Insurance Indicators in the California AIDS Drug Assistance Program

By Monica Ghaly, MS; Genevieve Kray, MPH; Glorietta Kundetti, MPH; Kelly Wu, MS

Facebook
Twitter
LinkedIn

Citation

Ghaly M, Kray G, Kunddetti G, Wu K. Insurance indicators in the California AIDS drug assistance program. HPHR. 2022;70.

Insurance Indicators in the California AIDS Drug Assistance Program Insurance Indicators in the California AIDS Drug Assistance Program

Background

Insurance coverage is associated with better health outcomes for people living with HIV (PLWH). The California AIDS Drug Assistance Program (ADAP) provides funding for life-saving medications, insurance premium payments, and medical-out-of-pocket costs for PLWH. Considering the services offered by the program and the benefits of health insurance, there is programmatic interest in navigating uninsured clients into coverage. Prior studies explored the relationships between Spanish language preference, marital status, age, region of residence, and income with insurance status in the general population.

 

Insurance coverage is associated with better health outcomes for people living with HIV (PLWH). The California AIDS Drug Assistance Program (ADAP) provides funding for life-saving medications, insurance premium payments, and medical-out-of-pocket costs for PLWH. Considering the services offered by the program and the benefits of health insurance, there is programmatic interest in navigating uninsured clients into coverage. Prior studies explored the relationships between Spanish language preference, marital status, age, region of residence, and income with insurance status in the general population. 

Aims/Purpose

This study aims to determine the utility of these factors in identifying clients most likely to be uninsured to bolster navigation efforts. Methods Utilizing data from the 19,358 clients active in the California ADAP Enrollment System (AES) database as of October 2022, we constructed a logistic regression model to estimate the odds ratios of being insured vs uninsured given Spanish language preference, marital status, age, region of residence, viral suppression, gender, and income. Moreover, a random forest model using the same factors applied an adjusted sampling scheme to reduce the rate of misclassification observed in the confusion matrix.

Results

An analysis of the insurance status and demographics of the active 19,358 ADAP clients under age 65 shows that individuals whose preferred language is Spanish (OR 6.52, p less than 0.001), reside in the Inland Empire (OR 1.21, p = 0.02) or Southern California regions (OR 1.98, p less than 0.001), are between the ages of 58 to 64 and have an income between 200% and 300% of the Federal Poverty Level (FPL) (OR 1.82, p less than 0.001), or are unpartnered (OR 1.42, p less than 0.001) are more likely to be uninsured than insured. A confusion matrix analysis demonstrates that a random forest model that over-samples gender and language preference populations and under-samples insured clients has a higher recall rate (73.5% versus 61.1%) than one with unadjusted sampling. These results will be used to identify clients likely to be uninsured to inform navigation efforts.

Discussion Discussion Discussion

This study starkly shows that ADAP clients whose preferred language is Spanish have the highest odds of being uninsured. However, those whose preferred language is Spanish and are 27 to 49 are less likely to be uninsured. Additionally, as shown in prior research, unpartnered people are more likely to be uninsured, and virally suppressed ADAP clients are less likely to be uninsured. The random forest model trained on the adjusted sampling scheme is more nuanced and minimizes the most egregious mis-categorization error.

Conclusion

Further research is needed to identify the barriers that those whose preferred language is Spanish face in the health insurance system; more granular preferred language categories may pinpoint the drivers for this disparity. Future studies may also retain distinct insurance types to explore insurance trends.

About the Author