Investigating the mechanistic relationship between smoking and sleep to inform a tailored digital sleep intervention for smokers.

Lead Research Organisation: University of Bristol
Department Name: Electrical and Electronic Engineering


Hypothesis: There is good evidence of a reciprocal relationship between sleep and smoking. For example, we have previously found strong evidence for positive genetic correlations between insomnia and both smoking initiation and smoking heaviness. Using Mendelian randomization, we also found evidence that insomnia causally increases smoking heaviness (inverse-variance weighted fixed effects meta-analysis: beta 1.21, 95% CI 0.20 to 2.22) and causally decreases likelihood of smoking cessation (odds ratio 0.80, 95% CI 0.65 to 0.97). These data indicate that sleep disturbance may play an important role in the continuation of smoking. In addition, smoking has been linked to sleep disturbance and changes to sleep architecture, suggesting there may be a negative cycle of sleep disturbance and smoking that is difficult to break. Several well validated sleep interventions exist that could be tailored for smokers, and we hypothesise that delivering these in a targeted way to smokers utilising digital technology may reduce smoking heaviness and improve cessation rates.
Aims: This PhD project will involve three work packages to inform future post-doctoral work in this area. These will assess: (1) the mechanisms through which sleep disturbance or poor sleep quality may affect smoking behaviour and cessation in order to better tailor interventions; (2) machine learning and modelling techniques to predict withdrawal, smoking urges, abstinence self-efficacy, attentional bias and smoking topography ; (3) the development of a digital intervention tailored to smokers based on current sleep interventions.
Experimental Studies: Study 1 will be a within-subjects design with one factor of sleep disturbance (vs. normal sleep). Smokers will complete a smoking-related test battery (self-reported nicotine withdrawal, smoking urges, abstinence self-efficacy, attentional bias and smoking topography) after one night of disturbed sleep (achieved by text messages during the night that are repeated until responded to) and after one night of normal sleep (at least one week apart).
A study investigating predictors of relapse in a smoking cessation study (Nakamura et al. 2014) identified an effect size of d = 0.2 when comparing smoking urges in smokers who relapsed (M = 0.6, SD 0.8) and those who did not (M = 0.4, SD 0.6). To observe a comparable effect with 80% power at an alpha level of 5%, we would require a total sample size of 140. As we will use a within-subjects design, we will recruit 70 smokers in total.
Machine learning and modelling:
Study 2 would include the development of trigger algorithms using third party (existing) app technology. A machine learning model will be developed as a predictive tool to predict instances of high risk of smoking relapse based on data from study 1 .
Co design: Study 3 utilises a co -design approach with current and quitting smokers This study will inform the design of the digital intervention. It will also address any additional barriers to good quality sleep or implementing sleep interventions in smokers, to inform development of a novel digital sleep intervention for smokers.
Digital Intervention development: The aim of a sleep intervention is to improve sleep quality generally, but this does not negate the occurrence of poor sleep quality on some nights. These situations are likely to put quitting smokers at heightened risk of smoking relapse. We propose that the smoking-tailored intervention would therefore include a trigger intervention (informed by study 2), whereby poor sleep is identified by a smartwatch. This in turn initiates a JITAI. The support delivered would be informed by study 1 (which will identify what smoking-related variables are particularly sensitive to sleep disturbance). The design of the intervention will be generated by the co-design process in study 3. Together these studies would support a future funding bid to assess the feasibility and acceptability of the developed digital JITAI.


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/S023704/1 01/04/2019 30/09/2027
2383145 Studentship EP/S023704/1 23/09/2019 22/09/2023 Joe Matthews