Condition-Linked Opioid Prescribed-Medicine Events and Subsequent Emergency Department Use Among Adults With Surgical or Procedural Events: A Longitudinal Medical Expenditure Panel Survey Study.

Background We examined whether an observed opioid prescribed-medicine event linked to the same condition as a surgical/procedural event in year 1 was associated with all-cause emergency department (ED) use in year 2 and whether this association differed by poverty status. These medication events represent medicines reported as purchased or otherwise obtained; the linkage does not establish postoperative timing. Materials and methods We pooled four two-year Medical Expenditure Panel Survey (MEPS) panels: 2017-2018, 2018-2019, 2020-2021, and 2021-2022. Adults were eligible when a year-1 surgical/procedural event was linked to a condition record. The exposure was one or more observed year-1 opioid prescribed-medicine events linked to that same condition, compared to none. The primary outcome was the year-2 annual all-cause ED visit count, and the secondary outcome was any year-2 ED use. Survey-weighted quasi-Poisson and logistic models were adjusted for demographic characteristics, poverty, insurance, year-1 utilization, health status, surgical settings, surgical-event burden, and panel. Adjusted outcomes were standardized over the analytic population. A prespecified interaction compared people below with at or above 200% of the federal poverty level (FPL). Results Among 4,468 eligible adults, 4,454 (99.7%) were complete cases, and 816 of 4,454 (18.3%) had an observed condition-linked opioid prescribed-medicine event. Adjusted mean annual ED visits were 0.336 (95% confidence interval {CI}: 0.275-0.412) with and 0.345 (95% CI: 0.311-0.382) without a linked opioid prescribed-medicine event. The adjusted mean ratio was 0.975 (95% CI, 0.784-1.211; p=0.8165), and the mean difference was -0.009 visits (95% CI, -0.082 to 0.065; p=0.8150). Any year-2 ED use occurred in 194 of 816 exposed participants (survey-weighted, 22.5%; 95% CI, 18.9%-26.1%) and 845 of 3,638 comparators (survey-weighted, 21.5%; 95% CI, 19.7%-23.3%). The exposure-by-poverty interaction did not provide evidence of heterogeneity (ratio of ratios, 1.112; 95% CI, 0.751-1.649; F=0.284; p=0.5943). The adjusted risk difference for any ED use was -1.9 percentage points (95% CI, -5.7 to 1.8; p=0.3073). Conclusions No clear adjusted association was detected between an observed year-1 condition-linked opioid prescribed-medicine event and year-2 annual all-cause ED use. The estimates were imprecise and do not establish equivalence, the absence of an association, or socioeconomic equity. The exposure cannot be classified as postoperative because the procedure-medication sequence was not established.