Opioid Use Disorder and Treatment Among Pregnant and Postpartum Medicaid Enrollees
Medicaid Claims (T-MSIS) & State Exclusion Criteria
KFF used the 2016-2019 Transformed Medicaid Statistical Information System (T-MSIS): demographic eligibility base (DE) (2017 and 2018) and both header and line files from inpatient (IP), other services (OT), and long-term care, as well as the prescription (RX) line files from 2016-2019. The 2016-2018 files were Release 2 and the 2019 files were Release 1.
We use 39 states in the main analysis and for maternal age. States with “unusable” data based on state-level information available from the DQ Atlas were excluded. We further excluded states whose final sample in this analysis had a difference of 20% or more from the number of Medicaid-covered births in the CDC Wonder Natality Files and state birth reports. For the analysis on race/ethnicity, we used 24 states with race/ethnicity data that was not considered of “high concern” or “unusable” by the DQ Atlas. Data reported in this analysis are from states with useable and sufficient (n > 50) data. Please see Appendix Table 1 for more information on the measures we used from the DQ Atlas, CDC Wonder Natality Files, and state birth reports to assess states’ data quality.
T-MSIS Enrollee Sample Selection
The sample includes pregnant and postpartum Medicaid enrollees that had a live birth in 2017 and 2018. Enrollees were not included if they did not have a live birth within the study period. This analysis defined the prenatal period as 10 months prior to delivery and the postpartum period includes delivery and 60 days post-delivery. After enrollee and data quality exclusions, the sample includes 2,368,069 enrollees with a live birth from 39 states. Among those with a live birth, 65,092 enrollees had a documented OUD. Out of those with a clinically documented OUD, 35,612 received MOUD.
For the analysis on age, there were 970,458 live births among those 25 years old and younger, 1,084,572 live births among those 26-34 years old, and 313,039 live births among those 35 years old and older across 39 states. Out of those with a live birth, there were 15,234 25 years old and younger, 40,143 between 26-34 years old, and 9,715 35 years old and older with a documented OUD. There were 7,277 enrollees 25 years old and younger, 22,926 enrollees between 26-34 years old, and 5,409 enrollees 35 years old and older that had at least one claim for MOUD.
For the analysis on race/ethnicity, there were 633,477 White live births, 324,546 Black live births, and 527,147 Hispanic live births across 24 states. Out of those with a live birth there were 34,899 White enrollees, 3,447 Black enrollees, and 3,279 Hispanic enrollees who had a clinically documented OUD. Among those with a documented OUD, 19,960 White enrollees, 1,074 Black enrollees, and 1,749 Hispanic enrollees that had at least one claim for MOUD. Persons of Hispanic origin may be of any race but are categorized as Hispanic for this analysis. Other groups are considered non-Hispanic.
Identification of OUD and Treatment in T-MSIS
We link header and line files using BENE_ID and CLM_ID. DE files are linked to header and line files using BENE_ID and if BENE_ID is missing, MSIS_ID. To identify OUD diagnoses and MOUD, we used the Behavioral Health Service Algorithm (BHSA) reference codes provided by Urban Institute (Victoria Lynch, Lisa Clemans-Cope, Paul Johnson, Marni Epstein, Doug Wissoker, and Emma Winiski. Behavioral Health Services Algorithm. Version 3. Washington, DC: Urban Institute, 2022.) which uses ICD-9 and ICD-10 diagnosis codes, procedure codes, service codes, and National Drug Codes (NDCs).
The OUD definition that was used for this analysis included all F11 diagnosis codes, procedure codes for naltrexone unique to OUD (G2073, G2074, G2075), procedure codes for methadone maintenance (HZ91ZZZ, HZ81ZZZ, H0020, G2078, G2067), buprenorphine procedure and NDC codes, and other procedure codes unique to OUD (HZ82ZZZ, HZ92ZZZ, M1032, M1033, M1034, M1035, M1036). One or more occurrence of these codes was coded as an OUD. Modifications to the BHSA algorithm include the removal of NDC codes that were not specific to OUD. Approximately, 2% of pregnant and postpartum Medicaid enrollees had only an Rx prescription and were considered to have an OUD diagnosis.
MOUD treatment included enrollees with at least one of the following: procedure codes for methadone maintenance (HZ91ZZZ, HZ81ZZZ, H0020, G2078, G2067), buprenorphine procedure and NDC codes, and other procedure codes unique to OUD (HZ82ZZZ, HZ92ZZZ, M1032, M1033, M1034, M1035, M1036), procedure codes for naltrexone unique to OUD (G2073, G2074, G2075), and naltrexone NDC codes among those with an OUD diagnosis. Naltrexone is also used for the treatment of alcohol use disorder (AUD) and we did not include those that did not have a corresponding OUD diagnosis code. As previously mentioned, we used reference codes from the Behavioral Health Services Algorithm developed by the Urban Institute.