Validity of Self-Reported Health Service Utilization Data in the Homeless

BACKGROUND: Determining the validity of self-report data from homeless populations is critical to appropriate interpretation of that data. The validity of self-reported emergency room use and hospitalizations among 161 chronically medically ill homeless adults participating in a clinical trial was assessed.

METHODS: Patients who were homeless for at least 30 days prior to admission were enrolled from 2 hospitals and followed for 12 months. Concordance of self- report health service use data with electronic and paper medical records was assessed.

FINDINGS: 71% were men, the mean age was 46years (SD 8.6) and 81% self-identified as African-American. 81% (130/161) reported at least one hospitalization (exclusive of admitting emergency room visit) during the 12 months. The sensitivity of these reports was 97%, 18 participants gave false positive reports and 3 gave false negative reports. 75% (118/158) reported at least one emergency room visit during the 12 months. The sensitivity was 86%, 25 participants gave false positive reports and 12 gave false negative reports.

CONCLUSIONS: Rates of health service use were high in this population. Self-report data on emergency rooms visits was less valid than that of hospitalizations; recall bias may contribute to these findings. Self-report data on emergency room visits in the chronically ill homeless should be interpreted with caution.