Mechanistic modeling of social conditions into disease predictions for public health intervention-analyses: application to HIV

BackgroundAs social and economic conditions are key determinants of HIV, the National HIV/AIDS Strategy (NHAS), in addition to care and treatment metrics, aims to address mental health, unemployment, food insecurity, and housing instability, as a strategic plan for the Ending the HIV Epidemic initiative. Mechanistic models of HIV play a key role in identifying cost-effective intervention strategies, however, social conditions are typically not part of the modeling framework. HIV projections are typically simulated by modeling care and sexual behaviors, and transmissions as functions of those behaviors. MethodsWe developed a methodological framework, using Markov random field model to estimate joint probability distributions between social conditions and behaviors, to incorporate in a mechanistic model to simulate behaviors as functions of social conditions and HIV transmissions as a function of behaviors. As demonstration, we conducted two numerical applications in a national-level agent-based network model, Progression and Transmission of HIV (PATH 4.0). The first modeled care behavior (using viral suppression as proxy) as a function of depression, neighborhood, housing, poverty, education, insurance, and employment status. The second modeled sexual behaviors (number of partners and condom-use) as functions of employment, housing, poverty, and education status, using exchange sex as a mediator. Both simulated HIV transmissions and disease progression as functions of behaviors. We conducted what-if intervention analyses to estimate the impact of an ideal 100% efficacious intervention strategy. ResultsIf we intervene on HIV infected persons with the social needs modeled here, such that their care behavior increases to become equal to that among persons who do not have those social needs, the overall viral suppression in persons with diagnosed HIV infection increases from 65.5% to 80% (79% to 83%), resulting in a 10-year cumulative national incidence reduction of 29% (20% to 41%). If we address the social needs modeled here among persons who exchange sex, such that their sexual behavior becomes equal to that among those who do not exchange sex, we can expect a 10-year cumulative national incidence reduction of 6% (2.5% to 14%). ConclusionsWe developed a methodological framework for modeling social conditions into intervention decision-analytic models, using the limited data to present two demonstrative applications. Routinely monitoring quantitative data on associations between social conditions and HIV risk behaviors, and efficacy of structural interventions can help develop a comprehensive mechanistic model to identify cost-effective intervention combinations and inform public health strategic plans.