Development of a prediction model for postoperative pneumonia: A multicentre prospective observational study

Area of Science:

• Anesthesiology
• Critical Care Medicine
• Pulmonary Medicine
• Surgical Outcomes Research

Background:

• Postoperative pneumonia significantly increases patient morbidity, mortality, and healthcare costs.
• Existing pneumonia prediction models often rely on outdated data and administrative coding.
• There is a need for robust, prospective models to identify patients at risk of postoperative pneumonia.

Purpose of the Study:

• To identify independent predictors of postoperative pneumonia.
• To develop and validate a predictive model for postoperative pneumonia.
• To improve risk stratification and management strategies for patients undergoing surgery.

Main Methods:

• A prospective, multicenter observational study involving 5094 patients across 63 European hospitals.
• Data collected during a 7-day recruitment period from patients undergoing general and/or regional anesthesia.
• Postoperative pneumonia defined by specific clinical criteria including antibiotic treatment, sputum changes, lung opacities, fever, and leukocyte count.

Main Results:

• Postoperative pneumonia occurred in 2.4% of patients (120/5094), with a higher ICU admission rate (68.3% vs 8.0%).
• Five independent predictors identified: poor functional status, low pre-operative SpO2, intra-operative colloid administration, intra-operative blood transfusion, and open upper abdominal surgery.
• The predictive model demonstrated excellent discrimination (c-statistic 0.89) and calibration.

Conclusions:

• Five key variables were identified as independently associated with postoperative pneumonia.
• The developed model shows strong performance and potential for external validation.
• This validated model can be utilized for risk stratification and management of patients susceptible to postoperative pneumonia.