Prediction model for postoperative pneumonia in abdominal surgery: results of an observational multicenter study
R. V. Veyler,N. V. Trembach,T. S. Musaeva,M. A. Magomedov,A. S. Popov,V. V. Fisher,V. E. Khoronenko,A. I. Gritsan,P. V. Dunts,A. Zh. Bayalieva,A. M. Ovezov,K. M. Lebedinskii,D. V. Martynov,A. P. Spasova,V. V. Stadler,D. A. Levit,K. G. Shapovalov,V. N. Kokhno,Igor B. Zabolotskikh
DOI: https://doi.org/10.21320/1818-474x-2023-4-43-59
2023-10-30
Annals of critical care
Abstract:INTRODUCTION: Taking into account the prevalence of postoperative pneumonia and the increase in the number of surgical procedures, forecasting its development is an urgent task that allows taking measures to reduce the frequency of its occurrence by optimizing the perioperative period. Despite their value, the existing scales for predicting postoperative pneumonia do not provide domestic specialists with a reliable and consistent method by which to stratify the risk of developing postoperative pneumonia in our population. OBJECTIVE: To develop a model for predicting postoperative pneumonia based on the identification of risk factors for its development. MATERIALS AND METHODS: A multicenter prospective study of 6844 patients over 18 years of age undergoing elective abdominal surgery. 30-day mortality and postoperative pneumonia were assessed. In the first phase of the study, a comparison was made between the pneumonia and non-pneumonia group of baseline patient data, as well as factors associated with surgery and anesthesia. At the second stage of the study, a logistic regression analysis was performed to assess the contribution of factors to the development of postoperative pneumonia. At the third stage of the study, a model for predicting postoperative pneumonia was built according to the data of multivariate logistic regression analysis. At the final stage, the obtained model was compared with the forecasting models of other authors found in the world literature. RESULTS: Pneumonia was detected in 53 patients (0.77 %). A lethal outcome was observed in 39 patients: in patients with pneumonia in 15 cases (28.3 %), and without pneumonia in 24 cases (0.4 %). Retrospectively, taking into account the obtained model, 933 patients were assigned to the high-risk group for developing pneumonia, the incidence of pneumonia was 4.5 %. In the low-risk group for developing pneumonia — 5911 patients, the incidence of pneumonia was 0.19 %. CONCLUSIONS: Eight independent variables associated with postoperative pneumonia were identified: duration of surgery, smoking, complete functional dependence, perioperative anemia requiring iron supplementation, intraoperative use of vasopressors, American Society of Anesthesiologists classification 3 functional class, use of bronchodilators for chronic obstructive pulmonary disease, and high operative risk. The postoperative pneumonia prediction model has excellent predictive value (AUROC = 0.904).