Variation by default: cesarean section discharge opioid prescription patterns and outcomes in Military Health System hospitals: a retrospective longitudinal cohort study

Abstract

Abstract Background To examine factors associated with post-Cesarean section analgesic prescription variation at hospital discharge in patients who are opioid naïve; and examine relationships between pre-Cesarean section patient and care-level factors and discharge morphine equivalent dose (MED) on outcomes (e.g., probability of opioid refill within 30 days) across a large healthcare system. Methods The Walter Reed Institutional Review Board provided an exempt determination, waiver of consent, and waiver of HIPAA authorization for research use in the present retrospective longitudinal cohort study. Patient records were included in analyses if: sex assigned in the medical record was “female,” age was 18 years of age or older, the Cesarean section occurred between January 2016 to December 2019 in the Military Health System, the listed TRICARE sponsor was an active duty service member, hospitalization began no more than three days prior to the Cesarean section, and the patient was discharged to home < 4 days after the Cesarean section. Results Across 57 facilities, 32,757 adult patients had a single documented Cesarean section procedure in the study period; 24,538 met inclusion criteria and were used in analyses. Post-Cesarean section discharge MED varied by facility, with a median MED of 225 mg and median 5-day supply. Age, active duty status, hospitalization duration, mental health diagnosis, pain diagnosis, substance use disorder, alcohol use disorder, gestational diabetes, discharge opioid type (combined vs. opioid-only medication), concurrent tubal ligation procedure, single (vs. multiple) births, and discharge morphine equivalent dose were associated with the probability of an opioid prescription refill in bivariate analyses, and therefore were included as covariates in a generalized additive mixed model (GAMM). Generalized additive mixed model results indicated that non-active duty beneficiaries, those with mental health and pain conditions, those who received an opioid/non-opioid combination medication, those with multiple births, and older patients were more likely to obtain an opioid refill, relative to their counterparts. Conclusion Significant variation in discharge pain medication prescriptions, as well as the lack of association between discharge opioid MED and probability of refill, indicates that efforts are needed to optimize opioid prescribing and reduce unnecessary healthcare variation.