Introduction and objective:
Severe obstructive sleep apnea (OSA), defined by the Apnea Hypopnea Index as AHI ≥30, is associated with increased cardiovascular, metabolic, and neurocognitive risks but is frequently overlooked in primary care. The aim of this study is to identify the predictors of severe OSA, and to derive a simple, easy-to-use a clinical model as a guide for pre-specialist triage.

Material and methods:
A retrospective single-centre study was carried out of 278 consecutively hospitalized adults assessed for sleep-disordered breathing (Lublin, 2018–2020). Polygraphy/polysomnography provided AHI and oximetry indices. Receiver operating characteristics (ROC) analyses identified thresholds for continuous variables; predictors were tested in univariate and multivariable logistic regression. Primary outcome: severe OSA (AHI ≥30).

Results:
Severe OSA occurred in 56.8% of patients (158/278). ROC analysis suggested cut-offs at: age ≥55 years, body mass index (BMI) ≥30 kg/m², Epworth Sleepiness Scale (ESS) ≥10, and High Dense Lipoprotein (HDL) ≤53 mg/dL (best performance for BMI: area under curve (AUC) 0.756; sensitivity 70.9%, specificity 70.0%). In multivariable models, independent predictors included: BMI ≥30 (OR 5.37, 95% CI 2.70–10.70), age ≥55 (OR 2.66, 95% CI 1.36–5.20), and ESS ≥10 (OR 2.77, 95% CI 1.46–5.21). A three-item rule (age, BMI, ESS) produced predicted probabilities over 50% for several combinations, and overall discrimination was approximately 0.75.

Conclusions:
Obesity is the strongest predictor of severe OSA, with additional influences from older age and daytime sleepiness. A simple three-item rule may assist primary care in triaging referrals for sleep testing; external validation is necessary.