Objectives: Opioid prescribing trends in medical oncology are poorly defined past 2017, the year after the CDC updated opioid prescription guidelines in non-cancer settings. We aim to characterize opioid prescribing trends by medical oncologists from 2013 to 2019, identify physician-related factors associated with prescribing patterns, and assess whether CDC guidelines for non-oncological settings changed prescribing patterns. Study Design: Retrospective database review. Methods: The CMS Medicare Part D Prescribers-by Provider and CMS Medicare Physician National Downloadable files from 2013 to 2019 were merged by NPI. Sparsely missing demographic data were filled in with public information. The database included physicians’ genders, years of practice, regions, and practice settings. Pearson Chi squared and ANOVA tests with post-hoc Tukey test analysis identified significant differences in demographics and mean opioid prescribing between years. Multivariate binary logistic regression identified significant predictors of total opioid and total long-acting opioid prescriptions. Results: The assembled database consisted of 68040 and 48039 medical oncologists with reported total opioid prescriptions and total long-acting opioid prescriptions, respectively, in 2013-2019. Tukey test revealed no significant difference in mean total opioid prescriptions from 2013 to 2016. Total opioid prescriptions decreased significantly from 2016 to 2017 (p<0.05), as well as 2018 to 2019 (p<0.01). On multivariate binary logistic regression modeling (Table 1), greater total opioid prescriptions were associated with physician male gender (p<0.001), practicing over 10 years (p<0.001), and practice in non-urban areas (p<0.001). Total opioid prescriptions in the five regions were each significantly different; from greatest to least, they were: South, Midwest, West, Northeast, and US Territories (p<0.001). The same patterns were observed with total long-acting opioid prescriptions, except a significant decrease from 2017 to 2018 (p<0.05). Conclusions: Opioid prescriptions from medical oncologists decreased significantly beginning in 2017. Causative factors may include the 2017 designation of the opioid crisis as a public health emergency and the 2016 CDC update of opioid prescribing guidelines in non-cancer settings. These results may imply that the guidelines impacted how medical oncologists, particularly more junior physicians in urban settings, managed chronic cancer pain.
Binary logistic regression for physician characteristics associated with total opioid prescriptions above the 75th percentile.
