RESUMEN
OBJECTIVE: To determine which of 3 HIV testing and counseling (HTC) models in outpatient departments (OPDs) increases HIV testing and entry of newly identified HIV-infected patients into care. DESIGN: Randomized trial of HTC interventions. METHODS: Thirty-six OPDs in South Africa, Tanzania, and Uganda were randomly assigned to 3 different HTC models: (A) health care providers referred eligible patients (aged 18-49, not tested in the past year, not known HIV positive) to on-site voluntary counseling and testing for HTC offered and provided by voluntary counseling and testing counselors after clinical consultation; (B) health care providers offered and provided HTC to eligible patients during clinical consultation; and (C) nurse or lay counselors offered and provided HTC to eligible patients before clinical consultation. Data were collected from October 2011 to September 2012. We describe testing eligibility and acceptance, HIV prevalence, and referral and entry into care. Chi-square analyses were conducted to examine differences by model. RESULTS: Of 79,910 patients, 45% were age eligible and 16,099 (45%) age eligibles were tested. Ten percent tested HIV positive. Significant differences were found in percent tested by model. The proportion of age eligible patients tested by Project STATUS was highest for model C (54.1%, 95% confidence interval [CI]: 42.4 to 65.9), followed by model A (41.7%, 95% CI: 30.7 to 52.8), and then model B (33.9%, 95% CI: 25.7 to 42.1). Of the 1596 newly identified HIV positive patients, 94% were referred to care (96.1% in model A, 94.7% in model B, and 94.9% in model C), and 58% entered on-site care (74.4% in model A, 54.8% in model B, and 55.6% in model C) with no significant differences in referrals or care entry by model. CONCLUSIONS: Model C resulted in the highest proportion of all age-eligible patients receiving a test. Although 94% of STATUS patients with a positive test result were referred to care, only 58% entered care. We found no differences in patients entering care by HTC model. Routine HTC in OPDs is acceptable to patients and effective for identifying HIV-infected persons, but additional efforts are needed to increase entry to care.