This prospective cohort study was performed in an eight-bed pediatric intensive care unit (PICU) at Songklanagarind Hospital, Hat Yai, Songkhla. Children aged 1 day–18 years with congenital or acquired cardiac disease who underwent elective open cardiac surgery with CPB and were admitted to the PICU postoperatively were included in the study. The exclusion criteria were as follows: preterm infants (gestational age <37 weeks), weight <2 kg, inability to wean off CPB, requirement for extracorporeal membrane oxygenation (ECMO) before leaving the operating room, and absence of arterial and central lines postoperatively. This study was approved by the Institutional Review Board of the Faculty of Medicine, Prince of Songkla University (Institutional Review Board approval number 64-299-1-1; date of approval: August 3, 2021). The study was conducted in accordance with Good Clinical Practice and the Helsinki Declaration of 1975.

Anesthesia was initiated according to our standardized protocol. Following incision, heparin was administered intravenously at a dosage of 3 mg/kg to achieve an activated clotting time (ACT) exceeding 400 s. ACT was monitored at 30-min intervals throughout the procedure, with an additional heparin dose of 3 mg/kg provided if the ACT fell below 400 s. Intraoperative corticosteroids were administered intravenously, based on individual anesthesiologist preferences. The priming solution for the CPB circuit included either Ringer's lactate or normal saline with mannitol, along with 20% albumin for patients weighing less than 10 kg. Leukocyte-depleted packed red blood cells were utilized when the preoperative hematocrit was below 30%. CPB was performed using a CAPIOX® FX05 Oxygenator and Stockert S5 pump, following the α-stat strategy at a target temperature of 28°C. Hypothermia was induced to varying degrees; depending on the surgical procedure. Antegrade cold blood cardioplegia was administered at a volume of 20 ml/kg to achieve cardiac arrest, with an additional 10 ml/kg given if aortic cross-clamp time exceeded 20 min. Pump flow rates were maintained between 100 and 150 ml/kg/min for infants and 2.5–3.0 L/m²/min for older patients: adjusted according to age-appropriate mean arterial pressure. Arteriovenous modified ultrafiltration was selectively applied, based on the consensus of the surgical team and perfusionist, and performed for 5 min after separating from CPB. At the conclusion of CPB intravenous protamine sulfate was administered to reverse heparin anticoagulation.

After receiving written informed consent from the patients’ parents or legal guardians before operation, 0.3 ml blood samples were simultaneously collected from both the arterial and internal jugular central lines. This sampling was repeated at four different time points after patient arrival in the PICU: at PICU arrival (T0), 6-h (T6), 12-h (T12), and 24-h (T24) post operation. Arterial and venous blood gas readings were accepted if the samples were collected within 5 min of each other. A 1 h time gap in the collection of blood samples from the research schedule was allowed owing to unpredictable intensive care unit occupancy. Blood samples were analyzed within 1 min of blood collection using an arterial blood gas analysis machine (ABL 800 Basic Radiometer; Copenhagen, Denmark) located within the PICU. The attending staff independently provided standard post-cardiac surgery care, including fluid resuscitation, vasopressor and inotropic medication, and steroid administration without the researcher's involvement. If either the central or arterial lines were displaced within 24 h postoperatively, the remaining data were recorded as missing.

Data, including general baseline characteristics, intraoperative parameters, postoperative interventions, and outcomes, were collected until the patient was discharged. The Risk Adjustment for Congenital Heart Surgery tool (15) was used to classify the risk of mortality after congenital cardiac surgery. The vasoactive-inotropic score (VIS) was calculated using the formula by Gaies et al. (16) as follows:VIS=Dopaminedose(μg/kg/min)+dobutaminedose(μg/kg/min)+100×epinephrinedose(μg/kg/min)+10×milrinonedose(μg/kg/min)+100×norepinephrinedose(μg/kg/min)+10000×vasopressindose(U/kg/min)Acute kidney injury was diagnosed according to the Kidney Disease-Improving Global Outcomes guidelines, 2012 (17), based on the presence of any of the following: a >0.3 g/dl increase in the serum creatinine level within 48 h, >1.5 times increase in the serum creatinine level from a known baseline value or one recorded within the preceding 7 days, and urinary volume <0.5 ml/kg/h for 6 h. The 28-day ventilator-free days (VFDs), 28-day inotrope-free days, and 28-day ICU-free days (IFDs) were defined as the number of days that the patient survived without invasive ventilation, inotropic drugs, or ICU admission during the first 28 days postoperatively, with the day after the first postoperative night considered as day 1. These variables were counted as zero for non-survivors.

The independent variables studied included the bedside surrogate markers (VACO₂, O₂ER, and lactate level) that were measured postoperatively. ScvO₂ was not evaluated because the participants’ single-ventricle physiology might have interfered with ScvO₂ interpretation. VACO₂ was calculated as the central venous minus arterial CO₂ level and O₂ER as the ratio between the difference in arterial and central venous oxygen saturation divided by the arterial oxygen saturation.

The primary outcomes were LCOS-related outcomes, which was consisted of at least two of the following criteria within 24 h postoperatively: (1) VIS ≥20; (2) left ventricular ejection fraction <50% on echocardiography; (3) any unplanned surgery or intervention, cardiac arrest, or utilization of ECMO; and (4) death. The secondary outcomes included: 28-day VFDs, 28-day inotrope-free days, 28-day IFDs, percentage of morbidities (reintubation, significant arrhythmic events requiring medication or intervention, acute kidney injury, renal replacement therapy, and neurological complications), and mortality rate.

Statistical analyses involved descriptive analysis of means (standard deviations) for normally distributed continuous data, medians (interquartile ranges) for non-normally distributed continuous data, and percentages for categorical data. The Student's t-test or Mann–Whitney U test was used for intergroup comparison of continuous data, depending on the pattern of data distribution. Categorical data were compared using the chi-squared or Fisher's exact test, as indicated. Correlation analysis between bedside parameters was performed using Pearson's and Spearman's correlation coefficients for parametric and non-parametric variables, respectively. Linear mixed-model regression was used to compare the postoperative laboratory (lactate, VACO₂, and O₂ER) values between the LCOS and no-LCOS groups over time with adjusted confounder variables that might affect outcome (age, RACHS, type of repair, CPB time, intraoperative corticosteroid, intraoperative fluid balance). Receiver operating characteristic curves were used to evaluate performance of VACO₂ on discriminating LCOS-related outcomes. Subgroup analysis was performed according to patient's age, type of repair, and RACHS score. Statistical significance was set at p < 0.05. All analyses were conducted using R version 4.3.1 (The R Foundation for Statistical Computing, Vienna, Austria).

Ethical approval for this study was obtained from the Ethics Committee of the Faculty of Medicine, Prince of Songkla University, Songkhla, Thailand.

Between August 2021 and December 2023, 136 patients underwent open-heart surgery. Nine patients were excluded from the study, which included five patients who could not be weaned off CPB postoperatively and four patients who underwent emergency operations. The final analysis included 127 patients, of which 26 (20.4%) developed LCOS-related outcomes within 24 h postoperatively. Five patients (3.9%) required re-operation or reintervention, seven (5.5%) required ECMO, three (2.4%) developed cardiac arrest, and seven (5.5%) died. Factors significantly associated with LCOS-related outcomes included preoperative sepsis, a high Risk Adjustment for Congenital Heart Surgery score, long CPB time, long aortic clamp time, high amount of blood transfusion, and intraoperative corticosteroid administration (Table 1). The rates of postoperative intervention, including systemic steroid administration (60.4% vs. 24.3%, p < 0.01), blood transfusion [19.6 [10.0–34.6] vs. 6.4 [0–13.5] ml/kg, p < 0.01], and renal replacement therapy (13% vs. 0%, p = 0.002), as well as those of end-organ dysfunction and death, were higher in the LCOS group than in the no-LCOS group (Supplementary Table 1).

The patterns of laboratory values at the four time points are illustrated in Figure 1. The overall values of three parameters in the LCOS group were higher than those in the non-LCOS group at all four time points, with the exception for the VACO₂ at 24 h post operation. Univariate analysis performed to compare bedside surrogate marker values between patients with and without poor LCOS-related outcomes revealed that the LCOS group had significantly higher VACO₂ values at 12 h postoperatively, higher lactate levels at all four time points, and higher O₂ER values at PICU admission and at 6 h postoperatively than the no-LCOS group. Analysis of the overall unadjusted area under the curve (AUC) of VACO₂ was 0.58–0.64 which was inferior to serum lactate levels in predicting LCOS events (Table 2).

After performing linear mixed-model regression analysis, the relationship between lactate change and the LCOS occurrence was significantly at 6 h and 12 h postoperatively (p < 0.05), whereas the VACO₂ change was significant at 24 h postoperatively (p = 0.02) and there was no relationship between O₂ER change at all four time points (Figure 1).

To achieve the secondary study objective, we used a 6-mmHg cut-off, as described previously (9), to categorize the 6-h postoperative VACO₂ values into two groups. The results revealed significantly fewer 28-day inotrope-free days and 28-day IFDs in patients with a VACO₂ ≥6 mmHg than in patients with a VACO₂ <6 mmHg (24.0 [22.0–27.0] vs. 26.5 [23.0–28.0] days, p = 0.03 and 23.0 [18.2–26.0] vs. 24.0 [22.0–27.0] days, p = 0.03, respectively) (Table 3). Subgroup analysis in patients with biventricular physiology and with aged more than one-month old revealed insignificant different outcomes (Supplementary Tables 2, 3).

The correlations between VACO₂ and the other parameters (lactate level and O₂ER) at the four time points are shown in Figures 2, 3. While VACO₂ did not show a significant correlation with lactate levels, it demonstrated a moderate correlation with O₂ER at 6 h postoperatively (r = 0.58; p < 0.001) and a weak correlation with O₂ER at 12 and 24 h postoperatively (r = 0.22; p = 0.015 and r = 0.19; p = 0.045, respectively).