Peer advocacy and access to hospital care for people who are homeless in London, United Kingdom 2019-2023: a cohort study
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Objectives
To measures differences in hospital use between homeless adults using a peer advocacy service (clients) and non-clients in London.
Design
We conducted a cohort study with linkage to hospital episode statistics (HES) one year prior and post enrolment.
Setting
London, UK
Population
People who are homeless in London aged over 18 years residing in a hostel, attending a day centre, or being referred by a homelessness service; experiencing difficulties accessing healthcare; and speaking either English or Polish. Participants were required to provide consent for linkage to Hospital Episode Statistics (HES). To be classified as a client, individuals must have used the HHPA service at least once between January and July 2021; non-clients were those who had never used the service.
Intervention
Peer advocacy is the provision of support by trained advocates with lived experience of homelessness to individuals to overcome barriers to accessing health services.
Outcomes
The primary outcome was not attending a scheduled outpatient appointment (‘did-not-attend’, DNA) over 12 months post-recruitment, commencing from their baseline interview date. Secondary outcomes included the number of A&E and inpatient admissions (all and planned admissions) during that same period.
Methods
We estimated the probability of DNA using Poisson regression and the number of inpatient admissions and accident and emergency visits using linear regression models. Models adjusted for: i) propensity score weights; and ii) propensity scores and imbalanced confounders. Sensitivity analyses assumed that participants who did not link to HES had no hospital attendance. Secondary analyses examined differential effects by type of peer advocacy engagement (new vs. ongoing clients; supported vs. unsupported) and anxiety or depression symptom scores measured with the Public Health Questionnaire 4 (PHQ4).
Results
153 clients and 159 non-clients were recruited between July-December 2021. Most were male (77.5%) with median age of 48 years. Weighted regression models suggested no effect of peer advocacy on DNA(Rate Ratio (RR) 0.97 95% CI 0.67, 1.42), no difference in the mean number of A&E visits (0.86, 95% CI -0.06, 1.79) but a difference in inpatient admissions 1.14 (95% CI 0.52,1.75). Sensitivity analyses suggested a higher number of completed outpatient appointments (1.77 95% CI 0.13,3.40) among clients. Clients with PHQ4 scores of 9-12 had greater probability of DNA at outpatient appointments (RR 1.98 95% CI 1.0,3.89), those with scores of 6-8 had 3.38 (95% CI 0.05,6.71) more completed appointments and 1.09 (95% CI 0.56-1.63) more inpatient admissions, relative to non-clients.
Conclusions
In the context of COVID-related disruptions to the work of peer advocates and health services we found mixed evidence on the effect of peer advocacy: with no impact on outpatient appointments or use of emergency services; but increased inpatient admissions.
Article Summary
Strengths and Limitations of this study
-
Use of linked hospital data minimised recall bias by relying on objective outcome measures sourced from national Hospital Episode Statistics (HES), enhancing data accuracy and reliability.
-
Application of propensity score weighting addressed baseline imbalances between intervention and comparison groups, allowing more robust adjustment for confounding and strengthening causal inference in a non-randomised design.
-
Potential for residual confounding remains, given notable differences between peer advocacy clients and non-clients and the possibility of unmeasured variables (e.g. severity of illness or motivation to seek care) influencing outcomes.
-
Data linkage was incomplete, with 26% of participants not matched to HES records, introducing potential selection bias and limiting confidence in sensitivity analyses that assumed non-attendance among unlinked individuals.
-
Impact of COVID-19 disrupted both intervention fidelity and usual care, with reduced in-person peer support and altered healthcare access potentially diluting the observed effects of the intervention and affecting generalisability.
