[Feature request] Can exposures and outcomes be selected separately for gender?

[yin-huamin]

Thank you for your great work. It's very helpful for my research.

When I performed the cis-MR Analysis, I wanted to explore the potential causal association between female obesity and autism.

So if the exposure could select female and the outcome could select all conditions, would it make the results more reliable?

Thanks!
2024.11.26
Ethan

[legaultmarc]

Hi Ethan,

In the cis-MR platform, you can select a "Analysis subset" where you could use "Female only" if you want a sex-stratified analysis.

If you mean that you would like to consider sex as the exposure, I don't think this model would work as there are no instrumental variables for sex.

[yin-huamin]

Hi Ethan,

In the cis-MR platform, you can select a "Analysis subset" where you could use "Female only" if you want a sex-stratified analysis.

If you mean that you would like to consider sex as the exposure, I don't think this model would work as there are no instrumental variables for sex.

Hi, Thank you for your quick response and the helpful clarification about the "Analysis subset" option for sex-stratified analysis.

To clarify my question further: I am interested in exploring the potential causal association between female obesity as the exposure and autism in males as the outcome. Would it be feasible to use different sex-specific subsets for exposure and outcome in the cis-MR platform?

Specifically, could I select "female-only" for obesity (exposure) and "male-only" for autism (outcome) to conduct such an analysis?

Thank you again for your time and insights!
2024.12.06
Ethan

[legaultmarc]

Thank you for the clarification. It is not conventional to conduct such an analysis because it implies that an intervention on one group has causal effects on a different group. The causal estimand that such an analysis would estimate is not really well defined. For this reason, it is not implemented in ExPheWAS.