STOP-Bang Modification for Screening Treatment-Relevant Obstructive Sleep Apnea in Cluster Headache: A Retrospective Cohort Study.

Background

Cluster headache (CH) and obstructive sleep apnea (OSA) are both associated with elevated cardiovascular risk and unhealthy lifestyle factors, including smoking and alcohol use. This study investigated whether the conventional STOP-Bang questionnaire or a modified screening approach incorporating CH-specific clinical features is more useful for identifying treatment-relevant OSA.

Methods

We retrospectively analyzed 52 patients with CH who underwent polysomnography (PSG) between January 2020 and December 2025. An apnea-hypopnea index (AHI) of at least 10 events/hour was defined as treatment-relevant OSA. We assessed whether incorporating age at CH onset and the circadian and/or seasonal rhythmicity of CH attacks into the STOP-Bang questionnaire improved screening performance.

Results

Of 52 patients, 41 (78.8%) were diagnosed with OSA (AHI ≥ 5) and 25 (48.1%) had treatment-relevant OSA. Patients with treatment-relevant OSA were predominantly male and had a higher median age, body mass index, and age at CH onset. A modified model incorporating CH onset at or after age 30 showed modest improvement than conventional STOP-Bang. At a cut-off point of ≥3, the modified model demonstrated a numerically higher area under the curve (0.804 vs. 0.767) and greater sensitivity (84.0% vs. 76.0%; DeLong p = 0.125), with low specificity (48.1% vs. 55.6%). At a cut-off point of ≥4, the specificity increased to 77.8%. The addition of other clinical characteristics of CH did not further improve the performance.

Conclusions

Incorporating CH onset age into the STOP-Bang modestly improved screening for treatment-relevant OSA in CH patients, representing a practical screening refinement warranting external validation.