0443 Proposing a ‘projected apnea-hypopnea index (AHI)’ to mitigate intrinsic pitfalls of conventional AHI and testing its validity

Abstract

Abstract Introduction The night-to-night variability of apnea-hypopnea index (AHI) can be explained by the fact that a person with OSA tends to spend less time in supine position and potentially less REM sleep, during which obstructive events could be a lot worse than in the other form. The primary aim of this study is to propose and test a “projected AHI” (pAHI) with which AHI can be mathematically adjusted for sleep position and sleep stage spent in a night of recording. Methods Using mathematical modeling explained in a separate abstract, a final pAHI formula was generated from polysomnography data of 337 subjects who exhibited all four sleep-types [combination of REM vs NREM and supine (SUP) vs non-supine (nSUP)] in one overnight study. A confusion matrix was generated using cutoffs of (p)AHI of 5 and hypothesizing pAHI being the gold standard and AHI being the test. Among subjects with borderline results (AHI of 5±3), sleep-type time and sleep-type AHI were compared between 4 groups of the confusion matrix to test the validity of the pAHI model. Results Out of 337 subjects with all four sleep-types, 119 subjects with AHI of 5±3 were identified and classified into a confusion matrix of no-OSA (43 TN, 10 FP), and OSA (13 FN, 53 TP). Actual sleep-type AHIs were in the order of increasing pAHI (TN, FP, FN, TP): 15.3, 29.2, 32.0, 34.8 for REMSUP; 4.00, 4.53, 4.73, 9.40 for REMnSUP; 2.83, 2.87, 9.38, 10.0 for NREMSUP; 0.44, 0.1, 1.06, 1.98 for NREMnSUP. Sleep time parameters were significantly different between no-OSA and OSA groups only when pAHI was used for diagnosis: %REMSUP (9.32 vs 5.39, p = 0.000009), %REMnSUP (8.08 vs 11.31, p = 0.002), %NREMSUP (48.13 vs 36.53, p = 0.001). %NREMnSUP (34.46 vs 46.77, p = 0.0002). Conclusion Without having to repeat a sleep study, projected AHI, by taking the effect of sleep position and sleep stages into account, can produce an AHI that mitigates night-to-night variability stemming from position and stage differences between nights. Projected AHI can be particularly useful in interpreting a sleep study with a borderline AHI result. Support (if any)