I often read meta-analyses. What they do is:
- explain how they searched for relevant studies (i.e. which databases or bibliographies they used, sometimes even which keywords)
- give the number of studies they found
- list the studies in the reference list
They never list any sources when they state how many sources they found. Instead they explain how they searched.
If the databases, bibliographies and other sources you use for your search are not common knowledge in your field, list them in the reference list and cite them in text when you explain your search strategy. For example, if you use the bibliography in the back of a book published 1765, you would list that book and cite it where you mention it in your text. Don't list or cite common databases like PubMed (but mention if and how you used them).
Whenever a meta-analysis claims that there is no more than they found, or an author wants to delineate the development of a research topic from the first publication to its climax and beyond to its decline (i.e. when it is important to accurately know how much has been published on a certain topic at a certain time), the authors explain their search strategy in enough detail to allow a reader of their article to reconstruct that strategy and evaluate it. For example, if they used a databases, they list the keywords they used for their search.