Background. Persistent racial/ethnic disparities have been observed in the use of psychiatric emergency services (PES). This project explored how friend/family versus stranger involvement in PES admissions varies across racial/ethnic groups and is related to psychosocial functioning, and how neighborhood racial/ethnic composition is associated with the race-specific rate of hospitalizations and the level of psychosocial functioning amongst persons admitted involuntarily to PES.

Methods. Study staff abstracted medical and legal records from three months of involuntary admissions to San Francisco's public PES department and calibrated a measure of patient psychosocial functioning, the Global Assessment Scale (GAS) score. GAS scores were then predicted with multivariate regression; explanatory variables included gender, substance use, homelessness, racial/ethnic group (Asian, Black, Latino, White and other), friend/family involvement, and interactions between racial/ethnic group and friend/family involvement. Census data were used to estimate the racial/ethnic and socioeconomic composition of census tracts in which admission events occurred. Poisson and negative binomial regression were used to estimate race-specific admission rates for each tract.

Results. In 717 community-identified admissions, friend/family involvement was observed in significantly more Asian (56.4%) and Latino (47.4%) patients than white (28.5%) or Black (27.4%) patients. Functioning in Black (P<0.01) and Asian patients (p<0.01) was lower than in white patients, and friend/family involvement was associated with higher functioning (p<0.01). However, interactions between race/ethnicity and friend/family involvement revealed that racial/ethnic differences in functioning were due to white patient friend/family-involved admissions occurring at higher levels functioning than in any other group. Admission rates of Black patients by strangers were inversely associated with the proportion of Black residents, but functioning was not associated with neighborhood composition.

Conclusions. The relationship of the patient to those causing the admission was a stronger predictor of functioning than was neighborhood composition. Racial/ethnic differences in functioning at PES admission may be due in part to friends/family members of white patients seeking emergency treatment at less-severe levels of impairment, not due to strangers' tolerance varying by the patient's race/ethnicity. Assuming that the burden of care increases with lower patient functioning, Asian and Black patients' friends/family who delay treatment may experience a greater burden than their white counterparts.