Methodological issues in studying the epidemiology of mild to moderate chronic renal insufficiency

There is increasing interest in studying the epidemiology of subjects with mild to moderate chronic renal insufficiency (CRI), defined as reduced glomerular filtration rate (GFR) not requiring renal replacement therapy. This review discusses some of the methodological challenges presented by the epidemiological study of mild to moderate CRI that have not been adequately addressed in the literature. Issues that relate to defining the prevalence of CRI include between-laboratory differences in serum creatinine (SCr) assays, within-person measurement errors in SCr, and differences in SCr in different demographic groups that are independent of GFR. Issues that relate to examining CRI as an outcome include the choice between a "slope" or "threshold" analysis. Issues that relate to examining CRI as an exposure include the choice of renal function measure (for example, SCr vs. estimated GFR) in multivariable analysis, whether to normalize renal function to body surface area or other body size parameters, potential effect modification of the association between CRI and the outcome and the complex relation between CRI, adverse outcomes, potential confounders and intermediary variables. As we enter an era of more intensive study of mild to moderate CRI, recognition of these potential pitfalls should guide researchers toward improving the quality of epidemiological research in this field.