Explaining variation in long-term healthcare costs and utilisation of Stroke: Comparing findings from alternative statistical methods
Professor Julia Fox-Rushby
Professor of Heath Economics
School of Population Health and Environmental Sciences, King’s College London
Dr Marina Soley-Bori (King’s College London)
Every year there are 100,000 strokes in the U.K, with NHS and social care costs amounting to £1.7 billion approximately. More people are surviving stroke than ever before though, and stroke survival rates are expected to more than double in the next 20 years. As stroke becomes a long-term condition, care complexity increases as approximately half of stroke survivors report at least two other long-term conditions.
Little is known, however, about the long-term costs and healthcare use of stroke survivors, or which patient and provider factors explain variability across time. This is despite findings that rehabilitation and nursing costs are greatest (Rajsic et al 2019). To date only 12% of studies have considered costs longer than 12 months and the study covering the longest time period of 10 years (Gloede et al 2014) totalled costs from Australia but did not examine variation either over time or across patients.
This thesis will explore factors that influence long-term costs and post-stroke care use across the multimorbidity and stroke severity spectrum to inform future care improvement interventions. Different multivariate statistical methods, including machine learning, will be compared. Statistical modelling will be informed by a previous literature review.
The South London Stroke Register (SLSR) will be used to identify stroke survivors. SLSR has been recording and following up all first-ever strokes in people of all ages living in inner city South London since 1995. Information has been collected on over 5,900 patients who are followed up at three months and annually after stroke, for life. Examples of SLSR variables are demographic and clinical factors, social support network, and resource use before and after discharge. SLSR will be complemented with data linkages to the Hospital Health Statistics (HES) and Lambeth Data Net (LDN) to create a comprehensive patient resource use profile post-stroke.
This project provides a unique opportunity to further develop programming, statistical modelling, and health economics skills. A strong background in statistics and experience with a software package (R, STATA, or SAS) is required.
Gloede, T.D et al (2014) Long-term costs of stroke using 10-year longitudinal data from the North East Melbourne Stroke Incidence Study. Stroke J. Cereb. Circ. 45(11), 3389-3394
Rajsic S,. et al (2019) Economic burden of stroke: a systematic review on post-stroke care. Eur. J. Hlth Econ, 20:107-134
We suggest using SLSR, HES, and LDN. Data linkages have been successfully completed before within the department in the context of the stroke program.
Stroke, healthcare costs, data linkages, statistical modelling