Perioperative factors and radiographic severity scores for predicting the duration of mechanical ventilation after arterial switch operation

Abstract

Objectives:

Cardiac surgery on cardiopulmonary bypass (CPB) during the neonatal period can cause perioperative organ injuries. The primary aim of this study was to determine the incidence and risk factors associated with postoperative mechanical ventilation duration and acute lung injury following arterial switch operation (ASO). The secondary aim was to examine the utility of the Brixia score for characterizing postoperative acute lung injury (ALI).

Design:

A retrospective study.

Setting:

A single-center university hospital.

Participants:

A total of 93 neonates with transposition of great arteries with intact ventricular septum (dTGA/IVS) underwent ASO.

Interventions:

None

Measurements and Main Results:

From January 2015 to December 2022, 93 neonates with dTGA/IVS were included in the study. The cohort had a median age of 4.0 (3.0–5.0) days and a mean weight of 3.3 ± 0.5 kg. About 63% of patients had ≥ 48 hours postoperative mechanical ventilation after ASO. Risk factors included prematurity, post-CPB transfusion of salvaged red cells, platelets and cryoprecipitate, and postoperative fluid balance by univariate analysis. The larger transfused platelet volume was associated with the risk of ALI by multivariate analysis. The median baseline Brixia scores were 11.0 (9.0–12.0) and increased significantly in the postoperative day 1 in patients who developed moderate ALI 24 hours after admission to the intensive care unit (15.0 (13.0–16.0) vs. 12.0 (10.0–14.0), p = 0.046).

Conclusions:

Arterial switch operation results in a high incidence of ≥ 48-hour postoperative mechanical ventilation. Blood component transfusion is a potentially modifiable risk factor. The Brixia scores may also be used to characterize postoperative acute lung injury.

Introduction

Despite significant advancements in extracorporeal circulation and perioperative care, cardiac surgeries in neonates are associated with high morbidity and mortality ⁽¹⁾. Lung injury is one of the major organ injuries that frequently occur following cardiac surgery ⁽²⁾. Transposition of the great arteries (TGA) is the second most common cyanotic congenital cardiac lesion. One subtype, dextroposition TGA with intact ventricular septum (dTGA/IVS), requires definitive surgical correction (arterial switch operation, ASO) during the first days of life to ensure optimal outcomes ^{(3, 4)}. During the neonatal period, the lungs can be greatly compromised with significant pulmonary overcirculation. Mechanical ventilation, cardiopulmonary bypass (CPB) and blood transfusion further contribute to pulmonary insult, leading to delayed recovery ^{(5, 6)}. In recent years, enhanced recovery and early extubation have become increasingly common in pediatric cardiac surgery ⁽⁷⁾. Despite this, few studies address the perioperative factors that could potentially shorten postoperative mechanical ventilation and reduce lung injury in this patient population.

This study aims to investigate the incidence and potential predictors of prolonged mechanical ventilation and lung injury following ASO. Furthermore, the conventional diagnostic tool for lung injury, such as the PaO₂/FiO₂ (P/F) ratio, may be of limited utility in some of congenital heart diseases due to their intracardiac shunts, including parallel circulation found in dTGA/IVS preoperatively ⁽⁴⁾. Thus, the secondary aim of this study is to assess the utility of the Brixia score ^{(8, 9)} as a novel tool in characterizing lung injury in neonatal cardiac surgery.

Methods

Study Cohort

The Institutional Review Board at Boston Children’s Hospital approved this retrospective cohort study. Neonates who underwent ASO from January 2015 to December 2022 were identified. Only neonates diagnosed with dTGA/IVS were included in the study. Patient demographics, preoperative characteristics, operative variables, and postoperative outcomes were collected from the hospital’s electronic medical records. Arterial blood gas data from the intraoperative baseline, post-CPB, postoperatively at intensive care unit (ICU) admission, 6, 12, 24, and 48 hours after ICU admission were identified and used in statistical analysis. Patients diagnosed with tracheal, tracheobronchial, and laryngeal abnormalities, and those with missing pertinent data (mechanical ventilation, CPB duration) were excluded.

In this study cohort, pressure-controlled ventilation was the primary mode of intraoperative mechanical ventilation. All patients received general anesthesia with a combination of volatile and intravenous anesthetics. Cuffed endotracheal tubes were used routinely. CPB circuit was primed with reconstituted whole blood to maintain hematocrit ≥ 30%. Additional blood product transfusion and coagulation management were based on thromboelastogram (TEG) data and the discretions of the anesthesiologists. Salvaged blood and packed red cells were transfused where indicated to maintain hematocrit ≥ 30%. Methylprednisolone, 30 mg/kg, was administered during CPB in all patients. Postoperative and extubation management was under the discretion of cardiac intensivists and followed institutional protocols.

The duration of mechanical ventilation was determined as the interval from the end of operating room time (room-out time) and extubation time at the ICU. This did not include the duration of non-invasive bilevel and/or continuous positive ventilation if required after extubation. Perioperative and outcome data were compared between patients who were extubated within 48 hours and those extubated after 48 hours. The P/F ratio was calculated as PaO₂ divided by FiO₂ at the corresponding time, oxygenation index (OI) as (FiO₂ x mean airway pressure x 100)/PaO₂, and oxygen saturation index (OSI) as (FiO₂ x mean airway pressure x 100)/SpO₂. Acute postoperative lung injury (ALI) was defined as a P/F ratio < 300 at the 24-hour time point after postoperative ICU admission. Postoperative fluid balance was calculated as total fluid output – total fluid input from the corresponding postoperative day.

Brixia scores

We adopted the Brixia score ⁽⁸⁾, initially developed for grading SARS-CoV-2 pneumonia severity, to quantify lung abnormalities on perioperative chest radiographs. In this scoring system, the lungs were divided into six anatomical zones (upper, mid, and lower sections of the right and left lungs). The score in each lung zone was assigned based on the perceived radiographic abnormalities present in that zone: 0, no abnormality; 1, interstitial infiltrates; 2, interstitial predominance with alveolar infiltrates; and 3, alveolar dominance with interstitial infiltrates (Figure 1). The sum of the scores from these six lung zones was assigned as each radiograph’s Brixia score (minimum 0, maximum 18).

Figure 1.

Chest radiographs of a dTGA/IVS patient demonstrate the assigned score of each lung zone. (A) Preoperative Brixia score (1+1+0+1+0+0) = 3. (B) Postoperative Brixia score (3+3+2+1+1+1) = 11.

Patients’ Brixia scores were evaluated by three assessors (PL, KY, and WM), who were all blinded to patients’ perioperative and outcome data during scoring process. An experienced radiologist (AT) trained these three reading physicians to standardize the interpretation of the chest radiographs. As part of the training process and to further maximize reader agreement, intraclass correlation coefficient (ICC) was calculated in two separate practice sessions prior to evaluating the patients’ radiographs. For each patient, we identified three chest radiographs from the medical record, each corresponding to one of the following three time periods: preoperative, postoperative (at ICU admission), and postoperative (one day after surgery, POD1). In cases where multiple chest radiographs existed on the specified time period, the most recent preoperative image, the first image after ICU admission, and the first image on postoperative day 1 were selected for analysis. Chest radiographs were viewed using the hospital picture archiving and communication system (Synapse, Fujifilm Medical System, USA). Standard image displays using preset window settings for optimal assessment of lung parenchyma were used during the scoring process. Each assessor independently scored all images in four separate sessions. The scores from each lung zone by each assessor were compared and discussed in determining the consensus Brixia scores for all the chest radiographs. All assessors agreed to the final Brixia scores.

Statistical analysis

We used means and standard deviations as summary statistics for continuous variables with a normal distribution, and medians and interquartile ranges for variables with non-normal distributions. We utilized student t-test or Mann-Whitney U test to compare continuous variables, depending on the normality of the distribution. Categorical variables were presented as frequencies and percentages, and the chi-square test or Fisher’s exact test was used as appropriate. We conducted logistic regression analysis to determine the association between the duration of mechanical ventilation and patient factors. Multivariable analysis was performed to determine the adjusted odds ratios with 95% confidence intervals. We used Youden J-index to determine the variable cut-off points. The correlations between variables were provided by Pearson’s correlation coefficient. We calculated ICC (two-way random-effects model) to determine the assessors’ agreement of the Brixia scores. A p-value of < 0.05 was considered to be statistically significant. These statistical analyses were performed using software STATA 17 (STATA version 17; StataCorp; College Station, TX, USA) and Prism 10 (GraphPad Prism version 10.0.0, GraphPad Software, La Jolla, CA, USA).

Results

Characteristics of the study cohort

From January 2015 to December 2022, 2,911 cardiac surgeries with CPB were performed in neonates and infants under 1 year of age. Among them, 93 neonates with dTGA/IVS who underwent ASO were included in the study cohort. The perioperative data are shown in Table 1. Most were male (68%) with a median age of 4.0 (3.0–5.0) days, and a mean weight of 3.3 ± 0.5 kg. About 45% underwent urgent balloon atrial septostomy (BAS) before arriving for definitive surgery, and most (76%) of the cohort received preoperative mechanical ventilation support at least briefly. Intraoperative baseline P/F ratios were 55.9 (49.1–65.9), which improved following ASO.

Table 1.

Demographic and perioperative data

Demographic data	Total (n = 93)	ventilation < 48 hours (n = 34)	ventilation ≥ 48 hours (n = 59)	p-value	OR	95%CI		p-value
Age (day)	4.0 (3.0–5.0)	3.5 (3.0–5.0)	4.0 (3.0–5.0)	0.160
Gestational age (wk)	38.5 ± 1.7	39.0 ± 0.7	38.2 ± 0.2	0.016	1.26	0.98–1.62		0.07
Gestational age < 37 wks	10 (11.1%)	0 (0.0%)	10 (17.5%)	0.011	n/a
Male gender	64 (68.8%)	23 (67.6%)	41 (69.5%)	0.85
Weight (kg)	3.3 ± 0.5	3.4 ± 0.4	3.2 ± 0.6	0.150
Chromosome abnormality	11 (11.8%)	3 (8.8%)	8 (13.6%)	0.740
Preoperative mechanical ventilation	71 (76.3%)	28 (82.4%)	43 (72.9%)	0.450
Balloon atrial septostomy	42 (45.2%)	17 (50.0%)	25 (42.4%)	0.520
Preoperative Hb (g/dL)	14.1 ± 1.9	14.3 ± 1.6	13.9 ± 2.0	0.350
Preoperative platelet (per uL)	239.7 ± 66.4	245.4 (69.6)	236.4 ± 64.9	0.530
Preoperative lactate (mmol/L)	2.1 ± 0.7	2.1 ± 0.6	2.2 ± 0.8	0.510
Intraoperative and CPB data
CPB time (min)	180.9 ± 39.0	183.5 ± 40.9	179.4 ± 38.1	0.630
Cross-clamp time (min)	122.9 ± 23.8	130.5 ± 27.9	118.5 ± 20.1	0.018	0.98	0.96–1.00					0.02
Circulatory arrest	19 (20.4%)	5 (14.7%)	14 (23.7%)	0.420
Modified ultrafiltration	73 (78.5%)	30 (88.2%)	43 (72.9%)	0.120
CPB crystalloid (ml)	370.0 (250.0–520.0)	362.5 (292.5–520.0)	370.0 (245.0–540.0)	0.960
Surgical duration (hour)	5.8 ± 1.0	5.7 ± 0.9	5.8 ± 1.0	0.430
Delayed sternal closure	14 (15.1%)	0 (0.0%)	14 (23.7%)	0.002	n/a
Intraoperative lactate	1.7 ± 0.7	1.6 ± 0.7	1.8 ± 0.7	0.280
Post CPB lactate	3.2 ± 0.6	3.1 ± 0.6	3.2 ± 0.6	0.550
Tranexamic acid (mg/kg)	233.0 ± 40.8	234.4 ± 58.6	232.1 ± 26.0	0.790
Post CPB PRC (ml)	0.0 (0.0–0.0)	0.0 (0.0–0.0)	0.0 (0.0–0.0)	0.290
Post CPB salvaged blood (ml/kg)	22.8 (14.5–35.7)	17.1 (14.3–27.4)	29.6 (15.5–41.9)	0.007	1.04	1.01–1.08					0.01
post CPB FFP (ml/kg)	0.0 (0.0–0.0)	0.0 (0.0–0.0)	0.0 (0.0–0.0)	0.450
Post CPB cryoprecipitate (ml/kg)	7.8 (0.0–13.6)	5.2 (0.0–8.6)	8.3 (0.0–17.8)	0.014	1.09	1.02–1.16					0.01
Post CPB platelet (ml/kg)	19.8 (14.3–30.3)	18.5 (13.9–22.2)	22.5 (14.3–33.3)	0.043	1.04	1.01–1.08					0.03
Post-CPB transfused volume (ml/kg)	54.0 (36.2–74.2)	41.2 (34.4–55.0)	58.5 (42.3–86.2)	0.002	1.03	1.01	1.05	0.009
Post CPB epinephrine use	58 (62.4%)	20 (58.8%)	38 (64.4%)	0.660
Postoperative data, outcomes
Mechanical ventilation (hour)	66.8 (41.7–94.5)	33.5 (24.8–42.9)	90.3 (68.6–124.8)	<0.001
ICU length-of-stay (hour)	190.3 (141.6–249.8)	149.4 (118.2–183.6)	225.9 (174.7–323.7)	<0.001
Respiratory complications	10 (11.5%)	4 (13.3%)	6 (10.5%)	0.730
Postoperative infection (any)	7 (7.5%)	3 (8.8%)	4 (6.8%)	0.700
Pacemaker insertion	29 (31.2%)	10 (29.4%)	19 (32.2%)	0.820
Fluid balance POD0 (ml/kg)	3.2 (−9.7–18.4)	4.6 (−2.2–14.3)	2.6 (−12.7–20.3)	0.600
Fluid balance POD1 (ml/kg)	−5.2 (−44.8–30.6)	−30.5 (−50.3–12.0)	5.6 (−32.8–42.7)	0.024	1.00	1.00–1.01		0.03
Fluid balance POD2 (ml/kg)	−50.9 (−80.6-−28.2)	−44.2 (−75.3-−27.4)	−53.4 (−103.9-−31.9)	0.290
PRC transfusion on POD0	16 (17.2%)	5 (14.7%)	11 (18.6%)	0.780
PRC transfusion on POD1	15 (16.1%)	2 (5.9%)	13 (22.0%)	0.046	4.52	0.95–21.42		0.06
Postoperative to 24hr minimum Hb (g/dL)	13.4 (12.5–14.4)	14.0 (13.3–14.6)	13.0 (12.2–14.0)	0.004	0.67	0.49–0.92		0.01
24 to 48hr minimum (g/dL)	13.5 (11.7–14.7)	13.4 (11.7–14.4)	13.7 (11.6–14.9)	0.670
Thromboelastogram data
Angle
Rewarm	55.1 (50.1–62.2)	54.0 (49.8–63.3)	55.4 (50.4–61.6)	0.980
ICU admission	64.0 (49.9–71.1)	58.1 (49.5–70.5)	64.8 (53.2–71.4)	0.370
POD1	70.5 (66.7–74.2)	70.4 (62.7–72.6)	70.9 (66.9–75.1)	0.190
Reaction (R)
Rewarm	3.5 (2.8–5.1)	3.8 (2.8–5.1)	3.3 (2.8–5.0)	0.470
ICU admission	2.0 (1.3–4.6)	2.1 (1.3–6.7)	1.9 (1.2–4.5)	0.650
POD1	1.3 (1.1–1.8)	1.5 (1.2–5.2)	1.3 (1.0–1.7)	0.051
Maximum amplitude (MA)	45.7 (40.1–53.3)	47.0 (40.2–56.4)	45.4 (40.1–51.0)	0.300
Rewarm
ICU admission	52.2 (40.0–63.4)	45.1 (40.0–61.9)	54.8 (40.0–64.7)	0.250
POD1	66.5 (60.9–73.3)	62.6 (58.5–69.0)	67.6 (61.6–73.6)	0.060
Clotting (K)
Rewarm	6.9 (5.8–8.8)	6.6 (5.7–8.3)	7.2 (5.9–9.3)	0.360
ICU admission	6.3 (5.3–7.5)	5.9 (4.5–7.5)	6.7 (5.6–7.9)	0.200
POD1	6.5 (5.5–7.2)	5.9 (2.1–7.0)	6.7 (5.6–7.4)	0.097
Respiratory variables
P/F ratios
Intraoperative	55.9 (49.1–65.9)	53.8 (48.6–61.2)	57.7 (49.7–68.3)	0.087
Post CPB	198.0 (113.0–313.0)	188.0 (111.0–306.0)	216.0 (113.0–344.0)	0.630
ICU admission	255.0 (167.5–348.0)	197.0 (164.0–302.0)	271.5 (191.0–380.0)	0.029	1.00	1.00–1.01		0.03
6-hour	303.0 (251.0–368.5)	285.5 (252.0–380.0)	313.0 (245.0–360.0)	0.650
12-hour	335.0 (253.0–413.0)	341.5 (272.5–424.5)	303.0 (246.0–413.0)	0.410
24-hour	350.0 (271.0–410.0)	395.0 (310.0–420.0)	329.0 (255.5–386.5)	0.021	0.99	0.98–1.00		0.02
48-hour	347.5 (273.0–404.0)		347.5 (273.0–404.0)
P/F ratio <300
ICU admission	58 (63.0%)	25 (73.5%)	33 (56.9%)	0.120
6-hour	42 (47.7%)	19 (55.9%)	23 (42.6%)	0.280
12-hour	37 (43.5%)	11 (34.4%)	26 (49.1%)	0.260
24-hour	26 (34.7%)	2 (10.5%)	24 (42.9%)	0.012	6.38	1.34–30.27		0.02
48-hour	16 (33.3%)		16 (33.3%)
P/F ratio <200
ICU admission	33 (35.9%)	18 (52.9%)	15 (25.9%)	0.013	0.31	0.13–0.76		0.01
6-hour	10 (11.4%)	4 (11.8%)	6 (11.1%)	1.000
12-hour	9 (10.6%)	3 (9.4%)	6 (11.3%)	1.000
24-hour	6 (8.0%)	0 (0.0%)	6 (10.7%)	0.330
48-hour	2 (4.2%)		2 (4.2%)
Oxygenation index
ICU admission	3.6 (2.5–5.5)	4.1 (3.1–5.9)	3.3 (2.3–4.7)	0.067
6-hour	3.0 (2.2–3.7)	3.1 (2.1–3.5)	2.9 (2.3–3.7)	0.860
12-hour	2.3 (1.8–3.4)	2.1 (1.8–3.0)	2.6 (1.9–3.7)	0.200
24-hour	2.3 (2.0–3.2)	2.2 (1.9–2.5)	2.5 (2.0–3.4)	0.073
48-hour	2.6 (2.0–3.1)		2.6 (2.0–3.1)
Oxygen saturation index
ICU admission	5.0 (4.0–6.9)	4.7 (3.6–5.6)	5.1 (4.2–8.0)	0.088
6-hour	3.6 (3.2–4.4)	3.6 (3.2–4.1)	3.7 (3.2–4.4)	0.380
12-hour	3.2 (2.8–3.7)	3.2 (2.8–3.6)	3.2 (2.8–4.0)	0.160
24-hour	3.2 (2.8–3.6)	3.2 (2.5–3.6)	3.2 (2.8–3.6)	0.740
48-hour	3.2 (2.7–3.6)		3.2 (2.7–3.6)

Note: data are presented as n (%), mean ± SD, or median (IQR); OR, crude odds ratio

Abbreviations: CPB, cardiopulmonary bypass; PRC, packed red cells; FFP, fresh frozen plasma; POD, postoperative day; Hb, hemoglobin; P/F ratio, PaO₂/FiO₂ ratio

The median duration of postoperative mechanical ventilation for the entire cohort was 66.8 (41.7–94.5) hours. About 63% of patients required postoperative mechanical ventilation for 48 hours or more, with a median duration of 90.3 (68.6 to 124.8) hours. Postoperative infections and respiratory complications did not differ between the two groups (i.e., those with and without ≥ 48 hours of mechanical ventilation), and there was no in-hospital mortality in our study cohort.

Factors associated with the duration of mechanical ventilation

Patients with ≥ 48-hour mechanical ventilation had a higher incidence of prematurity, although the weight and age at surgery, preoperative mechanical ventilation, BAS, preoperative hemoglobin, and serum lactate were comparable with those without. Intraoperatively, these patients received a larger volume of salvaged red blood cells, cryoprecipitate, and platelet transfusions after CPB, Figure 2, and had a higher rate of delayed sternal closure. The transfusion volume of salvaged red blood cells was positively correlated with the volume of transfused platelets and cryoprecipitate (Pearson’s r = 0.60, p < 0.001, r = 0.45, p < 0.001, respectively).

Figure 2.

(A) Distribution of salvaged red blood cell transfusion volume. (B) Distribution of platelet transfusion volume.

Note: Y-axis, number of patients; X-axis, volume transfused (ml/kg).

At ICU admission, although most patients (63%) developed low P/F ratios (< 300), the ≥ 48-hour ventilation group had significantly higher P/F ratios (271.5 (191.0–380.0) vs. 197.0 (164.0–302.0), p = 0.029), Figure 3(A). However, at the postoperative 24-hour time, they later demonstrated significantly lower P/F ratios with a higher incidence of ALI (42.9 vs. 10.5%, p = 0.012). During ICU admission, they also had lower hemoglobin levels, received more packed red cell transfusions, and had more positive fluid balance. The fluid balance on postoperative day 1 was more positive in patients with ≥ 48 hours of mechanical ventilation (5.6 (−32.8–42.7) vs. −30.5 (−50.3−12.0) vs. p = 0.024) and was also a risk factor for prolonged mechanical ventilation (crude OR1.04 (1.01–1.08) p = 0.03). About 17% of patients received dexmedetomidine as part of ICU sedation. In those requiring ≥ 48-hour of ventilation, 13% were given dexmedetomidine, compared to 23% in those requiring < 48 hours (p = 0.26).

Figure 3.

(A) Comparisons of P/F ratios from ICU admission to 48 hours between patients with ≥ 48 hours and < 48 hours of mechanical ventilation. (B) P/F ratios from ICU admission to 48 hours between patients who developed acute lung injury (24-hour P/F ratio < 300) compared with those who did not.

Note: Data are presented as median, IQR; * represents a statistically significant interval difference (p-value < 0.05) between groups.

Abbreviation: P/F ratio, PaO₂/FiO₂ ratio, POD, postoperative day

The univariable and multivariable analyses are shown in Table 2. In the multivariable analysis, we did not find a significant association between the volume of salvaged red blood cells, platelets, or cryoprecipitate and prolonged mechanical ventilation. However, only the volume of platelet transfusion, but not salvaged red blood cells or cryoprecipitate, was associated with ALI (adjusted OR 1.04 (1.00–1.08, p = 0.042). To further predict the probability of ≥ 48-hour ventilation duration in dTGA/IVS patients, we developed a logistic regression model from relevant factors, demonstrating the relationship between the salvaged red blood cell volume, the 24-hour P/F ratio, and the duration of ventilation. This regression model had an acceptable area under the receiver operating characteristic curve (AUC-ROC) of 0.80 and had the Hosmer–Lemeshow probability of 0.52. From the model, it was demonstrated that for an equal volume of salvaged red blood cells received, patients with a 24-hour P/F ratio < 300 had a higher predicted probability of prolonged mechanical ventilation, Figure 4.

Table 2.

Univariable and multivariable analysis

Univariable analysis	Crude OR	(95%CI)	p-value
Gestational age (wk)	1.26	(0.98–1.62)	0.07
Cross-clamp time (min)	0.98	(0.96–1.00)	0.02*
Post-CPB salvaged blood (ml/kg)	1.04	(1.01–1.08)	0.01*
Post-CPB cryoprecipitate (ml/kg)	1.09	(1.02–1.16)	0.01*
Post-CPB platelet (ml/kg)	1.04	(1.01–1.08)	0.03*
Post-CPB transfused volume (ml/kg)	1.03	(1.01–1.05)	0.009*
Fluid balance POD1 (ml/kg)	1.00	(1.00–1.01)	0.03*
RBC transfusion POD1	4.52	(0.95–21.42)	0.06
Postoperative to 24hr minimum Hb	0.67	(0.49–0.92)	0.01*
P/F ratio ICU admission	1.00	(1.00–1.01)	0.03*
24-hour P/F ratio	0.99	(0.98–1.00)	0.02*
24-hour P/F ratio <300	6.38	(1.34–30.27)	0.02*
P/F ratio <200 at ICU admission	0.31	(0.13–0.76)	0.01*
Multivariable analysis	Adjusted OR	95%CI	p-value
Ventilation ≥ 48 hours †
24-hour P/F ratio < 300	3.49	0.81–19.73	0.157
Post-CPB salvaged blood (ml/kg)	1.04	0.98–1.10	0.148
Post-CPB platelet (ml/kg)	1.00	0.94–1.08	0.805
Post-CPB cryoprecipitate (ml/kg)	1.06	0.96–1.17	0.256
POD1 fluid balance	1.01	0.99–1.02	0.069
Gestational age (wk)	0.55	0.62–0.99	0.05
Acute lung injury †
Post CPB salvaged blood (ml/kg)	1.04	0.97–1.03	0.971
Post CPB platelet (ml/kg)	1.04	1.00–1.08	0.042*
Post CPB cryoprecipitate (ml/kg)	0.96	0.91–1.01	0.150
Logistic regression model for ventilation ≥ 48 hours
24-hour P/F ratio < 300	6.33	1.26–32.62	0.025*
Post CPB salvaged blood (ml/kg)	1.04	1.00–1.09	0.033*
POD1 fluid balance	1.01	0.99–1.01	0.152
Area under ROC curve = 0.795

^*Represents statistical significance

^†Post-CPB transfused volume was omitted because of collinearity.

Abbreviations: CPB, cardiopulmonary bypass; PRC, packed red cells; POD, postoperative day; Hb, hemoglobin; P/F ratio, PaO₂/FiO₂ ratio

Figure 4.

Predicted probability of prolonged mechanical ventilation based on the interaction between the volume of salvaged red blood cells received and 24-hour P/F ratios.

Note: date presented as predicted probability and 95% confidence interval

Abbreviation: P/F ratio, PaO₂/FiO₂ ratio

The cut-off value of salvaged red blood cell volume at the maximum Youden J-index was 32.5 ml/kg, (crude OR of 6.11 (1.90–19.60), p = 0.002), and platelets at 23.5 ml/kg, (crude OR 4.51 (1.63–12.50), p = 0.004). Patients who received > 32.5 ml/kg salvaged red blood cells had significantly longer mechanical ventilation duration (92.9 (68.56 – 125.8) vs. 48.1 (35.33 – 73.58) hours, p <0.001. Similarly, patients who received > 23.5 ml/kg platelets had 81.9 (52.9–117.8) hours of mechanical ventilation compared to 49.0 (30.5–90.3) hours (p = 0.004).

Correlation between OI and OSI

OI and OSI were moderately correlated at every observed interval in this patient cohort. Additionally, P/F ratios derived from the arterial PaO₂ were inversely correlated with OSI derived from peripheral SpO₂ readings; however, only weak correlations were observed.

Overall, the OI and OSI values were highest at ICU admission and decreased over time in the postoperative period. Unlike the P/F ratios, OI and OSI did not differ statistically between patients with and without ≥48 hours of mechanical ventilation, as shown in Table 3. However, patients who developed ALI had significantly higher OSI values observed at ICU admission and 12 hours postoperatively.

Table 3.

Brixia scores and ventilation duration

Brixia scores	Total (n = 93)	ventilation < 48 hours (n = 34)	ventilation ≥ 48 hours (n = 59)	p-value
Preoperative	11.0 (9.0–12.0)	10.0 (9.0–12.0)	11.0 (9.0–13.0)	0.350
Postoperative	12.0 (10.0–13.0)	11.0 (10.0–13.0)	12.0 (9.0–13.0)	0.970
POD1	12.0 (10.0–14.0)	12.0 (9.0–13.0)	12.0 (10.0–14.0)	0.180
Post – preop difference	1.0 (−2.0–3.0)	1.0 (−1.0–3.0)	0.0 (−2.0–3.0)	0.320
POD1 – postop difference	0.0 (−1.0–3.0)	0.0 (−2.0–2.0)	1.0 (0.0–3.0)	0.089
POD1 – preop difference	1.0 (−2.0–4.0)	1.0 (−2.0–4.0)	1.0 (−2.0–4.0)	0.740

Note: Data are presented as median (IQR)

Brixia scores, mechanical ventilation duration, and acute lung injury

Among the 75 patients for whom the 24-hour P/F ratio data were available, 26 patients (34.7%) developed ALI. Figure 3(B) compares the P/F ratios between patients who developed ALI and those who did not. The P/F ratios were comparable at the time of ICU admission but were significantly lower at the 12, 24, and 48-hour time points in patients with ALI. Furthermore, an increase in the 12-hour PF ratio was associated with a reduced probability of ALI after adjusting for other factors (adjusted OR 0.99 (0.98–.99), p = 0.002).

Of the 281 radiographs assessed for Brixia scores, the ICC showed a satisfactory agreement among assessors (ICC = 0.85 (95%CI 0.81–0.88), p = <0.001). Overall, patients had a considerably high median baseline score (11.0 (9.0–12.0)), and both upper lung zones had higher scores than the rest of the lungs, Figure 5(A). The median postoperative Brixia scores increased slightly from the preoperative baseline but did not differ between those with and without ≥ 48 hours of postoperative ventilation, Table 3, Figure 5(B). We further categorized patients into three groups based on their Brixia scores (0–6, 7–12, and 13–18) and found that patients with POD1 scores of 13–18 had significantly longer ICU length-of-stay compared to scores 7–12 (Figure 6(B)). Additionally, post-CPB transfused volume (platelet, cryoprecipitate, and salvaged red cells) differed among the three groups, with scores of 13–18 patients receiving the largest transfusion volume (Figure 7).

Figure 5.

(A) Comparisons of Brixia scores from each lung zone at preoperative baseline, postoperative ICU admission, and POD1, (B) Comparisons of Brixia scores between patients with ≥ 48 and < 48 hours ventilation, (C) Comparison of Brixia scores between patients with and without ALI, (D) Comparison of Brixia scores between patients with and without 24-hour P/F ratio < 200

Note: (A) Data are presented as mean, SD; (B) – (D) Data are presented as median, IQR;

* represents a statistically significant interval difference (p-value < 0.05) between groups

Abbreviation: P/F ratio, PaO₂/FiO₂ ratio, POD, postoperative day; LUZ, left upper lung zone; LMZ, left middle lung zone; LLZ, left lower lung zone; RUZ, right upper lung zone; RMZ, right middle lung zone; RLZ, right lower lung zone

Figure 6.

(A) Postoperative Brixia scores, mechanical ventilation duration, and ICU length-of-stay, (B) POD1 Brixia scores, mechanical ventilation duration, and ICU length-of-stay (LOS).

Note: Data are presented as median, IQR; * represents p-value = 0.002 (two-way ANOVA).

Figure 7.

(A) Postoperative Brixia scores and volume of post-CPB transfusions. Note: Data are presented as median, IQR; * represents p-value = 0.047; ** p-value = 0.002; *** p-value = 0.001 (two-way ANOVA with Tukey’s multiple comparisons). (B) POD1 Brixia scores and post-CPB transfusions. Note: Data are presented as median, IQR; * represents p-value < 0.001 (two-way ANOVA with Bonferroni’s correction).

Increased POD1 scores were associated with an increased risk of ALI (crude OR 1.22 (1.02–1.50), p = 0.026), Table 4, Figure 5(C). Both postoperative ICU and POD1 scores also increased significantly in patients who had 24-hour P/F ratios < 200, Table 4, Figure 5(D). Lastly, an inverse correlation existed between the 24-hour P/F ratios and the POD1 Brixia scores (Pearson’s r = −0.33, p = 0.004), Table 5.

Table 4.

Brixia score and severities of acute lung injury at preoperative baseline, postoperative at ICU admission, and postoperative day 1 (POD1)

Brixia scores	P/F ratio ≥ 300 (n = 49)	P/F ratio < 300 (n = 26)	p-value	P/F ratio ≥ 200 (n = 69)	P/F ratio < 200 (n = 6)	p-value
Preoperative	11.0 (8.5–12.5)	11.0 (9.0–13.0)	0.94	11.0 (9.0–13.0)	11.5 (11.0–12.0)	0.71
Postoperative	12.0 (10.0–13.0)	12.0 (10.0–13.0)	0.50	12.0 (10.0–13.0)	13.5 (12.0–16.0)	0.018*
POD 1	12 (9.0–14.0)	13.0 (12.6–16.0)	0.027*	12.0 (10.0–14.0)	15.0 (13.0–16.0)	0.046*

^*Represents statistical significance

Note: Data are presented as median (IQR)

Table 5.

Correlation analysis

Correlation analysis	Pearson’s coefficient	p-value
OI and OSI (ICU admission)	0.42	<0.001*
OI and OSI (6-hour)	0.65	<0.001*
OI and OSI (12-hour)	0.65	<0.001*
OI and OSI (24-hour)	0.67	<0.001*
OSI ICU and P/F ratio_ICU	−0.30	0.01*
OSI 6h and P/F ratio 6h	−0.45	<0.001*
OSI 12h and P/F ratio 12h	−0.30	0.008*
OSI 24h and P/F ratio 24h	−0.33	0.01*
Preoperative Brixia
P/F ratio intraoperative	0.14	0.204
P/F ratio post CPB	0.02	0.876
ICU admission Brixia
P/F ratio post CPB	−0.03	0.779
P/F ratio ICU admission	0.15	0.159
P/F ratio 6 hour	−0.10	0.366
P/F ratio 12 hour	−0.14	0.202
P/F ratio 24 hour	−0.16	0.164
POD 1 Brixia
P/F ratio first ICU	0.06	0.564
P/F ratio 6 hour	0.02	0.855
P/F ratio 12 hour	−0.15	0.161
P/F ratio 24 hour	−0.33	0.004*
P/F ratio 48 hour	−0.09	0.560

^*Represents statistical significance

Abbreviation: CPB, cardiopulmonary bypass; OI, oxygenation index; OSI, oxygenation saturation index; P/F ratio, PaO₂/FiO₂ ratio

Discussion

In this retrospective single-center study involving neonates diagnosed with dTGA/IVS who underwent ASO, we found: (i) high incidence of acute lung injury and ≥ 48-hour mechanical ventilation; (ii) perioperative factors that are associated with mechanical ventilation duration and acute lung injury; and (iii) utility of Brixia scores as a predictor of acute lung injury in neonatal cardiac surgeries.

In recent years, the duration of mechanical ventilation and utilization of enhanced recovery/fast-track protocols have emerged as important factors in improving outcomes and reducing healthcare costs ^{(7, 10, 11)}. With that notion, our study evaluated the incidence and risk factors associated with postoperative ventilation in neonates undergoing ASO, considered as a moderate- to high-risk cardiac surgery ⁽¹²⁾. We reported that approximately 60% of patients in our cohort required more than 48 hours of postoperative mechanical ventilation. This specific threshold was selected based on the characteristics of our patient group, institutional practice, and previous studies of similar surgical procedures. Previous studies reported varying durations of mechanical ventilation after ASO; interestingly, they ranged from on-the-table/early extubation to several days ^(13–19). Although immediate/early extubation has been successful after ASO in previous reports, our institution has not utilized such practice. Out of 93, only 8 patients in our cohort were extubated in less than 24 hours (median 23.0 (21.8–23.7)). It is challenging to compare one specific outcome (e.g., duration of ventilation) among institutions as there are various contributing patient and perioperative factors. Early extubation is encouraged in the European Association for Cardio-Thoracic Surgery (EACTS) dTGA/IVS guideline ⁽²⁰⁾. However, high-quality studies, although limited feasibility due to the nature of the cases, are needed to determine the benefits, risks, and optimal practice strategies to achieve the recommended target. Nonetheless, based on our institution’s data and by recognizing the need to enhance outcomes and to align with current and future practice trends in the field, we determined the 48-hour threshold as the reasonable outcome in this study.

Predictors of mechanical ventilation duration after neonatal cardiac surgeries have been previously reported in cohorts comprising a wide range of congenital heart diseases ^{(6, 21–25)}; these include prematurity, longer CPB time, surgical complexity, amount of transfused blood products, acute kidney injury, and inotropes use. Prematurity was reported as a preoperative risk factor for delayed extubation after ASO ⁽²⁴⁾; we found that none of the patients with <48 hours of ventilation were premature compared to 17% of their counterparts. The contribution of the immature lungs and increased pulmonary blood flow from intra- (PFO/ASD) and extra-cardiac shunts (PDA) in TGA might explain the findings. We also found that most patients had a brief period of mechanical ventilation preoperatively due to either needing cardiorespiratory support during BAS or apnea due to prostaglandin use. Patients were often extubated prior to surgery. Mechanical ventilation, although for a brief period, could potentially result in lung injury before surgery and lead to delayed lung function recovery ^{(5, 26)}.

It should be noted that even though the ≥ 48 hours ventilation group had significantly higher P/F ratios in the early postoperative period, they later became lower than the < 48 hours group. This is likely explained by the higher incidence of delayed sternal closure in this group (23.7% vs. 0%). Patients with delayed sternal closure had higher P/F ratios at ICU admission and at 6 hours (339.0 (275.0–414.0) vs. 239.5 (164.0–330.0), p = 0.002, 381.5 (285.0–513.0) vs. 293.5 (250.0–350.0), p =0.031, respectively). They also exhibited longer mechanical ventilation duration (131.4 (99.1–187.7) vs. 49.3 (37.1–90.2), p <0.001) with the median duration from ICU admission to chest closure procedure of 64.7 (min 18.85, max 108.18) hours. It is not uncommon to replace coagulation factors and platelets during the post-CPB period in neonatal cardiac surgery. In our cohort, similar to a recent study ⁽²⁷⁾, post-CPB platelet and cryoprecipitate transfusions were associated with prolonged mechanical ventilation. Furthermore, although salvaged blood is considered safe in pediatric cardiac surgeries ^(28–30) and routinely transfused at our institution, we found that larger transfused volume was associated with a longer ventilation duration. However, after adjusting for other variables, the salvaged blood or blood component volume was not associated with prolonged mechanical ventilation. Only platelet volume was associated with ALI, which was supported by a previous study in a pediatric cardiac surgery cohort ⁽²¹⁾. It is also worth mentioning that only few studies reported associations between transfused salvaged red blood cells volume and clinical outcomes in pediatric cardiac surgeries. Our median transfused volume was 22 (14.5–35.7) ml/kg for the entire cohort and 29 (15.5–41.9) ml/kg in the ≥ 48-hour ventilation group. In contrast, previous randomized trials involved pediatric cardiac patients who received an average volume of salvaged red blood cells up to 38 ml/kg ^{(28, 30)}. In these trials, there was no association with prolonged ventilation. However, these studies included various patient diagnoses and ages, with limited emphasis on neonates. Further studies to determine the effects of salvaged blood and postoperative adverse outcomes in neonates are still needed. Although platelet transfusions are common in neonatal cardiac surgery, similar to the volume of salvaged red cells, the associations between platelet volume and clinical outcomes are not commonly reported. This warrants further studies to investigate whether the adverse outcomes resulted from the inflammatory response after transfusions or simply due to the excess alveolar and/or interstitial fluid. Some notable concerns for salvaged blood are the quality of the washed red blood cells and the increase in free hemoglobin due to hemolysis. Intraoperative blood processing increases red blood cell concentration and removes inflammatory mediators but adversely increases red blood cell fragility with a higher tendency to hemolyze compared to non-washed red blood cells ^{(31, 32)}. Elevated free hemoglobin concentrations after pediatric cardiac surgery result in adverse outcomes, including infection, thrombosis, and death ⁽³³⁾. Furthermore, transfusing adult platelets to the very different physiological environment of neonates may cause a developmental mismatch, leading to adverse events ⁽³⁴⁾. Platelets interact with various immune cells. Adult platelets, but not neonatal platelets, promote inflammatory responses through monocyte migration and neutrophil extracellular traps (NET) formation. These responses are hypothesized to contribute to transfusion-related organ injury in neonates receiving adult platelets, with the lungs being one of the affected organs ^{(34, 35)}.

We also found that the ≥ 48-hour ventilation group had lower hemoglobin levels, higher positive fluid balance, and a higher incidence of blood transfusions after surgery. This may indirectly indicate that these patients experienced more blood loss, which resulted in more transfusions and worsened outcomes. Our findings were consistent with a large neonatal cardiac surgery cohort ⁽³⁶⁾, which found that patients with >10% weight-based fluid balance on postoperative day 2 had significantly longer postoperative ventilatory support and major postoperative complications. Another recent study on 643 pediatric cardiac surgery patients found that post-CPB red blood cells, cryoprecipitate, and platelet transfusions were associated with increased duration of mechanical ventilation, renal impairment, and mortality ⁽³⁷⁾. Thus, further emphasis should be placed on reducing transfusions through meticulous surgical hemostasis, CPB management, and optimal goal-directed transfusions. Interestingly, a recent study reported a transfusion-free rate of 52% for ASO in 100 consecutive neonates with dTGA. The combination of thromboelastometry and protocol-driven perioperative practice was employed successfully, which resulted in a low transfusion rate while maintaining optimal patient outcomes in this study. However, mechanical ventilation duration and other early postoperative outcomes were not reported ⁽³⁸⁾.

Dexmedetomidine, with known anti-inflammatory and organ-protective effects ⁽³⁹⁾, is commonly used in cardiac anesthesia and postoperative sedation. It has been shown to shorten ventilation duration after pediatric cardiac surgery in clinical trials and is considered a part of early extubation protocols ^{(40, 41)}. We found that the proportion of patients who received dexmedetomidine sedation was lower in the prolonged ventilation group, but the difference did not reach statistical significance.

Intra-cardiac mixing in congenital heart disease can make the diagnosis of lung injury with P/F ratio challenging. To better characterize postoperative lung injury and mechanical ventilation duration, we evaluated a novel chest radiographic scoring system, specifically the Brixia score ^{(8, 9)}. During the COVID-19 pandemic, Brixia score was developed for risk stratification of COVID-19 patients and has later been adopted in various settings ⁽⁴²⁾. These scores have been shown to correlate well with patient outcomes and are practical in routine patient care ^(42–44). We demonstrated that non-radiologists could report the scores with an acceptable interobserver agreement after training by a radiologist. We found that patients who later developed ALI had increased Brixia scores on the postoperative ICU admission and the first postoperative day. Following ASO, it is likely that normal circulation has returned, and intracardiac mixing is negligible. Thus, our findings suggest correlations between the Brixia score and alveolar gas exchange (POD1 Brixia and 24-hour P/F ratio had Pearson’s r = −0.3268, p =0.004). However, given the exploratory nature of our data, further studies are needed to evaluate its usefulness in characterizing respiratory outcomes, especially in those left with mixing of oxygenated and non-oxygenated blood after surgery. In the meantime, our observations suggested that increased Brixia scores in the early postoperative period can be used to inform providers of the increased likelihood of adverse outcomes after ASO (ALI and increased ICU length-of-stay) and warrant proactive measures to facilitate early recovery. It will also be interesting to compare the radiographic findings from other institutions where immediate/early extubation is commonly practiced. And we should emphasize that although the scoring method is straightforward, a brief training period is useful to ensure standardized readings.

Our study had the following limitations. Firstly, our data are from a single institution with a relatively small sample size. Practice variations and potential improvements in patient care over the eight-year study period should also be considered. Secondly, our analysis did not include coronary anatomy and other surgical factors. Although this is a homogeneous dTGA/IVS study group, these factors may contribute to postoperative outcomes. Thirdly, we did not analyze ICU sedation management and other provider-related factors that influenced the decision to extubate the patients. Finally, the presence of chest tubes and other devices may limit interpretations of the Brixia scores after cardiac surgeries.

Conclusions

This study concludes that most patients had ≥ 48-hour mechanical ventilation after ASO. The etiologies of postoperative lung injury and delayed recovery are likely multifactorial, with platelet transfusion being a potential modifiable risk factor. Thus, perioperative transfusion should be tailored to achieve satisfactory coagulation while maintaining optimal fluid balance. Further, we showed that Brixia score, a straightforward chest radiograph scoring system, can be effective in quantifying postoperative lung injury after neonatal cardiac surgery, although additional study is needed to determine its diagnostic clinical utility.