Different Causes of Missing Values in a Randomized Clinical Trial of Kidney Decline: Implications for the Statistical Analysis Plan

• Activity Number: 478 - Missing Data
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Andrzej Galecki* and Cathie Spino and Alessandro Doria and Michael Mauer
• Companies: University of Michigan and University of Michigan and Joslin Diabetes Center and University of Minnesota
• Keywords: RCT; longitudinal study; kidney decline; missing data

Abstract

Preventing Early Renal Loss (PERL) is an ongoing multi-center, double-blind, placebo-controlled, parallel-group randomized clinical trial (RCT) designed to study the effect of lowering serum uric acid on progression of renal decline in Type 1 diabetes (T1D) subjects. A de facto estimand of primary interest is an effectiveness of study treatment at the end of the trial, attributable to treatments assigned at randomization.The primary dependent variable is glomerular filtration rate measured longitudinally on a continuous scale by the plasma clearance of non-radioactive iohexol (iGFR). The primary end point is iGFR at the end of the 2-month drug wash-out period following the 3-year treatment, adjusted for the iGFR at baseline. The proposed primary analysis will be performed using ITT approach by means of a linear model for correlated errors with general/unstructured covariance using all available iGFR measures. During the course of the 3-year study, a number of subjects has discontinued, withdrew consent, died or developed end stage renal disease (ESRD). The iGFR measures will not be available at corresponding visits, resulting in missing values of the primary end point.