Browse the corpus

Balanced Fluid or 0.9% Saline in Children Treated for Septic Shock. BACKGROUND: Whether treatment with balanced crystalloid fluid leads to better outcomes than 0.9% saline in children treated for septic shock is debated. METHODS: In this pragmatic clinical trial conducted at 47 emergency departments in five countries, patients (2 months to <18 years of age) with suspected septic shock and abnormal perfusion were randomly assigned to receive fluid resuscitation with either balanced fluid or 0.9% saline for up to 48 hours. The primary outcome was a major adverse kidney event (a composite of death, new renal-replacement therapy, or persistent kidney dysfunction) at 30 days after enrollment or hospital discharge, whichever occurred first. RESULTS: Of 9041 enrolled patients, 277 (6.1%) in the balanced-fluid group and 282 (6.2%) in the 0.9%-saline group withdrew from the trial, leaving 4235 and 4247 patients, respectively, for analysis. A primary-outcome event occurred in 137 patients (3.4%) in the balanced-fluid group and in 124 (3.0%) in the 0.9%-saline group (difference, 0.4 percentage points; 95% confidence interval [CI], -0.5 to 1.3; risk ratio, 1.10; 95% CI, 0.88 to 1.40; P = 0.85). The median number of hospital-free days during 28 days after enrollment was 23 (interquartile range, 19 to 25) in both groups. Hyperchloremia occurred in 868 patients (31.4%) in the balanced-fluid group and in 1383 (49.0%) in the 0.9%-saline group; hypernatremia in 52 (1.8%) and 89 (3.1%), respectively; and hyperlactatemia in 260 (19.8%) and 228 (16.7%). No differences in other safety outcomes or adverse events were seen. CONCLUSIONS: Among children treated for septic shock, no significant difference was seen in the incidence of death, new renal-replacement therapy, or persistent kidney dysfunction when fluid resuscitation was administered with balanced fluid as compared with 0.9% saline. (Funded by Eunice Kennedy Shriver National Institute of Child Health and Human Development and others; PRoMPT BOLUS ClinicalTrials.gov number, NCT04102371.).

We conducted a multicenter, pragmatic, open-label, randomized interventional trial in which children with suspected septic shock treated in an emergency department (ED) between August 25, 2020, and October 31, 2025, were allocated to balanced crystalloids or 0.9% saline for resuscitation and maintenance fluids. The study was conducted at 47 sites in the United States (US), Canada, Australia, New Zealand, and Costa Rica. Human subjects regulatory oversight was administered separately within each country (see Supplementary Appendix). The protocol demonstrated feasibility in a pilot study,25 and methodologic details were previously published.26 The protocol and statistical analysis plan are available at NEJM.org. The first and last authors designed the study and wrote the initial draft of the manuscript. The first authors and biostatisticians (AA, JH) had full access to the data, vouch for it and analyzed and confirmed the data independently. An international steering committee oversaw all study activities, including data collection, and vouches for data accuracy and fidelity of protocol adherence. All authors reviewed the final version of the manuscript and participated in the decision to publish. The sponsors had no role in the design or conduct of the study, data analysis, or approval of the manuscript.

udy activities, including data collection, and vouches for data accuracy and fidelity of protocol adherence. All authors reviewed the final version of the manuscript and participated in the decision to publish. The sponsors had no role in the design or conduct of the study, data analysis, or approval of the manuscript. Patients from 2 months to <18 years-old being treated in an ED for suspected septic shock with fluid resuscitation for abnormal perfusion were eligible for enrollment provided that total volume of crystalloid fluid administration was confirmed as ≤40 mL/kg prior to enrollment. All participating ED clinicians screened patients during routine care, but only those trained on study procedures could enroll. Patients could be enrolled at any time prior to ED disposition so long as eligibility criteria were met. Enrolled patients could participate more than once. Patients for whom the treating clinician judged it unsafe to administer either fluid type were excluded (Supplementary Appendix for additional details). Due to the narrow therapeutic window to begin fluid resuscitation, enrollment adhered to “Exception from Informed Consent” (21 CFR 50.24) for emergency research in the US27 and ethically-approved alternative processes to prospective informed consent in Canada, Australia/New Zealand, and Costa Rica, as described in the Supplementary Appendix.28,29

c window to begin fluid resuscitation, enrollment adhered to “Exception from Informed Consent” (21 CFR 50.24) for emergency research in the US27 and ethically-approved alternative processes to prospective informed consent in Canada, Australia/New Zealand, and Costa Rica, as described in the Supplementary Appendix.28,29 Study patients were allocated to either balanced fluid or 0.9% saline using permuted-block randomization, stratified by site (Supplementary Appendix). Treatment allocation was concealed using serially numbered, opaque envelopes for efficiency of enrollment concurrent with clinical management. Fluid type allocation was revealed after eligibility was confirmed and the patient was enrolled. We compared treatment with a predominantly balanced fluid or 0.9% saline strategy, acknowledging that routine care often includes multiple fluid types (Supplementary Appendix, Fig. S1). Balanced fluids could be LR, Plasma-Lyte™, or Hartmann’s solution, depending on availability or clinician preference (Table S1), but the allocated study fluid was preferentially used for all bolus fluid and as the base fluid for maintenance hydration until 11:59 P.M. of the following day (Table S2). The intervention phase was timed to end so that all patients would receive study fluid for 24 to 48 hours, the time-window when most fluid resuscitation for septic shock is completed and which provided a pragmatic end to the intervention phase.14

r maintenance hydration until 11:59 P.M. of the following day (Table S2). The intervention phase was timed to end so that all patients would receive study fluid for 24 to 48 hours, the time-window when most fluid resuscitation for septic shock is completed and which provided a pragmatic end to the intervention phase.14 Each site established procedures to promote protocol adherence (Supplementary Appendix). Other than fluid type, all decisions about timing, volume, and rate of fluid administration remained at the discretion of the treating clinicians. Alternative fluid types were allowed for clinical indications (e.g., hyponatremia). Maintenance fluids were included because these constitute a substantial proportion of total crystalloid fluid administration.25,30 Non-isotonic fluids (e.g., 0.45% saline) are not recommended as maintenance fluid in children and were discouraged.31,32 Hospital fluid supplies were used without changes to labeling. Patients and clinicians were not blinded to treatment allocation as a practical necessity and to reduce risk that clinicians might attribute electrolyte changes to the study fluid and stop adhering to the protocol. However, the senior biostatistician (JH) and all investigators remained blinded to aggregate outcomes until enrollment was complete.

e not blinded to treatment allocation as a practical necessity and to reduce risk that clinicians might attribute electrolyte changes to the study fluid and stop adhering to the protocol. However, the senior biostatistician (JH) and all investigators remained blinded to aggregate outcomes until enrollment was complete. Data obtained during clinical care were extracted from the medical record (Supplementary Appendix) and monitored for quality within each network. A central Data Coordinating Center at CHOP collated data across networks for interim and final analyses. Site of infection and diagnosis of septic shock were ascertained by the lead investigator at each site using all available data. AKI at presentation was determined using serum creatinine cut-points recommended by Kidney Disease Improving Global Outcomes (KDIGO) guidelines.33,34 The primary outcome was one or more criteria for major adverse kidney events at 30 days (MAKE30)—a composite of death from any cause, initiation of RRT, or persistent kidney dysfunction—following study enrollment or hospital discharge, whichever occurred first.35 RRT included treatment with any renal replacement modality started during the hospitalization. Persistent kidney dysfunction was defined as a final serum creatinine at least 200% of baseline and a minimum increase of at least 0.3 mg/dL. Baseline serum creatinine was recorded as the lowest value available between 12 months and 24 hours prior to enrollment or, if missing, imputed using median creatinine values for age and sex (Supplementary Appendix).36

d as a final serum creatinine at least 200% of baseline and a minimum increase of at least 0.3 mg/dL. Baseline serum creatinine was recorded as the lowest value available between 12 months and 24 hours prior to enrollment or, if missing, imputed using median creatinine values for age and sex (Supplementary Appendix).36 Secondary effectiveness outcomes included components of MAKE30, hospital length of stay, hospital-free days out of 28, and all-cause mortality prior to hospital discharge and within 90 days after randomization. Safety outcomes included electrolyte abnormalities within three days of enrollment, arterial/venous thrombosis, and cerebral edema diagnosed during clinical care. Adverse events were collected for seven days after enrollment (Supplementary Appendix). Details regarding sample size determination were previously published.26 Briefly, enrollment of 8,800 participants was calculated as providing 95% power to detect an absolute risk reduction in MAKE30 from 6.0% for children treated with 0.9% saline (based on preliminary data36) to 4.3% for balanced fluid with type-I error of 0.05. To account for withdrawal after EFIC/deferred consent, the final enrollment target was increased by 4% to 9,178.

lated as providing 95% power to detect an absolute risk reduction in MAKE30 from 6.0% for children treated with 0.9% saline (based on preliminary data36) to 4.3% for balanced fluid with type-I error of 0.05. To account for withdrawal after EFIC/deferred consent, the final enrollment target was increased by 4% to 9,178. The primary analysis included all randomized patients, except those who withdrew use of their data. Patients later determined not to have met eligibility criteria were retained in the primary analysis to avoid bias from differential assessment of eligibility. Multiple imputation was used to account for missing outcome data, with the assumption that data were missing-at-random (Supplementary Appendix). MAKE30 and other binary outcomes were compared using the Cochran–Mantel–Haenszel test, stratified by study site. Continuous outcomes were compared using the Van Elteren test, stratified by site. Mortality within 90 days was estimated with the Kaplan–Meier method, and a mixed-effects Cox proportional-hazards model was used to compare groups, with random intercepts for study site. The proportional-hazards assumption was assessed by visual inspection of log–log survival plots and by testing Schoenfeld residuals, and was not violated.

ays was estimated with the Kaplan–Meier method, and a mixed-effects Cox proportional-hazards model was used to compare groups, with random intercepts for study site. The proportional-hazards assumption was assessed by visual inspection of log–log survival plots and by testing Schoenfeld residuals, and was not violated. Sensitivity analyses included patients with complete primary outcome data and a per-protocol analysis of patients who received at least 75% of their total crystalloid fluid volume during the intervention phase as the fluid to which they were randomized. We also conducted a tipping-point analysis to assess the robustness of the primary analysis of MAKE30 to departures from the missing-at-random assumption. For safety outcomes with more than 30% missing data, additional sensitivity analyses were performed (Supplementary Appendix). Prespecified subgroup analyses were conducted for age, sex, cancer comorbidity, AKI at presentation, total fluid volume during the intervention phase, and country of enrollment (Supplementary Appendix). Additional post-hoc analyses evaluated measured versus imputed baseline creatinine and initial severity of acidosis and hyperlactatemia. Subgroup analyses were performed using multiply imputed data.

KI at presentation, total fluid volume during the intervention phase, and country of enrollment (Supplementary Appendix). Additional post-hoc analyses evaluated measured versus imputed baseline creatinine and initial severity of acidosis and hyperlactatemia. Subgroup analyses were performed using multiply imputed data. A data and safety monitoring board (DSMB) oversaw the study. Interim analyses for efficacy were performed after enrollment of 15%, 40%, and 70% of participants using a group sequential design with symmetric two-sided O’Brien-Fleming boundaries to control the overall type I error rate at 0.05. A two-sided P-value of less than 0.044 at the final analysis indicated statistical significance for the primary outcome. Due to multiplicity, statistical significance is not reported for secondary effectiveness outcomes, and unadjusted P-values are reported for safety outcomes. All analyses were performed using SAS, Version 9.4 (SAS Institute, Cary, NC) or R, Version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

Patients from 2 months to <18 years-old being treated in an ED for suspected septic shock with fluid resuscitation for abnormal perfusion were eligible for enrollment provided that total volume of crystalloid fluid administration was confirmed as ≤40 mL/kg prior to enrollment. All participating ED clinicians screened patients during routine care, but only those trained on study procedures could enroll. Patients could be enrolled at any time prior to ED disposition so long as eligibility criteria were met. Enrolled patients could participate more than once. Patients for whom the treating clinician judged it unsafe to administer either fluid type were excluded (Supplementary Appendix for additional details). Due to the narrow therapeutic window to begin fluid resuscitation, enrollment adhered to “Exception from Informed Consent” (21 CFR 50.24) for emergency research in the US27 and ethically-approved alternative processes to prospective informed consent in Canada, Australia/New Zealand, and Costa Rica, as described in the Supplementary Appendix.28,29

Study patients were allocated to either balanced fluid or 0.9% saline using permuted-block randomization, stratified by site (Supplementary Appendix). Treatment allocation was concealed using serially numbered, opaque envelopes for efficiency of enrollment concurrent with clinical management. Fluid type allocation was revealed after eligibility was confirmed and the patient was enrolled.

We compared treatment with a predominantly balanced fluid or 0.9% saline strategy, acknowledging that routine care often includes multiple fluid types (Supplementary Appendix, Fig. S1). Balanced fluids could be LR, Plasma-Lyte™, or Hartmann’s solution, depending on availability or clinician preference (Table S1), but the allocated study fluid was preferentially used for all bolus fluid and as the base fluid for maintenance hydration until 11:59 P.M. of the following day (Table S2). The intervention phase was timed to end so that all patients would receive study fluid for 24 to 48 hours, the time-window when most fluid resuscitation for septic shock is completed and which provided a pragmatic end to the intervention phase.14 Each site established procedures to promote protocol adherence (Supplementary Appendix). Other than fluid type, all decisions about timing, volume, and rate of fluid administration remained at the discretion of the treating clinicians. Alternative fluid types were allowed for clinical indications (e.g., hyponatremia). Maintenance fluids were included because these constitute a substantial proportion of total crystalloid fluid administration.25,30 Non-isotonic fluids (e.g., 0.45% saline) are not recommended as maintenance fluid in children and were discouraged.31,32 Hospital fluid supplies were used without changes to labeling.

mia). Maintenance fluids were included because these constitute a substantial proportion of total crystalloid fluid administration.25,30 Non-isotonic fluids (e.g., 0.45% saline) are not recommended as maintenance fluid in children and were discouraged.31,32 Hospital fluid supplies were used without changes to labeling. Patients and clinicians were not blinded to treatment allocation as a practical necessity and to reduce risk that clinicians might attribute electrolyte changes to the study fluid and stop adhering to the protocol. However, the senior biostatistician (JH) and all investigators remained blinded to aggregate outcomes until enrollment was complete.

Data obtained during clinical care were extracted from the medical record (Supplementary Appendix) and monitored for quality within each network. A central Data Coordinating Center at CHOP collated data across networks for interim and final analyses. Site of infection and diagnosis of septic shock were ascertained by the lead investigator at each site using all available data. AKI at presentation was determined using serum creatinine cut-points recommended by Kidney Disease Improving Global Outcomes (KDIGO) guidelines.33,34

The primary outcome was one or more criteria for major adverse kidney events at 30 days (MAKE30)—a composite of death from any cause, initiation of RRT, or persistent kidney dysfunction—following study enrollment or hospital discharge, whichever occurred first.35 RRT included treatment with any renal replacement modality started during the hospitalization. Persistent kidney dysfunction was defined as a final serum creatinine at least 200% of baseline and a minimum increase of at least 0.3 mg/dL. Baseline serum creatinine was recorded as the lowest value available between 12 months and 24 hours prior to enrollment or, if missing, imputed using median creatinine values for age and sex (Supplementary Appendix).36 Secondary effectiveness outcomes included components of MAKE30, hospital length of stay, hospital-free days out of 28, and all-cause mortality prior to hospital discharge and within 90 days after randomization. Safety outcomes included electrolyte abnormalities within three days of enrollment, arterial/venous thrombosis, and cerebral edema diagnosed during clinical care. Adverse events were collected for seven days after enrollment (Supplementary Appendix).

Details regarding sample size determination were previously published.26 Briefly, enrollment of 8,800 participants was calculated as providing 95% power to detect an absolute risk reduction in MAKE30 from 6.0% for children treated with 0.9% saline (based on preliminary data36) to 4.3% for balanced fluid with type-I error of 0.05. To account for withdrawal after EFIC/deferred consent, the final enrollment target was increased by 4% to 9,178. The primary analysis included all randomized patients, except those who withdrew use of their data. Patients later determined not to have met eligibility criteria were retained in the primary analysis to avoid bias from differential assessment of eligibility. Multiple imputation was used to account for missing outcome data, with the assumption that data were missing-at-random (Supplementary Appendix). MAKE30 and other binary outcomes were compared using the Cochran–Mantel–Haenszel test, stratified by study site. Continuous outcomes were compared using the Van Elteren test, stratified by site. Mortality within 90 days was estimated with the Kaplan–Meier method, and a mixed-effects Cox proportional-hazards model was used to compare groups, with random intercepts for study site. The proportional-hazards assumption was assessed by visual inspection of log–log survival plots and by testing Schoenfeld residuals, and was not violated.

Of the 9,041 participants enrolled, 4,512 were randomized to balanced fluid and 4,529 to 0.9% saline (Figs. S2–S4 and Table S3). A total of 277 (6.1%) and 282 (6.2%) from the balanced fluid and 0.9% saline groups withdrew use of their data. Therefore, the analysis included 4,235 patients assigned to balanced fluid and 4,247 assigned to 0.9% saline. Patient characteristics were similar between groups (Tables 1 and S4–S5). Median age was 6.8 years (interquartile [IQR], 2.8–13), and 50.1% were male. The site of infection was most often respiratory (46.6%, Table S6), and only 5.8% were determined not to have sepsis (Table S7). At presentation, 1,172 (13.8%), 415 (4.9%), and 432 (5.1%) had stage 1, 2, or 3 AKI, respectively (Tables 1 and S8). Treatments for sepsis through the intervention phase are summarized in Table S9. Overall, 1,207 (14.2%) received vasoactive medications and 822 (9.7%) required invasive mechanical ventilation. Treatment with bicarbonate occurred in 5.2%, including 194 (4.6%) in the balanced fluid and 249 (5.9%) in the 0.9% saline groups. The volumes of total, bolus, and maintenance crystalloid fluids were similar between treatment groups prior to randomization and during the interventional phase (Fig. 1). The median total crystalloid volume, including both prior to randomization and during the intervention phase, was 85 mL/kg (IQR, 55–119) and 88 mL/kg (IQR, 57–123) for the balanced fluid and 0.9% saline groups, respectively (difference in medians, −2.3 mL/kg; 95% CI −4.3 to −0.3;Table S10 and Fig. S5).

phase (Fig. 1). The median total crystalloid volume, including both prior to randomization and during the intervention phase, was 85 mL/kg (IQR, 55–119) and 88 mL/kg (IQR, 57–123) for the balanced fluid and 0.9% saline groups, respectively (difference in medians, −2.3 mL/kg; 95% CI −4.3 to −0.3;Table S10 and Fig. S5). The median volume of 0.9% saline received was 20 mL/kg (IQR, 5.5–32) in the balanced fluid group compared with 79 mL/kg (IQR, 49–113) in the 0.9% saline group, while the median volume of balanced fluid was 58 mL/kg (IQR, 31–92) in the balanced group compared with 0 mL/kg (IQR, 0–0) in the 0.9% saline group (Table S10). Overall, 80% of patients in the balanced fluid and 88% in the 0.9% saline groups received at least 75% of total crystalloid as their randomized fluid type. MAKE30 occurred in 137 of 4,073 patients (3.4%) in the balanced fluid group and 124 of 4,068 (3.0%) in the 0.9% saline group (absolute difference, 0.4 percentage points; 95% CI, −0.5 to 1.3; risk ratio, 1.10; 95% CI, 0.88 to 1.40, P=0.85; Table 2 and Fig. S6). The results were similar in pre-specified sensitivity analysis of patients without missing components of MAKE30 and the per-protocol analysis (Table S11). A post-hoc analysis did not demonstrate effect modification from pre-randomization 0.9% saline administration (Table S12). Tipping-point analysis demonstrated that the findings for MAKE30 remained robust to all but extreme departures from the missing-at-random assumption (Fig. S7).

he per-protocol analysis (Table S11). A post-hoc analysis did not demonstrate effect modification from pre-randomization 0.9% saline administration (Table S12). Tipping-point analysis demonstrated that the findings for MAKE30 remained robust to all but extreme departures from the missing-at-random assumption (Fig. S7). Results for secondary effectiveness and safety outcomes are shown in Table 2. Median hospital-free days out of 28 were 23 (19–25) in both groups. Mortality was not different between groups at hospital discharge or within 90 days (Table 2 and Fig. S8). Among patients with follow-up laboratory values measured through study day 3, hyperchloremia and hypernatremia occurred less often, and hyperlactatemia more often, with balanced fluid (Tables 2 and S13 and Fig. S9). The lower rate of hyperchloremia with balanced fluid remained significant across sensitivity analyses (Table S14 and Fig. S10). Differences in follow-up blood chloride, bicarbonate, and creatinine were more pronounced between groups as crystalloid volume increased (Figs. S11–13). Thrombosis and cerebral edema did not differ between groups (Table 2), nor did other adverse events (Table S15). MAKE30 did not differ between balanced fluid and 0.9% saline within pre-specified subgroups (Fig. 2) or post-hoc analyses stratified by measured versus imputed baseline creatinine (Fig. S14) or initial bicarbonate or lactate levels (Fig. S15). A post-hoc analysis comparing MAKE30 across increasing total volumes of crystalloid fluid also did not demonstrate differences between treatment groups (Fig. S16).

Of the 9,041 participants enrolled, 4,512 were randomized to balanced fluid and 4,529 to 0.9% saline (Figs. S2–S4 and Table S3). A total of 277 (6.1%) and 282 (6.2%) from the balanced fluid and 0.9% saline groups withdrew use of their data. Therefore, the analysis included 4,235 patients assigned to balanced fluid and 4,247 assigned to 0.9% saline.

Patient characteristics were similar between groups (Tables 1 and S4–S5). Median age was 6.8 years (interquartile [IQR], 2.8–13), and 50.1% were male. The site of infection was most often respiratory (46.6%, Table S6), and only 5.8% were determined not to have sepsis (Table S7). At presentation, 1,172 (13.8%), 415 (4.9%), and 432 (5.1%) had stage 1, 2, or 3 AKI, respectively (Tables 1 and S8). Treatments for sepsis through the intervention phase are summarized in Table S9. Overall, 1,207 (14.2%) received vasoactive medications and 822 (9.7%) required invasive mechanical ventilation. Treatment with bicarbonate occurred in 5.2%, including 194 (4.6%) in the balanced fluid and 249 (5.9%) in the 0.9% saline groups.

The volumes of total, bolus, and maintenance crystalloid fluids were similar between treatment groups prior to randomization and during the interventional phase (Fig. 1). The median total crystalloid volume, including both prior to randomization and during the intervention phase, was 85 mL/kg (IQR, 55–119) and 88 mL/kg (IQR, 57–123) for the balanced fluid and 0.9% saline groups, respectively (difference in medians, −2.3 mL/kg; 95% CI −4.3 to −0.3;Table S10 and Fig. S5). The median volume of 0.9% saline received was 20 mL/kg (IQR, 5.5–32) in the balanced fluid group compared with 79 mL/kg (IQR, 49–113) in the 0.9% saline group, while the median volume of balanced fluid was 58 mL/kg (IQR, 31–92) in the balanced group compared with 0 mL/kg (IQR, 0–0) in the 0.9% saline group (Table S10). Overall, 80% of patients in the balanced fluid and 88% in the 0.9% saline groups received at least 75% of total crystalloid as their randomized fluid type.

MAKE30 occurred in 137 of 4,073 patients (3.4%) in the balanced fluid group and 124 of 4,068 (3.0%) in the 0.9% saline group (absolute difference, 0.4 percentage points; 95% CI, −0.5 to 1.3; risk ratio, 1.10; 95% CI, 0.88 to 1.40, P=0.85; Table 2 and Fig. S6). The results were similar in pre-specified sensitivity analysis of patients without missing components of MAKE30 and the per-protocol analysis (Table S11). A post-hoc analysis did not demonstrate effect modification from pre-randomization 0.9% saline administration (Table S12). Tipping-point analysis demonstrated that the findings for MAKE30 remained robust to all but extreme departures from the missing-at-random assumption (Fig. S7).

Results for secondary effectiveness and safety outcomes are shown in Table 2. Median hospital-free days out of 28 were 23 (19–25) in both groups. Mortality was not different between groups at hospital discharge or within 90 days (Table 2 and Fig. S8). Among patients with follow-up laboratory values measured through study day 3, hyperchloremia and hypernatremia occurred less often, and hyperlactatemia more often, with balanced fluid (Tables 2 and S13 and Fig. S9). The lower rate of hyperchloremia with balanced fluid remained significant across sensitivity analyses (Table S14 and Fig. S10). Differences in follow-up blood chloride, bicarbonate, and creatinine were more pronounced between groups as crystalloid volume increased (Figs. S11–13). Thrombosis and cerebral edema did not differ between groups (Table 2), nor did other adverse events (Table S15).

MAKE30 did not differ between balanced fluid and 0.9% saline within pre-specified subgroups (Fig. 2) or post-hoc analyses stratified by measured versus imputed baseline creatinine (Fig. S14) or initial bicarbonate or lactate levels (Fig. S15). A post-hoc analysis comparing MAKE30 across increasing total volumes of crystalloid fluid also did not demonstrate differences between treatment groups (Fig. S16).

In this trial of children treated for suspected septic shock in an emergency department, use of balanced fluid for up to 48 hours did not reduce MAKE30, hospital-free days, or mortality. Adequate treatment separation was evident in the distribution of fluid type between groups, as well as by lower rates of hyperchloremia and hypernatremia in the balanced fluid group. Hyperlactatemia was also more frequent with balanced fluids, but there were no differences in rates of thrombosis, cerebral edema, or other adverse events. Our data do not show benefit for the routine use of balanced fluid over 0.9% saline in children with suspected septic shock. As in prior studies, balanced fluid resulted in less hyperchloremia and hypernatremia,6,16 but these biochemical effects did not translate to improved patient-centered outcomes. Notably, our findings differ from the trial by Sankar et al. which found a 38% relative reduction in new/progressive AKI with balanced fluid among 708 children with septic shock.16 However, in that study, the outcome was only monitored for seven days and included any increase in serum creatinine of 0.3 mg/dL or more, whereas MAKE30 required at least a two-fold increase above baseline creatinine. Despite these differences, neither Sankar et al.16 nor our study found differences in mortality or hospital-free days.

study, the outcome was only monitored for seven days and included any increase in serum creatinine of 0.3 mg/dL or more, whereas MAKE30 required at least a two-fold increase above baseline creatinine. Despite these differences, neither Sankar et al.16 nor our study found differences in mortality or hospital-free days. A meta-analysis of six clinical trials in adult patients estimated an 89.5% probability that balanced fluids reduce mortality compared to 0.9% saline.37 However, a recent cluster-randomized trial of 43,626 patients did not find a beneficial effect of hospital-wide use of LR.19 Two prior adult studies demonstrated small, but significant, reductions in MAKE30, but neither demonstrated differences in hospital-free days.13,14 Strengths of our trial included a large sample size, which was needed to detect small differences in patient outcomes. Additionally, we only enrolled patients prior to large-volume fluid resuscitation because prior studies had suggested that this population would be most likely to benefit from balanced fluid.14,21

A meta-analysis of six clinical trials in adult patients estimated an 89.5% probability that balanced fluids reduce mortality compared to 0.9% saline.37 However, a recent cluster-randomized trial of 43,626 patients did not find a beneficial effect of hospital-wide use of LR.19 Two prior adult studies demonstrated small, but significant, reductions in MAKE30, but neither demonstrated differences in hospital-free days.13,14 Strengths of our trial included a large sample size, which was needed to detect small differences in patient outcomes. Additionally, we only enrolled patients prior to large-volume fluid resuscitation because prior studies had suggested that this population would be most likely to benefit from balanced fluid.14,21 Our study also had several limitations. First, it is not certain that our results are generalizable to low-resource settings or hospital-acquired sepsis. Second, we defined septic shock using immediately accessible signs of abnormal perfusion to capture patients near the onset of fluid therapy rather than wait for higher-risk criteria, which are often not available at presentation. Although this approach aligns with clinical practice and the study population was representative of children treated for septic shock in an ED (Table S16), the low event rate diminished statistical power to detect planned differences between groups. Moreover, while we did not observe heterogeneity of treatment effect in subgroup analyses, point estimates favoured balanced fluid among patients who received the highest fluid volumes and presented with more extreme acidosis and hyperlactatemia; thus, we cannot exclude a benefit of balanced fluid in children with the most severe illness. Third, although studies have shown that most crystalloid fluid is administered within the initial 48 hours of sepsis treatment,6,14,16,21 it is possible that unmeasured fluid administration after the intervention phase altered the outcomes. Fourth, our analysis could not distinguish patients enrolled more than once. Fifth, a higher proportion of patients withdrew prior to ascertaining the primary outcome than expected. However, withdrawal was unlikely to be related to the intervention, and the results were robust to all but unplausible conditions among patients with missing data. Finally, MAKE30 is composed of endpoints that may not be of equivalent value to patients; however, there were no differences in the sub-components or other secondary effectiveness outcomes.

to be related to the intervention, and the results were robust to all but unplausible conditions among patients with missing data. Finally, MAKE30 is composed of endpoints that may not be of equivalent value to patients; however, there were no differences in the sub-components or other secondary effectiveness outcomes. In this pragmatic, randomized clinical trial of approximately 9,000 children with suspected septic shock treated with fluid resuscitation for abnormal perfusion in an ED, there was no significant difference between balanced fluid and 0.9% saline in the composite outcome of death, new renal replacement therapy, or persistent kidney dysfunction.