The impact of body posture and sleep stages on sleep apnea severity in adults

Abstract

Study objectives: Determining the presence and severity of obstructive sleep apnea (OSA) is based on apnea and hypopnea event rates per hour of sleep. Making this determination presents a diagnostic challenge, given that summary metrics do not consider certain factors that influence severity, such as body position and the composition of sleep stages.

Methods: We retrospectively analyzed 300 consecutive diagnostic PSGs performed at our center to determine the impact of body position and sleep stage on sleep apnea severity.

Results: The median percent of REM sleep was 16% (reduced compared to a normal value of ~25%). The median percent supine sleep was 65%. Fewer than half of PSGs contained > 10 min in each of the 4 possible combinations of REM/NREM and supine/non-supine. Half of patients had > 2-fold worsening of the apnea-hypopnea index (AHI) in REM sleep, and 60% had > 2-fold worsening of AHI while supine. Adjusting for body position had greater impact on the AHI than adjusting for reduced REM%. Misclassification--specifically underestimation of OSA severity--is attributed more commonly to body position (20% to 40%) than to sleep stage (~10%).

Conclusions: Supine-dominance and REM-dominance commonly contribute to AHI underestimation in single-night PSGs. Misclassification of OSA severity can be mitigated in a patient-specific manner by appropriate consideration of these variables. The results have implications for the interpretation of single-night measurements in clinical practice, especially with trends toward home testing devices that may not measure body position or sleep stage.

Figure 1. Schematic of stage- and position-dependent OSA

(A) Hypnogram from a 7-h PSG, simplified to a single NREM sleep stage. The body position is shown as the dotted line. Tick marks indicate respiratory events. For visual clarity, each tick mark represents 2 apneas. Each row of tick marks is a different potential circumstance: Indep, position- and stage-independence; R-dom, REM dominance alone (2-fold); S-dom, supine dependence alone (2-fold); S&R-dom, supine dominance with worsening in REM only while supine (2-fold for each factor); S&R-dom (all supine), supine and REM dominance (2-fold each) assuming that the patient was supine for the whole night. (B) AHI values from each of the four conditions representing combinations of sleep stage and body position, as well as the total AHI (events/h).

Figure 2. Distribution of sleep stage and body position in diagnostic PSGs

(A) Box and whiskers plot of the percent of the TST spent in REM sleep, showing the median value (central line), mean value (+), 25–75th percentiles (box edges), and 5% to 95% range (whiskers). (B) Box and whiskers plot of the percent of the TST spent in the supine position. (C) Percentage of PSGs containing minimum amounts of time (X-axis in minutes) in each of the four possible combinations of REM/NREM and supine/non-supine.

Figure 3. Distribution of stage- and position-dominance: AHI ratios

(A) Frequency histogram showing the relative occurrence (Y-axis) of REM: NREM AHI ratios. Larger values indicate increasing REM-dominant OSA. (B) Frequency histogram showing the relative occurrence of supine: non-supine AHI ratios. Larger values indicate increasing supine-dominant OSA.

Figure 4. AHI underestimation related to time spent in REM and supine position

(A) Box and whiskers plots of the percent of the AHI (events/h) in our cohort in baseline conditions (“BL”) compared to that after adjustment for each individual who spent < 25% of TST in REM sleep (“corrected”). Median (central line), mean (+), 25–75th percentiles (box edges), and 5% to 95% range (whiskers) are shown. The groups are not different (Mann-Whitney rank test). (B) Box and whiskers plots of the AHI values related to body position. The baseline condition (“observed”) includes only those patients who spent 20% to 80% of the TST in the supine position. These values are compared to the supine AHI and the lateral AHI. *Significant difference by Kruskal-Wallis test with Dunn post-test.

Figure 5. OSA category misclassification: risk due to body position

(A) Bar plot of the baseline percentage (X-axis) of the cohort with AHI values in the following disease categories: none (AHI < 5; gray), mild (AHI 5–15; green), moderate (AHI 15–30; orange), and severe (AHI > 30; red). (B) Misclassification for those subjects with either < 25% or < 50% of the TST spent in the supine position (n = 32). The percentage of the cohort that could switch categories based on extrapolation of supine AHI is given.

Figure 6. Simulations of AHI underestimation due to body position

Contour plots illustrate observed AHI values (color gradient scale), when the supine AHI value is 10 (A), 22.5 (B), or 40 (C). In each panel, all possible combinations of the portion of TST spent supine (X-axis) and the relative protection of lateral position compared to supine in terms of AHI (Y-axis; L: S ratio of AHI values) are shown to provide the entire spectrum of possible underestimations in each case. The white contour lines indicate the clinical boundaries for OSA severity categories. In each panel, the observed AHI (color) equals the “true” AHI only when either there is no position dependence (top margin of each panel), or when the time spent supine is 100% (right margin of each panel). Underestimation of the AHI occurs whenever either of these conditions is violated, in proportion to the extent of position dependence of AHI and the proportion of the night spent supine.

Figure 7. OSA severity categories misclassification

Histogram of AHI values obtained from a simulated cohort of n = 1,000 subjects in each of 4 categories: AHI < 5 (purple), AHI 6-15 (pink), AHI 16-30 (gold), and AHI > 31 (green). See methods for the distribution assumptions in this and subsequent panels. Histogram of AHI values from the same cohort as in panel A after imposing position dependence to the AHI for each individual. Note the change in X-axis range. Confusion matrix showing misclassification of individuals according to the distribution in panel B. Each row is the “true” AHI assigned to each group. Each column is the OSA category assigned given stochastic variation as well as position dependence. Gray shading indicates the correctly classified members of each group. The percentage of individuals misclassified is also given based on the standard categories, as well as a single cutoff value of AHI = 15.