Related Experiment Video
Updated: Apr 16, 2026
Inverse Probability of Treatment Weighting Propensity Score using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
Homeless Patients in the ICU: An Observational Propensity-Matched Cohort Study
Naïke Bigé1, Gilles Hejblum, Jean-Luc Baudel
11Service de Réanimation Médicale, AP-HP, Hôpital Saint-Antoine, Paris, France. 2Faculté de Médecine Pierre et Marie Curie, UPMC Univ Paris 06, Sorbonne Universités, Paris, France. 3U1136, INSERM, Paris, France. 4UMR_S 1136, UPMC Univ Paris 06, Sorbonne Universités, Paris, France. 5Unité de Santé Publique, Hôpital Saint-Antoine, AP-HP, Paris, France. 6Service Social Hospitalier, Hôpital Saint-Antoine, AP-HP, Paris, France. 7Département de l'Information Médicale, Hôpital Saint-Antoine, AP-HP, Paris, France.
Objective:
To describe epidemiology and outcome of critically ill homeless patients, as compared with those of nonhomeless patients.
Design:
Homeless and nonhomeless admissions were matched on the basis of a 1:4 ratio, using a propensity score-based procedure involving age, sex, date, and main diagnosis at ICU admission.
Setting:
A 18-bed closed medical ICU of a French tertiary care university hospital.
Patients:
All consecutive admissions from July 2000 to December 2012.
Interventions:
None.
Measurements And Main Results:
There were 421 homeless and 9,353 nonhomeless admissions. Considering homeless admissions, 50% patients had no health insurance, 56% had no financial resource, 91% were socially isolated, and 69% lived in street. In a multivariable analysis of homeless admissions including age, sex, and Simplified Acute Physiology Score II, living in street was significantly associated with hospital mortality (odds ratio=2.94; 95% CI, 1.30-7.10; p=0.012). As compared with nonhomeless, homeless admissions more frequently concerned men (89% vs 57%; p<0.0001) and younger patients (49 yr [43-57] vs 62 yr [46-76]; p<0.0001), whereas Simplified Acute Physiology Score II (37 [24-50] vs 37 [25-52]; p=0.99) and distribution of the number of organ supports (p=0.49) were similar. ICU mortality concerned 19.1% and 18% of matched homeless and nonhomeless admissions, respectively. The corresponding figures for hospital mortality were 20.8% and 20.6%. In multivariable analysis, homeless status was associated with neither ICU (odds ratio=1.27 [0.92-1.73]; p=0.14) nor hospital mortality (odds ratio=1.07 [0.77-1.49]; p=0.68), while it was independently associated with longer ICU (means ratio=1.16 [1.01-1.34]; p=0.035) and hospital (means ratio=1.30 [1.12-1.49]; p=0.0002) stay of survivors.
Conclusions:
Critically ill homeless patients benefit from the same level of care and have globally the same prognosis than housed patients but experience longer lengths of stay. Most precarious patients living in street have a higher mortality rate. The study perspective is not ICU centered but also concerns the global organization of healthcare since homeless patients are referred by numerous sources and discharged to different wards.
