Availability of Outpatient Addiction Treatment and Use of Emergency Department Services Among Medicaid Enrollees

We examined whether the availability of outpatient addiction treatment programs that accept Medicaid in South Carolina’s 46 counties was associated with repeat use of the emergency department for the treatment of addiction-related conditions. Study data were for all adult (ages 18 and older) Medicaid enrollees (N=2,401) with at least one addiction-related treatment visit in an emergency department in any of South Carolina’s 73 hospitals from January 1, 2012, to December 31, 2013. Hospitals reported visit data to South Carolina’s health data administrative reporting system. The study was approved by the University of South Carolina Institutional Review Board.

The dependent variable was a dichotomous measure assessing whether an individual had repeat (two or more) addiction-related treatment visits in the emergency department. Unique patient identifiers permitted tracking of multiple visits for a single Medicaid enrollee over time and across hospitals. Addiction-related visits were those with primary ICD-9 codes for alcohol and drug use, dependence, and withdrawal identified in a subset of all treatment visits with a major diagnostic code of 20.x. The independent variable, drawn from the 2012 National Survey of Substance Abuse Treatment Services, measured the number of outpatient addiction treatment programs that accepted Medicaid in each county. Given the relatively narrow range of programs accepting Medicaid (one to three programs per county), we created a dichotomous variable indicating whether one treatment program or two or more treatment programs in the county accepted Medicaid enrollees.

We developed a multilevel regression model to assess variables at both the individual and county level. Enrollee-level variables included age, gender, race-ethnicity, primary substance of abuse at admission for the majority of enrollee visits (alcohol, opioid, or other [cocaine, methamphetamine, marijuana, and nonspecific]), and geographic region (rural or nonrural). We created two dichotomous variables to assess whether enrollees received a secondary diagnosis of a psychiatric disorder (depression, anxiety, or psychosis) or a chronic general medical condition associated with high rates of emergency department use (asthma, diabetes, coronary artery disease, stroke, or chronic kidney disease) during at least one addiction-related emergency department visit (14). County-level variables included total number of Medicaid enrollees, total number of outpatient addiction treatment programs in the county (including those that do not accept Medicaid), and a dichotomous measure indicating whether at least one outpatient treatment program in the county offered opioid treatment medication for addiction, such as methadone, buprenorphine, or naltrexone.

Descriptive statistics were calculated for all study variables. We used random-effects logistic regression, accounting for within-county effects, to assess the association of residing in a county with two or more outpatient addiction treatment programs that accept Medicaid with the odds of having a repeat addiction-related emergency service visit. Our model assumed the following form:where P(Γ_ij=1x_ij,U_0j) is the probability of any repeat visit (Γ) during the study period, given a vector of patient-level explanatory variables (x) and a vector of county-level explanatory variables (U), and where γ represents the model coefficients. The i represents each visit, the j represents county, and x_ij is the value of an independent variable for visit i for a patient living in county j. Patient, treatment, and county control variables were included in the regression model to account for individual- and county-level differences. Results are reported as odds ratios (ORs). All analyses were conducted using Stata SE 14.1.

Medicaid enrollees in this study had a total of 6,447 addiction-related emergency service visits during the study period, with an average of 1.8 visits per enrollee. Just over 30% (N=1,964) of all visits were repeat addiction-related visits. Most of the Medicaid enrollees in our study who had at least one addiction-related emergency service visit were white (N=4,133, 64%), and male (N=4,495, 70%). [A table in the online supplement presents data on enrollee- and county-level variables.] Alcohol was the most common substance of abuse reported (N=2,946, 46%). Almost half of all enrollees (N=3,147, 49%) resided in a predominantly rural county. There was an average of about four addiction treatment programs (including programs that do not accept Medicaid) in the counties in which enrollees lived, and approximately 60% offered at least some form of opioid treatment medication. Every county in which enrollees lived had at least one outpatient addiction treatment program that accepted Medicaid (including programs that offer opioid treatment medication and those that do not). Just over half (N=3,331, 52%) of South Carolina counties where enrollees resided had two or more programs accepting Medicaid.

Results of the logistic regression model indicated that the total number of addiction treatment programs that accept Medicaid in enrollees’ county of residence was inversely related to incidence of addiction-related emergency service visits (Table 1). Enrollees who lived in counties with two or more programs that accept Medicaid were less likely to have repeat addiction-related emergency service visits (OR=.75, SE=.91). Age was also a significant predictor of repeat emergency service visits, with older enrollees at increased risk of repeat use of the emergency department for addiction-related conditions.

In this study, the odds of repeat addiction-related emergency service visits were substantially lower among Medicaid enrollees who lived in counties with two or more outpatient addiction treatment programs that accept Medicaid. This finding suggests that geographic availability of treatment may play an important role in reducing the need for emergency care among Medicaid enrollees with addiction. It stands to reason that in places where addiction treatment is more plentiful, people who need it will be more likely to use it. In turn, use of regular outpatient addiction treatment may reduce the need for emergency services. Moreover, hospitals providing emergency services in counties with a greater number of local outpatient addiction treatment programs to which they can refer at discharge may be more successful in preventing repeat visits. However, further research is needed to test the validity of these possible explanations and explore causal mechanisms underlying the relationship between availability of addiction treatment, receipt of treatment, and subsequent use of emergency services.

The high incidence of addiction-related emergency services—roughly 2.5 visits per every 100 enrollees—suggests the need for targeted strategies to increase access to outpatient addiction treatment. Ideally, improving geographic availability and overall treatment access could reduce the number of Medicaid enrollees in crisis who require emergency services. Rural residents were heavily overrepresented among enrollees receiving addiction-related emergency services. Although rural residents made up only 35% of South Carolina’s Medicaid population in 2013 (15), they represented just under half (49%) of enrollees with an addiction-related emergency service visit in our study. Rural counties may be particularly critical areas for intervention. Providing better access to local outpatient addiction treatment for Medicaid enrollees in rural communities could improve access to treatment and subsequently reduce the use of emergency services for addiction.

Several study limitations should be noted. First, our study included only data for individuals who received addiction-related treatment in the emergency department. We did not have data for enrollees who did not have an addiction-related emergency service visit. Second, our study was unable to account for whether treatment programs were accepting new patients. It is possible that some treatment programs had limited capacity and a waiting list for treatment. As a consequence, a treatment program could reportedly accept Medicaid enrollees but may not have actually done so during the study period. Third, our findings are generalizable only to South Carolina. Finally, data on the number of outpatient treatment programs that accept Medicaid were available only from 2012; the 2013 data have not been made publicly available. Although estimates of the proportion of addiction treatment programs in South Carolina accepting Medicaid showed very little change from 2012 to 2013 (59.8% in 2012 and 60.4% in 2013), we note the lack of availability of county-level data from 2013 as a limitation in this study.

Despite these limitations, this study provides new evidence of a link between availability of addiction treatment for Medicaid enrollees and reduced odds of repeated use of the emergency department for addiction-related problems.