We performed a cross-sectional analysis of caregiver-reported data from the 2016–2019 National Survey of Children’s Health, a population-based survey of the physical and emotional health of non-institutionalized United States children 17 years or younger. Households were randomly sampled from all 50 states and the District of Columbia. Among households identified with children, households with indicators of receipt of Supplemental Security Income or households living in areas of high poverty were sampled first. One child per household was selected. For households with multiple children, children aged 0–5 years and children with special healthcare needs were oversampled at a rate of 60 and 80%, respectively. National Survey of Children’s Health oversamples children aged 0–5 and children with special healthcare needs to produce sufficient subgroup sizes to examine less common outcomes among them. Up to two primary caregivers were surveyed per child. Surveys were offered in Spanish and English in web-based, paper, and telephone-assisted formats. From 2016 to 2019, the weighted overall response rate ranged from 37.4 to 43.1%. Data were weighted to account for non-response bias and to produce population-based estimates.

All measures were caregiver-reported. Exposures of interest were child’s heart condition status and race/ethnicity. Children were considered to have a heart condition if their caregiver answered “yes” to the following survey question: “Has a doctor or other healthcare provider EVER told you that this child has a heart condition?” Child’s race/ethnicity was combined into the following categories: Hispanic, non-Hispanic White alone, and Non-Hispanic Black or African American alone. Non-Hispanic children of other race/ethnicities or multiple race/ethnicities were excluded due to low sample size among each group. (For non-Hispanic American Indian or Alaska Native, non-Hispanic Asian, non-Hispanic Native Hawaiian and Other Pacific Islander, non-Hispanic other, and non-Hispanic multiracial children, the total n was 15,562 [12.6%, range n = 273–7620] and among heart conditions was 327 [11%, range n = 8–187]) Missing data on race/ethnicity, representing <1% of children, were multiply imputed.

Covariates of interest included child’s age and sex at birth. We grouped child’s age into three categories: 0–6, 7–11, and 12–17 years. Child’s sex at birth was defined as male or female. Missing data on age and sex were multiply imputed. We also examined primary language spoken at home, which was dichotomised as English and a language other than English.

The outcomes of this analysis broadly fall under three social determinants of health: socio-economic status, healthcare access, and healthcare utilisation. For socio-economic status, we examined caregiver’s educational attainment and work status. We categorised educational attainment, based on information from up to two caregivers, as all surveyed caregivers having less than or equal to high school education and at least one having greater than high school education. Similarly, we defined caregiver’s work status as “employed” if at least one caregiver indicated that he/she was employed at least 50 of the past 52 weeks, and “unemployed” if no caregivers were employed for at least 50 of the past 52 weeks.

For healthcare access, we examined child’s insurance type, child’s usual place of care, and whether the caregiver had problems paying for the child’s healthcare in the past 12 months. We dichotomised child’s health insurance type as any private (private only or public and private) and public, other, or no insurance. To determine child’s usual place of care, caregivers were asked “Is there a place you or another caregiver USUALLY take this child when they are sick or you need advice about their health?” and “Where does this child USUALLY go first? … Mark ONE only”. Children whose reported usual place of care was a doctor’s office or clinic/health centre were considered to have a usual place of care. Children who were usually taken to a place other than a doctor’s office or clinic/health centre (e.g. an emergency room) or who had no usual place were considered not to have a usual place of care. The child’s family was considered to have problems paying for healthcare if caregivers responded yes to the following question:“DURING THE PAST 12 MONTHS, did your family have problems paying for any of this child’s medical or healthcare bills?” Caregivers who answered “no” or who paid $0 for their child’s healthcare in the past 12 months, and were therefore not asked the question on problems paying for care, were considered to have no problems paying for care.

For healthcare utilisation, we examined any preventive healthcare visits in the past 12 months, any specialty healthcare visits in the past 12 months, number of hospital emergency room visits in the past 12 months, and unmet need for care. Children who had no healthcare visits in the past 12 months were considered to have no preventive care or emergency room visits. To assess preventive healthcare in the past 12 months, caregivers were asked: “DURING THE PAST 12 MONTHS, how many times did this child visit a doctor, nurse, or other health care professional to receive a PREVENTIVE check-up? … A preventive check-up is when this child was not sick or injured”. We dichotomised this variable as zero and one or more. To assess number of emergency visits, caregivers were asked “DURING THE PAST 12 MONTHS, how many times did this child visit a hospital emergency room?” We defined number of emergency room visits as 0, 1, or 2+ emergency room visits. To measure specialty care, caregivers were asked “DURING THE PAST 12 MONTHS, did this child see a specialist other than a mental health professional? … Specialists are doctors like surgeons, heart doctors, allergy doctors, and others who specialize in one area of health care”. We defined needed specialty care as “Yes” and “No, but this child needed to see a specialist”. Children were considered to have unmet need for care if the parent answered “yes” to the following survey question: “DURING THE PAST 12 MONTHS, was there any time when this child needed health care but it was not received? … Health care includes medical care, dental care, vision care, and mental health services”.

We excluded children missing any variables of interest. Using chi square tests, we compared weighted prevalence of all social determinants of health outcomes, by heart condition status and race/ethnicity. Using the predicted marginal approach to multivariable logistic regression, adjusted for age and sex, we calculated adjusted prevalence ratios and 95% confidence intervals to examine the associations between heart condition status and each dichotomous outcome. For number of hospital emergency room visits, a three-level variable, we used the predicted marginal approach to multinomial logistic regression using a generalised logit link function. We also examined the associations between race/ethnicity and each outcome among children with and without heart conditions, separately. Among children with heart conditions, we examined effect modification in the association between race/ethnicity and outcomes by caregiver work status and educational attainment. As a sensitivity analysis, among children with heart conditions, we dichotomised Hispanic children into two groups: Hispanic children from primarily English-speaking households and Hispanic children from households primarily speaking a language other than English and compared outcomes between each group and non-Hispanic White children.

All analyses were conducted using SAS-callable SUDAAN. All estimates accounted for complex sampling and were weighted for non-response and to produce population-based national estimates. To create population-based estimates, the National Survey of Children’s Health assigns a weight for each child, comprised of the probability of the child’s household being sampled, adjustment for non-response, and adjustments for demographic characteristics of the sampled children relative to the general US population (e.g. state, race, Hispanic ethnicity, sex, and age). More detailed information on sampling and weighting can be found in the National Survey of Children’s Health Methodology Reports.^16–19 This analysis was exempt from human patients review due to the de-identified nature of this secondary dataset.

Almost one in five (19.7%) children with heart conditions compared to 23.5% of children without heart conditions did not have a usual place of medical care (p = 0.02; adjusted prevalence ratio [95% confidence interval]: 0.8 [0.7, 1.0]); less than one in ten children with heart conditions (9.9%) compared to 19.8% of children without heart conditions had not received preventive healthcare in the past 12 months (p < 0.001; 0.5 [0.4, 0.6]). However, for 18.0% of children with heart conditions compared to 10.5% of those without heart conditions caregivers reported having problems paying for the child’s medical care (p < 0.001; 1.7 [1.5, 2.0]) (Fig 1). Children with heart conditions were more likely to have one (22.4% versus 15.1%; (1.5 [1.3, 1.7]) and two or more (10.6% versus 4.5%; (2.4 [1.9, 3.0]) emergency room visits in the past 12 months (p < 0.001) and unmet need for healthcare (5.6% versus 2.9%; p < 0.001; 1.9 [1.4, 2.5]). Among children with heart conditions, 2.2% reported needing medical care, 1.9% dental care, 0.9% vision care, 1.8% mental health services, and 1.1% another type of care; 1.8% reported needing two or more types of care. The most frequently reported reasons contributing to children with heart conditions not receiving needed health services were issues related to cost (60.1%), difficulties getting an appointment (54.2%), ineligibility for services (27.5%), services unavailable in their area (24.4%), and transportation (16.3%).

Among the 1318 children with and 19,082 without heart conditions who needed to see a specialist, 5.3% of children with heart conditions and 11.2% of children without heart conditions did not receive specialty healthcare in the past 12 months (p < 0.001; adjusted prevalence ratio = 0.5 [0.3, 0.8]). We found only 3.2% of children with heart conditions had no insurance; therefore, we combined this category with public or no insurance. We found no difference in caregiver educational attainment and work status, and child’s health insurance type by heart condition status.

Among children with heart conditions, caregiver employment (p = 0.007) and education (p = 0.048) differed by race/ethnicity. Specifically, among children with heart conditions, Hispanic children, compared with non-Hispanic White children, were more likely to have no caregiver employed at least 50 of the past 52 weeks (15.4% versus 8.0%; 1.9 [1.0, 3.5]) and have caregiver(s) with a high school education or less (33.1% versus 22.7%; adjusted prevalence ratio [95% confidence interval]: 1.5 [1.0, 2.1]). Compared with non-Hispanic White children, non-Hispanic Black children were also more likely to have no caregiver employed at least 50 of the past 52 weeks (24.7% versus 8.0%; 3.2 [1.9, 5.2]) and have caregiver(s) with a high school education or less (35.7% versus 22.7%; 1.6 [1.1, 2.3]) (Fig 2; for exact estimates see supplemental materials).

Differences by race/ethnicity among children with heart conditions were also seen for insurance type (p < 0.001) and having a usual place of care (p = 0.001). Specifically, among children with heart conditions, Hispanic compared with non-Hispanic White children were more likely to have public or no insurance (54.8% versus 29.6%; adjusted prevalence ratio = 1.9 [1.5, 2.4]) and no usual place of medical care (26.3% versus 14.5%; 1.8 [1.2, 2.7]). Compared with non-Hispanic White children, non-Hispanic Black children were also more likely to have public or no insurance (62.1% versus 29.6%; 2.1 [1.7, 2.7]) and no usual place of medical care (33.9% versus 14.5%; 2.3 [1.6, 3.4]). Estimates for public insurance did not change substantially when 73 children with no insurance were removed from the model (Hispanic adjusted prevalence ratio = 2.0; non-Hispanic Black adjusted prevalence ratio = 2.3). There were no racial/ethnic disparities in having problems paying for child’s medical care among children with and without heart conditions.

Among children with heart conditions, no racial/ethnic disparities were seen in receipt of preventive or specialty healthcare visits in the past 12 months or unmet need for healthcare in the past 12 months (p > 0.05 for all). However, Hispanic and non-Hispanic Black children, compared with non-Hispanic White children, respectively, were more likely to have ≥ 2 emergency room visits in the past 12 months (14.1%, 18.3% versus 7.8%, p < 0.001; 1.8 [1.0, 3.2] and 2.3 [1.3, 4.3]).

The racial/ethnic disparities seen in socio-economic status and healthcare access among children with heart conditions were also seen among children without heart conditions (Supplementary Tables S2 and S3).

Among 287 children of Hispanic ethnicity with heart conditions and information on language primarily spoken at home, 27.1% primarily spoke a language other than English at home (n = 10 (3.4%) missing data on language spoken at home). Hispanic children from households who primarily spoke a language other than English were more likely than non-Hispanic White children not to have received preventive (21.7% versus 8.8%; 2.5 (1.2, 5.2)) or specialty (16.5% versus 4.0%; 3.9 (1.1, 13.7)) healthcare in the past 12 months, although chi square p-values were not statistically significant, likely due to smaller sample size (p > 0.05). No disparities were seen between Hispanic children from primarily English-speaking households and non-Hispanic White children in not receiving preventive (9.6% versus 8.8%; adjusted prevalence ratio 1.1 (0.5, 2.4)) or specialty (3.5% versus 4.0%; 0.8 (0.2, 2.8)) healthcare in the past 12 months. Associations between Hispanic ethnicity and other outcomes did not differ substantially by language spoken at home.

When limited to children with heart conditions whose caregiver(s) were employed throughout the year, differences by race/ethnicity were seen for emergency room visits (p = 0.006). Specifically, compared to non-Hispanic White children, Hispanic children were 2.4 [1.3, 4.4] times more likely to have ≥2 emergency room visits in the past 12 months (14.8% versus 6.2%). Compared to non-Hispanic White children, non-Hispanic Black children were no more likely to have ≥2 emergency room visits in the past 12 months (6.8% versus 6.2%; 1.0 [0.5, 2.1]). When limited to children with heart conditions who have caregiver(s) with greater than a high school education, differences by race/ethnicity were seen for emergency room visits (p = 0.002) and insurance type (p < 0.001). Specifically, compared to non-Hispanic White children, Hispanic children were 2.8 [1.6, 5.2] times more likely to have ≥2 emergency room visits in the past 12 months (16.4% versus 6.1%) and 2.8 [2.0, 3.8] times more likely to have public or no insurance (46.4% versus 16.9%). Compared to non-Hispanic White children, non-Hispanic Black children were 2.0 [0.9, 4.6] times more likely to have ≥2 emergency room visits in the past 12 months (12.1% versus 6.1%) and 3.1 [2.3, 4.1] times more likely to have public or no insurance (51.4% versus 16.9%).