Considering road traffic as an example, some of the studies report confounding in their associated health effects (for example; CVD etc.) induced by environmental stressors such as air and noise pollution. How can we reduce or at least try to minimize this confounding effect?
Thanks a lot @Frank for your answer and suggested articles. They are indeed useful and I appreciate your help. In addition to the above papers, Tétreault et al., 2013 also suggested some potential confounder factors. However, I am more looking forward for the suggestion/discussions directing towards "how to minimize or reduce this confounding effect". For example, what measures may be suggested to do so.
I think what Frank was suggesting was that these covariates would simply be included in your statistical model. You could then present something like the odds ratio for road traffic in its own univariate model alongside the adjusted statistic for road traffic in a model with whichever other covariates you'd think should be included.
I assume that Jibran wishes to study the determinants of specific health problem . In this case, a thorough literature review will help to identify main determinants and risk factors which can be listed and addressed in the study. Traffic induced pollution is one of the major sources of air pollution. The pollutants are multiple. These must be considered as confounders. Any data on the level of air pollution may be used as a proxy measure . If no data available. Density of traffic can be used. All these points need advice from experts and relevant literature.
Nice answers to an important query. I assume an observational study rather than a cohort. I would add to consider the theoretical basis of the diseases involved. These necessarily guide the choice of explanatory variables used in models such as regressions or proportional hazards. Else, the estimates obtained will be biased in the case where important omitted theoretical (namely, medical) risk factors (as explanatory variables) are omitted. The random error does help, BTW. Regarding DAGs (first answer) the questions still remain: what is the theoretically correct model; is there one? It seems that the answer will be hypothesis generating, rather than causa.
Hope this helps
Well, thanks a lot to all of you for your feedback. It has been indeed a fruitful discussion.
- Badges
- Science topic
- Similar topics
- Medicine
- Public Health