Developing quantitative physiological phenotypes of sleep apnea for epidemiological studies

Existing physiological databases have not been sufficiently detailed to provide relevant and important information for characterizing the pathophysiology of obstructive sleep apnea. Critical collapsing pressure (P_CRIT) is a standard method for determining upper airway patency during sleep, however is labor intensive and prohibits large-scale studies. Based on previously published data indicating R_US does not significantly vary between groups, our aim was to develop an approach to estimate the P_CRIT from airflow at atmospheric pressure (V_atm). In a dataset of 126 subjects, where P_CRIT and R_US were measured using standard techniques. We then determined the minimum sample size required to estimate the R_US mean and variance by utilizing a bootstrap procedure (30 times for n=3 to 126). We first estimated the minimum number of subjects needed for obtaining a group for a two-tailed (z=1.96) standard error for R_US in the population. Then in 75 individuals, quantitative estimates of airflow were obtained at atmospheric pressure. Using the estimated R_US and atmospheric, we determined an estimated P_CRIT (€P_CRIT). Bland-Altman plots were generated to determine the agreement between the measured P_CRIT and €P_CRIT. For the entire population the mean±SEM R_US was 23±1cmH₂O/L/s (±95% CI: 21, 25). ~40 subjects represent the minimum sample required to estimate the population variance within ±2 SEM. In the subsample with atmospheric flow measurements, a linear regression model (€P_CRIT [cmH₂O]=V_@PN [L/s]x-23[cmH₂O/L/s]), €P_CRIT ranged from 0 to -9.6cmH₂O. In the Bland-Altman analysis there was no mean difference between the measured P_CRIT and €P_CRIT (-0.01cmH_{2O; p=0.8) with upper and lower limits of agreement at ±2.3cmH2O. The variance of upstream resistance approaches a constant value in groups with approximately 40 subjects. Utilizing a fixed up-stream resistance to estimate €PCRIT from the airflow at atmospheric pressure agrees with the measured values. These data suggest that measurements of quantitative airflow during standard polysomnography can be used to determine upper airway properties in large cohorts.}