Elsevier

The Lancet HIV

Volume 4, Issue 5, May 2017, Pages e214-e222
The Lancet HIV

Articles
PrEP for key populations in combination HIV prevention in Nairobi: a mathematical modelling study

https://doi.org/10.1016/S2352-3018(17)30021-8Get rights and content

Summary

Background

The HIV epidemic in the population of Nairobi as a whole is in decline, but a concentrated sub-epidemic persists in key populations. We aimed to identify an optimal portfolio of interventions to reduce HIV incidence for a given budget and to identify the circumstances in which pre-exposure prophylaxis (PrEP) could be used in Nairobi, Kenya.

Methods

A mathematical model was developed to represent HIV transmission in specific key populations (female sex workers, male sex workers, and men who have sex with men [MSM]) and among the wider population of Nairobi. The scale-up of existing interventions (condom promotion, antiretroviral therapy, and male circumcision) for key populations and the wider population as have occurred in Nairobi is represented. The model includes a detailed representation of a PrEP intervention and is calibrated to prevalence and incidence estimates specific to key populations and the wider population.

Findings

In the context of a declining epidemic overall but with a large sub-epidemic in MSM and male sex workers, an optimal prevention portfolio for Nairobi should focus on condom promotion for male sex workers and MSM in particular, followed by improved antiretroviral therapy retention, earlier antiretroviral therapy, and male circumcision as the budget allows. PrEP for male sex workers could enter an optimal portfolio at similar levels of spending to when earlier antiretroviral therapy is included; however, PrEP for MSM and female sex workers would be included only at much higher budgets. If PrEP for male sex workers cost as much as US$500, average annual spending on the interventions modelled would need to be less than $3·27 million for PrEP for male sex workers to be excluded from an optimal portfolio. Estimated costs per infection averted when providing PrEP to all female sex workers regardless of their risk of infection, and to high-risk female sex workers only, are $65 160 (95% credible interval [CrI] $43 520–$90 250) and $10 920 (95% CrI $4700–$51 560), respectively.

Interpretation

PrEP could be a useful contribution to combination prevention, especially for under-served key populations in Nairobi. An ongoing demonstration project will provide important information regarding practical aspects of implementing PrEP for key populations in this setting.

Funding

The Bill & Melinda Gates Foundation.

Introduction

There is optimism regarding the possibility of reducing HIV transmission to low levels,1, 2, 3 and this is the aim of Kenya's recent Prevention Revolution.4 However, despite the availability of several proven efficacy interventions, limited resources raise the question of what should and should not be included in an optimal portfolio of combination prevention in a given epidemiological context.

Pre-exposure prophylaxis (PrEP) works when used,5, 6 but PrEP is expensive because of the cost of delivery and monitoring. Therefore, PrEP needs to be prioritised for individuals at high risk of infection, in periods of high risk, and for those who will adhere well. Recent WHO guidance7 states that oral PrEP should be offered as an additional prevention choice for people at substantial risk of HIV infection.

The spread of HIV in Narobi has been well documented. The first cases of HIV in Nairobi were identified in 1984.8 Since then, the importance of key populations has been recognised,9 with retrospective analyses of saved specimens revealing that prevalence in female sex workers increased from 4% in 1981 to 65% in 1985.8 Given the high prevalence of sexually transmitted infections and almost no condom use, the initial spread of HIV within the female sex worker population was rapid and approximately half of uninfected female sex workers followed up in a cohort study acquired HIV within a year (1985–86).10 A cohort study11 in female sex workers showed that prevalence peaked at 81% in the late 1980s and subsequently declined. Concomitantly, transmission became established in the wider population.12

The HIV epidemic in the population of Nairobi as a whole is in decline, but a concentrated sub-epidemic persists in key populations. Men who have sex with men (MSM) have been receiving increasing attention as a key population,13, 14 and within the Nairobi MSM population, an exceptionally high incidence of HIV of approximately ten per 100 person-years has been observed in male sex workers.15, 16 Among female sex workers, although large declines in incidence have been observed, incidence remained at about two per 100 person-years overall in the period 2008–11.17

Research in context

Evidence before this study

We searched PubMed for cost-effectiveness analyses of pre-exposure prophylaxis (PrEP) published before July 7, 2016, with the terms “HIV” AND “Africa” and “PrEP” and “cost”, with no language restrictions. Much attention has focused on the potential cost-effectiveness of PrEP per se, and several studies compare the cost-effectiveness of PrEP and earlier antiretroviral therapy. A 2016 modelling study looked at optimal use of existing and future prevention technologies, including PrEP, for South Africa and found that scaling up of existing interventions (ie, male circumcision and early antiretroviral therapy) followed by intravaginal rings targeted to sex workers and vaccines, if and when they become available, was the most cost-effective way to reduce HIV transmission. A 2014 modelling study illustrated the additional effect which can be achieved for the same level of spending by means of a geographically prioritised approach, by exploring optimal allocation of resources for combination prevention including PrEP across each of Kenya's counties and largest cities. Here, we focus on combination prevention in Nairobi, specifically, providing a high-resolution representation and analysis of the local epidemiological context in Nairobi. Before this study, a comprehensive review was done to identify all relevant epidemiological, behavioural, and programme data to inform model structure and parameterise the model. We searched PubMed for studies published up to Sept 30, 2013, with the following terms: (“HIV”[MeSH Terms] OR “HIV”[All Fields]) AND (“Nairobi”[All Fields]). The titles and abstracts of all identified papers were screened for potential relevance. Additionally, relevant reports and grey literature were identified and included as appropriate.

Added value of this study

We provide a detailed representation of the HIV epidemic and existing interventions in female sex workers, MSM, male sex workers, and the wider population of Nairobi to identify optimal allocation of resources for combination prevention. We use realistic estimates of the cost of delivering PrEP to key populations in this setting, informed by a primary costing study. We indicate the level of spending required for PrEP for key populations to be included in an optimal portfolio of interventions.

Implications of all the available evidence

These analyses indicate that an optimal prevention portfolio for Nairobi should focus initially on condom promotion in MSM and male sex workers in particular, followed by improved antiretroviral therapy retention, earlier antiretroviral therapy, male circumcision, and PrEP, as the budget allows. Furthermore, male sex workers were found to be a priority population for a PrEP intervention in this setting.

Demonstration of clinical efficacy is not sufficient in itself for an intervention such as PrEP to be implemented and included in combination prevention;18 practical considerations for implementation must follow confirmation of efficacy. Demonstration projects provide important information on how to deliver PrEP effectively outside clinical trial settings. Whether PrEP can compete for resources against other prevention options is an important question (ie, whether PrEP is included in an optimally allocated budget, as opposed to defining whether PrEP could be cost-effective based on an arbitrary cost-effectiveness threshold). If PrEP cannot compete for financial resources, then the fact that PrEP works if used and could even be cost-effective might be irrelevant for optimal allocation of limited resources in the context of combination prevention.

In addition to improving and expanding existing interventions, PrEP provided to key populations at highest risk of acquisition could be a useful new intervention for key populations. We developed a mathematical model to find the optimal portfolio of existing and new prevention interventions to reduce incidence for a given budget and assessed the conditions under which PrEP could be included in an optimally allocated budget.

Section snippets

Model design

We developed a compartmental deterministic model of population-level HIV transmission. We stratified the model according to behavioural risk group, sex, and circumcision status, as well as PrEP use. We estimated rates of progression between CD4 cell count categories using observational data.19 Antiretroviral therapy can be initiated from each stage of infection. Assumptions regarding survival and dropout varied depending on whether antiretroviral therapy was initiated early at a CD4 count of

Results

The model is calibrated to the HIV epidemic in specific key populations and in the wider population in Nairobi (figure 1). The steep decline in prevalence in women and men is captured, with prevalence peaking around the late 1990s. Among female sex workers, incidence peaked rapidly and subsequently decreased. Incidence is projected to continue decreasing and stabilise at a relatively low-level equilibrium (figure 1). HIV prevalence in female sex workers increased sharply in the 1980s, reaching

Discussion

PrEP should be prioritised for the most at risk and underserved populations of male sex workers, and then MSM, and finally female sex workers, as financial resources allow. These patterns reflect HIV incidence in each of these population groups. Exceptionally high incidence of 10 per 100 person-years has been observed in male sex workers in Nairobi and is partly attributable to low levels of condom use,15, 16 whereas in female sex workers, a lower incidence of 2·2 per 100 person-years has been

References (32)

  • Guideline on when to start antiretrovrial therapy and on pre-exposure prophylaxis for HIV

  • P Piot et al.

    Retrospective seroepidemiology of AIDS virus infection in Nairobi populations

    J Infect Dis

    (1987)
  • JK Kreiss et al.

    AIDS virus infection in Nairobi prostitutes. Spread of the epidemic to East Africa

    N Engl J Med

    (1986)
  • FA Plummer et al.

    Cofactors in male-female sexual transmission of human immunodeficiency virus type 1

    J Infect Dis

    (1991)
  • J Kimani et al.

    Reduced rates of HIV acquisition during unprotected sex by Kenyan female sex workers predating population declines in HIV prevalence

    AIDS

    (2008)
  • Kenya HIV prevention response and modes of transmission analysis

  • Cited by (32)

    • PrEP uptake and adherence in relation to HIV-1 incidence among Kenyan men who have sex with men

      2020, EClinicalMedicine
      Citation Excerpt :

      Counties with an estimated high or medium HIV-1 incidence rate in the general population and those with a high or medium number of KP (including MSM) are prioritized for PrEP roll-out [7]. Mathematical models using local epidemiological and behavioral data provide insights on which approach would yield the greatest reduction of HIV-1 transmission [8–10]. Results from a modelling study using data from Nairobi suggested that, in a setting with limited resources, decreasing overall HIV-1 infections could best be achieved when male sex workers were prioritized for PrEP and ART initiation, followed by prioritizing MSM not reporting sex work and FSW [10].

    • Cost-effectiveness of HIV Prevention Interventions in Sub-Saharan Africa: A Systematic Review

      2019, EClinicalMedicine
      Citation Excerpt :

      Out of the 146 articles, 60 met the full inclusion criteria (Fig. 1). These 60 peer-reviewed studies provided cost-effectiveness results for the following HIV prevention interventions: 14 studies on VMMC, 13 studies on PrEP, five studies on TasP, 15 studies on PMTCT, nine studies on other biomedical interventions, one study on behaviour change, and three studies on structural interventions [15–74]. Among PMTCT studies, 14 considered Prong III strategies, while one focused on Prong II.

    View all citing articles on Scopus
    View full text