Predictors of persistent opioid use in bladder cancer patients undergoing radical cystectomy: A SEER-Medicare analysis

Purpose To evaluate patient and provider characteristics that predict persistent opioid use following radical cystectomy for bladder cancer including non-opioid naïve patients. Methods Patients undergoing cystectomy between July 2007 and December 2015 were identified using the SEER-Medicare database. Opioid exposure was identified before and after cystectomy using Medicare Part D data. Multivariable analyses were used to identify predictors of the primary outcomes: persistent opioid use (prescription 3–6 months after surgery) and postoperative opioid prescriptions (within 30 days of surgery). Secondary outcomes included physician prescribing practices and rates of persistent opioid use in their patient cohorts. Results A total of 1,774 patients were included; 29% had prior opioid exposure. Compared to opioid-naïve patients, non-opioid naïve patients were more frequently younger, Black, and living in less educated communities. The percentage of persistent postoperative use was 10% overall and 24% in non-opioid naïve patients. Adjusting for patient factors, opioid naïve individuals were less likely to develop persistent use (OR 0.23) while a 50-unit increase in oral morphine equivalent per day prescribed following surgery nearly doubled the likelihood of persistent use (OR 1.98). Practice factors such as hospital size, teaching affiliation, and hospital ownership failed to predict persistent use. 29% of patients filled an opioid prescription postoperatively. Opioid naïve patients (OR 0.13) and those cared for at government hospitals (OR 0.59) were less likely to fill an opioid script along with those residing in the Northeast. Variability between physicians was seen in prescribing practices and rates of persistent use. Conclusions Non-opioid naïve patients have higher rates of post-operative opioid prescription than opioid-naïve patients. Physician prescribing practices play a role in persistent use, as initial prescription amount predicts persistent use even in non-opioid naïve patients. Significant physician variation in both prescribing practices and rates of persistent use suggest a role for standardizing practices.