The long-term effects of non-steroidal antiinflammatory drugs in osteoarthritis of the knee: a randomized placebo-controlled trial

Background Non-steroidal anti-inflammatory drugs (NSAIDs) are widely used t o treat osteoarthritis (OA), though their long-term efficacy is uncertain. We report a comparison of the symptomatic responses to therapy with tiaprof enic acid, indomethacin and placebo over 5 yr. Methods. A parallel-group, randomized, single-blind trial of patients with knee OA recruited 812 patients from 20 centres; 307 patients received tiapr ofenic acid (300 mg b.d.), 202 indomethacin (25 mg t.d.s.) and 303 matching placebo for up to 5 yr. At the end of the parallel-group study, patients r eceiving tiaprofenic acid or placebo entered a 4-week blinded cross-over st udy of tiaprofenic acid or placebo, both given for 2 weeks. Assessments wer e at baseline, 4 weeks, then at 6-month intervals for up to 5 yr in the par allel group study and at 2-week intervals in the cross-over study. They com prised pain scores, duration of morning stiffness, patients' global assessm ents, paracetamol consumption, adverse reactions, withdrawals and functiona l outcomes. Results. There were significant falls in overall pain scores in patients re ceiving NSAIDs compared with placebo at 4 weeks in the parallel-group phase . Thereafter there were no advantages favouring active therapy. In the cros s-over phase, pain scores were significantly lower in patients receiving ti aprofenic acid than placebo. Patients who had been receiving long-term tiap rofenic acid showed significant rises in their pain scores when receiving p lacebo therapy and vice versa. Adverse events were reported by 61% of patie nts receiving tiaprofenic acid, 63% on indomethacin and 51% on placebo. Pot entially severe side-effects were rare: for example, there were only three cases of gastrointestinal bleeding on NSAIDs. The pattern of withdrawal was similar in patients taking NSAIDs and placebo in the parallel-group study; at 48 weeks 53% of the patients remained on tiaprofenic acid, 50% on indom ethacin and 54% on placebo. Conclusions. NSAIDs significantly reduce overall pain over 4 weeks. This sh ort-term responsiveness is retained, and even after several years of therap y with tiaprofenic acid pain scores increased over 2 weeks when it was chan ged to placebo. Our results do not show long-term benefits from the use of NSAIDs in OA and the majority of patients had persisting pain and disabilit y despite therapy.