Prediction of Real-World Drug Effectiveness Prelaunch: Case Study in Rheumatoid Arthritis. (August 2018)

• Record Type: Journal Article
• Authors: Didden, Eva-Maria; Ruffieux, Yann; Hummel, Noemi; Efthimiou, Orestis; Reichenbach, Stephan; Gsteiger, Sandro; Finckh, Axel; Fletcher, Christine; Salanti, Georgia; Egger, Matthias
• Is Part Of: Medical decision making. Volume 38:Number 6(2018)
• Page Start: 719
• Page End: 729
• Publication Date: 2018-08
• Subjects: effect modifier -- efficacy-effectiveness gap -- prognostic factor -- prediction model -- rheumatoid arthritis -- treatment predictor; Medical policy -- Periodicals; Clinical medicine -- Decision making -- Periodicals; Medicine -- Periodicals; Médecine clinique -- Prise de décision -- Périodiques; 362.1
• DOI: 10.1177/0272989X18775975
• Languages: English
• ISSNs: 0272-989X

Abstract

Background. Decision makers often need to assess the real-world effectiveness of new drugs prelaunch, when phase II/III randomized controlled trials (RCTs) but no other data are available.Objective. To develop a method to predict drug effectiveness prelaunch and to apply it in a case study in rheumatoid arthritis (RA).Methods. The approach 1) identifies a market-approved treatment ( S ) currently used in a target population similar to that of the new drug ( N ); 2) quantifies the impact of treatment, prognostic factors, and effect modifiers on clinical outcome; 3) determines the characteristics of patients likely to receive N in routine care; and 4) predicts treatment outcome in simulated patients with these characteristics. Sources of evidence include expert opinion, RCTs, and observational studies. The framework relies on generalized linear models.Results. The case study assessed the effectiveness of tocilizumab (TCZ), a biologic disease-modifying antirheumatic drug (DMARD), combined with conventional DMARDs, compared to conventional DMARDs alone. Rituximab (RTX) combined with conventional DMARDs was identified as treatment S. Individual participant data from 2 RCTs and 2 national registries were analyzed. The model predicted the 6-month changes in the Disease Activity Score 28 (DAS28) accurately: the mean change was -2.101 (standard deviation [SD] = 1.494) in the simulated patients receiving TCZ and conventional DMARDs compared to -1.873 (SD = 1.220) in retrospectively assessed observational data. It was -0.792 (SD = 1.499) in registry patients treated with conventional DMARDs.Conclusion. The approach performed well in the RA case study, but further work is required to better define its strengths and limitations.