So those who may be prone to heart problems are advised to not drink caffeine, and then we find that those who don't drink caffeine are more prone to heart problems. Yep.
It undoubtedly has some effect, but hopefully the study has controlled for that.
Get funding for a study, run multiple regression against a large set of interviewees, find an outcome which satisfies the proponent, spin the results: people who drink coffee and do not have heart disease or diabetes get fewer heart disease or diabetes later, ergo, caffeine reduces the risk. People, this is not a double-blind study. It was biased from the start.