INTRODUCTION:First-hand smoking is a major cause of global morbidity and mortality. Exposure to environmental tobacco smoke (ETS; "second-hand" or "passive smoking") may also cause ill health, but establishing ETS as the cause is challenging, in part due to confounding and reverse causation. METHODS:We applied Mendelian randomization (MR) to investigate the causal effects of ETS. We use four approaches to instrument ETS exposure: The first and second used an index individual's parent's genetically predicted smoking, independent of the index individual's genetically predicted smoking to assess the effects of that parent's smoking on the index individual. The third and fourth used one index individual's parent's genetically predicted smoking, independent of the other parent's genetically predicted smoking to assess the effects of the first parent's smoking on the second parent. We then meta-analyze the four MR approaches. RESULTS:Our findings suggest a causal effect of genetically predicted ETS exposure on lung cancer and chronic obstructive pulmonary disease (PFDR < .001 for both). We did not find evidence supporting an effect on hypertension, depression, coronary heart disease, or stroke (PFDR = 1.000 for all four non-respiratory outcomes). CONCLUSION:These results support existing public health measures to limit exposure to ETS. IMPLICATIONS:We assess the causal effects of environmental tobacco smoking (ETS; "second-hand smoking" or "passive smoking") using a quasi-experimental method, Mendelian randomization, which is more robust to confounding than conventional epidemiological methods.To study the effects of ETS exposure, we used an index individual's parent's or spouses' genetically predicted smoking, independent of either the index's or the other parent's genetically predicted smoking when assessing the effect of that parent's smoking on the index individual or other parent, respectively. We then meta-analyze the effects of different relatives.This study extends the Mendelian randomization paradigm to assess ETS by examining the effect that one relative has on the other relative, independent of the other relative's smoking. In doing so, it adds a unique source of evidence that triangulates with prior research to indicate an effect of ETS exposure on lung cancer and chronic obstructive pulmonary disease.

Responsible conduct of research (RCR) is imperative for the quality and trustworthiness of research, and the proper functioning of the whole research system. However, the extent to which RCR differs across disciplines is not clear. Currently, many approaches to research and training in RCR are either presented as universally applicable (e.g., international frameworks on research integrity) or discipline-specific. This Delphi study aimed to expand the current (underspecified) frameworks of RCR to develop a more diverse and comprehensive concept of what constitutes RCR across disciplines. Relying on the expertise and knowledge of a carefully selected multidisciplinary panel of RCR scholars and practitioners, we conducted a modified reactive Delphi, in which panellists progressively refined their judgement of the importance of individual dimensions of RCR for their respective disciplines, starting from a provisional list of dimensions derived from previous literature and interviews. In total, 56 panellists from a wide variety of research disciplines took part. All dimensions were rated as important, although they varied substantially both in perceived level of importance and in the degree of dispersion of responses. We classified the dimensions into three broad groups. The first comprised a core set of dimensions (18/38) rated as clearly important with relatively low dissent; the second set (9/38) included dimensions whose importance was generally endorsed but accompanied by nontrivial dissent, often due to perceived lack of applicability in some fields; and the third set (11/38) consisted of more contested dimensions characterised by complex or polarised rating patterns. Exploratory comparisons across four broad disciplinary groupings suggested that panellists in, on average across dimensions, Medicine, Health, &amp; Life Sciences and in the Social Sciences assigned higher overall importance ratings than those in Arts &amp; Humanities and in Physical Sciences, Engineering, &amp; Mathematics. In addition, qualitative and quantitative evidence pointed to substantial heterogeneity or ratings within disciplines. Taken together, these findings indicate that although a shared core set of RCR dimensions can be identified, many aspects of responsible research practice are interpreted and prioritised in discipline- and approach-dependent ways, even susceptible to variation among experts in the same discipline. These findings offer an empirical basis for approaching RCR as a shared endeavour with multiple recognised expressions and provide a concrete mapping of where common principles hold and where discipline- and context-specific interpretation makes a greater difference.

BACKGROUND:Early adversities before and after birth can impact children's cognitive and socioemotional development by altering critical brain maturational and functional processes. Some of these processes may be linked to later neurodevelopmental and/or mental health conditions. While most research is conducted in high-income countries, the majority of children live in low- and middle-income countries (LMICs) where they are more frequently exposed to poverty-related adversities. To address this gap, well-characterised longitudinal pregnancy cohorts in LMICs are needed to track trajectories of neurodevelopmental and mental health conditions in children exposed to cumulative environmental adversities. The Safe Passage Study (SPS) originally enrolled 7060 pregnant women from socioeconomically disadvantaged peri-urban communities in Cape Town, South Africa, to investigate the association between prenatal alcohol, multiple environmental risk factors and pregnancy outcome. The Safe Passage-Biomarkers of Neurodevelopmental Outcomes (BONO) study aims to follow up 2000 SPS children, aged 4-16 years, to assess the role of pre- and postnatal environmental factors in cognitive, neurodevelopmental and mental health outcomes. This report outlines the design and results of a feasibility study with 100 children, primarily aimed to confirm recruitment, assess participant retention, select measures and establish criteria for a deep-phenotyping visit. METHODS:Between March and October 2019, 100 SPS children were screened during a "broad-phenotyping visit" for adverse childhood experiences and protective factors, autistic traits and socioemotional and behavioural symptoms, and they completed cognitive tests and eye-tracking and electroencephalography assessments to measure brain function. Criteria for a second "deep-phenotyping" visit were established based on autistic traits, internalising/externalising scores and/or cognitive difficulties, to assess children and their mothers in terms of clinical and neurocognitive profile. RESULTS:Recruitment was adequate with a 96% retention rate for the deep-phenotyping visit. Feasibility study participants resembled the larger SPS cohort in most demographic and prenatal factors, except for higher prenatal depression and overcrowding indices. Most clinical and experimental measures were deemed suitable with minor modifications, and acquisition rates were high. The nature and length of visits were acceptable to families and testers. Threshold scores were adjusted to include 30% of participants for deep phenotyping. CONCLUSIONS:The feasibility study fulfilled progression criteria for the planned multimodal study.

Introduction Neural cue reactivity is increasingly being investigated as a biomarker of treatment response and relapse prediction in addiction disorders. While aberrant brain responses to salient cues (e.g., drugs) have been widely reported in addiction, it is unclear whether these brain responses persist during longer-term abstinence and how they compare between substance use disorder and obesity and relate to potential differences in eating behaviors. As part of the Gut Hormones in ADDiction (GHADD) neuroimaging study, we investigated how salient cue reactivity to drugs or food, craving and eating behaviors compare in three clinical populations where alterations have been previously observed: abstinent nicotine use disorder (NUD), alcohol use disorder (AUD), and obesity. Methods This study compared group differences in salient cue reactivity and eating behaviors among ex-smokers ( n = 25, ExS), adults with alcohol dependence but who are abstinent ( n = 26, AAD), and adults with obesity who were actively dieting ( n = 26, OB). Participants completed a high-energy food, preferred alcohol, and cigarette functional magnetic resonance imaging (fMRI) cue reactivity task, along with eating behavior questionnaires, appetite visual analogues scales, and an ad libitum test meal. Results ExS exhibited greater blood oxygen level dependent (BOLD) signal to high-energy food pictures in several reward processing regions in both whole brain and region of interest (ROI) analyses, compared with the OB and AAD groups, with no difference in their appeal rating. Compared with the OB group, ExS exhibited greater BOLD signal to cigarette pictures in the frontal gyrus, orbitofrontal cortex, frontal pole, and insula, with no difference in their appeal rating. There were no group differences in preferred alcohol cue reactivity. The AAD group rated sweet taste as more pleasant and consumed more calories from sweet dishes in the ad libitum meal than the OB and ExS groups. Conclusions The presence of heightened cue reactivity to high-energy foods in ex-smokers could contribute to post-quitting weight gain after smoking cessation. Neuroimaging findings were consistent with the persistence of some salient drug cue reactivity, despite absence of craving, after medium term abstinence in ExS, but not in AAD. This study also adds to the body of evidence supporting a sweet taste preference endophenotype predisposing individuals to AUD. These changes in eating behavior in NUD and AUD may provide targets for treatments to reduce substance misuse and facilitate abstinence.

Background: Research suggests that a greater perception of hostility in social cues increases aggression, and alcohol influences perception of social cues. Taken together, this could explain some instances of alcohol-related aggression. This study investigated whether social drinkers interpret faces as more hostile following acute alcohol compared to placebo, and whether alcohol influences the tendency to approach or avoid emotional facial expressions.Methods: Regular non-dependent drinkers (N = 84) participated in a double-blind placebo-controlled experiment. Participants completed two sessions and were tested following an alcoholic drink (0.4 g/kg), and matched placebo. In each session, they completed tasks measuring hostile attribution bias (HAB) towards emotional faces (happy, sad, angry, disgust, surprise, and fear), and approach/avoidance tendencies towards emotional faces (angry, happy, sad and disgust).Results: Alcohol did not affect global hostility ratings of emotional facial expression (p = 0.342). However, it did increase global hostility ratings of ambiguous emotional faces after alcohol (drink by intensity interaction; p = 0.002). At an emotion-specific level, happy faces were seen as more hostile after alcohol when compared to placebo (p = 0.009; irrespective of emotional intensity). Alcohol did not affect approach/avoidance tendencies when seeing emotional faces following alcohol.Conclusions: These findings suggest that alcohol increases hostile judgements of ambiguous emotional faces. They also suggest that happy faces are perceived to be more hostile following alcohol. As an increased HAB when processing socially relevant information increases aggressive responding, this increased hostile perception of happy faces following alcohol may increase the likelihood of aggressive behaviour.

Loneliness and social isolation are important public health concerns due to their associations with a range of health outcomes. However, it is difficult to ascertain whether any effects are biased by confounding and reverse causation. We use a triangulation approach combining observational, sibling control, and Mendelian Randomisation analyses (a genetically informed analysis), to draw robust conclusions about these relationships. Using a combination of UK Biobank data (N = 8,004 to 414,432) and genome-wide association studies (N = 17,526 to 2,083,151), we examine relationships between loneliness and social isolation and outcomes related to physical health, mental health and wellbeing and general health (e.g., multimorbidity capturing both mental and physical health). We find evidence for effects of loneliness and social isolation on poorer mental health and wellbeing and of loneliness on poorer general health. We do not find evidence of effects on specific physical health outcomes; however, these effects cannot definitively be ruled out.

Improvements to scientific methods and approaches should, in theory, mean more robust evidence and inference, and more rapid advances in knowledge. In practice, they have led to an increasing number of poor-quality studies. Can we break this cycle?

BACKGROUND:Digital devices have become a major aspect of children's lives. Associations between screen time and mental health have been observed, but the causality remains unclear. This study aimed to investigate the associations between screen time and later depressive symptoms, and to test the robustness of these associations when accounting for genetic confounding. METHODS:This study used data from the Avon Longitudinal Study of Parents and Children-a prospective cohort of children born between 1991 and 1992 in the UK. Different forms of screen time and depressive symptoms at ages 16, 22, and 26 years were assessed through self-completion questionnaires. The average daily screen time was calculated. Depressive symptoms were measured by using the Short Mood and Feelings Questionnaire (SMFQ). Polygenic scores (PGSs) for depression were calculated. Linear regression models were used to examine the associations between standardized screen time at ages 16, 22, and 26 years, and standardized depressive symptoms at age 26 years, adjusting for sociodemographic confounders and PGSs. Genetic sensitivity analysis (Gsens) was used to test for genetic confounding in these associations. RESULTS:A total of 3003 participants were included in the analysis. Some, but not all, forms of screen time were associated with higher SMFQ scores, e.g. time spent using a phone, tablet, or e-book at age 22 years [β: 0.10, 95% confidence interval (CI) 0.07, 0.14 for weekdays; β: 0.08, 95% CI 0.04, 0.11 for weekends] and television time at age 26 years (β: 0.10, 95% CI 0.06, 0.14 for weekdays; β: 0.09, 95% CI 0.06, 0.13 for weekends). These associations persisted after adjustment for sociodemographic confounders and PGSs, but were attenuated in the Gsens (β = 0.03, 95% CI -0.01, 0.07 for the association with time spent using a phone, tablet, or e-book at age 22 years on weekends; β = 0.06, 95% CI 0.01, 0.10 for television time at age 26 years on weekends). CONCLUSION:For some measures of screen time, there were no associations with depressive symptoms. Where associations were seen, they were attenuated in the Gsens, implying that genetic confounding is present in the relationship between screen time and depressive symptoms in adolescents and young adults.

Aims: To understand the motives, goals, and strategies of attempts to reduce alcohol among people who drink at risky levels in the United Kingdom (UK). Methods: We ran six online focus groups with people in the UK who drink at risky levels (Alcohol Use Disorders Identification Test [AUDIT] score ≥ 8; n = 26). We asked about participants’ reasons for making a reduction attempt or not, any goals they had, and any strategies used or considered. Transcripts were analysed using a codebook thematic analysis approach. Findings: Barriers to making an alcohol reduction attempt included participants’ knowledge (e.g., not perceiving their drinking as harmful), motives to drink alcohol (e.g., coping, enhancement), and the physical (e.g., accessibility of alcohol) and social (e.g., cultural expectations) environment. Facilitators included their knowledge (e.g., realising consumption levels), motives to drink less alcohol (e.g., health-related), and social support. Goals included complete abstinence, reduction in units, improved mental health, and improved relationships. Strategies included changing drinking contexts or the behaviour within them (e.g., not drinking at home), abstinence-based strategies, and tracking units. Conclusions: People drinking at risky levels in the UK reported a range of motives for drinking alcohol and drinking less alcohol, goals and strategies. These goals were largely inconsistent with the UK low risk drinking guidelines. Future research should assess whether alcohol reduction interventions that incorporate goals relevant to the individual, with tailored strategies based on the individual’s motives and goals for making a reduction attempt, are more effective than non-tailored strategies.

Objective Accumulating research conducted in high-income countries has reported that early life environmental factors (ELF) are linked to emotional-behavioral problems. However, approximately 90% of the world’s children live in low- and middle-income countries, where research remains limited. We investigated the prospective associations between ELF and emotional-behavioral problems in South African children and adolescents. Method Data were drawn from the Safe Passage Study, a population-derived birth cohort (5,889 mother-child dyads). ELF (pre-, peri-, and postnatal) were collected by questionnaires and clinical assessments and medical records at inception. In this Biomarkers of Neurodevelopmental Outcomes (BONO) follow-up, emotional-behavioral problems (externalizing and internalizing problems) were measured once using caregiver-reported Strengths and Difficulties Questionnaire at age 5-16 years (n=1,284). Associations were examined using regression models adjusted for the child’s sex and age at assessment. Results Of the 104 ELF, prenatal factors associated with both increased externalizing and internalizing problems were maternal mental health i.e., self-harm (externalizing problems: beta coefficient (β)=0.69; internalizing problems: β=0.60), depression (externalizing problems: β=0.06, internalizing problems: β=0.08) and anxiety (externalizing problems: β=0.05; internalizing problems: β=0.04) and infections (externalizing problems: β=0.65; internalizing problems: β=0.45). Factors only associated with externalizing problems included prenatal maternal substance use, asthma, paternal education, and infant secondhand smoke exposure. Enlarged placenta was associated with internalizing problems. Fetal, birth, and other placental factors showed modest associations. Conclusion In this South African setting, prenatal maternal mental health and infections emerged as consistent risk factors for both externalizing and internalizing problems. Future studies are needed to elucidate the impact of co-occurring ELF and investigate the mechanisms through which these ELF contribute to the emergence of emotional-behavioral problems.

Existing regulation in the UK states that the term ‘milk’ can only be used in labelling to describe products that originate from animals. We conducted an observational study, which surveyed the availability and labelling of milk substitutes in UK supermarkets, and an online experimental study, which assessed the impact of using the term ‘milk’ on milk substitute labelling. In the experimental study, 352 UK adults were randomised to one of the two conditions where they saw milk substitutes that were either labelled with UK regulations (e.g., soya drink) or using the term ‘milk’ (e.g., soya milk). Our primary aims were to assess whether adding the term ‘milk’ to labels would (1) more accurately communicate the uses of milk substitutes or (2) confuse consumers about which products come from an animal source. In our observational study, milk substitutes were readily available and labelling varied significantly. In our experimental study, labelling products with the term ‘milk’ increased understanding of the product's use. However, participants who saw the term ‘milk’ on milk substitute labelling misidentified more milk substitutes as coming from an animal source. Future policy should consider the clarification of such labelling.

The UK Reproducibility Network (UKRN) was established to promote robust and transparent research and drive research quality. Here, Marcus Munafò, a founding member of the UKRN, outlines the role of Reproducibility Networks as methods and technology evolves.

Abstract Background The UK Government has introduced a range of measures to tackle the rise in youth vaping, including the power to regulate flavours. Evidence is needed to inform how these measures should be implemented, but communication between health researchers and policymakers can be difficult. Policymakers need a simple, easy-to-digest synthesis of evidence to make policy decisions that benefit the public. Methods We designed a policy decision aid to support policymakers (commissioned by Public Health England), whereby existing data can be inputted to create a report (the decision aid). The report estimates whether restricting flavours other than tobacco or menthol flavours in e-cigarettes would have a net benefit due to reductions in youth vaping and smoking or a net detriment due to the negative impact on smoking cessation and relapse rates. Results Using the available evidence on 13 November 2024, the model estimated that 125,034 non-smoking youth experiment with e-cigarettes as a result of flavoured e-liquid availability, and 841,302 smokers and ex-smokers do not smoke due to flavoured e-liquid availability. The model estimated that 48,764 non-smoking youth subsequently smoke as a result of flavoured e-liquid availability. The algorithm indicated that if only unflavoured, tobacco flavoured or menthol flavoured vapes remained on the market, there would be a detrimental impact on smoking rates in adults that would outweigh the number of young people protected from vaping and later smoking. Conclusions This output suggests that restricting flavoured e-liquids in the United Kingdom could have a detrimental impact on public health when considering both youth vaping and smoking uptake due to flavoured vape availability. To provide timely advice to policymakers, the algorithm used is intentionally simple. The decision aid should be used alongside existing relevant evidence (e.g. from countries with similar regulatory environments) to inform policy decisions and/or used to highlight areas for research that warrant support. The reports have been used in policy documents and discussions, demonstrating an appetite for this type of communication aid. The tool can be used to assess this policy question within subpopulations and in other localities, and the framework can be adopted to address other policy questions.

BACKGROUND:Pre- and postpartum environments and genetic effects influence childhood internalising problems, which increase depression risk. DNA methylation (DNAm) may capture some of these effects. We therefore investigated associations between child blood DNAm and internalising problems. METHODS:We meta-analysed probe and region-level epigenome-wide association studies using data from 3 European birth cohorts (ALSPAC, MoBa, Generation R; analytic N = 1,121-3,011) from the Pregnancy And Childhood Epigenetics (PACE) Consortium. DNAm was assessed at birth (cord blood) and age 6 years (peripheral blood). Internalising problems (ages 3 and 6) were reported by mothers using the Child Behaviour Checklist or Strengths and Difficulties Questionnaire. Models were adjusted for age at DNAm assessment, 20 surrogate variables, estimated cell proportions, maternal education, age, smoking and, in secondary analysis, maternal anxiety and depression. Public databases were searched for mental health associations with top CpG sites and regions. RESULTS:No significant probe-level associations were found between cord- or peripheral-blood DNAm and internalising problems. In region-level analyses, two differentially methylated regions (DMRs) in cord blood were associated with internalising problems at age 3 (annotated to STK32C, MIR886) and one at age 6 (PFKFB2). At age 6, peripheral-blood analyses identified two DMRs (C10orf26 (WBP1L), FAM125A). Several genes showed prior associations with psychiatric phenotypes, including depression. CONCLUSIONS:The higher-powered regional-level analyses revealed more associations than probe-level. Among others, we identified a region annotated to STK32C that has previously been linked to adolescent depression. Larger samples and refined phenotyping are needed to clarify the role of DNAm in internalising problems.

BACKGROUND AND AIMS:Recent large studies have established the genetic basis of several conceptually linked phenotypes of externalizing. Polygenic risk scores (PRSs) for these constructs are associated with a range of substance use and mental disorder phenotypes but have not been examined with both pharmacological and non-pharmacological addictive behaviors, or across a developmental window. This study identified biological pathways responsible for observed associations between PRSs and addiction phenotypes. DESIGN, SETTING, PARTICIPANTS:We selected genome-wide association studies of 22 phenotypes, including substance use, general factors of externalizing and addiction, impulsivity and psychiatric conditions. Using summary statistics, we constructed PRSs in the offspring from the Avon Longitudinal Study of Parents and Children (ALSPAC) (nmax = 4995). Participants were genetically confirmed to be unrelated and of European-like genetic similarity. MEASUREMENTS:We examined the associations between PRSs and addiction-related phenotypes including substance use, gambling, eating behaviors and internet use across different life stages, from adolescence to young adulthood. PRSs were partitioned by biological pathways to examine the common and unique mechanisms underlying the genetics of addiction-related phenotypes. FINDINGS:The PRS of externalizing factor (PRSEXT) showed the strongest association across phenotypes for substance use (minP = 2.6 × 10-31, adjusted R2 = 0.10-4.72%), gambling (minP = 1.0 × 10-9, adjusted R2 = 0.18-1.50%), eating behaviors (minP = 8.2 × 10-4, adjusted R2 = 0.11-0.65%) and internet use (minP = 1.4 × 10-7, adjusted R2 = 0.17-1.04%). Sensitivity analyses excluding a small subset of ALSPAC participants who also contributed to the externalizing summary statistics, yielded consistent association effect sizes (R2 = 0.98), suggesting minimal bias. The results also revealed several time-varying associations between several PRSs and addiction phenotypes. Notably, the genetic influence of externalizing factor on alcohol and tobacco use was significantly stronger at younger ages. Finally, we identified multiple biological pathways that contribute to the link between addiction-related phenotypes and PRSEXT, emphasizing the importance of synaptic functions and neuronal plasticity in the context of gambling and substance use. CONCLUSIONS:There appears to be genetic evidence implicating externalizing as a common mechanism of substance and behavioral addictive behaviors. These results support the shared genetic liability across substance misuse, problematic gambling and internet use, and demonstrate the potential utility of externalizing traits as a transdiagnostic dimension across diverse forms of psychopathology. Notably, the predictive power of externalizing genetic liability appears developmentally dynamic, supporting the view that externalizing represents a broad, non-time-invariant risk factor that may give way to more specific disorder-related influences over time.

Previous studies suggest that smoking and higher alcohol consumption are associated with greater type 2 diabetes (T2D) risk. However, studies examining whether this reflects causal relationships are limited and often do not consider continuous glycaemic traits. We conducted both two-sample and one-sample Mendelian randomisation (MR), using publicly available GWAS data and UK Biobank data, respectively, to examine the potential causal effects of lifetime smoking index (LSI) and alcoholic drinks per week (DPW) on T2D and continuous traits (fasting glucose, fasting insulin and glycated haemoglobin, HbA1c). Two-sample MR results suggested possible causal effects of higher LSI on T2D risk (OR per 1SD higher LSI: 1.42, 95% CI 1.22 to 1.64); however, sensitivity analyses did not consistently support this finding. There was no robust evidence that higher DPW influenced T2D risk (OR per 1 SD higher log-transformed DPW: 1.04, 95% CI 0.40 to 2.65). There was evidence of a potential causal effect on higher fasting glucose (difference in mean fasting glucose in mmol/l per 1SD higher log-transformed DPW: 0.34, 95% CI 0.09 to 0.59), though, this was attenuated when accounting for body mass index (BMI), suggesting BMI confounding might explain the potential effect. One-sample MR results suggested a possible causal effect of higher DPW on T2D risk (OR per 1 SD higher log-transformed DPW: 1.71, 95% CI 1.24 to 2.36), but lower HbA1c levels (difference in mean SD of log transformed HbA1c (mmol/mol) per 1 SD higher log-transformed DPW: −0.07, 95% CI −0.11 to −0.02). Our results suggest effective public health interventions to prevent and/or reduce smoking and alcohol consumption are unlikely to reduce T2D prevalence.

Parental smoking increases offspring smoking risk. Young people with mental health difficulties are more likely to have parents who smoke and start smoking themselves. Improving mental health in youth could help disrupt intergenerational cycles of nicotine use, but the extent to which depressive symptoms mediate the link between parental and offspring smoking remains unclear. We used data on 6,729 participants of the Avon Longitudinal Study of Parents and Children (ALSPAC) with mother-reported data on parental smoking. Counterfactual-based mediation analysis was used to estimate both total and direct effects of parental smoking at offspring age 12 on offspring smoking at age 16 and estimate the indirect effect via offspring depressive symptoms at age 14. We also accounted for early offspring substance use (cigarettes, alcohol, or cannabis) at age 13. Multivariable logistic regression was used to examine pairwise associations between these variables. Mediation analysis showed a total effect of parental smoking on offspring smoking (odds ratio [OR] = 1.44, 95% CI = 1.20-1.72). There was no indirect effect via depressive symptoms (OR = 1.00, 95% CI = 0.99-1.01) when adjusted for confounding. Regression analyses showed parental smoking was associated with early offspring substance use and later offspring smoking but not offspring depression. Early offspring substance use was associated with both offspring depression and later smoking. Higher depressive symptoms were associated with later smoking. Parental smoking in early adolescence increases the odds of offspring smoking in later adolescence, but this does not appear to act through offspring depression.

The alcohol harm paradox, whereby low socioeconomic position (SEP) groups experience greater alcohol-related harms at a given level of alcohol consumption, is not yet fully understood. In observational studies, key drivers are correlated and share similar confounding structures. We used multivariable Mendelian randomization (MVMR) to estimate the direct causal effect of alcohol (drinks per week) and education (years of schooling) on multiple health outcomes, accounting for the effect of the other. Previously published genome-wide association summary (GWAS) statistics for drinks per week and years of schooling were used, and outcome summary statistics were generated from individual-level data from UK Biobank (N = 462,818). Inverse variance weighted analyses demonstrated evidence for direct effects of alcohol and education on liver diseases (alcoholic liver disease: alcohol OR = 50.19, 95% CI 19.35 to 130.21 and education OR = 0.27, 95% CI 0.14 to 0.53; other liver diseases: alcohol OR = 1.82, 95% CI 1.12 to 2.94 and education OR = 0.42, 95% CI 0.30 to 0.58), mental and behavioural disorders due to alcohol (alcohol OR = 12.89, 95% CI 7.46 to 22.27 and education OR = 0.51, 95% CI 0.35 to 0.75), and stroke (alcohol OR = 1.94, 95% CI 1.30 to 2.89 and education OR = 0.73, 95% CI 0.55 to 0.97). There was evidence for direct effects of education on depression, anxiety, influenza/pneumonia, and heart disease. In contrast, there was evidence of total (without considering the effect of education), but not direct, effects of alcohol on depression, influenza/pneumonia, epilepsy, and injuries. Although caution is required when interpreting these results, given weak instruments for alcohol, these results provide some evidence that the alcohol harm paradox is partially due to the protective effect of additional years of education. Replication with strong genetic instruments for drinks per week would be necessary to draw causal inferences.