High-cost users of health care in Saskatchewan: A population health perspective

Background: A small proportion of the population consumes the majority of health-care resources. High-cost user research is complicated by heterogeneous populations, a natural tendency for a regression to the mean and episodic versus persistent spending patterns. Using two separate study cohorts, 1) a clinical sub-group known to be high-cost – mental health and addiction clients, 2) a general provincial population grouped into separate clinically meaningful sub-groups, this thesis seeks to understand the high-cost health care user population in the province of Saskatchewan, Canada. \n Methods: The first two quantitative studies focus on a specific disease sub-population known to be high-cost – mental health and addiction clients. First, a retrospective cohort study explores risk factors associated with high-cost use with a particular focus on individuals who are persistently high-cost year after year. The second study aims to predict individuals at risk of both episodic and persistent high-cost use in the future. Predictive models using Classification and Regression Tree (CART) methods were constructed. The last study takes an overall population segmentation approach to understanding high-cost use. Using the recently developed Canadian Institute for Health Information Population Grouping methodology, individuals were assigned to one of 16 mutually exclusive and clinically distinct health profile groups. Following univariate and bivariate analyses, logistic regression models were constructed for each of the costliest health profile groups to explore risk factors associated with high-cost health care use. \n Results: Study 1: Persistent high-cost mental health and addiction clients comprised a small proportion of the study cohort (n = 6,455; 5%) but accounted for 35% of total costs. Exploratory models of mental health and addiction high-cost patients found increased risk of persistent high-cost use with hospitalization(s), unstable housing, severity of the diagnosis (schizophrenia versus others), and multiple comorbidities. Good connection to a primary care provider was protective of high-cost use, particularly when individuals had multiple mental health conditions. \n Study 2: The most important variables for predicting one-year and persistent high-cost use were delineated visually in CART diagrams. My models had reasonable calibration and validation and take advantage of health care utilization and demographic information readily available in Canadian provincial administrative health care databases. The visual nature of the CART method assists to make complex data readily understandable to policy and decision-makers. \n Study 3: A provincial cohort (n = 1,175,147) was identified for study. High-cost users consumed 41% of total health care resources. The costliest health profile groups were ‘long-term care’, ‘palliative’, ‘major acute’, ‘major chronic’, ‘major cancer’, ‘major newborn’, ‘major mental health’ and ‘moderate chronic’. Both ‘major acute’ and ‘major cancer’ health profile groups were largely explained by measures of health care utilization and multi-morbidity. In the remaining costliest health profile groups modelled, ‘major chronic’, ‘moderate chronic’, ‘major newborn’ and ‘other mental health’, a measure of socio-economic status, low neighbourhood income, was statistically significantly associated with high-cost use. \n In each exploratory study, when controlling for a variety of factors including demographics, health care utilization and health status, baseline measures of socio-economic status – unstable housing and low neighbourhood income – were found to be statistically significantly associated with high-cost use. \n Conclusion: Interventions aimed at improving population health and reducing the health care costs associated with a ‘high-cost user’ population should consider segmenting the population into relevant homogenous sub-groups, defining high-cost use within sub-groups, and, exploring risk factors associated with high-cost use within each subgroup. Primary health care system transformation efforts could include a ‘high-cost user’ component. Given that having a good connection to a primary care provider was found to be protective of high-cost use, it is suggested that primary care providers, and, high-cost patients themselves design interventions aimed to reduce costs and improve population health. Lastly, socio-economic status must be considered in exploratory and predictive modelling of high-cost health care use; policy efforts to address socio-economic status in a high-cost population may result in health care system cost savings, but more importantly, improved population health. \n

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.