The difference between the Swedish and the Italian cohorts with respect to the anticoagulant thatwas used, for which we corrected in the statisticalmodels of the combined dataset by adjusting for cohort,
is therefore an unlikely explanation for the observed between-cohort
differences in our results.