Immediate fluid management of children with severe febrile illness and signs of impaired circulation in low-income settings: a contextualised systematic review

Objective To evaluate the effects of intravenous fluid bolus compared to maintenance intravenous fluids alone as part of immediate emergency care in children with severe febrile illness and signs of impaired circulation in low-income settings. Design Systematic review of randomised controlled trials (RCTs), and observational studies, including retrospective analyses, that compare fluid bolus regimens with maintenance fluids alone. The primary outcome measure was predischarge mortality. Data sources and synthesis We searched PubMed, The Cochrane Library (to January 2014), with complementary earlier searches on, Google Scholar and Clinical Trial Registries (to March 2013). As studies used different clinical signs to define impaired circulation we classified patients into those with signs of severely impaired circulation, or those with any signs of impaired circulation. The quality of evidence for each outcome was appraised using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Findings are presented as risk ratios (RRs) with 95% CIs. Results Six studies were included. Two were RCTs, one large trial (n=3141 children) from a low-income country and a smaller trial from a middle-income country. The remaining studies were from middle-income or high-income settings, observational, and with few participants (34–187 children). Severely impaired circulation The large RCT included a small subgroup with severely impaired circulation. There were more deaths in those receiving bolus fluids (20–40 mL/kg/h, saline or albumin) compared to maintenance fluids (2.5–4 mL/kg/h; RR 2.40, 95% CI 0.84 to 6.88, p=0.054, 65 participants, low quality evidence). Three additional observational studies, all at high risk of confounding, found mixed effects on mortality (very low quality evidence). Any signs of impaired circulation The large RCT included children with signs of both severely and non-severely impaired circulation. Overall, bolus fluids increased 48 h mortality compared to maintenance fluids with an additional 3 deaths per 100 children treated (RR 1.45, 95% CI 1.13 to 1.86, 3141 participants, high quality evidence). In a second small RCT from India, no difference in 72 h mortality was detected between children who received 20–40 mL/kg Ringers lactate over 15 min and those who received 20 mL over 20 min up to a maximum of 60 mL/kg over 1 h (147 participants, low quality evidence). In one additional observational study, resuscitation consistent with Advanced Paediatric Life Support (APLS) guidelines, including fluids, was not associated with reduced mortality in the small subgroup with septic shock (very low quality evidence). Signs of impaired circulation, but not severely impaired Only the large RCT allowed an analysis for children with some signs of impaired circulation who would not meet the criteria for severe impairment. Bolus fluids increased 48 h mortality compared to maintenance alone (RR 1.36, 95% CI 1.05 to 1.76, high quality evidence). Conclusions Prior to the publication of the large RCT, the global evidence base for bolus fluid therapy in children with severe febrile illness and signs of impaired circulation was of very low quality. This large study provides robust evidence that in low-income settings fluid boluses increase mortality in children with severe febrile illness and impaired circulation, and this increased risk is consistent across children with severe and less severe circulatory impairment.

children in low-income countries. This review was clearly prompted by the discussions and commentaries that followed the publication of the FEAST trial in 2011. This landmark, high quality randomised-controlled trial challenged fundamental and strongly-held perspectives on fluid resuscitation in children and drew substantial criticism primarily relating to the generalisability of its findings in high and middle-income countries. Unsurprisingly, many of the commentators have emanated from high income countries and include prominent members of guidelinewriting committees. Despite the validity and probity of the FEAST study and the publication of subsequent analyses defining potential biological mechanisms associated with bolus-induced mortality in this population, practice guidelines, including those from the WHO continue to recommend boluses for fluid resuscitation in this population. The evidence on which these guidelines have been based are largely of low quality, compromising observational and cohort studies, all of which are subject to high levels of bias. This review is therefore timely and clearly defines the evidence base, both in terms of quantity and validity. The objectives, methodology and internal validity of this systematic review is of the highest quality and conforms to all criteria defined in the PRISMA process. The FEAST study dominates the dataset and analyses -this is an important outcome from the review and places the weighting of lowquality evidence into perspective. The results are presented in three pre-defined strata and the conclusions are consistent with the level of evidence. The discussion is appropriately conservative and there is little editorial comment, which is commendable given the high level of emotive commentary on this subject. The authors are to be commended on conducting this review.

REVIEWER
Eddy Lang University of Calgary, Calgary, Alberta, Canada Member GRADE working group REVIEW RETURNED 17-Mar-2014

GENERAL COMMENTS
The authors present a systematic review focusing on the mortality effects of high volume versus maintenance levels of fluid resuscitation in children with sepsis who present in developing health care systems. The authors identify and include two RCTs with one being the dominant work in this area as well as a series of observational studies. The review addresses an important and controversial question and provides an important contribution to the medical literature. While the findings are driven by the Maitland trial there is still merit in exploring the impact of aggressive fluid intervention across other trials in the hopes of facilitating generalizability.
In regards to methodology, the authors are adherent to PRISMA guidelines (except for providing a protocol or registering with PROSPERO) and do a good job applying the GRADE framework to rating the quality of evidence.

Specific comments:
Abstract: Creating sub-divisions in the result section is an uncommon format and consolidation into a unified results section would be preferable.

Synthesis of results:
This section is confusing as the three categories identified do not appear to be mutually exclusive with significant overlap between the second and third. Perhaps a table would be helpful here. Summary of findings tables (GRADE) Please note that these should be updated. The definitions as they pertain to quality of evidence are no longer valid. Evidence is conceptualized as confidence in estimates of effect and not in terms of the likelihood of future research making a difference. Table 4 Foot note 3 is problematic. The basis of the indirectness is unclear i.e. why is the African setting problematic? Also serious indirectness usually implies a drop by two levels. GRADE evidence profiles would be useful to include as well to facilitate a better understanding of the limitations of the evidence base.

VERSION 1 -AUTHOR RESPONSE
Reviewer: 1 This study highlights the findings of the large RCT from East Africa (FEAST Trial) which has significantly improved the evidence base regarding fluid management in children in low-income settings with severe infection (particularly P. falciparum infection) and impaired circulation.
The study should revise its statement in the discussion section that the findings of the FEAST Trial important to East Africa can be extrapolated to affect "those involved in paediatric care in highresource settings." In these industrialized countries, children with septic shock likely present earlier and have relevant ventilator, inotropic, and monitoring support within an ICU.
Authors" responses Agree: statement on the applicability of FEAST trial findings to high-resource settings now omitted.

Reviewer: 2
This is a well conducted systematic review of fluid resuscitation in children in low-income countries. This review was clearly prompted by the discussions and commentaries that followed the publication of the FEAST trial in 2011. This landmark, high quality randomised-controlled trial challenged fundamental and strongly-held perspectives on fluid resuscitation in children and drew substantial criticism primarily relating to the generalisability of its findings in high and middle-income countries.
Unsurprisingly, many of the commentators have emanated from high income countries and include prominent members of guideline-writing committees. Despite the validity and probity of the FEAST study and the publication of subsequent analyses defining potential biological mechanisms associated with bolus-induced mortality in this population, practice guidelines, including those from the WHO continue to recommend boluses for fluid resuscitation in this population.
The evidence on which these guidelines have been based largely of low quality, compromising observational and cohort studies, all of which are subject to high levels of bias. This review is therefore timely and clearly defines the evidence base, both in terms of quantity and validity. The objectives, methodology and internal validity of this systematic review is of the highest quality and conforms to all criteria defined in the PRISMA process.
The FEAST study dominates the dataset and analyses -this is an important outcome from the review and places the weighting of low-quality evidence into perspective. The results are presented in three pre-defined strata and the conclusions are consistent with the level of evidence.
The discussion is appropriately conservative and there is little editorial comment, which is commendable given the high level of emotive commentary on this subject. The authors are to be commended on conducting this review.
Authors" responses Agree with the reviewer"s comments.
Reviewer: 3 The authors present a systematic review focusing on the mortality effects of high volume versus maintenance levels of fluid resuscitation in children with sepsis who present in developing health care systems. The authors identify and include two RCTs with one being the dominant work in this area as well as a series of observational studies.
The review addresses an important and controversial question and provides an important contribution to the medical literature. While the findings are driven by the Maitland trial there is still merit in exploring the impact of aggressive fluid intervention across other trials in the hopes of facilitating generalizability.
In regards to methodology, the authors are adherent to PRISMA guidelines (except for providing a protocol or registering with PROSPERO) and do a good job applying the GRADE framework to rating the quality of evidence.

Specific comments:
Abstract: Creating sub-divisions in the result section is an uncommon format and consolidation into a unified results section would be preferable.
Authors" response Agree, results now consolidated and presented in paragraphs.

Synthesis of results:
This section is confusing as the three categories identified do not appear to be mutually exclusive with significant overlap between the second and third. Perhaps a table would be helpful here.

Authors" response
We have now included a table (table 1) describing the clinical features of the three categories of circulatory impairment.

Summary of findings tables (GRADE)
Please note that these should be updated. The definitions as they pertain to quality of evidence are no longer valid. Evidence is conceptualized as confidence in estimates of effect and not in terms of the likelihood of future research making a difference.
Authors" response The definitions of GRADE quality of evidence have now been revised to reflect confidence in estimates of effect (Supplementary tables 2 and 4): High quality: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect. Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect. Table 4 Foot note 3 is problematic. The basis of the indirectness is unclear i.e. why is the African setting problematic? Also serious indirectness usually implies a drop by two levels.
Authors" response The main reason for downgrading was because of the indirectness of the patient populationthat is, children with severe hypotension were excluded (effects of treatments administered were likely to be substantially different in this higher risk group).
We have now omitted the following statements from the footnote: "The trial was conducted in hospital settings in Kenya, Tanzania and Uganda. Children were aged 2 months to 12 years. Children with severe malnutrition, gastroenteritis, or shock due to trauma surgery or burns were excluded." According to GRADE guidelines serious indirectness implies a drop by one level while very serious indirectness, for example due to problems in more than one PICO (Patient, Intervention, Comparison, Outcome) element suggest a need to rate down the quality of evidence by two levels: Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, Alonso-Coello P, Falck-Ytter Y, Jaeschke R, Vist G, Akl EA, Post PN, Norris S, Meerpohl J, Shukla VK, Nasser M, Schünemann HJ; GRADE Working Group. GRADE guidelines: 8. Rating the quality of evidence--indirectness. J Clin Epidemiol. 2011 Dec;64(12):1303-10.
GRADE evidence profiles would be useful to include as well to facilitate a better understanding of the limitations of the evidence base.
Authors" response Agree, we have now included GRADE evidence profiles documenting our judgement of the determinants of quality of evidence for each outcome, in addition to the summary of findings for each outcome (Supplementary tables 3 and 5).