Finally, individual data were analyzed to determine if some ranges or combinations of ranges of variables were highly predictive of either the presence or the absence of OSA.
Results are expressed as mean ± SD unless indicated. We considered a p value < 0.05 to be significant. Logistic regression analysis was repeated with an AHI threshold of 10 for definition of OSA. Statistical analysis was performed with BMDP (BMDP Statistical Software; Los Angeles, CA), SPSS (SPSS Inc; Chicago, IL) and SAS (SAS Institute; Cary, NC) statistical software. ROC curve analysis was performed with ROC Analyzer software (RM Centor and J Keightley; Richmond, VA).
Results
The study population consisted of 34 women and 68 men, whose characteristics are summarized in Table 1. BMI and FEV1/FVC were higher and tobacco smoking was less frequent in women than in men. OSA was more frequent in men (p < 0.01; Table 1). Airflow obstruction (as defined by the American Thoracic Society) was present in 28% of our male population vs 12% of women, but this difference did not reach significance (p = 0.08), The proportion of patients with OSA did not differ between subjects with or without airflow obstruction, among both men and women (data not shown; p = not significant [NS] for all comparisons). AHI did not correlate with FEV1 and FEV-/FVC (r = 0 0.402 and r = —0.265, respectively; p = NS). canadian-familypharmacy.com

Patients with OSA exhibited the following characteristics when compared with those who did not have OSA: higher CS, lower diurnal Pa02, lower nocturnal minSaO2, mSa02, and higher CT80 (Table 2). These differences were found for continuous as well as for categorical variables. Thresholds used for transformation of continuous variables into categorical variables, as determined by ROC curve analysis, are shown in Table 3. Table 3 also shows the sensitivity and specificity of these individual variables for the diagnosis of OSA.

Table 1—Anthropometric, Polysomnographic, and Pulmonary Function Data

Women Men
No OSA OSA No OSA OSA
No. of patients 28 6 34 34
Age, yr 53.5 ± 10.9 53.0 ± 17.8 51.7 ± 10.7 54.4 ± 10.1
BMI, kg/m2 43.4 ± 9.4 53.5 ± 13.3 36.3 ± 6.5 37.2 ± 7.9*
Tobacco smoking, No. (pack-yr) 8 (12 ± 24) 1(10) 27 (32 ± 29) 28 (31 ± 29)*
AHI, events/h 3.6 ± 4.1 45.8 ± 19.3 4.5 ± 3.8 42.2 ± 19.0
FVC, % pred 81.2 ± 19.1 91.0 ± 18.4 83.0 ± 18.5 82.4 ± 17.8
FEVj, % pred 83.9 ± 19.7 87.8 ± 14.1 76.7 ± 24.6 80.6 ± 21.4
FEV/FVC, % pred 85.1 ± 9.1 82.2 ± 10.7 73.9 ± 13.7 77.9 ± 9.5*

Table 2—CS, Arterial Blood Gases, Indices of UAO, and Oximetric Data

No OSA OSA p Value
No. of patients 62 40
CS 2.6 ± 1.0 3.3 ± 0.8 < 0.05
Pao2, mm Hg 76.7 ± 16.1 70.9 ± 11.7 < 0.05
Paco2, mm Hg 41.3 ± 5.7 42.8 ± 5.01 NS
Aa gradient, mm Hg 21.8 ± 13.9 25.6 ± 8.6 NS
UAO criteria
Saw-toothing, % 45 58 NS
FEF50/FIF50 1.2 ± 0.6 1.3 ± 0.6 NS
FEV1/FEV0.5 1.3 ± 0.2 1.3 ± 0.1 NS
pef/fef50 2.3 ± 1.7 2.1 ± 1.0 NS
Oximetry, %
minSao2 66.5 ± 12.7 58.5 ± 15.7 < 0.05
mSao2 87.9 ± 4.5 82.9 ± 10.2 < 0.05
CT90 55.4 ± 35.4 61.7 ± 31.8 NS
CT80 9.4 ± 17.2 23.9 ± 25.7 < 0.05

Table 3—Best Thresholds and Diagnostic Characteristics of Clinical, Oximetric, and Functional Variables

Threshold Sensitivity Specificity Area Under ROC Curve (Mean ± SEM)
mSao2 85% 0.48 0.80 0.67 ± 0.05
CT80 5% 0.63 0.66 0.67 ± 0.05
CS 3 0.80 0.45 0.64 ± 0.05
Pao2, mm Hg 70 0.53 0.64 0.64 ± 0.05
FEV1/FEV0.5 1.3 0.65 0.56 0.62 ± 0.06
FEV1 80% 0.65 0.53 0.59 ± 0.06
FEF50/FIF50 120 0.58 0.59 0.59 ± 0.06
pef/fef50 175 0.59 0.53 0.54 ± 0.06