Both interventions were adequately described, and the analysis included the usual care, where patients were given CT scans on the basis of their symptoms, using standard guidelines.
The effectiveness data were from several published studies. The authors did not give the details of these studies, but they were identified by a literature search and some inclusion criteria were reported. The full methods of the literature search were not reported making it unclear if it was a systematic review and if all the best available evidence was included. The key effectiveness estimates were provided in a table with references. The methods used to derive the measure of benefit were not stated. It is unclear how generalisable this benefit measure would be to other studies.
The authors reported the study perspective and appear to have included those costs relevant to this perspective. The sources of the cost estimates were relevant to the study and setting. The price year was stated and the key cost components and references were given in a table. The time horizon was chosen as the average period within which deterioration might occur. It might have been useful to extend this time horizon to seven days, as neurological deterioration in this period was rare, but still possible, and might have affected the cost estimates.
Analysis and results:
: The analytic approach appears to have been appropriate and the model structure was presented in a diagram. The effectiveness data were only used to derive the transition probabilities and therefore the only results presented were the net costs for the interventions. Appropriate one- and two-way sensitivity analyses were performed, but probabilistic sensitivity analysis could have more thoroughly explored the parameter uncertainty.
: There were a few limitations to the study and the results were not reported in detail. The authors’ conclusions should be interpreted with caution.