Creatinine Clearance, Cockcroft-Gault Formula and Cystatin C: Estimators of True Glomerular Filtration Rate in the Elderly?

BACKGROUND: The aim of this study was to assess the accuracy and precision of estimators of true glomerular filtration rate (GFR) (Cockcroft formula, measured creatinine clearance (CCR) and a cystatin-C-based estimation) in elderly patients attending a geriatric department. Additionally, parameters influencing GFR in the elderly were evaluated. METHODS: 30 patients aged 57-90 years treated in the Geriatric Department for pulmonary or cerebral diseases were included in the study. Nine patients were diabetic and 16 hypertensive. Exclusion criteria were advanced dementia, acute heart failure and primary renal disease. Inulin clearance (CINU), CCR and estimation by Cockcroft-Gault equation (CG) were performed on the same day. For comparison of the methods an analysis according to Bland and Altman was used, depicting the mean difference between the methods and the limits of agreement of the differences, representing their 95% interval of confidence. Furthermore, the influence of confounding variables on GFR estimation was analyzed by multiple regression. RESULTS: Baseline characteristics showed a median age of 74.5 years and a median body weight of 66.7 kg. Median values for serum creatinine 88.4 micromol/l, 5.74 mmol/l for urea and 1.57 mg/l for cystatin C. CCR (median: 51.6 ml/min) and CG (median: 63.0 ml/min) underestimated CINU (median: 83.3 ml/min). Both methods showed poor precision compared with CINU. The upper limit of agreement of the difference was 101.3 ml/min for CCR and 81.4 ml/min for CG, the lower limit was -33.8 ml/min for CCR and -24.6 ml/min for CG. Among frequently used variables to predict GFR, the reciprocal of serum creatinine and body weight revealed a significant influence but not age or gender. A cystatin-C-based estimation of GFR, derived from regression analysis, did not improve the precision of the estimation of GFR compared to CG. Additionally, the occurrence of diabetes mellitus disclosed a borderline influence on the estimation of GFR. CONCLUSION: CCR is not only inconvenient and time consuming, but also imprecise and inaccurate in the elderly, mainly due to reduced muscle mass and erroneous urine sampling. CG and a cystatin-C-based estimation are slightly more adequate, but overall there is no sufficiently precise formula for GFR estimation in the elderly.