Estimation of the Harvey Bradshaw Index from the Patient-Reported Outcome 2 in Crohn's Disease: Results Based on a Large Scale Randomized Controlled Trial

Abstract

Background: Many registrational trials in Crohn's disease assess treatment efficacy with the 2-item Patient-Reported Outcome (PRO2), while the Harvey-Bradshaw Index (HBI) is prominent in pragmatic trials and clinical practice. The translation between PRO2 and HBI has not been established.

Methods: Data from a Phase 3 trial of vedolizumab in Crohn's disease were used to determine the Pearson correlation between PRO2 and HBI. Linear regression was used to fit equations that estimate between indices; 95% prediction intervals were determined for HBI scores corresponding to PRO2 thresholds for disease activity. Internal validation of the conversion equations was performed using the bootstrap methods.

Results: PRO2 and HBI were highly correlated at baseline (r = 0.75 95% confidence interval (CI) 0.73-0.78; P < .001), induction (r = 0.87; 95% CI, 0.85-0.88; P < .001), and maintenance (r = 0.88; 95% CI, 0.85-0.90; P < .001). PRO2 and HBI change scores were moderately correlated (r = 0.72; 95% CI 0.69-0.75; P < .001) in induction and more strongly correlated during maintenance (r = 0.81; 95% CI 0.78- 0.84; P < .001). Regression equations for conversion of PRO2 to HBI from all cohorts (induction, maintenance, randomized, open-label) support an approximate conversion where HBI = 0.5 PRO2. As expected from the imperfect correlation between scores, the prediction intervals were generally wide. No evidence of overfitting was seen in bootstrap internal validation.

Conclusions: PRO2 and HBI correlate strongly and conversion between them is possible. These findings facilitate the practical application of trial results and clinical guidelines.

Keywords: Crohn’s disease; Harvey–Bradshaw Index; correlation; patient-reported outcome; symptom scores.