Article Text


Epidemiology poster session 4: Modelling
P1-S4.25 Using mathematical models to understand the causes of the ecological association seen between HIV and HSV-2 in female sex workers in Southern India
  1. K Mitchell1,
  2. P Vickerman1,
  3. M Pickles2,
  4. M Kaushik2,
  5. S Verma3,
  6. S Isac3,
  7. R Adhikary4,
  8. M Mainkar5,
  9. M Alary6,
  10. M C Boily2
  1. 1London School of Hygiene and Tropical Medicine, London, UK
  2. 2Imperial College London, London, UK
  3. 3Karnataka Health Promotion Trust, Bangalore, India
  4. 4Family Health International, New Delhi, India
  5. 5National AIDS Research Institute, Pune, India
  6. 6Centre Hospitalier Affilié Universitaire de Québec, Québec, Canada


Background Cross-sectional surveys of female sex workers (FSW) in multiple districts in Southern India show a strong ecological association between HIV and HSV-2 prevalence. Modelling was used to determine if this association is primarily due to shared behavioural risk factors (such as partner change rate) or if biological interactions between the two viruses also play a major role.

Methods Linear regression was used to find significant predictors of FSW HIV prevalence in 17 districts in Southern India (using data from Avahan, the India AIDS Initiative). Explanatory variables investigated included prevalence of different STIs in FSW and clients, mean duration of being an FSW or client, and mean reported number of clients per FSW or FSW visited per client. A deterministic HIV/HSV-2 transmission model without behavioural heterogeneity or transmission co-factors was parameterised using district-specific FSW and client behavioural data and fitted to HIV and HSV-2 prevalence data for each district by varying the per partnership HIV and HSV-2 transmission probabilities. The model was firstly fit to all districts simultaneously assuming a constant HIV and HSV-2 transmission probability to see whether district-level variations in behaviour were sufficient to explain the trends seen, and then fit separately for each district to test for trends in the estimated transmission parameters in different settings.

Results FSW HSV-2, syphilis, gonorrhoea and chlamydia prevalence, client HIV prevalence and mean number of clients per FSW per week were all correlated with FSW HIV prevalence, but only FSW HSV-2 prevalence remained a significant predictor in multivariate analysis. The dynamic model reproduced the HIV/HSV-2 association when simultaneously fit to all districts, but was a poor fit to data (Abstract P1-S4.25 figure 1). When individual transmission probabilities were fit for each district positive correlations were seen between the HSV-2 transmission probability and both the HIV transmission probability and HSV-2 prevalence, but not between the HIV transmission probability and HSV-2 prevalence or vice versa.

Abstract P1-S4.25 Figure 1


Conclusion These results suggest that differences in mean reported partner change rate or duration of commercial sex are not sufficient to explain the association between HIV and HSV-2 prevalence in FSW in Southern India. However, initial analyses do not show clear evidence for a biological interaction. More detailed models will be used to further investigate the association.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.