severe and mild to severe categories, respectively. Di-
agnosis status was based on what the participant's phy-
sician had concluded after evaluation, regardless of
whether the participant was evaluated in a sleep lab-
oratory. Requiring polysomnographic evidence for di-
agnosis may have eliminated false positives but could
have also eliminated true positives. The latter conse-
quence would seriously underestimate the true propor-
tion of clinically recognized SAS.
Comparison of clinically diagnosed and screen-de-
tected cases on sociodemographic characteristics sug-
gested that diagnosed cases were more likely to be
men, have a history of cardiovascular disease, be older,
be Caucasian, and have higher income and educational
attainment. The higher proportion of cardiovascular
disease in clinically diagnosed cases may reflect sleep
clinic referral as a result of medical care for comorbid
conditions (see Table 1).