Population-based retrospective cohort study on community-acquired pneumonia hospitalization in children with a ventricular septal defect

The cohort consisted of 9400 exposed children diagnosed with ventricular septal defect (VSD). The risk of community-acquired pneumonia (CAP) or asthma with VSD was assessed using the Cox proportional hazard model with an inverse probability of treatment weighting. During a mean follow-up of 6.67 years (starting from 12 months after birth), there were 2100 CAP admission cases among exposed patients (incidence rate: 33.2 per 1000 person-years) and 20,109 CAP admission cases among unexposed children (incidence rate: 29.6 per 1000 person-years), with hazard ration of 1.09 (95% CI 1.04–1.14).


Ethics approval and consent to participate
Conducted in alignment with the Declaration of Helsinki, the Korea National Institute for Bioethics Policy Institutional Review Board (IRB) approved this study (P01-201603-21-005) due to its retrospective nature.The Korea National Institute for Bioethics Policy IRB approved that all methods in this study were performed according to the relevant guidelines and regulations.Also, because of its retrospective nature, the Korea National Institute for Bioethics Policy Institutional Review Board waived informed consent.

Study population and study design
Figure 1 shows the study population's selection.The present study included 2,395,966 Korean children born between 2008 and 2012.Of these, 121,339 were excluded according to the exclusion criteria (chromosomal abnormalities, congenital malformation of the respiratory system, cleft lip and cleft palate, disorder related to gestation length and fetal growth), leaving 2,274,627 eligible patients for further analyses.The 12,378  aortic stenosis (AS), coarctation of the aorta (COA), total anomalous pulmonary venous return (TAPVR) plus partial anomalous pulmonary venous return (PAPVR), or tetralogy of Fallot (TOF) were excluded.Finally, 9400 children were assigned to the VSD group.After a 1:10 complete random sampling, 94,000 children were in the non-VSD group.Also, the VSD group was divided into two subgroups: VAD with heart failure (HF) (N = 513) and VAD without HF(N = 8,887).
The primary endpoint was hospital admission for CAP diagnosed when a child had ICD-10 codes from J12X to J18X from NHIS claims-based data of inpatients.CAP hospital admission was also analyzed when a child had ICD-10 codes from J12X to J18X for inpatients after 12 months.The VSD with HF was analyzed when a child had ICD-10 codes from I50X for inpatients after 12 months.
Asthma was diagnosed when a child had ICD-10 codes of J45X or J46X at least twice and a prescription history of asthma medication (inhalation steroid and anti-leukotriene) from NHIS claims-based data of outpatients or admission history with ICD-10 codes of J45X or J46X for inpatients.The secondary endpoint was hospital admission for asthma diagnosed with ICD-10 codes of J45X or J46X for inpatients after 24 months.Meanwhile, atopic dermatitis (AD) and allergic rhinitis (AR) were diagnosed when a child had ICD-10 codes of L20.9 and J30. 4, respectively, at least five times for outpatients from NHIS claims-based data of outpatients.

Covariate and follow-up
Birthday, birth weight, prematurity, and breast milk feeding information were obtained from answers (by parents) to questionnaire items in the first round of the national health-screening program.Birth residence was used in the eligibility database divided into three regions (Seoul/metropolitan vs. urban vs. rural).Household income was determined based on insurance co-payment amount in eligibility database quintiles.Comorbidity allergic diseases including AD and AR were used in the eligibility database and divided into two classes (yes vs. no).
We separately analyzed subjects' primary (hospital admission for pneumonia) and secondary (hospital admission for asthma) endpoints.All participants were followed up from the index date (= birth date) until the diagnosis of pneumonia or asthma, outpatient visit, admission to the hospital, or the end of the study (December 31, 2019).To reduce risks of reverse causality and surveillance bias, we excluded those with the first year of follow-up for pneumonia and those with the first two years of follow-up for asthma.Follow-up of unexposed children was additionally censored if they were later diagnosed with VSD.These children were then moved to the exposed group.

Statistical analysis
Continuous variables are expressed as mean ± standard deviation (SD) based on data normality.They were compared using an independent sample t-test or Mann-Whitney U test, as appropriate.Cumulative incidence of events at 10 years was calculated based on Kaplan-Meier censoring estimates.Clinical outcomes of two groups were compared with the log-rank test.
Primary analyze was performed after adjusting for confounding factors.First, a multivariable Cox regression model was used.Covariates were sex, birth weight (< 3.2 kg or ≥ 3.2 kg), age at index date (1-5 years or ≥ 5 years), calendar year at index date (2008-2009 or 2010-2012), allergic disease (AR, AD, and asthma), and comorbidity number of hospital admission with any reason or wheezing during the first year after study entry (yes or no).Second, to reduce selection bias and other potential confounding factors, we performed an analysis using the logistic regression model with an inverse probability of treatment weighting (IPTW) for sensitivity analyses 17 .Adjusted covariates in IPTW analysis included sex, birth weight, calendar year at birth date, birth residence, and income quintile.
We established a multivariable Cox proportional hazards model to identify independent predictors over a median follow-up of 7 years' pneumonia and major adverse events.Primary end point comparison according to the various exploratory subgroups was then performed.In all analyses, participating centers were included as random effects.All statistical analyses were performed using R (version 3.1.1)and SAS software ver.9.4 (SAS Institute, Cary, NC, USA).

Institutional review board
The Korea National Institute for Bioethics Policy Institutional Review Board approved this study (P01-201603-21-005) due to retrospective nature.

Baseline characteristics
A total of 103,400 children participated in this study, including 94,000 in the non-VSD group and 9400 in the VSD group.We first examined children's essential sociodemographic and clinical characteristics in VSD and non-VSD groups (Table 1).VSD and non-VSD groups had male gender in 46.26% and 51.23%, respectively.Among the VSD group, the two leading diagnosis types were Q21.09 (ventricular septal defect, unspecified, N = 2,967 (31.56%)) and Q21.0 (ventricular septal defect, N = 2,235 (23.78%)).
The final assessment was on December 31, 2019.The median follow-up was 6.67 (3.14) years for those with pneumonia and 7.65 (2.26) years for those with asthma in the VSD group.It was 6.96 (3.04) year for those with pneumonia and 7.78 (2.24) years for those with asthma.Compared with the control group, children with VSD were more likely to be male portion (P < 0.001), prematurity (P < 0.001), only breast milk feeding (P = 0.019), residency of birth (P < 0.001), and birth date (P < 0.001).However, children with VSD were no more likely to be income level (P = 0.436) and allergic disease comorbidity (AR (P = 0.429) and AD (P = 0.497).The allergic disease comorbidity was found in 2,974 (31.55%) children with AR and 1,359 (14.46%) children with asthma in the VSD group (Table 1).

Risk of CAP among children with VSD compared with matched non-VSD children
Results of comparing follow-up clinical outcomes of CAP between VSD and non-VSD groups are shown in Table 2.During a mean follow-up of 6.67 years (starting from 12 months after birth), we identified 2100 with a newly diagnosed CAP admission among exposed patients (incidence rate: 33.2 per 1000 person-years) and 20,109 CAP admission cases among unexposed children (incidence rate: 29.6 per 1000 person-years).This corresponded to an absolute rate difference of 3.65 (95% confidence interval (CI): 2.20-5.10)per 1000 person-years.After controlling for confounders, the risk for CAP during the second year after study entry was increased among patients with VSD (HR: 1.09, 95% CI 1.04-1.14)compared with matched children without VSD.Consistent results were found in IPTW analyses.After IPTW, the absolute rate difference was 2.96 (95% CI 1.42-4.49)per 1000 person-years.number of admission with wheezing during the second year after study entry (HR: 1.06 [95% CI 0.99-1.13]and 0.87 [95% CI 0.81-0.93]for 0 time and ≥ 1 time, respectively; P for interaction < 0.001) (Table 2).
After restricting participants to be more than five years of follow-up, children with VSD had a higher cumulative incidence of CAP than their matched non-VSD children across the follow-up period (log-rank test P < 0.001) (Fig. 2).

Risk of CAP among children with VSD according to HF and operation
During a mean follow-up of 6.79 years (control), 6.27 years (VSD with no HF), and 3.26 years (VSD with HF) (starting from 12 months after birth), we identified 20,109 CAP admission cases among control (incidence rate: 29.6 per 1000 person-years) and 8887 CAP admission cases among VSD with no HF and 513 CAP admission cases among VSD with HF, respectively (Table 3).After controlling for confounders, the risk for CAP during the second year after study entry was increased among patients with VSD with no HF (HR: 1.08, 95% CI 1.03-1.13)and among patients with VSD with HF (HR: 1.42, 95% CI 1.20-1.68),compared with matched children without VSD.Also, the risk for CAP during the second year after study entry was increased in patients with VSD with HF than in patients with VSD with no HF (Table 3).
In subgroup analysis for follow-up period admissions, these associations were stronger among children with VSD operation at the number of admissions during the second year after study entry (HR, 2.85 [95% CI 1.44-5.67]and 1.00 [95% CI 0.75-1.33]for before and after, respectively; P for interaction = 0.003) (Table 3).

Risk of asthma among children with VSD compared with matched non-VSD children
Results of comparing follow-up clinical outcomes in asthma between VSD and non-VSD groups are shown in Table 4.During a mean follow-up of 7.65 years (starting from 24 months after birth), we identified 1010 asthma admission cases among exposed patients (incidence rate: 14.0 per 1000 person-years) and 9820 cases among unexposed children (incidence rate: 13.4 per 1000 person-years).This corresponded to an absolute rate difference of 6.25 (95% CI − 2.81-15.31)per 1000 person-years.After controlling for confounders, the risk for asthma admission or ER visit cases during the second year after study entry was increased among patients with VSD (HR: 1.05 [95% CI 0.99-1.13])compared with matched children without VSD.Consistent results were also found in IPTW analyses, showing an absolute rate difference of 0.42(95% CI − 0.55-1.38)per 1000 person-years after IPTW.
In subgroup analysis for follow-up period admissions, these associations were not stronger among children with VSD at the number of admissions during the second year after study entry (HR, 1.00[95% CI 0.90-1.10]and 1.04[95% CI 0.95-1.14]for 0 and ≥ 1 time, respectively; P for interaction > 0.05) or the number of admission with wheezing during the second year after study entry (HR, 1.07[95% CI 0.98-1.15]and 0.95[95% CI 0.83-1.07]for 0 and ≥ 1 time, respectively; P for interaction > 0.05) (Table 4).

Discussion
In the present study, we investigated the association of VSD with CAP or asthma hospitalization in children with or without VSD in a large-scale, population-based nationwide pediatric cohort from Korea (n = 103,400).
At the beginning of this study, we assumed that VSD might have some associations with the severity of CAP and asthma in the pediatric population during childhood.Thus, the current study had the following principal findings: (1) 12,378 (5.47%) individuals received their first diagnosis of VSD between birth and December 31, 2019, and (2) children with VSD were at elevated risk of hospital admission for CAP but not asthma.Moreover, using population-based comparisons, this is the first study to address VSD and its associations with CAP and asthma in the pediatric population.CHD is a gross structural abnormality of the heart or intra thoracic great vessels that is actually or potentially of functional significance 6,7,18 .The incidence of CHD in the general population is about 1%, ranging from 4/1000 to 50/1000 live births 18 .In childhood, VSD, patent ductus arteriosus (PDA), atrioventricular septal defect (AVSD), and atrial septal defect (ASD) are common acyanotic CHD 6,18 .The incidence of isolated VSD is about 0.3% of newborns 6 , but the incidence is significantly lower in adults because > 90% may eventually close spontaneously 3 .The current study's prevalence of VSD (5.47%) was higher than in previous studies in the newborn registry 6 and under five years of age 18 because the study period was from birth to December 31, 2019, a short time to close spontaneously.
This study showed that children with VSD were at elevated risk of hospital admission for CAP but not asthma.CAP is the most common cause of death in children under five years old and is responsible for approximately 1.5 million ambulatory visits in children annually in the US 1, 19,20 .The leading cause of CAP is viral infection 1, 19,20 , and due to improvements in hygiene and successful vaccination programs, recent Korean statistics indicate that CAP is no longer the leading cause of death among children 21 .
There is pulmonary overcirculation and pulmonary edema 12 because of left to right shunting of blood in children with CHD, including VSD.Due to this mechanism, patients with CHD have a higher risk for complications with viral diseases 4,5 .It is well-known that influenza infections are associated with high autopsy-confirmed coronary deaths among respiratory infections 3 .Additionally, in the case-crossover study from a population-based cohort 4 , CHD and ischemic stroke risk is higher after both in-and outpatient infection.Adults with infection had higher odds of CHD and ischemic stroke up to 90 days after infections compared with equivalent control periods 1 and 2 years before the event 4 .Recurrent acute lower respiratory tract infection often occurs in children with CHD 18 .Also, some CHD may predispose their sufferers to bronchopneumonia 8,9,11 , and children with pneumonia and CHD stayed significantly longer in the hospital than those without CHD.These results agreed with the present study, showing that children with VSD were at elevated risk of hospital admission for pneumonia compared to non-VSD children, and children with VSD with HF were at elevated risk of hospital admission for pneumonia compared to VSD children with non-HF.Therefore, children with VSD with HF might be at high risk because of their limited cardiopulmonary reserve.
Meanwhile, asthma in children is a significant concern because it increases the number of hospital visits and economic burden more than asthma in adults 22 .In the present study, we found that children with VSD were not significantly more frequently admitting asthma than non-VSD.This disagreement might be because asthma results from complex gene-environment interactions with heterogeneity in clinical presentation 23,24 .One study has demonstrated a high prevalence of airway hyperresponsiveness in patients with ASD, suggesting that airway hyperresponsiveness might be a possible mechanism for recurrent attacks of cough variant asthma 25 .The mechanism for cough variant asthma caused by heart disease is bronchial vascular congestion, resulting in bronchial edema and thickening.Furthermore, left ventricular dysfunction can lead to abnormal pulmonary function, www.nature.com/scientificreports/such as airway hyperresponsiveness or restrictive and obstructive dysfunction 25 .A retrospective cohort study of children has documented that the hazard of recurrent cough variant asthma is higher in children with CHD, especially in children with complex congenital heart disease 26 .However, these two reports had no data about the relationship between VSD and asthma or data showing that CHD had how many diagnoses 25,26 .
The greatest strength of this study was that we performed various assessments of risk for VSD with CAP or asthma in a large population of children.However, this study has several limitations.First, children in our study were 11 years or younger.Thus, the generalizability of our findings to older children needs more long-term follow-up, and there may be differential misclassification where some children in the non-VSD group may have undiagnosed VSD, and the opposite is less likely.Second, because this was an observational study, measurement errors or misunderstandings of adjustment factors might have occurred.Additionally, surveillance bias could exist in this study.Third, no comparison exists between the influence of pre-and post-operative VSD on CAP or asthma hospitalization.

Conclusions
In summary, evidence supports that childhood VSD is related to a higher surge in admission rates in children with CAP, not asthma, compared with matched children without VSD.Children with VSD might be at high risk of admission because of their limited cardiopulmonary reserve and thus need more intensive care to prevent viral infection.Further studies are needed to better understand the underlying mechanisms.

Figure 1 .
Figure1shows the study population's selection.The present study included 2,395,966 Korean children born between 2008 and 2012.Of these, 121,339 were excluded according to the exclusion criteria (chromosomal abnormalities, congenital malformation of the respiratory system, cleft lip and cleft palate, disorder related to gestation length and fetal growth), leaving 2,274,627 eligible patients for further analyses.The 12,378 individuals received their first diagnosis of a VSD between birth and December 31, 2019.Of them, 2978 children have VSD with complex CHD (ex.Double outlet right ventricle (DORV), transposition of the great arteries (TGA), single ventricle (SV), atrioventricular septal defect (AVSD), pulmonary atresia (PA), pulmonary artery stenosis (PAS),

Figure 2 .
Figure 2. Kaplan-Meier curves for 10-year failure in patients with VSD children and matched non-VSD children.VSD ventricle septal defect.

Table 1 .
Baseline characteristics of subjects.Data are presented as n (%) or mean, SD.SD standard deviation.

Table 2 .
Risk of CAP among children with VSD compared with matched unexposed individuals (after 12 months, index date = birth date).CAP community acquired pneumonia, VSD ventricle septal defect, CI confidence interval, HR hazard ratio, ICU intensive care unit, NA not applied.* Cox models were adjusted for sex, birth weight, calendar year at birth date, birth residence, and income quintile.The first year of follow-up was excluded for all analyses.** P-value was derived from interaction test by incorporating an interaction term to the Cox model.***Inverse probability of treatment weighting-adjusted for sex, birth weight, calendar year at birth date, birth residence, and income quintile. No.

Table 3 .
Risk of CAP among children with VSD, according to HF and OP (after 12 months, index date = birth date).HF heart failure, OP operation, CAP community acquired pneumonia, VSD ventricle septal defect, CI confidence interval; HR, hazard ratio.* Cox models were adjusted for sex, birth weight, calendar year at birth date, birth residence, and income quintile.The first year of follow-up was excluded for all analyses.** P-value was derived from interaction test by incorporating an interaction term to the Cox model. No.

Table 4 .
Risk of asthma among children with VSD compared with matched non-VSD children after 24 months (index date = birth date).VSD ventricle septal defect, CI confidence interval, NA not applied.* Cox models were adjusted for sex, birth weight, calendar year at birth date, birth residence, and income quintile.First two years of follow-up were excluded for all analyses.** P value was derived from interaction test by incorporating an interaction term to the Cox model.***Inverse probability of treatment weighting-adjusted for sex, birth weight, calendar year at birth date, birth residence, and income quintile. No.