Predicting the Risk of End-stage Renal Disease in the Population-based Setting

A Retrospective Case-control Study

Eric S Johnson; David H Smith; Micah L Thorp; Xiuhai Yang; Juhaeri Juhaeri


BMC Nephrology. 2011;11 

In This Article


The National Kidney Foundation's (NKF) Kidney Disease Outcomes Quality Initiative (K/DOQI) encourages providers to conduct laboratory screening of patients to identify unrecognized chronic kidney disease (CKD) and stratify patients into stages of CKD based on their serum creatinine values--as an estimate of their glomerular filtration rate (GFR)--so that providers can intervene to slow the progression to end-stage renal disease (ESRD) and other outcomes, such as cardiovascular events and mortality.[1] The utility of the NKF guidelines is not limited to management of individual patients. The guidelines offer an opportunity to manage populations of patients, especially within large integrated healthcare systems. The use of patient demographic information, laboratory values, patient education, and comprehensive supportive services can maximize the impact of the NKF guidelines within a large population. Population-based CKD management may reduce the burden of CKD and prevent or delay patients' progression to renal replacement therapy (RRT).

Previous case-control studies that evaluated risk factors for progression to ESRD identified patients on the basis of elevated serum creatinine[2] or treatment with renal replacement therapy[3] and asked participants to recall their medical history with a focus on long-term analgesic use. Interviewing does not work well for detailed clinical characteristics, especially laboratory values measured years earlier.

Investigators using data from Kaiser Permanente Northern California conducted a cohort study among 177,570 patients who had participated in health screening during 1964 and 1973 to evaluate a range of potential baseline predictors of ESRD (as measured by RRT) during a 25-year follow-up.[4] Their study provides a wealth of explanatory insight into predictors, including some that have only recently been recognized, such as lower hemoglobin values and elevated serum uric acid levels. Because their characteristics (predictors) were collected according to a study protocol among consenting study volunteers, and their hazard ratios reflect an induction period of up to 25 years, decision-makers may still have pragmatic questions about predictors collected during routine care (without a protocol), among a broader spectrum of patients (without requiring consent for study participation), over a briefer period. The same pragmatic considerations hold for a large Japanese cohort study, which collected data during the early 1980s.[5]

Two ongoing prospective, observational studies measured detailed clinical characteristics at baseline and continue to follow patients to predict a range of outcomes, including ESRD.[6,7] Both CRIC and STRIDE enrolled patients with known CKD who were identified in the clinic-based setting. However, it's unclear whether either study will be sufficiently powered to identify multivariable predictors of ESRD because those analyses depend on the number of events, not the number of patients. That may be one reason why both CRIC and STRIDE will evaluate a broad range of endpoints (e.g., surrogate measures of renal function, composite clinical endpoints, quality of life, etc).

In earlier work within Kaiser Permanente Northwest (KPNW), we conducted a cohort study restricted to patients with stage 3 or worse CKD to predict ESRD and the start of RRT.[8] While most patients who develop ESRD have CKD, some do not (i.e. patients who suffer acute kidney injury). To address this limitation of our study, as well as the limitations of the other studies cited above, we conducted a retrospective, nested case-control study by sampling within the KPNW population. Our eligibility criteria did not consider whether patients met laboratory criteria for CKD or had clinically recognized (diagnosed) CKD, so we hope the findings on predictors of RRT will be more pragmatic (than explanatory), and consequently of interest to decision-makers in usual care settings.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.