Potential Underestimation of Sleep Apnea Severity by At-Home Kits: Rescoring In-Laboratory Polysomnography Without Sleep Staging

Abstract

Study objectives: Home sleep apnea testing (HSAT) is increasingly available for diagnosing obstructive sleep apnea (OSA). One key limitation of most HSAT involves the lack of sleep staging, such that the respiratory event index is calculated using the total recording time (TRT) rather than total sleep time (TST).

Methods: We performed a retrospective analysis of n = 838 diagnostic polysomnography (PSG) nights from our center; n = 444 with OSA (4% rule, apneahypopnea index (AHI) ≥ 5), and n = 394 with AHI < 5. We recalculated the AHI using time in bed (TIB) instead of TST, to assess the predicted underestimation risk of OSA severity.

Results: Of all the patients with OSA, 26.4% would be reclassified as having less severe or no OSA after recalculating the AHI using TIB rather than TST. Of the n = 275 with mild OSA, 18.5% would be reclassified as not having OSA. The risk of underestimation was higher in those with moderate or severe OSA. Of the n = 119 moderate OSA cases, 40.3% would be reclassified as mild, and of the n = 50 severe OSA cases, 36.0% would be reclassified as moderate. Age strongly correlated with the degree of underestimation of the AHI, because age was significantly correlated with time awake during PSG.

Conclusions: The risk of sleep apnea underestimation is predicted to be substantial in a tertiary sleep center population. Phenotyping errors included risk of falsely negative results (from mild to normal), as well as category errors: moderate or severe moving to mild or moderate severity, respectively. Clinicians should recognize this underestimation limitation, which directly affects diagnostic phenotyping and thus therapeutic decisions.

Commentary: A commentary on this article appears in this issue on page 531.

Keywords: diagnosis; prediction; underestimation.

Figure 1. Reclassification based on scoring polysomnography using time in bed as denominator for the apnea-hypopnea index calculation.

Bar chart shows the percentage of the cohort (n = 444) with obstructive sleep apnea during clinical polysomnography, with apnea-hypopnea index ≥ 5 using the objective total sleep time as the denominator (standard clinical scoring) that are reclassified after scoring based on time in bed as the denominator (as if undergoing home sleep apnea testing). 26.4% moved at least one category less severe (black bar). Approximately 10% of the cohort moved specifically from mild to none, which represents approximately 18% of mild cases (51 of 275). Approximately 10% of the cohort moved from moderate to mild, but this represents approximately 40% of the moderate cases (48 of 119). A smaller percent of the cohort moved from severe to moderate, but this represents 36% of severe cases (18 of 50). mod = moderate, sev = severe.

Figure 2. Reclassification based on scoring polysomnography using subjective total sleep time as the denominator for the apnea-hypopnea index calculation.

Scatterplots show the subjective total sleep time (sTST, y-axis) and objective total sleep time (oTST, x-axis) pairing for each subject in the cohort with obstructive sleep apnea (A), or without obstructive sleep apnea (B). The diagonal dotted line is the identity line for reference. (C) The percentage of the cohort misclassified by type of misclassification. Above each bar is the percentage of each severity category that was misclassified. mod = moderate, sev = severe.