SMART : a personalized, chat-supported sequential, multiple assignment, randomized trial for smoking cessation

Abstract

Background Mobile health (mHealth) is promising in developing personalized smoking cessation interventions. An adaptive approach that used stepped treatment to address participants’ unmet needs may benefit smokers who were not intended to quit. This thesis aimed to examine the effectiveness of personalized instant messaging (PIM) adapted with optional combined interventions (OCI) on smoking cessation, with a sequential, multiple assignments, randomized trial (SMART) design. Methods Prior to the intervention study, a secondary analysis was conducted to explore factors influencing changes in motivation to quit among proactively recruited community-based smokers during 2009-2018. The feasibility of chat-based intervention integrated with additional cessation support was tested via a pilot randomized controlled trial (ClinicalTrials.gov: NCT03574077). Formative interviews were conducted among those who received the chat-based intervention but failed to quit to inform the development of adaptive intervention strategies.

In the two-arm, parallel, accessor-blinded SMART, adult daily cigarette smokers were proactively recruited from 70 community sites in Hong Kong (ClinicalTrials.gov: NCT03992742). At baseline (phase 1), participants received brief cessation advice plus referral assistance to smoking cessation services and were randomized to receive chat-based personalized instant messaging (PIM) or unidirectional regular instant messaging (RIM). At 1-month (phase 2), PIM participants who remained smoking (non-responders) were randomized to receive either optional combined interventions [e.g., multi-media messages, nicotine replacement therapy sampling (NRT-S)] or continued-PIM. Non-responders in the RIM group were randomized to receive PIM or continued-RIM. Self-reported quitters (responders) in both groups received the initial intervention. The primary outcome was biochemical validated abstinence at 6 months post-treatment initiation with missing data coded as smoking (intention-to-treat). Perceptions towards the adaptive, personalized smoking cessation program were explored in semi-structured interviews. Results During 2009-2018, prevalence of intention to quit within next 30 days decreased from 87.3% in 2009 to 31.0% in 2018. Previous quit attempt and perceived self-efficacy on quitting also significantly declined. Mobile chat-based support was feasible and preliminarily effective in increasing quitting and NRT-S use. Additional individualized cessation supports were perceived necessary for those who were less motivated to quit. In the main trial, a total of 844 participants (82.3% men, 86.7% aged 18-59) were recruited and randomized (each group 422). 43.9% had no intention to quit within next 30 days and 29.4% had no previous quit attempt. More participants in the PIM group had validated abstinence compared with RIM group at 6 months (10.2% vs. 8.3%; RR 1.23; 95%CI 0.80 to 1.88; P=0.34). Among the non-responders, adapt the intervention for higher intensity did not outperform the maintaining strategy. Engagement in either the PIM (Adjusted RR 4.87; 95%CI 2.11 to 11.25; P<0.001) or RIM (Adjusted RR 3.32; 95%CI 1.43 to 7.74; P<0.001) intervention strongly predicted higher validated abstinence, but the RIM group had a doubled engagement rate as in the PIM group (53.3% vs. 23.0%). Conclusion This study was the first to show that conducting a SMART was feasible in informing an efficacious adaptive intervention for the community-recruited smoking population. Adaptive mHealth interventions showed promise in promoting smoking abstinence. Further studies are warranted to optimize intervention design for higher engagement.