I have reformulated the question
I mean a statistical approach for adjusting values using multiple linear regression which is often used in ecological studies (e.g. Urashima et al., 2020 Int J Environ Res Public Health 17). Please see the attached figure and text below for more details.
Is an adjustment for a confounding factor (roughly) a substitution of the actual values by Residuals (as defined below)?
Residual =Y-Yh. Where Y is an actual value, Yh - is predicted based (e.g. on linear regression) for all data points.
To clarify my question I’m attaching a graph with unadjusted and (presumably) adjusted values. Do I understand correctly that empty circles represent values adjusted in regard of percentage of rural population?
The practical solution to answer these questions is as follows:
When preparing a study proposal, it is better to first prepare a list of confounding variables according to the objectives of the study. Then we classify these variables into two groups of major and minor based on the evidence related to confounders, which can usually be extracted from previous studies and literature review. A desirable and practical strategy is to control the main confounders through study design solutions (constraint or matching) and the sub-confounders through data analysis solutions.
In most case studies, factors such as age and sex are the main confounders of the study. Therefore, with the matching solution and selecting the age and sex matched control group, we can control these two confounders well, even control other non-main or sub-confounders when analyzing the data.
Carlos A. Torres
Mehran ShahiThank you very much for the answers.
It appears that my question is not specifically formulated. I mean a statistical approach for adjusting values using multiple linear regression which is often used in ecological studies (e.g. Urashima et al., 2020 Int J Environ Res Public Health 17).
P.S. (Actually, I attached a graph to my question to specify it, but I’m not sure that you can see it).
We can adjust or control the confounding factors in epidemiology by applying some practical methods
1- Randomization of different crops
2-Restriction
3-Matching
To establish the relationship between exposure variable and outcome, we need to consider and take care of other factors (confounds) that might affect the association between those two variables.
These factors (confounds) will either influencing the association to be seen easily or by limiting it due to the relationship which the confounds might have with the exposure variable or outcome.
To deal with those factors (confounds) to ensure their effects on outcomes are minimized that is why the issue of adjustment for confounding needs to be considered. And this can be done by adding those factors (confounds) in the model to ensure that the outcomes are adjusted by considering cofounding effects.
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology