VanderWeele TJ, Vansteelandt S. Odds ratios for mediation analysis with a dichotomous outcome

Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA.
American journal of epidemiology (Impact Factor: 5.23). 10/2010; 172(12):1339-48. DOI: 10.1093/aje/kwq332
Source: PubMed


For dichotomous outcomes, the authors discuss when the standard approaches to mediation analysis used in epidemiology and the social sciences are valid, and they provide alternative mediation analysis techniques when the standard approaches will not work. They extend definitions of controlled direct effects and natural direct and indirect effects from the risk difference scale to the odds ratio scale. A simple technique to estimate direct and indirect effect odds ratios by combining logistic and linear regressions is described that applies when the outcome is rare and the mediator continuous. Further discussion is given as to how this mediation analysis technique can be extended to settings in which data come from a case-control study design. For the standard mediation analysis techniques used in the epidemiologic and social science literatures to be valid, an assumption of no interaction between the effects of the exposure and the mediator on the outcome is needed. The approach presented here, however, will apply even when there are interactions between the effect of the exposure and the mediator on the outcome.

1 Follower
20 Reads
  • Source
    • "[2] "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Mendelian randomization uses genetic variants, assumed to be instrumental variables for a particular exposure, to estimate the causal effect of that exposure on an outcome. If the instrumental variable criteria are satisfied, the resulting estimator is consistent even in the presence of unmeasured confounding and reverse causation. Methods: We extend the Mendelian randomization paradigm to investigate more complex networks of relationships between variables, in particular where some of the effect of an exposure on the outcome may operate through an intermediate variable (a mediator). If instrumental variables for the exposure and mediator are available, direct and indirect effects of the exposure on the outcome can be estimated, for example using either a regression-based method or structural equation models. The direction of effect between the exposure and a possible mediator can also be assessed. Methods are illustrated in an applied example considering causal relationships between body mass index, C-reactive protein and uric acid. Results: These estimators are consistent in the presence of unmeasured confounding if, in addition to the instrumental variable assumptions, the effects of both the exposure on the mediator and the mediator on the outcome are homogeneous across individuals and linear without interactions. Nevertheless, a simulation study demonstrates that even considerable heterogeneity in these effects does not lead to bias in the estimates. Conclusions: These methods can be used to estimate direct and indirect causal effects in a mediation setting, and have potential for the investigation of more complex networks between multiple interrelated exposures and disease outcomes.
    International Journal of Epidemiology 08/2014; 44(2). DOI:10.1093/ije/dyu176 · 9.18 Impact Factor
  • Source
    • "In this case, association between outcome and methylation reflects sharing of genetic controls rather than causality. One can use mediation analysis (30–33) to estimate how much SNP–outcome association is through methylation as an intermediate phenotype and Mendelian randomization (34) to assess the causal direction between SNP, methylation and outcome. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Platform technologies for measurement of CpG methylation at multiple loci across the genome have made ambitious epigenome-wide association studies affordable and practicable. In contrast to genetic studies, which estimate the effects of structural changes in DNA, and transcriptomic studies, which measure genomic outputs, epigenetic studies can access states of regulation of genome function in particular cells and in response to specific stimuli. Although many factors complicate the interpretation of epigenetic variation in human disease, cell-specific methylation patterns and the cellular heterogeneity present in peripheral blood and tissue biopsies are anticipated to cause the most problems. In this review, we suggest that the difficulties may be exaggerated and we explore how cellular heterogeneity may be embraced with appropriate study designs and analytical tools. We further suggest that systematic mapping of the loci influenced by age, sex and genetic polymorphisms will bring important biological insights as well as improved control of epigenome-wide association studies.
    Human Molecular Genetics 06/2014; 23(R1). DOI:10.1093/hmg/ddu284 · 6.39 Impact Factor
  • Source
    • "Finally, to assess how the total effect (sum of the direct and indirect effects) of exposure to a given neighbourhood-level SES-related variable is transferred through the proposed pathways, mediation analyses for binary-outcome models were conducted following the method described by VanderWeele and Vansteelandt [33]. This method estimates the direct and indirect effects on the odds ratio scale by combining logistic and linear regressions. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Although socio-economic factors have been identified as one of the most important groups of neighbourhood-level risks affecting birth outcomes, uncertainties still exist concerning the pathways through which they are transferred to individual risk factors. This poses a challenge for setting priorities and developing appropriate community-oriented public health interventions and planning guidelines to reduce the level of adverse birth outcomes. This study examines potential direct and mediated pathways through which neighbourhood-level socio-economic determinants exert their impacts on adverse birth outcomes. Two hypothesized models, namely the materialist and psycho-social models, and their corresponding pathways are tested using a binary-outcome multilevel mediation analysis. Live birth data, including adverse birth outcomes and person-level exposure variables, were obtained from three public health units in the province of Ontario, Canada. Corresponding neighbourhood-level socio-economic, psycho-social and living condition variables were extracted or constructed from the 2001 Canadian Census and the first three cycles (2001, 2003, and 2005) of the Canadian Community Health Surveys. Neighbourhood-level socio-economic-related risks are found to have direct effects on low birth weight and preterm birth. In addition, 20-30% of the total effects are contributed by indirect effects mediated through person-level risks. There is evidence of four person-level pathways, namely through individual socio-economic status, psycho-social stress, maternal health, and health behaviours, with all being simultaneously at work. Psycho-social pathways and buffering social capital-related variables are found to have more impact on low birth weight than on preterm birth. The evidence supports both the materialist and psycho-social conceptualizations and the pathways that describe them, although the magnitude of the former is greater than the latter.
    International Journal of Health Geographics 06/2013; 12(1):32. DOI:10.1186/1476-072X-12-32 · 2.62 Impact Factor
Show more


20 Reads
Available from