Opioid Use Disorder among Medicaid Enrollees: Snapshot of the Epidemic and State Responses
Data in this brief is from the Medicaid Outcomes Distributed Research Network (MODRN), an initiative of AcademyHealth.1 MODRN is a collaborative effort to analyze data across multiple states to facilitate learning among Medicaid agencies. Participants from AcademyHealth’s State-University Partnership Learning Network (SUPLN) and the Medicaid Medical Director Network (MMDN) developed MODRN to allow states to participate in multi-state data analyses while retaining their own data and analytic capacity.
MODRN is composed of multiple organizations using a common data model to support centralized development, but local execution, of analytic programs. Under MODRN, each state-university partnership adopts the Medicaid Common Data Model, contributes to a common analytic plan, and conducts analyses locally on their own Medicaid data using standardized code developed by the data coordinating center. Finally, the state-university partners provide aggregate results, not data, to the data coordinating center, which synthesizes the aggregate findings from multiple states for reporting. The Medicaid Common Data Model will be continually updated and expanded for future Medicaid research projects.
Eleven university-state partnerships now participate in an effort to provide a comprehensive assessment of opioid use disorder treatment quality in Medicaid. The findings presented in this report resulted from that project that at the time of this writing had been implemented by six university participants include the University of Kentucky, University of Maryland Baltimore County, The Ohio State University, University of Pittsburgh, Virginia Commonwealth University, and West Virginia University.
Below we detail the construction of the variables used in the data analysis across the six study states.
The data analysis covered years 2014 through 2016. Some measures pool data across two-year period per National Quality Forum Specifications.
This analysis includes non-dual, full-benefit Medicaid enrollees age 12-642 with at least one month of Medicaid eligibility in the calendar year.
For analysis by eligibility category, we group enrollees into categories using the following hierarchy:
- Pregnant women, which includes any adolescents or women who are pregnant at any time in the calendar year. We identify women as pregnant during the year either by measuring the gestational period prior a claim for giving birth or by identifying a claim for prenatal care.
- Children, which includes those under the age of 21. In states using Medicaid as the basis of their Children’s Health Insurance Program, this group also includes children qualifying for Medicaid through Title XXI
- Adults age 21-64 qualifying due to receipt of Supplemental Security Income (SSI)
- Adults not qualifying on the basis of disability through a traditional (non-ACA expansion) category
- Adults qualifying through the ACA expansion category
Prevalence of Opioid Use Disorder (OUD)
We identify people with OUD based on diagnosis codes in claims. Specifically, we identify those who had at least one encounter with any diagnosis (counting all diagnosis fields) of OUD in inpatient, outpatient, or professional claims at any time during the measurement period. We used National Quality Forum code sets to identify diagnosis codes for measuring OUD.3
Rates of Medication-Assisted Treatment (MAT) among Enrollees with OUD
After identifying the population with OUD as detailed above, we calculate utilization rates for MAT by identifying individuals with OUD who have at least one claim for medication-assisted treatment for OUD. Specifically, we include those who have at least one claim with a National Drug Code (NDC) or a HCPCS code for any of the following OUD medications during the measurement period:
- Naltrexone (oral or injectable)
- Methadone administration
We excluded claims for oral medications with negative, missing, or zero days’ supply.
Continuity of Pharmacotherapy for OUD
This measure is calculated for three rolling two-year periods from 2014 to 2016: 2014-2015, and 2015-2016, to allow for 180-day measurement of pharmacotherapy for enrollees whose treatment episodes span calendar years. For each two-year period, we limit the analysis to individuals who (1) had a diagnosis of OUD, as described above4 (2) had at least one claim for an OUD medication, as described above, and (3) who are 18-63 years of age5 for the duration of the first year during which they appear in the period. We only include individuals who received oral OUD medications during the two-year period with a date at least 180 days before the end of the final calendar year of the measurement period. Further, we only include individuals who were continuously enrolled in Medicaid for at least 6 months after the month with the first OUD medication claim in the measurement period, with no gap in enrollment. Individuals who are not enrolled for 6 months, including those who die during the period, are not eligible and are not included in this part of the analysis.
Within this group, we measure continuity of treatment by identifying individuals who have at least 180 days of continuous pharmacotherapy with a medication prescribed for OUD without a gap of more than seven days. We developed a set of decision rules for counting surplus for overlaps among prescription claims and for counting length of days for medications with different administration (e.g., prescription OUD medications, Naltrexone injections, and for licensed treatment center-dispensed methadone and office-dispensed buprenorphine/naloxone).6
Emergency department use and inpatient hospitalizations for OUD
We measure emergency department (ED) visits for OUD as distinct ED visits with OUD diagnosis in any diagnosis field. For each enrollee, we consider a distinct combination of billing provider ID and date of service as a distinct ED visit. Similarly, we measure distinct inpatient hospitalization episodes with OUD diagnosis in any diagnosis field. We exclude detoxification and partial hospitalization and count direct transfers from one facility to another (discharge from one inpatient setting and admission to a second inpatient setting within one calendar day or less) as a single hospitalization.
To facilitate comparison of ED visit and inpatient hospitalization rates, we calculate visits/admissions per 1,000 member-months in the time period.