[R] Categorical variable in MA

[Pier-Alexandre Tardif]

Good morning,

We're undertaking a meta-analysis in which we have an association between a dichotomous outcome (mortality) and a categorical variable with multiple choices (usually between 3 and 4). However, the reference isnt always the same, so we could have study A who present results with A/B/C/D categories with �A� as reference, but study B who presents results with the same A/B/C/D categories but with B as the reference. In the case of a dichotomous variable and an odds ratio, we could ensure a common denominator among the study using 1/OR. Is there any reliable way to do so with a categorical variable (with OR mostly, and possibly relative risks [RR])?

Thanks in advance,

Pier-Alex

	[[alternative HTML version deleted]]