Table.

Representative observational clinical studies published between 2010 and 2017 evaluating the association between early net fluid balance and mortality in adults with sepsis. RCT: randomized controlled trial; ICU: intensive care unit; CI: confidence interval; HR: hazard ratio; OR: odds ratio; US: United States

Publication	Study Design [centers]	Population [sample size]	Exposure (Predictor) Variable(s)	Primary Outcome	Main findings
Boyd et al. Crit Care Med 20119	Secondary analysis of a multicenter RCT [27 centers in Canada, Australia, USA]	Adults in ICU with septic shock on norepinephrine ≥5 mcg/min [n = 778]	Net fluid balance at 12 hours after initiation of resuscitation; patients classified according to quartile of net fluid balance	28-day mortality	Compared to patients in the highest quartile of fluid balance (median 8.2 L) those in the lower quartiles of fluid balance (quartile 1: 0.7 L; quartile 2: 2.9 L) had lower risk of mortality in adjusted proportional hazard models [quartile 1 vs quartile 4: aHR 0.57 (95% CI: 0.41, 0.80); quartile 2 vs quartile 4: aHR 0.58 (0.41, 0.82). A fluid balance of +3 L at 12 hours correlated with optimal survival.
Micek et al. Crit Care 201310	Retrospective cohort study [1 center in US]	Adults in ICU with septic shock (vasopressor use >12 hours) [n = 163]	Net fluid balance at 24 hours after shock recognition; patients classified according to quartile of net fluid balance	In-hospital mortality	In an adjusted proportional hazards model, patients in the highest quartile of positive fluid balance at 24 hours had increased in-hospital mortality compared to those in the first quartile (p=0.001) and second quartile (p=0.034).
Sadaka et al. J Intensive Care Med 201411	Retrospective cohort study [1 center in US]	Adults in ICU with septic shock [n = 350]	Net fluid balance at 24 hours after ICU admit; patients classified into 4 categories according to net fluid balance: <6L, 6–12 L, 12–18L, 18–24L.	In-hospital mortality	In an adjusted proportional hazards model, compared to patients with <6 L fluid balance, those with 6–12L, 12–18L, and 18–24 L positive fluid balance had higher mortality risk [aHR: 1.52 (1.35, 1.69), 1.74 (1.47, 2.01), 1.62 (1.20, 2.04), respectively].
Acheampong & Vincent. Crit Care 201512	Prospective cohort study [1 center in Belgium]	Adults in ICU >48 hours with sepsis (infection & ≥1 organ failure) [n = 173]	Net daily fluid balance for first 7 days of ICU stay; daily fluid balance analyzed on a continuous scale	ICU mortality	On a continuous scale, more positive daily fluid balance was associated with increased ICU mortality in an adjusted proportional hazards model [aHR 1.014 per ml/kg increase (95% CI: 1.007, 1.022)].
de Oliveira et al. J Crit Care 201513	Retrospective cohort study [1 center in Brazil]	Adults in ICU with sepsis (infection & ≥1 organ failure) [n = 116]	Net fluid balance between 24 and 48 hours after first recognition of organ dysfunction	In-hospital mortality	A net positive fluid balance >3 L was associated with increased hospital mortality in an adjusted logistic regression model [aOR 3.19 (1.19, 8.54)].
Kelm et al. Shock 201514	Retrospective cohort study [1 center in US]	Adults in ICU with sepsis (infection & ≥1 organ failure) [n = 405]	Signs of fluid overload on day 1 (new pitting edema, crackles, anasarca on exam or new vascular congestion, pulmonary edema or pleural effusion on CXR)	In-hospital mortality	Patients with at least one sign of fluid overload on ICU day #1 had higher risk of in-hospital mortality in an adjusted logistic regression model [aOR: 2.27 (95% CI: 1.31, 4.09)].
Sakr et al. Crit Care Med 201715	Prospective cohort study [multicenter, multinational audit over 10 days]	Adults in ICU with sepsis (infection & ≥1 organ failure) [n = 1,808]	Net fluid balance at 24 hours and 72 hours after ICU admission; patients classified according to quartile of net fluid balance	28-day in-hospital mortality	Fluid balance at 24 hours was not associated with mortality; however, higher fluid balance at 72 hrs was associated with increased mortality. Compared with patients in the lowest quartile of fluid balance at 72 hrs, adjusted hazard ratios for quartiles 2, 3, and 4 were 1.36 (1.03, 1.80), 1.47 (1.12, 1.92), and 1.63 (1.25, 2.12), respectively.