Abstract
CONTEXT: Opioid prescribing to cancer patients is declining, but it is unknown whether reductions have been tailored to those at highest risk of opioid-related harms. OBJECTIVES: Examine whether declines in opioid dispensing to patients receiving active cancer treatment are sharper in patients with substance use disorder (SUD) or mental health diagnoses. METHODS: We used 2008-2018 national, commercial healthcare claims data to examine adjusted and unadjusted trends in opioid dispensing (receipt of ≥1 fill; average daily dosage; receipt of high-dose opioids; receipt of concurrent opioids and benzodiazepines) to patients ages ≥18 receiving treatment for one of four cancer types (breast; colorectal; head and neck; sarcoma; N = 324,789 patients). To compare declines across subgroups with varying risk of opioid-related harms, we stratified by SUD and mental health diagnosis. To address potential confounding, we estimated subgroup-specific trends using generalized estimating equations, adjusting for covariates. RESULTS: Across groups, rate of ≥1 opioid fill per quarter fell 32.5% (95% CI: 31.8%-33.2%) from 2008 to 2018; daily dose among those receiving opioids fell 37.6% (95% CI: 36.7%-38.6%). In most cases, these declines were not sharper in subgroups at greater risk of opioid-related harms. For example, patients with opioid use disorder experienced the smallest declines in dispensing frequency, and there was no evidence that declines were sharper in patients with mental health diagnoses. CONCLUSION: Sharp declines in opioid prescribing during the drug overdose crisis have affected a wide range of patients undergoing cancer treatment and may not have been sufficiently tailored to patient characteristics. Research on implications for opioid-related harms and pain management is needed.