This study offers an in-depth examination of psychiatric deinstitutionalization and highlights the role of substance use—alcohol, marijuana, and inhalable drugs—in influencing outcomes. We found that individuals reporting any substance use exhibited notably lower quality of life and higher healthcare costs than non-users. Moreover, substance use correlated with greater readmission rates, longer hospital stays, and elevated relapse frequencies, suggesting it acts as a potent modifier of clinical course. Deinstitutionalization, while beneficial for social reintegration, may inadvertently leave some patients vulnerable if community-based services are insufficient or fail to incorporate robust addiction interventions. Implementing these evidence-based interventions is feasible within Romania’s current staffing envelope. National workforce data indicate that one addiction counsellor typically covers three inpatient wards (≈150 beds). Embedding 15 min MI check-ins at discharge and a week 4 follow-up would require an additional 0.2 FTE per 50 community beds. The cost of upskilling ten nurses per ward via a 2-day CBT micro credential is ≈ EUR 150 per nurse—well within hospital continuing-education budgets.

These findings align with global trends showing that dual-diagnosis patients often require more intensive resources. Psychiatric disorders alone can be challenging to manage, yet co-occurring substance use imposes additional burdens on both the patient and the healthcare system. Several factors might contribute to this phenomenon: unrecognized withdrawal symptoms, inadequate adherence to psychiatric medications due to substance misuse, or complicated behavioral issues that impede stable community living. While some participants in community or residential settings reported higher employment rates and marginally better social function, the overall picture remained complex, particularly for those with polysubstance use.

Notably, these results underscored that substance use, more than demographic or some clinical variables, significantly predicted lower QoL scores. This highlights an opportunity for targeted interventions focusing on addiction treatment within psychiatric-care frameworks. Integrating motivational interviewing, cognitive-behavioral therapy for substance abuse, and close monitoring of relapse signs could be a valuable approach. In parallel, systematically implementing cost analyses can guide policymakers in resource allocation, emphasizing that front-loaded investment in specialized programs may prevent the recurrent expenses tied to repeated admissions and prolonged lengths of stay. Overall, addressing substance use is key to optimizing the potential benefits of deinstitutionalization efforts.

To visualize the assumed causal structure, we constructed a directed acyclic graph (Figure 5). It positions prior substance use as a potential common cause of both (1) care-setting allocation and (2) psychiatric severity, which in turn influences quality of life and health-care costs. The DAG guided variable selection for multivariable models and underscores that our results indicate association, not causation.

The studies by Samartzis et al. and Fulone et al. contribute important insights into the process and challenges of psychiatric deinstitutionalization (PDI). Samartzis et al. conducted a scoping review that identified both barriers and facilitators affecting PDI efforts, such as inadequate planning and insufficient community-based alternatives as major obstacles, while effective facilitation involved strong advocacy and comprehensive community service [22,23]. Similarly, Fulone et al. utilized the SUPPORT tools to develop an evidence brief, which was instrumental in guiding policy dialogue aimed at enhancing care for deinstitutionalized individuals. Their study synthesized findings from 15 systematic reviews, highlighting strategies like psychoeducation and intensive case management that improved global status and reduced relapses and hospitalization [23]. Both studies emphasize the complexity of deinstitutionalizing psychiatric care and underscore the necessity for robust planning, stakeholder engagement, and sustained resources to successfully transition patients into community settings.

The studies by Gao and Olfson [24] and Peterson et al. [25] present compelling data on the economic burdens associated with mental-health care and substance-use disorders (SUD) in the United States, highlighting significant out-of-pocket (OOP) costs and hospital expenditures. Gao and Olfson [24] found that a significant portion of psychiatric outpatients, especially those below the federal poverty level, experience high OOP cost burdens despite federal policies intended to increase insurance coverage for mental-health care. Specifically, they reported that 2.4% of psychiatric outpatients had a high OOP burden, with this figure dramatically increasing to 12.8% among those below the poverty level. Patients with higher OOP costs were more likely to be uninsured or diagnosed with substance use or bipolar disorders. In a similar manner, Peterson et al. [25] conducted an economic evaluation revealing that the annual medical costs associated with SUD in U.S. emergency departments and inpatient settings exceeded USD 13 billion in 2017. Their analysis, based on over 158 million hospital encounters, identified specific costs by substance type, with alcohol-related disorders alone accounting for USD 7.6 billion.

When discussing the cost of healthcare, it is also important to assess the readmission risk among psychiatric patients with SUD. Nordeck et al. [26] focused on the rehospitalization patterns among patients who received consultation from a hospital-based SUD consultation-liaison team. Their findings indicated a significantly higher rate of rehospitalization among patients with opioid and cocaine use disorders compared to those without these diagnoses, with opioid use disorder alone showing a notably high adjusted odds ratio (AOR = 2.4) for rehospitalization. Similarly, Böckmann et al. [27] examined patient-level predictors of psychiatric readmission in a Swiss hospital, identifying several key factors such as previous admissions, comorbid psychiatric conditions like psychosis or mania, and higher clinical severity scores at discharge, which correlated with an increased likelihood of readmission within 12 months.

Similarly, Reif et al. [28] found that among the 30,439 Medicaid beneficiaries analyzed, two-thirds did not receive any follow-up services within 14 days of discharge. Those who did receive medication-assisted treatment (MAT) or residential treatment showed a statistically significant reduction in the risk of readmission within 90 days, with behavioral health admissions occurring in 29% of individuals with an index admission. This contrasts with the higher readmission risks linked to outpatient services and intensive outpatient services in specific models. On the other hand, Owusu et al. [29] reviewed 75 studies and found that interventions such as improved access to mental-health services, especially residential treatment, and effective crisis interventions substantially reduced readmission risks. This review highlighted that certain patient groups, especially those with learning disabilities, developmental delays, and substance abuse issues, were particularly prone to readmission, suggesting targeted interventions could significantly alter outcomes. Both studies underscore the importance of structured, accessible post-discharge care as a critical strategy to reduce healthcare costs and improve patient outcomes by directly addressing the factors that frequently lead to readmission.

A primary limitation of this study is its cross-sectional design, which restricts our ability to draw causal inferences about the relationships among psychiatric diagnoses, substance use, and outcomes. Longitudinal follow-up would better illuminate how substance-use patterns evolve and whether changes in use directly influence readmission rates and cost trajectories. Additionally, we relied on self-reported measures for substance use, which may be susceptible to social desirability bias or underreporting. Although we made efforts to verify participants’ psychiatric diagnoses and medications from medical records, inconsistencies in record-keeping could introduce classification errors. Our sample size of 95 participants, while sufficient for an exploratory analysis, limits the generalizability of the findings to wider national or global populations. Although our DAG informed regression adjusted for measured common causes, residual confounding cannot be excluded in cross-sectional data. Finally, the study took place in a single region of Romania, where local healthcare infrastructure and funding may differ substantially from other settings. Future research should encompass multiple sites and integrate robust longitudinal data collection.