Estimation of causal effects of a time-varying exposure at multiple time points through Multivariable Mendelian randomization

Mendelian Randomisation (MR) is a powerful tool in epidemiology to estimate the causal effect of an exposure on an outcome in the presence of unobserved confounding, by utilising genetic variants as instrumental variables (IVs) for the exposure. The effects obtained from MR studies are often interpreted as the lifetime effect of the exposure in question. However, the causal effects of many exposures are thought to vary throughout an individual's lifetime and there may be periods during which an exposure has more of an effect on a particular outcome. Multivariable MR (MVMR) is an extension of MR that allows for multiple, potentially highly related, exposures to be included in an MR estimation. MVMR estimates the direct effect of each exposure on the outcome conditional on all of the other exposures included in the estimation. We explore the use of MVMR to estimate the direct effect of a single exposure at different time points in an individual's lifetime on an outcome. We use simulations to illustrate the interpretation of the results from such analyses and the key assumptions required. We show that causal effects at different time periods can be estimated through MVMR when the association between the genetic variants used as instruments and the exposure measured at those time periods varies, however this estimation will not necessarily identify exact time periods over which an exposure has the most effect on the outcome. We illustrate the method through estimation of the causal effects of childhood and adult BMI on smoking behaviour.

Household composition, smoking cessation and relapse: results from a prospective longitudinal Australian cohort

AimsTo examine the association between household members and their tobacco smoking behaviour on patterns of smoking cessation and relapse.Design and participantsData was sourced from 19 waves (years 2001 to 2019) of the nationally representative Household Income and Labour Dynamics in Australia (HILDA) survey, with all household members 15 years or older completing the survey annually. The final sample included, on average, 3,056 ex-smokers and 2,612 smokers per wave.MeasurementsSelf-reported annual smoking status was used to construct measures of smoking cessation and relapse. Information on household structure and relationships was then used to develop variables describing the presence of household members and their smoking status by relationship to the individual (i.e., child, parent, spouse, sibling, or other). Multivariate regression analyses were then used to predict the likelihood of smoking cessation and relapse controlling for the presence of other household members and their smoking status, sociodemographic characteristics, number of cigarettes smoked per day, previous quit attempts, and years abstained from smoking.FindingsIndividuals that lived with non-smokers were more likely to quit [OR1.22 (95%CI 1.11;1.34)] relative to those living alone. However, this favourable association was negated if living with another smoker, which was associated with a reduced likelihood of smoking cessation [OR0.77 (95%CI 0.72;0.83)] and a higher likelihood of relapse [1.37 (95%CI 1.22;1.53)]. In particular, living with a spouse or parent that smoked reduced the likelihood of smoking cessation [OR0.71 (95%CI 0.65;0.78) and OR0.71 (95%CI 0.59;0.84), respectively] and increased the likelihood of relapse [OR1.47 (95%CI 1.28;1.69) and OR1.39 (95%CI 1.00;1.94) respectively] relative to living with their non-smoking counterparts.ConclusionsHousehold composition and intrahousehold smoking behaviour should be considered when delivering, or estimating the benefits of, smoking cessation interventions. Interventions which encourage smoking cessation at the household level may assist individuals to quit and abstain from smoking.

Characteristics of Aboriginal and Torres Strait Islander Smokers Exposed to Anti-smoking Marketing in South Australia

Background: Smoking is disproportionately prevalent among Aboriginal and Torres Strait Islander Australian peoples, with 39% of Indigenous Australians aged over 15 years smoking daily. Efforts to reduce this high prevalence include culturally focused media campaigns, designed using community consultation, highlighting the need to determine how such health messaging is received by smokers. This study aimed to examine Indigenous Australian smokers’ reactions to a culturally focused anti-smoking mass media campaign—'Give up Smokes’. Methods: Intercept surveying across health services and events used recorded demographics, smoking status, quit attempts, smoking health effects, anti-smoking campaign recall, social support, and campaign reactions. Participants rated campaign images in five domains: 1) whether it made them stop and think; 2) personal relevance; 3) believability of design and message; 4) prompting concern about smoking; and 5) motivation towards quitting. Cluster analysis was used to segment smoker types. Results: Smoking health effects knowledge was high, and did not differ by quit readiness, attempts, or social support. Cessation support access was higher among those with greater readiness to quit. Social smoking behaviour and confidence to support others quitting did not significantly differ between participants, however importance of others quitting did. Quit readiness, attempts, and social support were associated with reaction to campaign design, but not message recall. Four types of smokers were described, using smoking characteristics, who differed in campaign message reactions. Conclusions: Strategies using campaign-exposed smoker characteristics to inform culturally focused health promotion are discussed in relation to four identified types of smokers.

The absence of association between anorexia nervosa and smoking: converging evidence across two studies

Previous studies have found increased smoking prevalence amongst adults with anorexia nervosa (AN) compared to the general population. The current investigation explored bidirectional associations between AN and smoking behaviour (initiation and heaviness), to address questions surrounding causation. In Study One, logistic regression models with variance robust standard errors assessed longitudinal associations between AN and smoking, using data from adolescent participants of the Avon Longitudinal Study of Parents and Children (N = 5100). In Study Two, two-sample Mendelian randomisation (MR) tested possible causal effects using summary statistics from publicly available genome-wide association studies (GWAS). Study One provided no clear evidence for a predictive effect of AN on subsequent smoking behaviour, or for smoking heaviness/initiation predicting later AN. MR findings did not support causal effects between AN and smoking behaviour, in either direction. Findings do not support predictive or causal effects between AN and smoking behaviour. Previously reported associations may have been vulnerable to confounding, highlighting the possibility of smoking and AN sharing causal risk factors.

Cigarette smoking and misperceived norms among adults in rural Uganda: a population-based study

BackgroundLittle is known about perceived norms about cigarette smoking in Uganda or the extent to which perceptions drive personal cigarette smoking behaviour.MethodsWe conducted a cross-sectional study in 2016–2018 that targeted all adults who resided within eight villages in Rwampara District, southwestern Uganda. Personal cigarette smoking frequency was elicited by self-report. We also asked participants what they believed to be the cigarette smoking frequency of most other adult men and women in their villages (i.e., perceived norms). Frequent cigarette smoking was defined as 4+ times/week. We compared perceived norms to cigarette smoking frequency reports aggregated at the village level. We used multivariable Poisson regression to estimate the association between perceived norms and personal cigarette smoking behaviour.ResultsAmong 1626 participants (91% response rate), 92 of 719 men (13%) and 6 of 907 women (0.7%) reported frequent smoking. However, 1030 (63%) incorrectly believed most men in their villages smoked cigarettes frequently. Additionally, 116 (7%) incorrectly believed that most women in their villages smoked cigarettes frequently. These misperceptions were pervasive across social strata. Men who misperceived frequent cigarette smoking as the norm among other men in their villages were more likely to smoke frequently themselves (adjusted relative risk=1.49; 95% CI, 1.13 to 1.97).ConclusionsMost adults overestimated cigarette smoking frequency among village peers. Men who incorrectly believed that frequent smoking was the norm were more likely to engage in frequent smoking themselves. Applying a ‘social norms approach’ intervention by promoting existing healthy norms may prevent smoking initiation or motivate reductions in smoking among men in rural Uganda.

Modifiable traits, healthy behaviours, and leucocyte telomere length

Background: Telomere length is associated with risk of several age-related diseases and cancers. The extent to which telomere length may be modifiable through lifestyle and behaviour and whether this has any clinical consequences is unknown. Methods: In up to 422,797 participants in UK Biobank, we investigated associations of leucocyte telomere length (LTL) with 117 potentially modifiable traits, as well as two indices of healthy behaviours incorporating smoking, physical activity, diet, maintenance of a healthy body weight and alcohol intake. Associations were interpreted as age-related change in LTL by dividing the trait beta coefficients with the age-coefficient. We used Mendelian Randomisation (MR) to test causality of the observed associations of educational attainment and smoking behaviour with LTL. We investigated whether the associations of LTL with 22 diseases were modified by the number of healthy behaviours and the extent to which the associations of more healthy behaviours with greater life expectancy and lower risk of coronary artery disease (CAD) may be mediated through LTL. Results: 71 traits showed significant associations with LTL but most were modest, equivalent to <1 year of age-related change in LTL. In multivariable analyses of 17 traits with stronger associations (equivalent to ≥2 years of age-related change in LTL), five traits (oily fish intake, educational attainment, general health status, walking pace and current smoking) remained significant. MR analysis suggested that educational attainment and smoking behaviour causally affect LTL. Both indices of healthy behaviour were positively and linearly associated with LTL, with those with the healthiest behaviour having longer LTL equivalent to approx. 3.5 years of age-related change in LTL when compared with those with the least heathy behaviours (P<0.001). However, healthy behaviours only explained <0.2% of the total variation in LTL and did not significantly modify the association of LTL with risk of any of the diseases studied. Neither the association of more healthy behaviours on greater life expectancy or lower risk of CAD were substantially mediated through LTL. Conclusions: Several potentially modifiable traits and healthy behaviours have a quantifiable association with LTL, at least some of which are likely to be causal. However, these effects are not of a sufficient magnitude to substantially alter the association between LTL and various diseases or life expectancy.

Genetic insights into smoking behaviours in 10,558 men of African ancestry from continental Africa and the UK

Smoking is a leading risk factor for many of the top ten causes of death worldwide. Of the 1 billion smokers globally, 80% live in low- and middle-income countries, where the number of deaths due to tobacco use is expected to double in the next decade according to the World Health Organization. Genetic studies have helped to identify biological pathways for smoking behaviours, but have mostly focussed on individuals of European ancestry or living in either North America or Europe.Here we present a genome-wide association study of two smoking behaviour traits in 10,558 men of African ancestry living in five African countries and the UK. Eight independent variants were associated with either smoking initiation or cessation at p-value < 5 × 10−6. Of these, four were monomorphic or rare in European populations. Gene prioritization strategy highlighted five genes, including SEMA6D, previously described as associated with several smoking behaviour traits. These results confirm the importance of genetic epidemiological studies in underrepresented populations.