Frequently Asked Questions

During 201762019, annual SUD service expenditures averaged $31 billion across the examined payers, and Medicaid funded the largest share. Medicaid paid 52.7% of SUD service expenditures and accounted for 37.7% of SUD service utilizations, while private insurance paid 32.3% of expenditures and represented 12.9% of utilizations. Self-pay accounted for 5.1% of SUD service expenditures but represented 43.0% of SUD service utilizations.

County-level SUD service use varied markedly across the United States. Across 3,143 US counties during 201762019, the average annual county-level number of SUD health service utilizations was 5.2 per 1,000 county residents (95% CI, 5.15.3). The distribution was <3.0 in 23.3% of counties, 3.03.9 in 23.3%, 4.05.9 in 25.3%, 6.09.9 in 18.8%, and 610.0 in 9.3%.

Self-pay had the highest average annual county-level SUD service utilization rate among the examined payer groups. The mean was 8.6 utilizations per 1,000 county residents (95% CI, 8.48.6) for self-pay, compared with 6.9 per 1,000 county Medicaid enrollees (95% CI, 6.67.1) for Medicaid and 2.8 per 1,000 privately insured county residents (95% CI, 2.72.8) for private insurance.

The county-level percentage of uninsured adults was the strongest correlate of lower SUD service utilization. In the overall model, uninsured rate alone accounted for 23.0% of the variance, and in Model 4 it was negatively associated with county-level SUD health service utilization (b2 = 0.45, P<.0001). The same pattern was seen for self-pay (b2 = 0.40, P<.0001), Medicaid (b2 = 0.46, P<.0001), and private insurance (b2 = 0.14, P<.0001).

Counties in the top 20% of SUD service use had 4.2 times the utilization of counties in the bottom 20%. The average annual county-level number of SUD health service utilizations was 10.4 per 1,000 county residents (95% CI, 10.210.6) in the top 20% versus 2.5 per 1,000 (95% CI, 2.52.5) in the bottom 20%. Uninsured adult rates also differed substantially: 9.3% (95% CI, 9.1%9.6%) in top-use counties versus 19.7% (95% CI, 19.2%20.3%) in bottom-use counties.

The Medicaid gap was especially large: counties in the top 20% had 10.4 times the Medicaid-paid SUD utilization of counties in the bottom 20%. Average annual utilization was 16.6 per 1,000 county Medicaid enrollees (95% CI, 16.217.0) in the top 20% versus 1.6 per 1,000 (95% CI, 1.61.7) in the bottom 20%.

Yes, but the direction differed by payer. County-level unemployment was positively associated with SUD service utilization overall (b2 = .20, P<.0001), for self-pay (b2 = .20, P<.0001), and for Medicaid (b2 = .25, P<.0001). For private insurance, unemployment was negatively associated with utilization (b2 = 0.12, P<.0001).

Yes. County-level models explained much more of the variation in SUD service utilization than in overall health care utilization. Model 4 explained 57.5% of the variance in overall SUD service utilization, compared with 20.2% for utilization for overall health conditions. The authors note that uninsured adult rates had a disproportionate relationship to SUD service use compared with their more limited influence on overall health service utilization.

In this study, self-pay represented a large share of SUD service use but a small share of spending, largely because self-pay services were primarily ambulatory care and prescriptions. The authors state that self-pay covered SUD services primarily in ambulatory care and prescriptions and note that funding from other federal, state, and local governments, foundations, or charities may have provided some subsidies. They interpret the high self-pay utilization as consistent with barriers to insured care, including lack of insurance, inadequate coverage benefits, cash-only acceptance by some SUD treatment providers, stigma-related privacy concerns, and utilization management requirements.

The main limitations were lack of county-level SUD prevalence data, use of county-level rather than individual-level data, incomplete payer capture, and limitations in the mortality data. The authors could adjust for county-level overdose mortality but not for county-level SUD prevalence, so they compared top-versus-bottom utilization counties rather than counties ranked by SUD prevalence. They also caution about ecological fallacy because analyses were county-level, note that mortality data lacked information on decedents' insurance and employment status, and state that other payers such as the Substance Abuse Prevention and Treatment Block Grant could not be examined because of data limitations.