Dear Editor, We read with interest the recent paper published in Acta Neurochirurgica by Li et al. [14]. In this paper, the authors have compared the outcome of craniotomy versus decompressive craniectomy in patients who have had an acute subdural hematoma evacuated. This is a well-written paper that addresses the important and topical issue of the role of decompressive craniectomy in the management of patients with severe traumatic brain injury and the authors should be congratulated. We would however, like to raise a few issues for further discussion. In the first instance, the authors concluded that “This is the first time that the CRASH head injury prediction model has been applied in a cohort of patients with traumatic ASDH”. This statement was incorrect. A review of the recent literature published in Acta Neurochirurgica would not support this statement [8]. Indeed, a wider review of the literature would reveal that this methodology has been used in a number of studies not only when considering evacuation of a mass lesion but also in the context of diffuse cerebral swelling [5, 6]. In our previous work, we stratified patients according to injury severity using a method that is similar to the standard method by which a prediction model is calibrated [2]. By comparing the predicted risk of an unfavorable outcome with the observed outcomewe can actually see differences between the predicted and observed risks of unfavorable outcomes across the full spectrum of severity of illness, rather than just the overall difference by the standardized morbidity ratio. Secondly, we can see from their results that the patients who had decompressive craniectomy were significantly sicker than the patients who had had a craniotomy.We would recommend the authors conduct a multivariable analysis to assess whether decompressive craniectomy was better than craniotomy, after adjusting for the CRASH-predicted risks of unfavorable outcome. Although the confidence interval of the two standardized morbidity ratios overlapped with one another, this did not confirm that these two estimates were statistically different. Because the standardized morbidity ratios between the groups appeared different, the authors should analyze the differences in their standardized morbidity ratio by a formal statistical testing before concluding that they were not different. Finally, this study was underpowered to detect a difference in outcomes between the two groups. At the very least, the risk of a type II error should be emphasized and any extrapolation of the negative results should be taken with extreme caution. Unfortunately, the authors did not highlight this important limitation in their discussion of the limitations of the study. Notwithstanding these issues, we would agree with the authors that one of the major confounding issues when interpreting the results of the many studies that have investigated the role of decompressive craniectomy is the failure to adjust for initial injury severity. Using the CRASH model to stratify patients according to injury severity provides an objective assessment of outcome (Fig. 1) and while we would agree with the CRASH collaborators that the use of this model should only be used to support and not replace clinical judgment, we feel that it can provide supportive S. Honeybul (*) : C. R. P. Lind Department of Neurosurgery, Sir Charles Gairdner Hospital and Royal Perth Hospital, Perth, Western Australia, Australia e-mail: stephen.honeybul@health.wa.gov.au