The Impact of a Severe Psychiatric Illness on a Cancer Diagnosis, Treatment, and Survival

Abstract

Background: Individuals with a severe psychiatric illness (SPI) are vulnerable to cancer care disparities. The contributions of cancer-related mortality to lower life-expectancy in individuals with an SPI have been understudied. Methods: This thesis investigated the association between an SPI history and colorectal cancer (CRC) outcomes in a cohort of patients diagnosed in Ontario between 01/04/2007-31/12/2012 using provincial healthcare data. SPI history was determined using hospitalization, emergency department, and psychiatrist visit data and categorized as ‘no history of mental illness, ‘outpatient SPI history’, and ‘inpatient SPI history’. The thesis was organized around a causal framework, highlighting the complexity of the relationship between an SPI and worse cancer outcomes. The thesis investigated the SPI effect on an unknown cancer stage, non-receipt of surgical resection, non-receipt of adjuvant treatment, and on overall and CRC-specific survival. Multiple log-binomial, logistic, and modified Poisson regression were used to study dichotomous outcomes. Multiple Cox-Proportional hazards, Aalen additive hazards and Fine & Gray sub-distribution hazards regression were used to study time-to-event outcomes. Results: CRC patients with an inpatient SPI history had 1.45 (95% CI: 1.31-1.84) times the risk of an unknown stage cancer at diagnosis, 2.15 times the risk of not receiving surgical resection (95% CI: 1.07-4.33), and 2.07 times the risk of not receiving adjuvant treatment (95% CI: 1.72-2.50), compared to those with no history of mental illness. The outpatient SPI effect was smaller in magnitude and often non-significant. The hazard ratio (HR) of death from any cause was 1.91 times higher for individuals with an inpatient SPI (95% CI: 1.63-2.25) and 1.40 times higher for individuals with an outpatient SPI history (95% CI: 1.22-1.59); the HR of death from CRC was slightly lower and the HR of non-cancer death was higher, than the overall estimates. Some significant effect modification by age, sex and stage at diagnosis was noted; however, the study was not adequately powered to detect interaction. Conclusions: These three studies tell a consistent, cohesive story of cancer outcome inequalities for patients with an SPI across the cancer care trajectory. Future work to understand why these inequalities exist and how to best intervene are necessary.