Cardiovascular and thoracic surgical patients frequently require postoperative ICU admission for mechanical ventilation, haemodynamic monitoring, vasoactive support, and management of procedure-related complications. Most patients experience an uncomplicated recovery and can be transferred to a step-down unit within the first few postoperative days. However, a clinically important subgroup requires prolonged intensive care, consuming a disproportionate share of critical-care resources and experiencing substantially greater morbidity and mortality [1–3]. The reported incidence of prolonged ICU stay following cardiac surgery varies according to the patient population, surgical complexity, institutional practices, and definition employed. Previous investigations have used thresholds ranging from more than 48 hours to seven or more days [1–5]. Although no universally accepted definition exists, an ICU stay of at least five days represents a clinically meaningful deviation from routine postoperative recovery. This threshold identifies patients requiring sustained organ support or treatment of major complications while remaining sufficiently frequent for meaningful statistical evaluation. Advances in surgical techniques, myocardial protection, cardiopulmonary bypass, anaesthesia, and intensive care have expanded the eligibility of older and medically complex patients for cardiovascular surgery. Consequently, contemporary surgical populations have a greater prevalence of diabetes mellitus, chronic kidney disease, pulmonary disease, impaired ventricular function, anaemia, and frailty. These conditions may reduce physiological reserve and increase susceptibility to perioperative organ dysfunction [4–7]. Several preoperative characteristics have been associated with prolonged postoperative intensive care. Advanced age, female sex, chronic kidney disease, chronic pulmonary disease, poor left ventricular function, preoperative anaemia, previous cardiac surgery, and emergency operative status have been identified in observational studies and predictive models [1,3,5,8]. Risk-scoring systems such as the European System for Cardiac Operative Risk Evaluation were primarily developed to estimate operative mortality, and their ability to predict prolonged ICU stay has varied across populations [8,9]. Locally derived assessments may therefore provide clinically useful information not captured by mortality-based scores. Intraoperative events also influence the postoperative course. Extended cardiopulmonary bypass and aortic cross-clamp durations are associated with inflammatory activation, myocardial injury, coagulopathy, acute kidney injury, and delayed recovery [4,6,10]. Complex or combined procedures, emergency operations, excessive blood loss, transfusion, haemodynamic instability, and mechanical circulatory support may further increase the need for prolonged critical care. In thoracic surgical patients, extensive pulmonary resection, limited pulmonary reserve, major bleeding, and postoperative respiratory dysfunction may have similar consequences. Postoperative complications frequently form the direct pathway to extended ICU treatment. Prolonged mechanical ventilation, reintubation, low cardiac output syndrome, acute kidney injury, arrhythmia, neurological injury, infection, and re-exploration may delay discharge and worsen survival [5–7]. Differentiating true perioperative predictors from complications arising during an extended admission is important. Variables available before or during surgery are useful for early prediction, whereas postoperative complications explain why intensive care was prolonged. Prolonged ICU stay has consequences beyond the individual patient. It may reduce bed availability, delay scheduled operations, increase staffing requirements, and impose considerable financial costs. Accurate identification of high-risk patients may allow preoperative correction of anaemia and nutritional deficits, renal-protective strategies, improved glycaemic and pulmonary management, blood-conservation measures, and realistic planning of postoperative resources. Published evidence has predominantly originated from high-income countries and has frequently focused on isolated coronary artery bypass grafting or elective cardiac surgery. The findings may not be fully generalisable to mixed cardiovascular and thoracic surgical populations in Indian tertiary-care institutions. Differences in disease severity, timing of referral, comorbidity, procedure type, blood-product availability, and discharge practices may influence both the occurrence and determinants of prolonged ICU stay. The present study evaluated adults admitted to a mixed cardiovascular and thoracic surgical ICU over 16 months. It aimed to determine the frequency of prolonged ICU stay, identify independent perioperative predictors, and examine its relationship with postoperative complications and in-hospital outcomes.

Study design and setting A retrospective observational cohort study was conducted in the Department of Cardiovascular and Thoracic Surgery at KMCT Medical College, Manassery, Mukkam, Kerala, India. Medical records of patients admitted to the cardiovascular and thoracic surgical ICU between 1 September 2022 and 31 December 2023 were reviewed. The study was structured and reported according to the Strengthening the Reporting of Observational Studies in Epidemiology recommendations. Ethical clearance was obtained. Study population All consecutive adult patients admitted to the CVTS ICU following a cardiovascular or thoracic surgical procedure during the study period were screened. The cohort included patients undergoing coronary artery bypass grafting, valve surgery, combined coronary and valve surgery, aortic procedures, adult congenital cardiac surgery, major thoracic surgery, and other cardiovascular procedures requiring postoperative intensive care. Patients were included when they were aged 18 years or older, underwent an eligible cardiovascular or thoracic operation, and had adequate documentation of ICU admission and discharge. Patients younger than 18 years, those admitted only for observation without a surgical procedure, patients transferred from another institution after receiving substantial postoperative intensive care, and patients with insufficient records to establish ICU length of stay were excluded. In patients with more than one eligible operation, only the first qualifying admission was included. Sampling and sample size Consecutive sampling was employed. All eligible admissions during the 16-month study period were included. A total of 301 records were screened. Fifteen records were excluded because of incomplete ICU timing or substantial missing perioperative information, leaving 286 patients for analysis. With 58 prolonged-stay events, the number of variables included in the final multivariable model was restricted to avoid overfitting. Variable selection was based on clinical relevance, previous evidence, univariable associations, and availability of complete data. Outcome definition The primary outcome was prolonged ICU stay, defined as a cumulative stay of five days or longer during the index hospitalisation. ICU length of stay was calculated from the recorded time of postoperative ICU admission to final discharge from intensive care. Patients were classified into a non-prolonged-stay group with an ICU duration of less than five days and a prolonged-stay group with an ICU duration of at least five days. When ICU readmission occurred during the same hospitalisation, the durations of the initial and subsequent ICU episodes were combined. ICU readmission was also recorded separately as a postoperative outcome. Data collection Data were obtained from patient case records, preoperative assessments, anaesthesia charts, operative notes, perfusion records, ICU flow sheets, laboratory reports, blood-bank records, discharge summaries, and electronic hospital records. Information was entered into a structured data-collection sheet after removal of direct patient identifiers. Demographic and preoperative variables included age, sex, body mass index, smoking, diabetes mellitus, hypertension, chronic kidney disease, chronic pulmonary disease, previous cerebrovascular disease, previous cardiac or thoracic surgery, New York Heart Association functional class, left ventricular ejection fraction, preoperative haemoglobin, serum creatinine, serum albumin, operative diagnosis, procedure type, and operative urgency. Preoperative anaemia was evaluated using the last haemoglobin value recorded before surgery. For predictor analysis, clinically important anaemia was defined as haemoglobin below 10 g/dL. Impaired left ventricular systolic function was defined as an ejection fraction below 40%. Chronic kidney disease was identified from the documented diagnosis or a sustained estimated glomerular filtration rate below 60 mL/min/1.73 m². Intraoperative variables included surgical duration, cardiopulmonary bypass duration, aortic cross-clamp duration, estimated blood loss, packed red-cell transfusion, administration of other blood components, difficulty separating from cardiopulmonary bypass, requirement for multiple vasoactive drugs, intraoperative arrhythmia, and use of an intra-aortic balloon pump or other mechanical circulatory support. Cardiopulmonary bypass and cross-clamp durations were treated as not applicable for procedures undertaken without cardiopulmonary bypass. Postoperative variables included mechanical ventilation duration, ventilation exceeding 48 hours, reintubation, low cardiac output syndrome, arrhythmia, acute kidney injury, renal-replacement therapy, neurological complications, pneumonia, bloodstream or surgical-site infection, sepsis, re-exploration, ICU readmission, total hospital stay, and in-hospital mortality. Acute kidney injury was classified from postoperative serum creatinine measurements according to the Kidney Disease: Improving Global Outcomes criteria where sufficient measurements were available. Statistical analysis Data were analysed using a standard statistical software package. Continuous variables were assessed for normality using histograms, Q–Q plots, and the Shapiro–Wilk test. Normally distributed variables were expressed as mean and standard deviation and compared using the independent-samples t test. Variables with skewed distributions were presented as median and interquartile range and compared using the Mann–Whitney U test. Categorical variables were expressed as frequencies and percentages. Associations were evaluated using the chi-square test or Fisher’s exact test, as appropriate. Effect estimates for candidate predictors were initially calculated using univariable binary logistic regression. Variables with clinical relevance or a univariable P value below 0.10 were considered for multivariable logistic regression. Postoperative events that could have occurred as consequences of an extended ICU admission were excluded from the principal predictive model. Multicollinearity was assessed using correlation matrices and variance inflation factors. Adjusted odds ratios were reported with 95% confidence intervals. Model discrimination was assessed using the area under the receiver operating characteristic curve. Calibration was examined using the Hosmer–Lemeshow goodness-of-fit test. All tests were two-sided, and P<0.05 was considered statistically significant.

Patient characteristics During the study period, 301 patient records were screened. Fifteen records were excluded because ICU admission or discharge timing was unavailable or essential perioperative information was incomplete. The final analysis included 286 patients. Of these, 228 patients (79.7%) remained in the ICU for less than five days, while 58 (20.3%) had an ICU stay of five days or longer. The mean age of the cohort was 60.2±11.9 years, and 194 patients (67.8%) were male. Patients with prolonged ICU stay were older than those without prolonged stay (65.1±10.6 versus 58.9±11.8 years; P<0.001). Chronic kidney disease, diabetes mellitus, hypertension, preoperative haemoglobin below 10 g/dL, left ventricular ejection fraction below 40%, advanced New York Heart Association functional class, and emergency surgery were more frequent in the prolonged-stay group (Table 1). Table 1. Preoperative characteristics according to ICU length of stay Variable ICU stay <5 days (n=228) ICU stay ≥5 days (n=58) P value Age, years, mean±SD 58.9±11.8 65.1±10.6 <0.001 Age ≥65 years 73 (32.0) 33 (56.9) <0.001 Male sex 158 (69.3) 36 (62.1) 0.293 BMI, kg/m², mean±SD 25.4±3.8 25.8±4.1 0.479 Current or former smoker 76 (33.3) 23 (39.7) 0.365 Diabetes mellitus 92 (40.4) 32 (55.2) 0.043 Hypertension 136 (59.6) 43 (74.1) 0.042 Chronic kidney disease 21 (9.2) 17 (29.3) <0.001 Chronic pulmonary disease 16 (7.0) 9 (15.5) 0.044 Previous stroke/TIA 10 (4.4) 6 (10.3) 0.098 Previous cardiac or thoracic surgery 11 (4.8) 7 (12.1) 0.048 NYHA class III or IV 39 (17.1) 24 (41.4) <0.001 LVEF <40% 29 (12.7) 22 (37.9) <0.001 Haemoglobin, g/dL, mean±SD 12.5±1.6 11.3±1.8 <0.001 Haemoglobin <10 g/dL 24 (10.5) 20 (34.5) <0.001 Serum creatinine, mg/dL, median (IQR) 0.9 (0.8–1.1) 1.2 (0.9–1.7) <0.001 Serum albumin, g/dL, mean±SD 3.8±0.5 3.5±0.6 <0.001 Emergency surgery 12 (5.3) 13 (22.4) <0.001 Values are presented as number (percentage) unless otherwise stated. BMI: body mass index; ICU: intensive care unit; IQR: interquartile range; LVEF: left ventricular ejection fraction; NYHA: New York Heart Association; SD: standard deviation; TIA: transient ischaemic attack. Operative and intraoperative characteristics Coronary artery bypass grafting was the most frequently performed procedure, accounting for 149 operations (52.1%). Valve surgery was performed in 72 patients (25.2%), combined coronary and valve surgery in 23 (8.0%), aortic surgery in 12 (4.2%), and major thoracic surgery in 30 (10.5%). Complex cardiac procedures, particularly combined and aortic operations, were relatively more frequent in the prolonged-stay group. Median surgical duration, cardiopulmonary bypass duration, and aortic cross-clamp duration were significantly greater among patients with prolonged ICU stay. These patients also had greater intraoperative blood loss, received more packed red-cell transfusions, and more frequently required intra-aortic balloon pump support (Table 2). Table 2. Operative and intraoperative characteristics Variable ICU stay <5 days (n=228) ICU stay ≥5 days (n=58) P value Type of procedure 0.003 Isolated CABG 124 (54.4) 25 (43.1) Isolated valve surgery 58 (25.4) 14 (24.1) Combined CABG and valve surgery 14 (6.1) 9 (15.5) Aortic surgery 5 (2.2) 7 (12.1) Major thoracic surgery 27 (11.8) 3 (5.2) Redo surgery 11 (4.8) 7 (12.1) 0.048 Surgery duration, minutes, median (IQR) 250 (210–300) 330 (275–405) <0.001 CPB duration, minutes, median (IQR)† 92 (72–116) 139 (105–179) <0.001 CPB duration >120 minutes† 39 (19.4) 31 (56.4) <0.001 Cross-clamp duration, minutes, median (IQR)† 58 (44–76) 91 (66–122) <0.001 Estimated blood loss, mL, median (IQR) 550 (400–750) 900 (600–1300) <0.001 Packed red-cell transfusion ≥3 units 24 (10.5) 28 (48.3) <0.001 Multiple vasoactive drugs 31 (13.6) 29 (50.0) <0.001 Difficulty separating from CPB† 10 (5.0) 14 (25.5) <0.001 Intra-aortic balloon pump use 8 (3.5) 12 (20.7) <0.001 Values are presented as number (percentage) unless otherwise stated. †Calculated among 256 patients who underwent procedures involving cardiopulmonary bypass: 201 in the ICU-stay <5 days group and 55 in the ICU-stay ≥5 days group. CABG: coronary artery bypass grafting; CPB: cardiopulmonary bypass; ICU: intensive care unit; IQR: interquartile range. Postoperative complications and outcomes Patients with prolonged ICU stay experienced substantially greater postoperative morbidity. Their median duration of mechanical ventilation was 61 hours compared with nine hours in patients with shorter ICU stays. Mechanical ventilation exceeding 48 hours occurred in 55.2% of patients with prolonged ICU stay but in only 2.6% of patients without prolonged stay. Low cardiac output syndrome, postoperative arrhythmia, acute kidney injury, renal-replacement therapy, reintubation, infection, re-exploration, and ICU readmission were significantly more frequent among patients with prolonged ICU stay. The median hospital stay was 21 days in the prolonged-stay group and nine days in the non-prolonged group. In-hospital mortality was 10.3% and 0.9%, respectively (P<0.001) (Table 3). Table 3. Postoperative complications and clinical outcomes Outcome ICU stay <5 days (n=228) ICU stay ≥5 days (n=58) P value Mechanical ventilation, hours, median (IQR) 9 (7–14) 61 (31–118) <0.001 Mechanical ventilation >48 hours 6 (2.6) 32 (55.2) <0.001 Reintubation 3 (1.3) 10 (17.2) <0.001 Low cardiac output syndrome 12 (5.3) 24 (41.4) <0.001 Postoperative arrhythmia 29 (12.7) 22 (37.9) <0.001 Acute kidney injury 19 (8.3) 27 (46.6) <0.001 Renal-replacement therapy 2 (0.9) 12 (20.7) <0.001 Neurological complication 3 (1.3) 7 (12.1) <0.001 Pneumonia or other major infection 7 (3.1) 18 (31.0) <0.001 Re-exploration 5 (2.2) 12 (20.7) <0.001 ICU readmission 4 (1.8) 8 (13.8) <0.001 Hospital stay, days, median (IQR) 9 (7–12) 21 (15–31) <0.001 In-hospital mortality 2 (0.9) 6 (10.3) <0.001 Values are presented as number (percentage) unless otherwise stated. ICU: intensive care unit; IQR: interquartile range. Predictors of prolonged ICU stay Univariable analysis identified older age, diabetes mellitus, chronic kidney disease, chronic pulmonary disease, advanced NYHA class, preoperative anaemia, left ventricular ejection fraction below 40%, emergency surgery, complex surgery, cardiopulmonary bypass duration exceeding 120 minutes, and transfusion of at least three packed red-cell units as potential predictors. After adjustment, emergency surgery, chronic kidney disease, preoperative haemoglobin below 10 g/dL, left ventricular ejection fraction below 40%, cardiopulmonary bypass duration exceeding 120 minutes, and transfusion of at least three units of packed red cells remained independently associated with prolonged ICU stay. Age showed a smaller independent association, with a 3% increase in the odds of prolonged stay for every additional year of age (Table 4). The final model demonstrated acceptable discrimination, with an area under the receiver operating characteristic curve of 0.86 (95% CI 0.81–0.91). The Hosmer–Lemeshow test indicated satisfactory calibration (P=0.61). Table 4. Logistic regression analysis of predictors of prolonged ICU stay Predictor Unadjusted OR (95% CI) P value Adjusted OR (95% CI) P value Age, per one-year increase 1.05 (1.02–1.08) <0.001 1.03 (1.00–1.06) 0.049 Diabetes mellitus 1.82 (1.02–3.25) 0.044 1.31 (0.66–2.60) 0.438 Chronic kidney disease 4.09 (1.98–8.45) <0.001 2.48 (1.13–5.44) 0.024 Chronic pulmonary disease 2.43 (1.00–5.91) 0.050 1.69 (0.61–4.67) 0.311 NYHA class III or IV 3.42 (1.84–6.34) <0.001 1.64 (0.76–3.53) 0.207 Haemoglobin <10 g/dL 4.47 (2.25–8.88) <0.001 2.39 (1.15–4.98) 0.020 LVEF <40% 4.19 (2.18–8.06) <0.001 2.57 (1.23–5.39) 0.012 Emergency surgery 5.20 (2.22–12.19) <0.001 3.06 (1.37–6.84) 0.006 Combined or aortic procedure 3.79 (1.82–7.89) <0.001 1.88 (0.78–4.54) 0.159 CPB duration >120 minutes 5.37 (2.85–10.13) <0.001 3.18 (1.55–6.52) 0.002 Packed red-cell transfusion ≥3 units 7.93 (4.13–15.21) <0.001 2.74 (1.31–5.75) 0.008 CI: confidence interval; CPB: cardiopulmonary bypass; LVEF: left ventricular ejection fraction; NYHA: New York Heart Association; OR: odds ratio.

Approximately one-fifth of the study population required an ICU stay of five days or longer. Prolonged stay was associated with greater preoperative comorbidity, more complex operations, longer cardiopulmonary bypass exposure, increased blood-product utilisation, and a markedly higher burden of postoperative complications. Chronic kidney disease, clinically important anaemia, impaired left ventricular function, emergency surgery, prolonged cardiopulmonary bypass, and transfusion of at least three packed red-cell units were independently associated with prolonged ICU stay. The proportion of prolonged-stay patients observed in this cohort was within the broad range reported in previous cardiac-surgery populations [1–5]. Direct comparison between studies requires caution because definitions have varied considerably. Some investigators defined prolonged stay as more than 48 or 72 hours, whereas others used thresholds of five, seven, or ten days. A lower threshold captures patients with moderately delayed recovery, while a threshold of five days identifies a smaller group with more substantial organ-support requirements and complications. Increasing age was independently associated with prolonged stay, although the magnitude of the association was smaller after adjustment for comorbidities and operative factors. Older patients frequently have reduced physiological reserve, frailty, renal impairment, pulmonary dysfunction, and a greater burden of vascular disease. Age may therefore function partly as a marker of accumulated vulnerability rather than as an isolated causal determinant. Risk assessment based solely on chronological age may be less informative than assessment combining age with organ function, frailty, and operative complexity. Chronic kidney disease was an important independent predictor. Patients with impaired renal function are more susceptible to perioperative haemodynamic disturbances, nephrotoxic exposure, fluid overload, electrolyte abnormalities, and acute kidney injury. Cardiopulmonary bypass may further compromise renal perfusion through non-pulsatile flow, haemodilution, inflammation, haemolysis, and microembolisation [11,12]. Even modest preoperative renal dysfunction has been associated with postoperative renal-replacement therapy, mortality, and extended hospitalisation. Identification of renal impairment before surgery permits attention to volume status, medication adjustment, perfusion pressure, avoidance of nephrotoxins, and early postoperative monitoring. Preoperative haemoglobin below 10 g/dL independently predicted prolonged ICU stay. Anaemia reduces oxygen-carrying capacity and may increase vulnerability to myocardial and renal hypoxia during cardiopulmonary bypass. It is also associated with a greater likelihood of allogeneic transfusion. Previous studies have linked preoperative anaemia with acute kidney injury, infection, prolonged ventilation, and mortality after cardiac surgery [13,14]. Where surgery is elective, investigation and treatment of iron deficiency or other reversible causes may represent a practical risk-reduction strategy. Left ventricular ejection fraction below 40% was another significant predictor. Patients with impaired ventricular function have limited cardiac reserve and may require extended vasoactive support, mechanical circulatory assistance, and longer ventilation. They are also at increased risk of low cardiac output syndrome and secondary renal, hepatic, or neurological dysfunction. Preoperative assessment of ventricular performance, optimisation of heart-failure therapy, and preparation for postoperative haemodynamic support are therefore central to perioperative planning. Emergency surgery increased the adjusted odds of prolonged ICU stay approximately threefold. Emergency procedures allow limited time for correction of anaemia, renal dysfunction, infection, pulmonary impairment, or metabolic abnormalities. These patients may present with haemodynamic instability, evolving myocardial injury, acute aortic disease, or other life-threatening conditions requiring immediate intervention. The association persisted after adjustment, suggesting that operative urgency captured risk beyond the measured baseline characteristics. Cardiopulmonary bypass lasting more than 120 minutes was among the strongest independent predictors. Prolonged bypass reflects operative complexity and exposes patients to a greater inflammatory, haemolytic, and coagulation-related burden. Extended bypass and cross-clamp durations have consistently been associated with acute kidney injury, low cardiac output, pulmonary dysfunction, bleeding, neurological complications, and mortality [4,6,10]. Although operative complexity cannot always be reduced, careful procedural planning, efficient surgical coordination, myocardial protection, and perfusion strategies may limit avoidable bypass exposure. Transfusion of at least three packed red-cell units remained associated with prolonged ICU stay after adjustment. Transfusion may indicate severe bleeding, haemodilution, complex surgery, or preoperative anaemia. It may also contribute to pulmonary complications, circulatory overload, inflammatory responses, infection, and organ dysfunction [15,16]. Because observational analyses cannot fully separate the effect of transfusion from the severity of bleeding and illness, this relationship should not be interpreted as proof of direct causation. Nevertheless, the finding supports patient blood-management measures, including correction of preoperative anaemia, meticulous haemostasis, antifibrinolytic therapy where appropriate, cell salvage, and evidence-based transfusion thresholds. Postoperative complications were concentrated in the prolonged-stay group. More than half required mechanical ventilation beyond 48 hours, and substantial proportions developed low cardiac output syndrome, acute kidney injury, infection, or required re-exploration. These events were not included in the main predictive model because they could occur after ICU stay had already begun and may be intermediates on the pathway between perioperative risk and prolonged admission. Their distribution nevertheless illustrates the clinical mechanisms through which ICU treatment becomes extended. In-hospital mortality was more than tenfold higher among patients with prolonged ICU stay. Similar associations have been reported in earlier studies, with prolonged intensive care serving as a marker of severe complications, multiorgan dysfunction, and reduced physiological reserve [1,2,17]. The longer median hospital stay observed in this group further demonstrates the broader resource implications. Strategies that reduce preventable complications may improve outcomes while increasing the availability of critical-care beds. The study incorporated preoperative, intraoperative, and postoperative information and included a mixed CVTS population reflective of routine tertiary-care practice. The use of a clinically meaningful five-day threshold provided a clear outcome for risk stratification. The multivariable analysis distinguished variables available for early prediction from postoperative complications that may have resulted from an already complicated course. Several limitations should be acknowledged. The retrospective design depended on the accuracy and completeness of clinical documentation. Unmeasured factors, including frailty, nutritional status, detailed perfusion parameters, clinician-specific discharge decisions, and bed availability, may have influenced ICU duration. The inclusion of different cardiovascular and thoracic procedures introduced clinical heterogeneity. Some laboratory and physiological variables were unavailable for all patients. The number of prolonged-stay events limited the complexity of the regression model, and residual confounding could not be excluded. As a single-centre study, the findings may not be generalisable to institutions with different patient populations or ICU-discharge practices. External validation would be required before using the model as a formal prediction tool.

Prolonged ICU stay occurred in approximately one-fifth of adults admitted after cardiovascular and thoracic surgery and was associated with markedly higher postoperative morbidity, hospital stay, and in-hospital mortality. Chronic kidney disease, preoperative haemoglobin below 10 g/dL, left ventricular ejection fraction below 40%, emergency surgery, cardiopulmonary bypass exceeding 120 minutes, and transfusion of at least three packed red-cell units independently predicted an ICU stay of five days or longer. Preoperative optimisation of anaemia and renal risk, appropriate planning for patients with ventricular dysfunction, efficient operative conduct, and patient blood-management strategies may reduce avoidable postoperative resource utilisation. Prospective multicentre studies are required to validate these findings and develop broadly applicable risk-prediction models. Strengths and Limitations This study evaluated a broad range of preoperative, intraoperative, and postoperative factors in an unselected cardiovascular and thoracic surgical ICU population. Consecutive inclusion reduced the likelihood of selection bias, while the use of a five-day threshold identified patients with clinically meaningful resource utilisation. The multivariable model was restricted to variables available before or during surgery, improving its relevance for early risk assessment. The study had several limitations. Its retrospective design depended on the completeness and accuracy of medical records. Potentially important variables, including frailty, nutritional status, socioeconomic factors, detailed cardiopulmonary bypass parameters, and reasons for delayed ICU discharge, could not be evaluated consistently. The inclusion of different cardiovascular and thoracic procedures introduced clinical heterogeneity. The number of prolonged-stay events limited the number of variables included in the regression model. Residual confounding could not be excluded, and the findings from a single centre may not be directly generalisable to other institutions. External validation is required before the model can be adopted as a formal clinical prediction tool.

1. Hein OV, Birnbaum J, Wernecke K, England M, Konertz W, Spies C. Prolonged intensive care unit stay in cardiac surgery: risk factors and long-term survival. Ann Thorac Surg. 2006;81(3):880–885. doi:10.1016/j.athoracsur.2005.09.077. 2. Mahesh B, Choong CK, Goldsmith K, Gerrard C, Nashef SA, Vuylsteke A. Prolonged stay in intensive care unit is a powerful predictor of adverse outcomes after cardiac operations. Ann Thorac Surg. 2012;94(1):109–116. doi:10.1016/j.athoracsur.2012.02.010. 3. Kapadohos T, Angelopoulos E, Vasileiadis I, Nanas S, Kotanidou A, Karabinis A, et al. Determinants of prolonged intensive care unit stay in patients after cardiac surgery: a prospective observational study. J Thorac Dis. 2017;9(1):70–79. doi:10.21037/jtd.2017.01.18. 4. Almashrafi A, Elmontsri M, Aylin P. Systematic review of factors influencing length of stay in ICU after adult cardiac surgery. BMC Health Serv Res. 2016;16:318. doi:10.1186/s12913-016-1591-3. 5. Almashrafi A, Alsabti H, Mukaddirov M, Balan B, Aylin P. Factors associated with prolonged length of stay following cardiac surgery in a major referral hospital in Oman: a retrospective observational study. BMJ Open. 2016;6(6):e010764. doi:10.1136/bmjopen-2015-010764. 6. Zhang X, Zhang W, Lou H, Luo C, Du Q, Meng Y, et al. Risk factors for prolonged intensive care unit stays in patients after cardiac surgery with cardiopulmonary bypass: a retrospective observational study. Int J Nurs Sci. 2021;8(4):388–393. doi:10.1016/j.ijnss.2021.09.002. 7. Shah V, Ahuja A, Kumar A, Anstey C, Thang C, Guo L, et al. Outcomes of prolonged ICU stay for patients undergoing cardiac surgery in Australia and New Zealand. J Cardiothorac Vasc Anesth. 2022;36(12):4313–4319. doi:10.1053/j.jvca.2022.08.026. 8. Ettema RGA, Peelen LM, Schuurmans MJ, Nierich AP, Kalkman CJ, Moons KGM. Prediction models for prolonged intensive care unit stay after cardiac surgery: systematic review and validation study. Circulation. 2010;122(7):682–689. doi:10.1161/CIRCULATIONAHA.109.926808. 9. Nashef SAM, Roques F, Sharples LD, Nilsson J, Smith C, Goldstone AR, et al. EuroSCORE II. Eur J Cardiothorac Surg. 2012;41(4):734–745. doi:10.1093/ejcts/ezs043. 10. Tunç M, Şahutoğlu C, Karaca N, Kocabaş S, Aşkar FZ. Risk factors for prolonged intensive care unit stay after open heart surgery in adults. Turk J Anaesthesiol Reanim. 2018;46(4):283–291. doi:10.5152/TJAR.2018.92244. 11. Lannemyr L, Bragadottir G, Krumbholz V, Redfors B, Sellgren J, Ricksten SE. Effects of cardiopulmonary bypass on renal perfusion, filtration, and oxygenation in patients undergoing cardiac surgery. Anesthesiology. 2017;126(2):205–213. doi:10.1097/ALN.0000000000001461. 12. Vives M, Hernandez A, Parramon F, Estanyol N, Pardina B, Muñoz A, et al. Acute kidney injury after cardiac surgery: prevalence, impact and management challenges. Int J Nephrol Renovasc Dis. 2019;12:153–166. doi:10.2147/IJNRD.S167477. 13. Kulier A, Levin J, Moser R, Rumpold-Seitlinger G, Tudor IC, Snyder-Ramos SA, et al. Impact of preoperative anemia on outcome in patients undergoing coronary artery bypass graft surgery. Circulation. 2007;116(5):471–479. doi:10.1161/CIRCULATIONAHA.106.653501. 14. Williams ML, He X, Rankin JS, Slaughter MS, Gammie JS. Preoperative hematocrit is a powerful predictor of adverse outcomes in coronary artery bypass graft surgery: a report from the Society of Thoracic Surgeons Adult Cardiac Surgery Database. Ann Thorac Surg. 2013;96(5):1628–1634. doi:10.1016/j.athoracsur.2013.06.030. 15. Murphy GJ, Reeves BC, Rogers CA, Rizvi SIA, Culliford L, Angelini GD. Increased mortality, postoperative morbidity, and cost after red blood-cell transfusion in patients having cardiac surgery. Circulation. 2007;116(22):2544–2552. doi:10.1161/CIRCULATIONAHA.107.698977. 16. Tibi P, McClure RS, Huang J, Baker RA, Fitzgerald D, Mazer CD, et al. STS/SCA/AmSECT/SABM update to the clinical practice guidelines on patient blood management. Ann Thorac Surg. 2021;112(3):981–1004. doi:10.1016/j.athoracsur.2021.03.033. 17. Mazer CD, Whitlock RP, Fergusson DA, Hall J, Belley-Cote E, Connolly K, et al. Restrictive or liberal red-cell transfusion for cardiac surgery. N Engl J Med. 2017;377(22):2133–2144. doi:10.1056/NEJMoa1711818. 18. Brown CH IV, Probert J, Healy R, Parish M, Nomura Y, Yamaguchi A, et al. Cognitive decline after delirium in patients undergoing cardiac surgery. Anesthesiology. 2018;129(3):406–416. doi:10.1097/ALN.0000000000002253. 19. Hu BY, Laine GA, Wang S, Solis RT. Combined central venous oxygen saturation and lactate as markers of occult hypoperfusion and outcome following cardiac surgery. J Cardiothorac Vasc Anesth. 2012;26(1):52–57. doi:10.1053/j.jvca.2011.07.021. 20. Mehta RH, Grab JD, O’Brien SM, Bridges CR, Gammie JS, Haan CK, et al. Bedside tool for predicting the risk of postoperative dialysis in patients undergoing cardiac surgery. Circulation. 2006;114(21):2208–2216. doi:10.1161/CIRCULATIONAHA.106.635573.