Commentary: Incidence and independent risk factors for postoperative delirium in ICU patients after cardiopulmonary bypass cardiac surgery: a retrospective cohort study
Frontiers in Cardiovascular Medicine · 04.11.2026 · 13
DOI: 10.3389/fcvm.2026.1991451 ↗Kurz gefasst
Kohortenstudie in der Zeitschrift Frontiers in Cardiovascular Medicine, online erschienen am 04.11.2026. Verfasst von Antonio Salsano und 1 weiteren Autor:innen. Die Arbeit nennt 8 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Università degli Studi di Genova.
Themen: Intensivpflege, Delir
Abstract (Originalsprache)
Postoperative delirium (POD) remains a common complication after cardiac surgery. Zhang et al. investigated POD in a single-center retrospective cohort of 302 patients, reporting 85 cases (28.15%) (1). Their use of structured intensive care unit (ICU) nursing records offers a practical basis for exploring perioperative associations. Several considerations clarify how these findings can inform future research.Advanced age, longer aortic cross-clamping and mechanical ventilation, higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores, and lower albumin levels were independently associated with POD. Statistical independence denotes adjustment within the fitted model, not causality. APACHE II incorporates age and neurological status, sharing information with other severity measures. Residual confounding remains possible. Albumin may reflect inflammation, dilution, or nutritional vulnerability, while cross-clamp duration may indicate operative complexity.Temporal ordering is particularly important. Delirium may itself delay extubation; prolonged ventilation can therefore be a consequence as well as an antecedent of POD. The report does not establish whether ventilation duration preceded the first positive Intensive Care Delirium Screening Checklist (ICDSC) assessment. APACHE II scoring within 24-48 hours may overlap with delirium onset, and albumin sampling time is unclear. These variables may partly represent concurrent severity markers or consequences of the postoperative course rather than purely antecedent predictors. This potential reverse causality should be addressed by defining a prediction time and excluding information obtained afterward. The combined-model AUC of 0.876 is encouraging, but its interpretation requires reproducible internal validation. Section 2.4 describes univariate screening at P < 0.05 followed by forward stepwise selection. Figure 5A includes a bias-corrected curve, yet the resampling method, number of repetitions, and repetition of variable selection within resamples are not described. Consistent with TRIPOD reporting principles, these details and optimism-corrected discrimination would clarify potential overfitting (3).Calibration slope and intercept, with uncertainty estimates, would complement the calibration plot and reported mean absolute error of 0.076. The latter does not describe systematic overprediction or prediction extremity. Decision curves should relate thresholds to a specified intervention and its consequences. They estimate potential net benefit, not demonstrated outcome improvement. Planned external validation remains necessary (3).Operative duration measures do not fully characterize cerebral exposure. Under oxygen delivery-guided perfusion, Mukaida et al. found associations involving transfusion, intubation duration, and time with hematocrit below 25%; this did not establish that the perfusion strategy prevents POD (4).Their pooled delirium estimate favored cerebral oximetry-guided management, but the cardiac subgroup was inconclusive (RR 0.56; 95% CI 0.28-1.10) (5). Zhang et al.'s hypotension variable concerned the first postoperative 24 hours, not intraoperative exposure. Physiological measurements should therefore complement operative times in future studies.Trained ICU nurses performed daily ICDSC assessments, and investigators retrospectively extracted the scores. At least one score of 4 or higher defined POD. This structured approach is a strength and uses a validated instrument (6). Nevertheless, hypoactive episodes may escape detection, especially when reduced arousal is attributed to sedation. Their frequency and any under-detection cannot be quantified from the reported data.Assessment completion rates, sedation depth at scoring, and delirium subtype distribution are not reported. Non-adherence is therefore unverified, rather than demonstrated. Future reports should state scheduled versus completed assessments, missed assessments and reasons, time-stamped arousal scores such as the Richmond Agitation-Sedation Scale, sedative exposure, and hypoactive, hyperactive, or mixed subtypes. Unassessable states should be distinguished from negative screens. Arousal can affect delirium assessment (7).Multicomponent ICU prevention strategies involving analgesia and sedation review, mobilization, sleep preservation, and family engagement have supportive, although variable, evidence (8). Zhang et al. did not test such an intervention. Their associations generate potential targets; they do not establish that albumin supplementation or earlier extubation prevents POD.Research priorities should be sequential. First, establish time-stamped screening and a prediction time preceding POD. Second, prioritize baseline cognition, validated frailty assessment, functional dependence, prior cerebrovascular disease, and standardized nutritional screening. Depressive symptoms are a useful additional candidate. During bypass, prioritize hematocrit, transfusion exposure, indexed oxygen delivery, and the depth and duration of hypotension and cerebral desaturation. These are proposed collection priorities, not predictors established by this cohort. Validate prespecified models with resampling and independent multicenter data before testing model-guided prevention, including safety and delirium duration and severity.Zhang et al. provide useful exploratory evidence and demonstrate the research value of structured nursing records. Clearer temporal ordering, reproducible validation, and richer characterization of vulnerability would build on this contribution and identify associations that offer reproducible predictive value and inform individualized prevention.
Förderung
- Università degli Studi di Genova
Zitieren
APA
Salsano, A., & Santini, F. (2026). Commentary: Incidence and independent risk factors for postoperative delirium in ICU patients after cardiopulmonary bypass cardiac surgery: a retrospective cohort study. Frontiers in Cardiovascular Medicine, 13. https://doi.org/10.3389/fcvm.2026.1991451
Vancouver
Salsano A, Santini F. Commentary: Incidence and independent risk factors for postoperative delirium in ICU patients after cardiopulmonary bypass cardiac surgery: a retrospective cohort study. Frontiers in Cardiovascular Medicine. 2026;13. doi:10.3389/fcvm.2026.1991451
BibTeX
@article{Salsano2026,
author = {Salsano, Antonio and Santini, Francesco},
title = {Commentary: Incidence and independent risk factors for postoperative delirium in ICU patients after cardiopulmonary bypass cardiac surgery: a retrospective cohort study},
journal = {Frontiers in Cardiovascular Medicine},
year = {2026},
volume = {13},
doi = {10.3389/fcvm.2026.1991451}
}
EUVIMED