Study design and setting
A single-center, prospective cohort study was conducted at King Faisal Specialist Hospital and Research Center in Saudi Arabia-Jeddah between August 01, 2021, and April 30, 2022. The hospital institutional review board approved the study.
Patients
Enrollment criteria include patients 18 years old and older admitted to the medical intensive care unit (MICU) or surgical intensive care unit (SICU) and received IV fluids, including colloids and/or crystalloids, as a replacement or resuscitation. Exclusion criteria included the following: Age < 18 years old, patients with end-stage renal disease, patients on hemodialysis, patients on diuretics upon ICU admission, patients on maintenance IV fluids > 72 hours before ICU admission, patients started on renal replacement therapy 24 hours post ICU admission, patients started on DKA/HHS protocol or those with an inability to measure their fluid output (Appendix figure e1).
Data Collection
Data screening occurred each Sunday of each week, excluding holidays. All patients who fulfilled the inclusion criteria were considered for inclusion. We followed patients until the day of de-resuscitation by diuretics, RRT, or ICU discharge, whatever comes first. We collected the total amount of fluid administered from IV resuscitation fluids, IV maintenance fluid, IV medications, and total parenteral nutrition (TPN) at 24 hours, 48 hours, 72 hours, and until the time of de-resuscitation or ICU discharge. Moreover, we collected information on patients' comorbidities, severity score (APACHE II), the occurrence of acute kidney injury (AKI), use of mechanical ventilation (MV), the number of organs failure, and fluid balance. Each patient's data were collected and handled using Research Electronic Data Capture (REDCap®) software hosted by King Faisal Specialist Hospital and Research Center (KFSHRC).
Outcomes
This study's primary outcome of interest is the median percent fluid accumulation on the day of fluid de-resuscitation or ICU discharge, whatever comes first. Secondary outcomes include the mean or median fluid accumulation at 24, 48, and 72 hours of ICU admission, the mean or median fluid balance at 24, 48, and 72 hours of ICU stay, the duration of mechanical ventilation, length of hospital and ICU stay, and ICU and hospital mortality. Primary exposures of interest included an APACHE II score, a diagnosis of ARDS upon ICU admission, and a need for mechanical ventilation. Other characteristics of interest included: age, gender, weight, BMI, source of ICU admission, other ICU admission diagnoses, comorbidities, number of organ failures, need for vasopressors, use of nephrotoxic drugs, and whether a Foley catheter was removed before ICU discharge or starting diuretics or RRT.
Definitions:
Cardiovascular failure was defined by a systolic blood pressure of 90 mm Hg or less or the need for a vasopressor. Coagulation abnormality was defined by a platelet count of 80,000 per cubic millimeter or less. Hepatic failure was defined by a serum bilirubin level of at least 2 mg per deciliter (34 µmol per liter). Renal failure was defined by a serum creatinine level of at least 2 mg per deciliter (177 µmol per liter) (18). Percentage of fluid overload was defined as a calculation of total cumulative fluid balance from ICU admission to fluid de-resuscitation or ICU discharge and dividing the cumulative balance in liters by the patient's baseline weight and multiplying by 100 (11)
Sample Size And Statistical Analysis
As our study is the first to look into the percent fluid overload as a predictor of the need for fluid de-resuscitation, we included all eligible patients who were screened during the study period. Our study is not powered to detect the association between mean percent overload and the need for de- resuscitation.
The normality of the continuous predictor variables was assessed using Shapiro-Wilk and Kolmogorov-Smirnov tests by comparing median and mean values and observing the data graphically. The association between exposure variables, baseline and clinical characteristics, and fluid de-resuscitation status was assessed using Student's t-test, Wilcoxon two-sample test with normal approximation, Chi-Square test, and Fisher's Exact Test. The association between predictor variables and hospital mortality was also assessed using these bivariate tests, as was the association between patient fluid and mortality outcomes by fluid de-resuscitation status. The fluid outcomes were also assessed at 24, 48, and 72 hours post ICU discharge. A multivariable logistic model was constructed to examine the association between the main exposures of interest and the outcome of hospital mortality. Analyses were performed using SAS software, version 9.4, and statistical significance was determined at an α = 0.05 level.