To the extent of my knowledge, it depends on the degree of overlapping populations. If the degree of overlapping population is nearly 100% (e.g. when article A is published in both preprint and peer-reviewed versions), it is plausible to exclude one of the study. Alternatively, if you detect any scientific misconduct (e.g. plagiarized content where both patient characteristics and analysis methods are same), you can contact the journal to investigate this issue (tips: you can identify them by comparing the patients' characteristics, outcome of interests, and analysis methods. If both studies are sufficiently different, but there are possibilities of overlapping populations (e.g. studies with overlapping study location and recruitment period), I usually include both of them in the systematic review, but exclude one of them from the meta-analysis

It is always better to contact the corresponding authors of the potential overlapping studies to confirm with them. If contacting the authors is not possible (or if the authors did not reply), I usually judge potential overlapping population from recruitment period and study location. I tend to prioritize analysis for larger-sized studies, although it is true that there is currently no formal guidance on this (as far as I know).

I agree with Davide Donelli that it is important for meta-analysts to identify overlapping population as it may introduce significant heterogeneity to the final model.

yes you can

I agree with Davide Donelli . In addition to overlapping populations, I would consider the degree of homogeneity/heterogeneity among these studies based on outcome variable(s) of interest to include them in meta analysis.

No I shouldn't

It is good practice not to duplicate subjects in metanalysis. If you are conducting a solely qualitative systematic review, you can and in my opinion include both, and assess the data taking into account the limitation of possible overlapping population. If you also want to do a metanalysis, you should only include non-overlaping samples in that part of your analysis (you can include all of them for the qualitative synthesis but you should only use one for the metanalysis). In terms of which one to choose, chose either the longest followup or the biggest sample, depending on what you want to study.

If the percentage of overlap is a relatively small percentage of each study sample, you can conduct secondary (exploratory/sensitivity) analyses including all studies, however, it should not be the primary analysis, because it artificially increases the weight of those subjects (an extreme example of this would be the same study data being published 3 times and someone including all three publications in a metanalysis, which would almost triple the weight of those subjects/events).

Rachael Beale Hope this helps!