Full, Partial, or Modified: Understanding Relationships Between Permutations of the Mayo Score in Real-World Ulcerative Colitis Patients

Abstract

Introduction: The Mayo Score is a composite instrument that consists of four sub-scores: the stool frequency (SF), rectal bleeding (RB), endoscopic subscores (ES), and the physician's global assessment (PGA). A recent ES may not be available at clinical visits, while the PGA is no longer accepted by the FDA for efficacy assessment in clinical trials. This analysis aims to understand the relationships between individual components and various permutations of the Mayo Score in patients with ulcerative colitis (UC). Methods: Data from a 2015 cross-sectional, multi-national survey of Inflammatory Bowel Disease (IBD) patients and gastroenterologists (GIs) conducted in Europe (EU) and the United States (US) were analyzed. GIs completed patient record forms related to treatment practice and clinical characteristics for adult UC patients. Mayo Scores were calculated from physician reported SF, RB, ES, and PGA. The Partial Mayo (pMayo) was calculated omitting ES, and the modified Mayo (mMayo) omits the PGA. Correlations between Mayo permutations and sub-scores were conducted using Pearson's R and Spearman's Rho. Ordered logistic regression was used to predict the omitted item from the pMayo and mMayo and weighted kappa analysis was used to measure agreement between predicted and observed values. Results: Data were available for 2,723 UC patients (EU 2035, US 688). Patients had a mean age of 41.5 years and 53.5% were male. Strong positive associations between the Mayo items (Table 1) and strong positive correlations between Mayo permutations were observed (Table 2). Ordered logistic regression using the proportional odds model shows an increase in pMayo score is associated with increased odds of a higher ES (OR= 4.606, p<0.001). A high degree of agreement was observed between predicted and observed endoscopic results from this model (kappa = 0.692, p<0.001). An increase in mMayo score was associated with increased odds of a higher PGA (OR= 6.172, p<0.001). A strong level of agreement was also found between predicted and observed PGA (kappa = 0.728, p<0.001). Similar significant patterns were observed in the US and EU cohorts. Conclusion: Significant correlation exists between the pMayo, mMayo and full Mayo Score. Moreover, the remaining components of the pMayo and mMayo may be used as predictors of the ES and PGA, respectively. These findings support the use of the pMayo in clinical practice and the mMayo score in clinical trials, as proxies for the full Mayo Score.684_A Figure 1 No Caption available.684_B Figure 2 No Caption available.