Prognosis of cognitive change following stroke: Identifying predictors, investigating the nature of between-variable relationships, and accounting for individual variability

Drozdowska, Bogna Anna (2021) Prognosis of cognitive change following stroke: Identifying predictors, investigating the nature of between-variable relationships, and accounting for individual variability. PhD thesis, University of Glasgow.

Full text available as:
[img]
Preview
PDF
Download (3MB) | Preview

Abstract

Background: Stroke survivors are at high risk of experiencing cognitive problems, which can severely compromise independence in daily activities, social participation, and quality of life. With limited evidence to support current interventions, finding ways to improve cognitive function is recognised as a priority for research relating to life after stroke. Prognosis research may contribute to this endeavour by informing the development and implementation of preventive and therapeutic strategies - from describing the natural history of post-stroke cognitive change, through identifying its relevant predictors and developing methods of estimating individual outcome probability, to supporting the application of stratified medicine. Prognosis research into post-stroke cognition is still developing, with little evidence regarding some of its more fundamental questions. These relate to the relevance of: i) potentially modifiable factors, ii) differential effects of risk factors, depending on paths of influence and co-occurrence, and iii) population heterogeneity in the trajectory of post-stroke cognitive change. Through focusing on these three topics, the purpose of this thesis is to improve our understanding of the cognitive change that occurs following stroke and its associations with individual characteristics.

Methods: Firstly, to gain a better insight into current advances in prognosis research in post-stroke cognition, I performed a systematic review of prognostic rules for predicting cognitive impairment and delirium following stroke. I considered these findings in specifying the aims and design of my subsequent, observational studies. I conducted two cross-sectional investigations in a sample of stroke survivors from the UK Biobank. Through a series of regression analyses, I assessed the associations of performance on four cognitive tasks with two groups of predictors of particular interest: 1) self-reported physical activity and sedentary behaviour, and 2) proxies of social engagement.
Using data from consecutive patients admitted to a hyper-acute stroke unit, I then investigated the influence of cardiovascular risk factors on acute post- 3 stroke cognitive performance. In a moderated mediation analysis, I tested the assumptions that the effects of these factors are partially mediated by stroke severity and prior dementia, and may be dependent on comorbidity. In my final, longitudinal study, based on the Assessing Post-Stroke Psychology Longitudinal Evaluation (APPLE) dataset, I conducted a latent class growth analysis to identify and describe differential trajectories of cognitive change, occurring over one year following stroke. Through subsequent regression analyses, I then explored factors that predicted trajectory class membership.

Findings: Through a systematic review of the literature, I identified seven prognostic rules predicting post-stroke cognitive impairment (including dementia) and four predicting post-stroke delirium. The most commonly incorporated predictors were: demographics, imaging findings, stroke type, and symptom severity. Among seven studies that assessed in the original sample how well a prognostic rule discriminated between participants who developed the outcome of interest and those who did not, performance was reported as being good to excellent. Only one rule had been validated in an independent dataset, showing fair discriminatory power. In the first of two UK Biobank studies, I found relatively consistent, although weak associations for two types of sedentary behaviour, where the daily duration of watching TV was associated with poorer cognitive performance, while duration of computer use was associated with better performance. Some effects remained significant after adjusting for demographic, health-related, and lifestyle factors. Physical activity, however, was not independently associated with performance on any of the considered tasks. In the second study, reported loneliness was the only proxy of social engagement to be associated with most cognitive tasks, consistently predicting poorer performance.
Findings from my analysis of data from a hyper-acute stroke unit setting supported the mediatory role of stroke severity and prior cognitive impairment in the effects of specific cardiovascular risk factors on acute cognition. Poorer cognitive performance was associated with atrial fibrillation through increased stroke severity, and with previous stroke through an increased risk of prevalent dementia. Conversely, through an association with reduced stroke severity, better performance seemed predicted by vascular disease (in the presence of hypertension and absence of diabetes) and by previous transient ischaemic attack. In the APPLE dataset, I identified four distinct trajectories of cognitive change: i) with high early cognitive function, improving over following weeks and thereafter declining; ii) with some early cognitive deficits, followed by improvement in function and then relative stability; iii) with comparatively poor initial function, which after a stage of steeper improvement continued to improve at a slower rate; and iv) with severe cognitive deficits, followed by improvement at a near-constant rate. Overall, participants representing the two trajectories with greatest initial cognitive deficits were characterised by older age, lower education, higher prevalence of pre-stroke cognitive impairment, and greater stroke severity.

Conclusions: In summary, my findings speak to the complex nature of cognitive change following stroke and its associations with individual characteristics. This is apparent on more than one level. What can be considered a single variable, such as sedentary behaviour, may be multifaceted. Entailing distinct properties, particular variable components are likely to have differential effects on post-stroke cognitive function. The effects of specific factors may moreover differ depending on the path of influence and the constellation of coexisting variables. Finally, post-stroke cognitive change is a heterogenous process, both on a between- and within-individual level. These observations suggest that it is important to consider how, in what form, under what conditions, and for whom, a possibly causal factor can affect post-stroke cognitive outcome. A lack of evidence-based assumptions regarding these aspects to inform the development of a statistical model may lead to misidentification of relevant associations. This is in turn likely to have implications at the stage of intervention development and implementation, limiting application. Recognising and at least partly accounting for the complexities I observed in my series of studies could contribute to bridging a gap between the potential and actual impact of prognosis research on improving cognitive function following stroke.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Additional Information: Supported by funding from the Stroke Association.
Subjects: R Medicine > R Medicine (General)
Colleges/Schools: College of Medical Veterinary and Life Sciences > Institute of Cardiovascular and Medical Sciences
Supervisor's Name: Quinn, Dr. Terry and Langhorne, Professor Peter
Date of Award: 2021
Depositing User: Theses Team
Unique ID: glathesis:2021-82413
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 30 Aug 2021 11:17
Last Modified: 30 Aug 2021 11:18
Thesis DOI: 10.5525/gla.thesis.82413
URI: http://theses.gla.ac.uk/id/eprint/82413
Related URLs:

Actions (login required)

View Item View Item