Introduction Smoking in pregnancy negatively impacts the mother’s and offspring’s health. Mobile health, especially mobile phone-based approaches, shows promise in supporting pregnant women to quit smoking. Our previous research indicated that support from laypersons, such as family, friends and coworkers, can increase smoking cessation. Building on these findings, we developed an innovative smartphone application (app) that combines the power of social support with artificial intelligence. The app leverages the unique functionality of smartphones and reinforcement learning (RL) to provide a deeply tailored intervention that continuously adapts while emphasising positive support through reciprocal interactions between the smoker and the support person. Herein, we report the methods used to develop, implement and test the feasibility and effectiveness of a novel adaptive mobile pregnancy tobacco cessation app-based intervention using deep tailoring and a self-nominated supporter.

Methods and analysis The study is a type II hybrid effectiveness-implementation randomised controlled trial. In total, 300 pregnant smokers and their supporters will be randomised to (1) the SmokeFree Together 2.0 app and an RL-based intervention that continuously decides if a counselling call is needed using a fixed algorithm measuring motivation and craving or (2) a control group. The intervention will be administered throughout pregnancy up to 1 month after giving birth, with a 5-month follow-up period and outcome assessments at 3-month and 6-month post partum. Cessation (primary outcome) is defined as 7-day point biochemically verified prevalence of tobacco use and abstinence since birth, assessed at 6 months post partum. The central hypothesis is that the intervention will show evidence of feasibility and effectiveness in increasing social support, pregnancy cessation and postnatal abstinence.

Ethics and dissemination The Institutional Review Board of Michigan State University (#IRB00000297) approved this trial. The findings will be disseminated through peer-reviewed publications and participation in scientific conferences.

Trial registration number NCT05337722. Protocol version: 1.3 from 15.05.2024.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Tobacco remains the leading global cause of preventable disease and death, killing more than 8 million people a year around the world.1 Tobacco smoking among women in Romania increased dramatically by close to 30% between 2011 and 2018, and the share of smokers with quit attempts decreased by 37% during this period.1 Smoking among reproductive-aged women is especially harmful because, in addition to its effects on the smoker’s mortality and morbidity, it negatively impacts pregnancy,2 3 birth4 and the health of the offspring.5 6 In Romania, 50% of women smokers continue during pregnancy, and most experience moderate-to-high nicotine dependence,7 with no improvement in the past decade.8 9 Nearly 60% of those who quit during pregnancy relapsed post partum.

Access to tobacco cessation services in Romania is limited, including services for pregnant smokers. There is no automatic referral of pregnant smokers to a quitline or a cessation programme. Many pregnant smokers in Romania are not asked about smoking, are not counselled to quit and receive mixed information from their primary care physician (PCP)10 or obstetrics-gynaecology (OB-GYN) regarding the benefits of quitting smoking.11 Some providers advise against cessation, arguing that quitting may stress the fetus and that smoking a few cigarettes a day during pregnancy is acceptable.10 Nonetheless, most pregnant smokers understand the benefits of quitting and want to stop smoking.10 Research suggests that support from laypersons, including relatives, friends, coworkers or partners, may encourage cessation and abstinence.12 There is relatively little research on the role of social support in quitting smoking during pregnancy, and the available data have shown limited success,12 most likely because eliciting effective support proved difficult. Similar to pregnancy-specific studies, most general population interventions that assessed peer or partner support showed that the interventions did not actually increase the support for cessation.12

The potential of mobile health, especially mobile phone-based strategies, to increase the effectiveness and reach of behaviour change interventions is enormous, given the large and expanding smartphone ownership.13 Even in resource-constrained countries, mobile phone ownership is very high. In Romania, a low- and middle-income country at the start of this study, the smartphone penetration rate was close to 70% in 2017.14 Applications (apps) leverage the unique functionality of smartphones to deliver messages and content deeply tailored using momentary assessments, including readiness to quit, cravings, stress and need for support.15 To our knowledge, there is only one large-scale tested pregnancy-specific app for smoking cessation16 17 and no tested app for smoking cessation with lay support in pregnancy nor in the general population.18

This manuscript presents the protocol of the Smoke-Free Together 2.0 (SFT2.0), a type II hybrid effectiveness implementation randomised controlled trial (RCT) targeting tobacco smokers during pregnancy and the first month post partum. The RCT evaluates a user-centred, artificial intelligence-driven mobile app for tobacco cessation during pregnancy. The app features deep tailoring of messages and the inclusion of a self-nominated support person to enhance social support. The intervention is grounded in self-determination theory (SDT)19–21 and uses motivational interviewing (MI),22–24 a counselling style that is effective in assisting people to quit smoking. The central hypothesis is that the intervention will show evidence of feasibility and effectiveness in increasing social support, pregnancy smoking cessation and postnatal abstinence, while achieving these outcomes more efficiently through reinforcement learning that optimises the delivery of MI counselling sessions.

The study protocol and all study-related materials received ethical approval from Michigan State University Institutional Review Board (IRB). Before study enrolment, all participants provide electronic informed consent through the secure Qualtrics platform. Additionally, participants complete GDPR consent forms as required by European Union data protection regulations. Protocol modifications are documented as amendments and validated by the Michigan State University IRB.

The intervention is delivered by counsellors trained in MI techniques for smoking cessation. To ensure intervention quality, counsellors receive ongoing supervision and participate in regular team meetings to discuss complex cases. Their sessions are also coded for fidelity to MI. For participants identified with serious mental health conditions, referral protocols are in place to connect them with certified psychotherapists while maintaining their study participation when appropriate.

Trial findings will be disseminated to participants through a plain-language summary and to the broader Romanian medical community through our healthcare provider network.

Tobacco use remains a leading global cause of preventable disease and death,32 with significant challenges in cessation, particularly among pregnant women.33 34 App-based interventions to promote quitting during pregnancy are scarce, with mixed findings.17 While counselling has been shown to be effective, it can be costly in developing economies, thus strategies to employ counselling more judiciously are needed. This RCT introduces an innovative approach to addressing smoking cessation during pregnancy that leverages smartphone technology, peer support and reinforcement learning to determine when counselling calls are employed.

The SFT 2.0 intervention goes beyond the state-of-the-art pregnancy smoking cessation interventions as: (1) it expands the breadth of supporters to any lay person (eg, family, relative, friend, coworker, spouse) nominated by the pregnant smoker; (2) it adds deep tailoring and reinforcement learning to adapt the intervention continuously; (3) it uses formative data from the target population to increase engagement with the intervention; (4) it shifts current approaches by building implementation and cost considerations into intervention development and study design in a hybrid trial approach; and (5) the intervention’s theoretical foundation in SDT and MI provides a robust conceptual framework.

These innovative features directly inform the study’s methodological approach. The hybrid effectiveness and implementation design are particularly noteworthy. Unlike traditional RCTs, this approach allows for simultaneous evaluation of intervention effectiveness and implementation strategies to better understand how to optimise and accelerate the translation of the SFT2.0 app-based intervention into practice.35

On successful completion of the RCT, we expect a feasible and acceptable app-based pregnancy smoking cessation intervention with lay support that is effective in increasing cessation and abstinence. This will be significant because it will provide strong evidence of lay support’s effectiveness in using smartphones’ functionalities to increase smoking cessation. In addition, this research has broader implications beyond immediate tobacco cessation, including a wide range of public health interventions potentially relevant to all women of reproductive age, including nutrition, physical activity and childcare. It demonstrates a potential model for developing user-centred, technology-driven health interventions that can be adapted to various behavioural health challenges. The approach could be particularly valuable in low- and middle-income countries with limited healthcare resources.

Several limitations should be acknowledged. First, the context-specific nature of the study, focused on Romania, may limit the generalisability of the findings. There may be unique cultural and systemic factors that limit generalisability even to other Eastern European cultures. Second, the reliance on smartphone technology, broadband internet access and Android-operated devices could induce access barriers for some users.

Our rules-based RL system used a fixed algorithm based on motivation to quit and risk of relapse for quitters measured every week to assign who gets a counselling call. The goal is to increase the efficiency of counselling resources by targeting those who need it the most. More sophisticated RL systems can learn and adapt the allocation algorithms based on how they impact short-term functions (eg, motivation), and they can be tailored to user characteristics (eg, age, baseline smoking).36 37 Time and budget reductions precluded the use of a more adaptive ‘living’ RL intervention.

In summary, given the impact of tobacco use during pregnancy and the limited effectiveness of current interventions, this trial will provide evidence on the feasibility and effectiveness of a novel, user-centred mobile app, using basic reinforcement learning, for pregnancy tobacco cessation that leverages lay support, deep tailoring, and adaptive MI counselling to increase positive support, pregnancy cessation and postnatal abstinence.

We are grateful to our participants and the healthcare partners throughout Romania, including gynaecologists, primary care physician, community health mediators and prenatal educators who support participant recruitment. We especially acknowledge our programme ambassadors for their efforts in disseminating the project.