Alternative scoring models of STOP-bang questionnaire improve specificity to detect undiagnosed obstructive sleep apnea

Abstract

Background: Obstructive sleep apnea (OSA) is common among surgical patients. The STOP-Bang questionnaire is a validated screening tool with a high sensitivity. However, its moderate specificity may yield fairly high false positive rate. We hypothesized that the specific combinations of predicting factors in the STOP-Bang questionnaire would improve its specificity.

Methods: After research ethics approval, consented patients were asked to complete the STOP-Bang questionnaire and then underwent sleep studies. The predictive performance of the STOP-Bang alternative scoring models was evaluated. Five hundred sixteen patients with complete data on the STOP-Bang questionnaire and polysomnography were reported.

Results: When the STOP-Bang score was ≥ 3 (any 3 positive items), the sensitivity and specificity for identifying moderate-severe OSA was 87% and 31%, respectively. The specificity for any 2 positive items from the 4 STOP questions plus BMI > 35 kg/m(2), male gender, or neck circumference > 40 cm for identifying moderate-severe OSA was 85%, 77%, and 79%, respectively. Compared with STOP-Bang score ≥ 3, the predicted probability for severe OSA of the specific combinations of STOP score ≥ 2 + male and STOP score ≥ 2 + BMI increased by 36% and 42%, respectively. For severe OSA, the specific combination of STOP score ≥ 2 + BMI + male demonstrated a specificity of 97% and 86% increase in predicted probability versus any 4 positive items of STOP-Bang questionnaire.

Conclusions: The specific constellations of predictive factors improved the specificity of STOP-Bang questionnaire. For patients with STOP score ≥ 2, male gender, and BMI > 35 kg/m(2) were more predictive than age ≥ 50 and neck circumference > 40 cm.

Keywords: Obstructive Sleep Apnea; perioperative care; polysomnography; preoperative screening; sleep apnea questionnaire.

Figure 1. Flow chart of patient recruitment.

Figure 2

(A) Post-test probability of moderate-severe OSA (AHI > 15 events/h) and (B) severe OSA (AHI > 30 events/h) by using a STOP-Bang 3 or greater and specific combination of 3 items as the cutoff value. Data were represented as probability with the 95% confidence interval for the specific combinations of STOP ≥ 2 + one item from Bang.

Figure 3

(A) Post-test probability of moderate-severe OSA (AHI > 15 events/h) and (B) severe OSA (AHI > 30 events/h) by using a STOP-Bang score 4 or greater and specific combination of 4 items as the cutoff value. Data were represented as probability with the 95% confidence interval for the specific combinations of STOP ≥ 2 + two various items from Bang.

Figure 4. Flow chart for two-step OSA screening strategy with STOP-Bang questionnaire.

Patient number, the probability of moderate-severe OSA (AHI > 15 events/h) and severe OSA (AHI > 30 events/h) were presented in each group.