Long-term cost-effectiveness of Housing First for homeless people with mental illness
Bibliographic record
Abstract
Background.Homelessness has been expanding in Canada and internationally.It significantly increases mortality and thus is a public health concern.Housing First (HF), an approach that involves providing immediate access to permanent housing and individualized support services, is a key component of strategies to end homelessness.A two-year costeffectiveness study found that HF resulted in significant cost-offsets but did not fully pay for itself.Objective.The objective of this project was to develop a simulation model to project HFs effects on costs and housing stability, from a societal perspective, compared to treatment-asusual, over a ten-year horizon. Approach.A novel Excel-based platform, Discretely Integrated Condition Event (DICE), was used to build a Markov simulation model.Cost and outcome data were ascertained from the Montreal At Home/Chez Soi randomized controlled trial.Individuals were divided into eight subgroups based on need level, homelessness history, and intervention received.Nine possible housing states including street, shelters, psychiatric hospitalization, and prison, were defined.Daily transition probabilities between states were calculated by subgroup.Costs for healthcare, social, and justice services, and income were calculated for each housing state using generalized least squares regression.Days in stable housing was used as the outcome measure.One-way sensitivity analyses were conducted on the discount rate, the rate of "autonomization" for services provided by HF (i.e., the proportion of participants who after the second year need only the rent supplement, but not the support of a clinical support team), and the death rate obtained from outside sources. IVResults.Data from 425 (257 in the HF group and 168 in the TAU group) of the 463 individuals randomized at the beginning of the study were included for analysis.Results indicate that HF is both cost-saving and more effective than treatment as usual (TAU).Over ten years, HF participants averaged an additional 1,501 days in stable housing compared to TAU, while costing $26,527 less.Individuals who had a longer history of homelessness and higher need level had the largest cost savings.Savings stem from individuals in HF transitioning and staying in HF apartments at a higher rate.TAU groups tend to spend more time in expensive forms of unstable housing such as emergency housing and substance abuse treatment.Housing First continued to be more effective and less costly over plausible ranges of the parameters selected for sensitivity analyses.Conclusion.This model illustrates the differences in effectiveness of HF based on clients' need level and homelessness history.Overall findings suggest long-term costeffectiveness of HF is even greater than suggested by the two-year findings. V RésuméContexte.L'itinérance est en expansion au Canada et à l'étranger.Elle augmente considérablement le risque de mortalité et constitue donc un problème de santé publique.Le logement d'abord (LA), une approche qui consiste à fournir l'accès immédiat à un logement permanent et l'accès à des services de soutien individualisés, est un élément-clé des stratégies pour mettre fin à l'itinérance.Une étude de sa coût-efficacité pendant les deux premières années a conclu que le LA entraînait des compensations de coûts importantes mais n'était pas entièrement rentabilisé.Objectif.L'objectif de ce projet était de développer un modèle de simulation pour projeter les effets de l'approche HF sur les coûts, d'un point de vue sociétal, et la stabilité du logement, en comparaison avec les services habituels, sur un horizon de dix ans.Approche.Un modèle de simulation de Markov été construit utilisant une nouvelle plateforme, Discretely Integrated Condition Event (DICE), construite avec le logiciel Excel.Les données sur les coûts et les résultats ont été déterminées à partir du site de Montréal de l'essai contrôlé à répartition aléatoire At Home/Chez Soi.Les individus ont été divisés en huit sousgroupes selon le niveau de besoin, les antécédents d'itinérance et le groupe d'intervention.Neuf états de logement possibles, tels que la rue, les abris d'urgence, l'hospitalisation psychiatrique et la prison, ont été définis.Les probabilités quotidiennes de transition entre les états de logement ont été calculées par sous-groupe.Les coûts des soins de santé, des services sociaux et de la justice ainsi que les revenus moyens ont été calculés pour chaque état de logement en utilisant la régression généralisée des moindres carrés.Les jours dans un logement stable ont été utilisés comme mesure d'efficacité.Des analyses de sensibilité ont été menées sur le taux d'actualisation, le taux d' « autonomisation » des services fournis par LA (c'est-à-dire, le pourcentage de participants qui, chaque année après la seconde année, n'ont plus besoin du soutien d'une équipe
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.018 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".