Skip to main content
Download PDF

From conceptualising to modelling structural determinants and interventions in HIV transmission dynamics models: a scoping review and methodological framework for evidence-based analyses

BMC Medicine volume 22, Article number: 404 (2024) Cite this article

Abstract

Background

Including structural determinants (e.g. criminalisation, stigma, inequitable gender norms) in dynamic HIV transmission models is important to help quantify their population-level impacts and guide implementation of effective interventions that reduce the burden of HIV and inequalities thereof. However, evidence-based modelling of structural determinants is challenging partly due to a limited understanding of their causal pathways and few empirical estimates of their effects on HIV acquisition and transmission.

Methods

We conducted a scoping review of dynamic HIV transmission modelling studies that evaluated the impacts of structural determinants, published up to August 28, 2023, using Ovid Embase and Medline online databases. We appraised studies on how models represented exposure to structural determinants and causal pathways. Building on this, we developed a new methodological framework and recommendations to support the incorporation of structural determinants in transmission dynamics models and their analyses. We discuss the data and analyses that could strengthen the evidence used to inform these models.

Results

We identified 17 HIV modelling studies that represented structural determinants and/or interventions, including incarceration of people who inject drugs (number of studies [n] = 5), violence against women (n = 3), HIV stigma (n = 1), and housing instability (n = 1), among others (n = 7). Most studies (n = 10) modelled exposures dynamically. Almost half (8/17 studies) represented multiple exposure histories (e.g. current, recent, non-recent exposure). Structural determinants were often assumed to influence HIV indirectly by influencing mediators such as contact patterns, condom use, and antiretroviral therapy use. However, causal pathways’ assumptions were sometimes simple, with few mediators explicitly represented in the model, and largely based on cross-sectional associations. Although most studies calibrated models using HIV epidemiological data, less than half (7/17) also fitted or cross-validated to data on the prevalence, frequency, or effects of exposure to structural determinants.

Conclusions

Mathematical models can play a crucial role in elucidating the population-level impacts of structural determinants and interventions on HIV. We recommend the next generation of models reflect exposure to structural determinants dynamically and mechanistically, and reproduce the key causal pathways, based on longitudinal evidence of links between structural determinants, mediators, and HIV. This would improve the validity and usefulness of predictions of the impacts of structural determinants and interventions.

Peer Review reports

Background

Structural determinants of HIV are the social, economic, political, cultural, organisational, and environmental factors that shape HIV acquisition and transmission risks across individuals and populations (Table 1) [1,2,3]. Socio-ecological frameworks have been applied to understand how structural determinants influence HIV transmission dynamics among populations most vulnerable to HIV (i.e. key populations) [4, 5]. Key populations include people who inject drugs (PWID), men who have sex with men (MSM), transgender people, and female sex workers (FSW) [6]. Inequitable access to essential resources such as education, employment, and health care, coupled with the criminalisation of certain behaviours, including sex work, drug use, and same-sex relationships concentrates HIV vulnerabilities within these groups [4, 7,8,9]. This compounding effect is exacerbated by pervasive stigma, discrimination, racism, homophobia, and sexism [10].

Table 1 Definitions of key terms used in this paper
Full size table

Recognising the importance of structural determinants, the Global AIDS Strategy 2021–2026 includes the 10–10-10 targets [10]. These targets aim to reach < 10% of key populations and people living with HIV (PLHIV) experiencing stigma and discrimination, < 10% of women and key populations encountering gender-based inequalities and violence, and < 10% of countries having punitive laws and policies that limit access to HIV-related services by 2025. The global strategy commits to supporting community-led organisations to deliver 60% of HIV programmes on societal enablers (structural interventions that improve the effectiveness of HIV services) including those to reduce stigma and discrimination, support enabling legal environments, and eliminate gender-based violence [10]. However, quantitative evidence of the population-level contribution of structural determinants and the impact of structural interventions on HIV and other outcomes is sparse (although increasing), partly because these impacts are often difficult to evaluate empirically [11]. Estimating the population-level impact of structural determinants is required to inform effective policies and interventions to mitigate their impacts on HIV outcomes. It builds the evidence base on their importance and can inform resource allocation—through complementary economic evaluations—tailored to the most important epidemic drivers. Mathematical models of HIV transmission that carefully triangulate information on structural determinants can provide a means to estimate their population-level impacts and quantitatively account for uncertainty in their individual-level effects, even with sparse observed data, to generate evidence on the potential benefits of structural interventions [12]. A key benefit of these models is their ability to project non-linear dynamics, including both direct and indirect effects of structural determinants and interventions on HIV over relatively longer time horizons than statistical models when quantifying population-level impacts.

Transmission dynamic models that describe the acquisition and transmission of HIV have long been used to quantify the population-level impact of biomedical and behavioural interventions [13,14,15,16]. However, few mathematical models have so far considered structural determinants, in part due to the inherent complexity of incorporating these upstream factors, limited understanding of their causal pathways, and uncertainty in the benefits of associated interventions [11]. Unlike individual-level risk factors that directly influence HIV transmission, structural determinants influence HIV risks through multiple intervening mechanisms [2, 17]. Given the importance of structural determinants, a new generation of evidence-based mathematical models is needed to better inform public health and decision-making on ending HIV/AIDS, and to evaluate the cost-effectiveness of different intervention strategies. These models need to explicitly represent structural determinants in a way that adequately captures the patterns of exposure and their influence on individual-level HIV risks through different causal pathways, while being firmly grounded in robust empirical evidence.

The overarching objective of this paper is to develop an evidence-based methodological framework to improve the design and analysis of dynamic HIV transmission models of structural determinants. Using our experience of modelling structural determinants [4, 18,19,20,21,22,23] and a scoping review evaluating previous models that represented structural determinants of HIV, we develop recommendations for the next generation of models and data needs. Although our framework focuses on HIV, it can also be applied to other infectious diseases.

Conceptual framework: causal pathways linking structural determinants to HIV in models

Structural determinants often have diffuse effects, in that exposure to structural determinants may impact multiple outcomes, through diverse causal pathways and mediators, which will differ by structural determinant and setting (see Table 1 for definitions of key terms) [17]. Exposure to some structural determinants may also increase exposure to other structural determinants (e.g. incarceration may increase exposure to stigma), and mediators and outcomes may themselves impact future exposure to structural determinants (e.g. HIV acquisition leading to illness, loss of income, and financial hardships) [24, 25]. Transmission dynamics models allow us to reproduce these complex relationships.

To model exposure to structural determinants, we need to translate the main features of exposures into their mechanistic components. This requires identifying and defining the patterns of exposure to the structural determinants that can be modelled, based on available evidence of their prevalence and frequency in the populations and settings of interest. We then need to simulate the main causal pathways, including mediators, needed to adequately reproduce the effects of exposure on HIV outcomes (Table 1, Fig. 1). Ideally, this requires strong empirical evidence on the causal pathways, including mediators, and the magnitudes and durations of causal effects (e.g. relative risks) linking structural determinants, mediators, and HIV outcomes.

Fig. 1
figure 1

Conceptual framework illustrating the causal pathways connecting exposure to structural determinants to HIV transmission and population-level HIV outcomes, via mediators, in dynamic mathematical models. Exposure to distal structural determinants such as laws and policies and proximate structural determinants such as stigma and discrimination (e.g. homophobia, racism, sexism, transphobia) impact HIV outcomes through their effects on intermediate variables (mediators). How exposure to structural determinants may impact HIV transmission within a modelled population can be conceptualised by considering the effects of exposure to structural determinants and interventions on key parameters that determine the basic reproduction number, 𝓡0, and the force of infection, λ (i.e. HIV incidence). In a simplified model that assumes a homogeneous population and therefore random mixing patterns, these parameters include contact rates (c), transmission probabilities (β), and the duration spent virally unsuppressed among PLHIV (D). Important mediators to account for include those affecting these parameters. In a more realistic heterogeneous population and models with non-random mixing, additional complexity can be considered. The exact way in which this is modelled will differ by model. \({}^{I}\!\left/ \!{}_{N}\right.\) = the prevalence of virally unsuppressed HIV among partners of those not living with HIV

Full size image

Structural determinants may be distal or proximate (Table 1, Fig. 1) [4, 31, 32, 38]. Distal structural determinants include macro-level aggregate exposures that affect whole populations, communities, or groups, such as laws and policies, social norms, and gender inequality [4, 31, 32]. Proximate structural determinants are individual-level consequences of distal exposures, such as incarceration, discrimination, and violence [4, 31, 32]. Some researchers advocate for focusing on proximate structural determinants, as they may be more easily modified by social programmes and policies [39]. They may also be more easily measured and thus operationalised in models, and their evidence base may be stronger than for distal structural determinants [5, 7, 40].

Models need to specify and quantify how exposure to structural determinants affects HIV outcomes, based on evidence of their effects. How these effects are captured in models will depend in large part on the model structure and the choice of mediators represented. In a simplified example modelling a homogeneous population with random mixing, important parameters that determine levels of HIV transmission include the probability of HIV transmission (β) per effective contact, the average duration of transmissibility among PLHIV (D; the time spent virally unsuppressed), and the contact rates between people (c; e.g. sexual or sharing injecting partners) (Fig. 1) [26]. Changes in these parameters influence the force of infection (λ) and the basic reproduction number (𝓡0)—concepts central to transmission dynamics models (Fig. 1, Table 1). In reality, populations are not homogeneous and both population heterogeneity and mixing patterns by relevant population subgroups will impact 𝓡0 and λ and intersect with structural determinants [41, 42].

Scoping review: existing models of structural determinants and HIV

Search methods and studies identified

To develop our framework and recommendations, we conducted a scoping review of HIV transmission dynamic modelling studies to appraise previous approaches. We included studies that modelled exposures to structural determinants and/or interventions, and their mediators, and estimated their impacts on HIV acquisition and onward transmission in any population and setting. We conducted the search on August 28, 2023, for studies published since January 1, 1980, using Ovid Embase and MEDLINE online databases (Additional file 1: Text S1, Additional file 2: Tables S1 and S2). We adopted a three-way classification to characterise studies: (a) static approaches where the proportion of individuals exposed to the structural determinants and its effects on the assumed mediators and/or HIV acquisition or transmission risks were accounted for by applying fixed relative rates or probabilities to relevant model parameters influenced by the structural determinants; (b) stratification-based approaches where the modelled population could experience one level of exposure, with some movement between exposed and non-exposed states; and (c) stratification-based approaches with movement between multiple exposure history states (e.g. recent, non-recent). Our scoping review was reported using the PRISMA extension for scoping reviews (Additional file 2: Table S3) [43]. Additional details on the scoping review’s methods are provided in Additional file 1: Text S1.

We identified 17 modelling studies based on 13 models that assessed the impact of structural determinants and/or interventions on HIV (Table 2, Additional file 1: Text S2, Additional file 2: Table S4, Additional file 3: Fig. S1). Most studies modelled proximate structural determinants (number of studies [n] = 12) [4, 44,45,46,47,48,49,50,51,52,53,54] and/or structural interventions (n = 14) [4, 5, 44, 46,47,48,49,50,51,52, 54,55,56,57], primarily affecting key populations including PWID (n = 8) [5, 46,47,48, 50, 53,54,55], FSW (n = 5) [4, 49, 55,56,57], and MSM (n = 3) [5, 46, 55]. Four models of PWID were not gender-stratified [47, 50, 53, 54]. Studies were primarily published since 2015 (n = 13) [4, 44,45,46,47,48, 50,51,52,53,54,55, 58] and largely modelled settings in Western and Central Europe and North America (n = 8) [4, 5, 44,45,46,47, 49, 57] and Eastern and Southern Africa (n = 7) [4, 5, 49, 51, 52, 57, 58]. Seven studies modelled multiple settings in different regions [4, 5, 49, 53,54,55, 57], including two studies that modelled 58 and 77 countries, respectively [53, 55]. One study modelled hypothetical settings with moderate to high HIV prevalence [50].

The modelling objectives of studies were primarily to estimate the impact of structural interventions on new HIV acquisitions (n = 15) [4, 5, 44, 46,47,48,49,50,51,52, 54,55,56,57,58] or to assess the contribution of structural determinants to HIV epidemics (n = 6) [4, 45, 48, 51,52,53] (Table 2). Most studies estimated impacts by predicting the fraction of new HIV acquisitions occurring or averted under different scenarios (n = 11 [4, 5, 44, 45, 47, 49, 51, 54,55,56,57]; Table 2).

Table 2 Characteristics of HIV mathematical modelling studies identified in our scoping review
Full size table

Structural determinants and interventions examined

Exposure to proximate structural determinants included incarceration of PWID (n = 5) [46,47,48, 50, 54] and African American men (n = 2) [44, 45], client- and police-perpetrated violence against FSW (n = 2) [4, 49], intimate partner violence against women (n = 1) [52], HIV stigma (n = 1) [51], and housing instability among PWID (n = 1) [53]. Few studies modelled distal exposures (Table 2, Additional file 2: Table S4) [58]. One study modelled “positive and negative attitudes” among Kenyan youth, which reflected a combination of proximate and distal exposures (e.g. health worker confidentiality, poverty, peer influences, stigma, and more) [58]. The modelled structural interventions included reducing/eliminating incarceration of PWID (n = 5) [46,47,48, 50, 54], reducing/eliminating violence against women and FSW (n = 5) [4, 5, 49, 52, 55], community mobilisation and empowerment for FSW (n = 3) [4, 56, 57], and HIV stigma reduction [51]. Most of these modelled several interventions or delivery strategies. One study considered the impacts of achieving the UNAIDS 10–10-10 targets [55]. Another modelled structural changes, including eliminating police beatings in Ukraine [5]. Six studies, five of which modelled incarceration, also modelled scale-up of biomedical interventions such as prison- or community-based opioid agonist therapy, PrEP, or ART for prisoners or their partners [5, 44, 47, 48, 50, 54].

Representations of exposure to structural determinants

The static representation category included studies that did not explicitly represent structural determinants (e.g. as compartments; n = 7; Table 2 (a), Additional file 2: Table S4) [5, 49, 51, 55,56,57,58]. For instance, Strathdee and colleagues modelled the impact of eliminating police beatings among PWID in Ukraine by comparing the baseline to a scenario with reduced sharing of injection equipment by a factor that was informed by empirical analyses showing greater sharing frequency if ever beaten by police and assuming that the reduction in sharing was due to the elimination of beatings [5]. Studies in this category included others that represented structural determinants as parameters that influenced HIV transmission or behaviours (n = 3) [5, 51, 58], and studies using the Goals models (n = 3) [49, 55, 57].

The stratification-based representation category included studies that stratified the population into mutually exclusive compartments or states, with transitions between them, to represent one current or recent exposure history to structural determinants (n = 2; Table 2 (b), Additional file 2: Table S4) [52, 53] or that represented multiple different exposure histories (n = 8; Table 2 (c), Additional file 2: Table S4) [4, 44,45,46,47,48, 50, 54]. For example, Shannon and colleagues’ model among FSW in Canada, Kenya, and India was the first to represent several structural determinants and exposure histories dynamically (Fig. 2a provides a simplified adaption of their Vancouver model flowchart) [4]. FSW transitioned between compartments of never, recent, and non-recent physical and sexual client violence and police harassment, which differed by settings. Similarly, all studies of incarceration represented multiple exposure histories (e.g. current, recent, non-recent incarceration; Table 2, Additional file 2: Table S4) [44, 45, 47, 48, 50, 54].

Fig. 2
figure 2

Dynamically representing exposure to structural determinants and their causal pathways in HIV models, with multiple different exposures and exposure histories. a Model flowchart (adapted from Shannon et al., 2015) [4] showing how exposure to different types of violence among FSW in different work environments and their impacts on HIV were represented in their model, and b a hypothetical model flowchart based on Shannon’s approach representing how exposure to stigma among MSM in settings could be modelled. Evidence suggests that in settings where sex between men is criminalised, MSM experience more stigma [59]. Enacted stigma, such as denial of care, and anticipated stigma, such as fear of discrimination, are linked to lower and slower uptake of HIV testing and treatment [60]. These could be represented by stratifying the population based on type of stigma, and criminalisation of sex between men, with multiple exposure histories for stigma to reflect short and long-term effects of exposure on HIV risks, and interactions reflecting links between the different exposures (purple arrow, incidence rate ratio for exposure; IRR > 1)

Full size image

Dynamically representing structural determinants offers more flexibility to capture both long and short-term effects of exposure, including cumulative, gradual, waning, or lagged effects, by varying HIV risks associated with each exposure level. It also allows for the consideration of different rates of re-exposure. Granular exposure histories also facilitate a wider range of interventions to be explored. For instance, Shannon’s model differentiated the smaller impact of an intervention that reduces the incidence of violence versus an intervention that additionally removes the persisting negative effects of ever having experienced violence [4]. It showed that tackling all forms of violence would have a greater impact on HIV given high levels of co-exposures and interactions. Other structural determinants could be modelled similarly (Fig. 2b provides an example for stigma among MSM). Nevertheless, the stratification-based approach can be complex and data-intensive, making the static approach perhaps more practical for situations with sparse data, such as initial assessments. However, the stratification-based approach can also be simplified by using fewer stratifications.

Most studies represented the indirect effects of exposure to structural determinants through mediators related to sexual behaviours and HIV services access (Table 2). For instance, Shannon’s study modelled the effects of exposure to violence on HIV through lower condom use, and feedback loops between the types of violence, since recent police harassment increased exposure to recent client violence and vice versa (Fig. 2a) [4]. The most common mediators across studies were contact patterns (n = 8) [44,45,46,47,48, 50, 52, 54], the frequency or number of sexual/injecting partners (n = 9) [5, 44,45,46,47,48, 50, 52, 54], condom use (n = 6) [4, 49, 52, 56,57,58], and ART use (n = 5) [44,45,46, 51, 58]. Some studies considered upstream mediators, such as binge drinking or harm reduction services [4, 46,47,48]. Three studies among PWID (two on incarceration [47, 54] and one on housing instability [53]), modelled both the total effect of exposure (by changing the transmission probability among those exposed based on empirical estimates that implicitly captured indirect pathways involving injection drug use) and indirect effects through changes in mixing patterns (e.g. no contact between those in prison and the community; Table 2).

Use of empirical evidence

In all studies, empirical evidence was used to inform model development. The information used included the proportion of the population exposed to the structural determinants (n = 9) [44,45,46, 48,49,50,51,52, 54], rates of exposure (n = 5) [4, 44, 45, 47, 48, 50, 52, 54], durations of exposure (n = 7) [44,45,46,47, 50, 53, 54], and estimates of the effect size of exposure to structural determinants or interventions on mediators or HIV risks (n = 10) [4, 5, 44, 45, 48, 50, 52,53,54,55,56,57] (Additional file 2: Table S5). All models were calibrated to different HIV outcomes (e.g. HIV prevalence, ART coverage) (Additional file 2: Table S6). Seven studies also calibrated or cross-validated models using structural determinants data including the proportion exposed or exposure rates (n = 4) [47, 51,52,53], and the effect size of exposure on HIV or HIV prevalence or incidence stratified by exposure histories (n = 4) [47, 48, 50, 54] (Additional file 2: Table S6).

Most model assumptions on structural determinants and their effects on mediators and HIV risks were based on empirical evidence—mostly from surveillance data or cross-sectional surveys, and largely from the same settings and risk populations as those modelled (Additional file 2: Table S5). In these modelling studies, the effects of exposures on mediators and HIV risks were based on various designs, each with limitations, including cross-sectional studies, cohort studies, trials, and some systematic reviews and meta-analyses, although these mostly included cross-sectional studies and sometimes pooled data from multiple settings. Cross-sectional effect sizes may limit the strength of evidence of a causal link, due to reverse causation. Single parameters were often informed by multiple sources (Additional file 2: Table S5). Only one study (Shannon et al.) that represented structural determinants dynamically with multiple different exposure histories was informed by longitudinal data on the effects of exposure on its mediators (condom use) for all exposure histories, and only in one of the three settings modelled [4]. Data used to parameterise exposures, transitions, and effect sizes were sometimes derived from different studies and settings, meaning that estimates informing the same model were not always based on standardised exposure definitions, potentially reducing the external validity of some model findings. Few studies validated their model predictions for structural determinants against observed estimates, perhaps due to insufficient validation data [4].

Methodological framework: improving models of structural determinants and HIV

Given existing limitations, we propose a generalised framework of recommendations for modelling structural exposures and their causal pathways and discuss data needs for this next generation of models (Fig. 3, Table 3). For simplicity, we focus on deterministic compartmental models, but the framework can also be applied to individual-based models.

Fig. 3
figure 3

Methodological framework for modelling structural determinants. a Recommendations for the next generation of models focused on structural determinants and HIV, and b the future data needed to improve models of structural determinants, including the strength of quantitative evidence that could be used to inform the effects of exposures on mediators and HIV outcomes in models. SD, structural determinant

Full size image
Table 3 Recommendations for developing, analysing, and describing models of exposure to structural determinants and interventions
Full size table

Recommendations

First, models should consider dynamic and granular representations of structural determinants within the model, while being cautious not to add complexity when there is not strong evidence to support it (Fig. 3a). Models should represent the key dimensions of exposure, including exposure histories, duration, frequency, intensity, as well as co-exposures with other structural determinants and important feedback loops linking them. To connect exposure to HIV outcomes, the key causal pathways should be considered, including the mediators required to adequately capture the effects of exposure in the model.

When deciding on parameters related to structural determinants, it is important to weigh up the strengths and validity of available evidence and their relevance to the specific research question and context. Even if the model perfectly represents the mechanistic process linking structural determinants to HIV outcomes, using biased inputs, or inputs from different populations and settings, could bias model outputs [61]. Ideally, modellers should consider evidence for effect modification, cumulative effects, and interactions [62]. If parameters are uncertain and the internal validity is weak, transparently conducting detailed uncertainty and sensitivity analyses is warranted [63]. In some instances, modellers may need to decide whether to try and incorporate uncertainty in the appropriate parameter value, explore assumptions in additional scenarios, or not to model the research question at all. Attention should be paid to the external validity (i.e. generalisability and transportability) of parameters [64]. At the fitting stage, data on HIV epidemiological and intervention outcomes should be used, ideally stratified by exposure history to the structural determinant. Efforts should be made to fit or validate model predictions to the prevalence of exposure to structural determinants, and levels of mediators by exposure history, if available and relevant. Ideally, the fitting method should allow uncertainty in parameter assumptions to be reflected (e.g. using a Bayesian framework), including uncertainty in estimates related to structural determinants [26].

Finally, when conducting model analyses, the modelling scenarios, including the counterfactuals, used to assess the contribution of structural determinants to HIV incidence or to evaluate future changes due to introducing structural interventions should be clearly specified. Sensitivity analyses should be used to explore how impacts change if short-term reductions in exposure to structural determinants are not sustained long-term [63]. Data on HIV outcomes, mediators, and structural determinants not used at the fitting stage should be used to validate predictions, which can help indicate whether the model predicts the impact of structural interventions well or not [65]. Similarly, predictions from older models considering the same structural determinants could be compared to observed estimates, to identify strengths and weaknesses in their model structures and/or parameterisations that can inform newer models.

Future data needs

Ultimately, the extent of model complexity will be determined by the research question and the availability of data on structural determinants, mediators, confounders, and HIV or other outcomes (Fig. 3b). Our set of recommendations (Table 3, Fig. 3a) can help outline data issues to consider. Ideally, exposures to specific structural determinants would be consistently measured to facilitate comparisons across studies and from the same settings and populations modelled. However, currently, exposure measurements (i.e. the survey questions) can vary considerably. For example, a global systematic review among sex workers and MSM in 2017 found that studies measuring stigma used various metrics that were not necessarily developed for the populations of interest and were largely not validated [66]. Additionally, most stigma measures among MSM addressed stigma based on sexual orientation rather than behaviour, limiting the generalisability of the measures to other settings where understandings of sexual orientation and identities may differ. Additional estimates of prevalence that reflect the different exposure histories are needed. These could come from cross-sectional studies and population surveillance exploring exposure over different recall periods. Furthermore, rates of exposure from longitudinal studies would be useful to inform models. In the absence of these, or if estimates from longitudinal studies may be limited (e.g. if there is substantial loss-to-follow-up), rates could be estimated by fitting the model to good quality cross-sectional data measured at different time points.

Despite increasing recognition of the importance of structural determinants for HIV transmission, estimation of the total effect of structural determinants on HIV outcomes has generally been overlooked in epidemiological analyses, except for socioeconomic status (e.g. income, education, employment) [67,68,69]. Previously, many estimates have been based on cross-sectional studies and ecological analyses, which despite being useful, may have limited value for causal inferences given the risk for reverse causation, confounding, and ecological fallacy [70]. To improve the strength of evidence linking structural determinants, mediators, and HIV outcomes, causal analyses of longitudinal studies are needed (Fig. 3b). A challenge is the potential abundance of confounding factors that may or may not be measured, but which may need to be adjusted for [71]. Ignoring this background heterogeneity could risk biasing the contribution of the structural determinant to HIV outcomes in the model. Empirical evidence (e.g. reviews of quantitative studies) can help identify the confounders to consider and directed acyclic graphs (DAGs) can help choose which to control for [72]. Estimates from path-specific inferences such as causal mediation analyses could be used to parameterise effect sizes [73]. Mediation analyses can be used to estimate causal estimands of exposure to structural determinants, including natural direct and indirect effects, path-specific effects, controlled direct effects, and proportions mediated, using longitudinal data (Additional file 1: Text S3) [74]. To improve the validity of model predictions, effect sizes should ideally be based on the same exposure definitions and settings as the other parameters (e.g. proportions exposed, exposure rates) that inform the model.

Although it may not be possible to randomise (at the individual or cluster-level) some structural determinants (e.g. criminalisation), evidence on the causal effects and impacts of structural interventions should ideally come from randomised controlled trials (RCTs)—often considered the gold standard for causal inference analyses (Fig. 3b). For example, there have been several RCTs of individual and community-level interventions to address inequitable gender norms [75,76,77,78,79,80,81,82]. However, even with RCTs, additional analyses might be needed to identify and quantify specific causal pathways. For example, RCT data has also been used in causal mediation analyses to estimate the effects of exposure to interventions on inequitable gender norms along specific pathways [83, 84].

Given the challenges associated with obtaining causal estimates, evidence on structural determinants and causal pathways should be complemented with information from additional sources, including qualitative evidence, social theory, and inputs and involvement in the research from people with lived experience, ideally from the same or similar settings and populations as the ones modelled [85]. In our review, 11 of the studies that modelled specific settings included co-authors from those settings; however, it was generally unclear if people with lived experienced from those settings were involved in the studies. Finally, modellers should aim for transparency in reporting the strengths of evidence on model assumptions related to structural determinants and attempt to triangulate all relevant data to help identify and quantify sources of uncertainty using distributions of parameter values.

Discussion

In this paper, we introduce conceptual and methodological frameworks to assist investigations of the population-level impacts of structural determinants on HIV outcomes, underpinned by a scoping review of previous models. Simultaneously, we advocate for strengthening the empirical evidence of the effects of structural determinants and interventions on HIV outcomes—an essential foundation for developing better models and prioritising interventions.

Previous models of structural determinants and interventions include notable efforts to represent structural determinants dynamically, with particularly complex representations of violence and incarceration, which were modelled in several studies with multiple exposures, exposure histories, and additional stratifications. Our recommendations aim to build upon these to help the next generation of models represent structural determinants dynamically and mechanistically and to portray the important causal pathways and mediators to produce useful, evidence-based estimates of the impacts of structural determinants and interventions. These insights could be useful to inform policy decisions for resource allocation [86]. Further, our methodological framework supports transparency in reporting of methods and assumptions to facilitate comparisons in approaches and results across studies, which differed among the studies identified in our review.

Others have considered how to represent social and structural determinants in transmission dynamic models of infectious diseases [71, 87, 88]. Although our framework was principally developed to support the design of HIV models, our recommendations have broad applicability and can be readily extended to models of other infectious diseases that may face similar limitations. Indeed, a previous review of tuberculosis models also found few models that represented structural determinants (e.g. undernutrition, wealth), which were limited by simple exposure representations and causal pathways, an almost exclusive focus on proximate structural determinants, and a lack of evidence on the exposures from the necessary contexts [88]. More generally, we advocate a mechanistic approach with an emphasis on understanding and reproducing the key causal pathways, which is adaptable yet applicable to multiple and diverse structural determinants, mediators, and outcomes, in various contexts.

Conclusions

Increasingly, transmission dynamic models are being used to explore how exposures to structural determinants influence social and health inequalities, and how structural interventions might mitigate these impacts. Models informed by strong evidence on the causal pathways linking structural determinants and interventions to changes in HIV outcomes—through their direct and indirect effects on downstream mediators—can be used to estimate the contribution of structural determinants to HIV epidemics and to predict the impacts of structural interventions. Our recommendations for the next generation of models can help modellers think about how to model exposure to structural determinants and interventions dynamically and mechanistically to improve estimation of their impacts. Future research should prioritise longitudinal studies designed to estimate the causal effects of structural determinants on mediators and HIV over suitable timeframes. This will not only contribute to a deeper understanding of structural determinants, but also facilitate greater use of models in exploring the impacts and economic feasibility of structural interventions, which will be critical in the next phase of the global HIV response.

Availability of data and materials

All information extracted and analysed during this study is included in this published article (and its supplementary information files).

Abbreviations

ART:

Antiretroviral therapy

AIDS:

Acquired immune deficiency syndrome

FSW:

Female sex workers

HIV:

Human immunodeficiency virus

MSM:

Men who have sex with men

PAF:

Population attributable fraction

PLHIV:

People living with HIV

PrEP:

Pre-exposure prophylaxis

PWID:

People who inject drugs

R0 :

Basic reproduction number

STI:

Sexually transmitted infection

tPAF:

Transmission population attributable fraction

UNAIDS:

Joint United Nations Programme on HIV/AIDS

References

  1. Sumartojo E. Structural factors in HIV prevention: concepts, examples, and implications for research. AIDS. 2000;14:S3–10.

    Article  PubMed  Google Scholar 

  2. Phelan JC, Link BG, Tehranifar P. Social conditions as fundamental causes of health inequalities: theory, evidence, and policy implications. Journal of health and social behavior. 2010;51(1_suppl):S28–40.

    Article  PubMed  Google Scholar 

  3. Marmot M. The influence of income on health: views of an epidemiologist. Health Aff. 2002;21(2):31–46.

    Article  Google Scholar 

  4. Shannon K, Strathdee SA, Goldenberg SM, et al. Global epidemiology of HIV among female sex workers: Influence of structural determinants. The Lancet. 2015;385(9962):55–71.

    Article  Google Scholar 

  5. Strathdee SA, Hallett TB, Bobrova N, et al. HIV and risk environment for injecting drug users: the past, present, and future. The Lancet. 2010;376(9737):268–84.

    Article  Google Scholar 

  6. Organization WH. Consolidated guidelines on HIV prevention, diagnosis, treatment and care for key populations: World Health Organization; 2016.

  7. Shannon K, Goldenberg SM, Deering KN, Strathdee SA. HIV infection among female sex workers in concentrated and high prevalence epidemics: why a structural determinants framework is needed. Curr Opin HIV AIDS. 2014;9(2):174–82.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Deering KN, Amin A, Shoveller J, et al. A systematic review of the correlates of violence against sex workers. Am J Public Health. 2014;104(5):e42–54.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Beyrer C, Sullivan P, Sanchez J, et al. The increase in global HIV epidemics in MSM. AIDS. 2013;27(17):2665–78.

    Article  PubMed  Google Scholar 

  10. UNAIDS. Global AIDS Strategy 2021–2026 — End Inequalities. End AIDS. Geneva, Switzerland, 2021.

  11. Buckee C, Noor A, Sattenspiel L. Thinking clearly about social aspects of infectious disease transmission. Nature. 2021;595(7866):205–13.

    Article  CAS  PubMed  Google Scholar 

  12. Pickles M, Boily M-C, Vickerman P, et al. Assessment of the population-level effectiveness of the Avahan HIV-prevention programme in South India: a preplanned, causal-pathway-based modelling analysis. Lancet Glob Health. 2013;1(5):e289–99.

    Article  PubMed  Google Scholar 

  13. Stover J. HIV models to inform health policy. Curr Opin HIV AIDS. 2011;6(2):108–13.

    Article  PubMed  Google Scholar 

  14. Anderson RM. The role of mathematical models in the study of HIV transmission and the epidemiology of AIDS. J Acquir Immune Defic Syndr. 1988;1(3):241–56.

    CAS  PubMed  Google Scholar 

  15. Rowley JT, Anderson RM, Ng TW. Reducing the spread of HIV infection in sub-Saharan Africa: some demographic and economic implications. AIDS. 1990;4(1):47–56.

    Article  CAS  PubMed  Google Scholar 

  16. Abbas UL, Anderson RM, Mellors JW. Potential impact of antiretroviral chemoprophylaxis on HIV-1 transmission in resource-limited settings. PLoS ONE. 2007;2(9): e875.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Organization WH. A conceptual framework for action on the social determinants of health. 2010.

  18. Godin A, Kronfli N, Cox J, Alary M, Maheu-Giroux M. The role of prison-based interventions for hepatitis C virus (HCV) micro-elimination among people who inject drugs in Montréal. Canada International Journal of Drug Policy. 2021;88: 102738.

    Article  PubMed  Google Scholar 

  19. Maheu-Giroux M, Sardinha L, Stöckl H, et al. A framework to model global, regional, and national estimates of intimate partner violence. BMC Med Res Methodol. 2022;22(1):159.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Boily M-C, Brisson M, Mâsse B, Anderson RM. The Role of Mathematical Models in Vaccine Development and Public Health Decision Making. Vaccinology; 2012: 480–508.

  21. Stone J, Fraser H, Young AM, Havens JR, Vickerman P. Modeling the role of incarceration in HCV transmission and prevention amongst people who inject drugs in rural Kentucky. International Journal of Drug Policy. 2021;88: 102707.

    Article  PubMed  Google Scholar 

  22. Lim AG, Stone J, Hajarizadeh B, et al. Evaluating the Prevention Benefit of HCV Treatment: Modeling the SToP-C Treatment as Prevention Study in Prisons. Hepatology. 2021;74(5):2366–79.

    Article  PubMed  Google Scholar 

  23. Stone J, Martin NK, Hickman M, et al. Modelling the impact of incarceration and prison-based hepatitis C virus (HCV) treatment on HCV transmission among people who inject drugs in Scotland. Addiction. 2017;112(7):1302–14.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Masanjala W. The poverty-HIV/AIDS nexus in Africa: a livelihood approach. Soc Sci Med. 2007;64(5):1032–41.

    Article  PubMed  Google Scholar 

  25. Bachmann MO, Booysen FL. Health and economic impact of HIV/AIDS on South African households: a cohort study. BMC Public Health. 2003;3:1–8.

    Article  Google Scholar 

  26. Mishra S, Fisman DN, Boily M-C. The ABC of terms used in mathematical models of infectious diseases. J Epidemiol Community Health. 2011;65(1):87–94.

    Article  PubMed  Google Scholar 

  27. Eaton JW, Hallett TB. Why the proportion of transmission during early-stage HIV infection does not predict the long-term impact of treatment on HIV incidence. Proc Natl Acad Sci. 2014;111(45):16202–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Hallett TB, Gregson S, Mugurungi O, Gonese E, Garnett GP. Assessing evidence for behaviour change affecting the course of HIV epidemics: a new mathematical modelling approach and application to data from Zimbabwe. Epidemics. 2009;1(2):108–17.

    Article  PubMed  Google Scholar 

  29. Wang L, Moqueet N, Simkin A, et al. Mathematical modelling of the influence of serosorting on the population-level HIV transmission impact of pre-exposure prophylaxis. AIDS. 2021;35(7):1113–25.

    Article  CAS  PubMed  Google Scholar 

  30. Boily M-C, Anderson RM. Sexual contact patterns between men and women and the spread of HIV-1 in urban centres in Africa. Mathematical Medicine and Biology: A Journal of the IMA. 1991;8(4):221–47.

    Article  CAS  Google Scholar 

  31. Rhodes T. The ‘risk environment’: a framework for understanding and reducing drug-related harm. International journal of drug policy. 2002;13(2):85–94.

    Article  Google Scholar 

  32. Gupta GR, Parkhurst JO, Ogden JA, Aggleton P, Mahal A. Structural approaches to HIV prevention. The Lancet. 2008;372(9640):764–75.

    Article  Google Scholar 

  33. Blankenship KM, Friedman SR, Dworkin S, Mantell JE. Structural interventions: concepts, challenges and opportunities for research. J Urban Health. 2006;83:59–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Schwartländer B, Stover J, Hallett T, et al. Towards an improved investment approach for an effective response to HIV/AIDS. The Lancet. 2011;377(9782):2031–41.

    Article  Google Scholar 

  35. Ng M, Gakidou E, Levin-Rector A, Khera A, Murray CJ, Dandona L. Assessment of population-level effect of Avahan, an HIV-prevention initiative in India. The Lancet. 2011;378(9803):1643–52.

    Article  Google Scholar 

  36. Cowan FM, Chabata ST, Musemburi S, et al. Strengthening the scale-up and uptake of effective interventions for sex workers for population impact in Zimbabwe. J Int AIDS Soc. 2019;22 Suppl 4(Suppl Suppl 4):e25320.

    Article  PubMed  Google Scholar 

  37. Vermeulen R, Chadeau-Hyam M. Commentary: Dynamic Aspects of Exposure History—Do They Matter? Epidemiology. 2012;23(6):900–1.

    Article  PubMed  Google Scholar 

  38. Auerbach JD, Parkhurst JO, Cáceres CF. Addressing social drivers of HIV/AIDS for the long-term response: Conceptual and methodological considerations. Glob Public Health. 2011;6(sup3):S293–309.

    Article  PubMed  Google Scholar 

  39. Harper S, Strumpf EC. Commentary: social epidemiology: questionable answers and answerable questions. Epidemiology. 2012;23(6):795–8.

    Article  PubMed  Google Scholar 

  40. Reeves A, Steele S, Stuckler D, McKee M, Amato-Gauci A, Semenza JC. National sex work policy and HIV prevalence among sex workers: an ecological regression analysis of 27 European countries. The Lancet HIV. 2017;4(3):e134–40.

    Article  PubMed  Google Scholar 

  41. Kiss IZ, Green DM, Kao RR. The effect of network mixing patterns on epidemic dynamics and the efficacy of disease contact tracing. J R Soc Interface. 2008;5(24):791–9.

    Article  PubMed  Google Scholar 

  42. Grassly NC, Fraser C. Seasonal infectious disease epidemiology. Proceedings of the Royal Society B: Biological Sciences. 2006;273(1600):2541–50.

    Article  PubMed Central  Google Scholar 

  43. Tricco AC, Lillie E, Zarin W, et al. PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med. 2018;169(7):467–73.

    Article  PubMed  Google Scholar 

  44. Adams JW, Khan MR, Bessey S, et al. Preexposure prophylaxis strategies for African-American women affected by mass incarceration. AIDS (London, England). 2021;35(3):453.

    Article  PubMed  Google Scholar 

  45. Adams JW, Lurie MN, King MRF, et al. Potential drivers of HIV acquisition in African-American women related to mass incarceration: an agent-based modelling study. BMC Public Health. 2018;18(1):1387.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Bernard CL, Rao IJ, Robison KK, Brandeau ML. Health outcomes and cost-effectiveness of diversion programs for low-level drug offenders: a model-based analysis. PLoS Med. 2020;17(10): e1003239.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Altice FL, Azbel L, Stone J, et al. The perfect storm: incarceration and the high-risk environment perpetuating transmission of HIV, hepatitis C virus, and tuberculosis in Eastern Europe and Central Asia. The Lancet. 2016;388(10050):1228–48.

    Article  Google Scholar 

  48. Borquez A, Beletsky L, Nosyk B, et al. The effect of public health-oriented drug law reform on HIV incidence in people who inject drugs in Tijuana, Mexico: an epidemic modelling study. The Lancet Public Health. 2018;3(9):e429–37.

    Article  PubMed  PubMed Central  Google Scholar 

  49. Decker MR, Wirtz AL, Pretorius C, et al. Estimating the impact of reducing violence against female sex workers on HIV epidemics in Kenya and Ukraine: a policy modeling exercise. Am J Reprod Immunol. 2013;69:122–32.

    Article  PubMed  Google Scholar 

  50. Dolan K, Wirtz AL, Moazen B, et al. Global burden of HIV, viral hepatitis, and tuberculosis in prisoners and detainees. The Lancet. 2016;388(10049):1089–102.

    Article  Google Scholar 

  51. Levy B, Correia HE, Chirove F, et al. Modeling the Effect of HIV/AIDS Stigma on HIV Infection Dynamics in Kenya. Bull Math Biol. 2021;83(5):55.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Rigby SW, Johnson LF. The relationship between intimate partner violence and HIV: A model-based evaluation. Infectious Disease Modelling. 2017;2(1):71–89.

    Article  PubMed  PubMed Central  Google Scholar 

  53. Stone J, Artenie A, Hickman M, et al. The contribution of unstable housing to HIV and hepatitis C virus transmission among people who inject drugs globally, regionally, and at country level: a modelling study. The Lancet Public Health. 2022;7(2):e136–45.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Ward Z, Stone J, Bishop C, et al. Costs and impact on HIV transmission of a switch from a criminalisation to a public health approach to injecting drug use in eastern Europe and central Asia: a modelling analysis. The Lancet HIV. 2022;9(1):e42–53.

    Article  CAS  PubMed  Google Scholar 

  55. Stover J, Glaubius R, Teng Y, et al. Modeling the epidemiological impact of the UNAIDS 2025 targets to end AIDS as a public health threat by 2030. PLoS Med. 2021;18(10): e1003831.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. Vassall A, Chandrashekar S, Pickles M, et al. Community Mobilisation and Empowerment Interventions as Part of HIV Prevention for Female Sex Workers in Southern India: A Cost-Effectiveness Analysis. PLoS ONE. 2014;9(10): e110562.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Wirtz AL, Pretorius C, Beyrer C, et al. Epidemic impacts of a community empowerment intervention for HIV prevention among female sex workers in generalized and concentrated epidemics. PLoS ONE. 2014;9(2): e88047.

    Article  PubMed  PubMed Central  Google Scholar 

  58. Ronoh M, Chirove F, Wairimu J, Ogana W. Evidence-based modeling of combination control on Kenyan youth HIV/AIDS dynamics. PLoS ONE. 2020;15(11): e0242491.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  59. Arreola S, Santos G-M, Beck J, et al. Sexual Stigma, Criminalization, Investment, and Access to HIV Services Among Men Who Have Sex with Men Worldwide. AIDS Behav. 2015;19(2):227–34.

    Article  PubMed  Google Scholar 

  60. Golub SA, Gamarel KE. The impact of anticipated HIV stigma on delays in HIV testing behaviors: findings from a community-based sample of men who have sex with men and transgender women in New York City. AIDS Patient Care STDS. 2013;27(11):621–7.

    Article  PubMed  PubMed Central  Google Scholar 

  61. Murray EJ, Robins JM, Seage GR, Freedberg KA, Hernán MA. A Comparison of Agent-Based Models and the Parametric G-Formula for Causal Inference. Am J Epidemiol. 2017;186(2):131–42.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Corraini P, Olsen M, Pedersen L, Dekkers OM, Vandenbroucke JP. Effect modification, interaction and mediation: an overview of theoretical insights for clinical investigators. Clin Epidemiol. 2017: 331–8.

  63. Pitman R, Fisman D, Zaric GS, et al. Dynamic transmission modeling: a report of the ISPOR-SMDM modeling good research practices task force-5. Value in health. 2012;15(6):828–34.

    Article  PubMed  PubMed Central  Google Scholar 

  64. Degtiar I, Rose S. A review of generalizability and transportability. Annual Review of Statistics and Its Application. 2023;10:501–24.

    Article  Google Scholar 

  65. Eaton JW, Bacaër N, Bershteyn A, et al. Assessment of epidemic projections using recent HIV survey data in South Africa: a validation analysis of ten mathematical models of HIV epidemiology in the antiretroviral therapy era. Lancet Glob Health. 2015;3(10):e598–608.

    Article  PubMed  Google Scholar 

  66. Fitzgerald-Husek A, Van Wert MJ, Ewing WF, et al. Measuring stigma affecting sex workers (SW) and men who have sex with men (MSM): a systematic review. PLoS ONE. 2017;12(11): e0188393.

    Article  PubMed  PubMed Central  Google Scholar 

  67. Singer M, Bulled N, Ostrach B. Whither syndemics?: Trends in syndemics research, a review 2015–2019. Glob Public Health. 2020;15(7):943–55.

    Article  PubMed  Google Scholar 

  68. Joy R, Druyts EF, Brandson EK, et al. Impact of neighborhood-level socioeconomic status on HIV disease progression in a universal health care setting. JAIDS Journal of Acquired Immune Deficiency Syndromes. 2008;47(4):500–5.

    Article  PubMed  Google Scholar 

  69. Wojcicki JM. Socioeconomic status as a risk factor for HIV infection in women in East, Central and Southern Africa: a systematic review. J Biosoc Sci. 2005;37(1):1–36.

    Article  PubMed  Google Scholar 

  70. Naimi AI, Kaufman JS. Counterfactual Theory in Social Epidemiology: Reconciling Analysis and Action for the Social Determinants of Health. Current Epidemiology Reports. 2015;2(1):52–60.

    Article  Google Scholar 

  71. Hogan JW, Galai N, Davis WW. Modeling the impact of social determinants of health on HIV. AIDS Behav. 2021;25:215–24.

    Article  PubMed  PubMed Central  Google Scholar 

  72. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008;8(1):70.

    Article  PubMed  PubMed Central  Google Scholar 

  73. Kaufman JS. Commentary: causal inference for social exposures. Annu Rev Public Health. 2019;40:7–21.

    Article  PubMed  Google Scholar 

  74. Jackson JW. On the interpretation of path-specific effects in health disparities research. Epidemiology. 2018;29(4):517–20.

    Article  PubMed  Google Scholar 

  75. Vaillant J, Koussoubé E, Roth D, Pierotti R, Hossain M, Falb KL. Engaging men to transform inequitable gender attitudes and prevent intimate partner violence: a cluster randomised controlled trial in North and South Kivu, Democratic Republic of Congo. BMJ Global Health 2020;5(5).

  76. Pettifor A, Lippman SA, Gottert A, et al. Community mobilization to modify harmful gender norms and reduce HIV risk: results from a community cluster randomized trial in South Africa. J Int AIDS Soc. 2018;21(7): e25134.

    Article  PubMed  PubMed Central  Google Scholar 

  77. Sharma V, Leight J, Verani F, Tewolde S, Deyessa N. Effectiveness of a culturally appropriate intervention to prevent intimate partner violence and HIV transmission among men, women, and couples in rural Ethiopia: findings from a cluster-randomized controlled trial. PLoS Med. 2020;17(8): e1003274.

    Article  PubMed  PubMed Central  Google Scholar 

  78. Baird SJ, Garfein RS, McIntosh CT, Özler B. Effect of a cash transfer programme for schooling on prevalence of HIV and herpes simplex type 2 in Malawi: a cluster randomised trial. The Lancet. 2012;379(9823):1320–9.

    Article  Google Scholar 

  79. Humphries H, Kharsany AB, Leask K, Ntombela F, Abdool Karim Q. The impact of conditional cash transfers in reducing HIV in adolescent girls and boys (RHIVA): the CAPRISA 007 matched pair, cluster randomised controlled trial. The CAPRISA Clinical Trials: HIV Treatment and Prevention. 2017: 77–89.

  80. Dunbar MS, Kang Dufour M-S, Lambdin B, Mudekunye-Mahaka I, Nhamo D, Padian NS. The SHAZ! project: results from a pilot randomized trial of a structural intervention to prevent HIV among adolescent women in Zimbabwe. PLoS ONE. 2014;9(11): e113621.

    Article  PubMed  PubMed Central  Google Scholar 

  81. Gupta J, Falb KL, Lehmann H, et al. Gender norms and economic empowerment intervention to reduce intimate partner violence against women in rural Côte d’Ivoire: a randomized controlled pilot study. BMC Int Health Hum Rights. 2013;13(1):46.

    Article  PubMed  PubMed Central  Google Scholar 

  82. Gibbs A, Corboz J, Chirwa E, et al. The impacts of combined social and economic empowerment training on intimate partner violence, depression, gender norms and livelihoods among women: an individually randomised controlled trial and qualitative study in Afghanistan. BMJ Glob Health. 2020;5(3): e001946.

    Article  PubMed  PubMed Central  Google Scholar 

  83. Boyce SC, Minnis A, Deardorff J, et al. Effect of a gender-synchronized family planning intervention on inequitable gender norms in a cluster randomized control trial among husbands of married adolescent girls in Dosso, Niger. medRxiv 2023: 2023.09. 28.23296292.

  84. Boyce SC, Minnis A, Deardorff J, et al. Mediating effects of inequitable gender norms on intimate partner violence and contraceptive use in a cluster randomized control trial in Niger: A causal inference mediation analysis. medRxiv. 2023.

  85. Jones NL, Gilman SE, Cheng TL, Drury SS, Hill CV, Geronimus AT. Life course approaches to the causes of health disparities. Am J Public Health. 2019;109(S1):S48–55.

    Article  PubMed  PubMed Central  Google Scholar 

  86. McGillen JB, Anderson S-J, Dybul MR, Hallett TB. Optimum resource allocation to reduce HIV incidence across sub-Saharan Africa: a mathematical modelling study. The lancet HIV. 2016;3(9):e441–8.

    Article  PubMed  Google Scholar 

  87. de Oliveira RB, Rubio FA, Anderle R, et al. Incorporating social determinants of health into the mathematical modeling of HIV/AIDS. Sci Rep. 2022;12(1):20541.

    Article  PubMed  PubMed Central  Google Scholar 

  88. Pedrazzoli D, Boccia D, Dodd PJ, et al. Modelling the social and structural determinants of tuberculosis: opportunities and challenges. Int J Tuberc Lung Dis. 2017;21(9):957–64.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This work was supported by the Wellcome Trust [226619/Z/22/Z] and the McGill Interdisciplinary Initiative in Infection and Immunity–McGill Global Health Programmes (Mi4-GHP). JS1 is supported by a doctoral award from the Fonds de Recherche du Québec – Santé (FRQ-S). MCB acknowledges funding from the MRC Centre for Global Infectious Disease Analysis [MR/X020258/1], funded by the UK Medical Research Council (MRC). This UK-funded award is carried out in the frame of the Global Health EDCTP3 Joint Undertaking. MCB also acknowledges funding from the HIV Prevention Trials Network (HPTN) Modelling Centre, which is funded by the U.S. National Institutes of Health (NIH UM1 AI068617) through the HPTN. MM-G’s research programme is funded by a Canada Research Chair (Tier II) in Population Health Modeling. AP was partly funded by the Bill & Melinda Gates Foundation (INV-007145). The funders had no role in the conceptualisation, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Author information

Authors and Affiliations

  1. Department of Epidemiology and Biostatistics, School of Population and Global Health, McGill University, Montréal, Canada

    James Stannah, Jorge Luis Flores Anato & Mathieu Maheu-Giroux

  2. MRC Centre for Global Infectious Disease Analysis, School of Public Health, Imperial College London, London, UK

    Michael Pickles, Kate M. Mitchell & Marie-Claude Boily

  3. HPTN Modelling Centre, Imperial College London, London, UK

    Michael Pickles & Marie-Claude Boily

  4. Centre Population et Développement, Institut de Recherche pour le Développement, Université Paris Cité, Inserm, Paris, France

    Joseph Larmarange

  5. Department of Nursing and Community Health, Glasgow Caledonian University, London, UK

    Kate M. Mitchell

  6. Population Health Sciences, University of Bristol, Bristol, UK

    Adelina Artenie, Jack Stone & Peter Vickerman

  7. Ukrainian Institute On Public Health Policy, Kiev, Ukraine

    Kostyantyn Dumchev

  8. Programme PAC-CI, CHU de Treichville, Site ANRS, Abidjan, Côte d’Ivoire

    Serge Niangoran

  9. Department of Social and Environmental Health Research, London School of Hygiene and Tropical Medicine Faculty of Public Health and Policy, London, UK

    Lucy Platt

  10. UNAIDS, Geneva, Switzerland

    Fern Terris-Prestholt

  11. The Johns Hopkins University School of Medicine, Delhi, India

    Aditya Singh

  12. Institute for Global Health, University College London, London, UK

    Andrew Phillips

  13. Centre for Infectious Disease Epidemiology and Research, University of Cape Town, Cape Town, South Africa

    Leigh Johnson

Authors
  1. James Stannah
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jorge Luis Flores Anato
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Michael Pickles
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Joseph Larmarange
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Kate M. Mitchell
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Adelina Artenie
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Kostyantyn Dumchev
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Serge Niangoran
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Lucy Platt
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Fern Terris-Prestholt
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Aditya Singh
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Jack Stone
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Peter Vickerman
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Andrew Phillips
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Leigh Johnson
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Mathieu Maheu-Giroux
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Marie-Claude Boily
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

JS1, MM-G, and M-CB conceptualised the study. JS1 conducted the scoping review and drafted the initial manuscript. M-CB, MM-G, JLFO, MP, JL, KMM, AA, KD, SN, LP, FTP, AS, JS2, PV, AP, and LJ provided substantive edits to the manuscript. All authors read and approved the manuscript and agreed to submit for publication.

Authors’ Twitter handles

Twitter handles: @JamesStannah (James Stannah), @epi_arepa (Jorge Flores Luis Anato), @J_Larmarange (Joseph Larmarange), @adelinaartenie (Adelina Artenie), @Mat__MG (Mathieu Maheu-Giroux), @MCBOILY2 (Marie-Claude Boily), @LeighJohnson9 (Leigh Johnson)

Corresponding author

Correspondence to Marie-Claude Boily.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

JL reports grants from Unitaid and ANRS|MIE, consulting fees from Inserm, presidency of the scientific committee of ANRS|MIE evaluating projects submitted for funding, and membership of a scientific committee at Inserm, all outside the submitted work. KMM reports consulting fees from the University of North Carolina, and payments from Pfizer for teaching, all outside the submitted work. The other authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

12916_2024_3580_MOESM1_ESM.docx

Additional file 1: Text S1. Additional scoping review methods. Text S2. Additional results of scoping review. Text S3. Definitions of effects that can be estimated using causal mediation analysis.

12916_2024_3580_MOESM2_ESM.docx

Additional file 2: Table S1. Examples of structural determinants, societal enablers, and structural interventions identified in the UNAIDS Global AIDS Strategy 2021-2026 that are important for HIV transmission, and the mechanisms through which they impact HIV. Table S2. Scoping review search terms and hits. Table S3. Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) Checklist. Table S4. Additional information on the modelling of structural determinants and interventions in the studies identified in the scoping review. Table S5. Empirical evidence used to parameterise models of exposure to structural determinants. Table S6. Data on HIV epidemiology and structural determinants used to calibrate the models.

Additional file 3: Figure S1. PRISMA-ScR checklist for the scoping review.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Stannah, J., Flores Anato, J.L., Pickles, M. et al. From conceptualising to modelling structural determinants and interventions in HIV transmission dynamics models: a scoping review and methodological framework for evidence-based analyses. BMC Med 22, 404 (2024). https://doi.org/10.1186/s12916-024-03580-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12916-024-03580-z

Keywords

BMC Medicine

ISSN: 1741-7015

Contact us