Aims: To evaluate the quality of bedside estimation of glomerular filtration rate (GFR) in hospitalized elderly patients.
Methods: We evaluated common estimators of GFR in 29 women and 32 men aged 60 and older hospitalized in a geriatric ward: creatinine clearance (CCR), the Cockcroft-Gault formula (CG), the modification of diet in renal disease formula (MDRD), Baracskay formula (BAR), and a newly developed formula derived recently by us (GCM). Inulin clearance (CINU) was used to assess GFR. Exclusion criteria were mental illness and urinary incontinence.
Results: According to Bland and Altman accuracy and precision of all estimators were low and there was an underestimation of actual GFR: CCR 38.9 ml/min; CG 39.7 ml/min; MDRD 19.8 ml/min; BAR 34.0 ml/min, and GCM: 24.7 ml/min. The accuracy and precision of all methods were even lower in patients with a GFR of >90 ml/min and in patients with diabetes. In receiver-operating characteristics (ROC analysis) all formulas were superior to serum creatinine and overall MDRD disclosed the best results in detecting both a GFR of <90 ml/min and <60 ml/min.
Conclusions: In general, estimation errors are large in an acute care setting. However, formula estimation is clearly superior to serum creatinine and CCR. MDRD gave the best results but may be replaced by the more simple CG and GCM formulas, whereas BAR was inferior.