Clinical risk factors for postoperative infection in adult cardiac surgery with cardiopulmonary bypass: a retrospective study.

BACKGROUND

Postoperative infection remains a serious problem for patients undergoing open-heart surgery and is associated with poor prognosis and mortality.

AIM

To determine the incidence, characteristics and associated risk factors for nosocomial infections in adult cardiac surgery patients and to develop a nomogram prediction model.

METHODS

Data were retrospectively collected from patients who underwent cardiac surgery with cardiopulmonary bypass (CPB) at a tertiary hospital in 2023. Patients were divided into an infected group ( = 130) and a non-infected group ( = 192). Multivariate logistic regression analysis was used to analyse the independent risk factors for healthcare-associated infections after cardiac surgery under CPB.

RESULTS

Of the 1584 patients, 130 (8.21%) developed postoperative infections (infection group). Lower respiratory tract was the most common site of infection ( = 74, 56.9%), while Gram-negative bacteria were the predominant isolates overall ( = 81, 62.3%). Among the Gram-negative bacteria, was the most frequently identified, whereas was the leading strain among Gram-positive bacteria. Multivariate logistic regression analysis of the 322 patients included in the study revealed that CPB duration, American Society of Anaesthesiologists score, procalcitonin concentration on the first postoperative day, monocyte:lymphocyte ratio, preinfection mechanical ventilation duration, and preinfection central venous catheterization duration were the six independent predictors of postoperative infection. The area under the receiver operating characteristic curve was 0.824 (0.778-0.870), and the model showed good predictive performance.

CONCLUSION

A nomogram has been developed to predict postoperative infection via commonly available data. This tool could assist clinicians in optimising the perioperative care of patients undergoing cardiac surgery with CPB, but further external validation is needed.