Length/serum creatinine ratio does not predict measured creatinine clearance in critically ill children.

TitleLength/serum creatinine ratio does not predict measured creatinine clearance in critically ill children.
Publication TypeJournal Article
Year of Publication1995
AuthorsFong J, Johnston S, Valentino T, Notterman D
JournalClin Pharmacol Ther
Date Published1995 Aug
KeywordsAdolescent, Adult, Aging, Child, Child, Preschool, Creatinine, Critical Illness, Female, Glomerular Filtration Rate, Humans, Infant, Infant, Newborn, Linear Models, Male

OBJECTIVE: Information regarding renal function is important in critically ill children to adjust the dosage of drugs that are eliminated by the kidneys. Methods for estimating glomerular filtration rate (GFR) based on age and serum creatinine level have shown good agreement with measured creatinine clearance (CLCR) in children without critical illness but have not been examined in critically ill children.

METHODS: CLCR (24 hours) was measured (CLCR-measured) in 100 individuals (aged 5.6 years [range, 0.1 to 20.8 years]) admitted to a pediatric intensive care unit. Urine was collected by indwelling bladder catheters. Serum levels were determined. CLCR was calculated (CLCR-measured) according to the standard formula. GFR was estimated (CL-estimated) according to a published method, in which GFR is based on serum creatinine levels, patient length, and a constant that varies with the age and sex of the child. For each patient, the percentage difference between methods was calculated as the difference between the methods divided by the average obtained by the two methods and expressed as a percentage. Bias was calculated as the absolute value of the percentage difference.

RESULTS: CLCR-measured and CL-estimated were significantly correlated (CLCR-measured = 0.57 CL-estimated + 16.8; r = 0.68; p < 0.001). However, CL-estimated was greater than CLCR-measured in 84 patients. The difference ranged from -230 to +123 ml/min/1.73 m2 (mean -25.9 ml/min/1.73 m2 [95% confidence interval, -18.1 to 33.7 ml/min/1.73 m2]). The mean percentage difference between the methods was also large (-38.1% [95% confidence interval, -47.1% to 29.2%]) and ranged from -153.2% to 102.1%. The mean bias was 45.2% (95% confidence interval, 37.7% to 52.8%). In 36 of 100 patients the discrepancy between the two methods was greater than 50%. Adjusting for weight percentile, as a proxy for abnormal muscle mass, did not improve the model.

CONCLUSION: A method to estimate GFR in children that is based on age and sex, but not critical illness, does not correspond with measured 24-hour CLCR. Use of this method to adjust dosage of drugs eliminated by the kidney might result in significant overdosage in most critically ill children.

Alternate JournalClin. Pharmacol. Ther.
PubMed ID7648769