Cite this article as:
Team DFTB. PREDICT: from knowledge generation to knowledge translation – Meredith Borland at DFTB19, Don't Forget the Bubbles, 2020. Available at: https://doi.org/10.31440/DFTB.21930
Meredith Borland is the current chair of PREDICT and is well-known for her work on the use of steroids in croup. The aim of PREDICT is not just to generate multi-centre research but also to share it and measure its impact. That is where DFTB comes in.
One of the key pieces of work that has informed a lot of Australian practice over the last few years is the CRIB study.
Oakley E, Babl FE, Acworth J, Borland M, Kreiser D, Neutze J, Theophilos T, Donath S, South M, Davidson A, Paediatric Research in Emergency Departments International Collaborative (PREDICT. A prospective randomised trial comparing nasogastric with intravenous hydration in children with bronchiolitis (protocol) The comparative rehydration in bronchiolitis study (CRIB). BMC pediatrics. 2010 Dec 1;10(1):37.
By challenging the long held belief that children with bronchiolitis should get an IV they have saved many of us countless hours.
This talk was recorded live at DFTB19 in London, England. With the theme of “The Journey” we wanted to consider the journeys our patients and their families go on, both metaphorical and literal. DFTB20 will be held in Brisbane, Australia.
This post is based on a talk I presented at the RCEM Spring Conference in April 2019 – Top 5 papers in PEM.
Kylie and Jason are enjoying their time at home with their first baby. The highs of being new parents is at its peak and true sleep deprivation is yet to set in. Jayden is two weeks old and is simply perfect. They spend hours staring at him each day marvelling at the perfect human they have created.
As we follow Jayden through his journey to adulthood, we’ll encounter some common paediatric problems. The 5.5 papers I have chosen were selected because: they cover common presentations; they use large patients groups; and they were conducted by well-respected and highly regarded research groups. But back to our story…
One night Jayden seems a bit more unsettled than normal. When they check his temperature it’s 38.4. They get in the car and bring Jayden to ED
Febrile neonates are a huge source of concern – we know that they can deteriorate quickly and we usually err on the side of caution by doing a full septic screen, IV antibiotics, and admission. Actually many of these babies don’t have a serious bacterial infection. Is there a way to tell which ones do?
When you see Jayden in your ED, you ask yourself is…should I do a full septic screen?
This paper aimed to derive and validate a highly accurate prediction rule to identify infant at low risk of SBI. The patients were febrile infants 60 days and younger (who had a rectal temp of >38 in the ED or a fever at home within the preceding 24 hours)
They excluded those who were critically ill, who had antibiotics in the preceding 48 hours, those born premature, and those with other medical conditions.
There were 1821 febrile infants included.
The authors considered clinical suspicion of SBI. They then look at various markers: blood culture; urine culture and urinalysis; CSF; FBC; and procalcitonin levels. The outcomes considered were serious bacterial infection – that is bacterial meningitis, bacteraemia, or urinary tract infection.
Overall, the rates of SBI in this group was 9%. The authors formulated a rule with a very high sensitivity (97.7%) for identifying those at low risk of serious bacterial infection. They were low risk if they fulfilled three criteria:
neutrophil count of less than 4/mm3
procalcitonin of less than 0.5ng/ml
61.3% of their patient group were low risk.
Interestingly their low risk rule does not include use of lumbar puncture – 67.4% of the low risk group had a lumbar puncture that would not have been necessary.
Key take away: There may be some febrile neonates that are low risk, and therefore we could avoid a lumbar puncture and full work up. In practical terms, this is unlikely to change our practice at the moment. Many of us cannot send a procalcitonin in the ED, and we might have to wait several hours to get a neutrophil count back. However this does bode well for the future in identifying which of these well febrile neonates are low risk.
Jayden does get a full septic screen. He has IV antibiotics for 48 hours and remains well. His blood cultures are negative so his antibiotics are stopped and he is discharged.
Jayden is growing well. At 7 months of age, he is looking great and developmentally normal. Dad, Jason, smokes, but reassures you that he never does so in the house. Jayden develops a cough and two days later starts breathing very quickly and noisily. They head to the emergency department.
Jayden has bronchiolitis. This is very common and your departments and wards have no doubt been filled with these children over the winter. We know that little works with these children. So you force yourself to hold back the ‘trial of salbutamol’ because it won’t make any difference.. But high flow does seem to be the one thing (along with oxygen) that might make a difference.
You ask yourself the question...should I start high flow?
This study looks at infants under 12 months old with a clinical diagnosis of bronchiolitis and a need for supplemental oxygen. 1472 were included (after exclusions). Patients were excluded if: they had an alternative diagnosis; they had cyanotic heart disease; or they were on home oxygen.
Patients were randomised to either high flow or low flow. The high flow group were given heated humidified high flow oxygen – 2L/kg/min via Optiflow. The oxygen was then weaned to achieve target saturations, and they were taken off high flow once they had been on air for four hours. The low flow group were given wall oxygen via nasal cannulae at 2L/min max.
The outcome was escalation of care. This meant who in the low flow group was escalated to high flow, and who in the high flow group was escalated to BiPAP or was intubated. Treatment failure was based on: an increase in heart rate; if the respiratory rate increased or didn’t drop; if they were needing oxygen in >2L/min of flow or >0.4 FiO2 to maintain their saturations; or if they achieve a high early warning score. Clinicians could also escalate care themselves (34% were escalated in this way).
Escalation of care occurred much more commonly in the low flow group – with 12% being escalated in the high flow group and 23% in the low flow group.
Interestingly there was no difference in the length of stay between the two groups.
Key take away: High flow does reduce the need for escalation. Escalation itself is significant – it requires increased nursing attention for low flow patients while they are transferred onto Optiflow. There may be less medical staffing on the wards if the child deteriorates on high flow overnight. Although they aren’t comparing like with like, escalation itself is an important clinical event. They also demonstrated that high flow does not increase the number of adverse events (for example there was no difference in the number of pneumothoraces between the groups). High flow is safe to use and we should consider starting it early in ED.
You start Jayden on high flow in ED and he stabilises. 12 hours later he is weaned off on the ward and is discharged the following day.
Jayden is now a healthy 3 year old boy. He loves Paw Patrol. He hates vegetables and won’t eat any food that is the colour green or yellow. Kylie and Jason are expecting their next child, and Jason has finally quit smoking. Unfortunately Jayden is prone to wheezy episodes and now has his very own inhaler which he hates using. The change in weather in London, from quite cold to…colder, seems to have triggered something and he’s now pretty wheezy and short of breath. They head into their favourite emergency department.
Jayden is now firmly in the realm of viral-induced wheeze. Yes, it’s all on a spectrum, but he’s now 3 years old with an inhaler. You asses him and think he should have a salbutamol burst.
As you are writing the salbutamol up, your SHO asks you – should I give him steroids?
This paper aimed to assess the efficacy of oral prednisolone in children presenting to an ED with viral wheeze.
The patients included were 2-6 years old. They were excluded if: saturations were less than 92% in air; they had a silent chest; they had sepsis; there was a previous PICU admission for wheeze; they had prematurity; or they had recently had steroids.
605 patients were included and they were randomised to receive either prednisolone or placebo. The prednisolone group received 1mg/kg prednisolone once a day for three days. The placebo group received a placebo medication (matched for volume and taste to prednisolone) once a day for three days.
Patients were assessed for their wheeze severity using a validated pulmonary score.
The outcome measures were length of stay (until clinically fit for discharge). They also considered re-attendance, readmission, salbutamol usage, and residual symptoms.
The results are tricky to interpret. Those who were discharged from ED within four hours did not benefit from prednisolone. However there may be some benefit in the mild to moderate wheeze group, and some in those who used salbutamol at home prior to presenting to ED. Interestingly this paper did not support our previously held belief that those children with atopy respond better to prednisolone.
Key take homes: Some pre-schoolers are steroid responsive, but identifying which ones is a challenge. As Damian Roland discusses here, it is likely that we are seeing lots of children presenting with the same symptoms (wheeze) but with different pathology behind it. Once we can identify the pathology we can start to target specific groups of patients with management that works.
You decided not to give Jayden prednisolone and after his salbutamol burst he stretches to 4 hours and is discharged home.
Jayden is 5 years old and in his excitement of building the new Hogwarts Lego castle he accidentally swallows a Lego head. Kylie and Jason aren’t sure whether to worry or not? So they take him into ED.
Children ingesting random objects is a common presentation to ED.
When you see Jayden in the department, his parents ask you…should I search through his poo?
Myself and 5 of my fearless, and brave, paediatric colleagues swallowed a Lego head each to see how quickly it passed. The paper was generously published in the Journal of Paediatrics and Child Health.
To ensure serious scientific rigour, we put together some scoring systems.
The Stool Hardness and Transit time (the SHAT score) took into account how hard our stools were, and whether that impacted (no pun intended) on the time to retrieve the Lego head.
And out main outcome was the Found And Retrieved Time (the FART score). This was the time to get our Lego heads back, and the average FART score was 1.71 days.
Unfortunately one of the six of us didn’t find his Lego head. After valiantly searching through his own faeces for two weeks, he gave up. And it may still be up there.
Key take home: Don’t search through poo, it’s gross.
Jayden heads home happily to finish building his Lego Castle.
Jayden is 6 years old. He thinks Paw Patrol is for losers. Fortunately he still loves Lego and Harry Potter. He also enjoys climbing. Unfortunately, two days ago he fell off the ladder coming down from his bunk bed. He seemed okay at the time, and Kylie and Jason had other plans that evening, so they decided to keep him at home. Now, two days later, he has a massive egg on his head and has been complaining of a headache. He also vomited yesterday. They bring him to ED.
We have fabulous head injury guidance for kids thanks to PECARN, CHALICE, and CATCH. But actually PECARN and CATCH specifically exclude injuries more than 24 hours old, and CHALICE doesn’t publish data on this group. So, for Jayden you need to put the NICE guideline away because it doesn’t apply. This is a common grey area.
This paper aimed to establish the prevalence of traumatic brain injuries in children presenting more than 24 hours after the head injury.
Traumatic brain injury (TBI) was defined as: intracranial haemorrhage; contusion; cerebral oedema; diffuse axonal injury; traumatic infarction; shearing injury; or a sigmoid sinus thrombosis.
The also looked a clinically significant traumatic brain injury (cTBI) – this included death, intubation for more than 24 hours, neurosurgery, or admission for 2 or more nights to hospital.
The patients were from the Australian Paediatric Head Injury Study Cohort which was 20,137 patients. 5% of these presented over 24 hours after the injury. 981 children were included in this study.
The authors considered the injury characteristics and demographics, trying to find an association between mechanism and delay in presentation. Those presenting were more likely to have: a non-frontal scalp haematoma; headache; vomiting; and assault with NAI concern. Those with loss of consciousness and amnesia were more likely to have presented within the first 24 hours.
The CT rates were much higher in the late presentation group – 20.6% being scanned in the delayed group and only 7.9% in the early group. This probably reflects the lack of evidence in this area, and therefore we feel safer doing more scans.
But the rates of TBI also varied. 3.8% in the delayed presentation group had a TBI, whereas only 1.2% in the early presentation group did.
The rates cTBI were the same between the groups at 0.8%
Key take homes: There is an increased risk of TBI when presenting more than 24 hours after a head injury injury. The authors found that risk is increased if the patient has a non-frontal scalp haematoma or a suspicion of a depressed skull fracture.
You decide to scan Jayden’s head, but it turns out to be normal and he is discharged home.
Jayden is 8. He’s been drinking a LOT of water over the last few weeks and seems to be weeing constantly. His clothes seem a bit big for him too. He looks so bad one day (and has vomiting and abdominal pain) that Jason finally reneges and takes him into ED.
Jayden has DKA. The debate about over-zealous fluid administrations and its relationship to the dreaded cerebral oedema is long-standing. Previous research suggested a link but only by association, not causality.
You ask yourself…how fast should I give IV fluids?
The study examines the causal effect between fluid resuscitation and cerebral oedema.
They included 1389 episodes of DKA. Exclusions were mainly due to too much management prior to contact with the study team, as well as children with a GCS<12. The median age was 11. It should be noted that the very young and the very sick are probably lost in this cohort.
Patients were randomised to received either fast or slow rehydration, and then were split again into received either 0.9% NaCl or 0.45% NaCl.
The fast rehydration group received 20ml/kg bolus and then replacement of 10% deficit, half over 12 hours and rest over next 24 hours. The slow rehydration group received a 10ml/kg bolus and then replacement of 5% deficit over 48 hours. Maintenance fluids and insulin were given in addition.
The outcomes looked at were deterioration of neurological status within first 24 hours of treatment. They also assessed short term memory during treatment, and IQ 2-6 months after the episode of DKA.
In short, they found no difference between the groups. There was a 0.9% rate of brain injury overall and it didn’t matter which type of fluids or how fast. Patients were more likely to get hyperchloraemic acidosis in the 0.9% NaCl group but this is of debatable clinical significance.
Key take homes: The evidence does not support our traditionally cautious approach to DKA. The speed of IV fluids does not seem to be the cause of brain injury in DKA.
You resuscitate Jayden and send him off to the ward. He is discharged a few days later with good support from the Endocrine team for management of his diabetes.
Jayden is now 16 years old and next time he comes to ED, he’ll be in the harsh world of Adult ED. We have navigated him through his common childhood presentations to ED and answered the key questions we ask ourselves every day in the Paeds ED.
Should I do a full septic screen on this hot baby?
Should I start high flow on this infant with bronchiolitis?
Should I give prednisolone to this 2 year old with wheeze?
Should I scan this child with a head injury?
How fast should I give fluids to my DKA?
And most importantly, do I ever need to sift through my child’s poo, or my own ever again?
Cite this article as:
Tessa Davis. Delayed presentation of head injuries – should we be worried?, Don't Forget the Bubbles, 2019. Available at: https://doi.org/10.31440/DFTB.17874
We have a clear algorithm for when to CT children who present with head injuries immediately after the injury. But, when children present more than 24 hours after an injury, we aren’t really sure what is best practice. This paper, by the PREDICT group, look at the rates of traumatic brain injury in this patient group.
Borland M, Dalziel SR, Phillips N, Lyttle M, Bressan S, Oakley E, Hearps SJC, Kochar A, Furyk J, Cheek J, Neutze J, Gilhotra Y, Dalton S, Babl F. Delayed Presentations to Emergency Departments of Children With Head Injury: A PREDICT Study, Annals of Emergency Medicine, DOI: https://doi.org/10.1016/j.annemergmed.2018.11.035
Why is this study needed?
We have a range of decision rules to help guide us for children presenting immediately after a head injury. PECARN and CATCH clinical decision rules specifically exclude children who present with a head injury more than 24 hours after the injury. CHALICE doesn’t specifically exclude this group, but there is no published data on this group of patients.
What we worry about is missing a traumatic brain injury, and in particular one that will need surgical intervention. If a child presents after the initial 24 hours, are they more likely to have a traumatic brain injury and should we therefore have a lower threshold to CT scan these patients?
This is a really common dilemma in Paediatric Emergency, and in my own experience, most people have a lower threshold for scanning children presenting late with head injury concerns because of the lack of guidance and evidence in this group.
The authors’ aim is to look at the prevalence of traumatic brain injury in this group, and to identify any factors in these patients that would make a traumatic brain injury more likely.
Who were the patients?
This was a secondary analysis of an existing cohort – the Australian Paediatric Head Injury Study cohort. This was children with a head injury who presented to one of ten paediatric EDs in Australia/New Zealand over a 3.5 year period.
For this secondary analysis, the cohort was split into those presenting within 24 hours, and those presenting later than 24 hours after the head injury. 5% of the cohort presented >24 hours after the injury.
Children were excluded if they had GCS<14, and were also excluded for representations of the same injury.
The original APHIRST cohort included 20,137 head injury presentations.
352 were excluded due to GSC<14 and 20 were excluded due to unknown time to presentation.
Of the 19,765 left, 981 children presented >24 hours after the injury.
Traumatic brain injury on CT (TBI) – intracranial haemorrhage or contusion, cerebral oedema, traumatic infarction, diffuse axonal injury, shearing injury, sigmoid sinus thrombosis, signs of brain herniation, midline shift, diastasis of the skull, pneumocephalus, and depressed skull fracture.
Clinically important traumatic brain injury (cTBI) – death, intubation >24 hours, neurosurgery, or a traumatic brain injury-related admission to hospital of two or more nights.
What were the authors looking at?
The paper examined any associations between a delay in presentation and the mechanism of injury.
It also looked at the injury characteristics and demographics for patients presenting within and after 24 hours of the injury.
Who presented more than 24 hours after a head injury?
Those presenting >24 hours after the injury were significantly more likely to have had a non-frontal scalp haematoma, headache, vomiting, and assault with non-accidental injury concerns.
Loss of consciousness and amnesia were more likely to present within 24 hours of the injury.
Were the late presentations more likely to have a head CT and a brain injury than those presenting within 24 hours?
203 of the 981 patients had a head CT in the late group. This is 20.6% compared to 7.9% in the early presentations.
37 of these children had a TBI on head CT. This is 3.8% compared to 1.2% in the early presentations group. The most common injuries were a depressed skull fracture, intracranial haemorrhage, or contusions.
Eight children had a cTBI (0.8% – which is the same as in the early group) and two required neurosurgical intervention (also not significantly higher than in the early group).
Who were the eight children with clinically important traumatic brain injuries?
The children ranged from six months to 15 years.
Five of them had a low-level fall (<1 m) – one of these required neurosurgical intervention
One was struck by a high speed object
One sustained a blunt injury with a bat during sport – required neurosurgical intervention
One fell out of bed more than two days earlier
Of note in the late group…
No children with amnesia had a traumatic brain injury on head CT
Suspicion of a depressed skull fracture and a non-frontal scalp haematoma were significantly associated with a cTBI
No children with loss of consciousness had a cTBI
What can we take from this?
There may be many reasons why our scanning rate in delayed head injury presentations is so much higher – including the lack of previously existing evidence, and our clinical concern that a TBI is more likely if the symptoms are persisting.
The authors conclude that presenting >24 hours after the injury (with a GSC>14), significantly increases the risk of a TBI. Suspicion of depressed skull fracture or a non-frontal scalp haematoma increase the risk of TBI and cTBI in this group.
Commentary from Damian Roland:
This is a useful sub-analysis of a very good research data set prepared by the PREDICT group which has good face validity and is likely to be externally reproducible in other developed nations.
The question I ask myself when reviewing head injury patients with a ‘delayed’ presentation is ‘why are you delayed?’. The sheer size of this data set is testament to the fact that lots of children present to Emergency Departments because of parental concern following a fall or blunt trauma. If a parent chooses not to present initially it’s usually because they thought the injury was not that significant (not a very high bar to reach usually!) and symptoms have evolved or perhaps the initial circumstances weren’t clear or un-witnessed. For the former case this ‘evolution’ of disease is (not surprisingly) significant. The ‘delayed’ group more likely to demonstrate relevant pathology because the symptoms that pathology were producing were becoming more apparent. For the latter “historical’ muddying is either sinister (note the relationship with non-accidental injury concerns) or perhaps critical information which may have resulted in earlier attendance has been missed.
It is important to note that while the post 24 hour group demonstrated increased risks for many features and outcomes, the absolute numbers are still low. Just because you present 24 hours down the line doesn’t mean do a CT. Just think that bit more carefully than if the child had presented straight after the injury. As this same group have also recently shown, our individual decision making capacity is probably just as good as any rule so we can still trust our own clinical judgement
This year I was asked to give not one, but two talks at the Victorian branch of the Australasian College of Emergency Medicine Annual Scientific Meeting. It takes place on the picturesque Mornington peninsula. With specific streams for doctors in training as well as fully fledged FACEMs I have tasked with providing a little bit of a paediatric update. This post will focus on the talk I am giving to the consultant stream.