Sleep Breathing Disorders VALIDATION OF A NEW CRITERION FOR CHARACTERIZATION OF POSITIONAL OBSTRUCTIVE SLEEP APNEA D. Levendowski 1 , A. Oksenberg 2 , C. Vicini 3 , D. Dawson 4 , M. Levi 5 , P. Westbrook 1 . 1 Advanced Brain Monitoring, Inc., Carlsbad, United States; 2 Rehabilitation Center, Loewenstein Hospital, Raanana, Israel; 3 Head and Neck Department, Morgagni-Pierantoni Hospital, Forlì, Italy; 4 Department of Anaesthesia and Sleep Medicine, Bradford Teaching Hospitals, Bradford, United Kingdom; 5 Dr. Levi's Sleep Clinic, West Gosford, Australia Introduction: Positional Obstructive Sleep Apnea (POSA) has been conventionally defined by the Cartwright criterion i.e., supine 2 times the non-supine severity. An alternative criterion based on the overall divided by the non-supine (NS) severity was used as enrollment criterion for a POSA research study. This study compares the two criteria with a range of Overall/NS-AHI thresholds. Materials and methods: This retrospective analysis was conducted with studies acquired using the Sleep Profiler-PSG2™ (Advanced Brain Monitoring, Carlsbad, CA). The self-applied, in-home recordings were made with electroencephalography acquired from three frontopolar, airflow using a nasal cannula and pressure transducer, head movement/ position by actigraphy, snoring with an acoustic microphone, pulse from the forehead and finger, wireless wrist oximetry, and thorax and abdomen effort by respiratory induced plethysmography. The 73 females and 69 males had a mean age of 45±13.5 years and body mass index of 32±7.2. The same technician performed a focused review of the full disclosure recordings on the cloud-based portal to confirm the sleep staging accuracy, insert apnea or hypopnea events when the amplitude of the airflow signal was extremely low, or remove events due to artifact/ movement (average of 12.7±4 minutes per record). OSA severity was measured using the 3% desaturation rule with total sleep time and AASM 2012 criteria for the apnea-hypopnea index (AHI) and total recording time with a oxygen desaturation index (ODI-3%). POSA prev- alence was based on 142 records with an AHI>5 (16 with an AHI5 and ODI-3%< 5). The Cartwright and Overall/NS-AHI criteria (with thresholds of 1.3, 1.35 and 1.4) were compared for both AHI and ODI-3%. Sensitivity and specificity was based on the reference standard the NS severity 25% the overall severity. Results: Strong correlations were observed between Overall/NS-AHI and Cartwright ratios derived from AHI/ODI-3% measures (r ¼ .86/.90, p< 0.00001). POSA prevalence decreased as the Overall/NS-AHI thresholds increased (i.e., 1.3, 1.35 and 1.4) for both the AHI (58%, 62% and 63%) and ODI-3% (58%, 56%, and 52%), respectively. By comparison, the POSA prevalence by Cartwright was AHI ¼ 65% and ODI-3% ¼ 57%. Based on a 25% reference standard across all subjects, the AHI sensitivi- tyjspecificity was .93j.71 for Cartwright compared to 1.00j1.00, .99j1.00, and .93j 1.00 for the Overall/NS-AHI thresholds. For ODI-3%, the sensitivi- tyjspecificity was .92j.87 for Cartwright versus 1.00j.96, .99j.96, and .93j1.00 for the Overall/NS-AHI thresholds. The sensitivityjspecificity by Cartwright compared to the Overall/NS-AHI thresholds for 5AHI< 15 ¼ .95j.87 versus 1.00j1.00, 1.00j1.00, .89j1.00; for 15  AHI  30 ¼ .93j.67 versus 1.00j1.00, .97j1.00 .97j1.00; and for AHI  30 ¼ 91j .81 versus 1.00j1.00, .97j1.00 and .96j1.00. Comparisons of Cartwright to the Overall/NS-AHI thresholds for the ODI-3% yielded a sensitivityjspecificity of .88j.95 versus 1.00j1.00, .97j1.00, .91j 1.00 for ODI-3%< 15; 1.00j.74 versus 1.00j.89, 1.00j0.89, .95j 1.00 for 15ODI- 3% 30; and 91j .81 versus 1.00j1.00, 1.00j1.00, and .94j 1.00 for ODI-3%30. Conclusions: A strong correlation was observed between the Cartwright and Overall/NS-AHI ratios, both estimated POSA prevalence >60% by AASM 2012 criteria and >55% by ODI-3%. When the NS was >25% the overall severity, the sensitivity and specificity of Overall/NS-AHI was superior to Cartwright across both measures and all ranges of OSA severity. Sleep Breathing Disorders AGREEMENT BETWEEN AUTO-SCORED VS. EDITED UNATTENDED IN- HOME POLYSOMNOGRAPHY D. Levendowski 1 , D. Dawson 2 , M. Levi 3 , P. Westbrook 4 . 1 Advanced Brain Monitoring, Inc., Carlsbad, United States; 2 Department of Anaesthesia and Sleep Medicine, Bradford Teaching Hospitals, Bradford, United Kingdom; 3 Dr. Levi's Sleep Clinic, West Gosford, Australia; 4 Advanced Brain Monitoring, Carlsbad, United States Introduction: Auto-scored in-home polysomnography accuracy can be impacted by the signal quality associated with self-application. This study evaluated when the auto- and edited-findings were equivalent. Materials and methods: This retrospective analysis was conducted in 218 consecutive Sleep Profiler-PSG2 TM studies (Advanced Brain Monitoring) that included 54% males, mean age 45±13.5 years, and BMI 31±7.0 kg/m 2 . Patients self-applied a forehead worn device to affix three frontopolar electroencephalography (EEG) sensors and nasal cannula, a wireless wrist oximeter, and thorax/abdomen RIP belts. After watching a 5-min video, patients briefly practiced affixing the device. During the night voice mes- sages alerted the patient if the SpO2 sensor probe was not affixed and up to four times/night when the cannula was misplaced. Auto-scored staged sleep and detected OSA severity using methods to emulate the AASM 2007 criteria with a 4% desaturation (AHI 2007 ) and AASM 2012 criteria with a 3% desaturation (AHI 2012 ). The same technician performed a review to confirm the sleep staging accuracy, insert apnea or hypopnea events during periods with airflow or SpO2 loss, or remove events due to artifact/movement. Failed studies included those with recording times with <50% airflow, < 90% SpO2 and/or < 90% good EEG in all three channels. The impact of editing on auto-scored results was evaluated by stratifying non-rejected studies into three groups; 1) “High Quality” i.e., airflow90%, three EEG90% and WristOx95% (n ¼ 156), 2) “Marginal EEG” i.e., only one EEG90% (n ¼ 24), and 3) “Airflow/SpO2” with airflow 50-89% or SpO2 90-95% (n ¼ 26) of recording time. Results: The PSG2 failure rate of 5.5% resulted from 8 records with extensive airflow loss and 4 records with EEG loss. For the remaining 206 studies, the percentage with study times 6, 7 and 8 h were 91%, 71% and 44%, respectively. The percentage of sleep times 4, 5 and 6 h was 84%, 74% and 46% for auto-staging, and 90%, 79% and 55% after editing. The median time to edit the records was 10-min (Inter-quartile range 10-15); increased data loss did not significantly impact editing time. Editing decreased the percent-time Stage N1 by 2.9±4.7% and increased REM by 2.1±5.1%. High Quality studies had significantly greater edited sleep time vs. Marginal EEG and Airflow/SpO2 (p<0.001, 6.37±1.27, 5.28±1.49, 4.82±1.39 h, respectively). Comparisons of the edited vs. auto-scored AHI 2007 resulted in sensitivity, specificity, PPV and NPV agreements exceeding 0.95 for clinical cutoffs 5, 10 and 15 in the High Quality and Marginal EEG groups. For the Airflow/ SpO2 group, the accuracy values were >0.90 for AHI5 and 1.00 for AHI>15. The accuracy measures with AHI 2012 exceeded 0.95 for clinical cut-offs 10 and 15 in High Quality studies, with lower specificities in the Marginal EEG group (i.e., 0.75 and 0.86 for cut-offs10 and 15), and compromised accuracies with the Airflow/SpO2 group. Conclusions: With AHI 2007 and 4% desaturation, edited and auto-scored OSA severities were equivalent across all clinical cut-off and quality group. With AHI 2012 and 3% desaturation, the auto-scored OSA severities were reliable with clinical cut-offs>10, except when airflow or SpO2 loss exceeded 10% or 5% of recording time. Restless Legs Syndrome (RLS) ASSOCIATION OF BTBD9 AND MAP2K5/SKOR1 WITH RESTLESS LEGS SYNDROME IN CHINESE POPULATION G. Li , H. Tang, C. Wang, X. Qi, J. Chen, S. Chen, J. Ma. Neurology, Ruijin Hospital Affiliated to Shanghai Jiaotong University School of Medicine, Shanghai, China Introduction: Genetic factors play an important role in the pathogenesis of RLS. Linkage studies identified eight suspected genetic loci accounting for RLS. In addition, genome wide association study (GWAS) showed that several single nucleotide polymorphisms (SNPs) were associated with the risk of RLS, including MEIS1, BTBD9, PTPRD, MAP2K5, TOX3, and Intergenic region of 2p14. A low prevalence of RLS is reported in Asian Abstracts / Sleep Medicine 40 (2017) e186ee363 e189