Children bump their heads. A lot.
(This article was first posted on the excellent DrGdH blog, make sure you pay it a visit – EMManchester – Ed)
Maybe they’re small, and just getting the hang of this walking business ….
…. or maybe they’re old enough to start doing daft stuff like this:
That top heavy head on top of an overexcited child seems very prone to getting bashed. Frequently, this means a trip to the ED.
Now most of the time, they’re fine. We know they’re fine after 5 seconds with the child.
After watching them tear around the waiting room for 3 hours and wolf down the chips and gravy from the hospital canteen, even the parents are starting to suspect the child is fine.
Unfortunately there is another thing kids do a lot of, which can make things a little more complicated:
(couldn’t find a picture of mine being sick)
Now, kids vomit. A lot. They vomit cause they’ve got cold. They vomit cause they’ve got a stomach bug. Sometimes they vomit just cause they’re upset and we look at them the wrong way.
Unfortunately they also vomit if they’ve got a brain injury:
The trick is to work out which ones are puking cause they’ve got a serious injury, and which one are just puking. Which children should we be sending to CT?
We can’t scan all of them. CT scans are not good for growing brains. There are small, but real risks with bombarding your child with radiation, (not just transforming them into a gamma ray fuelled super hero, which some may see as a plus)
Children also wriggle, and cry, and many of them will need sedation for a scan. This carries some risks of its own.
So how do we decide who gets scanned?
We are all familiar with one of the well know guidelines (for a brilliant summary of the literature and the guidance out there, try the excellent empem.org podcast). This EMJ article from earlier this year compares the three main decision rules.
Here in the UK, we have NICE guidance. I’m not going to go through it in detail, but with regards to vomiting they state:
“If 3 or more discrete episodes of vomiting….. request CT scan immediately”
Simple enough, clear unambigious (once you tease out the whole ‘discrete vomits’ thing) advice.
The only thing is….
We don’t do it. I don’t do this. My consultants don’t do this. I get the impression from the interweb that a lot of people don’t do this. We seem very happy to observe these children and see how they do, rather than scan them straight away as NICE would recommend.
This option is even written into local guidance. The relevant bit of protocol from Royal Manchester Childrens Hospital goes like this:
You can see, there is that all important phrase. For the children you are at ‘not low’ risk (this includes the vomiting ones), there is provision for a ‘period of observation’ instead of immediate CT scan.
How do we justify this, when the guidance from NICE is clear?
Lets take a closer look:
CHUNDERING IN CHALICE
The guidance from NICE is almost entirely based on the CHALICE study, which was done right here in my own stomping grounds; the North West of England. If we look specifically at the vomiting kids we see that:
Out of a population of 22772 children (<16yrs, all head injuries included)
857 vomited more than 3 times (3.8%)
56 of these children had a significant brain injury on CT. 801 did not.
Using vomiting as a screening test, and significant brain injury as our disease, we can plug these numbers into a 2×2 table.
It is the positive predictive value that we are most interested in. If a child vomits 3 or more times after their head injury, then their risk of a significant brain injury is 6.5%.
This is a fairly significant number. So why aren’t we scanning all these children then?
There are some caveats to this. The first is that the available data in the CHALICE study does not detail how many of these children were vomiting but had no other risk factors, it is the isolated vomiters that we are interested in.
CHALICE was intended to identify a low risk group we could safely not scan. It was not designed to inform management for those designated ‘high risk’
So on we go. Where else could we go looking? Surely there couldn’t be another massive cohort of head injured children we could examine?
PUKING UP PECARN
We all love PECARN. Their head injury rule was derived from a cohort of 42412 (!). I especially love the fact the low risk group has a lower risk of clinically significant head injury than CT induced malignancy. That’s the kind of reassuring fact you can use.
This massive cohort was comprised of children who had sustained a head injury. Only those with a GCS of 14-15 on presentation were included in the analysis (unlike CHALICE who included everybody).
But what does it have to say about vomiting children?
Interestingly, they do not consider vomiting a risk factor in small children (<2 yrs). It just does not come up in their rule (pathway A on the chart)
In bigger children (>2 yrs) it is included. They consider a history of any vomiting a risk factor, and isolated vomiting puts a child in their ‘intermediate’ risk category. They recommend observation or CT depending on the opinion of the doctor.
Sounds familar right? But what are the numbers? For children over 2 years of age who vomit more than twice, what is the risk of significant TBI?
Without going through it all again, the PPV is low at 2.3%. Low, but low enough to reassure us? Maybe not.
Once again, from the article it is not possible to work out how many of these kids had vomiting as their only symptom.
Fortunately, this time, someone has done it for us. This abstract (page S175) was published by the same team at the SAEM annual meeting 2008. They looked at the PECARN cohort and identified 1228 children with vomiting as their only symptom. Of this group, only one child required neurosurgical intervention, 0.1%. Reassuring, even if it comes from a conference abstract.
(running out of vomit slang now..)
Derived in 2010 by Osmond et al (and a lot of the same people as the Ottawa rules), the CATCH rule is another go and deriving a decision rule to help guide our management of head injured kids. A cohort of 3866 kids were looked at from 10 different centres. They included only symptomatic head injuries, so not the really trivial stuff.
So what did they think about vomiting? It got looked at, but didn’t make it into the decision rule:
Once again, it’s not possible from the initial paper to work out which of these kids had isolated vomiting.
But once again, they have asked themselves the same question. The very next abstract (S176) after the one mentioned above looks at the CATCH cohort.
In this group there were 3866 kids.
226 had vomiting >2 times as their only symptom
2 of them had positive findings on CT
16 (1%) needed neurosurgery.
Don’t know about you, but I’m reassured. Although vomiting is mentioned in two of the 3 major decision rules, it looks like that when its an isolated finding, we can be reassured. My practice is to observe these children, and I’m happy I can back that up if challenged.
Despite this, I’m still technically not following our national guidance. In view of what we have found…. time for an update?
Its all well and good if you can observe these children yourself, but in most hospitals these children we need to go to inpatient paediatrics. From my own experience, persuading the paediatricians that a immediate scan is not required can be tricky. I can quote PECARN and CATCH at them until I go blue in the face, but the fact remains that NICE says we should be scanning these kids.
There are more questions to answer here. More evidence needed! The rules above identify our low risk children, but don’t give us any guidance on what to do with those who are not low risk.
Which symptoms are more predictive of injury than others? For example, it has been shown that if the only high risk feature is the mechanism of injury, then the chance of having a serious injury is low. What about the other symptoms and signs?
If I’m observing them… what then? How long do I need to keep them until the risk of deterioration is acceptable? How many vomits? If not 3, then 5? 10? Or can I happily watch them puking away for days as long as no other symptoms develop?
Answers on a post-card please, preferably backed up by a big prospectively identified cohort….