Supplementary Table 1 Cardiovascular Procedures Excluded from Study

Supplementary Table 1Cardiovascular procedures excluded from study

Supplementary Table 2Summary of patient demographics by resuscitation fluid volume received

APS = acute physiology score

a Gender not specified for 54 patients

Supplementary Table 3Summary of patient diagnoses and surgical procedures by resuscitation fluid volume received

Supplementary Table 4. Summary of multivariate regression analysis. A multivariate regression model accounting for patient demographics, hospital characteristics, Elixhauser comorbidities, APS, and volume-adjusted chloride load was constructed to assess mortality OR, and modeled the change in in-hospital mortality OR associated with a 10 mmol/L incremental increase in volume-adjusted chloride load

Supplementary Table 5. Summary of mortality ORs associated with a 1 mmol/L incremental increase in volume-adjusted chloride load across three defined ranges. Analysis was carried out using continuous data. The 98-110 mmol/L range was defined as the normal range based on mean normal serum chloride ranges reported by hospitals in the EHR database. ORs overlapping with the null value (OR=1) suggest the absence of a meaningful association between mortality odds and increasing volume-adjusted chloride load. The multivariate regression model accounted for patient demographics, hospital characteristics, Elixhauser comorbidities, APS, and volume-adjusted chloride load

Supplementary Fig.1 Flow chart summarizing patients meeting study inclusion criteria. Among patients meeting SIRS criteria and receiving fluid resuscitation, 19,092 were excluded as a result of having <2 serum chloride measurements

Supplementary Fig. 2Relationship between baseline serum chloride concentration and in-hospital mortality (a) and peak serum chloride concentration and in-hospital mortality (b) in patients meeting SIRS criteria and receiving >500 mL intravenous crystalloid within two days of SIRS qualification. Data in both panels are fitted with a cubic polynomial function. Fit lines (solid lines) are weighted based on the number of patients with given baseline or peak chloride concentrations. Dashed lines represent 95% confidence interval

Supplementary Fig. 3 Relationship between sodium load received via IV resuscitation fluids within 72 h following SIRS qualification and in-hospital mortality, stratified by total resuscitation fluid volume received. In the <1500 mL group, there was a weak association between increasing sodium load and reduced mortality (weighted linear fit; slope=-0.012, R2=0.99). For the 1500-3000 mL group, weighted regression analysis showed a minimal association between the two variables (slope=-0.001; R2=0.01). For the 3000-4500 through 6000-7500 mL groups, there was an increasingly positive association (increased mortality with increasing sodium load) for each incremental volume increase (3000-4500 mL: slope=0.006, R2=0.81; 4500-6000 mL: slope=0.016, R2=0.61; 6000-7500 mL: slope=0.028, R2=0.85; >7500 mL: slope=0.007, R2=0.34)

Supplementary Fig. 4 Relationship between volume-adjusted sodium load received within 72 h following SIRS qualification and in-hospital mortality. The majority of patients (87%) received either 135-145 or 145-155 mmol/L sodium. Mortality was lowest (2.7%) for patients receiving 135-145 mmol/L sodium, while volume-adjusted sodium loads <115mmol/L were associated with >5% mortality. Regression line (solid line) represents a cubic polynomial fit and is weighted based on the number of patients receiving each given amount of volume-adjusted sodium. Dashed lines represent 95% confidence interval