Assessing the reliability of obstructive sleep apnea screening instruments in isolation or in combination

Abstract

Study objectives

A) Compare the standard obstructive sleep apnea (OSA) screening instruments to each other and to the judgement of experienced sleep clinicians. B) Evaluate if a novel OSA screening instrument constructed from components of the standard OSA screening instruments would improve the accuracy of screening.

Methods

We screened 344 subjects using the most commonly used OSA screening instruments- Berlin sleep questionnaire, STOP, STOP-BANG, Mallampati upper airway score and Epworth sleepiness scale. A clinical impression score (CIS) reflected the clinician's likelihood of OSA being present. All subjects underwent nocturnal polysomnography. The sensitivity and specificity of all screening instruments and the CIS for any OSA and for moderate-to-severe OSA were calculated. We constructed a logistic regression model incorporating all the components of these OSA screening instruments in an effort to develop a more accurate OSA screening metric.

Results

STOP-BANG was the most sensitive instrument (sensitivity of 93% and 97% for any OSA and moderate-to-severe OSA respectively). Its ability to exclude OSA was on par with the CIS (sensitivity of 95% and 97% respectively). The logistic regression model incorporating all the individual items of the OSA screening instruments did not meaningfully improve accuracy.

Conclusions

A comprehensive clinical evaluation and STOP-BANG both possess sufficiently high sensitivity to be useful screening tools. Combining the existing OSA screening instruments into a single instrument did not appreciably improve upon the accuracy of the STOP-BANG. Further refinements of the existing OSA screening instruments would require identifying new, easily recognizable risk factors.