The outcomes of interest included the HRQOL assessed using the Pediatric Quality of Life Inventory (PedsQL) for children aged at least 2 years and the Pediatric Quality of Life Inventory Infant Scale for children aged < 2 years. The Generic Core Scales Version 4.0 PedsQL includes 23 items. The quality-of-life score was evaluated separately according to each parent and child. The questionnaire for the parent-proxy report was used for children aged between 1 and 18 years whereas the questionnaire for child self-report was used for children aged between 5 and 18 years. Both questionnaires were the same and included questions concerning physical function (8 items), emotional function (5 items), social function (5 items), and school function (5 items). Each question was answered on a scale of 0–4 (0 = never a problem [100 points]; 1 = almost never a problem [75 points]; 2 = sometimes a problem [50 points], 3 = often a problem [25 points], and 4 = almost always a problem [0 point]). These scales were subsequently converted into a total score of 0–100. Physical health referred to physical function (P1/C1). Psychosocial health was the combination of emotional function (P3/C3), social function (P4/C4), and school function (P6/C6). A high score indicated a good quality of life. The PedsQL Infant Scale consisted of a questionnaire for the parent only (assessed in infants aged < 2 years). The questionnaire was evaluated using five topics to evaluate the physical function (P1), physical symptoms (P2), emotional function (P3), social function (P4), and cognitive function (P5). Physical health was the combination of physical function (P1) and physical symptoms (P2). Psychosocial health was the combination of emotional function (P3), social function (P4), and cognitive function (P5). This questionnaire has been validated by Varni et al. [8] in children and passed the reliability test. Its translation and cultural adaptation into Thai were performed by the Mapi Research Institute (www.mapi-trust.org) according to the international guidelines.

Total repair/progression to the next stage was defined as having undergone the Glenn/Fontan procedure, total correction, Rastelli procedure, or arterial switch procedure depending on the cardiac diagnosis. The palliative shunt group was defined as children who received an MBTS as a definite treatment or who were waiting for the next stage of repair.

The variables collected were categorized into patient-related factors, anesthesia-related factors, and surgery-related factors. Patient-related factors included the age during the MBTS procedure, age at the time of the assessment of the quality of life (QOL), weight during the operation, preterm (gestational age < 37 weeks), cardiac diagnosis (categorized as Tetralogy of Fallot [TOF], univentricular heart [UVH], pulmonary atresia with ventricular septal defect [PA/VSD], and transposition of great arteries [TGA]), non-cardiac abnormalities (categorized as a syndrome, such as dextrocardia, asplenia, delayed development, or chronic lung disease), history of preoperative hypoxic spells, and socioeconomic status. Socioeconomic status was assessed using the Hollingshead two factors index of social position (level of education and parents’ income) [9]. Anesthesia-related factors included the American Society of Anesthesiologists (ASA) classification, intraoperative hypoxemia (arterial oxygen saturation [SpO₂] < 75%), and bradycardia (heart rate below the baseline according to the child’s age, which required atropine). Surgery-related factors included differential oxygen saturation, which was defined as the difference in the oxygen saturation between preoperative and postoperative assessments, the ratio of the shunt size to the body weight, duration of postoperative mechanical ventilator use, duration of stay in the intensive care unit (ICU), re-operative thoracotomy at admission, postoperative complications (shunt thrombosis, congestive heart failure, sepsis, pneumonia, and coagulopathy), history of shunt revision at admission, total repair status, number of MBTS procedures and their complications, length of hospital stay, and readmission within 30 days. In the total repair subgroup, age at total repair and surgery-related factors, such as type of surgery, intraoperative complications, and intraoperative blood loss were included in the analyses.

Data record forms were created, and information was extracted from the electronic medical records. They were then double entered into a database using EpiData software version 3.1 (The EpiData Association, Pascal, Denmark). R software version 4.1.1 (R Core Team, Vienna, Austria) was used to analyze the data. Continuous variables are presented as the median and interquartile range while categorical variables were compared using Fisher’s exact test or Chi-square test where appropriate. Continuous variables were compared using the Wilcoxon rank sum test. A univariate analysis was performed to assess the association between the risk factors and outcomes. Any factor with a p-value < 0.2 in the univariate analysis was included in the multivariate linear regression analysis. The subgroup analysis was performed similarly in patients who underwent total repair. Age at the HRQOL assessment was included in the initial multivariate model regardless of its p-value in the univariate analysis.

The standardized HRQOL scores were compared using both parent-proxy reports and child self-reports. However, we considered significant predictors based on the parent assessment only since the proportion of missing scores was lower in the parent-proxy reports. Appendix C shows potential predictors of physical and psychosocial health evaluated by parents only. Appendix D summarizes the univariate linear regression results assessing factors associated with physical health evaluated by the parents and children. Sixteen variables in the parent-proxy report (history of prematurity, delayed development, chronic lung disease, ASA classification, intraoperative anesthetic complication, ratio of shunt size to body weight, duration of mechanical ventilator, duration of stay in the intensive care unit, re-operative during admission, postoperative sepsis/coagulopathy, history of shunt revision in admission, total number of MBTS procedures and other procedures related to the patient’s condition, length of hospital stay, readmission and received total repair) and five variables (age at QOL test, history of prematurity, delayed development, chronic lung disease, and received total repair) in the child self-report with p-value < 0.2 were included in the initial multivariate linear regression analysis. After adjusting for covariates, the multivariate model used for both parent-proxy reports and child self-reports is summarized in Table 3. Significant predictors of lower physical standardized HRQOL score included chronic lung disease (p = 0.035), longer duration on a mechanical ventilator (p = 0.02), and longer duration of hospital stay (p < 0.001). Total repair was a predictor of improvements in the physical standardized HRQOL scores in both parent-proxy reports (p = 0.007) and child self-reports (p < 0.001).

Appendix E summarizes the univariate linear regression analysis for factors associated with the psychosocial health evaluated by the parents and children. Sixteen variables in the parent-proxy report (age at MBTS, age at QOL test, delayed development, chronic lung disease, ratio of shunt size to body weight, duration of mechanical ventilator, duration of stay in the intensive care unit, re-operative during admission, postoperative congestive heart failure sepsis/coagulopathy, history of shunt revision in admission, total number of MBTS procedures and other procedures related to the patient’s condition, length of hospital stay, readmission and received total repair) and five variables in the child self-report (age at QOL test, history of prematurity, delayed development, chronic lung disease, and received total repair) with p < 0.2 were included in the initial multivariate linear regression. The multivariate model used for both parent-proxy reports and child self-reports is summarized in Table 4. Significant predictors of lower psychosocial standardized HRQOL score included older age at QOL assessment (p = 0.012), delayed development (p < 0.001), and chronic lung disease (p = 0.016). Younger age at the time of the MBTS procedure was a predictor of improvements in the psychosocial standardized HRQOL scores (p = 0.027).

Among the 110 children who underwent total repair, 74 underwent the operation at our hospital and 36 were transferred to other hospitals. We recorded the surgery-related and anesthesia-related variables according to the total repair. Appendix F shows the univariate linear regression analysis for factors associated with the physical health and psychosocial health evaluated by the parents. Table 5 summarizes the multivariate linear regression analysis for factors associated with the physical health and psychosocial evaluated by the parents. Predictors of lower physical standardized HRQOL score included age 5–7 years (vs. 13–18 years) at QOL assessment (p = 0.023 by t-test), history of sepsis (p = 0.010), and children who underwent the Fontan procedure/total correction (vs. Glenn procedure) (p = 0.021/p = 0.046 by t-test). Higher socioeconomic status and higher postoperative oxygen saturation were predictors of improvements in the physical standardized HRQOL score (p = 0.039 and p = 0.021, respectively). Predictors of lower psychosocial standardized HRQOL score included age 5–7 years (vs. 13–18 years) at QOL assessment (p = 0.006 by t-test), UVH (vs. TOF) (p = 0.037 by t-test), total correction (vs. Glenn procedure) (p = 0.030 by t-test); and intraoperative complications (p = 0.048). Predictor of improvement in psychosocial health standardized HRQOL score was higher socioeconomic status (p = 0.048). Higher socioeconomic status was a common predictor of improvements in both physical and psychosocial standardized HRQOL scores.

Both univariate and multivariate linear regression analysis revealed that total repair/progression to the next stage (vs. palliative shunt) was one of the important predictors of improvements in physical HRQOL (p = 0.007) but not psychosocial HRQOL scores in the parent-proxy reports (p = 0.083). Nevertheless, approximately 80% of the children survived total repair at our hospital, while the others died from intraoperative and postoperative complications. The children who survived total repair can experience improvements in physical functions, such as walking, running, and doing household chores without developing dyspnea. However, the child self-reports in the total repair group revealed improvements in social function (p = 0.031) as well as the overall psychosocial HRQOL score in the multivariate analysis (p < 0.001). Therefore, physical HRQOL can be improved with total repair; when physical health improves, it may promote better emotional as well as social and school functions [12,13]. Consequently, psychosocial HRQOL may eventually improve following total repair. However, children with a complex cardiac condition might still have poorer psychosocial HRQOL regardless of their diagnosis or type of operation compared to simple conditions [14]. Uzark et al. [5] reported that patients with moderate cardiac disease who underwent curative surgery had higher physical and psychosocial HRQOL scores than those with severe cardiac disease and uncorrectable defects (89 vs. 76 and 82 vs. 74, respectively).

We observed that younger age at the time of the MBTS procedure can help improve the psychosocial HRQOL score. Since all children had cyanotic heart disease in this study, good left-to-right shunt flow could improve the child’s oxygenation and promote fair living in terms of emotional and social functions but not improve complete physical function if total repair is not performed. This may be why younger age at the time of the MBTS procedure could improve the psychosocial but not physical HRQOL score. Moreover, Wang et al. [15] reported that adolescents’ knowledge of their cardiac condition incorporated with an increase in the sense of well-being improved the psychosocial quality of life of the affected individual. In our routine preoperative care, our disciplinary team, especially the pediatric cardiologist, take a reasonable amount of time to explain to the patient their cardiac abnormality condition and discusses the possible morbidities and prognosis with the parent or adolescents themselves at the pediatric outpatient clinic at least 1 month before surgery or at the pediatric ward at least one day before surgery.

In the total repair subgroup, UVH was associated with worse psychosocial HRQOL scores compared with TOF (p = 0.037 using t-test). Previous studies have shown that a diagnosis other than TOF is the principal risk factor for early mortality [16,17], which could be associated with poorer HRQOL if the patients survive. After total repair, the HRQOL score in these children could be similar [11] or lower than those of healthy peers [18]. Higher HRQOL scores following corrective surgery can be due to several factors, such as residual heart defect/chronic disease [19,20], socioeconomic reasons [20,21], and familial relationships [22].

Higher socioeconomic status was not an independent predictor of our primary outcome but was one of the predictors of improvements in both physical and psychosocial HRQOL scores following total repair. Landolt et al. [22] reported that socioeconomic status was not associated with physical and psychosocial HRQOL scores following open-heart surgery, while adverse familial relationships negatively affected the parent-related HRQOL scores. Socioeconomic status includes the family’s income and parents’ education. Families with higher income or higher education would have better capability to take care of sick children than those with poorer socioeconomic status. Since some of the children with UVH (29%) received only the Glenn shunt and not definite Fontan’s procedure, they may still experience cyanotic manifestations. Therefore, parental support in terms of both financial and psychosocial aspects can improve the child’s physical and psychosocial health, which has been previously reported [20,21].

Common risk factors of poor physical and psychosocial HRQOL scores included chronic lung disease and delayed development. Patients with chronic lung disease, such as bronchopulmonary dysplasia, can develop worsening of cyanotic symptoms in conjunction with postoperative shunt problems, which can result in poor physical and psychosocial HRQOL scores. Similarly, children with cyanotic heart defects and intellectual developmental problems can also develop worsening of both physical and psychosocial functions due to shunt malfunction. Both diseases are classified as chronic diseases, which can affect the overall HRQOL in pediatric patients. If children with CHD also have other chronic diseases, they will have a poorer QOL [19]. Such children require a sense of coherence or a sense of appreciation for life and expectations earlier than healthy children, which may affect their HRQOL [23].

Longer duration of mechanical ventilation and longer duration of hospital stay were risk factors for poor physical HRQOL. The duration of mechanical ventilation and the duration of hospital stay may represent the severity of cyanotic heart disease, its treatments, and postoperative MBTS complications, such as shunt thrombosis, coagulopathy, and sepsis. Our previous study has shown that postoperative complications following MBTS procedure, such as shunt thrombosis (5-fold), bleeding (4.5-fold), and renal failure (4.1-fold), were strong predictors of time-to-death <90 days postoperatively [7]. Therefore, children who survive such complications would require prolonged mechanical ventilation, which results in a longer duration of hospital stay. Landolt et al. [22] also reported that prolonged duration of hospital stay was related to impaired HRQOL in children following open-heart surgery based on parent-proxy reports, which was consistent with the findings of our study. The poor physical health might also be related to complications due to shunt problems or partial recovery from infections after prolonged mechanical ventilation and admission. Secondary hypothyroidism in the immediate postoperative period has also been reported in children who underwent cardiac surgery. Furthermore, Bettendorf et al. [24] showed that lower thyroid hormone concentrations were significantly associated with prolonged mechanical ventilation and length of hospital stay. Therefore, the effects of thyroid function on long-term cognitive function during the postoperative period after congenital heart surgery remain to be investigated.

Older age at QOL assessment was also one of the risk factors of poor physical HRQOL in child self-reports and psychological HRQOL in both child and parent reports, irrespective of the status of total repair (Tables 3 and 4). However, older age at QOL assessment in the total repair subgroup was associated with a higher psychological HRQOL score. Therefore, irrespective of total repair status, in children who receive only a palliative shunt (MBTS), increasing waiting time for total repair is associated with worsening physical and psychosocial functions at the time of QOL assessment. In contrast, in children who undergo total repair, the physical and psychosocial functions will improve with time. Longer time after surgery was associated with older age at QOL assessment. We observed that adolescents (aged 13–18 years) had much better QOL scores than children aged 5–7 years (Table 5).

We found that total repair/progression to the next stage can improve physical HRQOL scores, whereas higher psychological HRQOL scores was associated with younger age during MBTS surgery. In total repair subgroup, higher socioeconomic status and older age at QOL assessment were associated with higher physical and psychological HRQOL scores. Although total repair/progression to the next stage can improve physical HRQOL scores in children with cyanotic heart disease, the overall in-hospital mortality at our hospital following Glenn’s procedure/Fontan procedure/total repair was relatively high (20%) compared with those in other studies, which varied from 2%–12% depending on the severity, co-morbidities and experience of the surgeon [25–30]. Therefore, patient selection for total repair is critical for minimizing postoperative mortality. Competing risk analysis in our previous study revealed that the proportion of children who underwent Glenn’s procedure/total repair increased at approximately 5 months of age; however, only 31.5% had undergone Glenn’s procedure/total repair at 5 years after the MBTS procedure due to various limitations [7]. Therefore, we propose the following two solutions to improve HRQOL. First, Glenn’s procedure/total repair should be considered in young children with chronic conditions that affect HRQOL, such as delayed development and chronic lung disease. Second, the MBTS procedure should be performed as soon as possible after the diagnosis of UVH/biventricular heart or other cyanotic heart conditions to prevent worsening of symptoms and, perhaps, improve HRQOL while awaiting definite repair. Since individual HRQOL courses are unpredictable regardless of the surgery, repeated HRQOL assessments over time might benefit children with congenital heart disease [31].

There are a number of strengths of this study. First, we used a valid Thai translated questionnaire and only one pediatric psychologist for the assessment of HRQOL, which ensured the validity and reliability of the results. Second, we used multivariate linear regression analysis to adjust for potential independent predictors. Third, the results in the total repair subgroup supported the primary findings, which may also reflect the accuracy of the study. Despite these strengths, a few limitations need mentioning. First, the duration between surgery and HRQOL evaluation in each child varied (1–11 years); for example, only five patients were of 1–2 years of age, which might have affected the HRQOL score for that age group. Second, one-third of the patients underwent total repair at other institutions, which resulted in a small sample and, possible oversight related to some predictors of HRQOL.