Abstract
RATIONALE, AIMS AND OBJECTIVES: Emergency department (ED) clinicians account for approximately 13% of all opioid prescriptions to opioid-naïve patients and variability in the rates of prescribing have been noted among individual clinicians and different EDs. This study elucidates the amount of variability within a unified health system (the U.S. Military Health System [MHS]) with the expectation that understanding the sources of variability will enable health system leaders to improve the quality of decision making.
The design was a retrospective cohort study examining variation in opioid prescribing within EDs of the US MHS. Participants were Army soldiers who returned from a deployment and received care between October 2009 and September 2016. The exposure was ED encounters at a military treatment facility. Key measures were the proportion of ED encounters with an opioid prescription fill; total opioid dose of the fill (morphine milligram equivalent, MME); and total opioid days-supply of the fill.
The mean proportion of ED encounters with an opioid fill across providers was 19.7% (SD 8.8%), median proportion was 18.6%, and the distribution was close to symmetric with the 75th percentile provider prescribing opioids in 24.6% of their ED encounters and the 25th percentile provider prescribing in 13.4% of their encounters. The provider-level mean opioid dose per encounter was 113.1 MME (SD 56.0) with the 75th percentile (130.1) 50% higher than the 25th percentile (87.4). The mean opioid supply per encounter was 6.8 days (SD 3.9) with more than a twofold ratio between the 75th percentile (8.3) and the 25th (4.1). Using a series of multilevel regression models to examine opioid fills associated with ED encounters and their dose levels, the variation among providers within facilities was much larger in magnitude than the variation among facilities.
Among ED encounters of Army soldiers at military treatment facilities, there was substantial variation among providers in prescribing opioid prescriptions that were not explained by patient case-mix. These results suggest that programmes and protocols to address less than optimal prescribing in the ED should be initiated to improve the quality of care.