Factors influencing dementia identification, management, and patient outcomes: A retrospective cohort study of hospital patients with dementia in Australia

<p dir="ltr"><b>Background:</b> Dementia is an important population health issue in Australia. In 2022, the estimated prevalence of Australians living with dementia was 487,500. This is expected to increase to approximately 1.1 million by 2058, which will have an increasing impact on Australia’s aged care and health systems. Hospital admissions for people with dementia are associated with adverse inpatient and post-hospital outcomes, high healthcare costs due to long lengths of stay, high mortality, and high likelihood of re-admission. Optimising health system outcomes and exploring if there is capacity to improve the care of hospitalised people with dementia requires longitudinal data. Historically, research in this area relied predominantly on short-term or cross-sectional studies of individual patient episodes. There remains a lack of longitudinal research following individuals over time, and a lack of knowledge around complex factors associated with the identification of dementia, which may result in adverse patient outcomes. Due to clinical coding issues within administrative data sets and the high prevalence of under-coding of dementia, there are no reliable estimates of how many people are living with dementia in the community, or when they were diagnosed. Utilising longitudinal data to assess the first diagnosis of dementia in hospital (index admission) provides a “starting point” to study a patients’ diagnosis and management journey. Medical records can also provide insight into discrepancies between identification of dementia and diagnosis in hospital. The overall aim of this thesis was to understand factors that influence dementia identification and patient outcomes, and to inform recommendations for improving identification and management.</p><p dir="ltr"><b>Method:</b> A multiple methods study design was employed. This included a novel retrospective cohort study of linked, longitudinal Emergency Department and Admitted Patient data from 7919 people with an index hospitalisation with dementia and a matched cohort between 1 July 2006 to 30 June 2015 (and a five-year look-back period from 1 July 2001) in a regional health district in Australia. This research also included a medical records audit, and ongoing consultation from a collaborative partnership. Statistical methods comprised descriptive analysis of patterns of hospital utilisation before and after the index hospitalisation, regression modelling to assess the impact of dementia coding (as a proxy for clinical management) on patient outcomes, and trend analysis to investigate the impact of a change in Australian coding rules for chronic conditions.</p><p dir="ltr"><b>Results:</b> Longitudinal data highlighted persistent problems across the care trajectory from pre-hospital diagnosis through to under-identification and under-management of dementia in subsequent admissions post-diagnosis. This includes lost opportunity for earlier diagnosis, consistent misidentification, and poor health outcomes (e.g., readmission). Findings also identified opportunities to improve identification and management across the care trajectory – including target interventions for specific conditions in the lead-up to a diagnosis (e.g., falls, urinary tract infections), across specific specialty areas (e.g., emergency medicine), and at the point of discharge or transfer (i.e., transitional care). Finally, evaluation of a change to coding rules in a local setting demonstrated that the quality of dementia data within hospital datasets can be improved, and clinician behaviour can be impacted.</p><p dir="ltr"><b>Conclusion:</b> This PhD assessed patterns of hospitalisation in the lead-up to and following an index admission coded with dementia. It then investigated the impact of dementia identification and management on patient outcomes, and finally, assessed the impact of a change to coding rules through a medical records audit. Overall, this research contributed new understandings into the trends and challenges associated with dementia identification and management in hospital settings, and highlighted opportunities across the dementia trajectory (pre- and post-diagnosis) to improve identification and clinical management. In the short-term, key clinical findings can be utilised by clinicians to inform improvement to dementia care and outcomes, including through targeting specific clinical presentations to increase opportunities for earlier identification. There is also scope to implement findings more broadly at a systems level, such as through updates to hospital policy, and within local intervention studies. Finally, this PhD highlights that use of highly contextualised data and longitudinal findings is invaluable and will provide a foundation to inform future interventions to improve the experience of hospitalised people with dementia.</p>