Economic evaluation of HIV pre-exposure prophylaxis strategies: protocol for a methodological systematic review and quantitative synthesis

In 2016, an estimated 1.8 million people became infected with the human immunodeficiency virus (HIV) [1]. Pre-exposure prophylaxis (PrEP) refers to the use of antiretroviral medications in HIV-uninfected persons to reduce their risk of acquiring HIV and thus reduce population-level transmission [2]. Common PrEP regimens include oral tenofovir (TDF) or a combination of tenofovir and emtricitabine (FTC), and investigation into other agents and formulations is underway [3, 4]. Despite the effectiveness of PrEP, drug and clinical monitoring costs prevent access for many at-risk individuals and limit the scale of PrEP delivery and uptake required to control HIV epidemics [5‐9]. Thus, to integrate and deliver PrEP as a scalable component of an HIV prevention program, policy-makers and program implementers must decide about whether, how, and under what conditions resources are best allocated—or reallocated—to PrEP in order to achieve maximum health benefits at affordable costs [6].

Economic evaluations involve a systematic quantification of the trade-offs in costs and consequences of different strategies including single interventions, combination of interventions, or entire programs [10]. Synthesizing outcomes of economic evaluations can be challenging because there are many approaches to the evaluations—each with its strengths and limitations and often driven by data availability, the perspectives under which strategies are evaluated, and the research question and outcomes of interest [11, 12]. However, there are no guidelines for synthesizing the results of economic evaluations.

There exist several recommendations for the conduct and reporting of individual economic evaluations [11, 13‐21]. Most recommendations focus on health technology assessments [11, 13‐21]—for example, comparison of PrEP regimens. Yet a comparison of PrEP implementation strategies may extend beyond assessing the usual data and modeling decisions needed for health technology assessments. For example, an assessment of PrEP economic evaluations would involve examining whether processes such as the downstream benefits of HIV testing during eligibility screening for PrEP were included in the model [22]; how the fraction of undiagnosed HIV or antiretroviral treatment coverage in persons living with HIV influenced outcomes [23]; how potential ascertainment biases in identifying individuals deemed eligible for PrEP were addressed [24, 25]; and whether the model was calibrated to, and thus reproduced, key epidemiological features of the HIV epidemic pertinent to the strategies evaluated. An example of the latter includes the calibration of high- and lower-risk subgroups in comparisons of PrEP implementation strategies prioritized to subgroups at higher risk of HIV.

The methods, inputs, and assumptions in existing PrEP economic evaluations and their influence on outcomes have yet to be assessed with a systematic review. One systematic review summarized findings from 13 economic evaluations of PrEP that used transmission dynamic models published before 2013 [26]. The review provided a detailed and narrative synthesis of each study stratified by epidemic context (high-prevalence generalized epidemics vs. lower prevalence concentrated epidemics) and noted that PrEP strategies prioritized to key populations (such as men who have sex with men) with the highest risk of HIV exposure were most cost-effective [26]. However, the review was restricted to transmission dynamic models and did not describe the methodological strengths and limitations nor how assumptions may have influenced the results [26]. Another systematic review of pre-2014 studies identified only two economic evaluations and briefly summarized their main results [27]. In 2016, an expert review described in narrative detail the methods and findings from the seven most recent (published after 2014) cost-effectiveness studies of introducing PrEP [28]. The findings were stratified by geography, and authors outlined an in-depth discussion of how model parameters may influence outcomes—noting that conditions such as high baseline HIV incidence and PrEP adherence led to cost savings [28]. The expert review also identified key gaps, including a need to understand how real-world PrEP implementation factors influence cost-effectiveness and budgetary impact [28]. However, the two reviews did not examine every single scenario simulated in each primary study nor the various perspectives and funding systems for PrEP. Thus, the reviews could not quantitatively explore potential sources of variability in the projected outcomes [26, 28].

We aim to estimate the cost-effectiveness of all PrEP and PrEP implementation strategies evaluated to date in the published literature and systematically appraise and quantify the methodological and other sources of heterogeneity in outcomes. In the absence of guidelines for the synthesis of economic evaluations, we developed a protocol for a systematic review of economic evaluation studies for PrEP by drawing on best practices in systematic reviews and the conduct and reporting of economic evaluations.

Economic evaluations are important for healthcare policy decisions in an era of financial constraints. Thus, policy decisions could be best informed by a clear understanding of how heterogeneity in methods, data and structural assumptions, and implementation conditions influence the outcome of interest. In the case of HIV prevention, few systematic reviews of economic evaluations have been conducted [26], and even fewer have attempted to quantify potential sources of variability in the reported outcomes [26]. We developed a protocol for a systematic review of economic evaluations of HIV PrEP strategies that aims to assess the methodological quality of economic evaluations of PrEP, quantify sources of variability in projected cost-effectiveness results, and identify key knowledge gaps.

By 2016, two of eight clinical guidelines and position statements for PrEP mentioned the importance of considering cost when generating recommendations [32, 33]. As guidelines begin to utilize economic evaluations [28, 34], a more nuanced and quantitative synthesis of the sources of variability in cost-effectiveness ratios could better equip decision-makers to determine the generalizability of existing economic evaluations to their local context or to call for specific economic evaluations to meet their needs. To aid this process, our dissemination platform will enable end-users to intuit the key implications of the findings by selecting the features that are relevant to their needs. For example, some policy-makers may only wish to use estimates that capture onward transmission (i.e., methods that used transmission dynamic models), or health providers may wish to draw upon estimates based on high PrEP adherence if that is most relevant to their patient population. However, our synthesis will not include an assessment of the strength of the evidence as the process includes a comprehensive and joint review of the empirical (i.e., effectiveness) and modeled evidence (e.g., cost-effectiveness) and is best suited to clinical and public health guidelines [28, 34].

Systematic reviews of economic evaluation and of mathematical modeling studies are emerging in the published literature [35]. However, we identified several challenges in developing a high-quality systematic review and quantitative synthesis of economic evaluations. In the absence of recommendations for the synthesis of economic evaluations, there remains a lack of consensus surrounding the best instrument for assessing the quality of economic evaluation as part of a systematic review—especially given the heterogeneity in the types of models that can be appropriately and justifiably used. To address this gap, we will leverage CHEERS [15], the Second Panel of Cost-effectiveness in Health and Medicine [14], the Phillips checklist [11], and the ISPOR recommendations for health economic evaluations using various types of models, including transmission dynamic models [16‐21]. Second, the prior reviews suggest substantial heterogeneity in clinical assumptions and methodological features across studies and PrEP interventions [26, 28]. We will turn the challenge into an opportunity by performing a quantitative sensitivity analyses that stratify outcomes according to variables anticipated to influence cost-effectiveness, such as subgroups, baseline epidemic context, study designs, and type of outcomes. Formal meta-analyses using measures of statistical heterogeneity are not validated for use in outcomes projected from simulation models. Thus, we will apply traditional, non-parametric approaches to sensitivity analyses developed for within-model examination and to our between-model analyses [31]. Understanding the sources of variability in cost-effectiveness outcomes will provide a more robust basis for decision-making and generalizability.

Deliberations regarding public, governmental, or insurance provider funding for PrEP are currently underway in several countries [36‐40]. Findings from this systematic review can be used across jurisdictions to inform decisions on adoption and integration of PrEP implementation strategies into public health programs across epidemiologic and health system contexts. The review can also help identify key knowledge gaps to help prioritize future modeling studies, ensuring that modeling efforts are themselves optimized to generate new knowledge and to meet the needs of decision-makers, and in turn, to inform the effective and efficient delivery of PrEP as part of the HIV prevention response.