Study should have had more patients or longer time scale

EDITOR,—We welcome Richard J Davenport and colleagues' reminder of the importance of correcting outcome data for case mix.1 Given the size of their study, however, they should not be surprised or disappointed that they were unable to show a beneficial impact of their stroke unit on mortality. The 12 month mortality before the introduction of their stroke unit in Edinburgh was 39%.1 A meta-analysis of the effect of stroke …