Healthcare utilisation related to ambulatory-care-sensitive conditions amongst individuals with dementia : associations with the extent of cognitive impairment

Background: Dementia poses significant challenges for economies and health services worldwide; people with dementia are more likely to require an acute hospital admission and experience poorer in-hospital outcomes than age-matched controls without dementia. Reducing avoidable hospitalisations in people with dementia is a global healthcare priority, and within the UK specifically, there is an increasing focus on improving interventions and causes of hospitalisation which may be prevented by optimised primary care i.e., hospitalisations related to ambulatory care sensitive conditions (ACSC) in this population. Despite this, there is a lack of literature in the UK context describing the risk of ACSC-related hospitalisations and differences in the management or control of ACSCs in the primary care setting in those with dementia. This topic is of particular importance since it is expected that hospitalisations will put an increasing strain on health and social care infrastructures due to the projected increase in the number of cases of dementia over the next few decades. Further, in the absence of effective disease modifying therapies for the treatment of dementia, it is expected that a greater proportion of cases will comprise of those with more advanced dementia, yet there remains limited literature describing associations between the extent of cognitive impairment and the risk of hospitalisations or differences in the primary care management of ACSCs. This evidence could be used to help pre-empt future care needs and identify opportunities to tailor care in order to reduce the risk of ACSCrelated hospitalisations. This thesis seeks to explore the relationship between the extent of cognitive impairment amongst individuals with a confirmed dementia diagnosis and i.) hospitalisations (allcause and ACSC-related) and ii.) management of type 2 diabetes, as an example of an ACSC, within primary care. Methods: Individuals with a confirmed diagnosis of dementia were identified from the Clinical Record Interactive Search (CRIS), a de-identified electronic patient records data resource within the South London and Maudsley (SLaM) NHS Trust. Data relating to acute hospitalisations and encounters within primary care were obtained from existing linkages between CRIS and Hospital Episode Statistics (HES), and between CRIS and Lambeth DataNet (LDN) respectively. The general methods applied to these data sets are outlined in Chapter 4, and more specifically in the subsequent chapters, as follows. Chapter 5 uses age and sex standardised admissions ratios to determine whether individuals with dementia have a greater risk of hospitalisation relative to individuals residing in the same catchment area. Common causes of hospitalisation in those with dementia, and differences in the risk of hospitalisation by dementia severity are explored. Chapters 6 and 7 evaluate associations between ACSC-related admissions by stage of dementia. Hospital admissions in people with dementia within 6 months of a randomly selected date of cognitive function assessment are analysed; the subset of hospitalisations deemed to be ACSC-related are identified using published approaches. Chapter 6 uses data-driven approaches to explore whether stage of dementia can be inferred from hospitalisation patterns. Stage of dementia is first evaluated as a binary variable (mild and moderate-severe) using a logistic regression, then, in an attempt to improve model fit, as an ordinal variable (mild, moderate, severe) using a generalised ordinal logistic regression. To address a specific gap in the literature, Chapter 7 describes ACSC-related admissions amongst individuals with dementia and investigates differences by stage of dementia using a generalised ordinal logistic regression. Finally, Chapter 8 uses the CRIS-LDN linkage to evaluate the management of type 2 diabetesas an example of an ACSC within the primary care setting in individuals with dementia and a matched group of individuals without dementia. Comparisons are made for HbA1c trajectories (multilevel linear mixed-effects models), markers of diabetes-related management (logistic regression), mortality (Cox regression) and healthcare utilisation (multilevel mixed-effects binomial regression) based on presence of dementia, or on its severity at dementia diagnosis. Results: Hospitalisation amongst individuals with newly diagnosed dementia was common, with around half of patients having at least one admission in the 12-month period from first diagnosis. Compared to the catchment population, individuals with newly diagnosed dementia had a higher risk of hospitalisation. Among the common causes of hospitalisation in those with dementia, most were more common in those with more advanced dementia. Similarly, ACSC-related admissions were common in people with dementia, accounting for around 40% of those hospitalised particularly in those with greater cognitive impairment. However, there was no evidence that stage of dementia could be inferred from hospitalisation patterns. In the analysis of management of T2D in primary care, marked differences in the monitoring, management and control of diabetes were observed in people with dementia. HbA1c trajectories differed by dementia status; HbA1c increased over time for mild dementia and non-dementia, but the increase was greater in the mild dementia group. For those with moderate-severe dementia, HbA1c decreased over time. Despite individuals with dementia having increased healthcare utilisation around the time of dementia diagnosis, they were less likely to have received routine diabetes-related management. Patients with dementia had a higher prevalence of macrovascular complications and diabetes foot morbidity at dementia diagnosis and a higher mortality risk than those without dementia; these relationships were most marked in those with moderate-severe dementia. Conclusions: This thesis evaluated ACSCs from different perspectives, namely through primary care management of diabetes as an example of an ACSC and hospitalisations for ACSCs. By using linked clinical datasets, it was possible to address some of the evidence gaps highlighted in the literature. The findings indicate potential areas of suboptimal care amongst the population with dementia and prompts the need for further research. The observed differences in the management and control of T2D in those with dementia, relative to a matched comparator group, particularly those with more advanced dementia, suggests that care in this population could be tailored. Whether these differences explain the higher rates of hospitalisation and ACSC-related hospitalisation reported here and in the published literature could be the subject of future research. Although the population studied is not representative of the UK population, these results may be generalisable to the broader population, given that standardised care frameworks exist within the UK, and healthcare is universal and free at the point of delivery.