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

Model Selection for Exposure-Mediator Interaction

In mediation analysis, the exposure often influences the mediating effect, i.e., there is an interaction between exposure and mediator on the dependent variable. When the mediator is high-dimensional, it is necessary to identify non-zero mediatorsM and exposure-by-mediator ( X -by- M ) interactions. Although several high-dimensional mediation methods can naturally handle X -by- M interactions, research is scarce in preserving the underlying hierarchical structure between the main effects and the interactions. To fill the knowledge gap, we develop the XMInt procedure to select M and X -by- M interactions in the high-dimensional mediators setting while preserving the hierarchical structure. Our proposed method employs a sequential regularization-based forward-selection approach to identify the mediators and their hierarchically preserved interaction with exposure. Our numerical experiments showed promising selection results. Further, we applied our method to ADNI morphological data and examined the role of cortical thickness and subcortical volumes on the effect of amyloid-beta accumulation on cognitive performance, which could be helpful in understanding the brain compensation mechanism.

Multi-Trait Body Shape Phenotypes and Breast Cancer Risk in Postmenopausal Women: A Causal Mediation Analysis in the UK Biobank Cohort

Body shape phenotypes combining multiple anthropometric traits have been linked to postmenopausal breast cancer (BC). However, underlying biological pathways remain poorly understood. This study investigated to what extent the associations of body shapes with postmenopausal BC risk is mediated by biochemical markers. The study included 176,686 postmenopausal women from UK Biobank. Four body shape phenotypes were derived from principal component (PC) analysis of height, weight, body mass index, waist and hip circumferences, and waist-to-hip ratio (WHR). The four-way decomposition of the total effect was used to estimate mediation and interaction effects simultaneously as well as the mediated proportions. After 10.9 years median follow-up, 6,396 incident postmenopausal BC were diagnosed. There was strong evidence of positive associations between PC1 (general obesity) and PC2 (tall, low WHR), and BC risk. The association of PC1 with BC risk was positively mediated by testosterone and negatively by insulin-like growth factor-1 (IGF-1), with the overall proportion mediated (sum of the mediated interaction and pure indirect effect (PIE)) accounting for 11.4% (95% confidence intervals: 5.1 to 17.8%) and -12.2% (-20.5% to -4.0%) of the total effect, respectively. Small proportions of the association between PC2 and BC were mediated by IGF-1 (PIE: 2.8% (0.6 to 4.9%)), and sex hormone-binding globulin (SHBG) (PIE: -6.1% (-10.9% to -1.3%)). Our findings are consistent with differential pathways linking different body shapes with BC risk, with a suggestive mediation through testosterone and IGF-1 in the relationship of a generally obese body shape and BC risk, while IGF-1 and SHBG may mediate a tall/lean body shape-BC risk association.

Racial and ethnic disparities in preterm birth: a mediation analysis incorporating mixtures of polybrominated diphenyl ethers

Background

Racial and ethnic disparities persist in preterm birth (PTB) and gestational age (GA) at delivery in the United States. It remains unclear whether exposure to environmental chemicals contributes to these disparities.

Objectives

We applied recent methodologies incorporating environmental mixtures as mediators in causal mediation analysis to examine whether racial and ethnic disparities in GA at delivery and PTB may be partially explained by exposures to polybrominated diphenyl ethers (PBDEs), a class of chemicals used as flame retardants in the United States.

Methods

Data from a multiracial/ethnic US cohort of 2008 individuals with low-risk singleton pregnancies were utilized, with plasma PBDE concentrations measured during early pregnancy. We performed mediation analyses incorporating three forms of mediators: (1) reducing all PBDEs to a weighted index, (2) selecting a PBDE congener, or (3) including all congeners simultaneously as multiple mediators, to evaluate whether PBDEs may contribute to the racial and ethnic disparities in PTB and GA at delivery, adjusted for potential confounders.

Results

Among the 2008 participants, 552 self-identified as non-Hispanic White, 504 self-identified as non-Hispanic Black, 568 self-identified as Hispanic, and 384 self-identified as Asian/Pacific Islander. The non-Hispanic Black individuals had the highest mean ∑PBDEs, the shortest mean GA at delivery, and the highest rate of PTB. Overall, the difference in GA at delivery comparing non-Hispanic Black to non-Hispanic White women was -0.30 (95% CI: -0.54, -0.05) weeks. This disparity reduced to -0.23 (95% CI: -0.49, 0.02) and -0.18 (95% CI: -0.46, 0.10) weeks if fixing everyone's weighted index of PBDEs to the median and the 25th percentile levels, respectively. The proportion of disparity mediated by the weighted index of PBDEs was 11.8%. No statistically significant mediation was found for PTB, other forms of mediator(s), or other racial and ethnic groups.

Conclusion

PBDE mixtures may partially mediate the Black vs. White disparity in GA at delivery. While further validations are needed, lowering the PBDEs at the population level might help reduce this disparity.

Assessment of the prognostic value of preoperative high-sensitive troponin T for myocardial injury and long-term mortality for groups at high risk for cardiovascular events following noncardiac surgery: a retrospective cohort study

Background

Few studies explored the association between high-sensitive cardiac troponin T (hs-cTnT) and long-term mortality for patients after surgery. This study was conducted to assess the association of hs-cTnT with long-term mortality and to investigate the extent to which this association is mediated via myocardial injury after noncardiac surgery (MINS).

Methods

This retrospective cohort study included all patients with hs-cTnT measurements who underwent non-cardiac surgery at Sichuan University West China Hospital. Data were collected from February 2018 and November 2020, with follow-up through February 2022. The primary outcome was all-cause mortality within 1 year. As secondary outcomes, MINS, length of hospital stay (LOS), and ICU admission were analyzed.

Results

The cohort included 7,156 patients (4,299 [60.1%] men; 61.0 [49.0-71.0] years). Among 7,156 patients, there were 2,151 (30.05%) with elevated hs-cTnT(>14 ng/L). After more than 1 year of follow-up, more than 91.8% of mortality information was available. During one-year follow-up after surgery, there were 308 deaths (14.8%) with a preoperative hs-cTnT >14 ng/L, compared with 192 deaths (3.9%) with a preoperative hs-cTnT <=14 ng/L(adjusted hazard ratio [aHR] 1.93, 95% CI 1.58-2.36; p < 0.001). Elevated preoperative hs-cTnT was also associated with several other adverse outcomes (MINS: adjusted odds ratio [aOR] 3.01; 95% CI, 2.46-3.69; p < 0.001; LOS: aOR 1.48, 95%CI 1.34-1.641; p < 0.001; ICU admission: aOR 1.52, 95%CI 1.31-1.76; p < 0.001). MINS explained approximately 33.6% of the variance in mortality due to preoperative hs-cTnT levels.

Conclusion

Preoperative elevated hs-cTnT concentrations have a significant association with long-term mortality after noncardiac surgery, one-third of which may by accounted for by MINS.

Bayesian estimation of potential outcomes for mediation analysis of racial disparity for infant mortality

Background

There is a need for novel methods to determine preventable causes of racial health disparities. This need has been met with the development of improved methods for mediation modeling. Current mediational analysis methods call for an evaluation of statistical interaction or effect modification between the investigated cause and mediator. For racial disparity, this approach facilitates the estimation of racially specific risks for infant mortality. However, current methods for evaluating multiple interacting mediators are inadequate. The first objective of the study was to compare Bayesian estimation of potential outcomes to other approaches to mediation analysis that included interaction. The second objective was to evaluate three potentially interacting mediators of racial disparity for infant mortality by modeling the large dataset from the National Natality Database using Bayesian estimation of potential outcomes.

Methods

A random sample of observations from the 2003 National Natality Database was used to compare the currently promoted methods for mediation modeling. Racial disparity was modeled as a separate function for each of three potential mediators, (i) maternal smoking, (ii) low birth weight and (iii) teenage maternity. As a second objective, direct Bayesian estimation of potential outcomes modeled infant mortality as function of the interactions among the three mediators and race using the full National Natality Database for the years 2016 to 2018.

Results

The counterfactual model was inaccurate in estimating the proportion of racial disparity that was attributable to either maternal smoking or teenage maternity. The counterfactual approach did not accurately estimate the probabilities defined by counterfactual definitions. The error was a result of modeling the excess relative risk instead of the risk probabilities. Bayesian approaches did estimate the probabilities of the counterfactual definitions. Results showed that 73% of the racial disparity for infant mortality was attributed to infants born with low birth weight.

Conclusions

Bayesian estimation of potential outcomes could evaluate whether proposed public health programs would affect races differently and decisions could include consideration of the causal effect the program may have on racial disparity. The large contribution of low birth weight to racial disparity for infant mortality should be further investigated to identify preventable factors for low birth weight.

Residential greenness and dyslipidemia risk: Dose-response relations and mediation through BMI and air pollution

BACKGROUND

Evidence on associations of residential greenness with dyslipidemia is limited, particularly regarding dose-response relations and mediation.

OBJECTIVES

To investigate associations between greenness and dyslipidemia, non-linear dose-response relationships and mediators.

METHODS

This cross-sectional study draws on the 2018 Fujian Behavior and Disease Surveillance (FBDS) cohort that used multi-stage stratified random sampling from the general population of Fujian Province, China. Participants with one or more abnormities in total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), or low-density lipoprotein cholesterol (LDL-C) were classified as having dyslipidemia. Residential greenness was operationalized as 3-year average of the normalized difference vegetation index (NDVI_500m) and enhanced vegetation index (EVI_500m). A doubly robust approach was used for effect quantification. Dose-response relations were studied with natural cubic splines. Mediation via physical activity (PA), body mass index (BMI), PM_2.5, PM₁₀, SO₂, and NO₂ was also examined.

RESULTS

Data from 43,183 participants were analyzed. Increases in NDVI_500m and EVI_500m residential greenness were associated with decreased dyslipidemia risk and improved blood lipids. Non-linear dose response relationships were discovered. Significant reduction of dyslipidemia risk was observed at levels of EVI_500m > 0.48 and NDVI_500m > 0.65. Joint mediation effects of PA, BMI, PM_2.5, PM₁₀, NO₂, and SO₂ on the associations of NDVI_500m and EVI_500m with dyslipidemia risk were 49.74% and 44.64%, respectively.

CONCLUSIONS

Increased residential exposure to greenness was associated with decreased risk of dyslipidemia. A non-linear dose-response relationship between greenness and dyslipidemia suggests that specific thresholds of greenness need to be reached in order to achieve effects. BMI, PM_2.5, and PM₁₀ partially mediated the association.

Associations and pathways between residential greenness and metabolic syndromes in Fujian Province

Background

Greenness exposure is beneficial to human health, but its potential mechanisms through which the risk for metabolic syndrome (MetS) could be reduced have been poorly studied. We aimed to estimate the greenness-MetS association in southeast China and investigate the independent and joint mediation effects of physical activity (PA), body mass index (BMI), and air pollutants on the association.

Methods

A cross-sectional study was conducted among the 38,288 adults based on the Fujian Behavior and Disease Surveillance (FBDS), established in 2018. MetS was defined as the presence of three or more of the five components: abdominal obesity, elevated triglyceride, reduced high-density lipoprotein cholesterol (HDL-C), high blood pressure, and elevated fasting glucose. The residential greenness exposure was measured as the 3-year mean values of the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) within the 250, 500, and 1,000 meters (m) buffer zones around the residential address of each participant. Logistic regression models were used to estimate the greenness-MetS association. The causal mediation analysis was used to estimate the independent and joint mediation effects of PA, BMI, particulate matter with an aerodynamic diameter of 2.5 μm (PM_2.5), particulate matter with an aerodynamic diameter ≤ 10 μm (PM₁₀), nitrogen dioxide (NO₂), and sulfur dioxide (SO₂).

Results

Each interquartile range (IQR) increase in greenness was associated with a decrease of 13% (OR = 0.87 [95%CI: 0.83, 0.92] for NDVI_500m and OR = 0.87 [95%CI: 0.82, 0.91] for EVI_500m) in MetS risk after adjusting for covariates. This association was stronger in those aged < 60 years (e.g., OR = 0.86 [95%CI: 0.81, 0.92] for NDVI_500m), males (e.g., OR = 0.73 [95%CI: 0.67, 0.80] for NDVI_500m), having an educational level of primary school or above (OR = 0.81 [95%CI: 0.74, 0.89] for NDVI_500m), married/cohabitation (OR = 0.86 [95%CI: 0.81, 0.91] for NDVI_500m), businessman (OR = 0.82 [95%CI: 0.68, 0.99] for NDVI_500m), other laborers (OR = 0.77 [95%CI: 0.68, 0.88] for NDVI_500m), and non-smokers (OR = 0.77 [95%CI: 0.70, 0.85] for NDVI_500m). The joint effect of all six mediators mediated about 48.1% and 44.6% of the total effect of NDVI_500m and EVI_500m on the MetS risk, respectively. Among them, BMI showed the strongest independent mediation effect (25.0% for NDVI_500m), followed by NO₂ and PM₁₀.

Conclusion

Exposure to residential greenness was associated with a decreased risk for MetS. PA, BMI, and the four air pollutants jointly interpreted nearly half of the mediation effects on the greenness-MetS association.

Applied causal inference methods for sequential mediators

BACKGROUND

Mediation analysis aims at estimating to what extent the effect of an exposure on an outcome is explained by a set of mediators on the causal pathway between the exposure and the outcome. The total effect of the exposure on the outcome can be decomposed into an indirect effect, i.e. the effect explained by the mediators jointly, and a direct effect, i.e. the effect unexplained by the mediators. However finer decompositions are possible in presence of independent or sequential mediators.

METHODS

We review four statistical methods to analyse multiple sequential mediators, the inverse odds ratio weighting approach, the inverse probability weighting approach, the imputation approach and the extended imputation approach. These approaches are compared and implemented using a case-study with the aim to investigate the mediating role of adverse reproductive outcomes and infant respiratory infections in the effect of maternal pregnancy mental health on infant wheezing in the Ninfea birth cohort.

RESULTS

Using the inverse odds ratio weighting approach, the direct effect of maternal depression or anxiety in pregnancy is equal to a 59% (95% CI: 27%,94%) increased prevalence of infant wheezing and the mediated effect through adverse reproductive outcomes is equal to a 3% (95% CI: -6%,12%) increased prevalence of infant wheezing. When including infant lower respiratory infections in the mediation pathway, the direct effect decreases to 57% (95% CI: 25%,92%) and the indirect effect increases to 5% (95% CI: -5%,15%). The estimates of the effects obtained using the weighting and the imputation approaches are similar. The extended imputation approach suggests that the small joint indirect effect through adverse reproductive outcomes and lower respiratory infections is due entirely to the contribution of infant lower respiratory infections, and not to an increased prevalence of adverse reproductive outcomes.

CONCLUSIONS

The four methods revealed similar results of small mediating role of adverse reproductive outcomes and early respiratory tract infections in the effect of maternal pregnancy mental health on infant wheezing. The choice of the method depends on what is the effect of main interest, the type of the variables involved in the analysis (binary, categorical, count or continuous) and the confidence in specifying the models for the exposure, the mediators and the outcome.

Four-way decomposition of effect of cigarette smoking and body mass index on serum lipid profiles

Objective

Smoking and obesity are established risk factors of dyslipidemia, however, the interplay between them has not been well studied. This study aims to explore the joint effect of smoking and body mass index (BMI) on serum lipid profiles.

Methods

The study consisted of 9846 Chinese adults (mean age = 49.9 years, 47.6% males, 31.2% ever smokers), based on the China Health and Nutrition Survey. Serum lipid profiles included total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), apolipoprotein A (APO-A), and apolipoprotein B (Apo-B). The joint effect of smoking and BMI on serum lipids were examined by the four-way decomposition analysis and multivariate linear regression models.

Results

The four-way decomposition showed that the interplay between smoking and BMI was complicated. There was only indirect effect (the mediated effect) between smoking and BMI on TC, LDL-C and APO-B. The pure indirect effect was -0.023 for TC, -0.018 for LDL-C, and -0.009 for APO-B. For TG, HDL-C and APO-A, the interaction effect was dominant. The reference interaction (the interactive effect when the mediator is left to what it would be in the absence of exposure) was 0.474 (P < 0.001) for TG, -0.245 (P = 0.002) for HDL-C, and -0.222 (P < 0.001) for APO-A, respectively. The effect of BMI on TG, HDL-C and APO-A were significantly higher in smokers than in nonsmokers (TG: 0.151 in smokers versus 0.097 in nonsmokers, HDL-C: -0.037 versus -0.027, APO-A: -0.019 versus -0.009, P for difference < 0.001 for all).

Conclusion

These findings illustrate the joint effects of smoking and BMI on serum lipid profiles. There were significant interaction effects of smoking and BMI on TG, HDL-C and APO-A, while BMI maybe a mediator for the association of smoking with TC, LDL-C and APO-B. The effects between them were rather complex. Smoking cessation is necessary, especially for those overweight.

Associations between long-term drought and diarrhea among children under five in low- and middle-income countries

Climate change is projected to intensify drought conditions, which may increase the risk of diarrheal diseases in children. We constructed log-binomial generalized linear mixed models to examine the association between diarrhea risk, ascertained from global-scale nationally representative Demographic and Health Surveys, and drought, represented by the standardized precipitation evapotranspiration index, among children under five in 51 low- and middle-income countries (LMICs). Exposure to 6-month mild or severe drought was associated with an increased diarrhea risk of 5% (95% confidence interval 3-7%) or 8% (5-11%), respectively. The association was stronger among children living in a household that needed longer time to collect water or had no access to water or soap/detergent for handwashing. The association for 24-month drought was strong in dry zones but weak or null in tropical or temperate zones, whereas that for 6-month drought was only observed in tropical or temperate zones. In this work we quantify the associations between exposure to long-term drought and elevated diarrhea risk among children under five in LMICs and suggest that the risk could be reduced through improved water, sanitation, and hygiene practices, made more urgent by the likely increase in drought due to climate change.

Path-specific effects in the presence of a survival outcome and causally ordered multiple mediators with application to genomic data

Causal multimediation analysis (i.e. the causal mediation analysis with multiple mediators) is critical for understanding the effectiveness of interventions, especially in medical research. Deriving the path-specific effects of exposure on the outcome through a set of mediators can provide detail about the causal mechanism of interest However, existing models are usually restricted to partial decomposition, which can only be used to evaluate the cumulative effect of several paths. In genetics studies, partial decomposition fails to reflect the real causal effects mediated by genes, especially in complex gene regulatory networks. Moreover, because of the lack of a generalized identification procedure, the current multimediation analysis cannot be applied to the estimation of path-specific effects for any number of mediators. In this study, we derive the interventional analogs of path-specific effect for complete decomposition to address the difficulty of nonidentifiability. On the basis of two survival models of the outcome, we derive the generalized analytic forms for interventional analogs of path-specific effects by assuming the normal distributions of mediators. We apply the new methodology to investigate the causal mechanism of signature genes in lung cancer based on the cell cycle pathway, and the results clarify the gene pathway in cancer.

Assessing proximate intermediates between ambient temperature, hospital admissions, and mortality in hemodialysis patients

BACKGROUND

Typical thermoregulatory responses to elevated temperatures among healthy individuals include reduced blood pressure and perspiration. Individuals with end-stage kidney disease (ESKD) are susceptible to systemic fluctuations caused by ambient temperature changes that may increase morbidity and mortality. We investigated whether pre-dialysis systolic blood pressure (preSBP) and interdialytic weight gain (IDWG) can independently mediate the association between ambient temperature, all-cause hospital admissions (ACHA), and all-cause mortality (ACM).

METHODS

The study population consisted of ESKD patients receiving hemodialysis treatments at Fresenius Medical Care facilities in Philadelphia County, PA, from 2011 to 2019 (n = 1981). Within a time-to-event framework, we estimated the association between daily maximum dry-bulb temperature (TMAX) and, as separate models, ACHA and ACM during warmer calendar months. Clinically measured preSBP and IDWG responses to temperature increases were estimated using linear mixed effect models. We employed the difference (c-c') method to decompose total effect models for ACHA and ACM using preSBP and IDWG as time-dependent mediators. Covariate adjustments for exposure-mediator and total and direct effect models include age, race, ethnicity, blood pressure medication use, treatment location, preSBP, and IDWG. We considered lags up to two days for exposure and 1-day lag for mediator variables (Lag 2-Lag 1) to assure temporality between exposure-outcome models. Sensitivity analyses for 2-day (Lag 2-only) and 1-day (Lag 1-only) lag structures were also conducted.

RESULTS

Based on Lag 2- Lag 1 temporal ordering, 1 °C increase in daily TMAX was associated with increased hazard of ACHA by 1.4% (adjusted hazard ratio (HR), 1.014; 95% confidence interval, 1.007-1.021) and ACM 7.5% (adjusted HR, 1.075, 1.050-1.100). Short-term lag exposures to 1 °C increase in temperature predicted mean reductions in IDWG and preSBP by 0.013-0.015% and 0.168-0.229 mmHg, respectively. Mediation analysis for ACHA identified significant indirect effects for all three studied pathways (preSBP, IDWG, and preSBP + IDWG) and significant indirect effects for IDWG and conjoined preSBP + IDWG pathways for ACM. Of note, only 1.03% of the association between temperature and ACM was mediated through preSBP. The mechanistic path for IDWG, independent of preSBP, demonstrated inconsistent mediation and, consequently, potential suppression effects in ACHA (-15.5%) and ACM (-6.3%) based on combined pathway models. Proportion mediated estimates from preSBP + IDWG pathways achieved 2.2% and 0.3% in combined pathway analysis for ACHA and ACM outcomes, respectively. Lag 2 discrete-time ACM mediation models exhibited consistent mediation for all three pathways suggesting that 2-day lag in IDWG and preSBP responses can explain 2.11% and 4.41% of total effect association between temperature and mortality, respectively.

CONCLUSION

We corroborated the previously reported association between ambient temperature, ACHA and ACM. Our results foster the understanding of potential physiological linkages that may explain or suppress temperature-driven hospital admissions and mortality risks. Of note, concomitant changes in preSBP and IDWG may have little intermediary effect when analyzed in combined pathway models. These findings advance our assessment of candidate interventions to reduce the impact of outdoor temperature change on ESKD patients.

Decomposition of the total effect for two mediators: A natural mediated interaction effect framework

Mediation analysis has been used in many disciplines to explain the mechanism or process that underlies an observed relationship between an exposure variable and an outcome variable via the inclusion of mediators. Decompositions of the total effect (TE) of an exposure variable into effects characterizing mediation pathways and interactions have gained an increasing amount of interest in the last decade. In this work, we develop decompositions for scenarios where two mediators are causally sequential or non-sequential. Current developments in this area have primarily focused on either decompositions without interaction components or with interactions but assuming no causally sequential order between the mediators. We propose a new concept called natural mediated interaction (MI) effect that captures the two-way and three-way interactions for both scenarios and extends the two-way MIs in the literature. We develop a unified approach for decomposing the TE into the effects that are due to mediation only, interaction only, both mediation and interaction, neither mediation nor interaction within the counterfactual framework. Finally, we compare our proposed decomposition to an existing method in a non-sequential two-mediator scenario using simulated data, and illustrate the proposed decomposition for a sequential two-mediator scenario using a real data analysis.

Mediation analysis methods used in observational research: a scoping review and recommendations

Background

Mediation analysis methodology underwent many advancements throughout the years, with the most recent and important advancement being the development of causal mediation analysis based on the counterfactual framework. However, a previous review showed that for experimental studies the uptake of causal mediation analysis remains low. The aim of this paper is to review the methodological characteristics of mediation analyses performed in observational epidemiologic studies published between 2015 and 2019 and to provide recommendations for the application of mediation analysis in future studies.

Methods

We searched the MEDLINE and EMBASE databases for observational epidemiologic studies published between 2015 and 2019 in which mediation analysis was applied as one of the primary analysis methods. Information was extracted on the characteristics of the mediation model and the applied mediation analysis method.

Results

We included 174 studies, most of which applied traditional mediation analysis methods (n = 123, 70.7%). Causal mediation analysis was not often used to analyze more complicated mediation models, such as multiple mediator models. Most studies adjusted their analyses for measured confounders, but did not perform sensitivity analyses for unmeasured confounders and did not assess the presence of an exposure-mediator interaction.

Conclusions

To ensure a causal interpretation of the effect estimates in the mediation model, we recommend that researchers use causal mediation analysis and assess the plausibility of the causal assumptions. The uptake of causal mediation analysis can be enhanced through tutorial papers that demonstrate the application of causal mediation analysis, and through the development of software packages that facilitate the causal mediation analysis of relatively complicated mediation models.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01426-3.

Mediational g-formula for time-varying treatment and repeated-measured multiple mediators: Application to atorvastatin's effect on cardiovascular disease via cholesterol lowering and anti-inflammatory actions in elderly type 2 diabetics

Modern causal mediation theory has formalized several types of indirect and direct effects of treatment on outcomes regarding specific mediator variables. We reviewed and unified distinct approaches to estimate the "interventional" direct and indirect effects for multiple mediators and time-varying variables. This study was motivated by a clinical trial of elderly type-2 diabetic patients in which atorvastatin was widely prescribed to control patients' cholesterol levels to reduce diabetic complications, including cardiovascular disease. Among atorvastatin's preventive side-effects (pleiotropic effects), we focus on its anti-inflammatory action as measured by white blood cell counts. Hence, we estimate atorvastatin's interventional indirect effects through cholesterol lowering and through anti-inflammatory action, and interventional direct effect bypassing these two actions. In our analysis, total effect (six-year cardiovascular disease risk difference) estimated by standard plug-in g-formula of -3.65% (95% confidence interval: -10.29%, 4.38%) is decomposed into indirect effect via low-density lipoprotein cholesterol (-0.90% [-1.91%, -0.07%]), via white blood cell counts (-0.03% [-0.22%, 0.11%]), and direct effect (-2.84% [-9.71%, 5.41%]) by the proposed parametric mediational g-formula. The SAS program and its evaluation via simulated datasets are provided in the Supplemental materials.

First trimester mechanisms of gestational sac placental and foetal teratogenicity: a framework for birth cohort studies

BACKGROUND

The function of the gestational sac (GS) and the placenta in the closely related processes of embryogenesis and teratogenicity in the first trimester has been minimally described. The prevailing assumption is that direct teratogenic effects are mediated by the critical extraembryonic organ, the placenta, which either blocks or transfers exposures to the foetus. Placental transfer is a dominant mechanism, but there are other paradigms by which the placenta can mediate teratogenic effects. Knowledge of these paradigms and first trimester human developmental biology can be useful to the epidemiologist in the conduct of biomarker-based studies of both maternal and child health.

OBJECTIVE AND RATIONALE

Our aim is to provide a causal framework for modelling the teratogenic effects of first trimester exposures on child health outcomes mediated by the GS and placenta using biomarker data collected in the first trimester. We initially present first trimester human developmental biology for the sake of informing and strengthening epidemiologic approaches. We then propose analytic approaches of modelling placental mechanisms by way of causal diagrams using classical non-embryolethal teratogens (diethylstilboestrol [DES], folic acid deficiency and cytomegalovirus [CMV]) as illustrative examples. We extend this framework to two chronic exposures of particular current interest, phthalates and maternal adiposity.

SEARCH METHODS

Information on teratogens was identified by a non-systematic, narrative review. For each teratogen, we included papers that answered the five following questions: (i) why were these exposures declared teratogens? (ii) is there a consensus on biologic mechanism? (iii) is there reported evidence of a placental mechanism? (iv) can we construct a theoretical model of a placental mechanism? and (v) can this knowledge inform future work on measurement and modelling of placental-foetal teratogenesis? We prioritized literature specific to human development, the organogenesis window in the first trimester and non-embryolethal mechanisms.

OUTCOMES

As a result of our review of the literature on five exposures considered harmful in the first trimester, we developed four analytic strategies to address first trimester placental mechanisms in birth cohort studies: placental transfer and direct effects on the foetus (DES and maternal adiposity), indirect effects through targeted placental molecular pathways (DES and phthalates), pre-placental effects through disruptions in embryonic and extraembryonic tissue layer differentiation (folic acid deficiency), and multi-step mechanisms that involve maternal, placental and foetal immune function and inflammation (DES and CMV).

WIDER IMPLICATIONS

The significance of this review is to offer a causal approach to classify the large number of potentially harmful exposures in pregnancy when the exposure occurs in the first trimester. Our review will facilitate future research by advancing knowledge of the first trimester mechanisms necessary for researchers to effectively associate environmental exposures with child health outcomes.

Meta-analysis of Total Effect Decomposition in the Presence of Multiple Mediators: The Example of Schizophrenia Treatment

BACKGROUND

Causal mediation analysis addresses mechanistic questions by decomposing and quantifying effects operating through different pathways. Because most individual studies are underpowered to detect mediating effects, we outlined a parametric approach to meta-analyzing causal mediation and interaction analyses with multiple mediators, compared it with a bootstrap-based alternative, and discussed its limitations.

METHODS

We employed fixed- and random-effects multivariate meta-analyses to integrate evidence on treatment-mediators and mediators-outcome associations across trials. We estimated path-specific effects as functions of meta-analyzed regression coefficients; we obtained standard errors using the delta method. We evaluated the performance of this approach in simulations and applied it to assess the mediating roles of positive symptoms of schizophrenia and weight gain in the treatment effect of paliperidone ER on negative symptoms across four efficacy trials.

RESULTS

Both simulations and the application showed that the meta-analytic approaches increased statistical power. In the application, we observed substantial mediating effects of positive symptoms (proportions mediated from fixed-effects meta-analysis: (Equation is included in full-text article.)). Weight gain may have beneficial mediating effects; however, such benefit may disappear at high doses when metabolic side effects were excessive.

CONCLUSIONS

Meta-analyzing causal mediation analysis combines evidence from multiple sources and improves power. Targeting positive symptoms may be an effective way to reduce negative symptoms that are challenging to treat. Future work should focus on extending the existing methods to allow for more flexible modeling of mediation.

Latent class mediator for multiple indicators of mediation

This paper demonstrates the utility of latent classes in evaluating the effect of an intervention on an outcome through multiple indicators of mediation. These indicators are observed intermediate variables that identify an underlying latent class mediator, with each class representing a different mediating pathway. The use of a latent class mediator allows us to avoid modeling the complex interactions between the multiple indicators and ensures the decomposition of the total mediating effects into additive effects from individual mediating pathways, a desirable feature for evaluating multiple indicators of mediation. This method is suitable when the goal is to estimate the total mediating effects that can be decomposed into the additive effects of distinct mediating pathways. Each indicator may be involved in multiple mediation pathways and at the same time multiple indicators may contribute to a single mediating pathway. The relative importance of each pathway may vary across subjects. We applied this method to the analysis of the first 6 months of data from a 2-year clustered randomized trial for adults in their first episode of schizophrenia. Four indicators of mediation are considered: individual resiliency training; family psychoeducation; supported education and employment; and a structural assessment for medication. The improvement in symptoms was found to be mediated by the latent class mediator derived from these four service indicators. Simulation studies were conducted to assess the performance of the proposed model and showed that the simultaneous estimation through the maximum likelihood yielded little bias when the entropy of the indicators was high.

Joint and interactive effects between health comorbidities and environmental exposures in predicting amyotrophic lateral sclerosis

BACKGROUND

Amyotrophic lateral sclerosis (ALS) is a rare yet devastating neurodegenerative condition. The mechanisms leading to ALS are most certainly complex and likely involve a joint contribution of several factors with possible synergistic or antagonistic interactions. To provide a better understanding of the association between non-genetic factors and ALS, we evaluated the joint exposure to multiple health and environmental factors linked with ALS in our previous studies, also screening for high-dimensional interactions.

METHODS

We used data from a nested case-control study within the Danish population, with 1086 ALS cases from 1982 to 2009, jointly investigating 4 hospital-based diagnoses - diabetes, obesity, physical/stress trauma, cardiovascular disease (CVD) during 1977-2009; and 4 environmental exposures - lead, formaldehyde, diesel exhaust, and solvents, assessed from individual occupational history. All covariates were evaluated as ever/never exposed, and we used targeted machine learning techniques to screen for important joint predictors and interactions. These were then evaluated in a final logistic regression model adjusting for potential confounders (age, SES, geography). All analyses were stratified by sex.

RESULTS

Among men, trauma and solvents were associated with higher odds of ALS (OR = 1.55, 95% CI: 1.08-2.23; OR = 1.49, 95% CI: 1.17-1.89, respectively), and presented a negative interaction (OR = 0.49, 95% CI: 0.30-0.80). A positive diesel/CVD interaction was observed (OR = 1.56, 95% CI: 0.94-2.60). Among women, solvents, trauma, lead, and CVD were associated with higher odds of ALS, and a negative lead/solvents interaction was documented (OR = 0.52, 95% CI: 0.42-0.63).

CONCLUSIONS

This study is one of the first attempts to evaluate joint and interactive effects of multiple risk factors on ALS, identifying potential synergistic and antagonistic mechanisms.

A toolkit for the application of placental-fetal molecular biomarkers in epidemiologic studies of the fetal origins of chronic disease

Purpose of review

In this review, we provide essential background knowledge and an analytical framework for the application of placental-fetal molecular biomarkers in fetal origins chronic disease epidemiology. The widely available and highly quantitative placental hormone human chorionic gonadotropin (hCG) is used as an example. hCG is currently used for diagnosing fetal genetic disorders; yet it can and should be expanded to understanding the fetal origins of chronic diseases. We provide justification and methods to do this.

Recent findings

Ten papers published in the last 5 years were identified with supportive findings relevant to the application of biomarkers of hCG in epidemiologic studies on the developmental origins of health and disease (DOHaD).

Summary

There is increasing and consistent evidence that placental-fetal biomarkers may be highly informative in observational studies, as exemplified by hCG, with the correct approaches for measurement and data analysis.

Causal mediation analysis in presence of multiple mediators uncausally related

Mediation analysis aims at disentangling the effects of a treatment on an outcome through alternative causal mechanisms and has become a popular practice in biomedical and social science applications. The causal framework based on counterfactuals is currently the standard approach to mediation, with important methodological advances introduced in the literature in the last decade, especially for simple mediation, that is with one mediator at the time. Among a variety of alternative approaches, Imai et al. showed theoretical results and developed an R package to deal with simple mediation as well as with multiple mediation involving multiple mediators conditionally independent given the treatment and baseline covariates. This approach does not allow to consider the often encountered situation in which an unobserved common cause induces a spurious correlation between the mediators. In this context, which we refer to as mediation with uncausally related mediators, we show that, under appropriate hypothesis, the natural direct and joint indirect effects are non-parametrically identifiable. Moreover, we adopt the quasi-Bayesian algorithm developed by Imai et al. and propose a procedure based on the simulation of counterfactual distributions to estimate not only the direct and joint indirect effects but also the indirect effects through individual mediators. We study the properties of the proposed estimators through simulations. As an illustration, we apply our method on a real data set from a large cohort to assess the effect of hormone replacement treatment on breast cancer risk through three mediators, namely dense mammographic area, nondense area and body mass index.

Direct and indirect effects of marijuana use on the risk of fatal 2-vehicle crash initiation

BACKGROUND

Marijuana and alcohol each play a significant role in fatal crash initiation. We decomposed the total effect of marijuana use in the presence or absence of alcohol on fatal crash initiation into direct and indirect effects.

METHODS

Pair-matched data on 5856 culpable drivers (initiators) and 5856 nonculpable drivers (noninitiators) involved in the same fatal 2-vehicle crashes recorded in the Fatality Analysis Reporting System between 2011 and 2016 were analyzed using the conditional logistic regression model and the unified mediation and interaction analysis framework.

RESULTS

Crash initiators were more likely than noninitiators to test positive for marijuana (16.1% vs. 9.2%, P < 0.001), alcohol (28.6% vs. 9.7%, P < 0.001) and both substances (6.3% vs. 1.6%, P < .0001). Adjusted odds ratios of fatal 2-vehicle crash initiation revealed a positive interaction on the additive scale between marijuana and alcohol. Of the total effect of marijuana use on fatal 2-vehicle crash initiation, 68.8% was attributable to the direct effect (51.5% to controlled direct effect and 17.3% to reference interaction effect with alcohol) and 31.2% to the indirect effect (7.8% to mediated interaction effect and 23.4% to pure indirect effect through alcohol).

CONCLUSION

Our results indicate that the increased odds of fatal 2-vehicle crash initiation associated with marijuana use is due mainly to the direct effect.

Reply to Comment on "Environmental and Occupational Risk Factors of Amyotrophic Lateral Sclerosis: A Population-Based Case-Control Study"

We much appreciate the positive comments and interest concerning our study on the environmental and occupational risk factors of amyotrophic lateral sclerosis (ALS) [...].

Mechanisms underpinning the association between physical activity and mental health in adolescence: a 6-year study

BACKGROUND

Physical activity (PA) can promote mental health, but the mechanisms underpinning this association are not well-established. This study examined if perceptions of three basic psychological needs (autonomy, competence, relatedness) and moderate-to-vigorous physical activity (MVPA) mediate the association between number of years participating in PA and mental health in adolescence.

METHODS

Participants included 937 children (55% female) age 10-11 at inception of the longitudinal MATCH study, who provided data every 4 months over 6 years. Mediation analyses were used to assess the natural direct effect of number of years of PA participation (cycles 1-15) during late childhood and adolescence on later mental health (cycle 16), measured with the Mental Health Continuum-Short Form (MHC-SF), and the natural indirect effect through each of self-perceived autonomy, competence and relatedness, and self-report MVPA (cycle 15).

RESULTS

In single mediator models, indirect effects of autonomy, competence, relatedness and self-report MVPA were statistically significant. In joint mediation models (each of three models including one basic psychological need and MVPA), autonomy, competence and relatedness mediated 71, 27, and 51% of the association respectively; MVPA mediated 27-31% of the association. In the mediation model including all four mediators, relatedness mediated the largest proportion of the association, followed by autonomy and MVPA.

CONCLUSION

Results support developing strategies to encourage adolescents to engage and remain involved in PA. This could foster perceptions of autonomy, competence, and relatedness as well as MVPA, which in turn may enhance mental health.

Approaches for incorporating environmental mixtures as mediators in mediation analysis

Mediation analysis offers an essential and rapidly expanding tool in environmental health studies to investigate the contribution of environmental factors towards observed associations between risk factors and health outcomes. When evaluating environmental factors, there may be particular interest in quantifying the impact of exposure to environmental mixtures on human health. In this context, evaluating the joint effect of multiple chemicals or pollutants, rather than individual examination, allows accurate identification of risk factors, assessment of interactions, and ultimately development of more targeted public health interventions. While mediation analysis has been extended to incorporate several methodological complexities specific to environmental factors, little attention has been given to integrating the analysis of environmental mixtures. The aim of this review is to present some of the available methods for environmental mixtures, and discuss how these methods can be integrated within a mediation analysis framework. By incorporating these methods into a mediation framework, investigators will be able to evaluate the contribution of environmental mixtures as mediators of exposure-outcome associations, based on methodologies that are currently available. While standard regression-based methods for multiple mediators can be used, these can easily become unstable as the number of mixture components increases. Summary and classification methods, or hierarchical modeling, can reduce the number of mediators by creating scores or possibly uncorrelated subgroups. This approach allows retrieving indirect effects due to the mixture or to a specific subgroup, but makes identification of component-specific effects and interactions complicated. Finally, one can use various approaches for analyzing mixtures in a two-stage fashion, selecting relevant mediators to be included in the final model. We focused this review on techniques that have been presented to the environmental health community and that can be conducted with major statistical software. We encourage researchers to move beyond the evaluation of one environmental factor at a time to the assessment of the joint effects of environmental mixtures when a mediation model is of interest. Available methods target different aspects related to environmental mixtures and the choice of the suitable approach will depend on data structures and the research question of interest.

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

RATIONALE

Intersectionality 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.

OBJECTIVE

We 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.

METHOD

To 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 CONCLUSIONS

We 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.

Including Phenotypic Causal Networks in Genome-Wide Association Studies Using Mixed Effects Structural Equation Models

Network based statistical models accounting for putative causal relationships among multiple phenotypes can be used to infer single-nucleotide polymorphism (SNP) effect which transmitting through a given causal path in genome-wide association studies (GWAS). In GWAS with multiple phenotypes, reconstructing underlying causal structures among traits and SNPs using a single statistical framework is essential for understanding the entirety of genotype-phenotype maps. A structural equation model (SEM) can be used for such purposes. We applied SEM to GWAS (SEM-GWAS) in chickens, taking into account putative causal relationships among breast meat (BM), body weight (BW), hen-house production (HHP), and SNPs. We assessed the performance of SEM-GWAS by comparing the model results with those obtained from traditional multi-trait association analyses (MTM-GWAS). Three different putative causal path diagrams were inferred from highest posterior density (HPD) intervals of 0.75, 0.85, and 0.95 using the inductive causation algorithm. A positive path coefficient was estimated for BM → BW, and negative values were obtained for BM → HHP and BW → HHP in all implemented scenarios. Further, the application of SEM-GWAS enabled the decomposition of SNP effects into direct, indirect, and total effects, identifying whether a SNP effect is acting directly or indirectly on a given trait. In contrast, MTM-GWAS only captured overall genetic effects on traits, which is equivalent to combining the direct and indirect SNP effects from SEM-GWAS. Although MTM-GWAS and SEM-GWAS use the similar probabilistic models, we provide evidence that SEM-GWAS captures complex relationships in terms of causal meaning and mediation and delivers a more comprehensive understanding of SNP effects compared to MTM-GWAS. Our results showed that SEM-GWAS provides important insight regarding the mechanism by which identified SNPs control traits by partitioning them into direct, indirect, and total SNP effects.

A decade lost: does educational success mitigate the increased risks of premature death among children with experience of out-of-home care?

Background

Past research has consistently identified children with out-of-home care (OHC) experience as a high-risk group for premature mortality. While many have argued that educational success is a key factor in reducing these individuals’ excessive death risks, empirical evidence has hitherto been limited. The aim of the current study was therefore to examine the potentially mitigating role of educational success in the association between OHC experience and premature mortality.

Methods

Drawing on a Stockholm cohort born in 1953 (n=15 117), we analysed the associations among placement in OHC (ages 0–12), school performance (ages 13, 16 and 19) and premature all-cause mortality (ages 20–56) by means of Cox and Laplace regression analyses.

Results

The Cox regression models confirmed the increased risk of premature mortality among individuals with OHC experience. Unadjusted Laplace regression models showed that, based on median survival time, these children died more than a decade before their majority population peers. However, among individuals who performed well at school, that is, those who scored above-average marks at the age of 16 (grade 9) and at the age of 19 (grade 12), the risks of premature mortality did not significantly differ between the two groups.

Conclusion

Educational success seems to mitigate the increased risks of premature death among children with OHC experience.

The role of PANSS symptoms and adverse events in explaining the effects of paliperidone on social functioning: a causal mediation analysis approach

To date, no study has evaluated the joint role of symptoms and adverse events as mediators of the effect of second-generation antipsychotics on patients' social functioning. We used recently developed methods for mediation analysis with multiple mediators to clarify the interplay of adverse events and symptoms in explaining the effects of paliperidone (R code for implementing the mediation analysis for multiple mediators is provided). We used data from 490 participants in a 6-week randomized dose-response trial that assigned three fixed dosages of ER OROS paliperidone (3, 9, and 15 mg/day). The primary outcome was an individual's score on the social performance scale assessed after 6 weeks. The sum of Positive and Negative Syndrome Scale (PANSS), weight gain, and extrapyramidal symptoms measured via the Simpson-Angus Scale after 5 weeks were investigated as potential mediators and effect modifiers of treatment effects. Results from mediation analyses showed that the improvements in social functioning are partly explained by reduction in PANSS symptoms. Suggestive evidence that adverse events could play a role as mediators was found. In particular, weight gain displayed a non-linear relationship with social functioning, whereby beneficial effects observed at small levels of weight gain were reduced in the presence of excessive weight gain. In conclusion, we found that the short-term effects of paliperidone on social functioning were dependent on the successful reduction in PANSS symptoms and possibly the occurrence of excessive weight gain, thus suggesting future directions for treatment and interventions.

Multiple mediators approach to study environmental chemicals as determinants of health disparities

A major goal of health disparities research is to identify and intervene upon modifiable risk factors that help explain the observed associations between social factors and adverse health outcomes. To this end, statistical methods incorporating mediation analysis have shown promise, as they quantify the contribution of an intermediate variable in an exposure-outcome association. A growing body of literature suggests that environmental chemicals can contribute to health disparities. However, evaluating environmental chemicals as an important component of health disparities introduces methodological complexities that may make standard mediation approaches inadequate. Specific to environmental health is the issue of evaluating both the source and biomarker of the environmental toxicant in order to calculate the proportion of the disparity that would remain had we intervened on the modifiable factors. Recent methodological developments on multiple mediators can improve efforts to integrate both source and biomarker of exposure into epidemiological studies of health disparities. We illustrate a conceptual framework and present how mediation techniques can be used to address environmental health disparities questions. With this, we provide a methodological tool that has the potential to advance this growing field, while simultaneously informing public health prevention and policy surrounding the impact of environmental factors on health disparities.