Measuring Exposure to Opioids Using Self-Reported Medication Use Data Versus General Practitioner Prescription...

The ongoing global increase in opioid use necessitates studies examining long-term health impacts. Prospective cohorts frequently rely on self-reported medication use data which may be subject to several types of bias compared to more objective measurements. We evaluated the agreement between two opioid exposure measures in the UK Biobank (UKBB)-self-reported regular use and prescription-based indicators using linked general practitioner (GP) records.

Our analysis included 171 813 UKBB participants with linked prescription records. At baseline, participants reported medications taken regularly (weekly, monthly, every 3 months). We assessed agreement between self-reported regular opioid use and opioid prescription records prior to enrollment across various look-back periods and prescription counts. Logistic regressions assessed factors associated with omission and commission.

Agreement was moderate to substantial between self-reported opioid use and prescription records. The strongest agreement was observed for ≥ 3 prescriptions in the past 365 days (Cohen's Kappa = 0.66). Subgroup analysis showed better agreement for analgesic opioids (Kappa = 0.43-0.63) than for non-analgesic opioids (Kappa = 0.25-0.34). Omission odds were highest with increasing months since the last record and lowest for individuals with chronic pain. Commission odds were highest for individuals with chronic pain and lowest in married/partnered individuals.

This analysis indicates that self-reported regular opioid use in the UKBB could be a valid indicator for identifying individuals with repeated prescriptions for analgesic opioids in the past year, while also capturing opioid use from non-linked sources. However, agreement was low for non-analgesic opioids, suggesting limited utility of self-report for capturing use of these medications.