(T) Intended and Unintended Consequences: Changes in Opioid Prescribing Practices Following Two Policies in North Carolina, 2012–2018 – A Controlled Interrupted Time Series Analysis

Publication Number: 644

Background: The potential misapplication of current opioid prescribing policies remains understudied and may have substantial adverse implications for patient safety.

Objectives: To understand the extent to which unintended prescribing consequences followed implementation of two statewide opioid prescribing policies among opioid-naïve individuals in North Carolina from 2012 to 2018.

Methods: We conducted controlled interrupted time series analyses using autoregressive integrated moving average models to assess level and trend changes in monthly 1) prescribing rates, 2) mean days’ supply, and 3) mean daily morphine milligram equivalents (MME) of incident opioid prescriptions for acute, postsurgical, and chronic pain, relative to two statewide prescribing policies. We examined prescribing patterns for benzodiazepines to control for temporal trends in prescribing of similar medications. Policies included 1) a state medical board initiative to reduce high-dose and high-volume opioid prescribing (May 2016), and 2) legislation to limit initial opioid prescriptions for acute and postsurgical pain (January 2018). We examined outcomes overall and by cancer history using claims data from a large single private health insurance provider. Eligible persons included North Carolina residents, aged 18-64, insured by the provider at any point from January 2012 to August 2018.

Results: Following the investigative initiative, we observed no change in prescribing patterns for patients with chronic pain; conversely, among acute and postsurgical pain patients, we observed immediate declines in mean daily MME (acute: -0.70 [95% CI: -1.33, -0.07]; postsurgical: -2.39 [95% CI: -3.44, -1.34] mean daily MME). Post-legislation, prescription rates did not decline for those with acute, postsurgical, and non-cancer pain, but instead declined among cancer patients with chronic pain (-12.18 [95% CI: -24.03, -0.33] per 10,000 person months). We observed modest trend declines in mean days’ supply for acute and postsurgical pain, but greater declines for chronic pain (-7.25 [95% CI: -9.29, -5.21] days’ supply per year).

Conclusions: We found mixed evidence on the potential impact of two statewide opioid prescribing policies. Some prescribing indicators declined after the policy while others did not; where observed, declines were most often in a group not intended to be impacted by the policy. This study provides evidence of the need for clearer opioid prescribing policies, by pain indication and cancer status, to ensure impacts on intended populations and avoid impacts on unintended populations.