Modelling HIV transmission amongst high-risk groups to determine the extent that HIV transmission is driven by key populations and risk behaviours in Tijuana, Mexico

Background

Tijuana, Mexico, is situated on a major transit route where drugs including heroin and methamphetamine are trafficked to the U.S. It has one of the fastest growing drug-using populations in Mexico (estimated in literature at 10,000 people who inject drugs (PWID)), and sex-work is quasi-legal for female sex workers (FSW). Tijuana also has a large population of men who have sex with men (MSM), meaning there are overlapping high-risk populations who can contribute to the HIV epidemic. We used dynamic modelling to ascertain the extent that HIV transmission is driven by specific risk groups and behaviours including injecting drug use.

Methods

We developed and calibrated an overlapping dynamic HIV transmission model among PWID, FSW, clients of FSW and MSM using data from several completed and ongoing studies. We assumed HIV prevalence was 3.5% (2.3-5.1%) in 2012, 2.7% (1.2-5.2%) in 2013 and 17.3% (12.2-23.4%) in 2012 among PWID, FSW and MSM, respectively. Based on survey data we assumed the frequency of injecting decreased between 2012-2015, from 120 injections per month to 86 injections per month. We projected incident HIV infections over 1 and 10 years from 2019, and proportions attributable to different risk groups and sexual or injecting behaviours.

Results

Our model projects HIV incidence of 0.4 per 100 person-years (95% Credible Interval (95%CrI:0.1-1.0/100pyrs) in 2019. Between 2019-2020 the model projects there will be 76 (95%CrI:54–116) new HIV infections, increasing to 549 (95%CrI:373–893) over 2019-2029. Unprotected sex between men accounted for 39.5% (95%CrI:22.8-53.9%) of HIV infections over 10 years, reducing the number of infections to 332 (95%CrI:119-664) over 10 years. Eliminating transmission due to injecting drug use is projected to reduce incident infections to 56 (95%CrI:43–79) and 383 (95%CrI:283–579) over 1 and 10 years respectively, indicating that 24.3% (95%CrI:6.1–47.7%) and 27.5% (95%CrI:6.7–56.0%) of all HIV infections amongst key populations in Tijuana are attributable to needle sharing over 1 and 10 years. Amongst PWID, between 2019-2020 the model projects there will be 28 (95%CrI: 13-69), increasing to 203 (95%CrI: 90-552) infections over 2019-2029. Amongst these, needle sharing contributed 62.3% (95%CrI: 26.0-79.3%) or 68.7% (95%CrI:28.3-87.4%) of incident infections over 1 or 10 years, respectively, reducing the number of infections to 11 (95%CrI:9-16) or 64 (95%CrI:47-103) over 1 and 10 years, respectively. Removing sexual transmission between PWID only reduces the number of infections among PWID to 180 (95%CrI: 71-487) over 10 years, contributing 13.7% (95%CrI: 7.7-23.2%) of infections.

Conclusions

The majority of HIV transmission among key risk-groups in Tijuana is due to unprotected sex between MSM. However, amongst PWID allocating resources to reduce needle sharing would avert a significant number of HIV infections in Tijuana over the next decade.