Authors’ reply to Southall

In 2011, Maitland and colleagues published the FEAST study in which they concluded that febrile, unwell children in East Africa (excluding those with malnutrition, gastroenteritis, burns or surgical conditions) should not be administered intravenous fluid boluses for shock because it increases mortality.[1] Though the authors accept that this finding questioned many years of practice and experience,[2] they point out that the trial was a robust, large, multinational, randomised controlled study, with concealed treatment assignments, high adherence, and small loss to follow up.[1] Recently, the same group have criticised the World Health Organisation for not altering their new guidelines in the light of their trial findings.[3] However, as clinicians with an extensive combined experience of treating acutely ill children in Mbarara, Uganda, we have three major concerns about the FEAST trial that we feel need to be addressed.

First, we agree with Duke[4, 5] and Southall[6] that its definition of shock had such low specificity that many children who were not shocked may have been enrolled. In Mbarara, well over 90% of well nourished children without gastroenteritis who present unwell with fever have either acute malaria, pneumonia or meningitis. Few of these have hypovolaemic shock, but virtually all are tachycardic, and either have some alteration of consciousness or respiratory signs as part of their primary infective pathophysiology. Thus, they all fulfil the FEAST shock criteria, and would have been randomised to possibly receive a bolus of saline or plasma if they had presented to their study hospitals. Many would be made worse by an abrupt fluid overload, which might increase the mortality rate, as Maitland found, with a total of approximately 70 extra deaths in the fluid bolus groups.

In a further analysis designed to explain the FEAST study’s ‘unexpected findings’, Maitland compounds her mis-definition of hypovolaemic shock (where administering a fluid bolus would make physiological and clinical sense) by further defining a category of ‘severe shock’ which includes having biochemical test results[7] that we would expect to see in approximately half of severely ill, febrile, non-shocked children. A child without shock does not become one with severe shock by having a blood lactate concentration of >5 mmol/l, or a base deficit of >8 mmol/l.[7] Shock should be ascertained by physical signs that indicate impaired perfusion, such as the WHO definition of having all three of the following: a capillary refill time (CRT) of >3 seconds, cold hands and feet, and a weak fast pulse. Maitland does not provide this information in any of her publications.[1, 7] What is clear, however, is that no more than 26.1% of the children could possibly justify the diagnosis of shock at all (let alone severe shock) as this is the fraction who had a CRT of >2 seconds, a far less specific test than using a cut-off of 3 seconds. It could be that all or none of these children also had cold peripheries or a weak pulse.

Second, we are concerned that the FEAST study authors have taken insufficient account of having excluded up to 108 shocked children from the bolus fluid trial (Group A). As well as 29 hypotensive children (Group B), 79 were excluded as they had already been given fluid resuscitation by the receiving physician before they could be considered for enrolment, which could indicate that they needed urgent life-saving treatment. Whilst these children were obviously managed correctly, it was not correct for these authors to take so little account of their exclusion in their analysis and discussion. It could be that these were the only genuinely shocked children in the whole study.

Third, since some children died of cerebral oedema,[1] and since we have learnt that 89.6% of the children were subsequently administered intravenous water without salt (5% dextrose),[8] we are concerned that the published data do not allow us to determine how these two groups were linked.

We have asked Professor Maitland to further analyse the FEAST data by (a) diagnosing shock by using the combined physical signs that indicate that a child has an impaired circulation, (b) considering fully the impact of excluding 108 children who probably did have shock, and (c) identifying whether the children who died of cerebral oedema received hypotonic maintenance fluids. However, as with Professor Southall,[8] she has told us that she has already made sufficient data available.[9] We disagree, and therefore support the WHO in continuing to advise treating properly diagnosed hypovolaemic shock in normally nourished children with a fluid bolus.

Malcolm G Coulthard
Unni K Wariyar

Email to malcolm.coulthard@nuth.nhs.uk

References

1. Maitland K, Sarah Kiguli S, Opoka R, O, Engoru C, Olupot-Olupot P, Akech SO, et al. Mortality after fluid bolus in African children with severe infection. NEJM 2011;364:2483-1495.

2. Maitland K, Babiker AG, Kiguli S, Molyneux E. The FEAST trial of fluid bolus in African children with severe infection. Lancet 2012;379:613.

3. Kiguli S, Akech SO, Mtove G, Opoka R, O, Engoru C, Olupot-Olupot P, et al. WHO guidelines on fluid resuscitation in children: missing the FEAST data. BMJ 2013;347:f7003.

4. Duke T. What the African fluid-bolus trial means. Lancet 2011;378:1685-87.

5. Duke T, Mason Ewi. WHO guidelines on fluid resuscitation in children with shock. Lancet 2014;383:411-12.

6. Southall DP, Samuels MP. Treating the wrong children with fluids will cause harm: response to ‘mortality after fluid bolus in African children with severe infection’. Arch Dis Child 2011;96:905-06.

7. Maitland K, George EC, Evans JA, Kiguli S, Olupot-Olupot P, Akech SO, et al. Exploring mechanisms of excess mortality with early fluid resuscitation: insights from the FEAST trial. BMC Medicine 2013;11:68.

8. Southall DP. Concerns about intravenous fluids given to critically ill children. BMJ 2014;348:g1617.

9. Kiguli S, Akech SO, Mtove G, Opoka R, O, Engoru C, Olupot-Olupot P, et al. WHO guidelines on fluid resuscitation in children: authors' reply to Southall. BMJ 2014;348:g1619.