Prediction of Glomerular Filtration Rate Using Aminoglycoside Clearance in Critically Ill Medical Patients

Abstract

OBJECTIVE: The aim of this preliminary investigation was to evaluate the use of aminoglycoside serum concentrations as a surrogate measure of the glomerular filtration rate (GFR) in comparison with other measured and empiric methods against inulin, the criterion standard measure of GFR. DESIGN: A consecutive sample of all eligible patients. SETTING: An eight-bed medical intensive care unit in a university-affiliated tertiary-care teaching hospital. PATIENTS: Ten critically ill medical patients receiving gentamicin or tobramycin for presumed or documented gram-negative bacillary infection were enrolled in the study. The patients were mechanically ventilated and had underlying organ system dysfunction. All ten patients completed the study. INTERVENTION: Patients underwent renal functional assessment by measured inulin (Clin) and 24-hour urinary creatinine clearance (Clcr). Aminoglycoside serum concentrations were used to estimate GFR and were compared with the two measured methods and a creatinine clearance calculated with the Cockcroft-Gault method (ClCG). All evaluations were performed the same day. RESULTS: Clin averaged 51.6 ± 35.0 mL/min and serum creatinine ranged from 0.3 to 5.4 mg/dL (26.5 to 477.3 μmol/L). Steady-state peak and trough aminoglycoside concentrations were 6.1 ± 1.4 and 1.3 ± 0.9 μg/mL, respectively. There were no statistically significant differences between the various methods, although the aminoglycoside-calculated GFR (Clamg) 95 percent confidence intervals were smaller man Clcr and ClCG compared with Clin. Mean absolute errors were smaller with Clamg than with Clcr and ClCG. Regression results indicated that only Clang and ClCG demonstrated agreement with Clin (lines not different from y=x). However, the Clamg showed closer agreement, with a mean square error almost half that of ClCG (9.6 vs. 18.1). CONCLUSIONS: Clamg can be used routinely as an estimate of GFR in critically ill patients, with less error man empiric methods.