Epidemiological research in the HIV field

Abstract

In 1988, AIDS published the first issue of ‘a year in review’. Nearly 25 years later, the epidemiology section makes very interesting ‘re-reading’ [1]. By the end of the 1980s, we had a fairly good understanding of the mechanisms underlying the spread of HIV. As Michael Adler and Peter Piot pointed out, progress in epidemiological research was slowing down and we entered a period of ‘fine tuning of our knowledge’. They identified a number of gaps in our understanding. These included mechanisms of transmission, the natural history of HIV infection in adults and in children, the discrepancy of the HIV epidemics in the heterosexual populations in Europe and North America and Africa, the efficiency and risk factors of mother-to-child transmission of HIV, and the ‘incapacity and/or unwillingness’ of societies to deal rationally with drug use in the face of fulminant HIV epidemics among injecting drug users. Although this latter challenge still is topical in 2012, it is now difficult to imagine that there was a time when it was thought that ‘control of perinatal transmission will only be achieved through effective programs aimed at decreasing the spread of infection to women’. Yet, there was some hope already for a treatment that would prevent progression to AIDS, although it was going to be expensive and pose major challenges for health services. We undoubtedly have made a lot of progress since 1988. We have come to realize that ‘high-risk sexual behavior’ is much more complex than just ‘high numbers of lifetime sex partners’. Much more attention is now being paid to contextual factors that shape sexual behavior. We have a better understanding of factors that influence the efficiency of transmission of HIV. Numerous randomized trials have been conducted to test the effectiveness of interventions that have been developed based on findings from observational epidemiological and behavioral research. The most spectacular trials in terms of results were the ones testing the protective effect of male circumcision and the trials of the use of antiretroviral compounds in preventing mother-to-child transmission and sexual transmission. Trials of behavioral interventions have, by and large, given disappointing results. This has led to discussions about the suitability of community randomized controlled trials for evaluating the effectiveness of nonbiomedical prevention interventions [2]. The good news, however, is that in many parts of sub-Saharan Africa, HIV prevalence has been decreasing, apparently due to changes in behavior. The challenge is to try and understand what exactly happened in these populations and whether these favorable trends can be replicated elsewhere. Classical epidemiological methods will not allow us to come to grips with the complexities of epidemic trends and mathematical modeling has become an indispensable tool to try and understand the many factors that shape HIV epidemics. What are the major challenges for epidemiological research in 2012? We urgently need a reliable, easy-to-use tool to measure incidence at a population level. We still do not fully understand why the spread of HIV has been (and still is) so different in sub-Saharan Africa compared with heterosexual populations in other parts of the world and why the incidence of HIV infection in young women in southern Africa is so high. We need to revise what we know about the efficiency of HIV transmission in populations in different parts of the world. We also do not fully understand why in some gay communities HIV incidence has declined, whereas in other communities there is no decreasing trend despite high coverage with testing and antiretroviral treatment. And last but not least, how can we evaluate complex behavioral interventions in a reliable way? There is a consensus that there is no magic bullet (yet) to control HIV epidemics, and that we have to rely on combination prevention, including interventions at the structural and the individual level. Evaluating the effectiveness of combination prevention will have to rely on combination evaluation approaches using different data sources and mixed methods to build a plausible case. This will require interdisciplinary collaborations involving epidemiologists, mathematical modelers, and social scientists, and close collaboration between researchers and program implementers. Acknowledgements Conflicts of interest There are no conflicts of interest.