Search

BackgroundInflammation is implicated in depression, but the issue of causality remains unclear.ObjectivesTo investigate potential causality and direction of effect between inflammation and depression.MethodsUsing data from the ALSPAC birth cohort (n=4021), we used multivariable regression to investigate bidirectional longitudinal associations of GlycA and depression symptoms score and diagnosis, assessed at ages 18y and 24y.We used two-sample Mendelian randomization (MR) to investigate potential causality and directionality. Genetic variants for GlycA were obtained from UK Biobank (UKBB) (N=115,078); for depression from the Psychiatric Genomics Consortium and UKBB (N=500,199); and for depressive symptoms (N=161,460) from the Social Science Genetic Association Consortium. In addition to the Inverse Variance Weighted (IVW) method, we used sensitivity analyses to strengthen causal inference. We conducted multivariable MR adjusting for body mass index (BMI) due to known genetic correlation between inflammation, depression and BMI.ResultsAfter adjusting for potential confounders we found no association between GlycA and depression symptoms score orvice versa. We observed an association between GlycA and depression diagnosis (OR=1.18, 95% CI: 1.03-1.36).MR suggested no causal effect of GlycA on depression, but there was evidence of a causal effect of depression on GlycA (mean difference in GlycA = 0.09; 95% CI: 0.03-0.16), which was maintained in some, but not all, sensitivity analyses.ConclusionWe found no consistent evidence for an effect of the inflammatory marker GlycA on depression. There was some evidence that depression may increase GlycA, but this may be confounded/mediated by BMI.

Background Non-random selection of analytic subsamples could introduce selection bias in observational studies. We explored the potential presence and impact of selection in studies of SARS-CoV-2 infection and COVID-19 prognosis. Methods We tested the association of a broad range of characteristics with selection into COVID-19 analytic subsamples in the Avon Longitudinal Study of Parents and Children (ALSPAC) and UK Biobank (UKB). We then conducted empirical analyses and simulations to explore the potential presence, direction and magnitude of bias due to this selection (relative to our defined UK-based adult target populations) when estimating the association of body mass index (BMI) with SARS-CoV-2 infection and death-with-COVID-19. Results In both cohorts, a broad range of characteristics was related to selection, sometimes in opposite directions (e.g. more-educated people were more likely to have data on SARS-CoV-2 infection in ALSPAC, but less likely in UKB). Higher BMI was associated with higher odds of SARS-CoV-2 infection and death-with-COVID-19. We found non-negligible bias in many simulated scenarios. Conclusions Analyses using COVID-19 self-reported or national registry data may be biased due to selection. The magnitude and direction of this bias depend on the outcome definition, the true effect of the risk factor and the assumed selection mechanism; these are likely to differ between studies with different target populations. Bias due to sample selection is a key concern in COVID-19 research based on national registry data, especially as countries end free mass testing. The framework we have used can be applied by other researchers assessing the extent to which their results may be biased for their research question of interest.

ObjectivesMultimorbidity, typically defined as having two or more long-term health conditions, is a common patient characteristic that is associated with reduced wellbeing and life expectancy. Understanding the determinants of multimorbidity may help with the design and prioritisation of interventions to prevent multimorbidity. This study examined potential causal determinants (education, BMI, smoking and alcohol consumption) of multimorbidity, and assessed the extent to which BMI, smoking and alcohol consumption explain observed educational inequalities in multimorbidity.DesignMendelian randomization study; an approach that uses genetic variants as instrumental variables to interrogate causality.Participants181,214 females and 155,677 males, mean ages 56.7 and 57.1 years respectively, from UK Biobank.Main outcome measuresMultimorbidity status (2+ self-reported health conditions); secondary analyses considered complex multimorbidity defined as 3+ or 4+ conditions, and a continuous multimorbidity score.ResultsMendelian randomization suggests that lower education, higher BMI and higher levels of smoking causally increase the risk of multimorbidity. For example, one standard deviation (equivalent to 5.1 years) increase in years of education decreases the risk of multimorbidity by 9.0% (95% CI: 6.5 to 11.4%). A 5 kg/m2 increase in BMI is associated with a 9.2% increased risk of multimorbidity (95% CI: 8.1 to 10.3%) and a one SD higher lifetime smoking index is associated with a 6.8% increased risk of multimorbidity (95% CI: 3.3 to 10.4%). Evidence for a causal effect of alcohol consumption on multimorbidity was less strong; an increase of 5 units of alcohol per week increases the risk of multimorbidity (2+ conditions) by 1.3% (95% CI: 0.2 to 2.5%). The proportions of the association between education and multimorbidity explained by BMI and smoking are 20.4% and 17.6% respectively. Collectively, BMI and smoking account for 31.8% of the educational inequality in multimorbidity.ConclusionsEducation, BMI, smoking and alcohol consumption are intervenable risk factors that our results suggest have a causal effect on multimorbidity. Furthermore, BMI and lifetime smoking make a considerable contribution to the generation of educational inequalities in multimorbidity. Public health interventions that improve population-wide levels of these risk factors are likely to reduce multimorbidity and inequalities in its occurrence.SUMMARY BOXSECTION 1: WHAT IS ALREADY KNOWN ON THIS TOPIC-Multimorbidity has several known lifestyle and anthropometric risk factors and is associated with deprivation.-The effect of education (proxying deprivation) on multimorbidity is likely mediated by some of these intervenable risk factors.-These associations are likely to be confounded and their causality is not well understood.SECTION 2: WHAT THIS STUDY ADDS-Analyses using genetically predicted effects suggest that education, BMI, smoking and alcohol consumption each have a causal effect on multimorbidity and that 32% of the educational inequality in multimorbidity is attributable to BMI and smoking combined.

BackgroundStructural barriers to testing may introduce selection bias in COVID-19 research. We explore whether changes to testing and lockdown restrictions introduce time-specific selection bias into analyses of socioeconomic position (SEP) and SARS-CoV-2 infection.MethodsUsing UK Biobank (N = 420 231; 55 % female; mean age = 56·3 [SD=8·01]) we estimated the association between SEP and i) being tested for SARS-CoV-2 infection versus not being tested ii) testing positive for SARS-CoV-2 infection versus testing negative and iii) testing negative for SARS-CoV-2 infection versus not being tested, at four distinct time-periods between March 2020 and March 2021. We explored potential selection bias by examining the same associations with hypothesised positive (ABO blood type) and negative (hair colour) control exposures. Finally, we conducted a hypothesis-free phenome-wide association study to investigate how individual characteristics associated with testing changed over time.FindingsThe association between low SEP and SARS-CoV-2 testing attenuated across time-periods. Compared to individuals with a degree, individuals who left school with GCSEs or less had an OR of 1·05 (95% CI: 0·95 to 1·16) in March-May 2020 and 0·98 (95% CI: 0·94 to 1·02) in January-March 2021. The magnitude of the association between low SEP and testing positive for SARS-CoV-2 infection increased over the same time-period. For the same comparisons, the OR for testing positive increased from 1·27 (95% CI: 1·08 to 1·50), to 1·73 (95% CI: 1·59 to 1·87). We found little evidence of an association between both control exposures and all outcomes considered. Our phenome-wide analysis highlighted a broad range of individual traits were associated with testing, which were distinct across time-periods.InterpretationThe association between SEP (and indeed many individual traits) and SARS-CoV-2 testing changed over time, indicating time-specific selection pressures in COVID-19. However, positive, and negative control analyses suggest that changes in the magnitude of the association between SEP and SARS-CoV-2 infection over time were unlikely to be explained by selection bias and reflect true increases in socioeconomic inequalities.FundingUniversity of Bristol; UK Medical Research Council; British Heart Foundation; European Union Horizon 2020; Wellcome Trust and The Royal Society; National Institute of Health Research; UK Economic and Social Research Council

BackgroundA strong association between the proportion of Native American ancestry and the risk of gallbladder cancer (GBC) has been reported in observational studies. Chileans show the highest incidence of GBC worldwide, and the Mapuche are the largest Native American people in Chile. We set out to investigate the causal association between Native American Mapuche ancestry and GBC risk, and the possible mediating effects of gallstone disease and body mass index (BMI) on this association.MethodsMarkers of Mapuche ancestry were selected based on the informativeness for assignment measure and then used as instrumental variables in two-sample mendelian randomization (MR) analyses and complementary sensitivity analyses.ResultWe found evidence of a causal effect of Mapuche ancestry on GBC risk (inverse variance-weighted (IVW) risk increase of 0.8% for every 1% increase in Mapuche ancestry proportion, 95% CI 0.4% to 1.2%, p = 6.6×10-5). Mapuche ancestry was also causally linked to gallstone disease (IVW risk increase of 3.6% per 1% increase in Mapuche proportion, 95% CI 3.1% to 4.0%, p = 1.0×10-59), suggesting a mediating effect of gallstones in the relationship between Mapuche ancestry and GBC. In contrast, the proportion of Mapuche ancestry showed a negative causal effect on BMI (IVW estimate -0.006 kg/m2 per 1% increase in Mapuche proportion, 95% CI -0.009 to -0.003, p = 4.4×10-5).ConclusionsThe results presented here may have significant implications for GBC prevention and are important for future admixture mapping studies. Given that the association between Mapuche ancestry and GBC risk previously noted in observational studies appears to be causal, primary and secondary prevention strategies that take into account the individual proportion of Mapuche ancestry could be particularly efficient.

Dietary factors are assumed to play an important role in cancer risk, apparent in consensus recommendations for cancer prevention that promote nutritional changes. However, the evidence in this field has been generated predominantly through observational studies, which may result in biased effect estimates because of confounding, exposure misclassification, and reverse causality. With major geographical differences and rapid changes in cancer incidence over time, it is crucial to establish which of the observational associations reflect causality and to identify novel risk factors as these may be modified to prevent the onset of cancer and reduce its progression. Mendelian randomization (MR) uses the special properties of germline genetic variation to strengthen causal inference regarding potentially modifiable exposures and disease risk. MR can be implemented through instrumental variable (IV) analysis and, when robustly performed, is generally less prone to confounding, reverse causation and measurement error than conventional observational methods and has different sources of bias (discussed in detail below). It is increasingly used to facilitate causal inference in epidemiology and provides an opportunity to explore the effects of nutritional exposures on cancer incidence and progression in a cost-effective and timely manner. Here, we introduce the concept of MR and discuss its current application in understanding the impact of nutritional factors (e.g., any measure of diet and nutritional intake, circulating biomarkers, patterns, preference or behaviour) on cancer aetiology and, thus, opportunities for MR to contribute to the development of nutritional recommendations and policies for cancer prevention. We provide applied examples of MR studies examining the role of nutritional factors in cancer to illustrate how this method can be used to help prioritise or deprioritise the evaluation of specific nutritional factors as intervention targets in randomised controlled trials. We describe possible biases when using MR, and methodological developments aimed at investigating and potentially overcoming these biases when present. Lastly, we consider the use of MR in identifying causally relevant nutritional risk factors for various cancers in different regions across the world, given notable geographical differences in some cancers. We also discuss how MR results could be translated into further research and policy. We conclude that findings from MR studies, which corroborate those from other well-conducted studies with different and orthogonal biases, are poised to substantially improve our understanding of nutritional influences on cancer. For such corroboration, there is a requirement for an interdisciplinary and collaborative approach to investigate risk factors for cancer incidence and progression.

ObjectiveIdentify whether participants with lower education are less likely to report taking statins for primary cardiovascular prevention than those with higher education, but an equivalent increase in underlying cardiovascular risk.MethodsUsing data from a large prospective cohort study, UK Biobank, we calculated a QRISK3 cardiovascular risk score for 472 097 eligible participants with complete data on self-reported educational attainment and statin use (55% female participants; mean age 56 years). We used logistic regression to explore the association between (i) QRISK3 score and (ii) educational attainment on self-reported statin use. We then stratified the association between QRISK3 score and statin use, by educational attainment to test for interactions.ResultsThere was evidence of an interaction between QRISK3 score and educational attainment. Per unit increase in QRISK3 score, more educated individuals were more likely to report taking statins. In women with ≤7 years of schooling, a one unit increase in QRISK3 score was associated with a 7% higher odds of statin use (OR 1.07, 95% CI 1.07 to 1.07). In women with ≥20 years of schooling, a one unit increase in QRISK3 score was associated with an 14% higher odds of statin use (OR 1.14, 95% CI 1.14 to 1.15). Comparable ORs in men were 1.04 (95% CI 1.04 to 1.05) for ≤7 years of schooling and 1.08 (95% CI 1.08, 1.08) for ≥20 years of schooling.ConclusionPer unit increase in QRISK3 score, individuals with lower educational attainment were less likely to report using statins, likely contributing to health inequalities.

ObjectiveTo use the example of the effect of body mass index (BMI) on COVID-19 susceptibility and severity to illustrate methods to explore potential selection and misclassification bias in Mendelian randomisation (MR) of COVID-19 determinants.DesignTwo-sample MR analysis.SettingSummary statistics from the Genetic Investigation of ANthropometric Traits (GIANT) and COVID-19 Host Genetics Initiative (HGI) consortia.Participants681,275 participants in GIANT and more than 2.5 million people from the COVID-19 HGI consortia.ExposureGenetically instrumented BMI.Main outcome measuresSeven case/control definitions for SARS-CoV-2 infection and COVID-19 severity: very severe respiratory confirmed COVID-19 vs not hospitalised COVID-19 (A1) and vs population (those who were never tested, tested negative or had unknown testing status (A2)); hospitalised COVID-19 vs not hospitalised COVID-19 (B1) and vs population (B2); COVID-19 vs lab/self-reported negative (C1) and vs population (C2); and predicted COVID-19 from self-reported symptoms vs predicted or self-reported non-COVID-19 (D1).ResultsWith the exception of A1 comparison, genetically higher BMI was associated with higher odds of COVID-19 in all comparison groups, with odds ratios (OR) ranging from 1.11 (95%CI: 0.94, 1.32) for D1 to 1.57 (95%CI: 1.57 (1.39, 1.78) for A2. As a method to assess selection bias, we found no strong evidence of an effect of COVID-19 on BMI in a ‘no-relevance’ analysis, in which COVID-19 was considered the exposure, although measured after BMI. We found evidence of genetic correlation between COVID-19 outcomes and potential predictors of selection determined a priori (smoking, education, and income), which could either indicate selection bias or a causal pathway to infection. Results from multivariable MR adjusting for these predictors of selection yielded similar results to the main analysis, suggesting the latter.ConclusionsWe have proposed a set of analyses for exploring potential selection and misclassification bias in MR studies of risk factors for SARS-CoV-2 infection and COVID-19 and demonstrated this with an illustrative example. Although selection by socioeconomic position and arelated traits is present, MR results are not substantially affected by selection/misclassification bias in our example. We recommend the methods we demonstrate, and provide detailed analytic code for their use, are used in MR studies assessing risk factors for COVID-19, and other MR studies where such biases are likely in the available data.SummaryWhat is already known on this topic-Mendelian randomisation (MR) studies have been conducted to investigate the potential causal relationship between body mass index (BMI) and COVID-19 susceptibility and severity.-There are several sources of selection (e.g. when only subgroups with specific characteristics are tested or respond to study questionnaires) and misclassification (e.g. those not tested are assumed not to have COVID-19) that could bias MR studies of risk factors for COVID-19.-Previous MR studies have not explored how selection and misclassification bias in the underlying genome-wide association studies could bias MR results.What this study adds-Using the most recent release of the COVID-19 Host Genetics Initiative data (with data up to June 2021), we demonstrate a potential causal effect of BMI on susceptibility to detected SARS-CoV-2 infection and on severe COVID-19 disease, and that these results are unlikely to be substantially biased due to selection and misclassification.-This conclusion is based on no evidence of an effect of COVID-19 on BMI (a ‘no-relevance control’ study, as BMI was measured before the COVID-19 pandemic) and finding genetic correlation between predictors of selection (e.g. socioeconomic position) and COVID-19 for which multivariable MR supported a role in causing susceptibility to infection.-We recommend studies use the set of analyses demonstrated here in future MR studies of COVID-19 risk factors, or other examples where selection bias is likely.

Aims/hypothesis Type 2 diabetes and atherosclerotic CVD share many risk factors. This study aimed to systematically assess a broad range of continuous traits to separate their direct effects on coronary and peripheral artery disease from those mediated by type 2 diabetes. Methods Our main analysis was a two-step Mendelian randomisation for mediation to quantify the extent to which the associations observed between continuous traits and liability to atherosclerotic CVD were mediated by liability to type 2 diabetes. To support this analysis, we performed several univariate Mendelian randomisation analyses to examine the associations between our continuous traits, liability to type 2 diabetes and liability to atherosclerotic CVD. Results Eight traits were eligible for the two-step Mendelian randomisation with liability to coronary artery disease as the outcome and we found similar direct and total effects in most cases. Exceptions included fasting insulin and hip circumference where the proportion mediated by liability to type 2 diabetes was estimated as 56% and 52%, respectively. Six traits were eligible for the analysis with liability to peripheral artery disease as the outcome. Again, we found limited evidence to support mediation by liability to type 2 diabetes for all traits apart from fasting insulin (proportion mediated: 70%). Conclusions/interpretation Most traits were found to affect liability to atherosclerotic CVD independently of their relationship with liability to type 2 diabetes. These traits are therefore important for understanding atherosclerotic CVD risk regardless of an individual’s liability to type 2 diabetes. Graphical abstract

Background Understanding the interplay between educational attainment and genetic predictors of cardiovascular risk may improve our understanding of the aetiology of educational inequalities in cardiovascular disease. Methods In up to 320 120 UK Biobank participants of White British ancestry (mean age = 57 years, female 54%), we created polygenic scores for nine cardiovascular risk factors or diseases: alcohol consumption, body mass index, low-density lipoprotein cholesterol, lifetime smoking behaviour, systolic blood pressure, atrial fibrillation, coronary heart disease, type 2 diabetes and stroke. We estimated whether educational attainment modified genetic susceptibility to these risk factors and diseases. Results On the additive scale, higher educational attainment reduced genetic susceptibility to higher body mass index, smoking, atrial fibrillation and type 2 diabetes, but increased genetic susceptibility to higher LDL-C and higher systolic blood pressure. On the multiplicative scale, there was evidence that higher educational attainment increased genetic susceptibility to atrial fibrillation and coronary heart disease, but little evidence of effect modification was found for all other traits considered. Conclusions Educational attainment modifies the genetic susceptibility to some cardiovascular risk factors and diseases. The direction of this effect was mixed across traits considered and differences in associations between the effect of the polygenic score across strata of educational attainment was uniformly small. Therefore, any effect modification by education of genetic susceptibility to cardiovascular risk factors or diseases is unlikely to substantially explain the development of inequalities in cardiovascular risk.

BackgroundNon-random selection into analytic subsamples could introduce selection bias in observational studies of SARS-CoV-2 infection and COVID-19 severity (e.g. including only those have had a COVID-19 PCR test). We explored the potential presence and impact of selection in such studies using data from self-report questionnaires and national registries.MethodsUsing pre-pandemic data from the Avon Longitudinal Study of Parents and Children (ALSPAC) (mean age=27.6 (standard deviation [SD]=0.5); 49% female) and UK Biobank (UKB) (mean age=56 (SD=8.1); 55% female) with data on SARS-CoV-2 infection and death-with-COVID-19 (UKB only), we investigated predictors of selection into COVID-19 analytic subsamples. We then conducted empirical analyses and simulations to explore the potential presence, direction, and magnitude of bias due to selection when estimating the association of body mass index (BMI) with SARS-CoV-2 infection and death-with-COVID-19.ResultsIn both ALSPAC and UKB a broad range of characteristics related to selection, sometimes in opposite directions. For example, more educated participants were more likely to have data on SARS-CoV-2 infection in ALSPAC, but less likely in UKB. We found bias in many simulated scenarios. For example, in one scenario based on UKB, we observed an expected odds ratio of 2.56 compared to a simulated true odds ratio of 3, per standard deviation higher BMI.ConclusionAnalyses using COVID-19 self-reported or national registry data may be biased due to selection. The magnitude and direction of this bias depends on the outcome definition, the true effect of the risk factor, and the assumed selection mechanism.Key messagesObservational studies assessing the association of risk factors with SARS-CoV-2 infection and COVID-19 severity may be biased due to non-random selection into the analytic sample.Researchers should carefully consider the extent that their results may be biased due to selection, and conduct sensitivity analyses and simulations to explore the robustness of their results. We provide code for these analyses that is applicable beyond COVID-19 research.

Background Education is inversely associated with cardiovascular disease (CVD). Several mediators of this have been established; however, a proportion of the protective effect remains unaccounted for. Mental health is a proposed mediator, but current evidence is mixed and subject to bias from confounding factors and reverse causation. Mendelian randomization is an instrumental variable technique that uses genetic proxies for exposures and mediators to reduce such bias. Methods and Results We performed logistic regression and 2‐step Mendelian randomization analyses using UK Biobank data and genetic summary statistics to investigate whether educational attainment affects risk of mental health disorders. We then performed mediation analyses to explore whether mental health disorders mediate the association between educational attainment and cardiovascular risk. Higher levels of educational attainment were associated with reduced depression, anxiety, and CVD in observational analyses (odds ratio [OR], 0.79 [95% CI, 0.77–0.81], 0.76 [95% CI, 0.73–0.79], and 0.75 [95% CI, 0.74–0.76], respectively), and Mendelian randomization analyses provided evidence of causality (OR, 0.72 [95% CI, 0.67–0.77], 0.50 [95% CI, 0.42–0.59], and 0.62 [95% CI, 0.58–0.66], respectively). Both anxiety and depression were associated with CVD in observational analyses (OR, 1.63 [95% CI, 1.49–1.79] and 1.70 [95% CI, 1.59–1.82], respectively) but only depression showed evidence of causality in the Mendelian randomization analyses (OR, 1.09; 95% CI, 1.03–1.15). An estimated 2% of the total protective effect of education on CVD was mediated by depression. Conclusions Higher levels of educational attainment protect against mental health disorders, and reduced depression accounts for a small proportion of the total protective effect of education on CVD.

Background Despite reductions in the rates of cardiovascular disease in high income countries, individuals who are the most socioeconomically deprived remain at the highest risk of disease. Although intermediate lifestyle and behavioural risk factors explain some of this, much of the effect remains unexplained. It is not known whether differences in risk adjusted use of statins between educational groups may contribute to these inequalities. Methods Using data from a large prospective cohort study, UK Biobank, we calculated a QRISK3 cardiovascular risk score for 472 097 eligible participants with complete data on self-reported educational attainment and statin use (55% female; mean age, 56). We used logistic regression to explore the association between i) QRISK3 score and ii) educational attainment on self-report statin use. We then stratified the association between QRISK3 score, and statin use by educational attainment to test for interactions. We then replicated analyses using QRISK or QRISK2 scores recorded in primary care data and statin prescriptions recorded in primary care prescription records. Results There was evidence of an interaction between QRISK3 scores and education. For an equivalent QRISK3 score, more educated individuals were more likely to report taking statins. In women with 7 years of schooling, a one unit increase in QRISK3 score was associated with a 7% higher odds of statin use (odds ratio (OR) 1.07, 95% CI 1.07, 1.07). In women with 20 years of schooling, a one unit increase in QRISK3 score was associated with an 14% higher odds of statin use (OR 1.14, 95% CI 1.14, 1.15). Comparable ORs in men were 1.04 (95% CI 1.04, 1.05) for 7 years of schooling and 1.08 (95% CI 1.08, 1.08) for 20 years of schooling. These inequalities were also present in analyses using primary care data. Conclusion For the same level of cardiovascular risk, individuals with lower educational attainment are less likely to receive statins, likely contributing to cardiovascular inequalities. The mechanisms leading to these differences are unknown, but both health seeking behaviours and clinical factors may contribute.

Rationale Associations between COVID-19 risk factors and COVID-19 outcomes change over time, likely due to selection into who receives a COVID-19 test. When studies do not account for the changes in testing criteria, the association between a risk factor and outcome is a joint estimate across time. The transportability of a joint estimate aggregated over multiple testing periods may be limited. To improve generalisability, it is desirable to estimate effects net of time-varying selection. Aim 1) Demonstrate variation in the association between covariates expected to associate with testing, and those which would not, on COVID-19 at different timepoints. 2) Apply methods to mitigate biases in empirical estimates. Methods Analyses will be carried out on up to 421,037 UK Biobank participants residing in England at baseline (mean age at baseline = 56; 55% female). Risk factors will be determined at baseline (from 2006 to 2010), and COVID-19 outcomes will be ascertained from linked Public Health England COVID-19 test data and mortality statistics.Univariate cox proportional hazard models will be used to explore how associations between time-varying and time-stable variables change over time with; i) having a test for COVID-19, ii) testing positive for COVID-19 and iii) dying with COVID-19.Time-varying risk factors will be based on measures of socioeconomic position (SEP) including education, Townsend deprivation index and income. ABO blood group will be considered as a time-stable risk factor. Distinct time periods will be defined based on changes in testing definitions and changes in lockdown restrictions.Inverse probability weights will then be calculated for each time period. These weights will then be applied to models estimating risk across all time periods. Expected Results Preliminary analyses show that the size of the association between SEP and i) COVID-19 testing and ii) testing positive for COVID-19, changes across the course of the pandemic. These differences may be due to differential testing and not time-varying causal effects of the risk factor. We expect inverse probability weights will provide estimates closer to the true value for the association between each risk factor and outcome, independent of selection pressures on receiving a COVID-19 test. Population Health Relevance. Where studies do not account for time-varying selection pressures, the causal interpretations and the validity of results may be distorted. Where these findings are to be translated into developing population level or pharmaceutical interventions to mitigate against COVID-19 outcomes, efforts may be diverted away from more important risk factors.

Background Non-random sampling could bias estimates of association in observational studies but the extent to which this occurs in studies of SARS-CoV-2 infection/COVID-19 severity is not well established. Methods Using ALSPAC and UK Biobank we investigated pre-pandemic predictors of selection (i.e. having data on SARS-CoV-2 infection and COVID-19 severity from self-report and/or health record linkage). We conducted empirical analyses and simulations to explore the potential presence, direction and magnitude of bias due to selection when estimating the association of BMI with SARS-CoV-2 infection and COVID-19 severity. Results A broad range of characteristics related to selection in both cohorts, sometimes in opposite directions. We found bias in all simulated scenarios, mostly of small magnitude. Both the direction and magnitude of bias was influenced by the presence of an effect of BMI on SARS-CoV-2 infection and COVID-19 severity and the control group definition used (e.g. assuming no effect of BMI on SARS-CoV-2 infection our main simulation showed bias equivalent to an estimated odds ratio of 0.99 when using non-infected controls but 1.16 when using controls combining non-infected and non-assessed). Conclusions Despite small amounts of bias in most scenarios, a control group definition including those non-assessed (e.g. non-tested) can induce more bias. In large samples such as UK Biobank the statistical power means incorrect conclusions could be made. Key messages Observational studies estimating associations of factors with SARS-CoV-2 or COVID-19 may be biased due to non-random selection into the analytic sample.

Background Mendelian randomisation uses genetic variants randomly allocated at conception as instrumental variables for an exposure. Methodological advances allow for mediation analysis to be carried out using Mendelian randomisation using either multivariable Mendelian randomisation or two-step Mendelian randomisation. Methods We use simulations and an applied example to demonstrate when multivariable Mendelian randomisation and two-step Mendelian randomisation methods are valid and how they relate to traditional phenotypic regression-based approaches to mediation. We demonstrate how Mendelian randomisation methods can relax assumptions required for causal inference in phenotypic mediation, as well as which Mendelian randomisation specific assumptions are required. We illustrate our methods in data from UK Biobank, estimating the role of body mass index mediating the association between education and cardiovascular outcomes. Results Both multivariable Mendelian randomization and two-step Mendelian randomization are unbiased when estimating the total effect, direct effect, indirect effect and proportion mediated when both confounding, and measurement error are present. Multivariable Mendelian Randomization can be used when multiple mediators are to be investigated in a single model. Conclusions Mendelian randomisation provides an opportunity to improve causal inference in mediation analysis. Although Mendelian randomisation specific assumptions apply, such as no weak instrument bias and no pleiotropic pathways, strong phenotypic assumptions of no confounding and no measurement error can be relaxed. Key messages Mendelian randomisation offers an opportunity to address bias by unmeasured confounding, measurement error and reverse causality in mediation analysis.

BackgroundType 2 diabetes and atherosclerotic cardiovascular disease share several risk factors. However, it is unclear whether the effect of these risk factors on liability to atherosclerotic cardiovascular disease is independent of their effect on liability to type 2 diabetes.MethodsWe performed univariate Mendelian randomization to quantify the effects of continuous risk factors from the IEU OpenGWAS database on liability to three outcomes: type 2 diabetes, coronary artery disease, and peripheral artery disease, as well as the effects of liability to type 2 diabetes on the risk factors. We also performed two-step Mendelian randomization for mediation to estimate the mediating pathways between the risk factors, liability to type 2 diabetes, and liability to the atherosclerotic cardiovascular disease outcomes where possible.ResultsWe found evidence for 53 risk factors as causes of liability to coronary artery disease, including eight which were causes of liability to type 2 diabetes only and four which were consequences only. Except for fasting insulin and hip circumference, the direct and total effects from the two-step Mendelian randomization were similar. This suggests that the combination of these risk factors with liability to type 2 diabetes was unlikely to alter liability to coronary artery disease beyond their individual effects. We also found 13 risk factors that were causes of liability peripheral artery disease, including six which were causes of liability to type 2 diabetes only and four which were consequences only. Again, the direct and total effects were similar for these ten risk factors apart from fasting insulin.ConclusionsMost risk factors were likely to affect liability to atherosclerotic cardiovascular disease independently of their relationship with liability to type 2 diabetes. Control of modifiable risk factors therefore remains important for reducing atherosclerotic cardiovascular disease risk regardless of patient liability to type 2 diabetes.RESEARCH IN CONTEXTWhat is already known about this subject?Type 2 diabetes, coronary artery disease and peripheral artery disease, share several risk factorsType 2 diabetes is also one of the strongest independent risk factors for both coronary and peripheral artery diseaseWhat is the key question?Which risk factors for atherosclerotic cardiovascular disease are mediated by liability to type 2 diabetes and which are independent?What are the new findings?Among 108 risk factors in this study, there was evidence to support: 10 risk factors as causes, 23 risk factors as consequences, and 34 risk factors as both causes and consequences of liability to type 2 diabetesIn addition, we found evidence for 53 risk factors as causes of liability to coronary artery disease and 42 risk factors as causes of liability to peripheral artery diseaseUsing two-step Mendelian randomization for mediation, we found most risk factors for atherosclerotic cardiovascular disease were likely to act independently of liability to type 2 diabetesHow might this impact on clinical practice in the foreseeable future?Our findings support continued control of modifiable risk factors as this is likely to reduce atherosclerotic cardiovascular disease, regardless of patient liability to type 2 diabetes

Background: Longitudinal studies are crucial for identifying potential risk factors for infection with, and consequences of, COVID-19, but relationships can be biased if they are associated with invitation and response to data collection. We describe factors relating to questionnaire invitation and response in COVID-19 questionnaire data collection in a multigenerational birth cohort (the Avon Longitudinal Study of Parents and Children, ALSPAC). Methods: We analysed online questionnaires completed between the beginning of the pandemic and easing of the first UK lockdown by participants with valid email addresses who had not actively disengaged from the study. We assessed associations of pre-pandemic sociodemographic, behavioural, anthropometric and health-related factors with: i) being sent a questionnaire; ii) returning a questionnaire; and iii) item response (for specific questions). Analyses were conducted in three cohorts: the index children born in the early 1990s (now young adults; 41 variables assessed), their mothers (35 variables) and the mothers’ partners (27 variables). Results: Of 14,849 young adults, 41% were sent a questionnaire, of whom 57% returned one. Item response was >95%. In this cohort, 78% of factors were associated with being sent a questionnaire, 56% with returning one, and, as an example of item response, 20% with keyworker status response. For instance, children from mothers educated to degree-level had greater odds of being sent a questionnaire (OR=5.59; 95% CI=4.87-6.41), returning one (OR=1.60; 95% CI=1.31-1.95), and responding to items (e.g., keyworker status OR=1.65; 95% CI=0.88-3.04), relative to children from mothers with fewer qualifications. Invitation and response rates and associations were similar in all cohorts. Conclusions: These results highlight the importance of considering potential biases due to non-response when using longitudinal studies in COVID-19 research and interpreting results. We recommend researchers report response rates and factors associated with invitation and response in all COVID-19 observational research studies, which can inform sensitivity analyses.

Background Mendelian randomisation uses genetic variants randomly allocated at conception as instrumental variables for a modifiable exposure of interest. Recent methodological advances allow for mediation analysis to be carried out using Mendelian randomisation. When genetic instruments are available for both an exposure and mediator, both multivariable and two-step Mendelian randomisation may be applied. Methods We use simulations and an applied example to demonstrate when multivariable Mendelian randomisation and two-step Mendelian randomisation methods are valid and how they relate to traditional phenotypic regression-based approaches to mediation. We demonstrate how Mendelian randomisation methods can relax assumptions required for causal inference in phenotypic mediation, as well as which Mendelian randomisation specific assumptions are required. We illustrate our methods in data from UK Biobank, estimating the role of body mass index mediating the association between education and cardiovascular outcomes. Results Both multivariable Mendelian randomization and two-step Mendelian randomization are unbiased when estimating the total effect, direct effect, indirect effect and proportion mediated when both confounding, and measurement error are present. Where both the exposure and mediator are continuous, in the presence of a rare or common binary outcome, we found little evidence of bias from non-collapsibility of the odds ratio. Conclusion Phenotypic mediation methods require strong, often untestable, assumptions. Mendelian randomisation provides an opportunity for improving causal inference in mediation analysis. Although Mendelian randomisation specific assumptions apply, such as no weak instrument bias and no pleiotropic pathways, strong assumptions of no confounding and no measurement error can be relaxed.

BackgroundUnderstanding the interplay between educational attainment and genetic predictors of cardiovascular risk may improve our understanding of mechanisms relating educational attainment to cardiovascular disease.MethodsIn up to 320 120 UK Biobank participants of White British ancestry (mean age = 57, female 54%), we created polygenic scores for nine cardiovascular risk factors or diseases: alcohol consumption, body mass index, low-density lipoprotein cholesterol, lifetime smoking behaviour, systolic blood pressure, atrial fibrillation, coronary heart disease, type 2 diabetes and stroke. We estimated whether educational attainment modified genetic susceptibility to these risk factors and diseases.ResultsOn the additive scale, higher educational attainment reduced genetic susceptibility to higher BMI, smoking, atrial fibrillation and type 2 diabetes, but increased genetic susceptibility to higher LDL-C and higher systolic blood pressure.On the multiplicative scale, there was evidence that higher educational attainment increased genetic susceptibility to atrial fibrillation and coronary heart disease, but no evidence of effect modification was found for all other considered traits.ConclusionsEducational attainment modifies the genetic susceptibility to some cardiovascular risk factors and diseases. The direction of this effect was mixed across traits considered and differences in associations between the effect of the polygenic score across strata of educational attainment was uniformly small. Therefore, any effect modification by education of genetic susceptibility to cardiovascular risk factors or diseases is unlikely to contribute substantially to the mechanisms driving inequalities in cardiovascular risk.Key MessagesThe role of educational attainment in modifying the effect of polygenic scores for a wide range of cardiovascular risk factors or diseases has not previously been studiedWe explore whether educational attainment modifies the effects of polygenic susceptibility to alcohol consumption, body mass index, low-density lipoprotein cholesterol, lifetime smoking behaviour, systolic blood pressure, atrial fibrillation, coronary heart disease, type 2 diabetes and strokeEffect modification by education was observed for some cardiovascular polygenic scores, but not all.Effects were not always in the hypothesised direction and were dependent on the scale of analysis.Modification of the effect of genetic susceptibility to cardiovascular risk factors or cardiovascular disease by educational attainment is unlikely to contribute substantially to the mechanisms driving inequalities in cardiovascular risk.

The sample size of a study is a key design and planning consideration. However, sample size and power calculations are often either poorly reported or not reported at all, which suggests they may not form a routine part of study planning. Inadequate understanding of sample size and statistical power can result in poor quality studies. Journals increasingly require a justification of sample size, for example through the use of reporting checklists. However, for meaningful improvements in research quality to be made, researchers need to consider sample size and power at the design stage of a study, rather than at the publication stage. Here we briefly illustrate sample size and statistical power in the context of different research questions and how they should be viewed as a critical design consideration.

Background We aimed to quantify the role of the plasma metabolic profile in explaining the effect of adiposity on cardiac structure. Methods and Results Body mass index (BMI) was measured at age 11 in the Avon Longitudinal Study of Parents and Children. Left ventricular mass indexed to height 2.7 (LVMI) was assessed by echocardiography at age 17. The metabolic profile was quantified via 1 H‐nuclear magnetic resonance spectroscopy at age 15. Multivariable confounder (maternal age, parity, highest qualification, maternal smoking, prepregnancy BMI, prepregnancy height, household social class, adolescent birthweight, adolescent smoking, fruit and vegetable consumption, and physical activity)–adjusted linear regression estimated the association of BMI with LVMI and mediation by metabolic traits. We considered 156 metabolomic traits individually and jointly as principal components explaining 95% of the variance in the nuclear magnetic resonance platform and assessed whether the principal components for the metabolic traits added to the proportion of the association explained by putative cardiovascular risk factors (systolic and diastolic blood pressures, insulin, triglycerides, low‐density lipoprotein cholesterol, and glucose). A 1 kg/m 2 higher BMI was associated with a 0.70 g/m 2.7 (95% CI, 0.53–0.88 g/m 2.7 ) and 0.66 g/m 2.7 (95% CI, 0.53–0.79 g/m 2.7 ) higher LVMI in males (n=437) and females (n=536), respectively. Putative risk factors explained 3% (95% CI, 2%–5%) of this association in males, increasing to 10% (95% CI, 8%–13%) when including metabolic principal components. In females, the standard risk factors explained 3% (95% CI, 2%–5%) of the association and did not increase when including the metabolic principal components. Conclusions The addition of the nuclear magnetic resonance‐measured metabolic traits appears to mediate more of the association of BMI on LVMI than the putative risk factors alone in adolescent males, but not females.

Background Low socio-economic position (SEP) is a risk factor for multiple health outcomes, but its molecular imprints in the body remain unclear. Methods We examined SEP as a determinant of serum nuclear magnetic resonance metabolic profiles in ∼30 000 adults and 4000 children across 10 UK and Finnish cohort studies. Results In risk-factor-adjusted analysis of 233 metabolic measures, low educational attainment was associated with 37 measures including higher levels of triglycerides in small high-density lipoproteins (HDL) and lower levels of docosahexaenoic acid (DHA), omega-3 fatty acids, apolipoprotein A1, large and very large HDL particles (including levels of their respective lipid constituents) and cholesterol measures across different density lipoproteins. Among adults whose father worked in manual occupations, associations with apolipoprotein A1, large and very large HDL particles and HDL-2 cholesterol remained after adjustment for SEP in later life. Among manual workers, levels of glutamine were higher compared with non-manual workers. All three indicators of low SEP were associated with lower DHA, omega-3 fatty acids and HDL diameter. At all ages, children of manual workers had lower levels of DHA as a proportion of total fatty acids. Conclusions Our work indicates that social and economic factors have a measurable impact on human physiology. Lower SEP was independently associated with a generally unfavourable metabolic profile, consistent across ages and cohorts. The metabolites we found to be associated with SEP, including DHA, are known to predict cardiovascular disease and cognitive decline in later life and may contribute to health inequalities.

BackgroundHigher body-mass index (BMI) and waist-to-hip ratio (WHR) increase the risk of cardiovascular disease, but the extent to which this is mediated by blood pressure, diabetes, lipid traits and smoking is not fully understood.MethodsUsing consortia and UK Biobank genetic association summary data from 140,595 to 898,130 participants predominantly of European ancestry, MR mediation analysis was performed to investigate the degree to which genetically predicted systolic blood pressure (SBP), diabetes, lipid traits and smoking mediated an effect of genetically predicted BMI and WHR on risk of coronary artery disease (CAD), peripheral artery disease (PAD) and stroke.ResultsThe 49% (95% confidence interval [CI] 39%-60%) increased risk of CAD conferred per 1-standard deviation increase in genetically predicted BMI attenuated to 34% (95% CI 24%-45%) after adjusting for genetically predicted SBP, to 27% (95% CI 17%-37%) after adjusting for genetically predicted diabetes, to 47% (95% CI 36%-59%) after adjusting for genetically predicted lipids, and to 46% (95% CI 34%-58%) after adjusting for genetically predicted smoking. Adjusting for all the mediators together, the increased risk attenuated to 14% (95% CI 4%-26%). A similar pattern of attenuation was observed when considering genetically predicted WHR as the exposure, and PAD or stroke as the outcomes.ConclusionsMeasures to reduce obesity will lower risk of cardiovascular disease primarily by impacting on downstream metabolic risk factors, particularly diabetes and hypertension. Reduction of obesity prevalence alongside control and management of its mediators is likely to be most effective for minimizing the burden of obesity.

IntroductionGenome-wide association studies (GWASs) often adjust for covariates, correct for medication use, or select on medication users. If these summary statistics are used in two-sample Mendelian randomization analyses, estimates may be biased. We used simulations to investigate how GWAS adjustment, correction and selection affects these estimates and performed an analysis in UK Biobank to provide an empirical example.MethodsWe simulated six GWASs: no adjustment for a covariate, correction for medication use, or selection on medication users; adjustment only; selection only; correction only; both adjustment and selection; and both adjustment and correction. We then ran two-sample Mendelian randomization analyses using these GWASs to evaluate bias. We also performed equivalent GWASs using empirical data from 306,560 participants in UK Biobank with systolic blood pressure as the exposure and body mass index as the covariate and ran two-sample Mendelian randomization with coronary heart disease as the outcome.ResultsThe simulation showed that estimates from GWASs with selection can produce biased two-sample Mendelian randomization estimates. Yet, we observed relatively little difference between empirical estimates of the effect of systolic blood pressure on coronary artery disease across the six scenarios.ConclusionsGiven the potential for bias from using GWASs with selection on Mendelian randomization estimates demonstrated in our simulation, careful consideration before using this approach is warranted. However, based on our empirical results, using adjusted, corrected or selected GWASs is unlikely to make a large difference to two-sample Mendelian randomization estimates in practice.

BackgroundThe most socioeconomically deprived individuals remain at the greatest risk of cardiovascular disease. Differences in risk adjusted use of statins between educational groups may contribute to these inequalities. We explore whether people with lower levels of educational attainment are less likely to take statins for a given level of cardiovascular risk.Methods and findingsUsing data from a large prospective cohort study, UK Biobank, we calculated a QRISK3 cardiovascular risk score for 472 097 eligible participants with complete data on self-reported educational attainment and statin use (55% female; mean age, 56). We used logistic regression to explore the association between i) QRISK3 score and self-report statin use and ii) educational attainment and self-report statin use. We then stratified the association of QRISK3 score, and statin use by strata of educational attainment to test for an interaction. In this sample, greater education was associated with lower statin use, whilst higher cardiovascular risk (assessed by QRISK3 score) was associated with higher statin use in both females and males. There was evidence of an interaction between QRISK3 and education, such that for the same QRISK3 score, people with more education were more likely to report taking statins. For example, in women with 7 years of schooling, equivalent to leaving school with no formal qualifications, a one unit increase in QRISK3 score was associated with a 7% higher odds of statin use (odds ratio (OR) 1.07, 95% CI 1.07, 1.07). In contrast, in women with 20 years of schooling, equivalent to obtaining a degree, a one unit increase in QRISK3 score was associated with an 14% higher odds of statin use (OR 1.14, 95% CI 1.14, 1.15). Comparable ORs in men were 1.04 (95% CI 1.04, 1.05) for men with 7 years of schooling and (95% CI 1.08, 1.08) for men with 20 years of schooling. Linkage between UK biobank and primary care data meant we were able to carry out a number of sensitivity analyses to test the robustness of our findings. However, a limitation of our study is that a number of assumptions were made when deriving QRISK3 scores which may overestimate the scores.ConclusionsFor the same level of cardiovascular risk, individuals with lower educational attainment are less likely to receive statins, likely contributing to health inequalities.SummaryWhat is already known on this topic?Despite reductions in the rates of cardiovascular disease in high income countries, individuals who are the most socioeconomically deprived remain at the highest risk.Although intermediate lifestyle and behavioural risk factors explain some of this, much of the effect remains unexplained.What does this study add?For the same increase in QRISK3 score, the likelihood of statin use increased more in individuals with high educational attainment compared with individuals with lower educational attainment.These results were similar when using UK Biobank to derive QRISK3 scores and when using QRISK scores recorded in primary care records, and when using self-reported statin prescription data or prescription data from linked primary care records.The mechanisms leading to these differences are unknown, but both health seeking behaviours and clinical factors may contribute.

BackgroundWe aimed to quantify the role of the plasma metabolic profile in explaining the effect of adiposity on cardiac structure.MethodsBody mass index (BMI) was measured at age 11 in the Avon Longitudinal Study of Parents and Children. Left ventricular mass indexed to height2.7 (LVMI), was assessed by echocardiography at age 17. The metabolic profile was quantified via nuclear magnetic resonance spectroscopy at age 15.Multivariable confounder (maternal age, parity, highest qualification, maternal smoking, pre-pregnancy BMI, pre-pregnancy height, household social class and adolescent birthweight, adolescent smoking, fruit and vegetable consumption, physical activity) -adjusted linear regression estimated the association of BMI with LVMI and mediation by metabolic traits. We considered 156 metabolomic traits individually, jointly as principal components (PCs) explaining 95% of the variance in the NMR platform, and assessed whether the PCs for the metabolic traits added to the proportion of the association explained by established cardiovascular risk factors (systolic and diastolic blood pressure, insulin, triglycerides, low density lipoprotein, and glucose).ResultsA 1kg/m2 higher BMI was associated with a 0.70 g/m2.7 (0.53, 0.88) and 0.66 g/m2.7 (0.53, 0.79) higher LVMI in males (N=437) and females (N=536), respectively. Established risk factors explained 3% (95% CI: 2% to 5%) of this association in males, increasing to 10% (95% CI: 8%, 13%) when including metabolic PCs. In females, the standard risk factors explained 3% (95% CI: 2%, 5%) of the association, and did not increase when including the metabolic PCs.ConclusionThe addition of the NMR measured metabolic traits appear to mediate more of the effect of BMI on LVMI than the established risk factors alone in adolescent males, but not females.

Background Studies have demonstrated causal effects of educational attainment on cardiovascular disease (CVD). We aimed to investigate the role of body mass index (BMI), systolic blood pressure (SBP) and smoking in explaining the effect of education on risk of CVD outcomes triangulating across multivariable regression analysis of observational data and one- and two-sample Mendelian randomisation (MR) analysis; an instrumental variable approach more robust to bias from confounding and reverse causation. Methods Individual level data from UK Biobank (N=217 013) was used for multivariable analyses and one-sample MR. Summary statistics from genome-wide association studies were used in two-sample MR. The total effect of education on risk of coronary heart disease (CHD), CVD (all subtypes), myocardial infarction (MI) and stroke (all measured in odds ratio, OR) was assessed using multivariable regression and univariable Mendelian randomization (MR). The degree to which this effect is mediated through BMI, SBP and smoking respectively (the indirect effect and proportion mediated) was estimated using the product of coefficients method, where the effect of education on each mediator, and each mediator on each outcome was assessed using multivariable regression and Network MR. The joint contribution of all three risk factors was assessed via the difference method, using multivariable regression or multivariable MR. Results Each additional standard deviation of education (3.6 years) associated with 13% lower risk of coronary heart disease (OR 0.87, 95% confidence interval [CI] 0.84 to 0.89) in observational analysis and 37% lower risk (OR 0.63, 95% CI 0.60 to 0.67) in MR analysis. As a proportion of the total risk reduction, BMI mediated 15% (95% CI 13% to 17%) and 18% (95% CI 14% to 23%) in the observational and MR estimates respectively. Corresponding estimates for SBP were 11% (95% CI 9% to 13%) and 21% (95% CI 15% to 27%), and for smoking, 19% (15% to 22%) and 34% (95% CI 17% to 50%). All three risk factors combined mediated 42% (95% CI 36% to 48%) and 36% (95% CI 16% to 63%) of the effect of education on CHD in observational and MR respectively. Similar results were obtained for risk of stroke, MI and all-cause CVD. Conclusion BMI, SBP and smoking mediate a substantial proportion of the protective effect of education on risk of cardiovascular outcomes. Intervening on these would reduce cases of CVD attributable to lower education. However, more than half of the protective effect of education remains unexplained.

ObjectivesTo investigate the role of body mass index (BMI), systolic blood pressure, and smoking behaviour in explaining the effect of education on the risk of cardiovascular disease outcomes.DesignMendelian randomisation study.SettingUK Biobank and international genome-wide association study data.ParticipantsPredominantly participants of European ancestry.ExposureEducational attainment, BMI, systolic blood pressure, and smoking behaviour in observational analysis, and randomly allocated genetic variants to instrument these traits in mendelian randomisation.Main outcomes measureThe risk of coronary heart disease, stroke, myocardial infarction, and cardiovascular disease (all subtypes; all measured in odds ratio), and the degree to which this is mediated through BMI, systolic blood pressure, and smoking behaviour respectively.ResultsEach additional standard deviation of education (3.6 years) was associated with a 13% lower risk of coronary heart disease (odds ratio 0.86, 95% confidence interval 0.84 to 0.89) in observational analysis and a 37% lower risk (0.63, 0.60 to 0.67) in mendelian randomisation analysis. As a proportion of the total risk reduction, BMI was estimated to mediate 15% (95% confidence interval 13% to 17%) and 18% (14% to 23%) in the observational and mendelian randomisation estimates, respectively. Corresponding estimates were 11% (9% to 13%) and 21% (15% to 27%) for systolic blood pressure and 19% (15% to 22%) and 34% (17% to 50%) for smoking behaviour. All three risk factors combined were estimated to mediate 42% (36% to 48%) and 36% (5% to 68%) of the effect of education on coronary heart disease in observational and mendelian randomisation analyses, respectively. Similar results were obtained when investigating the risk of stroke, myocardial infarction, and cardiovascular disease.ConclusionsBMI, systolic blood pressure, and smoking behaviour mediate a substantial proportion of the protective effect of education on the risk of cardiovascular outcomes and intervening on these would lead to reductions in cases of cardiovascular disease attributable to lower levels of education. However, more than half of the protective effect of education remains unexplained and requires further investigation.

BACKGROUNDThe interactive effect of two exposures on an outcome can be confounded. We demonstrate the use of Mendelian Randomization (MR) to estimate unconfounded additive interactions.METHODSUsing simulation, we test an extension to multivariable MR using two-stage least squares to estimate the additive interaction between two continuous exposures on a continuous outcome, including scenarios where one exposure has a causal effect on the other (mediation). The interaction parameters were set to be one third of the main effect’s parameters to impose a limit on the variance explained by interaction terms. We compare the performance of the two-stage least squares estimator to a Factorial MR design, in which genetic risk scores for each exposure are dichotomised to create four groups, akin to a factorial randomized controlled trial. As an illustrative example, we apply factorial MR and the 2SLS estimator to the interactive effect of education and BMI on systolic blood pressure in UK Biobank.RESULTSOur simulations demonstrate that factorial MR has very low statistical power; 5-7% at N=50,000 and 8-23% at N=500,000 across the range of parameters tested. The two-stage least squares estimator had higher power to detect interactions than factorial MR and a lower type I error. For N=500,000, the two-stage least squares estimator had a power ranging from 29.7-92.9% and a type I error ranging from 4-6%. 95% Monte Carlo confidence intervals suggested that the estimator was unbiased to a reasonable degree of accuracy at this sample size. In comparison, the power at N=50,000 was 7-55%.CONCLUSIONSA two-stage least squares estimator using genetic risk scores for each exposure is a more powerful alternative for detecting an unconfounded additive interaction of two exposures on an outcome than existing approaches that rely on dichotomising genetic risk scores, but it requires a large sample size and instruments of adequate strength.

Key Points Question Is there a joint association of body mass index and alcohol consumption with liver disease and liver injury biomarkers? Findings In this mendelian randomization study of a population-based cohort of 91 552 European adults, compared with individuals categorized as having both a high body mass index and weekly units of alcohol consumption, those in the low category for both of these risk factors were associated with lower circulating liver injury biomarkers compared with being high for either one or both of the risk factors. However, this association was less clear when considering cases of liver disease as the outcome. Meaning Interventions to reduce both body mass index and alcohol consumption may result in the greatest reduction in circulating liver injury biomarkers, but these results do not appear to be able to demonstrate whether these interventions will result in a reduction in cases of liver disease., Importance Individually, higher body mass index (BMI) and alcohol consumption increase the risk of liver disease. Evidence of a joint association is mixed; however, previous studies have not used causal inference methods robust to confounding and reverse causation. Understanding any true effect is key to developing effective interventions to reduce liver disease. Objective To investigate the joint association of BMI and alcohol consumption with liver injury biomarkers and incident liver disease using factorial mendelian randomization (MR). Design, Setting, and Participants A population-based cohort study (Copenhagen General Population Study) recruited a random sample of Copenhagen, Denmark, residents aged 20 years or older of white, Danish descent (N = 98 643) between November 25, 2003, and July 1, 2014. Data were also obtained from ongoing links to national registers, and then analyzed from September 30, 2016, to April 23, 2018. Exposures High and low BMI and alcohol consumption categories from baseline-measured or self-reported observational data and genetic variants predicting BMI and alcohol consumption. Main Outcomes and Measures Plasma biomarkers of liver injury (alanine aminotransferase [ALT] and γ-glutamyltransferase [GGT]) and incident cases of liver disease from hospital records were the outcomes. Results Of the 98 643 individuals recruited, 91 552 (54 299 [45.2%] women; mean [SD] age, 58 [13.05] years) with no baseline liver disease were included in main analyses. Individuals had a mean (SD) BMI of 26.2 (4.3) and consumed a mean (SD) of 10.6 (10.2) U/wk of alcohol. In factorial MR analyses, considering the high BMI/high alcohol group as the reference, mean circulating ALT and GGT levels were lowest in the low BMI/low alcohol group (ALT: −2.32%; 95% CI, −4.29% to −0.35%, and GGT: −3.56%; 95% CI, −5.88% to −1.24%). Individuals with low BMI/high alcohol use and high BMI/low alcohol use also had lower mean circulating ALT levels (low BMI/high alcohol use: −1.31%; 95% CI, −1.88% to −0.73%, and high BMI/low alcohol use: −0.81%; 95% CI, −2.86% to 1.22%) and GGT levels (low BMI/high alcohol use: −0.91%; 95% CI, −1.60% to −0.22%, and high BMI/low alcohol use: −1.13%; 95% CI, −3.55% to 1.30%) compared with the high BMI/high alcohol use reference group. These patterns were similar in multivariable factorial analyses. For incident liver disease (N = 580), factorial MR results were less conclusive (odds ratio of liver disease vs high BMI/high alcohol group: 1.01; 95% CI, 0.84 to 1.18, for the low BMI/high alcohol group, 1.22; 95% CI, 0.56 to 1.88 for the high BMI/low alcohol group, and 0.99 (95% CI, 0.41 to 1.56 for the low BMI/low alcohol group). Conclusions and Relevance Interventions to reduce both BMI and alcohol consumption might reduce population levels of biomarkers of liver injury more than interventions aimed at either BMI or alcohol use alone. However, it is not clear whether this intervention will directly translate to a reduced risk of liver disease., This population-based cohort study uses mendelian randomization to examine the association of body mass index (BMI) and alcohol consumption with levels of liver injury biomarkers and incident liver disease in adults in Denmark.

Key PointsQuestionWhat is the role of body mass index, systolic blood pressure and smoking in mediating the effect of education on cardiovascular disease risk?FindingWe find consistent evidence that body mass index, systolic blood pressure and smoking mediate the effect of education, explaining up to 18%, 27% and 33% respectively. Including all three risk factors in a model together explains around 40% of the effect of education.MeaningIntervening on body mass index, systolic blood pressure and smoking would lead to reductions in cases of CVD attributable to lower levels of education. Over half of the effect of education on risk of cardiovascular disease is not mediated through these risk factors.ImportanceLower levels of education are causally related to higher cardiovascular risk, but the extent to which this is driven by modifiable risk factors also associated with education is unknown.ObjectiveTo investigate the role of body mass index, systolic blood pressure and smoking in explaining the effect of education on risk of cardiovascular disease outcomes.DesignMultivariable regression analysis of observational data and Mendelian randomization (MR) analysis of genetic data.SettingUK Biobank and international genome-wide association study consortia.ParticipantsPredominantly individuals of European ancestry.Main outcomes and measuresThe effects of education (per 1-standard deviation increase, equivalent to 3.6 years) on coronary heart disease, cardiovascular disease (all subtypes), myocardial infarction and stroke risk (all measured in odds ratio, OR), and the degree to which this is mediated through body mass index, systolic blood pressure and smoking.ResultsEach additional standard deviation of education associated with 13% lower risk of coronary heart disease (OR 0.87, 95% confidence interval [CI] 0.84 to 0.89) in observational analysis and 37% lower risk (OR 0.63, 95% CI 0.60 to 0.67) in Mendelian randomization analysis. As a proportion of the total risk reduction, body mass index mediated 15% (95% CI 13% to 17%) and 18% (95% CI 14% to 23%) in the observational and Mendelian randomization estimates, respectively. Corresponding estimates for systolic blood pressure were 11% (95% CI 9% to 13%) and 21% (95% CI 15% to 27%), and for smoking, 19% (15% to 22%) and 33% (95% CI 17% to 49%). All three risk factors combined mediated 42% (95% CI 36% to 48%) and 36% (95 % CI 16% to 63%) of the effect of education on coronary heart disease in observational and Mendelian randomization respectively. Similar results were obtained when investigating risk of stroke, myocardial infarction and all-cause cardiovascular disease.Conclusions and relevanceBMI, SBP and smoking mediate a substantial proportion of the protective effect of education on risk of cardiovascular outcomes and intervening on these would lead to reductions in cases of CVD attributable to lower levels of education. However, more than half of the protective effect of education remains unexplained and requires further investigation.

BackgroundBody mass index (BMI) and alcohol consumption are suggested to independently and interactively increase the risk of liver disease. We assessed this combined effect using factorial Mendelian randomization (MR).MethodsWe used multivariable adjusted regression and MR to estimate individual and joint associations of BMI and alcohol consumption and liver disease biomarkers (alanine aminotransferase (ALT) y-glutamyltransferase (GGT)) and incident liver disease. We undertook a factorial MR study splitting participants by median of measured BMI or BMI allele score then by median of reported alcohol consumption or ADH1B genotype (AA/AG and GG), giving four groups; low BMI/low alcohol (-BMI/-alc), low BMI/high alcohol (-BMI/+alc), high BMI/low alcohol (+BMI/-alc) and high BMI/high alcohol (+BMI/+alc).ResultsIndividual positive associations of BMI and alcohol with ALT, GGT and incident liver disease were found. In the factorial MR analyses, considering the +BMI/+alc group as the reference, mean circulating ALT and GGT levels were lowest in the -BMI/-alc group (2.32% (95% CI: −4.29, −0.35) and −3.56% (95% CI: −5.88; −1.24) for ALT and GGT respectively). Individuals with -BMI/+alc and +BMI/-alc had lower mean circulating ALT and GGT compared to the reference group (+BMI/+alc). For incident liver disease multivariable factorial analyses followed a similar pattern to those seen for the biomarkers, but little evidence of differences between MR factorial categories for odds of liver disease.ConclusionsConsistent results from multivariable regression and MR analysis, provides compelling evidence for the individual adverse effects of BMI and alcohol consumption on liver disease. Intervening on both BMI and alcohol may improve the profiles of circulating liver biomarkers. However, this may not reduce clinical liver disease risk.

Adequate sample size is key to reproducible research findings: low statistical power can increase the probability that a statistically significant result is a false positive. Journals are increasingly adopting methods to tackle issues of reproducibility, such as by introducing reporting checklists. We conducted a systematic review comparing articles submitted to Nature Neuroscience in the 3 months prior to checklists (n=36) that were subsequently published with articles submitted to Nature Neuroscience in the 3 months immediately after checklists (n=45), along with a comparison journal Neuroscience in this same 3-month period (n=123). We found that although the proportion of studies commenting on sample sizes increased after checklists (22% vs 53%), the proportion reporting formal power calculations decreased (14% vs 9%). Using sample size calculations for 80% power and a significance level of 5%, we found little evidence that sample sizes were adequate to achieve this level of statistical power, even for large effect sizes. Our analysis suggests that reporting checklists may not improve the use and reporting of formal power calculations.

The presence of data in the “collection_date” field of a GenBank sequence record is of great assistance in the use of that sequence for Bayesian phylogenetics using “tip-dating”. We presentTempus et Locus(TeL), a tool for extracting such sequences from a GenBank-formatted sequence database. TeL shows that 60% of viral sequences in GenBank have collection date fields, but that this varies considerably between species. Primate erythroparvovirus 1 (human parvovirus B19 or B19V) has only 40% of its sequences dated, of which only 112 are of more than 4 kb. 100 of these are from B19V sub-genotype 1a and were collected from a mere 6 studies conducted in 5 countries between 2002 and 2013. Nevertheless, Bayesian phylogenetic analysis of this limited set gives a date for the common ancestor of sub-genotype 1a in 1990 (95% HPD 1981-1996) which is in reasonable agreement with estimates of previous studies where collection dates have been assembled by more laborious methods of literature search and direct enquiries to sequence submitters. We conclude that although collection dates should become standard for all future GenBank submissions of virus sequences, accurate dating of ancestors is possible with even a small number of sequences if sampling information is high quality.

Introduction Strong evidence shows that adiposity increases cardiovascular disease (CVD) risk, explained in part by blood pressure (BP), glucose, triglycerides and cholesterol. Metabolomics offers the potential to identify novel intermediate pathways. Methods Body mass index (BMI) was measured at age 11 in the Avon Longitudinal Study of Parents and Children. Measures of cardiac structure (precursors of CVD; age 17) were left atrial size indexed to height (LAI), left ventricular mass indexed to height2.7 (LVMI), relative wall thickness (RWT) and left ventricular internal diameter (LVIDD). Metabolic traits (mostly lipid and lipoprotein related) were quantified via high-throughput 1H-nuclear magnetic resonance spectroscopy (NMR) at age 15. Complete data was available for all exposures, mediators and outcomes (n = 772). Multiple imputation was used to deal with missing data in covariates. Multivariable linear regression was used to estimate associations of BMI with measures of cardiac structure. Mediation was assessed via controlled direct and natural indirect effects, firstly, considering 156 metabolic measures individually, and secondly considering all metabolites jointly (as principle components, npc = 17). Bootstrapping was used to calculate robust standard errors. Results A one-unit higher BMI was associated with 0.74 (0.54, 0.94) higher LVMI in males; 0.68 (0.52, 0.84) in females. Individually, each metabolite explained little of this association. Jointly, the PCs of the metabolites explained 8% of the association in males and 0.8% in females. Similar results were seen for LAI and LVIDD. There was weak evidence of an association of BMI on RWT. Conclusion In this adolescent population, individual metabolites measured by NMR contribute a small amount to the pathway from adiposity to cardiac structure. Considering them jointly indicates they may play a role in the pathway, particularly in males; further work is warranted to assess causality.