Acquired Immune Deficiency Syndrome Predictions for England and Wales (1992–97): Sensitivity Analysis, Information, Decision

The backcalculation method has been widely used to reconstruct the human immunodeficiency virus (HIV) infection process and to produce short‐term predictions for acquired immune deficiency syndrome (AIDS), and it is the central approach on which the latest (1993) projection results for England and Wales are based. Typically, specific assumptions on the incubation time from HIV infection to AIDS and on the structure (parametric or semiparametric) of the infection curve are made; use of knowledge on the observed AIDS cases allows an estimation of the 'parameters' of the infection curve, reconstruction of the past and current HIV infection process and inference about future AIDS incidence. Results are often crucially dependent on the specific assumptions made and, given the great uncertainty about some of the quantities involved in the estimation process, sensitivity analyses become essential. This paper explores in more detail some aspects of the latest projections which have only been hinted at in the report published by the Public Health Laboratory Service in 1993. The value of additional information on the HIV epidemic in discriminating between different, otherwise equally plausible, scenarios is demonstrated. The role of the backcalculation approach in determining whether, and how, the incubation distribution has been affected by increased uptake of pre‐AIDS prophylaxis and treatment is discussed.