I always wonder to what degree this kind of study might invert causality: that is, people who are subclinically "ill" may be less likely to exercise in the first place because it is harder or less pleasant for them. Someone with, for example, undiagnosed but significant cardiovascular disease (unfortunately very common: in many cases the first sign anyone recognizes as concerning is a heart attack) is likely to have some degree of decreased exercise tolerance and discomfort at longer durations and intensities -- and thus will be less likely to meet the recommendations.