POS0468 A PREDICTION MODEL FOR SWITCHING OF BIOLOGICS AND TSDMARDS IN RHEUMATOID ARTHRITIS

Background Patients with rheumatoid arthritis (RA) have many options for biologic therapy both within treatment classes and across differing mechanisms of action, but the lack predictability response failure can be frustrating patients physicians. Rheumatologists longitudinally collect patient physician reported measurements as a part routine clinical care that used prediction outcomes [1] . Harnessing data regularly collected to monitor RA inform decision making could improve care. Objectives To develop readily usable model in at preceding visits predict probability switching subsequent clinic visit Methods were drawn from CorEvitas (formerly CORRONA) registry adults over age 18 diagnosis RA. The study matched who switched biologics control had not biologics; cohort was divided into training test set development validation. Switchers: initiating biologic/tsdmard another biologic/tsDMARD least two prior switch while on drug. Visits between 2 12 months before. Control visit: Patient still initiated drug drug; next discontinuation or switch. A time initiation by class experience. Pairs control-switch (60%) (40%). Using set, best subset regression, lasso elastic net methods determine potential models. Area under ROC curve final selection estimated coefficients this applied switching. Change disease activity measures non-linear association fit using linear splines knot zero. Results total 5050 included, 3016 2034 dataset. average 59.6, majority female (3998, 79.2%), duration 12.8 years. included CDAI category, pain measurement, change baseline, group, number which significantly associated (either positively negatively) (Table 1). area 0.690 dataset (Figure then similar performance; 0.687 Conclusion We developed simple following visit. This incorporated electronic medical records an application give clinicians more information about their patient’s trajectory likelihood failing biologic. References [1]Novella-Navarro M et al. Predictive identify multiple biological arthritis. Ther Adv Musculoskelet Dis. 2022 Oct 6;14:1759720X221124028. Table 1. Prediction Predictor Odds Ratio 95% CI coefficient Remission (ref) 1 Low 1.75 (1.390 2.203) 0.559 Moderate 2.394 (1.845 3.107) 0.873 Severe 2.964 (2.141 4.103) 1.086 Pt Pain (0-100) 1.01 (1.006 1.013) 0.010 Age (yrs) ≤40 41-50 0.958 (0.684 1.341) -0.043 51-60 0.823 (0.603 1.123) -0.195 61-70 0.747 (0.549 1.017) -0.291 >70 0.606 (0.436 0.843) -0.501 Prior no. biologics/tsDMARDs 0 0.904 (0.749 1.090) -0.101 0.788 (0.630 0.984) -0.239 3+ 0.706 (0.558 0.893) -0.348 (spline zero) <0 1.011 (1.004 1.018) 0.011 >0 1.044 (1.022 1.067) 0.043 CDAI: index; DMARD: modifying anti-rheumatic drugs; ts: targeted synthetic; ref: reference Acknowledgements: NIL. Disclosure Interests Laura Cappelli Grant/research support from: Bristol-Myers Squibb, George Reed Consultant of: CorEvitas, Corrona Research Foundation, Joel Kremer Shareholder Lilly.