Chatbot-Delivered Real-Time Support to Improve HIV Self-Testing Rates: A Randomized Clinical Trial.

Abstract

HIV self-testing (HIVST) with online real-time support by human administrators (HIVST-OIC) is an evidence-based HIV intervention for men who have sex with men (MSM) but requires intensive resources to implement. To investigate whether a novel HIVST service using a chatbot to deliver online real-time instruction and counseling support (HIVST-chatbot) is noninferior to the HIVST-OIC in increasing HIVST uptake and counseling support for MSM. A partially blinded, parallel-group, and noninferiority randomized clinical trial was conducted between April 16, 2023, and May 31, 2024, in Hong Kong, China. Eligible participants were Hong Kong-based, Chinese-speaking MSM 18 years or older who reported anal intercourse with men in the past 6 months and had access to WhatsApp. Participants were recruited via gay venues, online recruitment, and peer referrals and then randomized to the HIVST-chatbot or HIVST-OIC groups. Both intention-to-treat (ITT) and complete case analyses were performed. The noninferiority margin was set at 10 percentage points. Standard-of-care pretest and posttest counseling was delivered by a chatbot for the HIVST-chatbot group and by trained human administrators in the HIVST-OIC group. Both groups received free HIVST kits. Primary outcomes included self-reported HIVST uptake and receiving any counseling support along with HIVST, validated by the research team. A total of 531 participants (mean [SD] age, 34.8 [9.3] years) were randomized to either the HIVST-chatbot (n = 266) or HIVST-OIC (n = 265) groups. In ITT analysis, the HIVST-chatbot was noninferior to the HIVST-OIC in increasing HIVST uptake (216 [81.2%] vs 227 [85.7%]; proportion difference, -4.5 percentage points; 95% CI, -9.8 to 0.8 percentage points; 1-sided P = .10) and proportion of HIVST users who received counseling support (197 [91.2%] vs 142 [62.6%]; proportion difference, 28.7 percentage points; 95% CI, 22.5 to 34.8 percentage points; 1-sided P < .001) at month 6. The HIVST-chatbot was more cost-effective than HIVST-OIC. In this randomized clinical trial, the HIVST-chatbot was noninferior to the HIVST-OIC in increasing HIVST uptake and counseling among HIVST users. These results suggest that the HIVST-chatbot might be an alternative to the evidence-based HIVST-OIC service for MSM. ClinicalTrial.gov Identifier: NCT05796622.