Could clinical inertia in part explain the unexpected association of insulin therapy with poorer cardiovascular outcomes in observational studies on diabetes?