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

BACKGROUNDPostoperative pneumonia is associated with increased morbidity, mortality and costs. Prediction models of pneumonia that are currently available are based on retrospectively collected data and administrative coding systems.OBJECTIVETo identify independent variables associated with the occurrence of postoperative pneumonia.DESIGNA prospective observational study of a multicentre cohort (Prospective Evaluation of a RIsk Score for postoperative pulmonary COmPlications in Europe database).SETTINGSixty-three hospitals in Europe.PATIENTSPatients undergoing surgery under general and/or regional anaesthesia during a 7-day recruitment period.MAIN OUTCOME MEASUREThe primary outcome was postoperative pneumonia. Definition: the need for treatment with antibiotics for a respiratory infection and at least one of the following criteria: new or changed sputum; new or changed lung opacities on a clinically indicated chest radiograph; temperature more than 38.3 °C; leucocyte count more than 12 000 μl-1.RESULTSPostoperative pneumonia occurred in 120 out of 5094 patients (2.4%). Eighty-two of the 120 (68.3%) patients with pneumonia required ICU admission, compared with 399 of the 4974 (8.0%) without pneumonia (P < 0.001). We identified five variables independently associated with postoperative pneumonia: functional status [odds ratio (OR) 2.28, 95% confidence interval (CI) 1.58 to 3.12], pre-operative SpO2 values while breathing room air (OR 0.83, 95% CI 0.78 to 0.84), intra-operative colloid administration (OR 2.97, 95% CI 1.94 to 3.99), intra-operative blood transfusion (OR 2.19, 95% CI 1.41 to 4.71) and surgical site (open upper abdominal surgery OR 3.98, 95% CI 2.19 to 7.59). The model had good discrimination (c-statistic 0.89) and calibration (Hosmer-Lemeshow P = 0.572).CONCLUSIONWe identified five variables independently associated with postoperative pneumonia. The model performed well and after external validation may be used for risk stratification and management of patients at risk of postoperative pneumonia.TRIAL REGISTRATIONNCT 01346709 (ClinicalTrials.gov).