Estimating the epidemic consequences of HIV prevention gaps among key populations

Abstract

Introduction: HIV epidemic appraisals are used to characterize heterogeneity and inequities in the context of the HIV pandemic and the response. However, classic measures used in appraisals have been shown to underestimate disproportionate risks of onward transmission, particularly among key populations. In response, a growing number of modelling studies have quantified the consequences of unmet prevention and treatment needs (prevention gaps) among key populations as a transmission population attributable fraction over time (tPAF_t ). To aid its interpretation and use by programme implementers and policy makers, we outline and discuss a conceptual framework for understanding and estimating the tPAF_t via transmission modelling as a measure of onward transmission risk from HIV prevention gaps; and discuss properties of the tPAF_t .

Discussion: The distribution of onward transmission risks may be defined by who is at disproportionate risk of onward transmission, and under which conditions. The latter reflects prevention gaps, including secondary prevention via treatment: the epidemic consequences of which may be quantified by the tPAF_t . Steps to estimating the tPAF_t include parameterizing the acquisition and onward transmission risks experienced by the subgroup of interest, defining the most relevant counterfactual scenario, and articulating the time-horizon of analyses and population among whom to estimate the relative difference in cumulative transmissions; such steps could reflect programme-relevant questions about onward transmission risks. Key properties of the tPAF_t include larger onward transmission risks over longer time-horizons; seemingly mutually exclusive tPAF_t measures summing to greater than 100%; an opportunity to quantify the magnitude of disproportionate onward transmission risks with a per-capita tPAF_t ; and that estimates are conditional on what has been achieved so far in reducing prevention gaps and maintaining those conditions moving forward as the status quo.

Conclusions: The next generation of HIV epidemic appraisals has the potential to support a more specific HIV response by characterizing heterogeneity in disproportionate risks of onward transmission which are defined and conditioned on the past, current and future prevention gaps across subsets of the population.

Keywords: HIV transmission; HIV/AIDS; Mathematical model; key populations; population attributable fraction.

Figure 1

Distribution of various estimates of the tPAF_t related to sex work in Yaoundé, Cameroon [13]. The Yaoundé transmission model included FSW, clients of FSW, men who have sex with men and lower activity males and females. The model was used to generate the measures referred to in the main text. (A) is the percent of new infections acquired by FSW in the status quo scenario (not a tPAF_t measure). (B and C) are the cumulative percentage of infections that stem from prevention gaps experienced by FSW, and estimated by interrupting acquisition and transmission across all partnerships among FSW from 2019 to 2020 (B, tPAF₁) and from 2019 to 2029 (C, tPAF₁₀). (D) is the same measure as C except that condom‐use among FSW declines after 2019 such that the conditions under which the tPAF₁₀ was estimated in (C), no longer holds. Examples of different counterfactuals are shown with E‐G for the tPAF₁₀: In E, the transmission was set to zero (“turned off”) in the context of sex work alone. In F, only acquisition risks among FSW were set to zero, whereas in (G) the acquisition and transmission across all partnerships of FSW were set to zero. The tPAF₁₀ of sex work (E) was smaller than the tPAF₁₀ of acquisition risks among FSW (F), because in 2019, levels of condom use between FSW and their non‐paying partners were lower than condom use in the context of sex work [13]. tPAF_t (transmission population attributable fraction); FSW (female sex workers).

Figure 2

Illustration of base‐case and counterfactual scenarios used to estimate the transmission population attributable fraction (tPAF_t) over one and ten years [13], based on an HIV model of Yaoundé, Cameroon. The counterfactual scenario represents “turning off” or interrupting transmission among all partnerships among FSW in the year 2010, and estimating the relative difference in cumulative infections in the total population up to 2011 and 2020 to generate the tPAF₁ and tPAF₁₀, respectively, of the prevention gaps among FSW. FSW (female sex workers).