The Determinants of Length of Homeless Shelter Stays: Evidence-Based Regression Analyses

The Impact of Specialized Medical Equipment Needs on Shelter Accessibility for Homeless Patients

Housing insecurity is a well-recognized social determinant of health, with poverty and homelessness significantly impacting health outcomes. When faced with unstable housing, health often becomes a lower priority for patients. We present the case of a 52-year-old homeless, morbidly obese female who arrived at the emergency department with worsening dyspnea. She required Bilevel Positive Airway Pressure (BiPAP) support for acute-on-chronic hypercapnic respiratory failure. Her discharge planning was complicated by the need for a BiPAP machine, which limited her options for placement in a receiving facility. The patient ultimately left against medical advice (AMA) after a seven-day hospitalization but was readmitted hours later with recurrent symptoms, highlighting the ongoing challenges posed by her chronic conditions and social determinants of health. This case highlights how the requirement for specialized medical equipment can serve as an additional barrier to securing shelter for individuals experiencing homelessness.

Health and Economic Outcomes of Offering Buprenorphine in Homeless Shelters in Massachusetts

Importance

Overdose is the leading cause of death among people experiencing homelessness (PEH), but engagement in medication treatment is low in this population. Shelter-based buprenorphine may be a strategy for increasing initiation and retention on lifesaving medications.

Objective

To estimate clinical outcomes and conduct an economic analysis of statewide shelter-based opioid treatment in Massachusetts.

Design, Setting, and Participants

This economic evaluation study in Massachusetts used a cohort state-transition simulation model. Two cohorts were modeled starting in 2013, including (1) a closed cohort of a fixed population of PEH with history of high-risk opioid use over their lifetimes and (2) an open cohort in which membership could change over time, allowing assessment of population-level trends over a 10-year period. Data analysis occurred from January 2023 to April 2024.

Exposures

Model exposures included (1) no shelter-based buprenorphine (status quo) and (2) offering buprenorphine in shelters statewide.

Main Outcomes and Measures

Outcomes included overdose deaths, quality-adjusted life-years (QALYs) gained, and health care and modified societal perspective costs. Sensitivity analyses were conducted on key parameters.

Results

In the closed cohort analysis of 13 800 PEH (mean [SD] age, 40.4 [13.1] years; 8749 male [63.4%]), shelter-based buprenorphine was associated with an additional 65.4 person-weeks taking buprenorphine over an individual's lifetime compared with status quo. Shelter-based buprenorphine was cost saving when compared with the status quo, with a discounted lifetime cost savings from the health sector perspective of $1300 per person, and 0.2 additional discounted QALYs per person and 0.9 additional life-years per person. In the population-level simulation, 254 overdose deaths were averted over the 10-year period with the shelter-based buprenorphine strategy compared with the status quo (a 9.2% reduction of overdose deaths among PEH in Massachusetts). Overdose-related and other health care utilization undiscounted costs decreased by $3.0 million and $66.4 million, respectively. Shelter-based opioid treatment generated $44.7 million in additional medication and clinical costs, but saved $69.4 million in overdose and other health costs.

Conclusions and Relevance

In this economic evaluation of clinical and economic outcomes among PEH, shelter-based buprenorphine was associated with fewer overdose deaths and was cost saving. These findings suggest that broad rollout of shelter-based buprenorphine may be an important tool in addressing the overdose crisis.

Pathways of Individuals Experiencing Serious Illness While Homeless: An Exploratory 4-Point Typology from the RASCAL-UP Study

The shifting age demographics of those experiencing homelessness in the United States expose shortcomings and barriers within homelessness response services and safety-net healthcare to address serious illness. The purpose of this study is to describe the common trajectories of patients concurrently experiencing homelessness and serious illness. As a part of the Research, Action, and Supportive Care at Later-life for Unhoused People (RASCAL-UP) study, the study uses patient charts (n = 75) from the only specialty palliative care program in the U.S. specifically for people experiencing homelessness. Through a thematic mixed-method analysis, a four-point typology of care pathways taken by people experiencing homelessness while seriously ill is introduced: (1) aging and dying-in-place within the housing care system; (2) frequent transitions during serious illness; (3) healthcare institutions as housing; and (4) housing as palliation. Implications of this exploratory typology include targeted, site-specific interventions for supporting goal-concordant patient care and assisting researchers and policy makers in appreciating heterogeneity in experience and need among older and chronically ill people experiencing homelessness and housing precarity.

Understanding Homelessness among Migrants to Thunder Bay using Machine Learning (Preprint)

BACKGROUND

Over the past years, homelessness has become a substantial issue around the globe. The largest social services organization in Thunder Bay, Ontario, Canada, has observed that a majority of the people experiencing homelessness in the city were from outside of the city or province. Thus, to improve programming and resource allocation for people experiencing homelessness in the city, including shelter use, it was important to investigate the trends associated with homelessness and migration.

OBJECTIVE

This study aimed to address 3 research questions related to homelessness and migration in Thunder Bay: What factors predict whether a person who migrated to the city and is experiencing homelessness stays or leaves shelters? If an individual stays, how long are they likely to stay? What factors predict stay duration?

METHODS

We collected the required data from 2 sources: a survey conducted with people experiencing homelessness at 3 homeless shelters in Thunder Bay and the database of a homeless information management system. The records of 110 migrants were used for the analysis. Two feature selection techniques were used to address the first and third research questions, and 8 machine learning models were used to address the second research question. In addition, data augmentation was performed to improve the size of the data set and to resolve the class imbalance problem. The area under the receiver operating characteristic curve value and cross-validation accuracy were used to measure the models' performances while avoiding possible model overfitting.

RESULTS

Factors predicting an individual's stay duration included home or previous district, highest educational qualification, recent receipt of mental health support, migrating to visit family or friends, and finding employment upon arrival. For research question 2, among the classification models developed for predicting the stay duration of migrants, the random forest and gradient boosting tree models presented better results with area under the receiver operating characteristic curve values of 0.91 and 0.93, respectively. Finally, home district, band membership, status card, previous district, and recent support for drug and/or alcohol use were recognized as the factors predicting stay duration.

CONCLUSIONS

Applying machine learning enables researchers to make predictions related to migrants' homelessness and investigate how various factors become determinants of the predictions. We hope that the findings of this study will aid future policy making and resource allocation to better serve people experiencing homelessness. However, further improvements in the data set size and interpretation of the identified factors in decision-making are required.