|dc.description.abstract||Background: Persons with severe mental illnesses are a small portion of the population that require a disproportionate amount of health and social services to meet their complex needs. This group is particularly vulnerable to experiencing marginalization and adverse social circumstances such as homelessness and incarceration. The literature recognizes that marginalization is a multidimensional social construct that influences mental health; however, its conceptualization and measurement remain unclear. In the mental health context, evidence suggest that both individual and contextual factors influence the use of services and mental health status of individuals. While most research on this area has focused on studying the individual level, the contextual level evidence is more limited.
Purpose: This dissertation aims to explore how the social context of where persons with mental illness live influences their mental health status and service use. Three distinct studies are employed to examine empirical patterns of area-level marginalization regarding mental health, the measurement and conceptualization of marginalization at the individual-level, and finally, the influence of context on inpatient mental health readmissions among marginalized persons.
Methods: This research linked data from a Canadian Census-derived index of marginalization, the Ontario Marginalization Index (ON-Marg), to clinical data from the Ontario Mental Health Reporting System (OMHRS); a dataset consisting of clinical and administrative data from every person admitted to a psychiatric hospital in Ontario. For the first study, bivariate and multivariate analyses on a sample admitted between January 1, 2006 and December 31, 2016 (N=150,600), examined the likelihood of residing in the most marginalized areas based on demographic, clinical, and service use characteristics using Statistical Analysis Software version 9.4 (SAS). For the second study, items that reflected the concept of marginalization were manually selected from the Resident Assessment Instrument-Mental Health (RAI-MH). Principal Component Analysis (PCA) and cluster analysis of these items was performed on a sample of patients admitted into psychiatric care between January 1, 2011 and December 31, 2016 (N=81,232) to identify dimensions being measured. Different weights and scoring methods were tested to assess convergent validity on multiple outcomes of marginalization. Receiver Operating Characteristic (ROC) curve analysis was utilized to determine optimal cut-offs for the index by modeling the likelihood of being homeless. For the third study, OMHRS data between January 1, 2006 and December 31, 2015, were used to identify persons with mental health conditions experiencing marginalization and who are at a high risk of homelessness (N=37,852). Standardized readmission rates at different points in time were calculated and mapped using the Forward Sortation Area geographic unit. Proximity to supportive housing services were measured using a 20-km radius buffer in ArcGIS software. Multilevel mixed-effects models were then built to test the influence of the different variables created, on readmissions to inpatient psychiatry in SAS.
Results: The first study found that the majority of persons admitted to inpatient psychiatry lived in the most marginalized areas of the Ontario. Those with little education, involved with the criminal system, on government assistance, diagnosed with schizophrenia, experiencing economic hardships, living alone, and those who lacked social support were the most likely to reside in areas with high marginalization. Patients in northern health regions were most likely to reside in areas with the most material deprivation while persons in resource intensive health regions like Toronto Central, resided in areas with the most residential instability. In the second study, 15 items were identified for the development of the Marginalization Index (MI). PCA and cluster analysis showed that these items measured 5 dimensions. ROC curve analysis for the most marginalized group, homeless individuals, identified an Area Under the Curve of 0.76 and an optimal cut-off of 5 on the MI. The frequency of homeless individuals, frequent mental health service users, persons with a history of violence and police intervention, and persons with addictions issues increased as scores on the MI increased, further confirming the convergent validity of the index. In the third study, readmission rates for those with high MI scores were 7.4% for short-term (within 30 days), 6.2% for the medium-term (31-90 days) and 13.1% for the long-term (91-365 days). While admissions to inpatient psychiatry occurred in 94% of Ontario’s FSAs, short term readmission only occurred in 20% of FSAs, medium-term in 11% of FSAs, and long-term in 41% of FSAs. Intraclass Correlation Coefficients showed that hospitals account for 3.8% of variance in readmissions within 30 days of discharge. Fixed effects β-parameter estimates of the models show that area level marginalization and proximity to supportive housing services increased the logs odds of readmissions.
Conclusion: This research identified factors that differentiated living in areas of low versus high marginalization among psychiatric inpatients. These findings are important for informing the equitable planning and distribution of evidence-based mental health services and supports to create social contexts that enable and support opportunities for improved mental health. Additionally, the Marginalization Index derived as part of this project proved to be a valid measure of marginalization and a strong predictor of risk of homelessness among psychiatric inpatients. The MI increases the visibility of the marginalized in inpatient psychiatry and provides a resource that can be used for supporting social and health policy decisions and evaluation. Finally, this research provided evidence that system structures influence readmissions in a variety of ways, while hospitals account for more variance among short-term readmission, area level marginalization accounts for more variance over longer-term readmissions. The findings contribute to the limited research that is currently available on the influence of contextual level factors on mental health service use by showing that contextual factors have various effects on readmissions at different points in time from discharge. These findings indicate that psychiatric readmissions may relate to social inequities at the area level and proximity to services.||en