Methods Study Design
Adult patients in the medical ICU who received a continuous furosemide infusion with and without 25% albumin were studied retrospectively. The study was conducted at Arizona Health Sciences Center, a tertiary care, academic medical center. The local institutional review board reviewed and approved the study. Primary end points were urine output and net fluid loss.
The diuretic effect of a continuous infusion of furosemide peaks approximately 3 hours after the infusion is started. Therefore, cumulative urine output at 6 hours was chosen as the initial and primary end point to ensure measurement of the full diuretic effect of the furosemide infusion.
Selection of Patients
Any adult patient admitted to the medical ICU between January 1, 2007, and August 31, 2010, who received a sequential continuous furosemide infusion for at least 6 hours and a combined furosemide plus 25% albumin infusion for at least 6 hours was included in the study. The order of infusion did not matter so long as no gap occurred between the 2 sequential infusions. Patients were excluded if they did not have sequential infusions of furosemide and furosemide plus albumin, received furosemide or furosemide plus albumin before the sequential infusions, had incomplete intake and output records, or had renal dysfunction (defined as serum creatinine level greater than 1.5 mg/dL; to convert to micromoles per liter, multiply by 88.4), or any current renal disease (eg, acute tubular necrosis, glomerulonephritis).
Data Collection
Baseline data collected included age, sex, height, weight, serum level of creatinine, serum level of albumin, diuretic medications, ICU diagnosis and underlying illnesses, ratio of PaO2 to fraction of inspired oxygen, and scores on the Acute Physiology and Chronic Health Evaluation II and the Sequential Organ Failure Assessment. In addition, daily measurements of serum levels of albumin, furosemide and albumin dosing, fluid intake, and urine output at the first 6 hours (0–6 hours), and every 6 hours up to 48 hours (if data were available) in each group were recorded.
Statistics
On the basis of data on urine output in critically ill medical patients, it was estimated that 22 patients would be needed to detect a 30% difference in urine output between the furosemidealone and the albumin-plus-furosemide groups with 80% power and α= .05. However, it was decided to enroll at least 30 patients to decrease the risk of a type II error.
Continuous data for comparison of the 2 groups of patients were evaluated by using paired t tests or repeated-measures analysis of variance. A 2-sample t test was used to evaluate unpaired data. Linear regression analysis was used to investigate relationships between the dependent variable urine output and demographic independent variables (age, weight, severity of illness) and other independent variables (ie, furosemide or albumin dose, fluid intake, and serum albumin concentrations) at 24 and 48 hours for both groups of patients. Significance was defined as P < .05 unless otherwise noted (Bonferroni correction for post hoc testing). All data are reported as mean and standard deviation.