Statistical methods in HIV/AIDS
Dennis O. Dixon · Statistics in Medicine · 2008
The worldwide HIV/AIDS epidemic of the past several decades has stimulated a very impressive amount of methodological innovation in biostatistics, epidemiology, mathematical biology, and related disciplines. These advances have in turn contributed to the considerable progress made in surveillance, prevention strategies, and therapeutics. From time to time, there have been attempts to inventory new methods and supporting theoretical developments related to HIV/AIDS, as in Fusaro et al. 1 and Foulkes 2. Do we need an update again? Perhaps, but the breadth of research summarized in 1998 was enormous (369 references listed); tackling another decade worth of research is a daunting prospect. As an alternative, it seems appropriate to call special attention to HIV-related research supported recently by the National Institute of Allergy and Infectious Diseases. The 10 papers that follow, all of which meet this criterion, cover a range of topics. Some address new versions of familiar problems: how to analyze time-to-event data when the events are observed only within intervals and with error (Zhang and Lagakos), how to employ novel modeling approaches to carry out and interpret covariate adjustments in the evaluation of treatment effects (Tsiatis, Davidian, Zhang, and Lu; Huang, Liang, and Wu; Robins, Orellana, Hernan, and Rotnitzky), and how to establish immune response-based surrogate markers for vaccine efficacy (Gilbert, Qin, and Self). Others, however, deal with problems involving new kinds of observations, including indices of viral dynamics (Perelson and Ribeiro) and HIV viral genotype (Schumi and DeGruttola; Pond, Poon, Zarate, Smith, Little, Pillai, Ellis, Wong, Brown, Richman, and Frost; Ahn, Seillier-Moiseiwitsch, and Koch). Yet another explores the use of modeling to gain understanding of the interplay of multiple outcome measures in studying the effects of certain HIV vaccine candidates (Wick). The research presented here nicely illustrates the difficulty in labeling science as either basic or applied. While the motivating problems relate directly to challenges in discovery of preventive or therapeutic interventions for HIV, solutions involve conceptual and theoretical advances with much wider applicability. It is this synergy that has attracted gifted and productive methodology researchers to the field for more than two decades.