The recent spread of cholera to Latin America, together with the persistent burden of this disease in Asia and Africa, have stimulated efforts to evaluate new cholera vaccines in field settings. Although the standard experimental paradigm for vaccine field trials is well established, the success of these trials will also depend on suitable consideration of the epidemiology of cholera and of cholera vaccination in the setting under study. Epidemiological studies done in Bangladesh emphasize the importance of appreciating the poorly predictable, multifocal occurrence of cholera in estimating a probable incidence of cholera for a field trial. They also underscore how the filtering effect of enrolling subjects into a prospective trial can dramatically reduce the available population for study, and can yield a study sample whose expected risk of cholera differs markedly from that for the source population. Finally, the data highlight the subtle effects that the mode of surveillance and the choice of an outcome definition can have upon protective efficacy, and emphasize the need for subgroup analyses that address the distinctive variations in vaccine protection that may occur in subjects differing in age and in ABO blood groups, and in subjects exposed to classical versus El Tor cholera.
PIP: It is important to predict cholera incidence for clinical trials, but it is hard to do so, even in areas where there is a regular endemic pattern. In Matlab, Bangladesh, cholera tends to occur seasonally, right before and after the summer monsoon. Various host facts affect cholera incidence, e.g., insufficient natural immunity and exposure to infected persons. Passive surveillance data in Matlab demonstrate that cholera occurs in a temporo-spatially clustered, multifocal pattern within a community. Ineligibility, refusal to participate, and absenteeism have a strong filtering effect on the population in a field trial. For example, in a field trial of killed oral cholera vaccines in Bangladesh, only 62.285 of the original 124,035 people completed the trial. As a result of nonparticipation, the expected incidence for nonvaccinees was overestimated. The definition of cholera used also determines incidence. For example, if a trial in endemic areas uses the classical definition (watery diarrhea with excretion of cholera vibrios but no copathogens), it would miss those cases with atypical symptoms (e.g., 21% of treated cholera episodes in the oral vaccine trial were clinically atypical or microbiologically mixed). Age related distinctions, ABO blood group, and biotype of the infecting organisms all influence vaccine efficacy. The above issues were considered when designing field trails of killed oral cholera vaccines in Colombia and Peru, but most people in these countries lack preexisting natural immunity, these areas have only E1 for cholera, and it will be very difficult to predict incidence among nonvaccines. This review of epidemiological studies of cholera in Bangladesh, including a large-scale field trial of killed oral cholera vaccines, describes the significance of these considerations.