Development of a minimum stroke dataset for electronic collection in routine stroke care

Improving stroke care is a national priority and adherence to national policy and guidelines is closely monitored by numerous organisations using a considerable number of overlapping indicators of stroke care processes. Demonstration that the processes of stroke care are linked to patient outcomes in empirical post-stroke populations is confounded by the complexities of patient case-mix.

Electronic, real-time, point of care data capture of care processes that are demonstrably linked to appropriately case-mix adjusted patient reported outcomes would increase confidence that the important aspects of patients’ care are
measured, monitored, and improved. This thesis aims to determine the best available case-mix adjuster, process measures and preferred patient reported outcome instruments and, through exploration of the relationships between these
factors, to develop a dataset for use within an electronic data system. The best available case-mix adjuster was identified through a systematic literature review as
the Six Simple Variable (SSV) model. Through group decision making workshops, and informed by a previous systematic review, the Subjective Index of physical and Social Outcome (SIPSO) was identified as the preferred postal outcome measure. I demonstrate how existing process markers for stroke lack variability, such that when recorded in their current format, their relative impact on patient outcome is
difficult to discern.

Process measures which feature as important predictors of patient outcome are shown to act as proxy measures of stroke severity. The SSV case-mix adjustment model is overshadowed by a simple univariable predictor (length of stay) which is also likely to be acting as a proxy for stroke severity. In this context, length of stay
may offer a pragmatic alternative to more complex case-mix adjustment models to examine the relationships between processes of care and outcome in populations of stroke survivors.