A new efficacy measure is developed for use in prevention trials of interventions which may affect both disease incidence and disease severity. We assign a severity score to each incident case and sum severity scores over all incident cases within each treatment group to create a burden-of-illness score for each treatment group. Efficacy is evaluated by the difference between the burden-of-illness per randomized subject in the two randomized treatment groups. Since the numbers of summands in each burden-of-illness score is a random variable, standard methods of analysis are not directly applicable. The asymptotic distribution and sampling properties of the net reduction in the burden-of-illness score are derived for trials designed to stop either after a fixed length of follow-up or after the occurrence of a fixed number of cases. We illustrate the method with data from a clinical trial of a human rotavirus vaccine.