Statistics from Altmetric.com
Editor,—Harris et al1 apply Rothrock's criteria2 to a UK population of non-trauma patients. Their abstract concludes “Simple criteria can be usefully applied to patients presenting to an A&E department in this country to target patients most likely to have clinically significant findings on urgent cranial computed tomography”. We believe that the method and findings of the study do not justify the change in practice implied by this conclusion.
Our methodological concerns are threefold. Information gathered retrospectively from notes and request forms casts doubt over the accuracy and completeness of the symptoms and signs (particularly the symptom of nausea). The inclusion criterion is ill defined (patients who are referred for computed tomography). There is no explanation for the inclusion of nausea (it is not one of Rothrock's original criteria).
There are also theoretical objections. To be useful, a clinical filter must be applied to unselected patients and include criteria that have a high inter-observer reliability. There is no logic in applying a clinical filter after the decision to investigate has been made.
Furthermore, both studies acknowledge that they do not tackle the problem of subarachnoid haemorrhage in young patients presenting with isolated headache. Surely this is a major consideration in formulating any criteria for computed tomography (CT)?
We applied Harris' criteria to our prospective series of patients attending A&E with non-traumatic headache (248 patients). Seventy two CT scans would have been performed. The criteria would have missed three (1.2%) patients with an abnormal CT scan.
Judging from the differing rates of CT abnormality in the two studies (35% v 6%), CT rates in the UK are well below those in the USA. Given that we accept a detection rate of 1 of 80 for patients with skull fracture and GCS 15, perhaps we should be scanning more patients with non-traumatic headache not fewer.
The authors reply
We are pleased that our study has prompted discussion about the use of clinical guidelines for emergency head computed tomography (CT) in the non-trauma population. This is a developing area where little evidence exists.
It was interesting to hear that our modified criteria (any of: (1) GCS<14, (2) focal neurology and (3) headache with nausea or vomiting) would have missed three patients with abnormal CT findings. It is unclear whether these were young patients with subarachnoid haemorrhage who, when presenting with isolated headache, we have already acknowledged as a problem population.
Our modified criteria are simple, commonsense suggestions that reflect current practice regarding requesting CT from the accident and emergency department. We acknowledge the retrospective nature of our trial. Having now demonstrated that the criteria could be useful, prospective validation is necessary.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.