Factors Associated With Ethnic Disparities in 1-Year Survival With Critical Congenital Heart Disease in New Zealand Open Access

Congenital heart disease (CHD) is the most common congenital abnormality and a significant contributor to the 1-year mortality rate internationally.^1,2 The prevalence of CHD is approximately 18 per 1000 live births,³ with critical CHD (CCHD), defined as those requiring life-preserving cardiac intervention within the first 28 days of life or those who died within the first 28 days of life from CHD without intervention,⁴ occurring less frequently at 2 per 1000 births.⁵ Ideally, CCHD is diagnosed prenatally to enable planning for postnatal treatment and, in cases in which treatment is not definitive and the outcome uncertain, to provide families with a choice of care pathway, including active surgical treatment, comfort care, or pregnancy termination, all of which are associated with high psychological and health care costs.^6–9 

In New Zealand (NZ), where universal publicly funded health care is available, pregnancy decisions and outcomes differ by ethnic group in CHD.¹⁰ For instance, termination decisions are made more frequently by European and Asian mothers with CHD pregnancies.¹⁰ More generally, worse perinatal outcomes occur in Pacific Islander and Indigenous Māori cases,¹¹ and in the specific case of CCHD with left heart obstruction, there are discrepancies in 1-year survival that are not fully explained by disease severity.^10,12,13 

Inequities in mortality and health outcomes by race and ethnicity are not isolated to NZ and occur globally at a high cost.^14–19 However, disentangling the complex interactions among ethnicity or race, socioeconomic status, and health outcomes is challenging, and evidence to assist health policy and system development to mitigate disparities is lacking in NZ.²⁰ A better understanding of the mechanisms underlying ethnic inequities in CCHD survival in NZ may provide a measure of health care system performance in the areas of maternal and infant access, investment in personnel and equipment, and team performance. Within this study, we aimed to identify factors associated with CCHD 1-year mortality rate, with a focus on factors responsible for ethnic disparity in NZ.

A population-based retrospective cohort study of CCHD live births in NZ was undertaken through the linkage of national fetal, cardiosurgical, and mortality datasets. The primary outcome was all-cause 1-year mortality rate. Further analysis reviewed the associations of maternal, infant, and clinical care factors with ethnicity and the primary outcome with inclusion and exclusion of stillbirths.

In NZ, there is universal publicly funded health care, with a single national congenital cardiac center and fetal cardiology service in Auckland. Antenatal ultrasonography examination is recommended at 20 weeks’ gestation to screen for fetal anomalies, including CHD. There is statutory reporting of all perinatal deaths from 20 weeks’ gestation to the National Mortality Review Committees, which record and monitor perinatal deaths from 20 weeks’ gestation. These include termination of pregnancy and stillbirth and all infant and child deaths.¹¹ 

Cases who were diagnosed with CCHD between January 1, 2006 and December 31, 2019, were included. CCHD was defined as those requiring life-preserving cardiac intervention within 28 days of birth or those who died within 28 days of birth from CHD without intervention. Those with congenital extracardiac and/or genetic anomalies (including variants of uncertain significance) were included and labeled as congenital comorbidities. Cases in which the pregnancy was terminated or stillbirth occurred were excluded as were cases with isolated patent ductus arteriosus, those with structurally normal hearts with myocardial disease, and those who were not NZ residents or who emigrated during the follow-up period. Because ethnicity was the key marker under investigation, where these data were unavailable, these cases were also excluded.

CCHD cases were identified from the following national datasets: hospital fetal cardiology, pediatric cardiac surgery and interventional cardiology records, and the perinatal and infant death records collected by the National Mortality Review Committees who report to the Quality and Safety Commission at governmental level.¹¹ Therefore, the merged dataset has near-complete case ascertainment nationally from 20 weeks’ gestation, missing only unreported or incorrectly reported cases. Of note, all deaths with CHD are included regardless of whether they were diagnosed before death.

Cardiac diagnoses were verified from fetal or postnatal echocardiogram reports and, where an echocardiogram was unavailable, from postmortem records and/or coding from the Mortality Review Committee dataset. CCHD cases were classified into diagnostic subgroups: hypoplastic left heart syndrome (HLHS), pulmonary atresia, conotruncal defects (CTD), left ventricular outflow tract obstruction, transposition of the great arteries (TGA), other simple, and other complex. The other simple group conditions were usually those in which a 2-ventricle circulation could be achieved, whereas the other complex group included conditions that most often resulted in a univentricular circulation (Supplemental Table 1).

The timing of diagnosis was obtained from the fetal cardiology and cardiac intervention datasets. The management pathway (comfort care or surgical intervention) was ascertained from the medical record, which was also used to cross-check the dataset and supplement missing or incomplete information.

Data from the above sources were matched and combined into a single dataset through unique mother and/or baby identifiers (NZ National Health Index number). Demographic information was obtained from the Ministry of Health’s National Maternity Collection by matching the unique identifiers to maternal and baby information. Data retrieved included date of birth, gestation at birth, weight at birth, self-identified and reported prioritized ethnicity, residential region at delivery (dichotomized into Auckland [location of the national pediatric cardiology service] or non-Auckland regions), and deprivation index. Birth weight z scores were calculated using the World Health Organization criteria.²¹ A single self-identified prioritized ethnicity was assigned to each case using NZ Ministry of Health recommended methodology: Indigenous Māori, Pacific Islander peoples, Asian, Middle Eastern, Latin American, and African (MELAA), and European.²² When ethnicity was unavailable in the maternity data, baby data were used if present. Sensitivity analysis revealed a difference in baby- and maternal-prioritized ethnicities of 6.4%.

The National Index of Deprivation (NZDep) is an area-based measure of deprivation calculated from census data in NZ.²³ This index includes measures of income, education, housing, welfare dependency, and access to communication and transportation. The NZDep is reported by geographic census area units or neighborhood areas, which are categorized into deciles, with decile 1 representing areas with the least-deprived residential households and decile 10 representing the most-deprived households. Deprivation was grouped into 2 groups of 1 to 7 (least deprived) and 8 to 10 (most deprived) for descriptive statistics, with the specific deprivation level (between 1–10) inputted as a continuous variable for multivariable analysis. NZDep2006 was used for infants born between 2006 and 2012, NZDep2013 was used for those born between 2013 and 2017, and NZDep2018 was used for those born between 2018 and 2019. When the deprivation index was not reported in the maternity data, baby data were used when available. All missing data were labeled as “unknown” values.

Survival analysis was used to compare the variables of interest to the survival probability of each CCHD live birth with differences in survival between groups compared using the log-rank test. Variables were chosen based on reasonable clinical or theoretical association with survival in CCHD. Cox proportional hazards regression was used to identify univariable and multivariable factors associated with the 1-year mortality rate. Statistical significance was assumed at the 5% level.

We first conducted univariable survival analyses to assess the association of each covariate with the survival outcome. Subsequently, we performed multivariable Cox regression, including all covariates to evaluate their combined effect. Additionally, we applied a backward selection method based on the Akaike Information Criterion (AIC), with the “Ethnicity” forced into the model, to further refine the model. Finally, we used elastic net regularization for variable selection, with the optimal α and λ parameters determined by 10-fold cross-validation. The selected variables were used to construct the final multivariable Cox regression model, and all hazard ratios (HR) with their 95% CIs were reported.

Three models are presented to illustrate how the independent variables influence the association between ethnicity and survival when adjusting for these different factors. This approach helped to avoid instability and overfitting, ensuring robust estimates of the effects of interest. Variance inflation factor (VIF) analysis was undertaken to ensure collinearity did not confound the multivariable analysis, with a VIF value below 5 indicating no significant multicollinearity. Some continuous variables were grouped for simplicity of descriptive reporting, but continuous variables were used when available for multivariable modeling to improve the reliability of outputs.

The mean (SD) or median (IQR) is reported for continuous variables as appropriate, with percentages used for categorical variables. The chi-square test was used to compare categorical variables, and analysis of variance, Student’s t test, or nonparametric tests were used to compare continuous variables, with the relationship of various factors analyzed in relation to ethnicity. Statistical analyses were performed using JMP Pro 17.0 software.

Of the 855 499 total births in NZ from 2006 to 2019, 1278 CCHD cases were identified (1.5 per 1000). There were 1121 fetal diagnoses (87.6%) with 237 pregnancy termination decisions (21.1%) and 64 stillbirths (5.7%). In those born alive (n = 975), there were 177 deaths by 1 year (18.2%) (Figure 1).

Of the patients with CCHD for whom surgical intervention was planned (n = 900), 793 survived to 1 year old (88.1%), with survival highest when cardiac surgery was completed as intended and survival lower when comorbidities were present (Table 1). The CCHD 1-year survival rate differed by ethnicity (log-rank P value = 0.027) (Figure 2).

In addition to ethnic group, factors significantly associated with 1-year survival included cardiac subgroup, congenital comorbidity, gestation at birth, birth weight z score, deprivation level, year of birth, and management pathway (Table 2).

Three models are presented (Table 2). All VIF analyses were less than 5 for the inputted variables, indicating no collinearity between inputted variables. Model 3 was the best fit with the lowest AIC value (Table 2). When deprivation level and management pathway are inputted into model 2 and 3, the adjusted HR (aHR) for all ethnic groups moves closer to 1 (Table 2).

The independent variables of significance identified in models 2 and 3 were year of diagnosis, surgical management, increasing deprivation, increasing birth weight z score, and cardiac diagnosis, which were all factors that were inequitably distributed by ethnic group. In model 2, for each z score increase in birth weight, the aHR for mortality fell by 16% (aHR, 0.84; 95% CI, 0.77–0.92). Similarly, active surgical management reduced mortality risk by 93% (aHR, 0.07; 95% CI, 0.05–0.11). Conversely, for each level increase in deprivation index from 1 to 10, the aHR for mortality increased by 7% (aHR, 1.07; 95% CI, 1.00–1.14). Cardiac subgroup was associated with 1-year mortality rate independent of other factors, with HLHS associated with the highest mortality risk. Results are similar when stillbirths are included in the models (Supplemental Table 4).

Ethnic group was not associated with congenital comorbidities, infant sex, timing of first intervention (if applicable), or stillbirth (Supplemental Table 2). However, ethnicity was associated with the following variables: gestational age at birth (lower in Asian peoples), birth weight z score (lower in Asian peoples), deprivation level (higher in Māori and Pacific Islander peoples); maternal age at birth (younger in Māori and MELAA peoples), and Auckland residence (more common in Asian, Pacific Islander, and MELAA peoples). Management pathway (surgical or comfort care) varied by ethnicity (P value < 0.001), with Māori, Pacific Islander, and Asian cases approximately twice as likely to follow comfort care pathways (8.4%–13.3%) compared with European cases (4.9%) (Supplemental Tables 2 and 3).

The distribution of cardiac subtypes also varied by ethnic group with an overall P value of <0.001 (Supplemental Table 2). There was a significant association between the cardiac subtype and receiving surgical intervention (P value < 0.001, Supplemental Table 4). The cardiac lesions that were managed with comfort care most often included HLHS, complex lesions, and CTD. There were significant differences in the rate of surgical intervention for HLHS by ethnic group (lower in Asian peoples than other ethnic groups, Supplemental Table 4). There were no cases of TGA managed with comfort care (Supplemental Table 4).

We demonstrate disparate 1-year CCHD survival by ethnic group with lower 1-year survival for Indigenous Māori, Asian, and Pacific Islander cases compared with European cases over a 14-year period in NZ. Once ethnicity, gestational age at birth, birth weight z score, cardiac subtype, comorbidities, deprivation index level, year of diagnosis, and management pathway were adjusted for, ethnicity was not independently associated with survival. These results suggest the burden of risk factors—birth weight, cardiac diagnosis, deprivation level, and plan for surgical management—all of which differ significantly by ethnic group for CCHD, explain the observed ethnic inequity.

The findings of this study extend the understanding of the interaction of ethnicity and other factors that influence outcomes in CCHD and are of relevance not only in NZ but also elsewhere where there are Indigenous or minoritized population groups.^10,13 We report comparable 1-year CCHD survival (81.8%) to a temporal trends report from 1994 to 2005 (82.5%).²⁴ Moreover, results confirm multiple CCHD mortality risk factors reported in the United States,^25–35 Panama,³⁶ England, and Wales.³⁷ They can be used to direct future research and target interventions to reduce the burden of independent mortality drivers for at-risk CCHD groups. These include reducing the burden of poverty and deprivation for families, optimizing birth weight and gestation, and understanding factors that increase the risk of CCHD subtypes with increased mortality rates (such as HLHS).

The findings of this study are in line with a recent Commonwealth Fund report, “Mirror, Mirror 2024,” which reviewed multiple markers of health system performance within 10 high income countries.³⁸ Although NZ scored third overall, it had the second-lowest rank for equity and access,³⁸ with “the highest income-related differences in reported cost-related access issues and instances of unfair treatment or feelings that health concerns were not taken seriously by health care professionals because of their racial or ethnic background.”³⁸ Unfair, unequal health care experiences are reported by non-European parents with children with CCHD in NZ. These may contribute to inequitable access and contribute to CCHD mortality risk that is not readily apparent when analyzing numerical data.³⁹ 

An association of care pathway (comfort care or surgical management) with ethnic groups independent of disease severity and other comorbidities supports similar findings in selected diagnostic CHD groups in NZ.^{10,1213,38,39} The marked reduction in mortality risk associated with surgical management is unsurprising. However, the attenuation of the aHR for Māori and Pacific Islander ethnic groups with the addition of management pathway implies a relationship between ethnicity and care pathway beyond the influence of other variables, including comorbidities and deprivation level. Given that disease severity did not appear to be associated with surgical intervention, there are a number of potential explanations for this finding, including ethnicity-related differences in parental choice of comfort care, perhaps mediated by cultural factors and/or the availability or acceptance of pregnancy termination and the availability of surgical care. Alternatively, severity of disease may not be fully captured using the data available. Qualitative research suggests that cultural values, institutional and systemic racism, or differing participation in shared decision-making may contribute to ethnic differences in the care pathway.³⁹ A detailed review of decision-making may identify reasons for the variation of management pathway by ethnicity.

Deprivation level was associated with the 1-year mortality rate in CCHD, with a 7% increase in mortality risk for each deprivation level in the adjusted model. These findings support the known impact of lower socioeconomic backgrounds on CHD mortality risk.^41–43 CHD appears to be overrepresented in Indigenous populations; the reasons for these ethnic disparities in incidence of types of CCHD are not clear and merit further investigation.^42–47 Modifiable factors such as access to care, education level, and health literacy may provide a means to advance equitable survival for minoritized ethnic groups.^36,48–51 Lower socioeconomic status is known to reduce access to health care through financial and other avenues, as well as being associated with individual and environmental risk factors for birth outcomes.^13,37,39 Hence, the findings of the current study suggest that reducing the impact of childhood poverty in NZ would likely improve survival for Indigenous Māori and other minoritized children as would also be the case for similar populations in other countries. It is likely this effect will be mediated through improved availability and access to health care.

A systematic review and meta-analysis of 65 studies demonstrated that out of the 546 981 patients with CHD assessed, the likelihood of death was influenced by social determinants of health, including socioeconomic deprivation, lower maternal age, lower maternal education levels, and public insurance after the neonatal period.⁵² The lower average maternal age at birth in the Māori and Pacific Islander CCHD cohorts, together with potential language and education challenges, could contribute to our findings. If so, there may be an opportunity for health service investment to be targeted appropriately to address modifiable aspects of these factors, particularly as they relate to culturally appropriate health care.^39,53 Māori and Pacific Islander cases with CCHD generally resided in areas of higher deprivation and were almost twice as likely to undergo comfort care compared with receiving critical cardiac surgery, compared with their non-Māori and Pacific Islander counterparts. This suggests education level, access to health care, and appropriate support may influence pregnancy choices in CCHD, a conclusion supported by a qualitative study of Māori and Pacific Islander parents’ experiences.³⁹ 

We report birth weight as a factor associated with the 1-year mortality rate, which was related to ethnicity. With each unit increase in z score, there was a 16% risk reduction in the infant mortality rate. Lower gestational age at birth, comorbidities, and more complex cardiac lesions are already known to be associated with a lower weight, stillbirth, and higher mortality rate in CCHD.^54,55 Both birth weight and gestational age at birth may influence the choice of management pathway due to mortality risk; thus, perhaps reducing this burden of disease within minoritized ethnic groups in NZ may influence ethnic gaps in the CCHD mortality rate.^18,56 Importantly, the acceptability of health care for Indigenous Māori peoples was not able to be determined from this study but may a contributor to birth weight and gestation due to robust evidence from Australia showing that Indigenous co-redesign of maternity services (Birthing on Country) was successful in reducing the rate of preterm birth compared with standard care.^57,58 A NZ equivalent is yet to be established but could hold promise.

A 2023 US study by Duong and colleagues reports disproportionately increased CHD mortality rate in minority populations (Black, Asian, and other racial groups), which was partially mediated by socioeconomic status.⁵⁹ In addition, delayed diagnosis, geographic location, and access to quality screening are interrelated with poverty in patients with CHD.^60–63 Targeting known modifiable mediators of CCHD mortality risk reported here and from other international cohorts could reduce ethnic health inequities. The findings of this study are important for advocacy, policy, and system development to close disparity gaps in survival by ethnic group in NZ. Results may also have a broader application in the pediatric and cardiology spaces by providing an evidence base for future equity interventions and research.

The main strength of this study is the use of a whole-of-population dataset over a long time frame, which enhances overall external validity and generalizability. However, study results may be influenced by the exclusion of terminations, the rate of which are associated with ethnicity.^11,15 In addition, results were limited by the demographic data available and were not sufficiently granular to include other frequently reported outcome measures such as length of stay or complications. In addition, it was sometimes unclear when comfort care decisions were made; the complexities involved in the management decisions are beyond the scope of this study. Furthermore, the deprivation index may not fully reflect an individual’s sociodemographic status; more in-depth markers may produce a more accurate representation.^64,65 Data not collected in this study but which may also be associated with outcome include the infant state at birth, given that neonatal depression at birth may be associated with a higher risk of CCHD mortality.³¹ Finally, in some instances, the number of cases within subgroups is small, and thus, conclusions should be interpreted with caution.

We present evidence that disparate upstream social determinants of health alongside differential surgical and comfort management pathways partially explain the ethnic disparity in CCHD 1-year survival in NZ. Policy implications include mitigating upstream health determinants, such as the impact of poverty on access to health care and health outcomes. This includes optimizing the quality, accessibility, and equity of pregnancy health care, as well as general maternal health and well-being. Improvements in these areas may enhance the detection of CCHD and the birth status of Indigenous and non-European infants affected by CCHD. Whether there are biases in the health care system, either at the systemic or individual level (including structural racism and its downstream effects), requires further investigation.

In this national population study of CCHD over a 14-year time frame in NZ, we found reduced survival in Indigenous Māori, Asian, and Pacific Islander CCHD cases compared with European cases. These ethnic differences were explained through variable management pathways, deprivation levels, cardiac subtype, and birth weight. There is scope for improving CCHD survival by reducing the impact of the independent mortality drivers, particularly for minoritized ethnic groups. Modifiable factors could also be used to direct policy and health care practice design to advance equity in CCHD outcomes. Further research in the areas of counselling and management decision-making is required to better understand the association of ethnicity and survival in CCHD.

aHR

adjusted hazard ratio

AIC

Akaike Information Criterion

CCHD

critical congenital heart disease

CHD

congenital heart disease

CTD

conotruncal defect

HLHS

hypoplastic left heart syndrome

HR

hazard ratio

MELAA

Middle Eastern, Latin American, and African

NZ

New Zealand

NZDep

National Index of Deprivation

TGA

transposition of the great arteries

VIF

variance inflation factor.

We acknowledge the assistance of the NZ Mortality Review Committees.