Impact of Initial Central Venous Pressure on Outcomes of Conservative Versus Liberal Fluid Management in Acute Respiratory Distress Syndrome

Abstract

Objectives: In acute respiratory distress syndrome, conservative fluid management increases ventilator-free days without affecting mortality. Response to fluid management may differ based on patients' initial central venous pressure. We hypothesized that initial central venous pressure would modify the effect of fluid management on outcomes.

Design: Retrospective analysis of the Fluid and Catheter Treatment Trial, a multicenter randomized trial comparing conservative with liberal fluid management in acute respiratory distress syndrome. We examined the relationship between initial central venous pressure, fluid strategy, and 60-day mortality in univariate and multivariable analysis.

Setting: Twenty acute care hospitals.

Patients: Nine hundred thirty-four ventilated acute respiratory distress syndrome patients with a central venous pressure available at enrollment, 609 without baseline shock (for whom fluid balance was managed by the study protocol).

Interventions: None.

Measurements and main results: Among patients without baseline shock, those with initial central venous pressure greater than 8 mm Hg experienced similar mortality with conservative and liberal fluid management (18% vs 18%; p = 0.928), whereas those with central venous pressure of 8 mm Hg or less experienced lower mortality with a conservative strategy (17% vs 36%; p = 0.005). Multivariable analysis demonstrated an interaction between initial central venous pressure and the effect of fluid strategy on mortality (p = 0.031). At higher initial central venous pressures, the difference in treatment between arms was predominantly furosemide administration, which was not associated with mortality (p = 0.122). At lower initial central venous pressures, the difference between arms was predominantly fluid administration, with additional fluid associated with increased mortality (p = 0.013).

Conclusions: Conservative fluid management decreases mortality for acute respiratory distress syndrome patients with a low initial central venous pressure. In this population, the administration of IV fluids seems to increase mortality.

Figure 1. Mortality by fluid strategy in each central venous pressure quartile

Among patients without baseline shock, a Kaplan–Meier plot of the cumulative incidence of death with liberal (black) versus conservative (gray) fluid management is displayed within each quartile of initial central venous pressure (CVP). By log-rank testing, outcomes were similar between fluid strategies in all quartiles except the lowest in which the incidence of death was higher with liberal fluid management (p=0.009).

Figure 2. Risk of death with conservative versus liberal fluid management by initial central venous pressure

In a multivariable logistic regression model adjusting for age, gender, race, mean arterial pressure (MAP), the ratio of the partial pressure of oxygen to the fraction of inhaled oxygen (PaO₂:FiO₂), and Acute Physiology and Chronic Health Evaluation (APACHE) III score, the relationship between assigned fluid strategy and 60-day in-hospital mortality was significantly different for patients with different initial central venous pressures. The main figure displays the risk of death when treated with liberal versus conservative fluid management for a prototypical patient with median values for all continuous covariates and mode values for all binary covariates. The inlaid forest plot displays, for each covariate in the model, the odds of death (and 95% confidence interval) relative to the reference group for categorical variables and moving from the 25th percentile to the 75th percentile for continuous variables.

Figure 3. Fluid balance by central venous pressure quartile and study group

Among patients without baseline shock, cumulative fluid balance over the 7 days of the study was greater for the lowest versus highest central venous pressure quartile in both the liberal arm (7.7±9.0L versus 3.0±8.0L, p=0.001) and conservative arm (0.1±8.5L versus -4.5±6.9L, p<0.001). CVP is central venous pressure in mmHg.

Figure 4. Fluid management and mortality by central venous pressure and study arm

The fluid management and in-hospital mortality experienced in the liberal and conservative arms are displayed relative to patients’ initial central venous pressure (CVP) using locally weighted scatterplot smoothing (LOWESS). The difference between arms in fluid output was greater for patients with higher initial central venous pressure, whereas the difference between arms in fluid input was greater for patients with lower initial central venous pressure. The difference in cumulative fluid balance between arms was similar at all CVPs, with more “net negative” fluid balance at higher CVPs in the conservative arm and “net positive” fluid balance at lower CVPs in the liberal arm. Mortality was highest for patients with lower CVPs treated with liberal fluid management.