Postoperative delirium (POD) is common, costly and associated with long-term morbidity and increased mortality. We conducted a cohort study to assess the contribution of cardiopulmonary bypass (CPB) to the development of POD by means of algorithm-based data processing.

A database was compiled from 3 datasets of patients who underwent cardiac surgery between 2014 and 2019: intensive care unit discharge files, CPB protocols and medical quality management records. Following data extraction and structuring using novel algorithms, missing data were imputed. Ten independent imputations were analysed by multiple logistic regression with stepwise deletion of factors to arrive at a minimal adequate model.

POD was diagnosed in 456/3163 patients (14.4%). In addition to known demographic risk factors and comorbidities like male sex, age, carotid disease, acute kidney failure and diabetes mellitus, cardiopulmonary parameters like total blood volume at the CPB [adjusted odds ratio (AOR) 1.001; confidence interval (CI) 1.1001–1.002] were independent predictors of POD. Higher values of the minimal blood flow were associated with a lower risk of POD (AOR 0.993; CI 0.988–0.997). Flow rates at least 30% above target did emerge in the minimal adequate model as a potential risk factor, but the confidence interval suggested a lack of statistical significance (AOR 1.819; 95% CI: 0.955–3.463).

CPB data processing proved to be a useful tool for obtaining compact information to better identify the roles of individual operational states. Strict adherence to perfusion limits along with tighter control of blood flow and acid–base balance during CPB may help to further decrease the risk of POD.