Background Few studies have tested whether ACP can sustain congruence in treatment preferences over time.

Aim To test the efficacy of an ACP intervention to sustain congruence in treatment preferences at 3-month post-intervention among adolescents with HIV and their families.

Methods A two-arm, randomised, controlled trial was conducted in five hospital-based clinics from July 2011–June 2014. Dyads (N = 91) were randomised to either the FACE intervention (Lyon Advance Care Planning Survey, Respecting Choices Interview, Advance Directive Completion) or Control (Developmental History, Nutrition, Safety Tips). Three, weekly, 60 min sessions were conducted. Outcome measure was Statement of Treatment Preferences.

Results Adolescents’ mean age was 17.9 years (range 14–20); 54% male; 92% Black. Immediately following the intervention, FACE dyads had significantly higher congruence in treatment preferences than controls (Situation 1- kappa 0.63 vs. –0.17 p < 0.000; Situation 2- kappa 0.59 vs. –0.07 p = 0.0002; Situation 3- kappa 0.76 vs. 0.16 p = 0.0003). At 3-month post-intervention, FACE group had significantly higher kappa for Situation 2 compared to controls (0.55 vs. 0.13, p = 0.0118). Agreement to discontinue treatment was always higher among FACE dyads than controls immediately following the intervention (Situation 1–15% vs. 0%; Situation 2–13% vs. 4%; Situation 3–24% vs. 4%) and at 3-month post-intervention (Situation 1–11% vs. 0%; Situation 2–23% vs. 6%; Situation 3–17% vs. 6%).

Discussion Congruence in treatment preferences was consistently higher for intervention dyads compared with controls. However, congruence among intervention dyads diminished over time, indicating a short-term intervention effect.

Conclusion Booster sessions may be needed to sustain congruence over time.