Mohsen Sadatsafavi, Paloma Burns, Mehrshad Mokhtaran, Jeenat Mehareen, Alexander J. Lin, Shelley Abercromby, Paul Gustafson, Chris Carlsten, Samir Gupta, Andrea Gershon, Nick Bansback, Annalijn I. Conklin, Mary A. De Vera, Janice M. Leung, Stephen van Eden, Kevin I. Duan, Nawaf Alotaibi, Wan Tan-Hogg, Don D. Sin
Introduction: Clinical guidelines are developed for widespread adoption. However, alignment with guideline recommendations is often low, partly because they are not embedded in routine care.
Aims: To evaluate whether a computerised decision support (CDSS) integrating COPD management guidelines and exacerbation risk prediction improves guideline-concordant pharmacotherapy.
Methods: We conducted a stepped-wedge cluster randomised trial involving 22 respirologists (clusters) across two teaching hospitals. The intervention was an on-screen CDSS that predicted 12-month exacerbation risk using the ACCEPT model and recommended pharmacotherapy according to the 2023 Canadian COPD guidelines. All clusters began in the control arm (routine practice) and crossed over in randomised order to the intervention arm. The primary endpoint was alignment between recommended and prescribed treatment classes.
Results: Between 2023/3 and 2025/6, clusters contributed 1,111 outpatient COPD visits (mean age 70.9y; 36% female). Guideline alignment was 38.6% during control and 49.9% during intervention periods. The intervention significantly improved alignment (adjusted odds ratio 1.70 [95% CI 1.08–2.70]). The largest improvement occurred when triple therapy was recommended (Figure). No significant effect modification was observed for age, sex, or GOLD severity grades.
Conclusions: Integrating risk prediction and guideline recommendations into routine care improves alignment with guideline-based pharmacotherapy in COPD.