Background:  2006 CDC guidelines recommend routine HIV screening in all health care settings and further suggest the use of existing resources and personnel performing routine clinical duties to meet this goal.  Our objective was to compare HIV test offer and acceptance rates in an urban emergency department (ED) when implemented by a dedicated HIV counselor vs. a current member of the ED staff.
Methods:  We conducted a single-center randomized controlled trial (the USHER Trial) of HIV testing strategies in an ED using the oral fluid OraQuick® test.  Eligible ED patients were 18-75 y/o, fluent in English/Spanish, not known to be HIV+, and triaged with an ED severity index (ESI) score 3, 4, or 5 (not acutely ill, competent to consent).  Trial enrollment occurred >60 h/wk (8a-MN). Consenting patients were randomized to HIV testing by a dedicated counselor (counselor arm) or by an ED provider (provider arm).  Counselors had all testing-related responsibilities (consent, counseling, delivery of both reactive [R] and non-reactive [NR] results). In the provider arm, a trained ED nursing assistant consented and tested the patient. NR results were delivered by house officers and R results by attending physicians. 
Results: Among 8,192 eligible patients approached, 4,859 (59%) agreed to participate and were randomized.  The mean age of participants was 37+14 yrs; 35% were male, 42% were white, and 37% had <HS education.  80% (1,960/2,447) of patients in the counselor arm were offered an HIV test, compared to 35% (863/2,412) in the provider arm (p<0.0001).  Acceptance of HIV testing was comparable in the provider (75%) and counselor arms (71%).  Among all patients randomized, the overall test rate favored the counselor arm (57% vs. 27%, p<0.0001).  Counselor offer rates remained similar across all demographic groups, ESI scores, and times of day.  Providers were significantly less likely to offer tests to patients >60 y/o compared to those 18-29 y/o (25% vs. 39%, p for trend = 0.0006).  Despite increases in trained personnel over time, provider offer rates were lower in the last 2 months of the study, compared to the first 3 months (27% vs. 70%, p for trend <0.0001), suggesting waning participation by providers over time.

Conclusions: Routine HIV testing in the ED setting is accomplished more frequently by dedicated HIV counselors than by ED staff in the course of routine clinical work.  Test acceptance rates do not appear to suffer when counselors are involved.