How Robust are our Methods of Detecting Impaired Glomerular Filtration Rate in the Intensive Care Unit?

BackgroundSerum creatinine is not a sensitive marker to assess early loss of renal function in acute kidney injury. Timed creatinine clearance and several formula used to predict glomreular filtration rate have not been validated.MethodsIn a prospective observational study in 50 adult patients admitted to the intensive care unit with apparent normal renal function, we assessed the glomerular filtration rate by the formula methods and timed creatinine clearance.ResultThe mean serum creatinine was 0.77mg/dl, SD ± 0.15 (range 0.5-1.14 mg/dl). The mean measured creatinine clearance was 87.15 ml/min/1.73m2, SD ± 20.5 (range 56.9-137 ml/min/1.73m2). In 25 (50%) patients, one hour urinary creatinine clearance was <80 ml/min/1.73m2 and in two (4%) patients, the creatinine clearance was <60 ml/min/1.73m2. Spearman correlation coefficient and regression analysis revealed a statistically significant correlation for the Cockcroft-Gault and predictive equations when compared with measured creatinine clearance. The differences between the predictive equations and creatinine clearance, as illustrated by the ±95% confidence interval in the Bland-Altman graphs was very significant [Cockcroft- Gault = −40.3 to 17.7 ml/min/ 1.73m2, Modification of Diet in Renal Disease equation = −46.2 to 30.6 ml/min/1.73m2 and the simplified Modification of Diet in Renal Disease equation = −72.8 to 24.8 ml/min/1.73m2].ConclusionFormula methods and creatinine clearance are more sensitive than serum creatinine in detecting early phase of acute kidney injury. However, there is no agreement between these methods of glomerular filtration rate estimation.