Limitations
This is a single centre retrospective study which presents inherent limitations. We have used propensity matching which although a robust technique, has its limitations and it was notable that there were some differences between the two groups in baseline demographics. We have defined stroke on the basis of a combined clinical suspicion and evidence of changes on CT. It is possible that patients with an ischaemic infarct may be missed due to early imaging – however, if clinical suspicion remains it is our policy to reimage the patients after a few days therefore reducing the likelihood of significantly underestimating the incidence. In this series we relied on CT for diagnosing stroke, however it is known that sensitivity and specificity of MRI is greater. Due to small numbers we have not been able to use this study to characterise risk factors for the development of stroke or examine differences between early and late stroke. A significant proportion of patients were transferred to stroke rehabilitation centres and it has not been possible to obtain information of the duration of stay at these centres, to more completely report on inpatient duration.