Mental Health Specialist Care: Impact on Emergency Department Visits and Hospitalizations of Adult Medicaid Beneficiaries with Co-occurring Diabetes and Depression

Background. Diabetes and depression are two of most common chronic conditions that tend to co-occur and affect each other’s courses. Co-occurring depression leads to higher ED and inpatient care use compared with those who were not depressed among those with diabetes. Whether mental health specialist care provides better quality of depression treatment has long been a major interest of research. While depression is now commonly treated in primary care, many individuals with depression are seen by mental health specialists, including prescribers and non-prescribers. Further, mental health specialists are not accessible in many areas, as over half of U.S. counties have shortages of mental health specialists. To date, there is only limited evidence on the differences in outcomes of individuals with depression treated in primary care settings and mental health specialty, especially among those with co-occurring physical conditions. It is unclear whether mental health specialist care increase or decrease ED and hospitalizations among those with co-occurring depression and diabetes. First, mental health specialist care could decrease ED/hospitalization by increasing the marginal returns to depression care on health. On the other hand, it is possible that mental health specialist care crowds out diabetes care due to the emphasis on depression issues in preventive outpatient care

Objectives. The objective of this study is to examine whether the use of mental health specialists for depression care affects overall ED visits and hospitalizations. The current study provides further evidence of the role of mental health specialist care on ED visits and hospitalizations by (1) focusing on individuals with co-occurring depression and diabetes and (2) addressing endogeneity caused by unobservable factors affecting both depression treatment modality and healthcare utilization.

Method. This study analyzed claims data from North Carolina Medicaid Analytic Extracts (MAX) during 2006-2011. The estimation sample was restricted to adult beneficiaries aged 18 or older with at least one inpatient or at least two outpatient diagnoses of diabetes and depression during the study period. The final estimation sample includes 22,392 unique persons with co-occurring diabetes and depression during 2006-2011, contributing 76,369 person-years to the analysis. Several instrumental variables (IV) that measure mental health specialist supply were used to estimate Local Average Treatment Effect (LATE). Panel-data analysis techniques, including person-fixed effect and Generalized Estimating Equations (GEE), were also implemented to account for correlation between different observations within the same individual. Outcome included both probabilities and numbers of ED visits and hospitalization days.

Results. Preferred GEE with IV specifications showed that mental health specialist care decreased numbers of ED visits and probability of hospitalization by 0.37 (p<0.01) and 7.0 % (p<0.01). The preferred specification failed to reject the null hypothesis at 5% level that mental health specialist reduces probability of ED visits (p=0.236) and number of hospitalization days (p=0.541).

Conclusions. Mental health specialist care could reduce ED/hospitalization utilization in this specific population. The results from LATE estimation further suggest that decrease in ED/hospitalization utilization might be achieved by increasing the local supply of mental health specialists. Therefore, future policies could target efforts at increasing such supply.