Examining the Connection: Traumatic Life Events, Substance Use, and Service Utilization Among Persons Admitted to Inpatient Psychiatry in Ontario

Abstract

Background Experiencing traumatic life events and the symptoms that follow have been associated with an increased risk for other mental health conditions. Among individuals who have experienced traumatic life events, comorbidities such as substance use disorder are particularly common. Individuals with co-morbid trauma and substance use may be less responsive to treatment, are prone to relapses, and increased hospitalizations. Gaps in care for individuals with co-occurring trauma and substance use reflect the growing need to understand associations between trauma experiences and substance use to identify opportunities for improving care and outcomes. Using data from persons who experienced trauma prior to admission to inpatient psychiatry, the purposes of this dissertation are to (1) identify the classifications of both trauma and substance use, (2) examine the service complexity received by persons with trauma, and (3) examine whether trauma classifications, and the presence of social relationships are associated with early leaves from inpatient care.

Methods A population based retrospective cohort was developed using interRAI Mental Health (RAI-MH) assessment data from all inpatient psychiatric assessments in Ontario, Canada between January 1, 2015, to December 31, 2019. The RAI-MH is a comprehensive assessment tool completed by clinical staff overseeing the care of the person. Completion of the assessment draws on multiple sources of information such as a review of the patient’s clinical records, interviews and observations, consultation with other clinical staff, family, and first responders (CIHI, 2023). The cohort included all persons in non-forensic and non-geriatric beds who were over the age of 18, and who had experienced a traumatic life event at some point prior to admission. Modelling and analyses were all conducted using SAS 9.4.

Study 1: Data were included for individuals with an index admission stay of 72 hours or longer during the observational window. Patients were excluded if they were admitted from another psychiatric hospital or if their first episode was not an admission assessment. Patients were included if they triggered the Traumatic Life events CAP of the RAI-MH (N=10,125). Latent class analysis was used to determine underlying subgroups of patients based on their patterns of traumatic life events and substance use behaviour. An 8-class solution was selected based on comparisons of Akaike information criteria, Bayesian information criteria, adjusted Bayesian criteria, and entropy values.

Study 2: Data from the Ontario Mental Health Reporting System were included for patients who triggered the Traumatic Life events CAP with no recent psychiatric assessments (i.e., no admissions within the last two years) (N=7,871). A service complexity variable was created based on length of stay (from date of admission to date of discharge, measured in days), the frequency of non-nursing formal care use, and nursing interventions in the prior 7 days. Descriptive statistics and bivariate associations between all demographic characteristics and level of service complexity were conducted. Logistic regression modelling was then used to assess the association between latent classes and the outcome (i.e., service complexity: low/moderate versus high service complexity). Odds ratios (unadjusted and adjusted), and 95% confidence intervals were reported for the initial and final models.

Study 3: All records for individuals who triggered the Traumatic Life events CAP with an index admission over 72 hours between January 1, 2015 and December 31, 2019 were included (N=11,043). Early leaves were defined based on discharge status. The variable was coded into three different levels including unplanned leaves (patients who were discharged due to an absence without an approved leave, and persons discharged against medical advice), early leave (patients with short length of stays), and no early leave. Chi-squared tests were used to understand associations between demographic and clinical characteristics, and early leaves from inpatient stays. Multinomial logistic regression modelling was then used to assess the association between latent classifications of trauma and substance use, Clinical Assessment Protocols, demographic and clinical characteristics, the multi-level outcome of early leaves (i.e., unplanned, or short length of stay), and those who did not discharge prematurely.

Results Study 1: Using latent class analysis, eight classifications of trauma and substance use were identified, ranging from low (i.e., Class 1: Interpersonal Issues, Without Substance use) to high (i.e., Class 8: Widespread Trauma, Alcohol & Cannabis Addiction) complexity patterns of traumatic life events and substance use indicators. Classes with similar profiles of trauma were differentiated by variations in substances use patterns. Furthermore, substance use patterns ranged from use of specific substances to widespread use and show variation in the presence of indicators of problematic use. Multinomial logistic regression models highlighted additional factors associated with class membership such as homelessness, where those who were homeless were estimated to be 1.71-3.02 more likely to be in Class 3: Safety & Relationship Issues, Alcohol & Cannabis use, and 2.09-4.02 times more likely to be in Class 6: Widespread Trauma & Substance Addiction.

Study 2: Service complexity ranged from 1 to 13, with the most common services being psychiatrist (84.3%), nurse practitioners or medical doctors (non-psychiatrists) (64.1%), and social workers (59.7%). High service complexity, defined as the upper quintile of formal care service use (scores of greater than or equal to 9), nursing interventions, and longer length of stay was observed in 18.1% of individuals with trauma. Compared to patients with few trauma experiences and no substance use, patients with more widespread trauma experiences and indicators of alcohol and cannabis addiction were 2.1 times (95% CI: 1.68-2.50) more likely to have high service complexity. Patients with safety and relationship traumas with alcohol and cannabis use, were less likely to have high service complexity compared to patients with interpersonal issues, without substance use (adj. OR: 0.70, 95% CI: 0.54-0.91). Characteristics such being female, having greater education, and being employed were associated with higher service complexity.

Study 3: Multinomial logistic regression revealed that individuals in latent classes with patterns of substance use (e.g., Class 6: Widespread Trauma & Substance Addiction) were more likely to have unplanned early leaves compared to those without substance use (adj. OR: 4.17, 95% CI: 2.72-6.39). Individuals with interpersonal conflict (i.e., conflict in relationships and widespread interpersonal conflict) had increased odds of having early leaves that were unplanned. Persons in Class 4: Immigration with Interpersonal Issues, Alcohol & Cannabis Addiction (adj. OR: 0.68, 95% CI:0.56-0.83), and Class 8: Widespread Trauma, Alcohol & Cannabis Addiction (adj. OR:0.73, 95% CIL 0.60-0.89) were less likely to have early leaves that were short length of stays compared to all other classes.

Discussion The findings highlight multi-dimensional experiences of both trauma and substance use. That is, experiences of trauma and patterns of substance use vary among patients with trauma admitted to inpatient psychiatry. Patterns of service use, and discharge status also varied. Differences identified suggest the need to consider the nuances of trauma to support patients, consider ongoing prevention of substance use, and address barriers in maintaining treatment.

Study 1: When considering traumatic life events across the latent classes, experiences of trauma were diverse among inpatients: from those with a few traumatic life experiences centered around health and loss (Class 1) to those with widespread experiences that include accidents, health challenges, grief and loss, and other social circumstances (Class 8). In Ontario, there are few specialized programs in place for supporting trauma, except for several tertiary hospitals. Advances in publicly funded services outlined in provincial strategic plans may hold promise, such as the introduction of structured psychotherapy programs and specific resources to support the military and first responders (Ministry of Health, 2022). Increasingly, dual treatment options for both trauma and substance use (e.g., Concurrent Treatment of PTSD and Substance Use Disorders (COPE)) should be further explored (Persson et al., 2017). Further research should explore patterns of trauma and substance use in community mental health settings, and supporting clinician confidence in discussing traumatic life events with patients.

Study 2: Nuances were observed when considering the relationship between latent classes of trauma and level of service complexity. Patients with indicators of substance addiction were more likely to have high service complexity. Findings highlight the importance of ensuring funding is allocated to public services for the continuation of care post-discharge. The RAI-MH can identify specific experiences and needs of persons with trauma that may be useful for informing further analyses on resource utilization and service planning. Given that economic costs data were not available, future research may consider the use of resource measurement and cost data to validate observed differences in service complexity.

Study 3: The results of study 3 point to differences between early leaves that are unplanned versus short length of stays. Both discharge statuses reflect an important period for providing treatment and recognizing substance use. Latent classes with the highest likelihood of unplanned early leaves generally included indicators of substance use. An eagerness to return to the community to utilize substances may reflect early unplanned discharges in this study. Inpatient admissions highlight an important timeframe to intervene in ongoing substance use. Unplanned early leaves may also reflect individuals with complex trauma that would better be supported in longer-term specialized treatment programs. Other factors such as interpersonal conflict, and eating disorders were associated with early discharge status. Future studies should assess the association between social relationships, formal supports, and early leaves.