Reduced diagnostic accuracy of screening questionnaires in older psychiatric patients at risk for obstructive sleep apnea: an observational cohort study

Background Obstructive sleep apnea (OSA) is highly prevalent among geriatric patients and is associated with increased mortality, reduced quality of life, depression, and cognitive decline. Early identification and treatment are particularly important in older psychiatric patients, in whom typical OSA symptoms overlap with psychiatric and aging-related symptoms. Screening questionnaires are used to identify patients at risk for OSA, but their performance in older psychiatric populations is largely unknown. This study compared the diagnostic accuracy of screening questionnaires between younger and older psychiatric inpatients at risk for OSA and assessed the discriminative value of chronological age alone. Patients and methods Observational study of consecutive admissions to a psychiatric ward at University Hospital Erlangen (June 2016–December 2022). 128 screened patients at risk for OSA with valid overnight polygraphy and questionnaire data were analyzed (72 aged 18–64 years; 56 aged ≥ 65 years). Risk scores were calculated for STOP-Bang, STOP-B28, BOAH, GOAL, NoSAS and No-Apnea questionnaires. The reference standard was apnea-hypopnea-index (AHI) measured by polygraphy. Primary outcome measures were the area under the receiver operating characteristic curve (AUC), sensitivity and specificity for any OSA (AHI ≥ 5 events/h). The discriminative value of age and individual questionnaire components was evaluated in the entire cohort. Results In the screened cohort of psychiatric inpatients, older patients had higher median AHI (7.7 vs. 1.8 events/h; p = 0.007), greater prevalence of any OSA (AHI ≥ 5: 73.2% vs. 47.2%; p = 0.003), deeper desaturations and longer apneas. In younger patients, questionnaire AUCs ranged 0.61–0.78. In the older subgroup, AUCs were 0.12–0.36 lower; BOAH had the highest AUC (0.59), while the other questionnaires showed AUCs of 0.38–0.49. The lower discrimination was mainly reflected in lower specificity. Chronological age yielded an AUC of 0.70 (95% CI, 0.61–0.80), whereas BMI, neck circumference, and symptom-based items showed poor discrimination. Conclusions In our sample of psychiatric inpatients at risk for OSA, commonly used screening questionnaires demonstrated markedly reduced diagnostic performance in patients aged ≥ 65 years, largely reflecting lower specificity. Chronological age demonstrated moderate discriminative value. Further research is warranted regarding the accuracy of OSA screening in older patients, and a lower threshold for device-based diagnostics may warrant consideration.