Article Text
Abstract
Background Deterministic models are widely used in simulating the potential effect of programmes for the prevention and treatment of HIV and other sexually transmitted infections (STIs). However, most deterministic models are frequency-dependent and do not model pair formation explicitly, which can lead to inaccuracies. We aim to quantify these inaccuracies by comparing a frequency-dependent deterministic model to a ‘gold standard’ microsimulation model of pair formation.
Methods An individual-based microsimulation model was created to represent as closely as possible the assumptions of a previously-developed deterministic model, which simulates heterosexual transmission of seven different STIs (HIV, genital herpes, syphilis, chancroid, gonorrhoea, chlamydia and trichomoniasis) as well as bacterial vaginosis and vaginal candidiasis, in the South African population. The microsimulation model was extended to simulate pair formation. For each STI, steady-state endemic prevalence levels were estimated using both models.
Results The ratio of the endemic STI prevalence in the microsimulation model to that in the deterministic model varied from 0.88 for HIV to 0.81 for genital herpes, 0.53 for chlamydia, 0.42 for trichomoniasis, 0.12 for gonorrhoea and 0.00 for both syphilis and chancroid. In contrast, the ratio was close to 1 for non-sexually transmitted infections (1.00 for vaginal candidiasis and 1.02 for bacterial vaginosis). The ratio was strongly negatively associated with the fraction of transmission occurring in the first 6 months of infection (r = –0.98).
Conclusion Frequency-dependent deterministic models of STIs tend to exaggerate the levels of transmission in the early stages of infection, because they ignore the period in which individuals remain in contact with the partner who infected them. This bias is particularly significant for non-viral STIs. Further work is required to assess whether microsimulation models of pair formation predict more accurately the effects of STI prevention and treatment programmes.
- HIV/AIDS
- mathematical model
- sexually transmitted infections