What is the recommended way to quantify the potential effect of unmeasured confounding, after constructing a multivariable ordinal logistic regression (i.e. proportional odds model) for an observational study?
I have come across one metric known as the “E-value”. As described in the JAMA Guide to Statistics and Methods, “an E-value analysis asks the question: how strong would the unmeasured confounding have to be to negate the observed results? The E-value itself answers this question by quantifying the minimum strength of association on the risk ratio scale that an unmeasured confounder must have with
both the treatment and outcome, while simultaneously considering the measured covariates, to negate the observed treatment– outcome association”. (Using the E-Value to Assess the Potential Effect of Unmeasured Confounding in Observational Studies | Research, Methods, Statistics | JAMA | JAMA Network)
The E-value can be calculated with an online calculator (How to use this website - E-value calculator), using a point estimate (e.g. odds ratio) and the upper/lower bounds of the 95% CI. However, there doesn’t seem to be a formula to calculate E-values for “common odds ratios” that would be produced by ordinal logistic regression. As such, I’m not sure where to go from here. Any guidance or suggestions would be much appreciated!