Letter to the Editor
To the Editor:
Orroth et al1 used a simulation model to assess the population attributable fraction (PAF) of HIV infections associated with STDs and concluded that published empirical PAF estimates underestimate the “true” PAF attributable to these conditions, mainly because the published PAF estimates only consider “STD effects on HIV acquisition” but do not account for such cofactor effects on transmission. We previously published estimates of the PAF of HIV acquisition and transmission associated with STD symptoms in HIV-concordant negative and discordant sexual partners in rural Rakai district of southwestern Uganda.2 This analysis found a PAF of incident HIV associated with genital ulcer disease (GUD) of 22.4% in concordant HIV-negative couples and a PAF for GUD of 14.3% in HIV-discordant couples. Since 93.8% of couples were concordantly HIV-negative in this population, the weighted PAF associated with GUD was approximately 21.7%. Orroth et al estimated that the PAF for curable STIs (mainly ulcerative infections) in Rakai was 20%.1 Thus, our couple-based estimates are compatible with those derived from the simulation model. However, analyses based on GUD symptoms and HIV acquisition in HIV-negative individuals, which ignored transmission effects, suggested a much lower GUD PAF of 8.8%.2 Thus, the empirical Rakai data support the argument that PAFs of STD cofactor effects confined to HIV acquisition underestimate the total PAF for STD effects on both transmission and acquisition within couples.
However, we must differ with Orroth et al in their conclusion that 40% of HIV infections in Mwanza were attributable to Haemophilus ducreyi and that chancroid might account for 4.7% to 11.7% of incident HIV in Rakai.1 The authors acknowledge that chancroid was not measured in the Mwanza trial and state that it was a “significant” cause of ulcers in an urban STD clinic. In Rakai, we detected H ducreyi by multiplex PCR of GUD.2 The proportion of ulcers due to H ducreyi was 2.5%, and the PAF of HIV acquisition associated with chancroid was <1.0%, which is much lower than the model estimates. The large putative PAF of chancroid generated by the simulation model is mainly due to the extremely high assumed cofactor effect (a relative risk of 25) and unsupported assumptions about the prevalence of chancroid in the Mwanza population. We believe the model seriously exaggerates the role of H ducreyi and this could distort strategies for future STD control efforts.
For the sake of full disclosure, we must acknowledge the fact that we were originally coauthors on this paper and had contributed data from Rakai to the simulation. However, we requested that our names be removed from the manuscript because we had profound reservations about the conclusions about the role of H ducreyi as well as other aspects of the simulation.
1. Orroth KK, White RG, Korenromp EL, et al. Empirical observations underestimate the proportion of human immunodeficiency virus infections attributable to sexually transmitted diseases in the Mwanza and Rakai sexually transmitted disease treatment trials: Simulation results. Sex Transm Dis 2006; 33:536–544.
2. Gray RH, Wawer MJ, Sewankambo NK, et al. Relative risks and population attributable fraction of incident HIV associated with symptoms of sexually transmitted diseases and treatable symptomatic sexually transmitted diseases in Rakai District, Uganda. AIDS 1999; 13:2113–2123.