We do not understand the choice of references 2 and 3 of Serin et al's letter, which are not related to the topic. Moreover, the last paragraph of their letter is confused and has some inaccuracies. Our risk model did not include BMI, FEV1/FVC, baseline arterial oxygen saturation (Sao2), or Epworth Sleepiness Scale. As shown in Table 3 of our paper, the PE recurrence risk model included as independent risk factors AHI > 10, nocturnal Sao2 (not baseline Sao2), D-dimer levels, and the percentage of total time study with Sao2 < 90% (CT90). The first two variables were clearly different between patients with and those without recurrence of PE (Table 2 in our paper). Recently, we reported that levels of D-dimer, a well-known PE recurrence risk factor, are increased in patients with OSA. Therefore, our univariate analysis probably failed to detect differences in the D-dimer levels between patients with and those without PE recurrence, but its role as a risk factor is evident when adjusted for AHI. Finally, we think the role of CT90 particularly remarkable in that it shows a negative relation with the risk of PE recurrence (Table 3 in our paper). Although the univariate analysis also found no difference in this variable between patients with and those without PE recurrence, this is probably due to its great dispersion. We interpret its incorporation into a multivariate model as an independent protective factor might to be due to its ability to discriminate between sustained intermittent hypoxia and continuous hypoxia. It might be speculated whether the incorporation of the CT90 into the risk model, along with that of AHI and mean nocturnal Sao2, constitutes indirect evidence of the contribution of intermittent hypoxia to the OSA-associated hypercoagulability state.