Efficacy of a bivalent killed whole-cell cholera vaccine over five years: a re-analysis of a cluster-randomized trial

February 20, 2018



An estimated 3 million cholera cases occur each year, about 100,000 of which are fatal [1, 2]. Although the disease is preventable with the provision of clean water and sanitation facilities, this would not be feasible in the near-term in most cholera-endemic regions. A short-term solution may be the use of oral cholera vaccine (OCV). Vietnam regularly uses OCV to control cholera, and other countries with endemic cholera are currently considering this option [3, 4, 5]. Improved estimates of the effectiveness of mass vaccination and the duration of protection, thought to be 3 to 5 years [6, 7], would be useful when planning mass OCV deployment in settings with limited resources.


The design and implementation of the OCV trial (ClinicalTrials.gov identifier: NCT00289224) has been described in detail. Briefly, the trial was conducted in three wards of the urban slums of Kolkata, India, with a population of about 109,000. Study clusters were defined by dwellings, each of which comprises one or more households with shared access to water and bathrooms or latrines. Clusters were randomized 1:1 to cholera vaccine or heat-killed Escherichia coli K12 placebo stratified by ward of residence (29, 30 and 33) and cluster size (20 or fewer individuals or more than 20 individuals during a pre-study census).

Statistical Analysis

In keeping with previous analyses we analyzed only the first episodes of cholera of each participant and excluded events that occurred less than 14 days after each individual’s second (and final) dose of OCV or placebo. Participants with events within 14 days of their second doses were removed from the analysis. Participants were right-censored when they had cholera, changed households, died, or when the study concluded in 2011. Also in keeping with previous analyses, time to event was treated as the time in days since the participant’s second dose of vaccine or placebo. To study the impact of the large outbreak that occurred in the fourth year of the study (March–April 2010) on the analysis, we also performed analyses in which the participants who are either censored or infected during the outbreak period were excluded from analysis (Fig. 1).

Fig. 1

Monthly number of culture-confirmed cholera cases during the study. The number of confirmed cases among all per-protocol study participants are indicated by the bars, with portions shaded gray to indicate the subset randomized to OCV. A large outbreak occurred in March–April 2010, and the bars representing these two months are indicated with black dots