SESSION TITLE: Sleep Disorders Poster Discussions
SESSION TYPE: Original Investigation Poster Discussion
PRESENTED ON: Wednesday, October 28, 2015 at 07:30 AM - 08:30 AM
PURPOSE: Using data collected at the initial clinic encounter we calculated PTP for obstructive sleep apnea using the Berlin Score. We also calculated PTP using two other well validated tools and compared the performance of the different PTP calculators. All patients had in lab polysomnography scored using 2012 AASM criteria.
METHODS: Using data collected at the initial clinic encounter we calculated PTP for OSA using the Berlin Score. We also calculated PTP using two other well validated tools and compared the performance of the different PTP calculators. All patients had in lab polysomnography scored using 2012 AASM criteria.
RESULTS: There were 265 patients who had data available for analysis. The mean age and BMI were 40.6±10.6 and 28.9±4.4 respectively, and the mean ESS was 9.9±9.0 and median AHI was 8.9 (4.3-20.5). The majority of patients (184 (71.0%)) had an AHI ≥ 5 and were considered positive for OSA and 170 (65.6%) had a high risk Berlin score. Using the Berlin score, the sensitivity and specificity for OSA were 24.7% and 68.8% respectively. Two other previously derived scores were also tested, and sensitivity and specificity ranged from 28.2-29.2% and 70.3-71.6% respectively.
CONCLUSIONS: The Berlin score did not perform well in our population, and two other validated scores were only slightly better. It is unclear whether poor score performance was due to a lack of generalizability to a military population or change in hypopnea scoring resulting in an increased OSA prevalence.
CLINICAL IMPLICATIONS: Pre-test probabality tools used to assess for sleep apnea may not be applicable to all populations. Scores should be re-calibrated using 2012 scoring criteria to see if recommendations for scoring respiratory events significantly alter sensitivity and specificity.
DISCLOSURE: The following authors have nothing to disclose: Aaron Holley, Karen Sheikh, Michael McMahon, Robert Walter
No Product/Research Disclosure Information