Screening for sleep-disordered breathing in the general population: predictive performance of four questionnaires

Abstract

Background : Several questionnaires have been used to screen for sleep-disordered breathing (SDB). Most of them have been validated in selected clinical populations, and little is known about their predictive performance in the general population. To investigate the accuracy of four questionnaires used for SDB screening in the general population: the Berlin Questionnaire (BQ), the Epworth Sleepiness Scale (ESS), the STOP-Bang (SB) and the adjusted-neck circumference (neck+). 2115 subjects (50.4% women, 58.4 ± 11.0 years old, BMI 26.2 ± 4.4 kg/m 2 ) participating in an ongoing population-based sleep cohort study (HypnoLaus, Lausanne, Switzerland) underwent complete polysomnographic recordings at home. Respiratory events were scored according to the AASM 2013 criteria. Prevalence of SDB was determined for apnea–hypopnea index (AHI) thresholds of ⩾15 and ⩾30. Interviewers administered the BQ and the ESS and calculated SB and neck+. SB includes the following items: snoring, daytime Tiredness, Observed apnea, high blood Pressure, Body-mass index ⩾35 kg/m 2 , Age ⩾ 50 yr, Neck girth ⩾ 40 cm, and male Gender. Neck + is obtained by measuring the neck girth in cm and adding 3 cm for the presence of snoring, 3 cm for observed apnea and 4 cm for high blood pressure. Subjects were considered at increased risk for SDB if BQ is positive in ⩾2 categories, ESS score ⩾11, SB score ⩾3 and neck+ ⩾43 cm. Assessment of questionnaire included their sensitivity, specificity, positive-predicted value (PPV) and negative-predicted value (NPV). Prevalence of SDB with AHI thresholds of ⩾15 and ⩾30 were respectively 36.4% and 15.0%. Percentages of positive BQ was 25.3%, ESS:13.4%, SB: 41.3% and neck+:29.0%. For an AHI ⩾ 15, sensitivity, specificity, PPV and NPV were respectively: BQ (39.4%, 82.7%, 56.6% and 70.5%), ESS (13.5%, 86.7%, 35.8% and 64.5%), SB (67.3%, 72.8%, 57.3% and 80.4%) and neck+ (52.3%, 83.8%, 64.0% and 76.1%). For an AHI ⩾ 30, sensitivity, specificity, PPV and NPV were respectively: BQ (48.7%, 78.8%, 28.8% and 89.7%), ESS (13.1, 86.6%, 14.2% and 85.4%), SB (81.7%, 65.6%, 28.7% and 95.5%) and neck+ (69.8%, 77.9%, 34.6% and 93.9%). Our population-based study suggests that questionnaires such as BQ, ESS, SB and neck+ could be used to rule out severe SDB considering their strong NPV. However they do not seem suitable to screen for moderate to severe SDB in the general population. Funding : Ligue pulmonaire vaudoise, Fondation Leenaards, Fond national Suisse and GSK.