Factors Associated with Difficult Discharge in Patients with Mental Disorders Following Prolonged Hospitalization: A Retrospective Study of 1254 Inpatients

Technical Summary: Factors Associated with Difficult Discharge in Patients with Mental Disorders Following Prolonged Hospitalization

Problem Statement
Patients with serious mental disorders (SMD) facing ultra-long-term hospitalization encounter significant barriers to deinstitutionalization and community reintegration. While prior research has identified clinical, social, and cognitive factors contributing to prolonged stays, existing literature treats "discharge difficulty" as a homogeneous outcome. This approach obscures the heterogeneous attributional structures of discharge barriers (e.g., distinguishing between clinical instability and lack of social placement). Furthermore, the specific role of China's guardian-signed discharge framework as a structural modifier of these barriers remains under-explored. There is a critical need to deconstruct discharge difficulties into specific attributional domains and understand how factors such as guardian status, physical comorbidities, and insurance types interact to influence discharge outcomes.

Methodology
This retrospective observational study analyzed data from 1,254 inpatients with mental disorders at a psychiatric specialty hospital in Jiangsu, China, discharged between January 2014 and December 2023.

• Study Population: Patients were categorized into a "successful discharge" group ($n=104$) and a "difficult discharge" group ($n=1,150$). For barrier attribution analysis, a subset of 1,147 patients with complete records was used.
• Variables:
  • Outcome: Discharge difficulty (binary).
  • Predictors: Sex, age, length of hospitalization (categorized as $\le$10, 10–20, and $\ge$20 years), number of physical comorbidities (0, 1, $\ge$2), guardian status (first-degree relatives, collateral/other relatives, social institutions, no guardian), and medical payment category (social assistance/zero out-of-pocket, employee insurance, resident insurance, self-pay).
  • Barrier Attribution: Discharge barriers were classified into four categories: illness-related factors, social support factors, cognitive/attitudinal factors, and other factors.
• Statistical Analysis:
  • Chi-squared or Fisher's exact tests were used for baseline comparisons.
  • Binary logistic regression identified independent risk factors for overall discharge difficulty.
  • Separate logistic regression models were constructed for each of the four barrier categories to identify specific predictors.
  • Interaction effects between guardian status and fee category were tested to examine structural modifications.

Key Results

1. Independent Risk Factors for Discharge Difficulty: Multivariable analysis identified three independent predictors of difficult discharge:

   • Male sex (OR = 2.284, $P < 0.001$).
   • Absence of a guardian (OR = 7.258, $P = 0.007$), which was the strongest predictor.
   • Two or more physical comorbidities (OR = 2.579, $P = 0.001$).
   • Age $\ge$ 60 years showed marginal significance ($P = 0.078$).

2. Attribution of Barriers: Among patients with discharge difficulties, barriers were distributed as follows:

   • Social support factors: 73.41% (largest proportion).
   • Illness-related factors: 25.98%.
   • Cognitive and attitudinal factors: 18.48%.
   • Other factors: 6.89%.
     (Note: Percentages sum to >100% due to multiple selections allowed.)

3. Heterogeneous Attribution Patterns:

   • Guardian Absence: Patients without guardians had the highest overall risk of discharge difficulty but the lowest risk of barriers attributed to illness-related factors (OR = 0.187 vs. first-degree relatives). Instead, their barriers were overwhelmingly driven by social support deficits or classified as "other" (attribution ambiguity).
   • Physical Comorbidities: Higher comorbidity burdens (especially $\ge$2 conditions) significantly increased the risk of illness-related barriers (OR = 7.971) but decreased the risk of social support barriers (OR = 0.199), suggesting a shift in attribution from social to medical causes as physical burden increases.
   • Cognitive/Attitudinal Factors: These were more prevalent in younger, female patients with first-degree relative guardians and zero out-of-pocket payment status.

4. Interaction Effects: A significant interaction was found between guardian status and fee category ($P = 0.033$).

   • Among patients without guardians, those with "family co-payment" (vs. zero out-of-pocket) showed a marginally significant reduction in discharge difficulty risk (OR = 0.058, $P = 0.055$), suggesting economic pressure might weakly facilitate discharge in this specific subgroup.
   • However, for patients with guardians, fee category did not significantly alter discharge difficulty, indicating that economic support policies are less effective when the primary barrier is the absence of a legal guardian.

Key Contributions

• Attribution Decomposition: The study moves beyond treating discharge difficulty as a single outcome by decomposing it into four distinct attributional domains (illness, social support, cognitive/attitudinal, other). This reveals that "difficult discharge" is not a monolithic phenomenon but comprises distinct mechanisms.
• Identification of Inverse Gradients: The research identifies a "social support needs-driven" pattern (characterized by guardian absence and low illness-driven barriers) versus a "medical needs-driven" pattern (characterized by high comorbidity and high illness-driven barriers). This challenges the assumption that prolonged hospitalization is always driven by clinical instability.
• Structural Insight: The study demonstrates that the absence of a guardian acts as a structural modifier, creating a "social support rupture" that economic coverage alone cannot resolve. It highlights that for patients without guardians, hospitalization often shifts from a treatment modality to a de facto placement modality.

Significance and Claims
The authors claim that this study provides the first systematic integration of attribution analysis with interaction modification in the context of ultra-long-term psychiatric hospitalization.

• Clinical Implication: The findings suggest that clinicians must distinguish between patients whose discharge is blocked by medical instability versus those blocked by social system failures. For the latter group (particularly those without guardians), medical stabilization is insufficient; the bottleneck is the lack of a legally designated community receiving entity.
• Policy Implication: The study argues that current fee exemption policies are inadequate for patients without guardians. Instead, a multidimensional discharge support system is required, prioritizing the reconstruction of social connections (e.g., community guardian agents) alongside economic support.
• Theoretical Contribution: The paper posits that ultra-long-term hospitalization erodes social support networks, creating a self-reinforcing cycle where longer stays lead to greater social disconnection, further increasing discharge difficulty. The identification of this "attribution blurring" (where social deficits are masked by medical comorbidities or recorded as "other") offers a new lens for understanding institutionalization.

The authors conclude that establishing a referral mechanism centered on social connection reconstruction and implementing differentiated interventions based on attribution classification are essential steps to breaking the cycle of prolonged hospitalization for this vulnerable population.