Materials and Methods
We identified three cohorts for study: 1) subjects randomized to FACTT Conservative; 2) subjects randomized to FACTT Liberal; and 3) subjects enrolled in two later ARDS Network studies that used FACTT Lite. We excluded patients on chronic dialysis from the FACTT Lite cohort to match the exclusions of the original FACTT study. Presence of a central venous catheter was an inclusion criterion for FACTT and for the studies included in the FACTT Lite cohort.
Comparison of subjects in the FACTT Conservative and Liberal cohorts was previously reported as part of the FACTT study that enrolled 1,000 subjects from June 2000 to October 2005, with a 60-day mortality of 25.2%. The ARDS Network studies that used FACTT Lite for cointervention control included a clinical trial of an aerosolized [beta]2-agonist for treatment of acute lung injury that enrolled 282 subjects from August 2007 to July 2008. The FACTT Lite cohort included 236 subjects from that study, with a 60-day mortality of 22%. The second study included in the FACTT Lite cohort was the Early Versus Delayed Enteral Feeding and Omega-3 Fatty Acid/Antioxidant Supplementation for ARDS study that enrolled 1,000 subjects between January 2008 and March 2011. The FACTT Lite cohort included 888 subjects from that study with a 60-day mortality of 22.2%.
We obtained data for each group of subjects from prospectively completed case report forms that included age, sex, enrollment Acute Physiology and Chronic Health Evaluation (APACHE) III score, baseline oxygenation variables (after randomization but before study interventions), and baseline CVP. We also collected data on daily fluid intake and output, creatinine, and receipt of vasopressors at baseline and for study days 1–7. Fluid intake included the total volume of crystalloid solutions, colloids, blood products, and enteral feeding solutions. Fluid output included urine, stool, gastric drainage, and any other fluid output.
The primary outcome was cumulative fluid balance over 7 days. In a sensitivity analysis, we excluded subjects in shock at baseline. We defined shock as receiving vasopressors to support blood pressure. Secondary outcomes included daily fluid balance, mean daily furosemide dose, ventilator-free days, and ICU-free days. Ventilator-free days are days alive and free from mechanical ventilation through day 28. ICU-free days are days alive and out of the ICU through day 28. Safety outcomes were the prevalence of acute kidney injury and new onset shock during the study. We also reported 60-day mortality adjusted for age and APACHE III score in order to control for differences in severity of illness among the different studies.
We defined acute kidney injury as an increase in serum creatinine of 50% or an absolute increase of more than 0.3 mg/dL over a 48-hour window during study days 1–7. Because of the significant difference in fluid balance between FACTT Conservative and FACTT Liberal, we used a creatinine adjusted for fluid balance to define acute kidney injury, similar to prior analyses of acute kidney injury in FACTT. For FACTT Lite, baseline creatinine was the lowest value from the 24 hours before randomization. To calculate adjusted creatinine, we estimated the volume of distribution for creatinine on the day of randomization, which is equal to total body water and assumed to be 60% of the subject's total body weight at the time of randomization. For each study day, we calculated cumulative on-study fluid balance using the 24-hour fluid intake and output. Then, we calculated adjusted creatinine = (measured serum creatinine) × (1 + [on-study cumulative net fluid balance/total body water]).
We prospectively collected daily furosemide dosing for the FACTT Conservative and Liberal groups during the FACTT study. For the FACTT Lite cohort, we retrospectively collected furosemide doses from 18 hospitals in 10 of the 12 ARDS Network sites. In a randomly selected subgroup of subjects who were on mechanical ventilation, had a systolic blood pressure greater than or equal to 90 mm Hg, were not on a vasopressor, and had received furosemide, we compared mean daily dose of furosemide administered among the three cohorts. We stratified sampling by sites. We calculated the number of FACTT Lite subjects needed from each site according to a 2.9:1 sampling ratio of 2.9 FACTT Conservative and Liberal subjects to each FACTT Lite subject in order to achieve adequate power. We then randomly sampled daily furosemide doses for that weighted sample of ARDS Network subjects from each participating site.
The Intermountain Healthcare Institutional Review Board, Salt Lake City, Utah, approved a waiver of informed consent for this retrospective study using data from already completed ARDS Network clinical trials and retrospectively collected furosemide dosing data for the FACTT Lite cohort.
Statistical Analysis
Comparisons of continuous variables, such as fluid balance, between FACTT Lite and FACTT Conservative or FACTT Liberal groups were made with a t test. Comparison of continuous variables among all groups was made with a one-way analysis of variance. Comparisons of proportions among FACTT Lite and FACTT Conservative or FACTT Liberal groups were made using a chi-square test. Data are reported as mean ± SEM unless otherwise indicated. A two-tailed α of 0.05 was used as the threshold for determining statistical significance. To account for baseline differences influencing 60-day mortality among the cohorts, we used a forward stepwise regression model to select independent variables important in predicting 60-day mortality (similar to an approach used in a prior ARDS Network study). The identified variables (age and severity of illness) were then used in a bivariate prediction model to generate adjusted 60-day mortality outcomes.