Opioid Analgesic Prescribing Prior to Opioid Use Disorder in British Columbia, Canada

Background: The prescription of opioid analgesics (OA) has contributed to the development of opioid use disorder (OUD) in many individuals. We aimed to characterize non-cancer OA prescribing for opioid-naïve individuals in British Columbia (BC), Canada, prior to identification of OUD.


Methods: We conducted a population-based retrospective cohort study using a linkage of six health administrative databases, including people with opioid use disorder (PWOUD) in BC between January 1, 2001 and September 30, 2018 who initiated OA therapy for non-cancer pain prior to OUD identification. Measurements included, dose (morphine milligram equivalent per day (MME/day)), days prescribed and measures of clinical guideline non-concordance for initial OA prescriptions (1) high-dose (≥90 MME/day); (2) long-term (≥7 days prescribed); (3) concomitant sedative prescription, across three time periods (2001-2006; 2007-2012; 2013-2018). We estimated the probability of non-concordant initial prescriptions according to health system contacts within 30 days of first prescription (1) inpatient; (2) non-inpatient acute care; (3) non-acute care, using logistic regression analysis, adjusting for individual characteristics and comorbidities.

Results: Among 66,372 individuals identified with OUD from 2001-2018, 21,331 (32.1%) received opioid analgesics prior to OUD identification. This proportion increased from 3.0% in 2001 to 41.0% in 2011, before decreasing to 34.2% in 2017. Roughly half of prescriptions were attributed to non-acute care visits, peaking at 56.8% in 2007, while the proportion of prescriptions attributable to inpatient visits increased from 19.7% in 2001, to 28.5% in 2017. The overall percentage of individuals receiving high-dose, long-term, and concomitant sedative prescriptions decreased across time periods, and in 2013-2018 were 43.8%, 14.5%, and 10.8%, respectively. In logistic regression analysis, adjusting for individual characteristics, the probability of receiving non-guideline concordant prescriptions declined across all three measures for inpatient and non-inpatient acute care, while remaining stable for non-acute care across time periods. In particular, the predicted probability of receiving ≥7 day initial opioid prescriptions following inpatient visits decreased from 53.3% (95% Confidence Interval: 50.9%, 55.8%) in 2001-2006 to 37.2% (33.9%, 40.5%) in 2013-2018, holding covariates at their mean values with non-acute care as reference group.

Conclusion: A substantial proportion of PWOUD from 2001-2018 in BC received opioid analgesics prior to OUD identification. While guideline concordance has improved in recent years among PWOUD, there remains room for improvement in opioid analgesic monitoring and surveillance, given the importance of initial prescription characteristics in mitigating the transition to non-medical use and use disorder.