^{1}

^{2}

^{*}

^{3}

^{4}

^{5}

^{3}

^{4}

IML, AN, MA, NS, and JDC designed the study and analyzed the data. All authors contributed to writing the paper. AN developed (i.e., researched, programmed, tested, and implemented) the stochastic simulator of cholera in Matlab (Bangladesh), and analyzed and interpreted the statistical and epidemiological results generated by the simulator. MY was involved as an investigator in the cholera vaccine trial that was conducted in the Matlab field area of the ICDDR,B from 1985 to 1989. NS helped analyze the simulation studies and transformed results into figures. Analysis of the data was based on a trial of cholera vaccines that JDC led in Bangladesh, and on his further analysis of herd protection in the trial data.

The authors have declared that no competing interests exist.

Although advances in rehydration therapy have made cholera a treatable disease with low case-fatality in settings with appropriate medical care, cholera continues to impose considerable mortality in the world's most impoverished populations. Internationally licensed, killed whole-cell based oral cholera vaccines (OCVs) have been available for over a decade, but have not been used for the control of cholera. Recently, these vaccines were shown to confer significant levels of herd protection, suggesting that the protective potential of these vaccines has been underestimated and that these vaccines may be highly effective in cholera control when deployed in mass immunization programs. We used a large-scale stochastic simulation model to investigate the possibility of controlling endemic cholera with OCVs.

We construct a large-scale, stochastic cholera transmission model of Matlab, Bangladesh. We find that cholera transmission could be controlled in endemic areas with 50% coverage with OCVs. At this level of coverage, the model predicts that there would be an 89% (95% confidence interval [CI] 72%–98%) reduction in cholera cases among the unvaccinated, and a 93% (95% CI 82%–99%) reduction overall in the entire population. Even a more modest coverage of 30% would result in a 76% (95% CI 44%–95%) reduction in cholera incidence for the population area covered. For populations that have less natural immunity than the population of Matlab, 70% coverage would probably be necessary for cholera control, i.e., an annual incidence rate of ≤ 1 case per 1,000 people in the population.

Endemic cholera could be reduced to an annual incidence rate of ≤ 1 case per 1,000 people in endemic areas with biennial vaccination with OCVs if coverage could reach 50%–70% depending on the level of prior immunity in the population. These vaccination efforts could be targeted with careful use of ecological data.

Using data from Bangladesh, Ira Longini and colleagues develop a mathematical model predicting that oral vaccination of 50%-70% of the population could control cholera transmission in an endemic region.

Throughout history, there have been devastating outbreaks of cholera—a gut infection characterized by diarrhea and severe dehydration—around the world. These days, cholera is mainly confined to developing countries where it disrupts social structures, impedes economic development, and probably causes about 100,000 deaths a year. People get cholera, which is caused by a bacterium called

The best way to control cholera is to ensure that everyone has access to safe water and good sanitation, but this is not possible in some poor countries, in refugee camps, or after natural disasters such as floods. Oral cholera vaccines (preparations given by mouth that stimulate the immune system to attack

The researchers built a large-scale model of cholera transmission using information about the population of Matlab, Bangladesh (a region where cholera is endemic), together with data on the biology of cholera and data from a large oral vaccine trial done in Matlab in the 1980s. They used this model to predict whether cholera would be controlled after vaccination of different proportions of the population. They found that cholera transmission would be controlled if half the population in the region was vaccinated. This level of vaccine coverage reduced the number of cholera cases among unvaccinated people by 89% and among the entire population by 93%. With only one-third of the population vaccinated, the number of cases of cholera still fell by three-fourths. The model also predicted that in areas where there is less natural immunity to cholera (the people in Matlab are constantly exposed to

These findings suggest that, because of herd immunity, vaccinating only half the population could control cholera transmission in endemic regions where there is a high level of natural immunity. Where there is less natural immunity, more of the population would need to be immunized. Although mass immunization of even 70% of a population should be achievable, for maximal protection against

Please access these Web sites via the online version of this summary at

The

Information is available from the World Health Organization on

The US Centers for Disease Control and Prevention provide information on

The UK National Health service provides simple information on

The global burden of cholera remains substantial. In 2005, 131,943 cases and 2,272 deaths were reported to the WHO, and recently major, sustained epidemics have been reported in West Africa [

This continuing high burden highlights the need for interventions to prevent cholera. While improved water and sanitation constitute the ultimate basis for the prevention of cholera, this is a far-off goal for the impoverished settings in which cholera thrives. Vaccines constitute near-term options for cholera control. During the past 20 y, killed oral cholera vaccines (OCVs) have been shown to be safe and protective in populations with endemic cholera [

In the mid-1980s, a randomized controlled vaccine trial with OCV in Matlab, Bangladesh, yielded 70% direct vaccine efficacy for up to two years [^{2} that are considered to be the geographic cholera transmission areas. The model (see

The modeled natural history of cholera is described in

Modeled natural history of cholera [

We base the indirect, overall, and total vaccine effectiveness on the reduction in infection rates when comparing the appropriate groups within a subregion with no vaccination to a comparable subregion with a fraction

Subregion 1 has a fraction _{1} > 0 people vaccinated, while the comparison subregion 2 has nobody vaccinated, i.e., _{2} = 0. We let r_{ij}_{12} = 1 − (_{01}/_{02}). The overall effect of vaccination is measured by comparing the average (over the vaccinated and unvaccinated groups) infection rates between the two subregions. Thus, the overall vaccine effectiveness, i.e., OVEF, is OVEF_{12} = 1 − (_{1}/_{2}), where the · indicates averaging over the vaccinated and unvaccinated. The total effect of vaccination is measured by comparing the infection rate in the vaccinated in subregion 1 to the unvaccinated in subregion 2. Thus, the total vaccine effectiveness, i.e., TVEF, is TVEF_{12} = 1 − (_{11}/_{02}). The direct effect of vaccination is measured by comparing the infection rates in the vaccinated and unvaccinated in the same subregion. The direct vaccine effectiveness, i.e., DVEF, is DVEF_{1} = 1 − (_{11}/_{01}). In general, these effectiveness measures could be computed across any gradient of coverage, |_{1} − _{2}|, other than those with _{2} = 0.

Vaccines would need to contain antigens that are reactive or have cross-reactivity to current circulating strains of cholera in endemic areas [

Since we are modeling endemic cholera, we assume that there is significant partial population-level immunity due to past exposure to cholera. The actual level of such protection is complex and difficult to determine. Past analysis of cholera incidence in Matlab reveals that there is partial waning of population-level immunity over six months [

We calibrated the simulation model using cholera incidence data observed in the first year of the vaccine trial (_{S}_{I}^{2} goodness-of-fit test for frequency data (

Vaccination Coverage, Average Incidence Rates, and Direct Effectiveness (Calibration Runs)

(A) No vaccination.

(B) 14% vaccination coverage of women and children.

(C) 38% vaccination coverage.

(D) 58% vaccination coverage.

Average Indirect, Total, and Overall Effectiveness of Vaccination, and Cases Prevented Per 1,000-Dose Regimens

From _{S}

These comparisons are for relative susceptibility in populations ranging from 1.5 to 2.5 times as susceptible as Matlab. The Matlab results are shown when the multiplier is 1. For populations that are 2–2.5 times as susceptible as Matlab, at least 70% vaccine coverage is needed to achieve an overall effectiveness of at least 80%. The vaccine efficacies are preset at VE_{S}_{I}

Since vaccine efficacy can vary for different field settings and vaccines, a sensitivity analysis was carried out on the VE_{S}_{I}_{S}_{I}_{S}_{I}

Sensitivity analyses also were performed for values of the seasonal boost factor, the relative infectiousness of symptomatic infectives, and varied subregion sizes. The baseline epidemic with no vaccination was calibrated to the simulated cholera incidence data for Matlab with no vaccination. Our result that in populations like Matlab 50% vaccine coverage should be sufficient to control cholera remains valid for variation in the season boost factor (^{2} each), 50% vaccine coverage was sufficient for control. But for very small subregions (0.04 km^{2} each), the average overall vaccine effectiveness approaches 75% with a vaccination coverage of 70% (

The results of this modeling study indicate that 50% coverage with OCV could control cholera transmission in endemic areas such as Matlab, Bangladesh through a combination of direct and indirect effects.. At this level of coverage, the model predicts an 89% reduction in cholera cases even among the unvaccinated, and a 93% reduction overall in the entire population. These results would apply only where cholera is endemic and population levels of immunity are relatively high. According to our simulations, areas where susceptibility is 2–2.5 as great as Matlab would need to have vaccine coverage of at least 70% to achieve cholera control. Since vaccine-induced protection with current OCV begins to wane after about two years, populations would have to be vaccinated biennially. This could be done in advance of the cholera season for regions that have clear seasonality. Alternatively, environmental predictors of cholera outbreaks could be sampled in regions where such a prediction capacity exists, and then vaccination could take place in advance of expected outbreaks. In environmental studies of cholera outbreak predictors in rural Bangladesh, increases in cholera incidence can be predicted several weeks in advance by water temperature, water depth, rainfall, conductivity, and copepod counts [

The results of our analysis are based on a mathematical model closely calibrated to cholera incidence data from a large-scale cholera vaccine trial in Matlab, Bangladesh. We would hope that the model captures the dynamics of endemic cholera sufficiently to allow us to evaluate the effectiveness of cholera vaccination at different coverage levels in other settings. However, a major limitation of this research is that the results are only as good as the data the model is based on and the veracity of the modeling assumptions. Verification of the results needs to take place through further community cholera vaccine trials with different coverage levels in a variety of geographic locations. A number of such vaccine trials are currently being planned. The modeling methods presented here can be used to help guide the design, analysis, and interpretation of such trials. Based on the results of such trials, such a model can be used to aid in designing successful cholera vaccination strategies. A second limitation of the model is that we have not explicitly modeled the level of natural immunity in populations with endemic cholera, but rather we alter the per-contact transmission probability to reflect different levels of immunity. This means that we can only approximate the effects of such immunity. More study is needed in this area.

We have not explicitly addressed how to target vaccines during a mass cholera vaccination campaign. In endemic cholera regions, partial immunity tends to increase with increasing age, as partially reflected in higher cholera incidence rates among the young [

The present analysis implies that mass immunization with killed, whole-cell based oral cholera vaccines could possibly confer a major protective impact against cholera in an endemic setting, even with modest levels of vaccine coverage, due to the combination of direct and indirect vaccine protective effects. It might, however, be questioned whether mass immunization with such vaccines is logistically feasible and affordable. Feasibility of delivery of two-dose regimens of B subunit–killed whole cell or whole cell-only oral cholera vaccines has been demonstrated in a stable refugee setting and a densely populated urban area in sub-Saharan Africa [

(19 KB PPT)

(168 KB PPT)

(35 KB PPT)

(A) Map of Bangladesh, showing location of Matlab. Matlab is in the Chandpur district of Bangladesh. It is located about 55 kilometers southeast of the country's capital, Dhaka at 23.38° north latitude and 90.72° east longitude.

(B) Close-up of Chandpur district, within which Matlab is contained.

(C) Rectangular grid mapped onto the Matlab region. The total area of the grid was approximately 384 km^{2}. This area was divided into 64 rectangular subregions of approximately 6 km^{2} each. Study baris were contained within 43 of these subregions, shown in yellow in the figure.

(182 KB PPT)

(40 KB PPT)

Calibration runs were performed with 0% (A), 30% (B), 50% (C), and 70% (D) vaccination coverage in the entire population 2 y and older in age.

(82 KB PPT)

These simulations are for scenarios with vaccine coverage in the entire population (2 y and older in age) ranging from 30% to 70%. The solid line shows the average incidence among unvaccinated people, the dashed line among vaccinated people.

(43 KB PPT)

The solid lines show the average incidence among unvaccinated people, the dashed lines among vaccinated people.

(53 KB PPT)

(41 KB PPT)

(37 KB PPT)

(41 KB PPT)

The solid lines show the average incidence among unvaccinated people, the dashed lines among vaccinated people.

(48 KB PPT)

(41 KB PPT)

(41 KB PPT)

(42 KB PPT)

(43 KB PPT)

(43 KB PPT)

(42 KB PPT)

(33 KB DOC)

(34 KB DOC)

(45 KB DOC)

(34 KB DOC)

Each animation is based on a single, representative run of the simulator. For each coverage level, the map shows red dots, indicating baris with at least one cholera case, and yellow dots, indicating baris in which cases occurred but have recovered or died. Beneath each animated map is the corresponding animated epidemic curve, showing the number of cholera cases occurring each day.

(6.3 MB AVI)

Each animation is based on a single, representative run of the simulator. For each coverage level, the map shows red dots, indicating baris with at least one cholera case, and yellow dots, indicating baris in which cases occurred but have recovered or died. Beneath each animated map is the corresponding animated epidemic curve, showing the number of cholera cases occurring each day.

(4.0 MB AVI)

confidence interval

International Centre for Diarrhoeal Disease Research, Bangladesh

Bangladesh OCV, oral cholera vaccine

vaccine efficacy