The purpose of this study is to see if a computerized counseling tool helps patients reduce their sexual transmission risk and improve their antiretroviral adherence.

Helping people living with HIV to remain healthy and reduce transmission to sexual partners ('positive prevention') can be accomplished by behavioral counseling and supporting adherence to infectivity-reducing antiretroviral therapy (ART). Positive prevention and ART adherence approaches that are not staff-dependent are urgently needed in Africa, the region hardest hit by HIV and decimated health workforce. Interactive health communication tools offer one approach. We will adapt a computerized counseling intervention found to be efficacious in reducing HIV-1 viral load and risk behaviors in the US ('CARE+') with the largest HIV provider in Kenya, the Academic Model for the Prevention and Treatment of HIV/AIDS (AMPATH). Aim 1: Adapt a theoretically driven computerized counseling intervention for use in Kenya ('CARE+_Kenya'). Conduct in-depth interviews with n≤ 50 urban and rural AMPATH patients to understand HIV support needs, and two staff focus groups to assess counseling practices and beliefs about computer use. Modify intervention content; translate and record audio into local Kiswahili. Adapt skill-building videos (e.g., on secondary prevention, HIV disclosure, ART adherence, reproductive health). Conduct software usability testing with n=20 patients and n=8 staff. Perform 3-day test-retest reliability assessment to establish psychometric performance of measures.

Aim 2: Establish biological and behavioral efficacy of CARE+_Kenya. Longitudinal randomized controlled trial (RCT) in one urban and one rural AMAPATH clinic. Randomly assign HIV-positive adults with any missed ART or unprotected sex in last 6 months, >1 sex partner in last year, or sexually transmitted infection (STI)diagnosis in last 3 years, to intervention (n=125) or risk-assessment control (n=125) for baseline, 3, 6, and 9 month sessions. HIV transmission risk will be measured by self-reported unprotected sex with HIV negative/unknown partner, and trends in Chlamydia trachomatis, Neisseria gonorrhoeae, and T. vaginalis. ART adherence will be measured by HIV-1 viral load, electronic monitoring, pharmacy refill, self-report, and clinic attendance. Aim 3: Establish cost-effectiveness of CARE+_Kenya. At baseline, follow 100 patients at each of the two clinics to evaluate standard of care counseling, and collect time-spent and facility data to determine costs and unmet counseling need. Conduct economic evaluation to compare CARE+_Kenya vs. standard of care. If the RCT shows that the intervention reduces viral load and transmission risk behaviors, we will use a Bernoulli transmission dynamics model to estimate number of secondary HIV infections prevented; then create a cost-effectiveness model to calculate 2 incremental cost-effectiveness ratios: 1) cost per HIV infection averted, and 2) cost per daily adjusted life year saved. If CARE+_Kenya is efficacious and efficient, we will develop a proposal for a cluster-randomized trial to assess translational effectiveness of CARE+_Kenya throughout the AMPATH system. This is directly responsive to PA-08-107's call for innovative, integrated interventions that leverage ART roll-out infrastructure in international settings to benefit people living with HIV.

Inclusion Criteria:

• Seen for care including ART at the clinics
• Able to understand spoken Kiswahili or English
• Reported less than "perfect" ART adherence/any missed dose or pill counts that indicate non-adherence with medication or delay in pharmacy refill; unprotected sex in the last 6 months, or >1 sex partner in last year, or any STI diagnosis in last 3 years
• Able to give consent (i.e., no evidence of inebriation or psychosis)

Exclusion Criteria:

• Not fluent in Kiswahili or English
• Has a thought disorder that precludes participation

The CARE Kenya counseling tool software randomizes the participant automatically to the control or intervention arms when the user first logs into the program, using a quasi-random number algorithm.