Is CKD in Your Future?
Abstract & commentary
By Allan J. Wilke, MD
Synopsis: These investigators derived and validated a risk score that predicts incident kidney disease.
Source: Kshirsagar AV, et al. A simple algorithm to predict incident kidney disease. Arch Intern Med 2008;168:2466-2473.
Building on their previous work, this group of investigators from the University of North Carolina's Kidney Center combined two data sets, the Atherosclerosis Risk in Communities (ARIC) Study and the Cardiovascular Health Study (CHS) to derive a risk score that would predict future chronic kidney disease (CKD) in the general population. ARIC enrolled 15,732 participants and followed them for a maximum of 9 years. CHS enrolled 5888 participants and followed them for a maximum of 10 years. The participants had the usual demographic information, clinical findings, and health history recorded. Some common laboratory tests, including hemoglobin, lipid profile, and serum creatinine were obtained. The subjects' glomerular filtration rate (GFR) was estimated with the formula from the Modification of Diet in Renal Disease Study, which uses as variables creatinine, age, race, and gender. CKD was defined as a GFR < 60 mL/min/1.73 m2.
After proper exclusion, which included missing data and baseline renal insufficiency, 14,094 participants were left in the combined data set. A total of 9470 were randomly assigned to the development data set and 4624 to the validation data set. A little more than 11% of the combined data set developed CKD during follow-up. Performing multiple logistic regressions on the development data set, age, diabetes mellitus, peripheral vascular disease, anemia, female sex, white race/ethnicity, systolic blood pressure, history of coronary vascular disease, history of heart failure, and low HDL-cholesterol concentration (≤ 40 mg/dL) were identified as predictors of CKD. The investigators looked at two models, one which used all 10 of the identified predictors and the other which dropped white race/ethnicity and low HDL-cholesterol level. Each identifier, except age, was assigned a risk score of 1. Age was divided into three intervals, 50-59, 68-69, and ≥ 70 years, and assigned 1, 2, and 3 points, respectively. When both models were applied to the validation data set, they performed similarly. The authors recommend using cut points of 5 for the full model and a cutoff of 3 for the simplified model. These cut points would identify 20% and 13% of participants progressing to CKD in the next 10 years, respectively.
Commentary
Back in 2007, Bang1 published (and Internal Medicine Alert2 reviewed) this team's development of a risk score for predicting CKD. So why are we looking at this again? Partly, it's my fascination with risk scoring tools, but more importantly, SCORED looks at prevalence and predicts who has CKD. SCORED also used a different set of variables, including the presence of proteinuria. The current algorithm tries to predict who will develop CKD. Presumably, this knowledge would allow patients to make lifestyle changes and physicians to intervene to address risk factors. The authors liken these risk score models to the Framingham risk score for predicting cardiovascular disease in the next 10 years. Interventions are based on a patient's risk score and those patients scoring > 20% are recommended for pharmaceutical and lifestyle interventions. The authors envision their CKD risk scores being used in physician offices, on medical web sites, and at health fairs to identify patients at risk.
In this combined data set, body mass index (BMI) was 27 kg/m2 in participants without CKD and 28 kg/m2 in those with CKD. In a recent study, researchers from Tufts University, using the same data sets as these authors, showed that waist-to-hip ratio (WHR), but not BMI, was associated with progression to CKD and mortality.3 Perhaps future refinements of these risk scores will incorporate WHR.
Of the risk factors listed, only anemia, hypertension, diabetes, and low HDL are modifiable and of these, only treatment of hypertension and diabetes has been shown to slow the progression to CKD. Anemia is more likely the result of CKD than a cause. Age is the most potent risk factor, and in the simplified model, being 70 years or older garners enough points by itself to identify a patient at risk. This suggests that we skip the simplified model and use the full one for our elderly patients. Of course, more than half of our elderly patients would get another point by virtue of being female, but in the full model that combination is still not enough to reach the cut point of 5. An elderly patient who has no other risk factors is likely to be healthy in general and unlikely to develop CKD in the next 10 years.
Family physicians and primary care internists are expected to manage chronic diseases, but we cannot manage what we do not detect. Risk scores like this one help us identify patients who may need our attention.
References
1. Bang H, et al. SCreening for Occult REnal Disease (SCORED): A simple prediction model for chronic kidney disease. Arch Intern Med 2007;167:374-381.
2. Wilke AJ. Score one for the kidneys. Intern Med Alert 2007;29:67-68.
3. Elsayed EF, et al. Waist-to-hip ratio, body mass index, and subsequent kidney disease and death. Am J Kidney Dis 2008;52:29-38.
These investigators derived and validated a risk score that predicts incident kidney disease.Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.