Rationale: There are no published data regarding use of the STOP-BANG sleep apnea questionnaire in populations referred to Veterans Affairs (VA) sleep facilities. If a particular STOP-BANG score cutpoint had high positive predictive value in this referral population, it could reduce the need for diagnostic sleep studies.
Methods: STOP-BANG questionnaires were prospectively administered to veterans undergoing unattended sleep studies at a single VA facility. We evaluated the sensitivity, specificity, positive predictive value, and area under the receiver-operating characteristic curve (ROC AUC) of STOP-BANG scores for identifying a Respiratory Disturbance Index (RDI) greater than 15/hour. We also recalibrated the STOP-BANG score to our referral population, using logistic regression models.
Measurements and main results: Of 1,196 consecutive veterans undergoing unattended sleep studies, the mean STOP-BANG score was 5.7 ± 1.4, and 67% had an RDI greater than 15/hour. Sensitivities were excellent at lower STOP-BANG scores, but sharply decreased at scores of 6 and above. Specificity improved in a linear fashion with increasing scores. The ROC AUC was 0.66 (95% confidence interval [CI], 0.64-0.69) and recalibrated models improved the ROC AUC to 0.74 (95% CI, 0.69-0.78). The highest STOP-BANG score of 8 was present in only 7.9% of the sample and had a positive predictive value of 85% (95% CI, 76-92%).
Conclusions: The STOP-BANG questionnaire alone is insufficient to confirm the presence of significant sleep apnea. A maximal score of 8 did not have a high enough positive predictive value to forego confirmatory sleep testing.