). Event-free survival for patients favoured vaccine over placebo (gure 3). When we stratied cumulative 5 year protection by age, protective ecacies did not dier signicantly for either the per-protocol or the intent-tovaccinate analyses (table 2). Nonetheless, the point estimates suggested reduced ecacy in individuals younger than 50 years at vaccination (table 2). The point estimate for children aged 1020 years was also low at 31% (95% CI 58 to 70; p=037) with eight cases in 349 recipients of vaccine and 16 cases in 439 recipients of placebo. However, more cases seemed to be prevented by vaccination (105 per 1000) for children aged 1050 years than were prevented in individuals aged 50150 years (55 per 1000) and aged at least 150 years (31 per 1000; table 2). We noted no signicant dierences in the protective ecacies for the trial population for study years 15 during follow-up (table 3). Occurrence of cholera episodes and protective ecacy of the oral cholera vaccine by age at vaccination and year of followup in per-protocol population are shown in the appendix.

Vaccine n Per protocol Year 1 Year 2 Year 3 Year 4 Year 5 Intent to vaccinate Year 1 Year 2 Year 3 Year 4 Year 5 33 127 31 714 30 065 28 664 27 148 18 10 21 29 4 31 932 30 532 28 976 27 642 25 964 11 9 18 27 4 Cases

Placebo n Cases

Crude protective Adjusted protective p value ecacy (95% CI) ecacy* (95% CI)

027 34 968 23 48% (11 to 75) 78% (52 to 90) 67% (41 to 82) 57% (26 to 75) 80% (40 to 93) 42% (6 to 68) 76% (50 to 89) 60% (31 to 77) 57% (27 to 75) 81% (44 to 93) 45% (19 to 74) 76% (47 to 89) 66% (40 to 81) 58% (29 to 75) 80% (40 to 93) 027 36 201 34 34 703 46 32 784 31 238 29 567 57 74 23 40% (10 to 67) 72% (42 to 87) 57% (26 to 75) 60% (33 to 76) 81% (42 to 94) 33 466 45 31 677 60 30 179 69 28 555 22

*Estimated after adjustment for the variables used to stratify for randomisation, and other covariates associated with the risk of cholera (listed in the appendix). p value was calculated by testing proportional hazards assumptions for the vaccine variable.

Table 3: Overall occurrence of cholera episodes, and protective ecacy of the oral cholera vaccine

Discussion
During 5 years of follow-up, a two dose regimen of the cholera vaccine reduced the incidence of clinically signicant cholera by about two-thirds in the study population of individuals vaccinated at the age of 1 year

or older. Although 95% of the cholera cases identied in the study were due to V cholerae O1 Ogawa, point estimates of protection against El Tor Inaba and Ogawa were much the same. The absence of cases attributable to V cholerae O139 precluded assessment of ecacy against this serogroup. Overall protection was sustained for 5 years follow-up. This level of sustained protection has not been reported for any previous cholera vaccine (panel). We did not detect signicant dierences in the cumulative 5 year vaccine protection among dierent age groups at vaccination. However, our sample size was not calculated with adequate power to assess ecacy by age group. Nevertheless, the seemingly low ecacy in individuals aged 1050 years and subsets of those aged 1020 years (who might have been immunologically naive at vaccination) suggests that vaccine might not protect children younger than 5 years for 5 years. Results

See Online for appendix

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

Panel: Research in context Systematic review In 2011, a Cochrane review15 of randomised or quasirandomised studies of oral cholera vaccines was published, assessing seven ecacy studies with ve variations of the killed whole-cell oral cholera vaccine. The overall vaccine ecacy during the rst year was 52% and during the second year was 62%. Protective ecacy was lower in children younger than 5 years (38%) than in individuals older than 5 years (66%). The review noted that for the present vaccine formulation, no data existed for protection beyond 2 years, but it was unlikely to last beyond 3 years. To supplement this review, we searched PubMed without date or language restriction with the terms killed oral cholera vaccine and ecacy or eectiveness, which revealed ve other publications not included in the Cochrane review. Four were case-control studies; two of these studies were done in Africa and showed 78% and 79% eectiveness against cholera with the recombinant B-subunit containing wholecell oral cholera vaccine, up to 6 months and 15 months after vaccination, respectively, after which no cases of cholera were detected.16,17 In one of these studies, eectiveness did not dier between individuals younger than 5 years of age and in an older age group.16 Two case-control studies with a previous version of the bivalent killed whole-cell oral cholera vaccine available only in Vietnam showed 76% eectiveness when used during an outbreak18 and 50% eectiveness up to 35 years after vaccination.19 Vaccine eectiveness by age group was not assessed separately in these studies. The fth study was the report on 3 years of follow-up of our trial. At 3 years of follow-up, the modied bivalent killed whole-cell oral cholera vaccine aorded 66% ecacy in all age groups; however, ecacy was somewhat reduced in children aged 10<50 years at 43%.7 Interpretation After 5 years of follow-up, we noted that the bivalent killed whole-cell oral cholera vaccine remained 65% protective in a cholera-endemic area in Kolkata, India. This level of sustained protection for 5 years has not been reported previously with other oral cholera vaccines. Although statistical comparisons were not signicant, vaccine ecacy seemed to vary by age group and by year of follow-up. Notably, a lower point estimate of ecacy that we noted in young individuals (1050 years) than in older individuals suggests that the vaccine might not provide equivalent levels of protection to younger children. In view of the increased cholera incidence in children younger than 5 years of age, and despite the possibility of lower levels of protection among young children, our study suggests an improved morbidity reduction in this age group. Our ndings support the use of oral cholera vaccine as a key component in public health programmes to protect against life-threatening and economically devastating cholera infections.

from a previous trial3 of a dierent killed oral cholera vaccine in Bangladesh are consistent with low levels of protection in young children. The Bangladeshi study reported that children vaccinated at 2050 years of age had reduced levels of protection against El Tor cholera in the rst year of observation after vaccination.3 However, even if low levels of protection were aorded to young children, our results suggest a potential reduction in morbidity because of the higher cholera incidence in the youngest age groups. Postlicensure assessments of the use of the vaccine studied in our trial in cholera-endemic populations should be done to address this issue more denitively. Although we noted variation in ecacy by year of followup, including a substantial increase in protection in follow-up year 5 (perhaps due a boosting of immunity20 from the large outbreak during March and April, 2010), the variations in vaccine ecacy by year of follow-up did
6

not dier signicantly. However, analysis by year of followup was not planned and the sample size was not calculated with adequate power to assess such trends. Thus, we cannot conclude that the dierences in protection by length of protection do not exist. Beyond limitations resulting from the limited power of our trial to address interactions of vaccine protection with baseline variables and to assess trends in protection, our assessment has additional important limitations. First, because the trial was done in a population that has endemic cholera at high rates every year, the protection that we noted reects protection from vaccine-induced immunity, together with the immunity induced by natural cholera infections occurring after baseline in recipients of vaccine and placebo. Second, individuals in our study population probably had previous exposure to cholera and some level of pre-existing natural immunity at baseline. Thus, our trial did not assess the protective ecacy of this vaccine in immunologically naive individuals who might be at risk in some cholera epidemics, such as that in Haiti. Third, we did not assess whether use of this vaccine prevents asymptomatic or mild disease but only whether vaccination prevents disease that is severe enough to seek treatment. These asymptomatic or mild cases could contribute to transmission. Our ndings conrm that this vaccine is ecacious in cholera-endemic settings, and could be adopted as part of a wider cholera control eort. Now prequalied by WHO, this vaccine does not require a buer (unlike Dukoral), making it easier to provide in areas where access to clean water is restricted. In 2010, WHO issued a position paper supporting the use of cholera vaccines in conjunction with other preventive and control strategies in areas where the disease is endemic and should be considered in areas at risk for outbreaks.21 Supported by a World Health Assembly resolution to expand cholera control eorts,22 a WHO Technical Working Group on Creation of an Oral Cholera Vaccine Stockpile issued a report in 2012 recommending that an oral cholera vaccine stockpile of 2 million doses should be maintained and used in response to cholera outbreaks, especially in low-income countries.23 Funding has been obtained to create and operate the stockpile. Cholera vaccines should be used in combination with health services that provide rapid detection and treatment of cholera cases with rehydration solutions and antibiotics, and increase access to safe water and sanitation, promote personal hygiene, and improve health education and community mobilisation.2123 Cholera is now endemic in more than 50 countries and causes substantial mortality and high economic costs in some of the worlds poorest nations.24 Our ndings support the use of oral cholera vaccine as a key component in an integrated public health programme to protect against life-threatening and economically devastating cholera infections.

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1

Articles

Contributors SKB, DS, MA, SK, BM, JLD, RC, AD, ALL, GBN, and JDC conceived and designed the study. MA, YAA, BM, JKP, MKP, DRK, AD, TFW, and JDC analysed the data. BSar, SKN, RC, MSD, and JH contributed reagents, materials, or analysis tools. SKB, DS, MA, YAY, BSah, JLD, JH, AD, GBN,ALL, TFW, and JDC wrote the paper. Conicts of interest MSD is an employee of Shantha Biotechnics. All other authors declare that they have no conicts of interest. Acknowledgments We are grateful to the people living in our study area in Kolkata, India, and our eld sta. We thank Roger Glass and Duncan Steele who served as our data safety monitoring board, Alok Deb, Anup Palit, and R Leon Ochiai who provided valuable support for this project. References 1 Ali M, Lopez AL, You YA, et al. The global burden of cholera. Bull World Health Organ 2012; 90: 20918A. 2 Kirigia JM, Sambo LG, Yokouide A, Soumbey-Alley E, Muthuri LK, Kirigia DG. Economic burden of cholera in the WHO African region. BMC Int Health Hum Rights 2009; 9: 8. 3 Clemens JD. Vaccines in the time of cholera. Proc Natl Acad Sci 2011; 108: 852930. 4 Clemens JD, Sack DA, Harris JR, et al. Field trial of oral cholera vaccines in Bangladesh: results from three-year follow-up. Lancet 1990; 335: 27073. 5 Sanchez JL, Vasquez B, Begue RE, et al. Protective ecacy of oral whole-cell/recombinant-B-subunit cholera vaccine in Peruvian military recruits. Lancet 1994; 344: 127376. 6 van Loon FP, Clemens JD, Chakraborty J, et al. Field trial of inactivated oral cholera vaccines in Bangladesh: results from 5 years of follow-up. Vaccine 1996; 14: 16266. 7 Sur D, Kanungo S, Sah B, et al. Ecacy of a low-cost, inactivated whole-cell oral cholera vaccine: results from 3 years of follow-up of a randomized, controlled trial. PLoS Negl Trop Dis 2011; 5: e1289. 8 Sur D, Lopez AL, Kanungo S, et al. Ecacy and safety of a modied killed-whole-cell oral cholera vaccine in India: an interim analysis of a cluster-randomised, double-blind, placebo-controlled trial. Lancet 2009; 374: 1694702. 9 Ansaruzzaman M, Bhuiyan NA, Nair BG, et al. Cholera in Mozambique, variant of Vibrio cholerae. Emerg Infect Dis 2004; 10: 205759.

10

11 12 13 14

15

16

17

18

19

21 22

23

24

WHO. The treatment of diarrhoea: a manual for physicians and other senior health workers, 4th revision. Geneva: World Health Organization, 2005. Reid N, Crepeau H. Inuence functions for proportional hazards regression. Biometrika 1985; 72: 19. Lin D, Wei L, Ying Z. Checking the Cox model with cumulative sums of Martingale-based residuals. Biometrika 1993; 80: 55772. Lin DY, Wei LJ. The robust inference for the proportional hazards model. J Am Stat Assoc 1989; 84: 107478. Lee EW, Wei LJ, Amato D. Cox-type regression analysis for large numbers of small groups of correlated failure time observations. Netherlands: Kluwer Academic, 1992: 23747. Sinclair D, Abba K, Zaman K, Qadri F, Graves PM. Oral vaccines for preventing cholera. Cochrane Database Syst Rev 2011; 3: CD008603. Lucas M, Deen JL, von Seidlein L, et al. High-level eectiveness of a mass oral cholera vaccination in Beira, Mozambique. N Engl J Med 2005; 352: 75767. Khatib AM, Ali M, von Seidlein L, et al. Eectiveness of an oral cholera vaccine in Zanzibar: ndings from a mass vaccination campaign and observational cohort study. Lancet Infect Dis 2012; 12: 83744. Anh DD, Lopez AL, Thiem VD, et al. Use of oral cholera vaccines in an outbreak in Vietnam: a case control study. PLoS Negl Trop Dis 2011; 5: e1006. Thiem VD, Deen JL, von Seidlein L, et al. Long-term eectiveness against cholera of oral killed whole-cell vaccine produced in Vietnam. Vaccine 2006; 24: 4297303. 20 Plotkin SA. Correlates of vaccine-induced immunity. Clin Infect Dis 2008; 47: 40109. Cholera vaccines: WHO position paper. Wkly Epidemiol Rec 2010; 85: 11728. WHO. Cholera: mechanism for control and prevention. WHA 64.15. Sixty-fourth World Health Assembly; Geneva, Switzerland; May 24, 2011. WHO. Meeting report: WHO technical working group on creation of an oral cholera vaccine stockpile. Geneva, Switzerland; April 2627, 2012. Holmgren J. A case for control of cholera in Africa by vaccination. Lancet Infect Dis 2012; 12: 81819.

www.thelancet.com/infection Published online October 18, 2013 http://dx.doi.org/10.1016/S1473-3099(13)70273-1