What drives prolonged hospital stays in schizophrenia‐spectrum disorders? A Canadian tertiary care review

Abstract

Aim

Schizophrenia imposes significant economic and social burdens, often resulting in prolonged inpatient length of stay (LOS), particularly for treatment‐resistant cases. This study aimed to identify clinical, demographic, and social factors associated with prolonged LOS in a Canadian tertiary schizophrenia inpatient unit to highlight modifiable drivers and inform targeted interventions.

Methods

A retrospective chart review of 120 consecutive discharges from January 2021 to May 2023 was conducted. Data on demographics, clinical history, and discharge delays were analyzed using ANOVA and t‐tests to identify factors associated with prolonged LOS. Prolonged LOS was examined while differentiating between discharge delays and extended hospital stays.

Results

The mean LOS was 200.43 days (SD = 285.67, median = 100 days). Fifty percent of patients (n = 60) experienced discharge delays, defined as >10 days between readiness and actual discharge. Housing instability was the leading cause (56.67%, n = 34), including patients seeking housing (40%) or awaiting long‐term care (16.67%). Longer illness duration (p = 0.048) and treatment‐resistant schizophrenia (TRS, indicated by clozapine use; p = 0.011) were significantly associated with length of delayed discharge. Sex, relationship status, and family or community support were not significantly associated with length of delayed discharge. Illness complexity/severity was the primary reason for extended hospital stay in all cases (100%, n = 12).

Conclusion

This exploratory study highlights distinct clinical and system‐level contributors to prolonged hospitalization. Housing instability, chronic illness, and TRS are key contributors to delays in discharge. Early discharge planning, enhanced housing support, and specialized TRS care pathways are essential to reduce LOS and optimize inpatient care.

BACKGROUND

Schizophrenia, a chronic and severe mental disorder, imposes substantial economic and social burdens on patients, families, and healthcare systems worldwide. ¹ In 2020, schizophrenia accounted for approximately 1.7% of total health expenditure in Canada, with prolonged inpatient length of stay (LOS) being a major cost driver. ² Schizophrenia patients occupy 1 in 12 hospital beds in Canada ² with an average LOS of 96.6 days in Ontario psychiatric hospitals from 2005 to 2015, reflecting the illness's chronicity and the complexities of stabilization. ³

A number of factors can influence the LOS for inpatients with schizophrenia. Clinical factors, such as illness severity and treatment resistance, often necessitate extended hospitalizations. ⁴ , ⁵ Institutional factors, including a higher number of beds, may further contribute to delays in discharge and prolonged hospitalization. ⁶ Social determinants also play an important role, particularly housing instability ⁷ , ⁸ and difficulty accessing appropriate post‐discharge resources such as long‐term care or outpatient follow‐up. ⁹ Overall, prolonged LOS in schizophrenia appears to reflect a complex interplay of clinical, institutional, psychosocial, and biological factors. Consistent with this, studies from different settings have identified several associated characteristics, including unmarried status, male gender, treatment resistance and clozapine use, as well as abnormal thyroid‐stimulating hormone levels. ¹⁰ , ¹¹

Balancing adequate inpatient care with timely discharge to facilitate community reintegration remains a persistent challenge. ¹² Both prolonged and premature discharges carry significant risks. Extended stays can create a difficult anticipated transition into the community ¹³ and increase costs, ¹⁴ while premature discharges are associated with higher readmission rates, suicide, and criminal activity. ¹⁵ , ¹⁶ , ¹⁷ At the same time, some studies suggest that shorter hospital stays may improve quality of life, ¹⁸ , ¹⁹ highlighting the need for nuanced and individualized discharge planning.

In our Schizophrenia Inpatient Unit, the mean LOS increased from 116 days in 2018–2019 to 176 days in 2022–2023, potentially due to an increase in treatment‐resistant cases or systemic delays post‐COVID, signaling a growing challenge. This exploratory study aims to identify factors associated with prolonged LOS in our unit, focusing on clinical, demographic, and social variables, with particular attention to treatment‐resistant schizophrenia (TRS). By elucidating these factors, we seek to inform targeted interventions to reduce LOS, enhance patient outcomes, and optimize resource allocation for schizophrenia care.

To our knowledge, this study is among the first Canadian investigations to examine prolonged hospital stays and simultaneously differentiate between discharge delays and extended hospital stays in a schizophrenia‐spectrum inpatient population, while integrating both clinical (e.g., treatment resistance, illness duration) and system‐level factors (e.g., housing instability) to better characterize drivers of prolonged hospitalization in a real‐world tertiary care setting.

METHODS

Study design and participants

This retrospective study examined factors associated with LOS in the Schizophrenia Inpatient Unit at the Royal Ottawa Mental Health Centre, Canada. We analyzed 120 consecutive discharges from January 2021 to May 2023, excluding the initial nine months of the COVID‐19 pandemic due to lockdowns and restricted discharges. The study was approved by the hospital's Research Ethics Board (REB#2023013).

Data collection

Data were extracted from electronic medical records using an investigator‐generated questionnaire. Variables included demographics (e.g., age, sex, and relationship status), clinical factors (e.g., diagnosis, age of onset, medications, and psychiatric and medical comorbidities), psychosocial factors (e.g., housing status, family/community support), and reasons for discharge delays.

Diagnostic categories were based on documentation in the electronic medical record and reflected treating clinician diagnoses; therefore, individuals with schizoaffective disorder and psychosis not otherwise specified (NOS) were included as part of a schizophrenia‐spectrum inpatient cohort, consistent with real‐world inpatient diagnostic practices.

Prolonged LOS was examined while differentiating between discharge delays and extended hospital stays. Delayed discharge was operationally defined as a period exceeding 10 days between readiness for discharge and actual discharge, based on clinical documentation. Readiness for discharge was determined by consensus of the treating clinical team that the patient had achieved sufficient symptom stability and met discharge criteria. The 10‐day threshold was selected by the investigators to capture delays extending beyond routine administrative or transitional processes and to identify clinically meaningful barriers to discharge. Extended stay was defined as LOS > 100 days in the absence of a documented discharge delay. This threshold was investigator‐defined but informed by contemporary Ontario‐based administrative data. Specifically, Putman et al. ²⁰ in a population‐based study of schizophrenia hospitalizations in Ontario, used ≥100 days as a key descriptive cut point to characterize the upper tail of the LOS distribution and identify a subgroup of markedly prolonged admissions. This supports the use of a 100‐day threshold as a marker of prolonged LOS within this population rather than typical inpatient care trajectories.

Clozapine use was recorded as a proxy indicator of TRS, defined as inadequate response to at least two adequate trials of different antipsychotics, ²¹ recognizing that this approach may under‐identify TRS cases where clozapine was not prescribed or initiated despite clinical eligibility. Housing instability included individuals who were homeless, lacked permanent housing, or were awaiting housing placement.

Statistical analysis

Data from 120 discharges were analyzed using IBM SPSS version 28.0. Descriptive statistics were used to summarize demographic, clinical, and psychosocial characteristics. Continuous variables were reported as means and standard deviations. Categorical variables were summarized as frequencies and percentages.

For the subset of patients that had a delay in discharge, independent samples t‐tests were used to compare mean length of delay in discharge (in days) between binary categorical variables (e.g., sex, clozapine use), while one‐way analysis of variance (ANOVA) was used for variables with more than two categories (e.g., relationship status, income source, and housing‐related categories where applicable). A number of continuous variables were categorized for analysis including age at discharge (5 categories: 18–29, 30–39, 40–49, 50–59, and 60 years and more), duration of illness (3 categories: 0–10, 11–20, and 21 years and more), and number of previous hospitalizations (5 categories: 0–5, 6–10, 11–15, 16–20, and 21 hospitalizations and more). All analyses examined univariable associations between the independent variable and the continuous outcome variable, length of delay in discharge (in days). Given the small sample size and exploratory nature of this retrospective chart review, multivariate analysis and formal adjustment for multiple testing were not carried out.

A p‐value of <0.05 was considered statistically significant. For patients with extended stays (LOS > 100 days without a documented discharge delay), descriptive analyses were conducted due to limited sample size (n = 12), which precluded reliable comparative statistical testing.

RESULTS

Patient and clinical characteristics

The study included 120 discharges from 102 unique patients (11 patients had 2 discharges, 1 had 3, and 1 had 6). Patient demographics and clinical characteristics are summarized below.

The mean age at discharge was 42.08 years (SD = 13.11, range: 20–70). Most patients were male (63.33%, n = 76), single (82.5%, n = 99), and receiving social assistance (77.5%, n = 93). Before admission, 41.67% (n = 50) lived independently, 28.33% (n = 34) with family, 12.5% (n = 15) in group homes, 5.8% (n = 7) in assisted living (including long‐term care), and 11.67% (n = 14) were homeless. Family support was reported by 83.33% (n = 100), and 65.83% (n = 79) had community support. The mean age of onset was 21.79 years (SD = 6.39, range: 11–54), with a mean illness duration of 18.35 years (SD = 13.00, range: 0–54). Clinical characteristics are summarized in Table 1. Reported psychiatric comorbidities included depression, anxiety, bipolar, post‐traumatic stress disorder (PTSD), personality disorder, and substance use, with substance use being the most common (n = 59, 49.16%). Other medical comorbidities included dental (n = 2, 1.67%), gastrointestinal (n = 12, 10%), hematologic (n = 3, 2.5%), infectious (n = 2, 1.67%), renal (n = 1, 0.83%), and rheumatologic (n = 2, 1.67%) conditions.

Table 1.

Clinical factors.

Main diagnosis	Schizophrenia	N = 84 (70.00%)
	Schizoaffective disorder	N = 34 (28.33%)
	Psychosis NOS	N = 02 (01.67%)
Number of psychiatric comorbidities	No comorbid condition	N = 21 (17.50%)
	One comorbidity	N = 53 (44.17%)
	Two comorbidities	N = 24 (20.00%)
	>Three comorbidities	N = 22 (18.33%)
Medical comorbidities	No medical comorbidities	N = 24 (20.00%)
	Cardio‐metabolic	N = 36 (30.00%)
	Pulmonary	N = 13 (10.63%)
	Endocrine	N = 13 (10.63%)
	Neurologic	N = 12 (10.00%)
	Others	N = 22 (18.34%)
Number of previous hospitalizations	Mean (SD)	8.34 (9.12)
	Range	0–43
On clozapine and/or LAI at discharge	Clozapine	N = 64 (53.33%)
	LAI	N = 62 (51.67%)
Number of antipsychotics at discharge	No antipsychotic	N = 01 (00.83%)
	One	N = 34 (28.33%)
	Two	N = 48 (40.00%)
	Three or more	N = 37 (30.83%)
ALC status at the time of discharge	Yes	N = 21 (17.5%)

Abbreviations: ALC, alternate level of care; LAI, long‐acting injectable; NOS, not otherwise specified.

Length of stay and discharge outcomes

The mean LOS was 200.43 days (SD = 285.67, median = 100 days, range: 7–1561 days), with four patients exceeding 1000 days. Patients were categorized into two groups:

• Delayed discharge group: Defined as >10 days between readiness for discharge and actual discharge. Fifty percent (n = 60) experienced a delay, with a mean delay of 151.87 days (SD = 269.64). Housing instability was the primary cause (56.67%, n = 34), including seeking housing (40%, n = 24) or awaiting long‐term care (16.67%, n = 10). Other reasons included legal/community treatment orders (11.67%, n = 7), medical complications (10.00%, n = 6), lack of community support (6.67%, n = 4), patient ambivalence (3.33%, n = 2), and administrative constraints (11.67%, n = 7).

• Extended stay group: Defined as LOS > 100 days without discharge delay. Twelve patients had an extended stay, with a mean LOS of 231.59 days (SD = 215.76, range: 102–861 days). Schizophrenia (66.67%, n = 8) and schizoaffective disorder (33.33%, n = 4) were common diagnoses. Illness complexity/severity was the primary reason for the extended stay in all cases (100%, n = 12).

Discharge destinations

At the time of discharge, 29.2% of patients (n = 35) were going to be living independently, 30.0% (n = 36) with family, 13.3% (n = 16) in group homes, 13.3% (n = 16) in assisted living, while 3.33% (n = 4) were homeless/going to a shelter. In total, 4.2% (n = 5) were discharged to an acute care or psychiatric hospital, while 5% (n = 6) were discharged to other destinations such as a retirement home, jail, and so forth. This information was unavailable or missing for 1.7% (n = 2).

Factors associated with length of delayed discharge

ANOVA and t‐tests identified significant associations with length of delayed discharge in univariable analyses (Table 2). Extended stay was examined descriptively and is reported separately.

• Illness duration (p = 0.048): Longer illness duration correlated with longer discharge delays.

• Treatment resistance (indicated by clozapine use, p = 0.011): Patients with TRS, as marked by clozapine use, experienced longer discharge delays.

• Non‐significant factors included sex (p = 0.883), relationship status (p = 0.860), income source (p = 0.546), family support (p = 0.392), community support (p = 0.809), prior hospitalizations (p = 0.083), number of antipsychotics (p = 0.583), and long‐acting injectable (LAI) use (p = 0.091). Age at discharge approached significance (p = 0.050).

Table 2.

Factors associated with length of delay in discharge (in days).

	Test used	n	Significance	Effect sizea	95% confidence intervals
				Point estimate	Lower bound	Upper bound
Age at discharge	ANOVA	60	p‐value = 0.050	0.156	0.000	0.282
Sex	t‐test	60	p‐value = 0.883	−0.039	−0.559	0.481
Relationship status	ANOVA	59	p‐value = 0.860	0.023	0.000	0.068
Source of income	ANOVA	42	p‐value = 0.546	0.054	0.000	0.173
Family support	t‐test	60	p‐value = 0.392	0.328	−0.420	1.073
Community support	t‐test	60	p‐value = 0.809	0.066	−0.465	0.596
Duration of illness	ANOVA	50	p‐value = 0.048	0.122	0.000	0.281
Number of previous hospitalizations	ANOVA	50	p‐value = 0.083	0.164	0.000	0.299
Currently on clozapine	t‐test	60	p‐value = 0.011	0.543	0.001	1.080
Number of antipsychotics at time of discharge	ANOVA	60	p‐value = 0.583	0.050	0.000	0.130
LAI use	t‐test	60	p‐value = 0.091	0.496	−0.025	1.012

Note: Bold values statistically significant.

Abbreviation: LAI, long‐acting injectable.

^a Effect size point estimates are reported as Eta‐squared for ANOVA and Cohen's d for t‐tests.

DISCUSSION

Interpretation of results

Our results offer critical insights into the clinical and systemic factors that prolong LOS within a tertiary schizophrenia inpatient unit. The study cohort included schizophrenia‐spectrum disorders (i.e., schizophrenia, schizoaffective disorder, and psychosis NOS), and findings are therefore interpreted within the context of schizophrenia‐spectrum inpatient populations, reflecting real‐world diagnostic heterogeneity in tertiary care settings.

This study extends existing literature by distinguishing between discharge delays occurring after clinical readiness and extended LOS in the absence of discharge delay, allowing for a more nuanced understanding of clinical versus system‐level contributors to prolonged hospitalization. In addition, the inclusion of a schizophrenia‐spectrum inpatient cohort and the integration of both clinical and social determinants provide a more comprehensive real‐world perspective than prior studies focused on narrower diagnostic groups or single‐domain factors.

Based on clinical documentation, housing instability was identified as the most prevalent barrier to timely discharge, affecting 56.67% of delayed cases. This aligns with findings from previous studies, where housing issues were identified as a systemic barrier to discharge. ⁷ , ⁸ These results highlight the urgent need for early identification of housing challenges and partnerships with social services to secure stable placements, facilitating smoother transitions to community care.

TRS, marked by clozapine use, was significantly associated with a longer delay in discharge, consistent with prior studies. ⁴ , ⁵ TRS affects approximately 20%–30% of schizophrenia patients and is characterized by persistent symptoms despite adequate trials of at least two antipsychotics. ²¹ Longer illness duration was also significantly associated with a longer delay in discharge, likely reflecting the chronicity of schizophrenia. In our tertiary care setting, patients with persistent, complex symptoms and extended illness histories often require prolonged stabilization, have frequent relapses, and/or accumulated comorbidities that may result in housing instability, legal/community treatment orders, and other complications that can delay discharge.

Though they did not experience a delay in their discharge, a subset of our patients (10%) experienced extended stays of over 100 days, primarily due to illness complexity and severity. The complexity of managing TRS, including clozapine's requirements for slow dose titration, frequent monitoring for side effects (e.g., agranulocytosis, metabolic syndrome), and the need for stabilization of severe symptoms, often necessitates extended inpatient care. Our findings suggest that TRS, rather than clozapine use alone, is associated with prolonged hospitalizations, as these patients often present with greater illness severity and chronicity. ²² , ²³ The interplay between chronicity and TRS further complicates care, as long‐standing illness may exacerbate treatment resistance, leading to prolonged hospitalization. ²² This underscores the need for targeted interventions to optimize TRS management and expedite discharge.

In contrast, demographic factors such as sex, marital status, and social support (family or community) were not significantly associated with length of delay in discharge, diverging from studies that found marital status protective due to enhanced psychosocial support. ¹⁰ , ¹¹ This discrepancy may reflect Canada's robust social safety net, which could mitigate the impact of family support on LOS compared to resource‐limited settings. However, social support remains critical for post‐discharge recovery, warranting its integration into treatment plans.

Reduced community services post‐2020 may have also indirectly prolonged LOS and contributed to delays in discharge. Systemic barriers, such as limited long‐term care availability and administrative delays, further contributed to prolonged LOS, emphasizing the need for a multifaceted approach to address both patient‐level and system‐level challenges.

Strengths and limitations

This study's strengths include its comprehensive data collection, relatively large sample size (n = 120), and focus on actionable clinical outcomes, particularly for TRS patients. By examining a broad range of clinical, demographic, and social factors, it provides a holistic view of LOS determinants in a real‐world setting.

Limitations include the retrospective design, which introduces potential biases from incomplete medical record documentation. The absence of a control group limits causal inferences, and the single‐site setting may reduce generalizability. Clozapine use was used as a proxy for TRS, which may have led to misclassification and underestimation of TRS prevalence, as some eligible patients may not have received clozapine due to clinical or system‐related factors. Unmeasured factors, such as specific TRS‐related comorbidities (e.g., cognitive deficits) or socioeconomic variables, may also influence LOS. Due to the small sample size and exploratory nature of this study, findings should be interpreted as hypothesis‐generating. Future studies should employ prospective, multicenter designs to enhance robustness and explore TRS‐specific interventions. Studies with larger samples should employ multivariable analysis to examine confounding among interrelated variables such as age, illness duration, prior hospitalizations, and clozapine use.

Clinical implications

These findings offer actionable strategies for optimizing schizophrenia inpatient care, with a particular focus on TRS:

• Addressing housing instability: Early identification of housing needs and partnerships with community housing organizations could reduce discharge delays. Streamlined referral systems for long‐term care placements are critical to alleviating systemic bottlenecks.

• Optimizing TRS management: Patients with TRS, as indicated by clozapine use, require specialized care pathways to address their complex needs. These include multidisciplinary teams (psychiatrists, pharmacists, nurses) to manage clozapine titration, monitor side effects, and integrate psychosocial interventions (e.g., cognitive behavioral therapy for psychosis). Early initiation of clozapine in suitable candidates may prevent prolonged hospitalizations by achieving faster stabilization. ²⁴ , ²⁵ Additionally, training staff in TRS management and leveraging outpatient clozapine clinics could facilitate timely discharge.

• Managing chronic illness: Patients with longer illness durations, often overlapping with TRS, benefit from integrated care that combines pharmacotherapy, psychosocial support, and rehabilitation to address chronicity‐related challenges.

• Systemic reforms: Enhancing access to community‐based services, including TRS‐specific outpatient programs, and addressing administrative delays could improve patient flow and reduce LOS.

• Holistic care: While family and community support were not significant factors associated with length of delay in discharge, they remain essential for post‐discharge stability, particularly for TRS patients who face higher relapse risks. Interventions like family education and social skills training could enhance long‐term outcomes.

• Early discharge planning: Initiating discharge planning at admission, with a focus on TRS‐specific barriers (e.g., clozapine monitoring needs, housing) and other factors (e.g., legal issues), could facilitate timely transitions.

By targeting these factors, healthcare providers could reduce LOS, improve outcomes for patients, and optimize resource utilization. These insights contribute to broader policy discussions on enhancing schizophrenia care, particularly for TRS patients.

CONCLUSION

Housing instability, longer illness duration, and TRS were associated with delays in discharge, necessitating targeted interventions. To reduce LOS, healthcare providers should prioritize early discharge planning, strengthen partnerships with community housing organizations, and develop specialized care pathways for TRS patients. This study provides novel insights into TRS‐specific barriers. While family and community support were not significant factors, their role in post‐discharge stability remains critical. Policymakers should prioritize funding for TRS outpatient programs and housing initiatives. By addressing these factors, clinicians and policymakers can optimize inpatient care and support community reintegration. Future research should explore causal mechanisms and evaluate TRS‐specific interventions.

AUTHOR CONTRIBUTIONS

Naista Zhand developed the methodology of the study and contributed to manuscript writing. Carrie Robertson analyzed the data and contributed to manuscript writing. Ali Manghi contributed to data collection, analyzed and interpreted the data, and wrote the manuscript. Fatima Iftikhar contributed to data collection and formatting of the manuscript. Rebecca Sunderland contributed to data collection. Yiling Zhu contributed to data collection. All authors read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS APPROVAL STATEMENT

The study was approved by the Research Ethics Board of the Royal Ottawa Mental Health Centre and carried out in accordance with all of their relevant guidelines and regulations.

PATIENT CONSENT STATEMENT

Given the retrospective nature of the study and that this is a chart‐based review that did not involve any direct patient contact, informed consent was waived. Approval was received from the Royal's Research Ethics Board to review the patient charts.

CLINICAL TRIAL REGISTRATION

N/A.

DATA AVAILABILITY STATEMENT

The datasets generated during and/or analyzed during the current study are not publicly available due to data protection ethics processes and privacy guidelines of the Royal Ottawa Mental Health Centre. Please contact the corresponding author, Dr. Naista Zhand, for data access requests.