Sleep apnea management effect sizes: Objective vs. Subjective measures

Introduction: It is well known that objective measures of sleep apnea severity only correlate modestly with self-reported measures of sleep apnea symptoms at baseline. This has been thought to be largely a function of issues associated with self-reported questionnaires. What has not been examined to date is the extent of this discrepancy over time while on treatment. Methods: Data from an existing clinical trial on a positive airway pressure (PAP) therapy adherence were examined. Baseline AHI was obtained from diagnostic sleep testing. Follow-up AHI was an average of the residual AHI provided by the PAP device over the treatment period. Epworth Sleepiness Scale was used as a proxy for sleep apnea symptoms and was self-reported. Effect size (ES) was calculated using Cohen's d. Results: Baseline AHI was 28.0 (SD = 7.5) and residual AHI was 1.9 (SD = 1.3). The effect size for AHI reduction was 4.85. Baseline ESS was 10.7 (5.2) and follow-up was 8.4 (4.9). The effect size for ESS reduction was 0.45. There are a number of ways to interpret effect size with one being to put it in terms of the correlation coefficient: an ES of 0.45 is equivalent to an r correlation of 0.22 while an ES of 4.85 is equivalent to an r correlation of 0.92. Conclusion: In this sample of OSA patients, sleep apnea was very well controlled (with mean residual AHI of 2.5). However, while the effect size for change in AHI is extremely large, the effect size for change in sleepiness level is modest in comparison. A number of reasons might account for this difference, including the time period over which the residual AHI applies and the baseline self-report response bias in sleep-deprived patients. Future research should determine the factors and their contribution in helping to explain this discrepancy in effect sizes.