Aims and objectivesTo analyse postoperative delirium (POD) in clinical practice after cardiac surgery, how it is detected and documented and if the use of a screening scale improves the detection rate.BackgroundCardiac surgery is considered a routine procedure with few complications. However, POD remains a concern, although often being overlooked in clinical practice.DesignRetrospective observational analysis.MethodsPatients 70 years and older with POD (n = 78) undergoing cardiac surgery were included in the study. Discharge summaries of both nurses and physicians were reviewed together with the clinical database for information about POD, to be compared with symptom screening using the Nursing Delirium Screening Scale (Nu‐DESC). A quantitative content analysis was used for the review of discharge summaries, with a coding scheme adopted from the Nu‐DESC method. The STROBE checklist was followed.ResultsIn discharge summaries, 41 of the 78 POD patients were correctly recognised, and 22 of these were identified in the clinical database. Screening by the Nu‐DESC identified delirium at a measurably higher rate, 56/78 patients. The review of discharge summaries showed that patients expressing "inappropriate behaviour" was the most easily identified sign for POD for both nurses and physicians.ConclusionsHealthcare professionals underdiagnose delirium after cardiac surgery, with a low detection rate described in both discharge summaries and in the clinical database. Recognition of delirium improved when Nu‐DESC was used for systematic screening.Relevance to clinical practiceThis study emphasises the need for better screening for the detection of delirium in daily clinical practice.