Abstract
PURPOSE: Although the Bacille Calmette-Guerin (BCG) vaccine is widely used in many developing countries, it has limited efficacy in preventing tuberculosis (TB). Of the newer TB vaccines which are currently being developed, some are intended to prevent disease in uninfected persons (pre-exposure vaccines) and others are intended to prevent the development of disease in latently infected persons (post-exposure vaccines.) We predict the impact that mass vaccination campaigns with moderate to high efficacy pre-exposure and post-exposure vaccines would have on the TB epidemic in India. We calculate the effect that these vaccines would have on the the incidence of active TB and the death rate from TB in India over twenty years. METHODS: We use mathematical models to make our predictions. Two models of TB epidemics are developed to represent the population effects of pre-exposure and post-exposure vaccines. Each model includes the various ways in which a vaccine may be fail: immunity may be conferred to only some of those vaccinated; immunity may only be partially protective to those who acquire it; immunity may wane over time. Time-dependent uncertainty and sensitivity analysis is performed using latin hypercube sampling, a stratified form of Monte Carlo sampling. This analysis enables us to include uncertainty in multiple parameters and transform this into prediction uncertainty. The sensitivity analysis is carried out over time to determine the vaccine characteristics and population parameters most important in determining the vaccines'impact at different time intervals. RESULTS: The model reveals that in general a post-exposure vaccine will have greater impact than a pre-exposure vaccine on the incidence of active TB and death from TB. For example, if the initial annual incidence of active TB is 187 cases per 100,000 persons then a post-exposure vaccine which is applied to 60-90% of the population and which has 50–80% efficacy and an average waning time of 10–30 years will reduce the annual incidence by a median of 33.6% (2.5th percentile 18.3% - 97.5th 62.8%) over 20 years. In contrast a pre-exposure vaccine applied to a population with the same initial incidence and the same efficacy, coverage rate, and waning rate will reduce the incidence by a median of 16.5% (2.5th percentile 11.6% - 97.5th 24.3%) over 20 years. A post-exposure vaccine will also be more effective in reducing the death rate from TB over 20 years. CONCLUSION: A post-exposure vaccine is likely to make the greatest impact on reducing the rate of active TB and death from TB in India.