Evaluating a Dengue Vaccine Trial in a Cambodian Province
A phase III randomised controlled trial of a dengue vaccine was conducted in 12,000 children aged 9–14 years in Kampong Cham province. Half were randomly assigned to receive the vaccine and half to placebo. Over two years of active surveillance, 120 laboratory-confirmed dengue cases occurred in the vaccinated group and 300 in the placebo group. Both groups had 6,000 children at enrolment. The trial reported a vaccine efficacy of 60% with a 95% confidence interval from 51% to 68%. The provincial health director wants to know whether the vaccine should be introduced into the routine immunisation schedule, what proportion of children would need to be vaccinated to interrupt transmission if the basic reproduction number (R0) for dengue in the province is 3.0, and how to explain the confidence interval to community leaders. The director also asks about the number needed to vaccinate to prevent one dengue case over two years, and whether the trial result is precise enough to guide policy.
1. Calculate the vaccine efficacy and the number needed to vaccinate (NNV) to prevent one dengue case over two years. Show your working.
Read the worked answer
Attack rate in the vaccinated group = 120/6000 = 0.020, or 2.0%. Attack rate in the placebo group = 300/6000 = 0.050, or 5.0%. Vaccine efficacy = (placebo attack rate − vaccinated attack rate) / placebo attack rate = (0.050 − 0.020) / 0.050 = 0.030 / 0.050 = 0.60, or 60%. The absolute risk reduction is 0.050 − 0.020 = 0.030, or 3 percentage points. The number needed to vaccinate is the reciprocal of the absolute risk reduction: 1 / 0.030 = 33.3. Rounding up, about 34 children must be vaccinated to prevent one laboratory-confirmed dengue case over two years. This calculation assumes the trial result is causal and applicable to the target population; it does not include indirect effects such as herd protection, which could reduce the NNV further.
2. Interpret the 95% confidence interval of 51% to 68%. Does it support a policy decision? What does it say about precision?
Read the worked answer
The 95% confidence interval means that if the trial were repeated many times, about 95% of such intervals would contain the true vaccine efficacy. Because the interval excludes 0%, the result is statistically significant. The lower bound of 51% is still a substantial efficacy, so the trial provides good evidence that the vaccine reduces dengue. However, the interval is moderately wide: the true efficacy could be as low as 51% or as high as 68%. For policy, this precision is enough to support a decision in principle, but cost-effectiveness, budget impact, and safety must be modelled across that range. The trial was conducted in children aged 9–14, so extrapolation to younger children or adults is uncertain. The confidence interval does not capture all uncertainty, such as serotype-specific efficacy, waning immunity, or trial population differences from the province. Community leaders should be told that the vaccine is effective but not perfect, and that a range is normal in research.
3. Using R0 = 3.0 and vaccine efficacy of 60%, calculate the vaccination coverage needed for herd immunity. What does the result imply, and what factors could change it?
Read the worked answer
The herd immunity threshold is H = 1 − 1/R0 = 1 − 1/3 = 0.667, or 66.7% of the population needing effective immunity. If the vaccine efficacy against infection is 60%, the required coverage is H / VE = 0.667 / 0.60 = 1.11, or 111%. Because coverage cannot exceed 100%, a 60% efficacious vaccine alone cannot interrupt dengue transmission with R0 = 3.0. This does not mean the vaccine is useless; it can still reduce disease burden substantially. To interrupt transmission, the programme would need additional vector control, a higher-efficacy vaccine, or targeting of high-transmission groups. The calculation could change if the vaccine has indirect herd effects, if efficacy is higher against infection than against disease, if serotype interactions alter R0, or if the R0 estimate is uncertain. It also assumes random mixing; in reality, dengue transmission is spatially heterogeneous, so targeted vaccination of hotspots may be more efficient than uniform coverage.