Derivation and Validation of a Renal Risk Score for People With Type 2 Diabetes

  1. Paul L. Drury, MA, MB, BCHIR4
  1. 1School of Population Health, University of Auckland, Auckland, New Zealand
  2. 2Department of Renal Medicine, Auckland District Health Board, Auckland, New Zealand
  3. 3Department of Endocrinology, Counties Manukau District Health Board, Auckland, New Zealand
  4. 4Auckland Diabetes Centre, Auckland District Health Board, Auckland, New Zealand
  1. Corresponding author: C. Raina Elley, c.elley{at}


OBJECTIVE Diabetes has become the leading cause of end-stage renal disease (ESRD). Renal risk stratification could assist in earlier identification and targeted prevention. This study aimed to derive risk models to predict ESRD events in type 2 diabetes in primary care.

RESEARCH DESIGN AND METHODS The nationwide derivation cohort included adults with type 2 diabetes from the New Zealand Diabetes Cohort Study initially assessed 2000–2006 and followed until December 2010, excluding those with pre-existing ESRD. The outcome was fatal or nonfatal ESRD event (peritoneal dialysis or hemodialysis for ESRD, renal transplantation, or death from ESRD). Risk models were developed using Cox proportional hazards models, and their performance was assessed in a separate validation cohort.

RESULTS The derivation cohort included 25,736 individuals followed for up to 11 years (180,497 person-years; 86% followed for ≥5 years). At baseline, mean age was 62 years, median diabetes duration 5 years, median HbA1c 7.2% (55 mmol/mol), 37% had albuminuria, and median estimated glomerular filtration rate (eGFR) was 77 mL/min/1.73 m2. There were 637 ESRD events (2.5%) during follow-up. Models that included sex, ethnicity, age, diabetes duration, albuminuria, serum creatinine, systolic blood pressure, HbA1c, smoking status, and previous cardiovascular disease status performed well with good discrimination and calibration in the derivation cohort and the validation cohort (n = 5,877) (C-statistics 0.89–0.92), improving predictive performance compared with previous models.

CONCLUSIONS These 5-year renal risk models performed very well in two large primary care populations with type 2 diabetes. More accurate risk stratification could facilitate earlier intervention than using eGFR and/or albuminuria alone.

  • Received January 22, 2013.
  • Accepted April 16, 2013.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See for details.