Neeraj Sood
Leonard D. Schaeffer Center for Health Policy and Economics, University of Southern California
American Journal of Health Economics 1(4): 515-536, 2015
According to the Centers for Disease Control (CDC), HIV/AIDS has claimed more than 500,000 lives in the United States and continues to infect about 50,000 people annually. Unfortunately, 18.1% of people living with HIV are undiagnosed and this ‘HIV-unaware’ population contributes to roughly half of all new HIV transmissionsi. Because people are significantly less likely to engage in risky sexual behavior if they are made aware of an HIV infection, providing HIV testing and counseling services to the HIV-unaware is a key tenet of the Obama Administration’s National HIV/AIDS strategyii. The expansion of health insurance coverage under the Affordable Care Act (ACA) is likely to have important effects on both the HIV testing rate and related health behaviors—however, exactly what these effects will be has remained unclear. In two recent papers, both coauthored with Zachary Wagner of the University of California, Berkeley and Yanyu Wu of University of Southern California, we sought to investigate the causal link between insurance coverage and HIV testing and use our results to predict changes in HIV testing behavior under the ACA.
There are several mechanisms through which the expansion of health insurance could influence HIV testing rates. First, insurance subsidizes HIV testing, lowering the cost incurred by the beneficiary, which likely increases utilization. Second, health insurance reduces the cost of treatment, which makes testing more valuable since a positive diagnosis is more likely to result in early initiation of treatment. Finally, health insurance increases utilization of health care and increased contact with the health system provides more testing opportunities. These three effects of health insurance suggest that efforts to expand HIV testing might be complemented by the substantial expansion in health coverage induced by the ACA.
With these mechanisms in mind, we developed our empirical model to estimate the effect of insurance coverage on HIV testing rates, allowing the effect to vary by risk status and over time to account for advancements in HIV treatment. To estimate the model, we looked at the period between 1993 and 2003 using data from the Behavioral Risk Factor Surveillance System (BRFSS)—a population-based telephone survey of U.S. adults that addresses various health behaviors associated with premature morbidity and asks for self-assessments of HIV risk status. With this data, we estimated recursive bivariate probit models with insurance coverage and HIV testing as the dependent variables. To control for potential bias, we used Medicaid eligibility and the distribution of firm size over time as instrumental variables for insurance coverage.
We had three questions in mind when we created this model: (1) How does health insurance affect HIV testing, (2) How does the effect of insurance on HIV testing differ for high-risk and low-risk individuals, and (3) How does the effect of insurance on HIV testing change over time with advancements in HIV treatment such as highly active antiretroviral therapy (HAART)?
Our results suggest that health insurance significantly increases the probability of testing for HIV in both the pre and post-HAART period. Among low risk populations, coverage increases the probability of testing by 2.6 percentage points in the pre-HAART period and by 1.8 percentage points in the post-HAART period. Among high risk populations, coverage increases the probability of testing by 2.7 percentage points in the pre-HAART period and by 4.8 percentage points in the post-HAART period.
These results suggest that insurance coverage increases HIV testing in both the high risk and low risk populations. More interestingly, the results suggest that the advent of HAART significantly increases the marginal effect on HIV testing for high risk populations and actually lowers it for low risk populations. With a highly effective but expensive treatment available, having insurance may be making HIV testing more valuable, since insurance will subsidize treatment—an effect that is compounded for high risk individuals.
With a better understanding of the impact of insurance on the likelihood of HIV testing, we next developed a model to estimate the impact of the ACA on HIV testing, diagnoses, and the percentage of people living with HIV/AIDS who are unaware of their status. To account for the uncertainty of policy adoption (the analysis was conducted prior to expansion of insurance under ACA), we used two different scenarios that provided lower and upper bounds of the ACA’s possible impact. In one scenario, we assumed that Medicaid expansion occurred only in the 18 states that had already committed to participating at the time of our writing in July 2013. In the second scenario, we assumed all 50 states expanded Medicaid.
Next, we designed a compartmental macrosimulation model and looked at the ACA’s impact for a five-year time horizon (2013-17). The model consisted of six “compartments,” which contained populations categorized by income, insurance status, level of HIV risk, probability of HIV testing, and HIV status. Incorporating estimates from our first model and again using BRFSS data on demographics, HIV risk, and insurance status, we estimated the number of people in each state who might be tested for HIV as a result of the ACA—specifically, by multiplying the number of people in each risk group expected to gain insurance by the relevant increase in testing probability that results from insurance coverage.
We found that by 2017 the ACA will substantially increase the number of people being tested for HIV. In simulations with only the subsample of 18 states expanding Medicaid, approximately 466,153 additional people are tested. In this scenario, we also found that between 2013 and 2017 there was a 22 percent reduction in HIV-unawareness among people who gained access to insurance through the ACA. If on the other hand all 50 states expand Medicaid, more than 603,024 additional people will be tested with a similar reduction in HIV-unawareness among this larger group of newly insured people.
Overall, we found that increased access to insurance dramatically increases HIV testing among the newly insured and reduces the population of HIV-unaware people—with a probable reduction in the number of HIV transmissions. It is also worth noting that our findings likely represent a lower bound given that our testing assumptions predate the new CDC recommendations for routine HIV testing for all adults, adolescents, and pregnant women. Either way, with millions more insured through the ACA, we may see a very different dynamic for the HIV/AIDS epidemic going forward.
* The article is co-authored with Zachary Wagner (School of Public Health, University of California, Berkeley) and Yanyu Wu (Precision Health Economics).
i Hall, H. Irene, David R. Holtgrave, and Catherine Maulsby. 2012. “HIV Transmission Rates from Persons Living with HIV Who Are Aware and Unaware of Their Infection.” AIDS 26 (7): 893—96.
ii Marks, Gary, Nicole Crepaz, J. Walton Senterfitt, and Robert S. Janssen. 2005. “Meta-analysis of High-Risk Sexual Behavior in Persons Aware and Unaware They Are Infected with HIV in the United States: Implications for HIV Prevention Programs.” JAIDS Journal of Acquired Immune Deficiency Syndromes 39 (4): 446—53.