Background

In humans, oxygen is vital for proper aerobic metabolism. The key metabolic reactions that take place in the mitochondria of cells involve the combination of a fuel (e.g., sugars, fats) with oxygen to create energy, as well as carbon dioxide (CO2) and water (H2O) [1]. Hypoxemia is characterized by a drop in the partial pressure of oxygen in the blood, whereas hypoxia is characterized by a decrease in tissue oxygenation [2]. In hypoxia, oxygen is not available sufficiently at the tissue level to maintain proper homeostasis. This can be caused by insufficient oxygen transport to tissues due to poor blood supply or low oxygen concentration in the blood as in hypoxemia [3]. Hypoxemia occurs frequently during the early postoperative period in infants who have undergone surgical repair of congenital heart defects. If hypoxemia is not corrected early, it can result in the need for prolonged mechanical ventilation (MV) support, intensive care unit (ICU) stay, and hospital stay. Additionally, uncorrected postoperative hypoxemia can cause multiple organ dysfunction [4]. Hypoperfusion resulting from circulatory deficiencies is characterized by inadequate oxygen delivery to tissues. The degree and duration of hypoperfusion are linked strongly to the development of organ failure [3]. It should be noted that global tissue hypoxia (GTH) is not detected through conventional monitoring of vital signs, such as heart rate, blood pressure, central venous pressure, and urine output [5].

Identifying patients at risk for postoperative complications is challenging, but it would assist physicians and nurses in monitoring and allocating additional resources to certain patients to prevent or quickly address and treat anticipated complications. Low central venous oxygen saturation (ScvO2) and high arterial lactate levels, which are indicators of circulatory problems, have been utilized by clinicians to initiate and improve early hemodynamic treatment [6]. The balance of oxygen supply and demand is measured by ScvO2; it decreases when cardiac output, hemoglobin, or arterial oxygen saturation decreases or when oxygen demand increases due to fever, shivering, agitation, or hypercatabolic state. GTH and lactic acidosis may be the result of disruption of this balance, which may manifest itself with clinical symptoms of hypoperfusion, such as hypotension or oliguria [6, 7]. In surgical patients at high risk for complications, elevated lactate levels have been linked to an increased risk of morbidity and mortality. Lactic acidosis is usually caused by systemic hypoperfusion in severely ill patients; however, it can also be caused by liver disease, thiamine deficiency, epinephrine, other drugs/toxins, or inborn metabolic abnormalities. These causes can affect the specificity of lactate levels to predict circulatory problems [8].

In patients undergoing general and cardiac surgery who are at high risk for complications, the association between low ScvO2 or elevated lactate levels and morbidity/mortality has been extensively explored. The critical values for ScvO2 and lactate in these studies were 60–70% and 2–4 mmol/L, respectively [9,10,11]. Serial blood lactate levels and the ScvO2/lactate ratio are useful predictors of death within 7 days after laparotomy [12]. Rivers et al. [13] used an increase in lactate levels to identify a subset of patients with severe sepsis and septic shock; this subset was treated early with medications aimed at improving ScvO2, and morbidity and mortality rates were reduced as a result. However, few researchers have investigated the prognostic significance of the combination of low ScvO2 and elevated lactate levels, especially in pediatric patients. We examined whether this combination could better predict complications after cardiac surgery in pediatric patients.

Methods

Sample size

We assumed that the area under the curve (AUC) of lactate levels and ScvO2 that would help predict GTH after cardiac surgery was 0.75 [10]; therefore, the minimal sample size using MedCalc software was 73 (on the basis of a significance level of 95%, a dropout rate of 20%, an alpha error of 0.05, 80% power, and null hypothesis of 50%). We randomly selected 73 children who underwent surgery at a cardiac tertiary referral center in Aswan, Egypt.

Study population and study setting

This prospective cohort study was conducted at a cardiac tertiary referral center in Aswan, Egypt, from August 2020 to March 2021. All patients aged 18 years or younger, on or off cardiopulmonary bypass, and with a central venous catheter in the internal jugular position after cardiac surgery were eligible for inclusion. To ensure that hemodynamic compromise was the only cause of a decrease in ScvO2 or an increase in lactate level, we excluded all children with the following conditions: diseases and conditions that cause impaired lactate metabolism, such as liver failure and sepsis; conditions that cause a high lactate level, such as high adrenaline dose or inotropic support before surgery; conditions that affect oxygen delivery, such as preexisting congestive cardiac failure, cardiomyopathy, hemodynamic instability, or cardiac arrest; conditions that affect oxygen content of the blood, such as perioperative bleeding; preoperative elevation of any laboratory value that indicated systemic hypoperfusion; conditions affecting outcomes of surgery, such as preexisting renal or organ failure; or the need for extracorporeal membrane oxygenation.

Study procedures

Sociodemographic data and anthropometric measures were collected from all study participants. The children’s ages were categorized as 1 month or younger, more than 1 month to 1 year, or older than 1 year. Weights of children were plotted on the Egyptian growth chart (weight for age) as being below the 5th percentile; in the 10th, 25th, 50th, or 75th percentile; and above the 95th percentile. After admission to the pediatric intensive care unit (PICU), all patients underwent a complete clinical examination, and the extensive medical history was collected from their caregivers or relatives. Information about operative details and events was obtained from surgeons and anesthesiologists. Routine laboratory data were obtained at PICU admission and every 24 h: complete blood cell count, coagulation profile, kidney function test and urinary output, liver function test, and C-reactive protein level. Serial lactate levels and ScvO2 were measured from arterial and central venous lines, respectively, immediately and 3, 6, 12, 18, and 24 h after PICU admission. Serial mean arterial blood pressure (MAP) was measured immediately and 3, 6, 12, 18, and 24 h after PICU admission. The vasoactive-inotropic score (VIS) was calculated for each patient according to the following formula [14]: VIS = dopamine (μg/kg/min) + dobutamine (μg/kg/min) + 100 × epinephrine (μg/kg/min) + 100 × norepinephrine (μg/kg/min) + 10 × milrinone (μg/kg/min) + 10,000 × vasopressin (mU/kg/min). Moderate GTH was defined as ScvO2 of < 70% and lactate level between ≥ 2 and < 4 mmol/L, and severe GTH was defined as ScvO2 of < 70% and lactate level of ≥ 4 mmol/L). Cryptic hypoperfusion was defined as moderate or severe GTH with normal macrohemodynamic parameters [15].

Outcomes

Complications included the need for prolonged PICU stay (≥ 14 days), prolonged MV (≥ 72 h), major adverse events (chest infection, sepsis, cardiac arrest, neurological deficit, acute kidney injury, and the need for surgical re-exploration), the need for extracorporeal membrane oxygenation, and inhospital mortality. Therefore, the patients were classified as those with and without complications.

Statistical analysis

Data for continuous numerical variables were calculated as means and standard deviations. In case of skewed data, medians and interquartile ranges were calculated instead. Categorical data were presented in the form of numbers and percentages. To predict postoperative complications, we used receiver operating characteristic curve (ROC) analysis and the Youden index to examine the prognostic value of serum lactate level, ScvO2, MAP, and inotropic score [16]. To assess the relationship between ScvO2, lactate level and complications, we performed Pearson’s correlation analysis. The chi-square test (χ2) was used to evaluate the associations between categorical variables. We performed binary logistic regression analysis to calculate odds ratio (OR) and 95% confidence interval (CI) for each variable to assess the variables that predicted complications (outcome); weight, lactate level, MAP, inotropic score, and ScvO2 were examined as independent covariates (predictors). We considered p-values of < 0.05 statistically significant. We created nomograms to allow for a preliminary visual assessment of the risk and severity of postoperative complications. For data analysis, we used SPSS Statistics, version 26 (IBM Corporation, Armonk, NY, USA), and Stata 17.0 (Stata Corp. LLC, College Station, TX, USA).

Results

Sociodemographic and laboratory findings

Of the 73 children included in this study, 51 (69.86%) were male, 33 (45.21%) were 1 month old or younger, 15 (20.55%) were older than 1 month and younger than 1 year, and 25 (34.24%) were older than 1 year. The weights of 25 (34.25%) were below the 5th percentile (Table 1).

Table 1 Patients’ sociodemographic features (n = 73)
Full size table

The complete blood cell count was within normal except for the mean platelet count (117.51 ± 9.25 × 103/μL), which was below the normal range. Similarly, the biochemical profile was normal except for the mean level of aspartate aminotransferase (121.29 ± 8.26 U/L), which was above the normal range. The mean complexity score was 8.86 ± 2.67, and the mean inotropic score was 12.73 ± 8.20. Total surgical repair was performed in 59 patients (90.8%), and 14 (19.2%) underwent palliative surgical intervention (Supplementary Table S1).

Complications after cardiac surgery

Complications developed in 35 patients (49.95%). The most common complication after cardiac surgery was the need for prolonged MV (38.36%), chest infection (30.14%), prolonged stay in the PICU (24.66%), and sepsis (9.59%) (Supplementary Figure S1).

Correlation between ScvO2, arterial lactate level, and complications after cardiac surgery

To study the correlation between lactate levels and different outcomes, different measures of lactate were compared with hospital stay (LOS), PICU LOS, MV duration, and inotropic duration. Baseline lactate levels were significantly and positively correlated with MV duration (r = 0.24, p < 0.5) and inotropic score (r = 0.25, p < 0.05) (Supplementary Tables S2, S3).

Role of ScvO2, arterial lactate level, vasoactive-inotropic score, and mean arterial blood pressure in predicting complications after cardiac surgery

ScvO2 measured 6 h after PICU admission had the highest AUC (85.5%; p < 0.001), with 81.6% sensitivity and 82.9% specificity for predicting complications, a positive predictive value (PPV) of 83.8%, and a negative predictive value (NPV) of 80.6%. Lactate measured 12 h after PICU admission had an AUC of 75.0% (p < 0.001), with 63.2% sensitivity and 82.0% specificity, a PPV of 80.0%, and an NPV of 67.4%. The ratio of ScvO2 6 h after admission to lactate level 12 h after admission had a sensitivity of 68.4% and specificity of 80.0% in predicting complications (AUC 83.0%, p < 0.001). MAP 18 h after admission had an AUC of 73.6% (p < 0.001), with 42.1% sensitivity, 97.1% specificity, a PPV of 94.1%, and an NPV of 60.7%. The inotropic score had the lowest AUC of (63.4%; p < 0.001), with 92.1% sensitivity, 40.0% specificity, a PPV of 62.5%, and a NPV of 82.4% (Table 2). The ROC showed that the best predictors were central venous oxygen saturation (ScvO2) 6 h after admission to the PICU (ScvO2_6) followed by arterial lactate level 12 h after admission to the PICU (Lactate_12), ScvO2/lactate ratio, mean arterial pressure (MAP) 18 h after admission to the PICU (MAP_18), and VIS (Fig. 1).

Table 2 Validity and performance of specific measurements and vasoactive-inotropic score in predicting complications after cardiac surgery in pediatric patients
Full size table
Fig. 1
figure 1

Receiver operating characteristic curves of predictors of complications after cardiac surgery. The best predictors were central venous oxygen saturation (ScvO2) 6 h after PICU admission (ScvO2_6) followed by arterial lactate level 12 h after PICU admission (Lactate_12), ScvO2/lactate ratio, mean arterial pressure (MAP) 18 h after PICU admission (MAP_18), and VIS

Full size image

Predictors of complications after cardiac surgery

The incidence of complications was highest among children between the ages of one month and one year (78.79%) in comparison with children between the ages of one month and one year (33.33%) and children older than 1 year (16.00%; p < 0.001). The mean ScvO2 measured 12 h after admission was significantly higher among patients with complications (64.74 ± 11.74) than among those without complications (49.14 ± 11.74; t = 5.86, p < 0.001). Furthermore, the mean arterial blood pressure measured after 18 h from admission (66.42 ± 13.33 mm Hg) was higher in patients without complications than in those with complications (56.06 ± 8.10 mm Hg; t = 4.03, p = 0.001). The mean inotropic score was lower in patients without complications (0.29 ± 0.46) than in those with complications (0.54 ± 0.51; t = \(-\) 2.24 p = 0.029), as was the mean lactate level (1.87 ± 0.90 mmol/L vs 2.62 ± 1.49 mmol/L, respectively; t = \(-\) 2.58, p = 0.012). Neither sex nor patient weight was significantly associated with complications between the studied groups (Table 3).

Table 3 Baseline laboratory findings as predictors of complications
Full size table

The regression model was well-calibrated (Table 4). The Hosmer–Leeshawn chi-square value was 4.61 (p = 0.59), and the overall model was statistically significant (χ2 = 48.29, p < 0.001). The predictive capacity of the model increased from 52.80 to 86.10%. ScvO2 measured 6 h after admission was a significant predictor of complications. The OR decreased by 0.87 (95% CI 0.798–0.948, p = 0.002) for each increase in ScvO2 by 1%. Weight was a statistically significant predictor of the incidence of complications (OR 0.01; 95% CI 0.001–0.689; p = 0.033). The AUC of the developed logistic regression model was 92%, p < 0.001 (Supplementary Figure S2). The nomogram for calculating the probability that complications would develop after cardiac surgery is depicted in (Supplementary Figure S3).

Table 4 Multivariable regression analysis of predictors of postoperative complications
Full size table

Discussion

We evaluated whether measuring arterial lactate level in addition to ScvO2 would improve the prediction of complications after pediatric cardiac surgery among 73 patients admitted to a cardiac tertiary referral center in Aswan, Egypt. In this study, 35 patients (49.95%) developed complications. The most common complications after cardiac surgery were prolonged MV, chest infection, prolonged PICU LOS, and sepsis. In addition to age, ScvO2 measured 6 h after PICU admission, blood lactate level measured 12 h after PICU admission, and MAP measured 18 h after PICU admission were significantly associated with complications. Furthermore, all of these markers had good validity in predicting complications after cardiac surgery. However, multivariate regression analysis revealed that ScvO2 and weight were the only predictors of complications after cardiac surgery.

The demand for tools to predict complications after cardiac surgery is not new. Research has focused on lactate clearance, duration of high lactate levels, duration of cardiopulmonary bypass, inflammatory markers, and other indicators, with an emphasis on predicting low cardiac output syndrome; however, no optimal prognostic method has been established [17]. A study of the role of pro-inflammatory cytokines in inducing cardiac injury after arterial switch procedure in 63 newborns revealed that cardiac troponin T, interleukin-6, and interleukin-8 levels were higher in patients who developed low cardiac output syndrome [18].

In our study, we observed that all measured lactate levels and ScvO2 during the first 24 h after admission were significantly correlated with hospital LOS, PICU LOS, duration of MV, and duration of inotropic support. To best predict clinical outcomes, the optimal times to measure lactate levels and ScvO2 are 12 and 6 h after admission to the PICU, respectively. Maillet et al. [19] found that compared to patients whose lactate levels were < 3 mmoL/L, those with lactate levels > 3 mmoL/L on ICU had considerably longer durations of both MV and LOS of the ICU, as well as an increased mortality rate. In a large study (n = 1820), Kogan et al. [20] found that a maximum lactate level of 4.4 mmol/L during the first 10 h after admission to the ICU was associated with prolonged MV, prolonged LOS in the ICU, and an increased rate of mortality. Similarly, Pearse et al. [9] found that the rate of complications was considerably higher among patients with a ScvO2 of < 64.4% during the first 8 h after surgery than among those with a higher ScvO2. On the other hand, lactate levels in patients with and without complications were < 2 mmoL/L; the two groups did not differ significantly. Of note, ScvO2 and lactate measurements play an important role in the diagnosis of occult hypoperfusion. In individuals with normal macrohemodynamics, occult hypoperfusion is characterized as moderate to severe GTH with a mean arterial pressure of 65 mmHg, a mean central venous pressure of 8 mmHg, and mean urine output of 0.5 mL/kg/h [15]. Hu et al. [10] found that on admission to the ICU, 19 patients (32%) had occult hypoperfusion. GTH was found in 21 individuals (35%): 13 (22%) had moderate GTH and 8 (13%) had severe GTH.

In our study, arterial lactate level 12 h after PICU admission had 63.2% sensitivity and 82.0% specificity for predicting postsurgical complications, ScvO2 had 81.6% sensitivity and 82.9% specificity, VIS had 92.1% sensitivity and 40.0% specificity, and MAP measured 18 h after PICU admission had 97.1% sensitivity and 94.1% specificity. Similarly, Gaies et al. [20] reported that the maximum VIS estimated during the first 24 h after cardiac PICU admission was significantly associated with morbidity and death. A maximum VIS of ≥ 20 indicates a higher risk of a bad composite clinical outcome. Seear and his colleagues [21] found that arterial lactate and ScvO2 levels were the only postoperative measures that could predict serious adverse outcomes after cardiac operations in pediatric patients. Moreover, the ScvO2/lactate ratio appeared to have better predictive value: when the ratio was < 5, the PPV for complications was 93.8%, with 78.9% sensitivity and 90.5% specificity. Each measurement separately had high specificity but limited sensitivity. Single-measure predictive power was of only fair quality, but it could have been increased for patients at high risk of complications by tracking repeated measures over time 0.986.

Rocha et al. (2021) [22] found that, overall, the ScvO2/lactate ratio distinguished patients with and without major adverse events very effectively (AUC = 84%), outperforming either variable alone, with 48% sensitivity, 94% specificity, 60% PPV, and 91% NPV. However, when we calculated the ratio of ScvO2 6 h after PICU admission and lactate level 12 h after PICU admission, the AUC did not improve. Moreover, when we included all significant variables in the bivariable analysis, in addition to weight, the regression model showed that ScvO2 and weight were the only significant predictors of complications, with an AUC of 92%. This finding may be attributable to the correlation between ScvO2 and lactate. Bisaya and his colleagues [23] found that when ScvO2 was 65%, arterial lactate level was weakly correlated with ScvO2 (r2 = 0.0431, p < 0.001), whereas they were strongly correlated in individuals with an O2 extraction ratio of 50% (r2 = 0.93, p = 0.0019). Of note, about one-third of patients studied by Bisarya et al. had a ScvO2 of 50% or less. In our study, approximately one-third of the patients had a ScvO2 6 h after PICU admission of ≤ 50 which explains the high correlation between arterial lactate and ScvO2.

In this study, we found that one of the main predictors of postoperative complications was the weight of the patient. Similarly, an examination of demographic data, preoperative results, and surgical details of the population, patients with serious postoperative morbidity: they were considerably with younger age, weighed less, and had higher Aristotle scores [24]. This finding can be explained by the fact that these conditions are associated with poor growth and development. In contrast, Seear et al. [21] reported that both body weight and duration of circulatory arrest were weak predictors of complications, but they were not predictors at every measured time point.

Strengths and limitations

To the best of our knowledge, this study is one of the first in pediatrics to determine the prognostic value of using combined markers in the prediction of complications after cardiac surgery. In this study, we obtained serial measurements of both markers and determined measurement times that were most strongly correlated with the outcomes. Additionally, the prospective nature of the study reduces the bias of missing data and allows intensive clinical and laboratory follow-up of the patients. However, this study had limitations. First, the significant heterogeneity of heart conditions and surgeries limits generalizability to other centers and populations; as a result, our findings must be interpreted with care. Actually, we included patients with different categories of congenital heart disease: some were acyanotic (VSD, ASD, coarctation of the aorta), and others were cyanotic of whom some have total correction (as arterial switch for patients with TGA and total correction of tetralogy of Fallot), and the others had only palliative repair (shunt for patients with complex congenital heart and Glenn for patients with single ventricle physiology); all these would have different baseline saturation and different oxygen delivery and consequently different ScVo2 and lactate owing to different degrees of mixing due to residual shunting, or amount of shunting is dependent on the patient’s hemodynamics as MAP and pulmonary vascular resistance PVR that are continuously changing according to different conditions and other variables (fever, sedation, over or under shunting). Another limitation was the different age categories; we had different age groups with different physiology, age, weight, and hemodynamic parameters as blood pressure that could not be accounted for, so we strongly recommend further research in every age category to avoid these limitations. Finally, this study was a single-center observational study; future multicenter studies are mandatory to provide more robust evidence.

Conclusion

In pediatric patients admitted to the PICU after cardiac surgery, ScvO2 measured 6 h after admission can predict the risk of complications and allow early initiation of therapy to achieve better outcomes. In this study, monitoring of lactate levels and ScvO2 facilitated rapid identification of hypoperfusion in pediatric patients after cardiac surgery. Our findings support the use of ScvO2 levels to detect hypoperfusion that would otherwise go unnoticed by conventional monitoring, thereby preventing the worsening of hypoperfusion and the development of organ failure. However, more research is necessary to determine the efficacy of lactate levels and ScvO2 in guiding hemodynamic management and their effect on morbidity and mortality in pediatric patients after cardiac surgery.