Statistics from Altmetric.com
The meta-analysis by Hofman et al (this issue pp416–22)1 opens again a running argument in both neurosurgery and neuroradiology. The argument is not unimportant: the outcome could affect the outlook for some people with an apparently mild head injury and have a significant impact on costs and workloads. The repercussions of these effects are on different people and it is unsurprising to find the merits of radiography usually being emphasised by surgeons and its demerits by radiologists. Two of Hofman's coauthors are radiologists and the third is an epidemiologist.
Their most valuable contribution is to refute the claim, oft repeated but not concordant with everyday experience, that demonstration of a skull fracture increases the risk of significant intracranial haemorrhage by a factor of 40. Their meta-analysis confirms an increased risk, but only fivefold, more in line with other studies.2 It has been suggested that radiographs can be used to obviate admission and observation in doubtful cases, but this may be inadvisable for several reasons. Not the least is the substitution of a test of questionable utility for proper evaluation. In 373 patients seen in Los Angeles with minor head injuries (excluded from the analysis of Hofman et al, presumably for methodological reasons), Feuerman et al 3 noted that, providing clinical assessment was adequate, nothing was gained from radiography. Indeed, they suggested that a patient with a Glasgow coma score of 15, shown to have a linear fracture of the skull, could be discharged to the care of a responsible companion. This may seem reasonable but, whereas in California the patient may well have had a detailed examination by at least a junior neurosurgeon, in the United Kingdom he may be seen by a junior non-specialist doctor, who has already been on duty for many hours.
The primacy of clinical assessment was borne out in the study of nine British hospitals by the Royal College of Radiologists.2Doctors, from housemen to consultants, asked to assess the likelihood of a skull fracture, were extremely good at correct negative predictions, although this is not surprising, as they formed the majority. However, one in 12 of the patients in whom a fracture was wrongly predicted had significant intracranial complications, suggesting that the predictions more accurately reflected an estimation of the severity of the injury rather than the presence of a fracture.
Guidelines issued by the Royal College of Radiologists4unequivocally reject both skull radiography and CT for patients thought to have a “low risk” of intracranial injury (although neither the low risk nor the degree of risk is defined). They also discard the triage value of the skull film, indicating that patients who cannot be placed in the care of a “responsible adult” may be admitted for observation rather than undergoing imaging. That appears to be borne out by the present meta-analysis. The recommendation for patients with a “medium risk” is indecisive, suggesting skull radiographyor CT. The presence of a skull fracture is said to transform the risk to “high”, thereby indicating CT, a recommendation still based on the presumed 40-fold increase in risk. It is to be hoped that the fourth edition of these guidelines, due next year, will omit reference to skull films entirely.
In the United States, even a decade ago, more than half the hospitals in a nationwide survey reported that they rarely used skull radiography for head injuries; CT was preferred, and where it was freely available, requests for skull films were few.5 The logical extrapolation is not, however, that all patients who might have had skull radiographs should undergo CT, but rather that CT is the examination of choice when the clinical condition reasonably raises concern about a treatable intracranial haemorrhage. Clearly, one would wish the threshold for suspicion to be such that there were few negative examinations, but that patients who needed scanning were not overlooked. The extra efforts required to organise emergency CT rather than skull films might effectively discourage poorly indicated requests.