Fig. 1: Validation of QR model: comparison of development and validation cohorts.
From: A standardized metric to enhance clinical trial design and outcome interpretation in type 1 diabetes
Placebo/control participants used for QR development (n = 5 studies) and validation (n = 8 studies). The published QR model (QRi = ln (Cp1year,i + 1) − 0.812⋅ln (Cp0,i + 1) − 0.00638⋅Agei + 0.191) from the development studies (n = 5) was applied to the validation studies (n = 8). Since the QR model was derived from the development cohort, it was expected that the QR distribution would be centered at zero. Thus, to assess how well the model fit the validation cohort, we compared the mean and distribution to that of the development cohort. No significant difference was observed between the development and validation cohorts when comparing group means using a two-tailed t-test (t = 0.8, DF = 346, p = 0.43); both cohorts have similar distributions as evaluated by a Kolmogorov-Smirnov test (KS = 0.04, p = 0.62). Mean and SD are indicated by black lines and dark shaded region; violin plots visualize data distribution.
