The first study you linked looked at 29518 people and that's how they controlled for comorbid illnesses: > As a measure of comorbid illness at the start of the

The first study you linked looked at 29518 people and that's how they controlled for comorbid illnesses: > As a measure of comorbid illness at the start of the study, we created a dummy variable ‘comorbidity’ identifying women who have been treated with antidiabetic [Anatomical Therapeutic Chemical Classification System (ATC) A:10] or anticoagulant (ATC B:01) drugs or medications for cardiovascular disease (ATC C:01–C:10) for more than 1 month. This is extremely simple, hardly careful controlling. If anything it's a good example of bad science. Grabbed whatever was easily available in the database and claims they've adequately controlled for comorbidity while doing no such thing. If you look at the difference in exercise, income, education and BMI there's a LOT going on between the two groups. You'd wish scientists would include the raw data and source code to easily calculate the adjustments yourself.