Storm Doris is raging here in the UK and departments are creaking up and down the land. It’s fair to say that 2017 has been off to a rocky start.
In amongst all the chaos I hope you’ve found time to read the recently published AHEAD paper published in the BMJ, and authored by some familiar names to those of us with a Yorkshire persuasion.
We were lucky enough to have Sue Mason join us for journal club which added a great amount of depth to discussion and also an insight into the way studies such as these are conducted, and some of the trials and tribulations of getting things published!!
Title of the Paper: AHEAD Study: an observational study of the management of anticoagulated patients who suffer head injury
Published: BMJ 2017
Talking of Doris, she’s your next patient in Majors. An 88 year old lady who appear frail, she has a history of ischaemic heart disease, hypertension, diabetes, COPD and AF (or A.fib for our cousins across the water). She has been on warfarin for a few years after the AF was discovered when being worked up for a TIA.
She fell in the kitchen at home, she bumped her head against the carpeted floor. She remembers falling, reaching for a tin of beans that she’d dropped. Daft really. Her son was in the next room. He could have got it for her. But she doesn’t want to appear dependent. She denies LOC, vomiting and certainly hasn’t had a seizure. She feels ‘just grand’ now and is keen to go home.
You dutifully examine her, and other than a graze on her forehead you don’t find anything at all of note. She’s been lucky.
Better to a CT though. She is on warfarin. And it’s in the guidelines, local and national (NICE 2014 Guidance here).
You talk to her about it and she’s accepting of your recommendation. you ring CT to organise it, the radiographer tells you its the ninth CT brain on someone over 80 he’s done already that morning, and they’ve all been ‘normal’, although he’s happy to scan Doris, as it’s recommended by NICE. Gets you though. What is the risk? Clinically she’s fine. And you probably wouldn’t “DO” anything if there was a bleed. She’s not really a neurosurgical candidate. She struggles to get out of the house. Hmmmm…..
Clinical question: In adults who are on warfarin and have sustained a head injury, without any worrying neurological findings on history or examination, what is the likelihood of finding a significant problem on CT scanning?
Population: Adults attending ED with a head injury whilst on warfarin
Outcome: Composite primary outcome measure of ‘adverse event rate’: the proportion who either died had a significant CT finding, required neurosurgery for their injury or who re-attended hospital within 10 weeks for a head injury related complication.
Secondary outcomes: What the effect of GCS, INR, or presence of neurological signs/symptoms was (Loss of consciousness, headache, amnesia or vomiting) was on the adverse event rate.
This was a retrospective cohort study across 33 EDs in England and Scotland before the latest iteration of the NICE guidance came out (for those of us old enough to remember practising then will recall that a CT wasn’t mandated for every head injury whilst on warfarin).
With it being a retrospective chart review it is possible patients were missed but the investigators did everything reasonable to ensure they got the majority by looking at everyone in ED who had a CT scan, everyone who had an INR check and everyone who’s attending complaint was of head injury. So they may have missed some, but I cant think of many. Perhaps those who went to the ward and were scanned there, with a booking complaint of ‘unwell’ for example? Hard to know the numbers there, but we agreed it was unlikely to be many given the strategy they employed.
Generalisability This is certainly generalisable to our patients, given that some of them were our patients. You could argue that the new CT scanning strategy recommended by NICE in 2014 means that the data are more difficult to interpret, and is therefore less generalisable to a ‘modern’ dept, given that more patients are probably being scanned now compared to then. However I think if you’re working in a UK ED then the patients studied here a probably as close as they can be to your patients.
They enrolled over 3500 patients to the study, meeting their power calculations sample size of 3000 in 19 months (remember the study was powered to a composite ‘adverse outcome’).
Around 2/3 of the patients had a head injury and were neuroloigcally ‘normal’ – i.e. GCS 15 an no amnesia, vomiting, headache or Los of consciousness at the time of the injury.
Around 60% had a CT head done as part of their care with around 5% of these having a significant injury related finding on the CT – so based on these numbers you’d need to scan 20 warfarinised head injury patients to detect one significant (radiologically) CT abnormality. I guess that poses the question about the 40% that didn’t have a head CT. What were their outcomes and what about them was it that meant they didn’t have a CT? too sick? too well? too co-morbid?
The most interesting group to me from this study was those with a GCS 15 (although only 91% of the cohort had a GCS documented which strikes me as worryingly low given they have head injuries!!). The authors tell us that if you have a GCS of 15 and no neurological symptoms of amnesia, vomiting, headache or LOC then your risk of adverse event is 2.7% (2.1-3.6%) and that the INR has no significant influence on relative risk 1.11 (0.95-1.18).
So what does that mean for me, for you, for your patients? Well I guess you have to decide where the line of acceptable risk lies – is 2.7% an acceptable risk to take not to do a CT (or between 2.1 and 3.6% given those 95% Confidence limits)? Given the adverse events included death? For me I’ll probably err on the side of recommending scan, but sharing that decision making with the patient and their family – we now have some amazingly useful numbers to talk to them about. Sue has resolved to go back to the original data and let us know what the breakdown within the adverse event group was – but it comes with a warning – the study wasn’t powered for the individual events so the significance of those individual results cant be relied upon as happening for anything other than chance.
I’m really interested to hear more on the evidence on the effect of INR in head injury as based on these numbers there seems to be no merit in performing one in those from this ‘low risk’ group. Let me knwo in the comments below if you routinely perform an INR test on all warfarinised patients.
If you want to drill down further into the subgroups of patients based on GCS and different neuro symptoms then make sure you read the whole paper, I’d also have a listen to Simon Laing discussing the paper on his podcast here.
But what about Doris and others like her, who probably aren’t neurosurgical candidates and have significant comorbidities? Whats the benefit of detecting something on CT? Well, for me, it’s the ability to tell her and her family she has a bleed that may worsen her quality of life and shorten her life, and that she may experience some neurological symptoms, even if insignificant to me as a clinician, i.e. not ‘testable in an ED – they may well be to her and her family, she may be more forgetful, or have personality change and impaired concentration that perpetuates.
There are a few questions for me coming out of this paper.
What about patients on NOACs? Can we extrapolate the findings to them? I’m not sure we know the answer to that yet, it may be that these patients find their way into AHEAD2 (if such a thing comes to exist!!).
Where would the level of risk be now in 2017 with a ‘scan al’ approach as dictated by NICE? perhaps lower? Maybe.
How reliable were the clinical notes that were retrospectively analysed. Only 91% had a GCS documented and significant number had missing data on neurological symptoms (especially amnesia) – which they tried to mitigate through multiple imputations (filling in the gaps). I dont think this taints the results but it is something that highlights a flaw in retrospective cohort studies.
In reality I think this is an incredibly useful paper and the data could and should be used as part of your shared decision making process with patients after they’ve suffered a head injury, but with an adverse event rate of around 2.7% in those who have no symptoms I don’t think I’ll push to avoid scanning yet.
You can download the one minute wonder
Thanks for reading
Keep fighting the good fight