Abstract
PURPOSE: (i) To report the prevalence of participants to a first-line intervention for OA in Sweden using over-the-counter (OTC) and/or prescribed NSAIDs; (ii) To estimate the accuracy of dispensed prescriptions of NSAIDs recorded in a Swedish health-care register to capture the use of NSAID considering clinician-report as reference standard. METHODS: Register-based study. We used data from OA individuals who participated in the Swedish first-line intervention recorded in the Swedish Osteoarthritis Register (SOAR). SOAR includes clinician-reported use of NSAIDs in the three months preceding the intervention. We used the Prescribed Drug Register to retrieve data on NSAID prescriptions dispensed in the same period. We estimated the prevalence of OTC users (individuals with clinicians-reported use of NSAID but no prescription dispensed), prescription users (individuals with clinicians-reported use of NSAID and a prescription dispensed) and non-users (neither of the previous). We calculated sensitivity, specificity, positive predictive value, and negative predictive value of dispensed prescriptions of NSAIDs vs clinician-report. RESULTS: We included 116,162 individuals (mean age [Standard Deviation]: 66 [9.6] years, 79% women, 77% knee OA). Overall, 24.7% (95% Confidence Intervals [CI] 24.5%; 25.0%) used OTC NSAIDs only, 18.2% (18.0%; 18.5%) used prescribed NSAIDs, 6.6% (6-4%; 6.7%) reported not using NSAIDs while having an NSAID prescription dispensed. Of the 49,913 individuals with clinician-reported use of NSAIDs, 21,190 had a prescription dispensed (sensitivity: 42.5% [95% CI 42.0%, 42.9%]; positive predictive value: 73.5% [73.0%, 74.0%]). Of the 66,249 individuals reporting not using NSAIDs, 58,617 did not have a prescription dispensed (specificity: 88.5% [88.2%, 88.7%]; negative predictive value: 67.1% [66.8%, 67.4%]). CONCLUSION: Overall, 24.7% of participants in a first-line intervention for OA used OTC NSAIDs only while 18.2% used prescribed NSAIDs. Dispensed prescriptions of NSAIDs have high specificity but low sensitivity and can correctly identify about 70% of both the non-users and users in this population.