Predictive factors for CPAP failure in obstructive sleep apnea patients

Area of Science:

• Respiratory Medicine
• Sleep Science
• Clinical Prediction Modeling

Background:

• Obstructive sleep apnea (OSA) affects millions, with Continuous Positive Airway Pressure (CPAP) as a primary treatment.
• A subset of OSA patients do not respond adequately to CPAP therapy.
• Bi-level Positive Airway Pressure (Bi-level PAP) serves as an alternative for CPAP treatment failures.

Purpose of the Study:

• To identify and validate predictors of CPAP failure in obstructive sleep apnea patients.
• To develop a predictive model for CPAP treatment non-response.
• To inform clinical decision-making for optimizing OSA management.

Main Methods:

• A theory-driven, hierarchical approach was used to select potential predictors.
• Data from OSA patients undergoing CPAP or Bi-level PAP titration (June 2015-October 2017) were analyzed.
• Five competitive logistic regression models were compared, with internal and external validation performed.

Main Results:

• A logistic regression model identified five significant predictors of CPAP failure.
• Predictors include: Age >60 years (OR=3.23), BMI >35 Kg/m² (OR=4.25), FEV <60% (OR=7.33), Apnea-Hypopnea Index >75 (OR=4.31), and T90 >30% (OR=6.67).
• The selected model demonstrated significant deviance reduction compared to the baseline.

Conclusions:

• The identified factors (BIPAP acronym) can predict CPAP failure in OSA patients.
• These predictors can assist clinicians in making timely decisions regarding treatment escalation.
• Improved prediction may lead to more vigilant and personalized care for OSA patients.