A vaccine trial in a flood-affected district
A district of 260,000 people has had a typhoid outbreak after flooding. A candidate conjugate vaccine is being considered. A trial in 6,000 children aged 6 months to 12 years randomised vaccine versus placebo, with 18 months of follow-up. Confirmed typhoid occurred in 12 of 3,000 vaccinated and 40 of 3,000 unvaccinated children. Attack rates were higher in children under 5. Several households moved away during the floods, and 8 percent of participants were lost to follow-up, more in the placebo arm. The manufacturer proposes introducing the vaccine nationally on the basis of this trial.
1. Calculate the vaccine efficacy from the trial data and state the confidence you can place in it, noting the effect of differential loss to follow-up.
Read the worked answer
Attack rates:
- Vaccinated: 12/3,000 = 0.40 percent.
- Unvaccinated: 40/3,000 = 1.33 percent.
- Relative risk = 0.0040/0.0133 = 0.30.
- Vaccine efficacy = (1 − RR) × 100 = 70 percent.
Interpretation: a 70 percent point estimate is meaningful, but the confidence interval matters. With these small counts, the 95 percent CI around efficacy is wide (roughly 35–87 percent), so the true effect could be substantially lower than 70 percent.
Differential loss to follow-up: 8 percent overall, but disproportionately from the placebo arm. If those lost were at higher risk (e.g., displaced, crowded shelters), their typhoid cases are missing from the placebo denominator, which inflates apparent efficacy. Conversely, if vaccinated children who moved were also high-risk, the effect could be underestimated. The direction is not automatically known, so a sensitivity analysis assuming all losses in each arm were cases or non-cases is required. Report efficacy under best-case and worst-case assumptions.
Conclusion: the trial supports a real protective effect but does not by itself justify national introduction.
2. List the additional criteria a national introduction decision requires beyond this efficacy estimate, and identify the subgroup issue.
Read the worked answer
Criteria beyond efficacy:
- Absolute burden and age targeting: efficacy in children does not automatically translate to adults; the trial enrolled only 6 months to 12 years.
- Duration of protection: 18 months of follow-up cannot show whether a booster is needed.
- Safety: local and systemic adverse events, and any serious adverse events, must be reviewed with adequate power.
- Cold chain and logistics: conjugate vaccines need consistent refrigeration across a flood-prone district.
- Cost-effectiveness and budget impact against other priorities such as water and sanitation, which address the root cause.
- Indirect effects and herd protection are not measured in an individually randomised trial.
- Local epidemiology: whether the outbreak strain matches vaccine serotypes.
Subgroup issue: attack rates were higher in under-5s, so efficacy should be reported by age band. If efficacy is lower in infants, a different schedule or timing may be needed. Do not assume the overall 70 percent applies uniformly.
Recommendation: conditional introduction in high-burden districts with a phase-in, accompanied by surveillance, a safety monitoring plan, and parallel investment in water and sanitation.