0419 Sensitivity of ORP to classify individuals with insomnia based on subjective complaints

Abstract

Abstract Introduction The diagnosis of insomnia is currently based on subjective complaints. Previous research has demonstrated limited benefits of polysomnography in improving diagnostic accuracy with no evident differences in sleep macro-architecture and traditional indices of sleep quality. In this study, we aimed to assess the ability of a novel index of sleep quality, the Odds Ratio Product (ORP), to correctly identify individuals with insomnia, based on subjective complaints. Methods We analyzed responses to a custom sleep survey from 510 participants (age 46.2±12.4; 229 females). We entered 13 outcomes to a principal component analysis (PCA). These included age, body mass index (BMI), Epworth Sleepiness Scale (ESS) scores, sex, questions about snoring, subjective hours of sleep, satisfaction with own sleep, movement symptoms, problems falling asleep, staying asleep and waking up early. Based on the correlations between the original outcomes, the PCA extracted 3 factors that were interpreted as 3 different phenotypes: 1) insomnia (e.g., sleep initiation and maintenance complaints, satisfaction with sleep, lower subjective sleep duration); 2) OSA (e.g., bed partner snoring complaints, snoring self-awareness, sleepiness, respiratory complaints); 3) movement disorder (e.g., movements, female, lower age, higher BMI). We then computed the pre-test probability of insomnia (factor 1 scores >80%) and conducted a series of ROC analyses to assess specificity of PSG indices traditionally used to evaluate the presence of insomnia, and of ORP metrics. Results The AUC values assessing the ability of ORP metrics to predict 80% pre-test probability of insomnia were: ORPwake (0.592, 95% CI 0.499-0.685), ORPNREM (0.465, 95% CI 0.362-0.558), ORPREM (0.542, 95% CI 0.447-0.636), and ORPTRT (0.567, 95% CI 0.475-0.659). When looking at traditional indices like total sleep time (TST), wake after sleep onset (WASO), and number of awakenings, the AUC values were 0.471 (95% CI 0.385-0.556), 0.612 (95% CI 0.528-0.697) and 0.509 (95% CI 0.421-0.597), respectively. Conclusion WASO, ORPwake and ORPTRT have the highest sensitivity to detect insomnia in participants with >80% pretest probability measured by subjective complaints. These analyses indicate the value of EEG metrics of sleep quality to correctly identify individuals with insomnia and the need to investigate both sleep and wake states. Support (if any)