Article ID: CJ-16-0201
To the Editor:
We read with great interest the paper by Satake et al recently published in the Journal.1 We agree with the conclusions made in the study, but we wish to raise some issues.
First, the comparability of the groups is not well established, despite all the variables in Table 1 not having significant differences. There may be other unmeasured variables that were not considered in the analysis that may have an effect on the result of the study; for example, it would be important to include history of atrial fibrillation, because the SchlaHF registry study concluded that atrial fibrillation was a clinical predictor for at least moderate sleep-disordered breathing (SDB).2 Also, it would be interesting to consider the existence of chronic kidney disease in the patients evaluated in this study. A systematic review concluded that chronic kidney disease independent of coexisting chronic heart failure could be associated with development of central sleep apnea (CSA).3
Second, the group of AHI <5 (non-SDB) also includes patients who had received therapy and responded to it and the group of AHI >5 (SDB group) includes patients who have received therapy, those who were recommended therapy but declined and also those who did not need it. This affects the comparability of the groups.
Third, the primary outcome establishes all the fatal arrhythmic events (ventricular tachycardia, ventricular fibrillation, sudden death) and the secondary outcome, all causes of death. However, you have considered 4 patients who died of ventricular tachycardia (2) and ventricular fibrillation (2) in the all-cause group of deaths.
Finally, the test used to diagnose obstructive sleep apnea (OSA) and CSA is not the gold standard for these diseases. In some cases, polygraphy is insufficient to rule out OSA when the respiratory events are mainly associated with arousals. Full-night polysomnography is recommended when polygraphy is “normal”.4 Some of the study patients diagnosed as “normal” may have had OSA but may not have been diagnosed correctly because of the test used in this study.
In conclusion, a study with a better design and larger sample size may solve these issues.
(Released online April 14, 2016)