The source data were sought from medical professionals and hospitals not individual patients, and the researchers asked the medical professionals to determine if the patient had prior heart disease (see question 7 in the link below). So this would not be the case that 50% of the dudes in this study failed to self evaluate their prior heart conditions. You are definitely identifying a potential observational uncertainty in the study, however the authors took steps to minimize the problem and there results show a strong association that it is hard to attribute the entire affect to misclassified prior status.
Here is a link to the source article in the scientific journal (I have no idea if this is a quality journal, what the impact factor or publication costs are):
https://academic.oup.com/ehjcimaging/article/doi/10.1093/ehjci/jeaa178/5859292?searchresult=1
Feel free to put your skepticism to test and review the paper.
Here is a link to the data used in the study:
https://oup.silverchair-cdn.com/oup...t~nO0QP3ug__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA
Feel free to analyze the data in anyway that you think is more appropriate.