Factors Influencing Clinician Decision Making Regarding Opioid Prescribing And Pain Management
Analytics
51 views ◎27 downloads ⇓
Abstract
Opioid overdose deaths have increased substantially over the past fifteen years. However, the position and beliefs of the medical community regarding opioids over time has not been described. Despite the proliferation of guidelines, interventions, and policies aimed at the medical community, the extent to which these actions have impacted clinical decision making and prescribing behavior has not been rigorously studied, accounting for simultaneous interventions and the nested structure of healthcare delivery systems. To address this gap, I characterized the experience of the medical community and measured the multi-level factors influencing opioid prescribing within the context of legislation and clinical decision support interventions. A content analysis of letters to the editor in the Journal of the American Medical Association between 2008 and 2018 demonstrated physicians overall valued balance between pain management and adverse effects of opioid prescribing. Physicians took ownership of their role in the epidemic, but called upon the government and community to help take action to address the issue. I found environmental context and resources to be a relevant theme. These findings framed and grounded my subsequent quantitative analyses. Among patients with an acute musculoskeletal injury in a large healthcare system (n=12,918), I found there was a 17.7% increase in prescriptions written for 7 days or less after the STOP Act was implemented (p<0.001), even after adjusting for the existing trend with an interrupted time series design. After implementation of the STOP Act, opioids were prescribed for less than 7 days in 77.1% of encounters, with 30% of variation accounted for by physician and another 9% by facility. I also assessed the impact of a clinical decision support on safe opioid prescribing, operationalized as a composite score of several behaviors in response to the intervention (e.g., prescribing naloxone, initiating a pain agreement). This intervention had a statistically significant but small impact on the percent of patients with chronic musculoskeletal conditions (n=1,290,746) receiving an opioid, with the percent being 1.6% lower than before the intervention. There was not a change in the average dose of opioid prescriptions associated with the intervention. Overall, the median safe opioid prescribing score in the post-intervention period was 77.1%, with 24% of the variation accounted for by practice site. Collectively, this research and resulting manuscripts present a sophisticated and nuanced understanding of the multi-level factors which influence guideline-concordant opioid prescribing. These data can be used to design and tailor additional interventions for populations where adherence to guidelines is low. The findings also demonstrate the necessity for more advanced modeling to account for the nested organization of healthcare delivery and team-based nature of clinical care, as well as rigorous research to explore the effect of interventions across sectors happening simultaneously to efficiently guide decision making and policy.