This Article  • Full text  • PDF  • Correction  • Send to a friend  • Save in My Folder  • Save to citation manager  • Permissions  Citing Articles  • Citation map  • Citing articles on HighWire  • Citing articles on ISI (4)  • Contact me when this article is cited  Related Content  • Similar articles in this journal  Topic Collections  • Renal Diseases, Other  • Anemias  • Women's Health  • Women's Health, Other  • Cardiovascular Disease/ Myocardial Infarction  • Alert me on articles by topic

A Simple Prediction Model for Chronic Kidney Disease

Heejung Bang, PhD; Suma Vupputuri, PhD; David A. Shoham, PhD; Philip J. Klemmer, MD; Ronald J. Falk, MD; Madhu Mazumdar, PhD; Debbie Gipson, MD, MSPH; Romulo E. Colindres, MD, MSPH; Abhijit V. Kshirsagar, MD, MPH

Arch Intern Med. 2007;167(4):374-381.

Background  Despite the wide availability and low cost of serum creatinine measurement, at-risk populations are not routinely tested for chronic kidney disease (CKD).

Methods  We used a cross-sectional analysis of a nationally representative, population-based survey to develop a system, SCORED (SCreening for Occult REnal Disease), that uses routinely available demographic and medical information to identify individuals with an increased likelihood of CKD. The analysis included 8530 adult participants in the National Health and Nutrition Examination Surveys conducted from 1999 to 2000 and 2001 to 2002 in the United States. Chronic kidney disease was defined as a glomerular filtration rate less than 60 mL/min per 1.73 m². Univariate and multivariate associations between a comprehensive set of risk factors and CKD were examined to develop a prediction model. The optimal characteristics of the model were examined with internal measures. External validation was performed using the Atherosclerosis Risk in Communities study. A model-based numeric scoring system was developed.

Results  Age (P<.001), female sex (P = .02), and various health conditions (hypertension [P = .03], diabetes [P = .03], and peripheral vascular disease [P = .008]; history of cardiovascular disease [P = .001] and congestive heart failure [P = .04]; and proteinuria [P<.001] and anemia [P = .003]) were associated with CKD. The multivariate model was well validated in the internal and external data sets (area under the receiver operating characteristic curve of 0.88 and 0.71, respectively). A score of 4 or greater was chosen by internal validation as a cutoff point for screening based on the diagnostic characteristics (sensitivity, 92%; specificity, 68%; positive predictive value, 18%; and negative predictive value, 99%).

Conclusion  This scoring system, weighted toward common variables associated with CKD, may be a useful tool to identify individuals with a high likelihood of occult kidney disease.

Author Affiliations: Department of Public Health, Division of Biostatistics and Epidemiology, Weill Medical College of Cornell University, New York, NY (Drs Bang and Mazumdar); and University of North Carolina Kidney Center (Drs Vupputuri, Shoham, Klemmer, Falk, Gipson, Colindres, and Kshirsagar), Department of Epidemiology, School of Public Health (Drs Vupputuri and Shoham), and Department of Medicine, Division of Nephrology and Hypertension, School of Medicine (Drs Klemmer, Falk, Gipson, Colindres, and Kshirsagar), University of North Carolina, Chapel Hill.

THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES

Principles of Screening for Chronic Kidney Disease
Jaar et al.
CJASN 2008;3:601-609.
FULL TEXT  

Validation and Comparison of a Novel Screening Guideline for Kidney Disease: KEEPing SCORED
Bang et al.
Arch Intern Med 2008;168:432-435.
FULL TEXT  

The practical implications of using standardized estimation equations in calculating the prevalence of chronic kidney disease
Quinn et al.
Nephrol Dial Transplant 2008;23:542-548.
ABSTRACT | FULL TEXT