Exposure-mediator Interaction Research Articles

Statistical Mediation Analysis for Models with a Binary Mediator and a Binary Outcome: the Differences Between Causal and Traditional Mediation Analysis

Mediation analysis is an important statistical method in prevention research, as it can be used to determine effective intervention components. Traditional mediation analysis defines direct and indirect effects in terms of linear regression coefficients. It is unclear how these traditional effects are estimated in settings with binary variables. An important recent methodological advancement in the mediation analysis literature is the development of the causal mediation analysis framework. Causal mediation analysis defines causal effects as the difference between two potential outcomes. These definitions can be applied to any mediation model to estimate natural direct and indirect effects, including models with binary variables and an exposure–mediator interaction. This paper aims to clarify the similarities and differences between the causal and traditional effect estimates for mediation models with a binary mediator and a binary outcome. Causal and traditional mediation analyses were applied to an empirical example to demonstrate these similarities and differences. Causal and traditional mediation analysis provided similar controlled direct effect estimates, but different estimates of the natural direct effects, natural indirect effects, and total effect. Traditional mediation analysis methods do not generalize well to mediation models with binary variables, while the natural effect definitions can be applied to any mediation model. Causal mediation analysis is therefore the preferred method for the analysis of mediation models with binary variables.

Does hypertension mediate the relationship between environmental noise and myocardial infarction ?

BACKGROUND AND AIM: Environmental noise is associated with cardiovascular diseases, including myocardial infarction (MI). We hypothesized that hypertension may be a mediator in the causal pathway linking noise exposure to MI. We aimed to assess the potential mediating effects of hypertension on the relationship between noise and MI, using causal mediation analysis in a population-based cohort. METHODS: The cohort derives from the linkage of health administrative databases. It includes all residents of Montreal (Canada) between 2000-2014, who were 45 years and older and free of MI and hypertension at time of entry. Incident cases of MI and hypertension were ascertained from validated case definitions. Long-term environmental noise exposure, expressed as level acoustic equivalent 24h (LAeq24h), was estimated from a land use regression model and assigned to participants using their residential postal code at entry in the cohort. We performed causal mediation analysis based on a counterfactual approach, using Cox proportional hazards model for the exposure-outcome model and logistic regression for the exposure-mediator model. RESULTS:The cohort comprised 912,632 individuals, with 26,966 incident cases of MI and 250,382 cases of hypertension. 36% of MI cases had developed hypertension. Environmental noise was associated with MI and hypertension; adjusted estimates of association were 1.07 (95% confidence interval (CI): 1.06-1.08) for MI and 1.07 (95% CI: 1.06-1.08) for hypertension, per interquartile range increase (54.9 to 60.5 decibel A) in the annual mean LAeq24h. In the causal mediation analysis, there was no evidence of exposure-mediator interaction. The estimated total effect was 1.08 (95% CI: 1.07-1.08), the direct effect of noise was 1.08 (95% CI: 1.07-1.08) and the indirect effect mediated by hypertension was null (1.00; 95% CI: 1.00-1.00), per interquartile range increase in LAeq24h. CONCLUSIONS:In this population-based cohort study, we found no evidence that the effect of long-term environmental noise exposure on the incidence of MI is mediated through hypertension. KEYWORDS: noise, mediation analysis, cardiovascular disease, hypertension, myocardial infarction, cohort

The impact of adjusting for pure predictors of exposure, mediator, and outcome on the variance of natural direct and indirect effect estimators.

It is now well established that adjusting for pure predictors of the outcome, in addition to confounders, allows unbiased estimation of the total exposure effect on an outcome with generally reduced standard errors (SEs). However, no analogous results have been derived for mediation analysis. Considering the simplest linear regression setting and the ordinary least square estimator, we obtained theoretical results showing that adjusting for pure predictors of the outcome, in addition to confounders, allows unbiased estimation of the natural indirect effect (NIE) and the natural direct effect (NDE) on the difference scale with reduced SEs. Adjusting for pure predictors of the mediator increases the SE of the NDE's estimator, but may increase or decrease the variance of the NIE's estimator. Adjusting for pure predictors of the exposure increases the variance of estimators of the NIE and NDE. Simulation studies were used to confirm and extend these results to the case where the mediator or the outcome is binary. Additional simulations were conducted to explore scenarios featuring an exposure‐mediator interaction as well as the relative risk and odds ratio scales for the case of binary mediator and outcome. Both a regression approach and an inverse probability weighting approach were considered in the simulation study. A real‐data illustration employing data from the Canadian Study of Health and Aging is provided. This analysis is concerned with the mediating effect of vitamin D in the effect of physical activity on dementia and its results are overall consistent with the theoretical and empirical findings.

Does positive affect mediate the association of multimorbidity on depressive symptoms?

Objectives Multimorbidity poses various challenges, and previous research has indicated a causal relation with depression. As multimorbidity is not curable, the underlying mechanisms are of great interest. Positive affect is a major resource for coping with chronic conditions and for the prevention of depression. Long-term multimorbidity, however, may deplete positive affect. The purpose of this paper is to investigate the role of positive affect in the association between multimorbidity and depressive symptoms. Method We used four consecutive waves (2008, 2011, 2014, 2017) of the nationally representative German Ageing Survey (DEAS) with a total of 1,558 older adults aged 40 and over. To account for time-varying confounding, exposure-induced mediator-outcome confounding, and reciprocities, we applied the mediational g-formula with inverse-probability weighting techniques. We also tested for exposure-mediator interaction to adjust for differences in mediation across the duration of multimorbidity. Results We confirmed a positive longitudinal relation between multimorbidity and depressive symptoms, both of which were negatively associated with while positive affect. The model without interaction indicated a share mediated of ca. 18.3% on the total effect of multimorbidity on depressive symptoms. Addition of interaction led to substantial differences for multimorbidity duration and levels of positive affect. Associations for long-term multimorbidity (at least two survey waves) were more substantial, and the share mediated doubled (>40%). Additionally, the direct effect of multimorbidity on depressive symptoms diminished for short-term multimorbidity. Conclusion Strengthening positive affect could reduce depressive symptoms in those facing multimorbidity. This study also discusses methodological challenges in performing longitudinal mediation analysis. We advise researchers to consider the mediational g-formula and exposure-mediator interaction.

Stunting and lead: using causal mediation analysis to better understand how environmental lead exposure affects cognitive outcomes in children

BackgroundMany children in Bangladesh experience poor nutritional status and environmental lead exposure, both of which are associated with lower scores on neurodevelopmental assessments. Recent studies have suggested that part of lead’s adverse effects on neurodevelopment are caused in part by lead’s effect on growth. New statistical methods are now available to evaluate potential causal pathways in observational studies. This study used a novel statistical method to test the hypothesis that stunting, a measure of linear growth related to poor nutrition, is a mediator and/or an effect modifier of the lead exposure’s adverse effect on cognitive development.MethodsParticipants were 734 children from a longitudinal birth cohort established in rural Bangladesh to study the health effects of prenatal and early childhood environmental metal exposures. Lead exposure was estimated using umbilical cord blood samples obtained at birth and blood obtained via venipuncture at age 20–40 months. Stunting was determined using the World Health Organization’s standards. Neurodevelopment was assessed at age 20–40 months years using the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III). We evaluated the effect of lead on stunting and whether the effect of lead on cognitive scores is modified by stunting status in multivariable regression analyses. We then conducted a novel 4-way mediation analysis that allows for exposure-mediator interaction to assess how much of the effect of lead on cognitive scores is explained by the pathway through stunting (mediation) and how much is explained by the interaction between lead and stunt (effect modification).ResultsStunting was not a mediator of the effect of lead in our analyses. Results suggested effect modification by stunting. In an area of Bangladesh with lower lead exposures (median umbilical cord blood lead concentration, 1.7 μg/dL), stunting modified the relationship between prenatal blood lead concentrations and cognitive score at age 2–3 years. A 1-unit increase in natural log cord blood lead concentration in the presence of stunting was associated with a 2.1-unit decrease in cognitive scores (β = − 2.10, SE = 0.71, P = 0.003). This interaction was not found in a second study site where lead exposures were higher (median umbilical cord blood lead concentration, 6.1 μg/dL, β = − 0.45, SE = 0.49, P = 0.360).ConclusionsWe used a novel method of mediation analysis to test whether stunting mediated the adverse effect of prenatal lead exposure on cognitive outcomes in Bangladesh. While we did not find that stunting acted as mediator of lead’s effect on cognitive development, we found significant effect modification by stunting. Our results suggest that children with stunting are more vulnerable to the adverse effects of low-level lead exposure.

The Use of Traditional and Causal Estimators for Mediation Models with a Binary Outcome and Exposure-Mediator Interaction

ABSTRACT An important recent development in mediation analysis is the use of causal mediation analysis. Causal mediation analysis decomposes the total exposure effect into causal direct and indirect effects in the presence of exposure-mediator interaction. However, in practice, traditional mediation analysis is still most widely used. The aim of this paper is to demonstrate the similarities and differences between the causal and traditional estimators for mediation models with a continuous mediator, a binary outcome, and exposure-mediator interaction. A real-life data example, analytical comparisons, and a simulation study were used to demonstrate the similarities and differences between the traditional and causal estimators. The causal and traditional estimators provide similar indirect effect estimates, but different direct and total effect estimates. Traditional mediation analysis may only be used when conditional direct effect estimates are of interest. Causal mediation analysis is the generally preferred method as its casual effect estimates help unravel causal mechanisms.

Direct and indirect pathways between low income status and becoming a high-cost health care user in Ontario, Canada: a mediation analysis of health risk behaviors

PurposeWe investigated natural direct and indirect pathways between low-income status and high-cost user (HCU) transitions considering health risk behaviors as potential mediators. MethodsWe analyzed data from respondents (aged 18+) from four pooled cycles of the Canadian Community Health Survey (2005–2010/2011) linked to administrative data in Ontario, Canada. HCUs were defined as the top 5% of the population, ranked by cost consumption in any of the five years after survey interview. Low-income status was defined from the provincial distribution of self-reported household income, with missing values imputed from neighborhood-level data. In mediation analyses based on marginal structural models, we quantified the contributions of smoking, physical inactivity, alcohol consumption, and body mass index to income-HCU associations. Results115,091 respondents (representative of 9,661,764 Ontarians) were included in the study, of which 7.2% became HCUs. The odds of becoming HCUs were 1.36 times (95% CI: 1.25–1.48) greater for low (vs. high) income status respondents. Smoking, physical activity, alcohol consumption, and body mass index contributed 9.4%, 6.5%, 10.6%, and 4.4% to this association, respectively. Tests for exposure-mediator interactions were not statistically significant. ConclusionsHealth risk behaviors only partially explain income inequalities in future HCU transitions.

1245-P: Maternal HbA1c Is Related to Childhood Fat Mass and Only Partially Mediated by Neonatal Fat Mass

Maternal diabetes during pregnancy impacts offspring body composition and has been associated with greater offspring adiposity across the life course. However, the extent to which the relationship between maternal glycemia and offspring adiposity during later life is independent of neonatal adiposity is unclear. In a prospective study of 544 mother-child pairs in the Healthy Start Study, we estimated the total effect of maternal HbA1c on childhood fat mass (FM) %, and the direct effect independent of neonatal FM %. We measured maternal fasting HbA1c at a gestational age of 20-34 weeks, and offspring FM % using air displacement plethysmography at birth and during childhood (age 4-7 y). We used multivariable linear regression with HbA1c tertiles (T1 [reference], T2, T3) as the exposure, neonatal FM % as the mediator, and childhood FM % as the outcome. A counterfactual-based approach was used to estimate total and direct effects, while also accounting for exposure-mediator interaction. Covariates were maternal race, education, gestational weight gain and smoking, and offspring sex and age. The medians (interquartile ranges) of the HbA1c tertiles were T1: 4.80 (3.70, 4.90), T2: 5.05 (5.00, 5.10), and T3: 5.30 (5.20, 6.30). The estimated total effect on child FM % was 1.27 (95% CI: 0.00, 2.54) units higher for T2 vs. T1, and 1.53 (0.23, 2.82) units higher for T3 vs. T1 (P-trend=0.02). After adjusting for pre-pregnancy BMI, the total effect was 1.18 (-0.09, 2.44) and 1.31 (0.00, 2.62) units higher for T2 vs. T1 and T3 vs. T1, respectively (P-trend=0.04). The direct effect after accounting for neonatal FM % was slightly attenuated (T2 vs. T1: 1.14 [-0.11, 2.39], T3 vs. T1: 1.20 [-0.11, 2.51]; P-trend=0.07). Excluding women with gestational diabetes mellitus (n=27) did not substantially impact the results. Our data suggest that maternal glycemia during pregnancy, even within normal levels, may influence childhood adiposity through pathways that are in part mediated by programming of neonatal adiposity. Disclosure E.C. Francis: None. B. Ringham: None. D.H. Glueck: None. W. Perng: None. D. Dabelea: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (5R01DK076648-10)

To What Extent Is the Association Between Race/Ethnicity and Fetal Growth Restriction Explained by Adequacy of Prenatal Care? A Mediation Analysis of a Retrospectively Selected Cohort

Race/ethnicity is associated with intrauterine growth restriction (IUGR) and small-for-gestational age (SGA) birth. We evaluated the extent to which this association is mediated by adequacy of prenatal care (PNC). A retrospective cohort study was conducted using US National Center for Health Statistics natality files for the years 2011-2017. We performed mediation analyses using a statistical approach that allows for exposure-mediator interaction, and we estimated natural direct effects, natural indirect effects, and proportions mediated. All effects were estimated as risk ratios. Among 23,118,656 singleton live births, the excess risk of IUGR among Black women, Hispanic women, and women of other race/ethnicity as compared with White women was partly mediated by PNC adequacy: 13% of the association between non-Hispanic Black race/ethnicity and IUGR, 12% of the association in Hispanic women, and 10% in other women was attributable to PNC inadequacy. The percentage of excess risk of SGA birth that was mediated was 7% in Black women, 6% in Hispanic women, and 5% in other women. Our findings suggest that PNC adequacy may partly mediate the association between race/ethnicity and fetal growth restriction. In future research, investigators should employ causal mediation frameworks to consider additional factors and mediators that could help us better understand this association.

Time series analysis of total and direct associations between high temperatures and preterm births in Detroit, Michigan

ObjectivesPreterm births (PTBs) represent significant health risks, and several studies have found associations between high outdoor temperatures and PTB. We estimated both the total and natural direct effects (independent of...

Depressive symptoms in the aftermath of major disaster: Empirical test of the social support deterioration model using natural experiment

IntroductionWe examined the extent to which the post-disaster deterioration of social support and social capital mediates the associations between disaster-related experiences and depressive symptoms among survivors to test the social support deterioration model. MethodsWe used unique natural experiment data (n = 3567) stemming from the 2011 Great East Japan Earthquake. The follow-up survey was conducted 2.5 years after the disaster. We examined associations between changes in depressive symptoms, disaster-related experiences (loss of loved ones, changes in household income, loss of job, drop in subjective economic status, and home loss), and post-disaster deterioration of four different types of social support (i.e., kin emotional, kin instrumental, non-kin emotional, and non-kin instrumental support), as well as two dimensions of individual-level social capital (i.e., informal socializing and social cohesion). We then performed causal mediation analysis. ResultsWe found that there was deterioration of social support and social capital among individuals with disaster-related experiences. We also found associations between deterioration of social support/capital and worsening of depressive symptoms for specific categories of disaster experiences (i.e., drop in subjective economic status, loss of job, and home loss). However, causal mediation analysis showed little evidence of mediation by post-disaster deterioration of social support/capital. There was some evidence to suggest exposure-mediator interaction such that the association between social support/capital deterioration and depressive symptoms was weaker among those with disaster experiences. DiscussionThe social support deterioration model was not empirically supported among older-adult disaster survivors of the 2011 Great East Japan Earthquake, after accounting for exposure-mediator interactions.

The association of stillbirth with placental abnormalities in growth-restricted and normally grown fetuses.

Stillbirth, defined as foetal death ≥20weeks' gestation, is associated with poor foetal growth and is often attributed to placental abnormalities, which are also associated with poor foetal growth. Evaluating inter-relationships between placental abnormalities, poor foetal growth, and stillbirth may improve our understanding of the underlying mechanisms for some causes of stillbirth. Our primary objective was to determine whether poor foetal growth, operationalised as small for gestational age (SGA), mediates the relationship between placental abnormalities and stillbirth. We used data from the Stillbirth Collaborative Research Network study, a population-based case-control study conducted from 2006-2008. Our analysis included 266 stillbirths and 1135 livebirths. We evaluated associations of stillbirth with five types of placental characteristics (developmental disorders, maternal and foetal inflammatory responses, and maternal and foetal circulatory disorders) and examined mediation of these relationships by SGA. We also assessed exposure-mediator interaction. Models were adjusted for maternal age, race/ethnicity, education, body mass index, parity, and smoking status. All five placental abnormalities were more prevalent in cases than controls. After adjustment for potential confounders, maternal inflammatory response (odds ratio [OR] 2.58, 95% confidence interval [CI] 1.77, 3.75), maternal circulatory disorders OR 4.14, 95% CI 2.93, 5.84, and foetal circulatory disorders OR 4.58, 95% CI 3.11, 6.74 were strongly associated with stillbirth, and the relationships did not appear to be mediated by SGA status. Associations for developmental disorders and foetal inflammatory response diverged for SGA and non-SGA births, and strong associations were only observed when SGA was not present. Foetal growth did not mediate the relationships between placental abnormalities and stillbirth. The relationships of stillbirth with maternal and foetal circulatory disorders and maternal inflammatory response appear to be independent of poor foetal growth, while developmental disorders and foetal inflammatory response likely interact with foetal growth to affect stillbirth risk.

The ‘double precarity’ of employment insecurity and unaffordable housing and its impact on mental health

This paper describes who is most likely to experience household employment insecurity and housing affordability stress – double precarity – and estimates the degree to which housing affordability mediates the effect of employment insecurity on mental health.We use a cohort of 24,201 participants in 2016 Household, Income and Labour Dynamics in Australia survey (6.2 repeated measures on average). We estimate the likelihood of onset of household employment insecurity, housing affordability stress and change in housing costs using longitudinal regression analyses for socio-demographic groups. We assess mediation by estimating how much exposure variable coefficients attenuate with inclusion of a mediator in fixed effects regression models. We also apply causal mediation methods to fixed-effects regression models to better account for exposure-mediator interaction and meet strict model assumptions.If people's households become insecurely employed, there are five times greater odds of them also experiencing housing affordability stress (OR 4.99 95%CI 4.21–5.90). Key cohorts within the population are shown to be especially vulnerable to double precarity – notably single parents (OR 2.91, 95%CI 1.94–4.35) and people who live alone (OR 4.42, 95% CI 3.03–6.45) (compared to couples), and people who are recently separated or divorced (OR 2.59, 95%CI 1.81–3.70). Mediation analysis confirms that household employment insecurity has a small, negative effect on mental health (Beta −0.24, 95%CI -0.38-0.11 on a 1 to 100-point scale with 10-point standard deviation). Estimates from casual mediation analyses suggest housing affordability accounts for 20% of the total effect; likely concentrated in the lowest and highest strata of income.Employment and housing insecurity represent a form of double precarity for people in households with a single income. When we consider the impact on mental health, we find evidence of a causal relationship between insecure employment onset and mental health, around one fifth of which is mediated by changing housing cost and onset of affordability stress.

Methods for analytic intercategorical intersectionality in quantitative research: Discrimination as a mediator of health inequalities

RationaleIntersectionality as a theoretical framework has gained prominence in qualitative research on social inequity. Intercategorical quantitative applications have focused primarily on describing health or social inequalities across intersectional groups, coded using cross-classified categories or interaction terms. This descriptive intersectionality omits consideration of the mediating processes (e.g., discrimination) through which intersectional positions impact outcome inequalities, which offer opportunities for intervention. ObjectiveWe argue for the importance of a quantitative analytic intersectionality. We identify methodological challenges and potential solutions in structuring studies to allow for both intersectional heterogeneity in outcomes and in the ways that processes such as discrimination may cause these outcomes for those at different intersections. MethodTo incorporate both mediation and exposure-mediator interaction, we use VanderWeele's three-way decomposition methodology, adapt the interpretation for application to analytic intersectionality studies, and present a step-by-step analytic approach. Using online panel data collected from Canada and the United States in 2016 (N = 2542), we illustrate this approach with a statistical analysis of whether and to what extent observed inequalities in psychological distress across intersections of ethnoracial group and sexual or gender minority (SGM) status may be explained by past-year experiences of day-to-day discrimination, assessed using the Intersectional Discrimination Index (InDI). Results and conclusionsWe describe actual and adjusted intersectional inequalities in psychological distress and decompose them to identify three component effects for each of 11 intersectional comparison groups (e.g., Indigenous SGM), versus the reference intersectional group that experienced the lowest levels of discrimination (white non-SGM). These reflect the expected inequality in outcome: 1) due to membership in the more discriminated-against group, if its members had experienced the same lower levels of discrimination as the reference intersection; 2) due to unequal levels of discrimination; and 3), due to unequal effects of discrimination. We present considerations for use and interpretation of these methods.

Indirect Effects of Body Mass Index Growth on Glucose Dysregulation via Inflammation: Causal Moderated Mediation Analysis

Objective: No existing studies have examined the mediating role of chronic inflammation between obesity and dysregulated glucose homeostasis in adolescent samples. This study evaluated whether C-reactive protein (CRP), an inflammation biomarker, mediated the effects of growth (annual increase) in body mass index (BMI) on glycated hemoglobin (HbA_1c). Methods: BMI and biomarker data were used from wave I to wave IV of the National Longitudinal Study of Adolescent to Adult Health (Add Health study; 4,545 adolescents; mean age = 14.9 years; 55.7% female) with valid CRP data. A causal moderated mediation analysis evaluated the direct and indirect effects of BMI slope on HbA_1c via CRP across gender, with demographic and clinical characteristics as model covariates. Results: The participants displayed a linear BMI growth of 0.53–0.58 kg/m²/year throughout adolescence, with substantial interindividual variation. The BMI slope showed positive direct and indirect effects on HbA_1c via CRP across gender, and there was a significant exposure-mediator interaction effect. A standardized increase in the BMI slope raised the probability of an abnormal HbA_1c value by 6.0–8.5% in participants with various profiles. The total natural indirect effect accounted for 13.3–15.9% of the total effect in males and 21.2–22.7% in females. Conclusions: The findings provide support for the inflammation mechanism in the effects of adiposity on glucose homeostasis. In adolescents, excess BMI growth was linked with a higher risk of glucose dysregulation either directly or indirectly via chronic inflammation.

High Temperature Associations with Preterm Birth and Mediation by Air Pollutants in Detroit, Michigan, 1991-2001.

Introduction: Preterm births (PTBs) cause significant infant health risks, and several studies have found associations between high temperatures and PTB. To understand this association in Detroit, Michigan and the extent to which temperature effects are mediated by air pollution, we evaluated both the total and natural direct effects (independent of particulate matter, ozone, and nitrogen dioxide) of two-day mean apparent temperature (AT) on PTB for the warm months (May to September), 1991-2001. We evaluated effect modification by maternal age, race, education, smoking status, and prenatal care.Methods: We used a time-stratified case-crossover design with splines of AT and wind speed (National Center for Environmental Information), solar radiation (National Renewable Energy Laboratory), and citywide average precipitation (Oregon State University’s Parameter-elevation Regression on Independent Slopes Model) to estimate total effects. To accommodate multiple mediators and exposure-mediator interactions, AT inverse odds weights, predicted by meteorological and air pollutant (Environmental Protection Agency AirData) covariates, were included in a subsequent case-crossover model to estimate direct effects.Results: At 24.9 deg. C (95th percentile of warm-season AT) vs. 18.6 deg. C (50th percentile of warm-season AT), 18.7% (95% CI: 1.3%, 31.4%) of PTBs were attributable to the total effects of AT and 11.1% (95% CI: 3.5%, 17.5%) to direct effects. Relative excess risks of interaction were null for each potential modifier.Conclusion: This additional evidence of a direct association between high temperature and PTB motivates public health interventions to reduce extreme heat exposures among pregnant women.

Decomposition of the Total Effect in the Presence of Multiple Mediators and Interactions.

Mediation analysis allows decomposing a total effect into a direct effect of the exposure on the outcome and an indirect effect operating through a number of possible hypothesized pathways. Recent studies have provided formal definitions of direct and indirect effects when multiple mediators are of interest and have described parametric and semiparametric methods for their estimation. Investigating direct and indirect effects with multiple mediators, however, can be challenging in the presence of multiple exposure-mediator and mediator-mediator interactions. In this paper we derive a decomposition of the total effect that unifies mediation and interaction when multiple mediators are present. We illustrate the properties of the proposed framework in a secondary analysis of a pragmatic trial for the treatment of schizophrenia. The decomposition is employed to investigate the interplay of side effects and psychiatric symptoms in explaining the effect of antipsychotic medication on quality of life in schizophrenia patients. Our result offers a valuable tool to identify the proportions of total effect due to mediation and interaction when more than one mediator is present, providing the finest decomposition of the total effect that unifies multiple mediators and interactions.

Does adiposity mediate the relationship between physical activity and biological risk factors in youth?: a cross-sectional study from the International Children's Accelerometry Database (ICAD).

To model the association between accumulating 60 daily minutes of moderate-to-vigorous physical activity and a composite score of biological risk factors into a direct and an indirect effect, using abdominal obesity as the mediator. Cross-sectional data from the International Children's Accelerometry Database (ICAD) including 6-18-year-old children and adolescents (N=3412) from 4 countries providing at least 3 days of accelerometry-assessed physical activity. A standardized composite risk score was calculated from systolic blood pressure and fasting blood samples of insulin, glucose, triacylglycerol and inverse HDL-cholesterol. Abdominal obesity was assessed by the waist-circumference:height ratio. Two-stage regression analysis, allowing for exposure-mediator interaction, was used for the effect decomposition. Participants achieving 60 daily minutes of moderate-to-vigorous physical activity had a 0.31 (95% CI: -0.39, -0.23) standard deviations lower composite risk score than those achieving less than 60 min. Modelling the associations suggested that 0.24 standard deviations (95% CI: -0.32, -0.16) was attributed to the direct effect and -0.07 (95% CI: -0.11, -0.02) to the indirect effect indicating that 22% of the total effect was mediated by central adiposity. Modelling 30 and 90 min of moderate-to-vigorous physical activity per day resulted in changes in the direct but not the indirect effect. One hour of daily moderate-to-vigorous physical activity was associated with clinically relevant differences in metabolic control compared to engagement in less than this minimally recommended amount. The majority of the difference was explained by the direct effect of physical activity.International Journal of Obesity advance online publication, 31 October 2017; doi:10.1038/ijo.2017.241.

Quantile causal mediation analysis allowing longitudinal data.

Mediation analysis has mostly been conducted with mean regression models. With this approach modeling means, formulae for direct and indirect effects are based on changes in means, which may not capture effects that occur in units at the tails of mediator and outcome distributions. Individuals with extreme values of medical endpoints are often more susceptible to disease and can be missed if one investigates mean changes only. We derive the controlled direct and indirect effects of an exposure along percentiles of the mediator and outcome using quantile regression models and a causal framework. The quantile regression models can accommodate an exposure-mediator interaction and random intercepts to allow for longitudinal mediator and outcome. Because DNA methylation acts as a complex "switch" to control gene expression and fibrinogen is a cardiovascular factor, individuals with extreme levels of these markers may be more susceptible to air pollution. We therefore apply this methodology to environmental data to estimate the effect of air pollution, as measured by particle number, on fibrinogen levels through a change in interferon-gamma (IFN-γ) methylation. We estimate the controlled direct effect of air pollution on the qth percentile of fibrinogen and its indirect effect through a change in the pth percentile of IFN-γ methylation. We found evidence of a direct effect of particle number on the upper tail of the fibrinogen distribution. We observed a suggestive indirect effect of particle number on the upper tail of the fibrinogen distribution through a change in the lower percentiles of the IFN-γ methylation distribution.

Interventional Approach for Path-Specific Effects

Abstract Standard causal mediation analysis decomposes the total effect into a direct effect and an indirect effect in settings with only one single mediator. Under the settings with multiple mediators, all mediators are often treated as one single block of mediators. The effect mediated by a certain combination of mediators, i. e. path-specific effect (PSE), is not always identifiable without making strong assumptions. In this paper, the authors propose a method, defining a randomly interventional analogue of PSE (rPSE), as an alternative approach for mechanism investigation. This method is valid under assumptions of no unmeasured confounding and allows settings with mediators dependent on each other, interaction, and mediator-outcome confounders which are affected by exposure. In addition, under linearity and no-interaction, our method has the same form of traditional path analysis for PSE. Furthermore, under single mediator without a mediator-outcome confounder affected by exposure, it also has the same form of the results of causal mediation analysis. We also provide SAS code for settings of linear regression with exposure-mediator interaction and perform analysis in the Framingham Heart Study dataset, investigating the mechanism of smoking on systolic blood pressure as mediated by both cholesterol and body weight. Allowing decomposition of total effect into several rPSEs, our method contributes to investigation of complicated causal mechanisms in settings with multiple mediators.