© 2009 American Thoracic Society
Sleep Study Predictors of Prevalent Cardiovascular DiseaseFrom the Authors:We appreciate Dr. Stradling's comments on our article describing the association between sleep-disordered breathing and prevalent cardiovascular disease (CVD) (1). The goal of our analysis was to examine the strength of associations between the frequency of disordered breathing events (i.e., hypopneas and apneas with an emphasis on the former) using different oxyhemoglobin desaturation thresholds and prevalent CVD. Due to analytical limitations of the data set, we did not compare the predictive values of hypopneas, defined by various reductions of airflow to the predictive value of oxyhemoglobin desaturation events identified by various thresholds of desaturation, independent of any flow-based criteria. However, Dr. Stradling raises an important issue of whether flow-based definitions are of any value by indicating that the reported adjusted odds ratios relating the hypopnea index (Table 2) to prevalent CVD are weaker than those reported for the oxyhemoglobin desaturation index (Table 3).
We submit that this is not a valid comparison. Table 2 shows the odds ratios for hypopneas with oxyhemoglobin desaturation within specific cut-points (e.g., 3.0–3.9%), whereas Table 3 shows the odds ratios for oxyhemoglobin desaturation events above a specific threshold (e.g., We also speculate that whereas the inclusion of arousals in the definition of hypopneas did not materially change the estimates of association, there are several reasons why the clinical impact of arousals should not be disregarded. First, visual scoring of arousals is fraught with poor to modest interscorer reliability. Second, the results from our article are based on cross-sectional analyses, which cannot explicitly support or refute causal inferences regarding the clinical effects of recurrent electroencephalographic arousals. Third, it is certainly possible that for some outcomes (e.g., sleepiness) arousals may carry pathophysiological significance independent of oxyhemoglobin desaturation, whereas for other outcomes (e.g., cardiovascular disease) the effects of oxyhemoglobin desaturation may be overwhelming. We agree that older indices of sleep-disordered breathing need to be critically appraised with regard to their predictive utility. There is obviously much to be learned, and we are entering an era in which many of the studies started in the last decade will yield robust findings on the clinical ramifications of sleep-disordered breathing and help refine the way we define, diagnose, and treat this prevalent and disabling condition.
Johns Hopkins University School of Medicine FOOTNOTES Conflict of Interest Statement: N.M.P. has participated as a speaker in scientific meetings, continuing medical education lectures, or clinical symposia sponsored by Respironics and ResMed. M.S. is a scientific consultant to Phillips-Respironics and has a financial interest in the BiPAP brand and related technologies used to treat sleep-disordered breathing by Phillips-Respironics. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
4%). Table 3 is more appropriately compared with Figure 1, which displays the odds ratio for hypopneas with an oxyhemoglobin desaturation above the same thresholds. An alternative comparison would be between Tables 2 and 4. Both of these display odds ratios for relatively comparable bins of oxyhemoglobin desaturation. Either comparison does not provide the necessary support for the notion that oxygen saturation dips alone are as good or better in predicting cardiovascular disease than flow-based hypopneas. Hence, the question of whether flow-based measures provide additive information for prevalent CVD or other health-related outcomes will require additional research.