Background/aim: Measurement of the renal function is critical to follow progression of kidney disease. Short-term and long-term variabilities in these measurements have significant impacts on clinical decision making and clinical trials. The goal of this study was to describe the variability in these measurements and to calculate minimum sample size estimates over varying time frames for clinical trials.
Methods: We studied 44 elderly men with diabetic nephropathy who participated in a clinical trial. Glomerular filtration rate and renal plasma flow were measured by continuous infusion technique with five urine collection periods on two occasions 4 months apart. Protein and creatinine excretion rates were measured in the same specimens. In addition, two consecutive 24-hour specimens every month for 4 months were collected to analyze urine protein, creatinine, urea nitrogen, and electrolytes. A hierarchical random effects model was used to analyze the reproducibility from hour to hour, from day to day, and from month to month.
Results: A total of 824 urine specimens were analyzed, of which 412 constituted specimens collected in the short term and 412 were 24-hour urine collections. Hour-to-hour variation accounted for 45% for urinary clearance of iothalamate, but for only 0.5% of the variability in plasma clearance of iothalamate. Day-to-day variability in 24-hour urinary excretion rates for creatinine was 46% and for protein 10%. Month-to-month variability in 24-hour excretion rates for creatinine was 11% and for protein 19%. The urine protein/creatinine ratio had a day-to-day variability of 2% and a month-to-month variability of 19%. Sample size requirements can be reduced by correcting for urine creatinine for some but not all urinary analytes.
Conclusions: In nephrotic men with diabetic nephropathy, the coefficient of variation in the month-to-month protein excretion rate is 36%. Approximately 28 patients in each arm of two groups are needed to detect a difference in protein excretion rate of 28% (1 g/day in this study). The coefficient of variation in plasma iothalamate clearance over 4 months is 16%. To detect a 10% change in glomerular filtration rate between two groups, 44 patients per group are needed. To be deemed statistically significant, a change in daily protein excretion rate of at least 72% over month(s) is needed in individual patients.
2007 S. Karger AG, Basel