Browse the corpus

Intravenous Rehydration for Severe Acute Malnutrition with Gastroenteritis. BACKGROUND: International recommendations advise against the use of intravenous rehydration therapy in children with severe acute malnutrition because of the concern about fluid overload, but evidence to support this concern is lacking. Given the high mortality associated with the current recommendations, the adoption of intravenous rehydration strategies might improve outcomes. METHODS: We conducted a factorial, open-label superiority trial in four countries in Africa. Children 6 months to 12 years of age with severe acute malnutrition with gastroenteritis and dehydration underwent randomization in a 2:1:1 ratio to one of three rehydration strategies: oral rehydration, plus intravenous boluses for shock; a rapid intravenous strategy that consisted of lactated Ringer's solution (100 ml per kilogram of body weight) administered over a period of 3 to 6 hours, with boluses for shock; or a slow intravenous strategy that consisted of the same solution administered over a period of 8 hours, with no boluses. The primary end point was death at 96 hours. RESULTS: A total of 272 children underwent randomization; 138 were assigned to the oral strategy, 67 to the rapid intravenous strategy, and 67 to the slow intravenous strategy. Participants were followed for 28 days. A nasogastric tube was used for oral rehydration in 126 of 135 participants (93%) in the oral group and in 82 of 126 (65%) in the intravenous groups. Intravenous boluses were administered at admission in 12 participants (9%) in the oral group, 7 (10%) in the rapid intravenous group, and none in the slow intravenous group. At 96 hours, 11 participants (8%) in the oral group and 9 (7%) in the intravenous groups (5 in the rapid group and 4 in the slow group) had died (risk ratio, 1.02; 95% confidence interval [CI], 0.41 to 2.52; P = 0.69). At 28 days, 17 participants (12%) in the oral group and 14 (10%) in the intravenous groups had died (hazard ratio, 0.85; 95% CI, 0.41 to 1.78). Serious adverse events occurred in 32 participants (23%) in the oral group, 14 (21%) in the rapid intravenous group, and 10 (15%) in the slow intravenous group. No evidence of pulmonary edema, heart failure, or fluid overload was noted. CONCLUSIONS: Among children with severe acute malnutrition and gastroenteritis, no evidence of a difference in mortality at 96 hours was noted between oral and intravenous rehydration strategies. (Funded by the Joint Global Health Trials scheme and others; GASTROSAM Current Controlled Trials number, ISRCTN76149273.).

We conducted an investigator-initiated open-label, superiority, multicenter, factorial randomized trial in six hospitals in Uganda (n=2), Kenya (n=2), Niger (n=1) and Nigeria(n=1), detailed in Supplementary Methods. Of note, participants from Niger and Nigeria comprise more than 90% of the cohort. The protocol21 was approved by the local ethics committees and is available at nejm.org. KM, DMG, POO and ECG designed the study; SO,HS,AC,AS,WO,DA,EM,EO gathered the data; OFO,TS,MH,POO,FA,CL,JED,IC,MEC and RP supervised the study teams. The first author wrote the first draft of the manuscript, which was reviewed and agreed on by all the authors. CM, RC and ECG vouch for the for the accuracy and completeness of the data. Eligible children aged 6 months to 12 years hospitalized with SAM (defined as either weight-for-height z-score <-3, mid-upper arm circumference (MUAC) <11.5cm or presentation with edematous malnutrition (kwashiorkor) with at least bilateral pedal edema7) with gastroenteritis (>3 loose stools per day) and signs of severe dehydration were included. Severe dehydration signs, following WHO criteria, include two or more of the following-- altered consciousness based on a score of <15 on the four-component AVPU score of 3-15, where Alert is 15, responsiveness to Verbal or Painful stimuli are next and Unresponsive is the final category, sunken eyes, reduced skin turgor (slow abdominal skin pinch return>2s) or unable to take or retain oral fluids. Children with known congenital or rheumatic heart disease or with non-acute diarrhea (lasting>14 days) were excluded.

sponsiveness to Verbal or Painful stimuli are next and Unresponsive is the final category, sunken eyes, reduced skin turgor (slow abdominal skin pinch return>2s) or unable to take or retain oral fluids. Children with known congenital or rheumatic heart disease or with non-acute diarrhea (lasting>14 days) were excluded. All children with SAM admitted to hospital with an acute history of gastroenteritis were screened for inclusion by study staff. In Niger and Nigeria participants were transferred to an intensive care ward (although assisted ventilation was not available) for management by a dedicated study team (see Supplementary Appendix, Table S1 for generalizability). In Uganda and Kenya participants were managed on general pediatric wards. When prior written consent from parents or legal guardians could not be obtained, ethics committees approved verbal assent with delayed written informed consent as soon as practical22. Otherwise, written informed consent was obtained from parents or guardians before randomization. The statistician in London generated the sequential randomization list, computer-generated using variably-sized permuted blocks. Randomization at sites used consecutively numbered opaque sealed envelopes containing the randomized allocation, opened in numerical order.

All children with SAM admitted to hospital with an acute history of gastroenteritis were screened for inclusion by study staff. In Niger and Nigeria participants were transferred to an intensive care ward (although assisted ventilation was not available) for management by a dedicated study team (see Supplementary Appendix, Table S1 for generalizability). In Uganda and Kenya participants were managed on general pediatric wards. When prior written consent from parents or legal guardians could not be obtained, ethics committees approved verbal assent with delayed written informed consent as soon as practical22. Otherwise, written informed consent was obtained from parents or guardians before randomization. The statistician in London generated the sequential randomization list, computer-generated using variably-sized permuted blocks. Randomization at sites used consecutively numbered opaque sealed envelopes containing the randomized allocation, opened in numerical order. Participants were randomly assigned 2:1:1 to control (WHO SAM strategy): oral rehydration solution 5ml/kg every 30 minutes for the first 2hours followed by 5–10 ml/kg per hour for the next 4–10h alternating hourly with F75 milk formula, with boluses of Ringer’s Lactate (15 ml/kg) for those with shock); liberal:rapid rehydration (WHO Plan C: 100 ml/kg Ringer’s Lactate over 3-6 hours according to age with boluses (20 ml/kg) for those with shock) or liberal:slow rehydration (100 ml/kg Ringer’s Lactate over 8h and no boluses) (Tables S2-S5; Fig. S1). Shock was defined as all of the following: cold peripheries (meaning hands and-or feet)_, a weak and fast pulse (rate not specified) and a capillary refilling time >3s15.

ith boluses (20 ml/kg) for those with shock) or liberal:slow rehydration (100 ml/kg Ringer’s Lactate over 8h and no boluses) (Tables S2-S5; Fig. S1). Shock was defined as all of the following: cold peripheries (meaning hands and-or feet)_, a weak and fast pulse (rate not specified) and a capillary refilling time >3s15. For oral rehydration management, all participants were simultaneously factorially randomized (1:1) either to (Rehydration Solution for Malnutrition (ReSoMal)) or WHO oral rehydration solution (ORS) (recommended for non-SAM) (comparison reported separately, Supplementary Methods). Basic infrastructural support for emergency care, patient monitors, bedside hemoglobin, glucose and lactate point-of-care tests were provided. Bedside observations were performed at admission and every 30 minutes for the first 2 hours, hourly to 8hours, then at 12, 24, 36 and 48hours after randomization. Clinical chemistry was assessed at 0, 8 and 24hours. Blood cultures were performed where facilities permitted. Participants unable to tolerate oral fluids had a nasogastric tube placed to administer oral rehydration fluids and nutritional milk (called F75); its correct positioning was checked at each administration. Participants were actively monitored for serious adverse events (SAEs), particularly suspected cardiac or pulmonary overload, at each clinical assessment. Participants were clinically assessed at 7-days and 28-days (trial exit) post-randomization. Study staff were unblinded throughout; laboratory tests were assayed blinded.

cipants were actively monitored for serious adverse events (SAEs), particularly suspected cardiac or pulmonary overload, at each clinical assessment. Participants were clinically assessed at 7-days and 28-days (trial exit) post-randomization. Study staff were unblinded throughout; laboratory tests were assayed blinded. The primary endpoint was mortality at 96hours. Secondary efficacy endpoints were mortality to day-28; change in weight, MUAC, at Day 3 and Day 7; urine output at 8hours. Safety endpoints were evidence of pulmonary edema or heart failure; change in sodium at 24hours compared with 8hours (post intravenous strategy completion); and correction of electrolyte abnormalities (severe hyponatremia <125 mmol/L or hypokalemia <2.5mol/L).

n weight, MUAC, at Day 3 and Day 7; urine output at 8hours. Safety endpoints were evidence of pulmonary edema or heart failure; change in sodium at 24hours compared with 8hours (post intravenous strategy completion); and correction of electrolyte abnormalities (severe hyponatremia <125 mmol/L or hypokalemia <2.5mol/L). Enrolling 272 children with severe dehydration was anticipated to provide 80% power to detect a 30% relative reduction in 96-hour mortality from 58% in the control group to 41% in the liberal strategies, assuming no lost-to-follow-up by 96hours (2-sided alpha=0.05) (Supplementary Appendix, Methods). An independent Data Monitoring Committee reviewed interim data (four meetings). This report presents the pre-specified primary comparison for the liberal intravenous rehydration randomizations (pooling liberal:rapid and liberal:slow intravenous rehydration arms) vs. control, and also analysing separately vs. control. Randomized groups were compared with an intention-to-treat analysis using a Mantel-Haenszel adjusted risk ratio for mortality at 96hours (primary endpoint), adjusted for pre-specified covariate of site (hospital), and Cox regression for mortality by 28 days (secondary endpoint). Continuous outcomes were compared using linear regression to estimate mean difference and confidence intervals at each time point, and proportions using chi-squared tests (prespecified in the Statistical Analysis Plan (SAP): presented as odds ratios and confidence intervals from logistic regression). Time to correction of hyponatremia and hypokalemia were calculated using competing risks regression taking into account death. Confidence intervals were not adjusted for multiplicity and may not be used in place of hypothesis testing. Complete case analyses are presented for primary and secondary outcomes following the SAP under a missing completely at random assumption as missingness was evenly distributed between arms and below the pre-defined threshold (10%); however, for secondary outcomes, where missingness was close to the threshold, multiple imputation was conducted under the missing at random assumption (Supplementary Appendix, Methods). Analyses used Stata v18.

We conducted an investigator-initiated open-label, superiority, multicenter, factorial randomized trial in six hospitals in Uganda (n=2), Kenya (n=2), Niger (n=1) and Nigeria(n=1), detailed in Supplementary Methods. Of note, participants from Niger and Nigeria comprise more than 90% of the cohort. The protocol21 was approved by the local ethics committees and is available at nejm.org. KM, DMG, POO and ECG designed the study; SO,HS,AC,AS,WO,DA,EM,EO gathered the data; OFO,TS,MH,POO,FA,CL,JED,IC,MEC and RP supervised the study teams. The first author wrote the first draft of the manuscript, which was reviewed and agreed on by all the authors. CM, RC and ECG vouch for the for the accuracy and completeness of the data.

Eligible children aged 6 months to 12 years hospitalized with SAM (defined as either weight-for-height z-score <-3, mid-upper arm circumference (MUAC) <11.5cm or presentation with edematous malnutrition (kwashiorkor) with at least bilateral pedal edema7) with gastroenteritis (>3 loose stools per day) and signs of severe dehydration were included. Severe dehydration signs, following WHO criteria, include two or more of the following-- altered consciousness based on a score of <15 on the four-component AVPU score of 3-15, where Alert is 15, responsiveness to Verbal or Painful stimuli are next and Unresponsive is the final category, sunken eyes, reduced skin turgor (slow abdominal skin pinch return>2s) or unable to take or retain oral fluids. Children with known congenital or rheumatic heart disease or with non-acute diarrhea (lasting>14 days) were excluded.

All children with SAM admitted to hospital with an acute history of gastroenteritis were screened for inclusion by study staff. In Niger and Nigeria participants were transferred to an intensive care ward (although assisted ventilation was not available) for management by a dedicated study team (see Supplementary Appendix, Table S1 for generalizability). In Uganda and Kenya participants were managed on general pediatric wards. When prior written consent from parents or legal guardians could not be obtained, ethics committees approved verbal assent with delayed written informed consent as soon as practical22. Otherwise, written informed consent was obtained from parents or guardians before randomization. The statistician in London generated the sequential randomization list, computer-generated using variably-sized permuted blocks. Randomization at sites used consecutively numbered opaque sealed envelopes containing the randomized allocation, opened in numerical order.

ith boluses (20 ml/kg) for those with shock) or liberal:slow rehydration (100 ml/kg Ringer’s Lactate over 8h and no boluses) (Tables S2-S5; Fig. S1). Shock was defined as all of the following: cold peripheries (meaning hands and-or feet)_, a weak and fast pulse (rate not specified) and a capillary refilling time >3s15. For oral rehydration management, all participants were simultaneously factorially randomized (1:1) either to (Rehydration Solution for Malnutrition (ReSoMal)) or WHO oral rehydration solution (ORS) (recommended for non-SAM) (comparison reported separately, Supplementary Methods).

Basic infrastructural support for emergency care, patient monitors, bedside hemoglobin, glucose and lactate point-of-care tests were provided. Bedside observations were performed at admission and every 30 minutes for the first 2 hours, hourly to 8hours, then at 12, 24, 36 and 48hours after randomization. Clinical chemistry was assessed at 0, 8 and 24hours. Blood cultures were performed where facilities permitted. Participants unable to tolerate oral fluids had a nasogastric tube placed to administer oral rehydration fluids and nutritional milk (called F75); its correct positioning was checked at each administration. Participants were actively monitored for serious adverse events (SAEs), particularly suspected cardiac or pulmonary overload, at each clinical assessment. Participants were clinically assessed at 7-days and 28-days (trial exit) post-randomization. Study staff were unblinded throughout; laboratory tests were assayed blinded.

The primary endpoint was mortality at 96hours. Secondary efficacy endpoints were mortality to day-28; change in weight, MUAC, at Day 3 and Day 7; urine output at 8hours. Safety endpoints were evidence of pulmonary edema or heart failure; change in sodium at 24hours compared with 8hours (post intravenous strategy completion); and correction of electrolyte abnormalities (severe hyponatremia <125 mmol/L or hypokalemia <2.5mol/L).

Enrolling 272 children with severe dehydration was anticipated to provide 80% power to detect a 30% relative reduction in 96-hour mortality from 58% in the control group to 41% in the liberal strategies, assuming no lost-to-follow-up by 96hours (2-sided alpha=0.05) (Supplementary Appendix, Methods). An independent Data Monitoring Committee reviewed interim data (four meetings). This report presents the pre-specified primary comparison for the liberal intravenous rehydration randomizations (pooling liberal:rapid and liberal:slow intravenous rehydration arms) vs. control, and also analysing separately vs. control. Randomized groups were compared with an intention-to-treat analysis using a Mantel-Haenszel adjusted risk ratio for mortality at 96hours (primary endpoint), adjusted for pre-specified covariate of site (hospital), and Cox regression for mortality by 28 days (secondary endpoint). Continuous outcomes were compared using linear regression to estimate mean difference and confidence intervals at each time point, and proportions using chi-squared tests (prespecified in the Statistical Analysis Plan (SAP): presented as odds ratios and confidence intervals from logistic regression). Time to correction of hyponatremia and hypokalemia were calculated using competing risks regression taking into account death. Confidence intervals were not adjusted for multiplicity and may not be used in place of hypothesis testing. Complete case analyses are presented for primary and secondary outcomes following the SAP under a missing completely at random assumption as missingness was evenly distributed between arms and below the pre-defined threshold (10%); however, for secondary outcomes, where missingness was close to the threshold, multiple imputation was conducted under the missing at random assumption (Supplementary Appendix, Methods). Analyses used Stata v18.

Between September 2nd 2019 and October 27th 2024, 272 participants were randomized--138 to control and 134 to liberal intravenous rehydration (67 liberal:rapid, 67 liberal:slow); all are included in all analyses (Fig.1;Fig.S2). Recruitment was halted between March 2020 through November 2021 due to COVID (Fig.S3). There were few imbalances in baseline characteristics between randomized groups (Table 1), fewer than expected by chance. Most children had three or more signs of dehydration (267(98%) sunken eyes, 242(89%) decreased skin turgor); 215(79%) were unable to take or retain oral fluids) and 76(29%) had moderate hypotension. Previously identified risk factors for mortality4,5 were highly prevalent including impaired consciousness (104:38%), bacteremia (largely gram-negative) (12/98 tested:12%); severe hyponatremia (sodium <125mmol/L) (137(52%)) and hypokalemia (potassium <2.5 mmol/L) (115(45%)). Few had kwashiorkor (11(4%)) or known HIV (2(1%)).

usly identified risk factors for mortality4,5 were highly prevalent including impaired consciousness (104:38%), bacteremia (largely gram-negative) (12/98 tested:12%); severe hyponatremia (sodium <125mmol/L) (137(52%)) and hypokalemia (potassium <2.5 mmol/L) (115(45%)). Few had kwashiorkor (11(4%)) or known HIV (2(1%)). Thirty-one (22%) participants in the control arm received intravenous fluids within 24h, starting median(IQR) 123(13-470) mins from randomization; including 12(9%) with shock receiving immediate boluses and 14(9%) following later development of shock or another serious adverse event (Table 2; Table S6). Sixty-six children (99%) in the liberal:rapid arm received intravenous fluids starting median(IQR) 16(10-28) mins from randomization; 7 of 8 with shock received an immediate bolus (1 died prior to bolus administration). Sixty-seven participants (100%) in the liberal:slow arm received intravenous fluids without boluses (5 with shock at baseline) starting median(IQR) 12(8-22) minutes from randomization). Oral rehydration started in 135/138(98%) participants in the control arm (2 died before starting; 1 missing form) (median(IQR) 0.3(0.2-0.5) hours), with 126 (92%) requiring a nasogastric tube. Post-intravenous rehydration ORS started in 64 children in the liberal:rapid arm (1 died before starting; 1 missing form) (median(IQR) 5.3(4.0-7.0) hours) (67% via nasogastric tube), and 62 in the liberal:slow arm (2 missing forms) (median 8.7 (8.4-9.2) hours) (63% via nasogastric tube). Vomiting and nasogastric tube insertion to administer oral rehydration were greater in the control group (Table 2).

died before starting; 1 missing form) (median(IQR) 5.3(4.0-7.0) hours) (67% via nasogastric tube), and 62 in the liberal:slow arm (2 missing forms) (median 8.7 (8.4-9.2) hours) (63% via nasogastric tube). Vomiting and nasogastric tube insertion to administer oral rehydration were greater in the control group (Table 2). At 96hours (primary endpoint) and day-28 (end of follow-up), 271 (99%) and 267 (98%) children, respectively, had known vital status. By 96hours, 11 (8%) in control arm vs. 9 (7%) in liberal rehydration (5 (7%) liberal:rapid; 4 (6%) liberal:slow) had died (Adjusted Risk Ratio (aRR) (liberal vs. control) =1.02 (95% CI 0.41,2.52); p=0.69, Table 3;Fig.1). At day-28 17(12%) control vs. 14(10%) liberal rehydration (8(12%) liberal:rapid, 6(9%) liberal:slow) participants had died (Hazard Ratio (HR)=0.85 (0.41,1.78)). Separate comparisons of 96-hour mortality were for liberal:rapid vs. control aRR=1.16 (0.40,3.40) and liberal:slow vs. control aRR=0.89 (0.28,2.80) (Tables S7-S8; Figs. S4-S8). Findings for the primary outcome were consistent across four prespecified subgroups: oral rehydration solution randomization, age (< or ≥1y), consciousness level and respiratory distress at randomization (Table S9).

trol aRR=1.16 (0.40,3.40) and liberal:slow vs. control aRR=0.89 (0.28,2.80) (Tables S7-S8; Figs. S4-S8). Findings for the primary outcome were consistent across four prespecified subgroups: oral rehydration solution randomization, age (< or ≥1y), consciousness level and respiratory distress at randomization (Table S9). No child developed pulmonary edema or signs consistent with heart failure in the trial. There was no evidence of a difference between arms for Serious Adverse Events (SAEs; Table 3 and Tables S10, S11). Deterioration in consciousness level or seizures occurred in 18(13%) control and 10(8%) liberal rehydration (odds ratio (OR) liberal vs. control=0.54 (95%CI 0.24-1.23)). Shock developed in 11(9%) vs. 6(5%) respectively (odds ratio liberal vs. control OR=0.55 (95%CI 0.19-1.53). By 8h and 24h significantly more children randomized to control had severe hyponatremia (58/126(45%) and 35/126(27%) respectively) than liberal rehydration (20/128(16%) and 21/127(17%) respectively) (liberal vs. control OR=0.23(95%CI 0.13-0.41) and OR=0.53 (95%CI 0.29-0.98) respectively) (Table 3). The liberal arms corrected their severe hyponatremia quicker (sub-HR=1.55 (1.14,2.09)). Potassium increased more slowly with liberal arms at 8h and 24h (liberal vs. control mean (95%CI) difference -0.3 (-0.5,-0.2) and -0.4 (-0.6,-0.2) respectively) but there was no evidence of difference in time to correction of severe hypokalemia (sub-HR=0.87 (0. 57,1.19)). Day-3 weight increased more with liberal rehydration (+0.5 (0.4,0.5) vs. +0.4 (0.3,0.4) kg, mean difference +0.1 (0.1,0.2)) but there was no evidence of difference by day-7 (Table 3). Similar findings were reflected in other anthropometric measures (Tables S7-8). Results from complete case analyses were not sensitive to the missing completely at random assumption (Table S12).

Thirty-one (22%) participants in the control arm received intravenous fluids within 24h, starting median(IQR) 123(13-470) mins from randomization; including 12(9%) with shock receiving immediate boluses and 14(9%) following later development of shock or another serious adverse event (Table 2; Table S6). Sixty-six children (99%) in the liberal:rapid arm received intravenous fluids starting median(IQR) 16(10-28) mins from randomization; 7 of 8 with shock received an immediate bolus (1 died prior to bolus administration). Sixty-seven participants (100%) in the liberal:slow arm received intravenous fluids without boluses (5 with shock at baseline) starting median(IQR) 12(8-22) minutes from randomization). Oral rehydration started in 135/138(98%) participants in the control arm (2 died before starting; 1 missing form) (median(IQR) 0.3(0.2-0.5) hours), with 126 (92%) requiring a nasogastric tube. Post-intravenous rehydration ORS started in 64 children in the liberal:rapid arm (1 died before starting; 1 missing form) (median(IQR) 5.3(4.0-7.0) hours) (67% via nasogastric tube), and 62 in the liberal:slow arm (2 missing forms) (median 8.7 (8.4-9.2) hours) (63% via nasogastric tube). Vomiting and nasogastric tube insertion to administer oral rehydration were greater in the control group (Table 2).

At 96hours (primary endpoint) and day-28 (end of follow-up), 271 (99%) and 267 (98%) children, respectively, had known vital status. By 96hours, 11 (8%) in control arm vs. 9 (7%) in liberal rehydration (5 (7%) liberal:rapid; 4 (6%) liberal:slow) had died (Adjusted Risk Ratio (aRR) (liberal vs. control) =1.02 (95% CI 0.41,2.52); p=0.69, Table 3;Fig.1). At day-28 17(12%) control vs. 14(10%) liberal rehydration (8(12%) liberal:rapid, 6(9%) liberal:slow) participants had died (Hazard Ratio (HR)=0.85 (0.41,1.78)). Separate comparisons of 96-hour mortality were for liberal:rapid vs. control aRR=1.16 (0.40,3.40) and liberal:slow vs. control aRR=0.89 (0.28,2.80) (Tables S7-S8; Figs. S4-S8). Findings for the primary outcome were consistent across four prespecified subgroups: oral rehydration solution randomization, age (< or ≥1y), consciousness level and respiratory distress at randomization (Table S9).

No child developed pulmonary edema or signs consistent with heart failure in the trial. There was no evidence of a difference between arms for Serious Adverse Events (SAEs; Table 3 and Tables S10, S11). Deterioration in consciousness level or seizures occurred in 18(13%) control and 10(8%) liberal rehydration (odds ratio (OR) liberal vs. control=0.54 (95%CI 0.24-1.23)). Shock developed in 11(9%) vs. 6(5%) respectively (odds ratio liberal vs. control OR=0.55 (95%CI 0.19-1.53). By 8h and 24h significantly more children randomized to control had severe hyponatremia (58/126(45%) and 35/126(27%) respectively) than liberal rehydration (20/128(16%) and 21/127(17%) respectively) (liberal vs. control OR=0.23(95%CI 0.13-0.41) and OR=0.53 (95%CI 0.29-0.98) respectively) (Table 3). The liberal arms corrected their severe hyponatremia quicker (sub-HR=1.55 (1.14,2.09)). Potassium increased more slowly with liberal arms at 8h and 24h (liberal vs. control mean (95%CI) difference -0.3 (-0.5,-0.2) and -0.4 (-0.6,-0.2) respectively) but there was no evidence of difference in time to correction of severe hypokalemia (sub-HR=0.87 (0. 57,1.19)). Day-3 weight increased more with liberal rehydration (+0.5 (0.4,0.5) vs. +0.4 (0.3,0.4) kg, mean difference +0.1 (0.1,0.2)) but there was no evidence of difference by day-7 (Table 3). Similar findings were reflected in other anthropometric measures (Tables S7-8). Results from complete case analyses were not sensitive to the missing completely at random assumption (Table S12).

This multicenter trial conducted in resource-poor conditions did not observe a reduction in mortality between standard and liberal rehydration strategies in patients with SAM. The use of liberal rehydration strategies, both intravenous and oral were not associated with cardiac or pulmonary complications and resulted in fewer patients requiring fluid boluses for shock and placement of nasogastric tubes.

ion in mortality between standard and liberal rehydration strategies in patients with SAM. The use of liberal rehydration strategies, both intravenous and oral were not associated with cardiac or pulmonary complications and resulted in fewer patients requiring fluid boluses for shock and placement of nasogastric tubes. The key limitation of our trial was the much lower overall mortality (9%) than we predicted from two small studies, which reported mortality at hospital discharge or 28 days as 68-82%14,15, at the high end compared with other observational data. A key reason for low mortality in our trial may be that most children were managed on high dependency wards (or step-down, intermediate care units) by dedicated clinical trial teams with very close and frequent monitoring to identify and treat complications (specifically fluid overload, shock or hypoglycemia) and to ensure protocol adherence. These measures were put in place to address concerns by ethics committees over the balance of risk to benefit for children participating in the trial, but may reduce the generalisability of the findings. The concerns resulted from longstanding national and international guidance recommending against IV rehydration in children with SAM arising from the perceived risk of incipient heart failure7. In routine practice in low-resourced, overcrowded pediatric wards in Africa, the close clinical monitoring afforded by our trial is not possible, as evidenced by the poor outcomes reported in SAM with severe dehydration under routine surveillance4–6. This underscores the need for simplified protocols for the management of dehydration. For example, at admission 79% of children were unable to take oral rehydration, resulting in 93% of the control arm requiring nasogastric tubes for oral rehydration solution administration. This is not a trivial low-risk procedure, especially in children with impaired consciousness and high purging rates. The current recommendations have resulted in additional demands on limited ward personnel, since oral rehydration solution cannot be given by the child’s caregiver (as noted in guidance documents). In contrast, slow intravenous rehydration, even compared to rapid rehydration, was simpler to implement requiring no calculation of volumes for boluses and adjustment for the rapid and slower rehydration phases depending on age.

dration solution cannot be given by the child’s caregiver (as noted in guidance documents). In contrast, slow intravenous rehydration, even compared to rapid rehydration, was simpler to implement requiring no calculation of volumes for boluses and adjustment for the rapid and slower rehydration phases depending on age. Another limitation was low participant numbers with kwashiorkor, a key group expected to be at significant risk of heart failure. However, our research group has previously demonstrated that myocardial function is preserved in children with SAM with no difference in fractional shortening (a global measure of myocardial function) in children with marasmus (severe wasting) and kwashiorkor16. Thus, results are likely to be generalizable to this subgroup. Relevant to the broader population of children hospitalized with acute diarrhea with severe dehydration (~10% weight loss), a study showed that approximately 20% temporarily fulfilled anthropometric criteria for SAM (MUAC<11.5cm) but were ‘reclassified’ as undernourished following rehydration.24 Thus, through ‘slippage’ the current recommendations may have wider implications, as potentially 20% of non-SAM children could be inappropriately diagnosed as malnourished and rehydrated. This may have contributed to the poor outcomes observed in the Global Enteric Multicentre (GEMS) study23.

d following rehydration.24 Thus, through ‘slippage’ the current recommendations may have wider implications, as potentially 20% of non-SAM children could be inappropriately diagnosed as malnourished and rehydrated. This may have contributed to the poor outcomes observed in the Global Enteric Multicentre (GEMS) study23. Currently, at the bedside, clinicians need to consider nutritional status, age and the presence of shock to determine which rehydration strategy to follow. Given the findings of this study, in the absence of other data, we would suggest that current guidance be reviewed to consider simplifying rehydration protocols, removing the distinction in management between SAM and non-malnourished children, which would be more pragmatic in the under-resourced settings where most children are managed. While there was no evidence of a difference in mortality at 96 hours between the rehydration strategies evaluated in our trial, power to detect modest differences was low. Specifically, there was no apparent signal of harm for the liberal intravenous rehydration strategies including no evidence of fluid overload nor evidence of sodium overload compared with the currently WHO-recommended oral rehydration strategy. In summary, we detected no difference in mortality among the rehydration strategies used in children in the present factorial open-label superiority trial. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.