Holmes JF, Borgialli DA, Nadel FM, Quayle KS, Schambam N, Cooper A, Schunk JE, Miskin ML, Atabaki SM, Hoyle JD, Dayan PS, Kuppermann N, TBI Study Group for the Pediatric Emergency Care Applied Research Network*. Do Children With Blunt Head Trauma and Normal Cranial Computed Tomography Scan Results Require Hospitalization for Neurologic Observation? Ann Emerg Med 2011 Jun. PMID 21683474
This is a reasonable question to be asking and a very tricky one to ever conclusively prove
Methods
- really an analysis of pts from the PECARN data set
- non-trivial head injury (whatever that means) and could have poly-trauma
- scanning was entirely at the docs discretion
- GCS of 14 or 15 who got a CT that was entirely -ve for trauma
- all were followed (whether admitted or not) to see whether they had a repeat scan or neurosurgery
- follow-up for discharged patients was a phone call, followed by a mailed questionnaire followed by a trawl of databases
This is where the real problem is (as it is in most similar studies) – if some of the patients lost to follow up have bad outcomes then your conclusions would be radically differently. we’ll come back to this.
Outcomes
- if it was done – a +ve second scan for traumatic findings
- neurosurgical intervention
Important to note that these are not the same thing. A neurosurgical intervention is a big deal. A small contusion is certainly less so. Unfortunately a small contusion is probably not entirely unimportant prognostically. If minor bumps cause long-term post-concussion cognitive problems then a bump big enough to cause a contusion is likely to be relevant. Never mind the potential for subsequent seizures. This is, at present, speculative.
Results
- roughly 13000 pts included
- most (80%) were discharged after their normal scan
- if you had a GCS of 15 and a -ve scan and were sent home
- 1 in 2000 had a repeat scan that was +ve
- none had neurosurgery
- if you had a GCS of 15 and a -ve scan and were admitted
- 1 in 200 had a repeat scan that was +ve
- none had neurosurgery
- the numbers for GCS 14 were higher but still none had a neurosurgical intervention
- 20% lost to follow-up (by phone or mail)
As I said this is the key problem with all these kind of studies (including the recent SAH studies by Perry et al). If some (or in fact any) of these kids were dead or had neurosurgery then it would probably matter. These guys did a thing called a sensitivity analysis to try to compensate for this.
Basically this means they assume a similar rate of event in the lost to follow-up group to see if this would change the numbers. That’s all well and good but there’s no reason to believe that the lost to follow-up group are the same as the group who did receive follow-up.
Statistical adjustment is no substitute for complete follow-up
Interpretation
Having said all that I still think that the authors conclusions are valid – that if you have a -ve CT in a kid who has a GCS of 14 or 15 then you don’t need to admit them.
I rarely get CTs of kids heads. This seems to be a big practice difference between Ireland/UK and the US – at least from the very distant perspective of reading the literature.
I rarely use the clinical decision aids to decide when to CT. Or at least I use a couple of them but always tempered by gestalt. Which is the great unknown, unproven difficult to articulate, bias ridden thing that we do every day.
I think we’re still waiting on the definitive word of what to do with anti-coagulated patients with a -ve CT scan as these guys definitely do deteriorate on occasion…