Estimating the contribution of key populations towards the spread of HIV in Dakar, Senegal

Abstract

Introduction: Key populations including female sex workers (FSW) and men who have sex with men (MSM) bear a disproportionate burden of HIV. However, the role of focusing prevention efforts on these groups for reducing a country's HIV epidemic is debated. We estimate the extent to which HIV transmission among FSW and MSM contributes to overall HIV transmission in Dakar, Senegal, using a dynamic assessment of the population attributable fraction (PAF).

Methods: A dynamic transmission model of HIV among FSW, their clients, MSM and the lower-risk adult population was parameterized and calibrated within a Bayesian framework using setting-specific demographic, behavioural, HIV epidemiological and antiretroviral treatment (ART) coverage data for 1985 to 2015. We used the model to estimate the 10-year PAF of commercial sex between FSW and their clients, and sex between men, to overall HIV transmission (defined as the percentage of new infections prevented when these modes of transmission are removed). In addition, we estimated the prevention benefits associated with historical increases in condom use and ART uptake, and impact of further increases in prevention and treatment.

Results: The model projections suggest that unprotected sex between men contributed to 42% (2.5 to 97.5th percentile range 24 to 59%) of transmissions between 1995 and 2005, increasing to 64% (37 to 79%) from 2015 to 2025. The 10-year PAF of commercial sex is smaller, diminishing from 21% (7 to 39%) in 1995 to 14% (5 to 35%) in 2015. Without ART, 49% (32 to 71%) more HIV infections would have occurred since 2000, when ART was initiated, whereas without condom use since 1985, 67% (27 to 179%) more HIV infections would have occurred, and the overall HIV prevalence would have been 60% (29 to 211%) greater than what it is now. Further large decreases in HIV incidence (68%) can be achieved by scaling up ART in MSM to 74% coverage and reducing their susceptibility to HIV by two-thirds through any prevention modality.

Conclusions: Unprotected sex between men may be an important contributor to HIV transmission in Dakar, due to suboptimal coverage of evidence-informed interventions. Although existing interventions have effectively reduced HIV transmission among adults, it is crucial that further strategies address the unmet need among MSM.

Keywords: HIV; clients; condom use; female sex workers; key populations; men who have sex with men; population attributable fraction.

Figure 1

Model schematic illustrating the (a) movement of individuals in and out of different sub‐populations (male and female low‐risk, clients, female sex workers (FSW), younger and older men who have sex with men (MSM), (b) stratification of the population with respect to HIV infection and (c) sexual interactions which can result in HIV transmission among female, male, FSW, their clients and MSM. Blue arrow in Figure 1c shows commercial sex and all other arrows show sex with main and casual partners.

Figure 2

Modelled condom use trends for (a) female sex workers (FSW) and (b) men who have sex with men (MSM). (a) FSW condom use with commercial partners for vaginal intercourse. Assume condom use for anal intercourse is half that of vaginal intercourse for all years. (b) MSM condom use with regular and casual male partners. We assume some bias in reporting so all rates have been multiplied by a bias factor of 0.7 to 1.0 – lower bound of 0.7 chosen to give overall lower bound of 0.5, as seen in figure.

Figure 3

A comparison of model fits with HIV prevalence estimates from 1985 to 2020 for (a) female sex workers (FSW) (b) clients of FSW, (c) younger men who have sex with men (MSM) (<30 years old), (d) older MSM (≥30 years old) and (e) female and (f) male overall adult populations. Continuous black line shows median projections from all the model fits, with dashed lines and grey shaded areas showing 95% credibility intervals. Red points and lines show data with 95% confidence intervals.

Figure 4

Ten‐year population attributable fraction (PAF) for sex between men (blue), commercial sex (red), heterosexual sex for men who have sex with men (MSM) (purple), non‐commercial sex for female sex workers (FSW) and their clients (grey), and sex between low‐risk groups (green). The PAF is estimated as the proportion of all HIV infections prevented over 10 years from 1995, 2005 or 2015 if the HIV transmission risk due to a specific type of sexual behaviour is removed. The box plots signify the uncertainty (middle line is median, limits of boxes are the 25% and 75% percentiles and whiskers are 2.5% and 97.5% percentile range) in the PAF estimates due to uncertainty in the model parameters.

Figure 5

Projections of the impact of removing existing levels of antiretroviral therapy (ART) and condom use on HIV prevalence trends for (a) female sex workers (FSW), (b) all men who have sex with men (MSM) and (c) overall general population prevalence. Figures show baseline (median with 95% credibility intervals) trends, and median trends with no effect of ART (median‐blue) or no condom use (median‐orange).

Figure 6

Effect of changes in antiretroviral therapy (ART) coverage or susceptibility to infection (due to pre‐exposure prophylaxis (PrEP) or increases in condom use) from 2017 to 2030 on overall projected HIV incidence per 1000‐person years in 2030. The plots show median and 95% credibility intervals of the projections. Scenarios included are additive on top of the previous scenarios from left to right, firstly including baseline projections, then projections with scale‐up of ART coverage in men who have sex with men (MSM) from 2017 such that 74% of MSM living with HIV are virally suppressed by 2020 (same as female sex workers (FSW)), reduce susceptibility of MSM by 33% and then 66% (similar to increase condom use or PrEP coverage in MSM), reducing susceptibility of FSW by 33%, scale‐up ART coverage in low risk such that 74% are virally suppressed by 2020 and reduce susceptibility of low‐risk groups by 33%.