SESSION TITLE: Sleep Disorders
SESSION TYPE: Original Investigation Slide
PRESENTED ON: Sunday, October 25, 2015 at 07:30 AM - 08:30 AM
PURPOSE: The STOP-Bang questionnaire has been validated as a screening tool for obstructive sleep apnea (OSA) in patients during preoperative assessment. It’s applicability to the general population is less well studied with only a few studies to date. We wanted to assess the ability of the questionnaire to identify patients with OSA in the primary care setting.
METHODS: The STOP-Bang questionnaire was completed prospectively by patients sent by their primary care physician for level 3 testing at our regional sleep laboratory. Testing was done by a registered sleep technologist who then manually scored the study. The raw data was then manually reviewed by a sleep physician to determine a diagnosis of OSA. The diagnostic performance of the STOP-Bang score was analyzed (sensitivity and specificity) using two distinct criteria - 1) interpreting sleep physician decision to treat - ‘positive study’ and 2) AHI of ≥15/hour.
RESULTS: A total of 312 patients completed the questionnaire and went on to have a level 3 study done. Analysis based on decision to treat found that a score of ≥2/8 had a sensitivity of 97.2%. Using an AHI cutoff of ≥15/hour the same score had a sensitivity of 98.0%. A score of ≥3/8 yielded sensivities of 88.3% and 94% respectively. The corresponding specificities were low. A score of ≥6/8 or higher had 95.9% specificity for OSA. Using an AHI cutoff of ≥15/hour, the same score had a specificty of 95%. A score of ≥7/8 increased the specificity to 100% and 98.1% respectively. Corresponding sensitivities were low.
CONCLUSIONS: A STOP-Bang score of <2/8 appears to rule out an AHI of ≥15/hour and a score of ≥6/8 rules in an AHI of ≥15/hour in this population.
CLINICAL IMPLICATIONS: In this primary care setting, a STOP-Bang score of less than 2 may be used to rule out significant OSA and a score of 6 or higher may be used to establish a diagnosis of OSA. Application of this model may allow for improved resource utilization and triage of diagnostic testing.
DISCLOSURE: The following authors have nothing to disclose: Mark Fenton, Sam Stewart, Robert Skomro, John Gjevre, John Reid
No Product/Research Disclosure Information