Khan, Kamran; (2024) Exploring factors associated with dementia diagnosis and post-diagnosis outcomes in the Electronic Health Care Record data. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text thesis_revised.pdf - Accepted Version Download (4MB) | Preview

Abstract

Importance: A better understanding of dementia progression will yield important information for future observational studies and clinical trials which will be essential for the approval of disease-modifying therapy (DMT). The knowledge gained will also inform guidelines for the treatment and management of dementia patients and improve resource allocation. Objectives: To comprehensively characterize longitudinal cohort studies and to identify what has been studied with regards to dementia progression after diagnosis in individuals aged  65, highlight limitations, any unexplored areas and identify opportunities to help inform new observational studies and clinical trials. To study dementia patients in the electronic health records data and provide an overview of their characteristics and select research-quality cohort. Conduct a time-to-event analysis for incident dementia and its risk factors based on the results of our preliminary analysis of our data. Describe and use multi-state modeling approach to study post-dementia outcomes using hospital admission and discharge data. Methods: For the review of literature, I searched OVID-MEDLINE for longitudinal studies with human participants from April 2008 till April 2019. Studies measuring outcomes of different domains important in dementia progression (clinical, health system utilization, biomarkers) were included. Preliminary analysis was performed using primary care general practice data and hospital episode statistics data and different summary statistics were performed and looked at the comorbidities in dementia cohort in these datasets. For the time-to-event analysis we used Cox proportional hazard model to study the factors for incident dementia in diabetic patients. We chose diabetes because it was the most prevalent comorbid conditions in dementia patients in the results of literature review and in the preliminary analysis of our datasets. For the post-dementia hospitalisation, institutionalisation, and mortality of dementia patients, I used a multi-state Cox model to study these outcomes and risk factors associated simultaneously in one model. Findings: I included 100 longitudinal studies comprising >2m individuals in the literature review. Mostly they had a small sample size (57% N<500 participants), short follow-up (66% <= 3 years), and dominance of AD (85% of the total sample in the selected studies was AD and only 9% was of vascular dementia (VaD)). Studies were mainly focused on measuring cognition (69% studies), while functioning and quality of life were less commonly measured (45% and 9% studies respectively). Studies were mainly measuring outcomes at 1 to 3 different time points and the follow-ups were shorter. The percentage of incident dementia in diabetic patients during the 10-years follow-up was 18.9 cases per 1000-person years. Increasing age, female gender and diabetes duration were associated with higher risk of dementia. From the results of the multi-state model, I found that home care availability influences mortality and institutionalisation from patients own home, and they were spending less time inside hospital and therefore, their rate of in-hospital mortality was low. Increasing age, frailty, and hospital admission due to injury were associated with higher rate of institutionalisation and death. Similarly, hospital stay  12 days was associated with hospital discharge to long-term care institutions and patients who were getting re-admitted within 30 days had a higher discharge rate and because of this had higher rate of rehospitalisation and subsequent institutionalisation and increased risk of death. Conclusion and relevance: The gaps identified by this review will help researchers design better observational studies that will inform future trials more comprehensively. It also provides an alternative way to study the intricate dynamics of hospitalisation, institutionalisation and mortality using multi-state model and provide a foundation for further research with appropriate data on formal and informal home care. The work serves as a cornerstone for further research.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item