To increase HIV testing among New York State (NYS) residents, a 2010 law requires that all persons aged 13 to 64 be offered HIV testing as part of routine medical care, and simplifies the informed consent and pre-test counseling processes. Although the NYS Department of Health (NYSDOH) is conducting an extensive policy evaluation using empirical data, traditional quantitative research methods (such as statistics and cost-effectiveness analysis) may not adequately address complexities in the system of HIV testing and care, involving multiple public and private payers, facilities, providers, and social services. Furthermore, the law is being implemented in the context of concurrent policies that may affect outcomes. Some of these related policies (such as national guidelines on HIV testing frequency) have similar goals to the state law, thereby attenuating its observed effect in empirical data. Other policies (such as funding cuts for STD services) may counteract the law’s potential effects. A third class of concurrent policies (such as upcoming financial changes from health reform) has unclear effects. Empirical data are limited to a short time horizon, whereas the law’s impact may occur over decades. Finally, empirical data are limited to outcomes that can be directly measured (such as new diagnoses, rather than new infections).

We collaborated with NYSDOH on a system dynamics model to supplement the policy evaluation. This branch of computer simulation modeling is useful for policy analysis and design for problems arising in complex social, managerial, economic, or ecological systems. Discussions with system experts were used to develop a conceptual model of the important variables and their relationships, and transform them into mathematical relationships. The model structure contains feedback loops to generate new infections over time using a “susceptible-infected” epidemic modeling framework. HIV patients’ disease transmission rates differ according to diagnosis (awareness may lead to behavior change), linkage to and engagement in medical care (antiretroviral therapy suppresses HIV viral load, thereby reducing infectivity), and stage of disease progression. To calibrate the model, estimate model parameters, and generate ranges for sensitivity analyses, empirical data were compiled from various sources including: the disease surveillance registry; claims from Medicaid and outpatient emergency department discharges; surveys to NYS labs that perform HIV testing, medical providers, and NYS residents (via modules appended to national health surveys); administrative data from NYSDOH-funded providers and programs; and published literature.

After model calibration, inputs will be modified to conduct counterfactual analyses of short- and long-term outcomes from the new HIV testing law, holding all other factors constant. Outcomes include: HIV testing rates, HIV diagnoses, linkage to care among newly diagnosed individuals, proportion of late diagnoses, and future infections. Additional simulation runs will assess how other related policies could affect these outcomes, and possible interactive effects. Results will be presented as graphs over time under different scenarios. In addition to predicting the law’s impact, output will help identify potential implementation challenges such as capacity shortages as more individuals are diagnosed and linked to care, or areas where NYSDOH may wish to redirect resources.