Background High operating costs challenge sustainability of successful U.S. Emergency Department(ED) HIV screening programmes. Free-standing registration kiosks could potentially reduce the marginal costs of ED HIV screening. We investigated a cost-effectiveness ratio(CER) as cost per new HIV diagnosis for a kiosk-based approach for offering screening, versus a counsellor-based approach.

Methods A rapid oral-fluid HIV screening programme, instituted in a U.S. ED since 2005, had a rate of new HIV diagnosis 0.16% in 2012. A two-phase quasi experimental design, including a counsellor-based approach to offer testing at the bedside (Phase I, August and September 2011) and a kiosk-based approach to offer testing at ED registration (Phase II, December 2011 and January 2012), was performed. CER per new HIV diagnosis was defined as total cost of the screening programme divided by number of newly diagnosed cases. Costs included screening programme personnel (study coordinator, testing staff, and kiosk helpers), diagnostic assays (rapid and confirmatory tests), and kiosks (2 kiosks, software, and IT consulting fees). Sensitivity analyses were performed.

Results Slightly higher rates of newly diagnosed HIV positivity were observed with kiosk approach [I: 0%(0/538); II: 0.47%(2/430; 95% CI: 0.08%, 1.53%)]. Compared to phase I, those tested via kiosk were more likely to be younger and report high-risk sexual behaviours and/or injection drug use(IDU) [phase I vs. II: age: 39 years, 35 years; high-risk sexual behaviours: 30%, 51%; IDU: 1%, 7% (all p < 0.05)]. Projected first-year CER for kiosk-facilitated screening was $12,470 versus $18,406 counsellor-based screening(using a new HIV diagnosis rate in 2012). Sensitivity analysis based on 95% CI of positivity(0.08%, 1.53%) estimated the 1-year CER for kiosk-facilitating screening to be $82,847 to $3,814, respectively.

Conclusions Our pilot data demonstrated that use of kiosks for offering HIV screening engaged more high-risk patients, identified more unrecognised infections, and was potentially more cost-effective than a counsellor-based bedside approach.