Abstract

Challenges in developing models to estimate the impact of male circumcision on the HIV epidemic

Presented by Nicolai Lohse, Switzerland.

N. Lohse¹, C. Hankins¹, J. Hargrove², B. Williams³, WHO/UNAIDS Male Circumcision Working Group

¹Joint United Nations Programme on HIV/AIDS (UNAIDS), Evidence, Monitoring, and Policy, Geneva, Switzerland, ²South African Centre for Epidemiological Modelling and Analysis (SACEMA), Stellenbosch, South Africa, ³World Health Organization (WHO), Geneva, Switzerland

Issues: Male circumcision (MC) reduces female-to-male HIV transmission by 60%. High HIV incidence, low male circumcision prevalence countries are considering introducing or expanding services. Mathematical models estimating the future impact of different scale-up strategies on the HIV epidemic can guide policy makers in optimal use of resources, but assumptions and results of different models should be compared.
Description: UNAIDS, WHO, and SACEMA brought together mathematical modellers in November 2007 to compare existing models and create a peer-learning environment. The models incorporated baseline variables such as HIV incidence and prevalence, prevalence of circumcised males, and condom use; programmatic variables such as age-group targets, speed of intervention rollout, and final level of MC coverage; biological variables such as risk of male-to-female HIV transmission and transmission risk during wound healing post-surgery; and behavioural variables such as condom use and number of sex partners following circumcision. Outcomes were HIV incidence, prevalence, and number of circumcisions necessary to avert one HIV infection.
Lessons learned: The methodologically diverse models produced consistently similar predictions of outcomes and their determinants. However, models should be improved by incorporating the effects of antiretroviral drugs on HIV transmission risk, sexual mixing patterns, associations between risk behaviour and willingness to be circumcised, temporal changes in age and sex distribution, updated disease progression projections, concurrent introduction of other prevention programmes, encouraging HIV testing before MC, and different service delivery models. Challenges include a paucity of evidence on the likelihood of risk compensation following circumcision, no knowledge of the potential interactions with other novel prevention technologies and the unknown time frame for their introduction, as well as evolving survival and HIV prevalence assumptions.
Next steps: Further refinement of model assumptions and outcomes will provide input for a programme planning tool to assist decision makers in assessing the potential impact of different programmatic choices.