Importance The established chronic kidney disease (CKD) progression endpoint, end-stage renal disease (ESRD) or doubling of serum creatinine (corresponding to a change in estimated glomerular filtration rate (eGFR) of −57% or greater) is a late event, limiting feasibility of nephrology clinical trials. Objective To characterize the association of decline in eGFR with subsequent progression to ESRD, with implications for using lesser declines in eGFR as potential alternative endpoints for CKD progression. Since most people with CKD die before reaching ESRD, we also investigated mortality risk. Data Sources Individual meta-analysis of up to 1.7 million participants with 12,344 ESRD events and 223,944 deaths from 35 cohorts. Study Selection Cohorts in the CKD Prognosis Consortium with a repeated measure of serum creatinine over 1-3 years and outcome data. Data Extraction and Synthesis Transfer of individual participant data or standardized analysis of outputs for random effects meta-analysis took place between July 2012 and September 2013 with baseline eGFRs during 1975-2012. Main Outcomes and Measures ESRD (initiation of dialysis or transplantation) or all-cause mortality risk related to percent change in eGFR over 2 years adjusted for potential confounders and first eGFR. Results The adjusted hazard ratios (HR) of ESRD and mortality were exponentially higher with larger eGFR decline. Among participants with baseline eGFR <60 ml/min/1.73m2, the adjusted HRs for ESRD were 32.1 (95% CI 22.3-46.3) and 5.4 (4.5-6.4) for −57% and −30% eGFR changes, respectively. However, changes of −30% or greater were much more common than changes of −57% (6.9% (6.4-7.4%) vs. 0.79% (0.52-1.06%) in the whole consortium). This association was strong and consistent across length of baseline (1 or 3 years), baseline eGFR, age, diabetes status, or albuminuria. Average adjusted 10-year risk of ESRD for eGFR changes of −57%, −40%, −30% and 0% were 99% (95-100%), 83% (71-93%), 64% (52-77%), vs. 18% (15-22%) respectively at baseline eGFR of 35 ml/min/1.73m2. Corresponding mortality risks were 77% (71-82%), 60% (56-63%), 50% (47-52%), vs. 32% (31-33%), showing a similar but weaker pattern. Conclusions and Relevance Declines in eGFR smaller than doubling of serum creatinine occur more commonly and are strongly and consistently associated with the risk of ESRD and mortality, supporting consideration of lesser declines in eGFR, such as 30% reduction over 2 years, as an alternative endpoint for CKD progression.

No abstract

Clinical prediction models are used frequently in clinical practice to identify patients who are at risk of developing an adverse outcome so that preventive measures can be initiated. A prediction model can be developed in a number of ways; however, an appropriate variable selection strategy needs to be followed in all cases. Our purpose is to introduce readers to the concept of variable selection in prediction modelling, including the importance of variable selection and variable reduction strategies. We will discuss the various variable selection techniques that can be applied during prediction model building (backward elimination, forward selection, stepwise selection and all possible subset selection), and the stopping rule/selection criteria in variable selection (p values, Akaike information criterion, Bayesian information criterion and Mallows’ Cp statistic). This paper focuses on the importance of including appropriate variables, following the proper steps, and adopting the proper methods when selecting variables for prediction models.