Prediction Aided Tapering In rheumatoid arthritis patients treated with biOlogicals (PATIO): a randomised controlled trial

The primary outcome was the flare rate per patient. Secondary outcomes included any flare, major flares (>12 weeks), DAS28-CRP over time, mean bDMARD dose fraction, infections, and patient-reported outcomes. Results: A total of 162 patients were included (81 per group).

Over 18 months, the mean number of flares per patient was 1.00 (95% CI 0.73–1.27) in the prediction-aided group and 0.73 (95% CI 0.51–0.95) in the DGDO group (rate ratio 1.32, 95% CI 0.90–1.957. Any flare occurred in 57% of prediction-aided and 44% of DGDO patients, and major flares in 22% and 15%, respectively. DAS28-CRP trajectories, mean bDMARD dose reduction, patient-reported outcomes, and infection rates were similar between strategies.

Non-adherence to tapering advice occurred in 23% of visits: 72% of these regarded advice to continue tapering, and 28% advice to stop tapering or escalate medication. Conclusion: Prediction-aided bDMARD tapering using this model did not improve flare rates or medication use. Prediction models require sufficient, robust accuracy and prospective evaluation before implementation.