Effectiveness of oral cholera vaccines in controlling outbreaks: a matched case-control study in Luapula Province, Zambia

¹Department of Public Health, Nchelenge District Health Office, Nchelenge, Zambia, ²Department of Public Health, University of Lusaka, Lusaka, Zambia, ³Department of Public Health, National TB and Leprosy Program, Lusaka, Zambia, ⁴Department of Clinical Services, Katete District Health Office, Katete, Zambia, ⁵Department of Pathology and Microbiology, Mulungushi University, Livingstone, Zambia, ⁶Department of Medicine and Clinical Sciences, Eden University, Lusaka, Zambia, ⁷University of London, London School of Hygiene and Tropical Medicine Collaboration, Foreign, Commonwealth and Development Office, Lusaka, Zambia, ⁸Department of Clinical Services, Chipangali District Health Office, Chipangali, Zambia

^&Corresponding author
Mulenga Cephas, Department of Public Health, Nchelenge District Health Office, Nchelenge, Zambia

Introduction    

Cholera is an acute watery diarrheal disease caused by toxigenic strains of the bacterium Vibrio cholerae serogroups O1 and O139 [1]. The disease is estimated to cause over 2.9 million cases and more than 95,000 deaths annually in cholera-endemic countries alone, with frequent epidemics occurring in settings characterised by poor water and sanitation infrastructure [2]. Global annual estimates range from 1.4 to 4.8 million cases and 28,000 to 142,000 deaths [3]. Cholera is characterised by the acute onset of watery diarrhoea, leading to rapid dehydration and death if not promptly treated with fluid replacement and, for severe cases, antibiotics [4].

Cholera is transmitted through the faecal-oral route, most frequently by way of contaminated water and food; hence, it primarily occurs in settings of poor water and sanitation infrastructure in sub-Saharan Africa, Asia, and, more recently, Haiti [5]. Additional complex humanitarian emergencies in recent times, with resulting infrastructure disruptions and population displacements, have added a significant cholera burden globally [6]. The prevention and control of cholera epidemics are typically based on a multidisciplinary integrated approach that may include community sensitisation, intensified epidemiological surveillance, improved access to clean water, hygiene, and sanitation, treatment of confirmed and suspected cases, and Oral Cholera Vaccination (OCV) [7]. In 2020, the World Health Organization (WHO) recommended the consideration of OCV alongside other traditional control measures, including health promotion, provision of safe water, and improved sanitation, for the prevention and control of cholera in endemic areas [3].

The use of vaccines in an outbreak is considered "preemptive" while vaccination after an outbreak begins is referred to as "reactive" [3]. Due to the lack of herd immunity and recurrence of cholera outbreaks, questions have arisen regarding the effectiveness of the cholera vaccine and why people continue to contract cholera despite being vaccinated [8]. Herd immunity (also known as population immunity) refers to the indirect protection from infection conferred to unvaccinated individuals when a sufficiently high proportion of the population is vaccinated, thereby reducing pathogen transmission. For cholera, a population-level herd immunity threshold of at least 80% vaccination coverage is required to achieve herd immunity against cholera [9]. In controlled trial settings, OCVs have demonstrated effectiveness of about 85-90% vaccination coverage over six months and 50-60% over two years [10-12].

This study had two primary objectives: to assess the effectiveness of the oral cholera vaccine administered during the 2021 mass vaccination campaign in preventing cholera during the 2023 outbreak in Luapula Province, Zambia, and to identify factors associated with the cholera outbreak, including vaccination coverage (single-dose vs. two-dose regimens) and sources of drinking water, particularly given the high vaccination coverage in the three study districts, which can contribute to reducing cholera transmission at a population level and may provide indirect protection to unvaccinated individuals.

Methods     

Study design and approach: we conducted a retrospective analysis of programmatic data from medical records in three districts of Luapula Province, Zambia (2024-2025). This study used an observational, quantitative research methodology employing a retrospective matched case-control design; predefined matching criteria were used, with a unique matched-pair identifier among cases and controls [13]. This design was selected to generate appropriate and relevant data efficiently, with reduced financial costs and within a relatively short time frame. Vaccination status was defined as having received at least one OCV dose; therefore, participants who received dose one or dose two were classified as vaccinated, while those with no documented OCV dose were classified as unvaccinated. To assess dose completion, dose-specific variables were retained separately from the original vaccination status.

Retrospective data were extracted from medical records of randomly selected health facilities in both rural and urban areas across three randomly selected districts. Random selection of facilities was performed to ensure representative sampling, minimise selection bias, and enable valid statistical inference [14]. An OCV campaign was implemented in Nchelenge, Mwansabombwe, and Chiengi districts from April 2021 to May 2021.

Data extracted from medical records included the following variables: vaccine doses received (single-dose oral cholera vaccine or two-dose oral cholera vaccine) of Euvichol-plus. The schedule for the vaccination campaign consisted of two doses administered 14 days apart. Vaccination status (vaccinated or unvaccinated), recorded for both cases and controls. Cholera status, categorised as infected (cases) or not infected (controls) and socio-demographic characteristics of participants

The primary limitations of this study were the absence of information in some source documents used for retrospective data abstraction. The planned sample size was achieved, though some study participants had incomplete source records for selected variables. Where possible, missing information was cross-checked against the surveillance records available and other documentation relevant to the study and subsequently completed in the data collection questionnaire. Information which could not be verified was treated as missing for specific variables and was excluded from the appropriate analysis.

Sample size calculation: sample size was calculated using the formula for matched case-control studies [15], assuming 80% power, a two-sided alpha of 0.05, an expected odds ratio of 2.5 (based on previous OCV effectiveness studies), and a 1:1 matching ratio. The calculated sample size was 127 cases and 127 controls, for a total of 254 participants.

Study setting: the study was conducted in Luapula Province, Zambia, targeting three districts: Nchelenge, Mwansabombwe, and Chiengi. These districts were selected because they participated in the 2021 mass OCV campaign and subsequently experienced a cholera outbreak in 2023. The 2021 OCV campaign administrative report indicated that Nchelenge district vaccinated 224 887 of 222,546 target individuals, yielding an OCV coverage of 96%, while Chiengi vaccinated 54272 of 181,038, yielding a coverage of 92%, and Mwansabombwe vaccinated 179, 132 of 56,043, giving OCV coverage of 94%. The target population included individuals aged five years and older residing in the catchment areas of both rural and urban health facilities within these districts [16].

Eligibility and recruitment: a total of 254 files were eligible and analysed. Participants were randomly selected from the catchment areas selected (rural and urban health facilities) across the three districts. Data were extracted from the medical records of patients seeking care at these facilities. The sample comprised two groups: cases (individuals diagnosed with cholera) and controls (individuals admitted for conditions other than cholera). A one-to-one matching strategy was used between cases and controls. with controls from the same specific health facility catchment area [17]. Furthermore, to evaluate oral cholera vaccine efficacy, controls were individually matched to cases by location of residence to reduce potential confounding related to geographical differences in cholera transmission, access to healthcare, and exposure to the OCV campaign. Whenever possible, controls were selected from among patients admitted for non-cholera causes during the same study period as the corresponding case [18].

Study variables

Dependent variable: this was cholera infection status, dichotomised as cases were defined according to the outbreak case definition: patients who met the clinical case definition and/or had cholera laboratory evidence. (Laboratory-confirmed using PRC: 5/62 (3.9%) and rapid diagnostic test 43 (34.0%), or clinically diagnosed cholera: 79/127 (62.1%) meeting the clinical and surveillance case definition in an endemic-declared area during the cholera outbreak) or control (admitted for non-cholera conditions during the same period) [19].

Independent variables: they included oral cholera vaccine (OCV) vaccination status (vaccinated vs. unvaccinated); OCV dose regimen (single dose vs. two doses); source of drinking water (borehole, tap water, well, or lake water); and sociodemographic factors including age, sex, education level, and district of residence [2,17].

Data collection: retrospective data abstraction was conducted from August 2024 to April 2025, with the aid of trained research assistants for data collection [13]. The medical records were used to extract information on participants who attended the health facilities and were sampled from the health facility medical records from the two rural health facility catchment areas in two rural districts and the two urban health facility catchment areas in the urban district. Data from the cholera line list and cholera registers were collected, including the number of vaccine doses received for both cases and controls (whether a single-dose or two-dose vaccination was received), vaccination status (vaccinated or not), and sociodemographic profile factors, which were also captured from medical records. The participant's cholera status was also collected and categorised as either infected (cases) or not infected (controls) [4,20].

Data analysis: data were entered into REDCap and exported to Microsoft Excel 2013 for cleaning and coding. Statistical analysis was performed using STATA 14 and StatCrunch. For statistical analysis, each matched pair was assigned a unique pair identification number.

Descriptive statistics: categorical variables were summarised using frequencies and percentages. Continuous variables (age) were summarised using mean and standard deviation or median and interquartile range, as appropriate. The Shapiro-Wilk test was used to assess data normality.

Bivariate analysis: associations between categorical variables were assessed using the chi-square test or Fisher's exact test (when expected cell counts were <5). Differences between two medians were evaluated using the Wilcoxon rank-sum test.

Multivariable conditional logistic regression analysis: variables with p < 0.20 in bivariate analysis were entered into a multivariable conditional logistic regression model to identify factors independently associated with cholera infection. Adjusted odds ratios (aOR) with 95% confidence intervals were reported. The regression models were adjusted for prespecified covariates, including age and sex where model estimation was possible. Where quasi-complete separation prevented estimation of a finite regression coefficient, adjusted estimates were not reported as a conventional OR; the limitation was instead explicitly documented.

Vaccine Effectiveness (VE) calculation: VE was calculated as (1 - odds ratio) × 100%, where the odds ratio compared the odds of vaccination among cases to that among controls [21]. A VE of 80% or higher was considered protective among cases compared to controls [21]. Because this study utilised a retrospective observational case-control study, VE was OR-driven and was interpreted as a measure of observed protective association rather than estimates of biological vaccine efficacy. OCV dose completion was assessed separately from the vaccination status. Those participants who had received at least one dose were classified as having incomplete vaccination; those who received a two-dose regimen were classified as having completed vaccination.

Sensitivity analysis: this was conducted to assess the robustness of the study by comparing participants who had completed two doses of OCV with those who had received fewer than two doses and by examining vaccination status.

Statistical significance: a p-value < 0.05 was considered statistically significant. All confidence intervals were reported at the 95% level.

Ethical considerations: approval and permission were obtained from the University of Lusaka Research Ethics Committee (Reference No: IORG0092). The National Health Research Authority (NHRA) (Reference No: NHRAR-R-3017/18/07/2025). Additional permission was secured from the Luapula Province Provincial Health Office and the respective District Health Offices. As this was a retrospective review of routinely collected, de-identified medical records, a waiver of individual informed consent was granted by the reviewing ethics committees. All data were anonymised, and confidentiality was maintained by assigning unique study identifiers and storing personal identifying information separately from analytical datasets, accessible only to the research team.

Results     

A total of 254 participants (127 cholera cases and 127 matched controls) were recruited and included in the analysis, while some source records were incomplete. The information that was missing was retrospectively reviewed and completed where possible. Participants were drawn from six health facilities across three districts, representing a catchment population of 482,508. Table 1 presents the distribution of cases and controls by health facilities. of 127 cases and 127 matched controls were identified, while some source records were incomplete. Missing information was retrospectively reviewed and completed where possible

Characteristics of the study population: Table 1 summarises the demographic characteristics of cases and controls. Across all districts, the mean age of participants was 27.6 years (SD = 17.8; range: 2-75 years). The sex distribution showed 120 females (47.2%) and 134 males (52.8%). Regarding education, 121 participants (47.6%) attended primary school, 70 (27.6%) attended secondary school, 24 (9.4%) had tertiary education, and 29 (11.4%) had no formal education. There were no statistically significant differences in age, sex, or education between cases and controls at baseline (p > 0.05 for all comparisons).

Source of drinking water: Table 2 presents combined district water source analysis for cases and controls. A striking difference in water source consumption was observed. A substantial proportion of cases reported consuming lake water; the corresponding control information was not sufficient to estimate a reliable matched OR. Conversely, across the three study districts, 72.4% of controls used borehole water compared to only 9.4% of cases; borehole water use was associated with lower odds of cholera (matched OR = 0.053; 95% CI: 0.014 - 0.140; p < 0.001). Similarly, shallow well water use demonstrated a protective association (matched OR = 0.022, 95% CI: 0.001-0.130; p < 0.001). These differences were statistically significant. Water source analysis among cases and controls therefore showed a significant difference in drinking water source, supporting the continued importance of WASH-related exposure.

Oral cholera vaccine coverage: Table 2 shows OCV coverage by district. Overall, among the vaccinated cases, only 15.7% (20/127) were vaccinated, whereas among controls, 89.0% (113/127) were vaccinated. Regarding the dose completion coverage, the combined district single-dose OCV completion rate was 15.7% (20/127) among cases, while controls had a dose-one completion rate of 89.0% (113/127) and a two-dose completion rate of 2.4% (3/127) among cases; controls had a two-dose completion rate of 81.1% (103/127).

Vaccine effectiveness analysis: overall vaccine effectiveness (any dose vs. no vaccination): in the matched analysis, 20 discordant pairs had vaccinated cases and an unvaccinated control, and about 107 discordant pairs had an unvaccinated case and a vaccinated control. The conditional logistic regression model yielded a matched OR of 0.021 (matched OR = 0.021; 95% CI: 0.005 - 0.085; P < 0.001), thus corresponding to an OR-derived VE of 97.9%. After adjusting for age and sex, the conditional logistic analysis yielded an adjusted OR of 0.019 (Adjusted [aOR] = 0.019; 95% CI: 0.007 - 0.0022; P = 0.003), corresponding to an OR-derived VE of 98.1% as an overall district combined VE based on vaccination status.

Combined assessment of drinking water and the incidence of cholera: the data for the combined assessment of drinking water and incidence of cholera showed that 9.4% of cholera cases had a history of consuming water from boreholes, 70.1% of the cases had a history of lake water consumption, and 20.5% had consumed shallow well water. In comparison with the controls, 27.6% consumed shallow well water, while 72.4% had a history of borehole water consumption, and 0% had a history of lake water consumption. Further, a chi-square test of independence demonstrated a statistically significant association between cholera status and water source (χ² = 184.62, df = 2, p < 0.001). Due to the presence of zero cells, Fisher's exact test confirmed significance (p < 0.001). Post-hoc pairwise comparisons with Bonferroni correction (α = 0.017) showed that cases were significantly more likely than controls to consume lake water compared to borehole water (OR = 68.1; 95% CI: 22.4-207.3; p < 0.001) and compared to tap water (OR = 14.2; 95% CI: 6.8-29.7; p < 0.001) (Figure 1).

Oral cholera vaccination distribution for cases and controls: in Nchelenge 54 participants out of 62 total cholera cases were not vaccinated, this showed, that 87.1% cholera cases recorded were not vaccinated with OCV, On the other hand, Mwansabombwe had 14 cases out of 19 recorded cases not vaccinated, representing 73.7% unvaccinated rate among cases, Chiengi had a total of 39 cases out of a total 46 who were not vaccinated, representing 84.7% unvaccinated coverage among cases. In comparison with the controls, Nchelenge had 95.2% of controls vaccinated and 4.8% unvaccinated, while Mwansabombwe had 13 (68.4%) of controls vaccinated and 6 (31.6%) unvaccinated. Chiengi, on the other hand, had a total of 41(89.1%) controls vaccinated and 5(10.9%) unvaccinated. Evidence from Figure 2 shows that controls who were vaccinated far outnumber those who were not vaccinated. The opposite is true among cases; cases that were not vaccinated far outnumber those who were vaccinated. A suggestion, perhaps, is that vaccination had some effect in controlling the epidemic of Cholera (Figure 2).

District-specific sample distribution of oral cholera vaccination among cases and controls: across the study districts, OCV receipt and dose completion were consistently more common among controls than cases. The dose vaccination analysis showed a total of 8 (12.9%) OCV doses were administered for Nchelenge; the dose-one completion rate was 8(12.9%), while the dose-two completion rate was 1 (2.0%) among cases. While Mwansabombwe had 5 (26.3%) total doses administered, the dose-one completion rate for cases was 5 (26.3%), and the dose-two completion rate was 1 (5.2%). Chiengi had a total of 7 (15.2%) doses; the dose-one OCV completion rate among cases was 7 (15.2%), and 1 (2.2%) for the dose-two completion rate. In comparison to controls, Nchelenge had 59 (95.2%) dose-one administered, representing a dose-one completion rate of 95.2%, while 57 had received a full two doses, representing a dose-two completion rate of 91.9%. Mwansabombwe had a total of 13 (68.4%) dose-one administered, representing a dose-one completion rate of 68.4%, and about 13 had received a full two doses, representing a two-dose completion rate of 68.4%. Chiengi, on the other hand, had 41 (89.1%) dose-one administrations among controls, representing a dose-one vaccine completion rate of 89.1%; the same coverage of 89.1% was observed in dose-two completion for Chiengi district. The summary provided in Figure 2 is that dose completion was consistently more common among controls than cases.

District-specific vaccine effectiveness: Table 2 presents the analysis by district. All districts showed a consistent association between being vaccinated and lower odds of cholera, for both unadjusted and adjusted conditional logistic regression models. In Nchelenge District, 8/62 (12, 9) cases and 59/62 (95.2%) controls had received at least one dose of OCV, yielding a matched OR of 0.021 (matched OR = 0.021, 95% CI: 0.003 - 0.154; P < 0.001) corresponding to OR-derived of 97.9%. The age- and sex-adjusted conditional logistic regression model for Nchelenge was (Adjusted [AOR] 0.018; 95% CI: 0.002 - 0.136; P < 0.001), yielding an adjusted OR-derived VE of 98.2%. Chiengi District, on the other hand, had 7/46 (15.2%) cases and 41/46 (89.1%) controls vaccinated, with a matched OR of 0.032 (matched OR = 0.032; 95% CI: 0.0004 - 0.236; P <0.001), corresponding to an OR-derived VE of 96.8%. An age- and sex-adjusted conditional logistic regression model for Chiengi district was (Adjusted [aOR] = 0.029; 95% CI: 0.004 - 0.240; p = 0.063). Thus, giving an adjusted OR-driven VE of 97.1%. Mwansabombwe district had about 5/19 (26.3%) cases and 13/19 (68.4%) controls vaccinated, yielding a matched OR of 0.165 (mOR = 0.165; 95% CI: 0.00317 - 0.8075; p = 0.0093) and a corresponding VE was 83.5%, age and sex-adjusted conditional logistic regression model for Chiengi district was (Adjusted [aOR] = 0.105; 95% CI: 0.025398-0.3375; P <0.0295) yielding a adjusted OR-driven of 89.5%.

Combined conditional logistic regression on cholera incidence and vaccination status between cases and controls: the analysis summarised in Table 2 revealed that vaccination with OCV was significantly associated with lower odds of cholera. In matched analysis, vaccinated participants across all three study districts had approximately 97.9% lower odds of cholera than unvaccinated participants (mOR = 0.021; 95% CI: 0.0052 - 0.0854). After adjusting for age and sex, the association remained unchanged (Adjusted [aOR] = 0.019; 95% CI: 0.007 - 0.022; P < 0.003). Yielding a corresponding adjusted VE of approximately 98.1%. The confidence interval is wider in the adjusted model than in the crude matched model; this is because of the strong relationship between vaccination status and water exposures. Consequently, the null hypothesis, which posited no significant difference in cholera incidence between vaccinated and unvaccinated individuals, was rejected at the 95% confidence level, as the p-value of 0.003 was less than the 0.05 significance threshold. These findings demonstrate a statistically significant association between vaccination status and protection against cholera, confirming that vaccinated individuals were substantially less likely to develop the disease compared to those who remained unvaccinated. Out of 127 cases that were included in the study, about 48 (37.9%) were laboratory-confirmed, about 5 (3.9%) were confirmed by PCR, and 43 (34.0%) by RDT, while 79 (62.1%) were clinically confirmed without laboratory confirmation. A sensitivity analysis was conducted to assess the robustness of the association between OCV vaccination and cholera. In the sensitivity analysis, eight pairs were excluded, containing records with inconsistencies between vaccination status reported and dose specification information; the association remained unchanged between vaccination with OCV and cholera ( [mOR] = 0.0217; 95% CI: 0.0054 - 0.0882; P <0.001), corresponding to OR-driven VE of 97.8, and the findings were consistent with the primary analysis including all eligible cases.

Comparing the OCV dosage between cases and controls in Nchelenge, Mwasabombwe, and Chiengi in 2021: a conditional logistic regression model was computed to assess the effectiveness of a single-dose oral cholera vaccine compared to a two-dose oral cholera vaccine in the prevention of cholera. Hypothesis: H0: EOR = 1 versus H1: EOR > 1. Where EOR = exposure odds ratio as defined below;

The analysis in Table 2 compared the district-specific effectiveness of a single dose to a two-dose oral cholera vaccine using a matched and adjusted exposure odds ratio, where dose-one was considered incomplete vaccination and dose-two as complete vaccination. Analysis demonstrated a consistently strong protective association between vaccination with a two-dose OCV and cholera in all three study districts. The protective association was consistent across Nchelenge, Chiengi, and Mwasabombwe districts; the wide confidence intervals across the three districts were due to the small number of informative discordant pairs.

In Nchelenge district, completion of two doses was strongly associated with approximately 98% lower odds of cholera (matched OR = 0.019; 95% CI: 0.005-0.112; P <0.001), yielding a matched OR-derived VE of 98.1%. The 95% confidence interval ranged from 0.005 - 0.112, which doesn’t contain the null value of one, and the p-value was <0.001, indicating the statistical significance of the finding; the null hypothesis was rejected, leading to the conclusion that the completion of the two-dose regimen was superior in offering protection against cholera in Nchelenge district compared to one-dose completion. It was further noted that sparse data may have compromised the validity of this finding, as some cells contained counts of less than five.

In Chiengi, completion of two-dose OCV was also associated with 97.6% lower odds of cholera (matched OR = 0.024; 95% CI: 0.007-0.165; P < 0.001), yielding a matched OR-derived VE of 97.6%. The 95% confidence interval ranged from 0.007 - 0.165, which doesn’t contain the null value of one, and the p-value was <0.001, indicating the statistical significance of the finding; the null hypothesis was rejected, leading to the conclusion that completion of the two-dose regimen was superior in offering protection against cholera in Chiengi district compared to a single-dose OCV. After adjusting for age and sex, the association for Chiengi district remained strong (adjusted [aOR] = 0.027; 95% CI: 0.0036 - 0.200; P <0.001), corresponding to an adjusted OR-derived VE of 97.3%. It was further noted that sparse data may have compromised the validity of this finding, as some cells contained counts of less than five.

Mwansabombwe as well demonstrated a strong association of 94.4% lower odds of cholera for the participants who completed a two-dose OCV (matched 0.056; 95% CI: 0.0013 - 0.352; P < 0.001). The 95% confidence interval ranged from 0.0013 - 0.352, which doesn’t contain the null value of one, and the p-value was <0.001, indicating the statistical significance of the finding; the null hypothesis was rejected, leading to the conclusion that completion of the two-dose regimen was superior in offering protection against cholera in Mwasabombwe District. It was further noted that sparse data may have compromised the validity of this finding, as some cells contained counts of less than five.

Combined conditional logistic regression analysis of single OCV dose completion vs. two-dose completion: the combined matched and adjusted odds ratio analysis shown in Table 2 revealed a statistically significant association between the completion of the two-dose schedule and lower odds of cholera for cases and controls. The 95% confidence interval ranged from 0.00501 to 0.1196. The completion of the two-dose OCV schedule was strongly associated with 98.3% lower odds of cholera than receipt of a single dose (matched OR = 0.0286; 95% CI: 0.00501 - 0.1196; P = 0.005). After adjusting for age and sex, the association remained strong (Adjusted [AoR] = 0.013; 95% CI: 0.001 - 0.265; P = 0.005), corresponding to an estimated adjusted OR-derived VE of 98.3%. The two-dose sensitivity analysis was also conducted, which indicated 98.3% lower odds of cholera with two-dose vaccination completion (Sensitivity OR = 0.017; 95% CI: 0.002 - 0.064; P <0.001). This finding was statistically significant at the 0.05 level, leading to rejection of the null hypothesis in favour of the alternative hypothesis. Consequently, the evidence demonstrates that the two-dose regimen offers superior protection against cholera compared to a single dose. It is noteworthy that the district-level analyses' age-and-sex-adjusted logistic regression model for Nchelenge and Mwasabombwe districts was not estimable because of sparse data/perfect or near-perfect separation. Almost all informative matched pairs strongly favoured two-dose vaccination, leaving insufficient discordant observations to estimate stable adjusted coefficients. The primary adjusted analysis indicates that participants who completed two-dose OCV had approximately 98.3% lower odds of being a cholera case compared with participants who received a single OCV dose (Table 2).

Discussion     

This matched case-control study assessed the effectiveness of the oral cholera vaccine (OCV) in three endemic districts of Luapula Province, Zambia, following a 2021 mass vaccination campaign and a subsequent outbreak in 2023 [13,16]. Our findings should be considered alongside evidence from Zambia by Ferrares. A single case-control matched evaluation study that was conducted in Lusaka, Zambia, reported an adjusted VE of 88.9% (95% CI: 42.7 - 97.8). Ferreras' study also examined the findings using an alternative study design that included a negative-test and case-cohort approach, which assessed the important potential control-selection bias. The study observed an estimate in the present study of 97.9% (95% CI: 0.0007 - 0.022), which was higher than Ferreras’s. The observed difference may reflect differences in study setting, OCV vaccination coverage, transmission intensity, exposure patterns, timing of vaccination relative to the outbreak, outcome assessment, vaccine schedule, and residual differences in the study population. Therefore, the present finding should not be interpreted as evidence that OCV provides a uniformly high level of protection across all outbreaks [22,23].

The proportion of controls reported to have received OCV was 89.0%, lower than the administrative vaccination coverage reported in the study districts. Nchelenge administrative OCV coverage was 96%, while Chiengi was 94%, and Mwansabombwe had 94% administrative OCV coverage in the August 2021 campaign. The findings show that receiving at least one dose of OCV provides significant protection against cholera, with an estimated OR-driven overall vaccine effectiveness (VE) of 98.1% among vaccinated individuals compared with unvaccinated counterparts after adjusting for age and sex. The association was consistent across Nchelenge, Chiengi, and Mwansabombwe. This aligns with the World Health Organization’s position that OCVs are effective tools for cholera control [3]. Notably, the study revealed a stark disparity in vaccination coverage between cases and controls; over 84.3% of cases were unvaccinated, whereas 89.1% of controls had received at least one dose of OCV. This pattern strongly suggests that the vaccine conferred direct protection to recipients, and the outbreak likely occurred due to gaps in population immunity rather than vaccine failure [24]. A critical finding pertains to the comparison of single-dose versus two-dose regimens. While the single-dose regimen was associated with a statistically significant risk reduction, the two-dose regimen demonstrated superior effectiveness.

The combined conditional logistic regression analysis revealed that the receipt of a two-dose OCV was strongly associated with lower odds of cholera. Among the participants who initiated the vaccination, completion of a two-dose OCV schedule was strongly associated with lower odds of cholera, yielding an OR-derived vaccine effectiveness of 98.7% after adjusting for age and sex. This is in line with a matched case-control study by [23], which evaluated two doses of Euvicho-plus during the 2017/18 cholera outbreak in Lusaka, Zambia [23]. This study in particular provided a benchmark because it evaluated OCV effectiveness using a matched case-control design. The finding was also consistent with global evidence indicating that while a single dose offers short-term protection, the two-dose schedule provides more durable and robust immunity [11,25,26]. The World Health Organization’s recent shift to a single-dose strategy due to global supply constraints may therefore have implications for long-term outbreak prevention in endemic settings [6].

Furthermore, the study identified significant environmental risk factors, notably the differences in reported drinking-water sources between cases and controls, highlighting that despite vaccination efforts, ongoing exposure to environmental and wash-related factors perpetuates transmission. These findings were consistent with OCV effectiveness among fishermen living around Lake Chilwa in Malawi; the study highlighted the importance of considering local transmission conditions and characteristics of the population when interpreting OCV effectiveness. This underscores a key limitation of relying solely on vaccination; OCVs are a crucial complement to, but not a substitute for, basic water, sanitation, and hygiene (WASH) interventions. The persistence of cholera in the study districts, despite achieving herd immunity thresholds, can thus be attributed to the convergence of incomplete vaccination (single-dose coverage) and continued high-risk environmental exposures [23].

Strengths and limitations of the study: This study used a matched case-control design with a 1:1 ratio, providing a robust assessment of oral cholera vaccine effectiveness in a real-world outbreak setting. The study achieved a 100% response rate and included 254 participants from three districts and six health facilities, enhancing representativeness. The study compared single-dose and two-dose regimens, providing critical evidence to inform vaccination policy in resource-limited settings. Limitations were that not all cases were confirmed by laboratory confirmation. Only 5 (3.9%) cases were PCR-confirmed, and 43 (34.0%) were confirmed using rapid diagnostic tests (RDTs), while 79 (62.1) were clinically confirmed. Clinical case definition and surveillance case definition may have lower specificity during an outbreak; therefore, some misclassification is possible and could attenuate the observed association between vaccination and cholera. The sensitivity analysis, however, was conducted that produced 97.8% lower odds of cholera (95% CI: 0.005 - 0, 0882; P < 0.001), which was consistent with 98.1 lower odds of cholera (95% CI: 0.007 - 0.022; P < 0.003) in the primary matched analysis, proving the important influence of clinical case classification on the study findings. Findings may not be generalizable beyond the 2023 outbreak or to other provinces in Zambia.

Conclusion     

The oral cholera vaccine remains highly effective in preventing cholera in Luapula Province. However, the sustained occurrence of outbreaks is likely driven by a combination of suboptimal two-dose coverage and persistent environmental contamination. To enhance cholera control, Zambia must prioritise the completion of the two-dose OCV schedule and strengthen integrated WASH interventions, particularly ensuring access to safe drinking water [23]. Future research should focus on long-term vaccine durability and strategies to improve vaccine uptake in hard-to-reach endemic communities.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Mulenga Cephas and Claude Mumba conceptualised the study. Mulenga Cephas, Howard Mushisha, Margaret Chishimba, Claude Mumba, and Alan Kachuka developed the methodology. Mulenga Cephas, Kaole Bwalya, Mulenga Chituma, and Chileleko Siakabanze curated the data. Mulenga Cephas, Claude Mumba, and Chileleko Siakabanze performed the formal analysis. Chileleko Siakabanze, Howard Mushisha, and Kaole Bwalya conducted the investigation. Mulenga Cephas and Claude Mumba handled visualisation. Mulenga Cephas and Claude Mumba wrote the original draft. Mulenga Cephas, Alan Kachuka, Howard Mushisha, Margaret Chishimba, Kaole Bwalya, Mulenga Chituma, and Claude Mumba reviewed and edited the manuscript. Claude Mumba and Alan Kachuka supervised the study. All authors read and approved the final manuscript.

Acknowledgments     

The authors acknowledge the Ministry of Health, Luapula Province Provincial Health Office, and the District Health Offices of Nchelenge, Mwansabombwe, and Chiengi for granting permission to conduct this study and access to routine health facility medical records. We are grateful to the management and staff of Kashikishi Hospital Affiliated Health Centre, Nchelenge Urban Clinic, Kalambwe Rural Health Centre, Puta Rural Health Centre, Mbereshi Rural Health Centre, and Mukamba Rural Health Centre for their support during data collection. Special thanks go to the Health Information Management System (HIMS) teams for their assistance with data management and reporting. We also extend our appreciation to the University of Lusaka Research Ethics Committee for reviewing and approving the study protocol, and to the National Health Research Training Institute (NHRTI) for their institutional support. We also thank Miss Yilan Gao for technical support provided on reviewing and writing. Finally, we thank all study participants whose medical records contributed to this research.

Tables and figures     

Table 1: distribution of oral cholera vaccine and demographic characteristics of cases and controls in a matched control study conducted in Luapula Province, Zambia (N = 254)
Table 2: conditional logistic regression analysis of oral cholera vaccine effectiveness, associated risk factors, district-specific and combined matched Odds Ratio and Adjusted Odds Ratio among cases and controls (N = 254)
Figure 1: source of drinking water among cases and controls
Figure 2: oral cholera vaccine and dose distribution among cases and controls in Nchelenge, Chiengi, and Mansabombwe Districts of Luapula Province, Zambia

References