Lapses and cues to smoke

Any smoking early on in a quit attempt (a ‘lapse’) is highly predictive of longer-term return to smoking (‘relapse’).^6,7 For example, one large prospective study found that 22% of smokers who lapsed early, a mean of 10 days after they started their quit attempt, were abstinent at 6 months compared to 71% who did not lapse early on in their quit attempt.⁷ This relationship also exists when a lapse is experimentally induced.⁸ Evidence, therefore, indicates that maintaining abstinence during the first few weeks of a quit attempt seems to be particularly crucial for long-term success.

Almost half of all lapses are estimated to be induced by cues to smoke from the environment or setting,⁹ through the generation of a type of craving referred to as ‘cue-induced craving’. Cues have been categorised as ‘proximal’, such as seeing unlit cigarettes or other people smoke, and ‘distal’, that is the actual environments that an individual smokes in or has smoked in.^10,11 Distal cues can be generic, such as a bar, or specific, such as someone’s back garden. Studies have shown that smokers will experience cravings after viewing generic images of environments associated with smoking selected by an experimenter, such as a bus stop, and that this effect is magnified when using images taken by participants of their own specific smoking locations.^12,13 Given that people attempting to quit smoking will very likely experience cravings driven by the exposure to both proximal and distal cues in their daily life, with a higher magnitude of cravings when both types are present simultaneously,¹¹ the effective management of cues could prevent early lapses and increase successful smoking cessation.

However, there is a lack of effective support available to help smokers manage cue-induced cravings. The most used medications for smoking cessation do not address cue-induced cravings. This includes steady-state medications such as bupropion, varenicline and nicotine patches.^14–17 While fast-acting nicotine replacement therapy (NRT) can help reduce these cravings,¹⁶ only a minority of NRT users use this type¹⁸ and the proportion of people attempting to quit in England who use NRT has been steadily decreasing, currently standing at 11% in 2022.¹⁹ Cognitive or behavioural lapse prevention strategies, such as using self-talk or avoiding other smokers, can help smokers avoid or manage cue-induced cravings,^16,20,21 though the strategies smokers are most likely to use are those least effective or with the least strong evidence base.²² An additional challenge with targeting cue-induced cravings with a behavioural response is that the window of opportunity for intervening is short. One of the only studies of its kind found half of lapses brought about by acute craving episodes occurred on average within 11 minutes after craving onset.⁹ Support to address cue-induced cravings therefore needs to be rapid, easily accessible in different locations and deliver evidence-based strategies.

Digital cessation interventions

While each year many hundreds of smoking cessation apps become publicly available, the vast majority demonstrate poor adherence to clinical guidance and evidence-based approaches²³ and even fewer have been evaluated.²⁴ Among those where evaluation research has been undertaken, there is a lack of clinical trials.²⁵ A 2019 Cochrane Review pooled five RCTs of cessation apps identified and found there was no evidence of effectiveness for cessation apps compared to a less intensive cessation intervention comparator. A more recent review (2021) identified 11 RCTs, 4 of which showed evidence of effectiveness.²⁶ However, smartphone apps are primarily delivery mechanisms for delivering support and so different therapeutic approaches delivered by apps are likely to have different success rates for cessation. RCTs have provided evidence that acceptance and commitment therapy can be effectively delivered through an app to promote cessation²⁷ but that mindfulness training for smoking cessation delivered via an app was not effective when compared to a monitoring-only intervention.²⁸ Furthermore, trials randomising smokers recruited after installing an app from an app store provide evidence that adding a chatbot feature to a cessation app increases short-term effectiveness²⁹ but that a multiple-features cessation app including craving feedback, counselling messages and NRT and e-cigarette advice was no more effective than a basic scaled-down cessation app.³⁰

Smartphone apps may be a potentially effective way of addressing cue-induced cravings. Unlike traditional in-person delivered cessation interventions, digital interventions, such as smartphone apps, have the advantage of portability and so are much more likely to be on hand when needed. Some mobile phone-based interventions have features that specifically aim to deliver support to reduce lapse risk in response to need identified by randomly triggered assessment prompts [via Ecological Momentary Assessment (EMA)]³¹ or on-demand requests for craving support.³² However, given the rapid time to lapse after the onset of a craving identified above, and evidence that some user-initiated on-demand craving tools are seldom used beyond a first try,^33–35 the efficacy of such features in addressing cue-induced craving may be limited.

Despite the potential of apps to deliver effective cessation support at the time of need, very few to date have been developed to achieve this and, among those that have,³¹ there have been a lack of RCTs to provide a better estimate of their impact.³⁶ Based on a recent systematic review,³¹ only one RCT has been undertaken testing a just-in-time support system for smoking cessation.³⁷ This pilot RCT (N = 81) tested an app (Smart-T2) where lapse risk was identified by randomly sent EMAs, which then triggered ‘in the moment’ support, and found that support could be feasibly delivered, although the study was underpowered to identify its impact on smoking and did not assess potential mechanisms of action.

The idea of developing cessation apps that passively ‘sense’ high-risk situations for smokers and respond in real time have been proposed for almost a decade,^38,39 with the expectation that they could reduce reliance and burden on the user to identify and respond to high-risk or high craving situations. Despite modern smartphones having the technological capability to support such apps, other than Quit Sense (see below), there have been no apps developed or evaluated that sense and adapt cessation support, so it is relevant to real-time context.

Quit Sense

Quit Sense was developed to address this important gap in digital quit smoking support and is a theory-guided ‘context aware’ smartphone app, also referred to as a Just-In-Time Adaptive Intervention (JITAI).³⁹ Quit Sense is trained by the individual smoker before their quit attempt starts in order to learn about their smoking habits, including the locations where they smoke and the smoking cues they are exposed to within these locations that precede their smoking. Then, once their quit attempt starts, Quit Sense delivers behavioural support triggered by and tailored to those smoking locations and the main cues within them. Alongside this, it provides additional components of behavioural support to prevent lapse and relapse that are not tailored to real time context.

Previous research has demonstrated that Quit Sense succeeds in providing ‘in the moment’ support to smokers, including the provision of lapse prevention strategies, and that it is both engaged with and found to be acceptable by users (N = 15).¹⁰ Logging a smoking episode, the primary method of training the app, took on average 13 seconds, including the identification of key smoking cues and location. There was an average of 70 sessions of app use per participant, the majority of which were to report smoking, complete a survey or view a support message, for the duration of use (up to 4 weeks). However, qualitative interviews identified barriers to logging smoking episodes due primarily to forgetting and environmental constraints, such as when driving or in the presence of others. These barriers led to under-reporting of smoking behaviour on approximately half of days pre-quit attempt. Participants overall were positive about having access to real-time support and had no privacy concerns with the collection of location data. Findings from this formative work were used to guide app optimisation. Optimisation included improved prompting and feedback to promote fuller recording of smoking episodes, the removal of the requirement for an internet connection to receive location-triggered tailored support, improved power efficiency of location sensors and enhancements to the app’s user interface. Changes were also made due to the COVID-19 pandemic, including ensuring support messages did not give recommendations that might contradict COVID guidance and ensuring the app was adapted to higher rates of working from home. A feature to record audio reports of users’ views of the app and how it could be improved was also added to the feasibility trial (‘auto-process evaluation’) to explore novel methods of evaluation.

Aims and objectives

The overarching aim of the proposed research was to establish the feasibility of a future RCT of the Quit Sense app.