2019 surveillance of intravenous fluid therapy in children and young people in hospital (2015) NICE guideline NG29 – summary of evidence
Closed for comments This consultation ended on at
A randomised controlled trial (RCT) (Liu et al. 2015) (n=44 children) compared 3% sodium chloride (6 ml/kg as a single bolus over 10–15 minutes, maximum 2 boluses) with 0.9% sodium chloride (guided by standard therapy) for early intravenous fluid resuscitation in septic shock in critical care. There were no significant differences between groups in oxygenation index at 1, 3, 6 and 24 hours after infusion. Plasma sodium was significantly higher with 3% than with 0.9% sodium chloride at 1 hour after infusion (though both groups were in the normal range), but not at 3, 6 and 24 hours after infusion. At 6 and 24 hours after treatment, fluid infusion volume was significantly less with 3% than with 0.9% sodium chloride.
An RCT (Allen et al. 2016) (n=100 children) across 8 emergency departments in the USA and Canada examined Plasma-Lyte A versus 0.9% sodium chloride as intravenous fluid for moderate to severe dehydration secondary to acute gastroenteritis. Both treatment groups received similar fluid volumes. Plasma-Lyte led to a significantly greater improvement in the primary outcome of 4-hour serum bicarbonate level from baseline (i.e. improvement in hyperchloraemic acidosis) than the 0.9% sodium chloride group. There was significantly less abdominal pain and better dehydration scores with Plasma-Lyte at hour 2 but not at hour 4. No patient experienced clinically relevant worsening of laboratory findings or physical examination, and hospital admission rates were similar. One patient in each group developed hyponatraemia, and 4 patients developed hyperkalaemia (1 with Plasma-Lyte and 3 with sodium chloride; between-group significance not reported in the abstract).
An RCT (Kartha et al. 2017) (n=68 children) examined Ringer's lactate versus 0.9% sodium chloride for intravenous resuscitation in acute severe diarrhoeal dehydration. Patients were given 100 ml/kg of the assigned fluid according to World Health Organization Plan C to treat severe dehydration quickly (infants under 1 year: 30 ml/kg for 1 hour, then 70 ml/kg for next 5 hours; children over 1 year: 30 ml/kg for 30 minutes, then 70 ml/kg for next 2.5 hours). There was no significant difference between groups in the primary outcome (improvement in clinical status and pH ≥7.35 at the end of 6 hours) or secondary outcomes (electrolytes, renal and blood gas parameters, time to start oral feeding, and hospital stay). No patients needed a second cycle of dehydration correction. The median total cost was significantly higher with Ringer's lactate then normal saline.
Topic experts queried whether balanced crystalloids were superior to 0.9% sodium chloride, and noted they were being used more widely in practice.
The new evidence found plasma sodium was higher with 3% than with 0.9% sodium chloride at 1 hour after infusion (though both groups were in the normal range), but not at 3, 6 and 24 hours after infusion. There were no differences between groups in oxygenation index at any timepoint after infusion (oxygenation index was not deemed an important outcome by the original guideline). Less fluid volume was needed with 3% sodium chloride, though benefits of this for patients were not apparent in the outcomes reported by the study. The evidence is unlikely to affect recommendation 1.3.1 to use glucose-free crystalloids for fluid resuscitation that contain sodium in the range 131–154 mmol/litre (of which 0.9% sodium chloride is an example).
New evidence is unlikely to change guideline recommendations.
From the new evidence, one study found Plasma-Lyte A for fluid resuscitation led to increased serum bicarbonate (i.e. less acidosis), less abdominal pain, and better dehydration scores versus 0.9% sodium chloride, but had no benefit for laboratory findings, physical examination, or hospital admission. Although there were some benefits of Plasma-Lyte A, the evidence is from a single study in a single condition (acute gastroenteritis). Evidence of benefit of balanced crystalloids from further studies in other conditions, and looking at other outcomes such as mortality and hospital stay that were deemed important by the original guideline, is therefore needed before considering any changes to current recommendations. The guideline does not currently name specific fluids for resuscitation – recommendation 1.3.1 allows for any glucose-free crystalloids that contain sodium in the range 131–154 mmol/litre (of which Plasma-Lyte A and 0.9% sodium chloride are both examples). Plasma-Lyte A may not be available in the UK, however it appears to have the same formulation as Plasma-Lyte 148 which is licensed in the UK.
The finding from a second study that Ringer's lactate had no benefit over 0.9% sodium chloride but was more costly is also unlikely to affect recommendation 1.3.1 to use glucose-free crystalloids that contain sodium in the range 131–154 mmol/litre (of which Ringer's lactate and 0.9% sodium chloride are both examples). Additionally, the fluid administration protocol in the study differed considerably from recommendation 1.3.1 which limits its relevance to the guideline.
New evidence is unlikely to change guideline recommendations
A Cochrane review (Li at al. 2018) of 3 RCTs (n=3,402 children) compared liberal and conservative fluid therapy in initial sepsis or septic shock. The liberal and conservative therapies compared by the 3 studies were:
20 ml/kg of 5% albumin bolus versus 20 ml/kg 0.9% sodium chloride bolus versus 1.2 ml/kg no bolus control
40 ml/kg fluid over 15 minutes versus 20 ml/kg over 20 minutes
80% of maintenance (liberal approach) versus fluid input based on calculated fluid overload (conservative approach).
The review permitted evidence from adults and children but found only paediatric studies. Liberal fluid therapy significantly increased the risk of the 2 primary outcomes: in-hospital mortality (2 studies, n=3,288 children; moderate-quality evidence) and mortality at 4-week follow-up (1 study, n=3,141 children; high-quality evidence). However, results for in-hospital mortality in the third study were inconclusive (very low-quality evidence). The effect on adverse events was uncertain because results were imprecise and low quality.
A UK pilot RCT across 13 NHS hospitals as part of a health technology assessment (Inwald et al. 2018) (n=75 children) compared restricted fluid (10 ml/kg bolus every 15 minutes for 4 hours) with current practice (20 ml/kg bolus every 15 minutes for 4 hours) in children presenting to an emergency department with clinical suspicion of infection and septic shock after 20 ml/kg of fluid. There were no deaths. Length of hospital stay, paediatric intensive care unit (PICU) transfers, and days alive and PICU free did not differ significantly between the groups. Two adverse events were reported in each group.
An RCT (Sankar et al. 2017) (n=96 children) assessed fluid resuscitation in septic shock in paediatric emergency and critical care settings. The study examined 40–60 ml/kg of fluids as fluid boluses in aliquots of 20 ml/kg each over 15–20 minutes versus over 5–10 minutes in the first hour of resuscitation. Compared with the 5–10 minutes group, significantly fewer children in the 15–20 minutes group needed mechanical ventilation, or had an increase in oxygenation index by 5 (lower value is better) from baseline, in the first 6 hours and 24 hours after fluid resuscitation. There was no difference in secondary outcomes of death, length of stay, or resolution of shock.
An RCT (Houston et al. 2019) in Uganda and Kenya (n=122 children) examined standard rehydration with World Health Organization Plan C for severe dehydration (100 ml/kg Ringer's lactate over 3 hours, or 6 hours if age less than 1 year, incorporating 0.9% sodium chloride boluses for shock), versus slower rehydration (100 ml/kg Ringer's lactate over 8 hours for all ages, without boluses). By 48 hours, there was no significant difference between groups in the primary outcome of number of serious adverse events (including cardiovascular, respiratory and neurological complications): 3 events with rapid versus 2 events with slower rehydration. There was also no significant difference in time to correction of dehydration or time to discharge between groups.
Topic experts queried whether a 20 ml/kg resuscitation bolus was still appropriate, but noted that this value was still used by Advanced Paediatric Life Support – a course run by the Advanced Life Support Group.
An enquiry to NICE raised concerns about the absence from the guideline of a specific maximum resuscitation bolus volume for children and young people (whereas NICE CG174 intravenous fluid therapy in adults in hospital gives a figure of 500 ml).
The Cochrane review found that liberal fluid therapy for resuscitation might increase mortality. However the largest of the 3 studies in the review (FEAST trial, n=3,141 children) was included when NG29 was originally developed, but the guideline committee disregarded it when developing recommendations because it was not directly applicable to the UK clinical setting. This evidence is therefore unlikely to affect recommendation 1.3.1 for a bolus of 20 ml/kg over less than 10 minutes.
A study of direct relevance to NG29 set in NHS hospitals comparing a 20 ml/kg bolus (in line with recommendation 1.3.1) with a smaller 10 ml/kg bolus found no difference in clinical outcome between groups. Along with the limitations of being a pilot study, the lack of benefit between groups is unlikely to affect the guideline, and also provides some evidence to address topic expert queries whether a 20 ml/kg resuscitation bolus is still appropriate. The authors of the study additionally noted that participants were not as unwell as expected, and that a larger trial is not feasible in its current design in the UK.
A separate trial of a bolus over less than 10 minutes (in line with recommendation 1.3.1) versus spreading the bolus over a longer period of 15–20 minutes, found that children receiving the longer bolus had less mechanical ventilation and improved oxygenation index. Although neither of these outcomes were noted as important by the original guideline, topic experts felt that they were likely to be important, though not critical. However, the lack of a difference in the study outcomes of mortality, length of stay or resolution of shock (outcomes deemed to be either critical or important by the original guideline), mean that results are unlikely to affect the guideline. Further evidence to confirm these findings, and which ideally includes outcomes deemed critical by the original guideline, is needed before impact on the guideline can be considered.
An additional trial found similar outcomes with World Health Organization Plan C for severe dehydration versus a slower rehydration approach. However, neither of these fluid administration strategies are recommended by the guideline and differ from the approach in recommendation 1.3.1, limiting the relevance of the study. This evidence is unlikely to affect the guideline.
Regarding the query to NICE about whether a maximum resuscitation bolus volume should be stated, this was discussed by the guideline committee during development of the guideline. At the time, the committee felt that including a maximum value would not be consistent with clinical practice or with other guidance (such as Advanced Paediatric Life Support – a course run by the Advanced Life Support Group). Recommendation 1.3.1 on fluid bolus is qualified by stating 'take into account pre-existing conditions as smaller fluid volumes may be needed', which allows healthcare professionals to reflect the needs of individual patients. Topic experts involved in the current surveillance review agreed that it was correct not to state a maximum bolus volume.
New evidence is unlikely to change guideline recommendations.