For this to be true they'd either need their own trackers on every single website online (or be ingesting from those that do but at which point it's not Elsevier's tracking) or be have software/browser addon installed on the clients' system.
Only a small handful of companies have approached the near-ubiquity of webpage presence of the former that made cross-internet tracking a realistic scenario (see eg. past critique of Google/Facebook embeds) but afaict the only thing being interpreted here is a line in a privacy policy that is very likely being misinterpreted.
Certainly the report doesn't make any claim about ubiquitous internet tracking in its key findings summaries. Contrast that to how Fight for the Future has run with it:
> Elsevier tracks more than what books someone reads—it surveils readers’ search and browsing history across the whole web, collects their locations, and builds profiles based on inferences from this data.