Distance from home to birth hospital, transfer, and mortality in neonates with hypoplastic left heart syndrome in California

Abstract

Background

Prior studies report a lower risk of mortality among neonates with hypoplastic left heart syndrome (HLHS) who are born at a cardiac surgical center, but many neonates with HLHS are born elsewhere and transferred for repair. We investigated the associations between the distance from maternal home to birth hospital, the need for transfer after birth, sociodemographic factors and mortality in infants with HLHS in California from 2006 to 2011.

Methods

We used linked data from two statewide databases to identify neonates for this study. Three groups were included in the analysis: “lived close/not transferred,” “lived close/transferred,” and “lived far/not transferred.” We defined “lived close” vs. “lived far” as 11 miles, the median distance from maternal residence to birth hospital. Log-binomial regression models were used to identify the association between sociodemographic variables, distance to birth hospital and transfer. Cox proportional hazards models were used to identify the association between mortality and distance to birth hospital and transfer. Models were adjusted for sociodemographic variables.

Results

Infants in the lived close/not transferred and the lived close/transferred groups (vs. the lived far/not transferred group) were more likely to live in census tracts above the 75^th percentile for poverty with relative risks 1.94 (95% confidence interval (CI) 1.41–2.68) and 1.21 (1.05–1.40), respectively. Neonatal mortality was higher among the lived close/not transferred group compared to the lived far/not transferred group (hazard ratio 1.77, 95% CI 1.17–2.67).

Conclusions

Infants born to mothers experiencing poverty were more likely to be born close to home. Infants with HLHS who were born close to home and not transferred to a cardiac center had a higher risk of neonatal mortality than infants who were delivered far from home and not transferred. Future studies should identify the barriers to delivery at a cardiac center for mothers experiencing poverty.

Introduction

Hypoplastic left heart syndrome (HLHS) affects 13 in 100,000 live births in the United States (Egbe et al., 2014). Despite advances in the field, HLHS is still associated with a high rate of neonatal mortality, between 21–30% in recent studies (Hirsch et al., 2011; Kane et al., 2016; Morris et al., 2014). Our own group recently published a rate of 28% neonatal mortality and 47% 1-year mortality for infants with HLHS in California from 2006 to 2011 (Purkey et al., 2021).

Many studies have linked a higher risk of mortality in infants with HLHS and other congenital heart diseases to specific sociodemographic factors, including race/ethnicity (Dean et al., 2013; Fixler et al., 2012; Nembhard et al., 2011), poverty (Bucholz EM, 2020; Hirsch et al., 2011; Kucik et al., 2014; Siffel et al., 2015), and insurance status (Kucik et al., 2014). These findings may be linked to inequities in health care access and quality or disease severity. Infant characteristics at birth, including prematurity (Mechak et al., 2018), low birth weight (Stasik et al., 2006; Tabbutt et al., 2012), genetic disorders (Tabbutt et al., 2012; Tweddell et al., 2012), and extracardiac anomalies (Jacobs et al., 2008; Stasik et al., 2006) have also been associated with mortality.

In 2014, Morris, et al., reported that birth far from a cardiac surgical center was associated with increased mortality in neonates with HLHS in Texas (Morris et al., 2014). Kaltman, et al., more recently found a 28% greater rate of death attributable to congenital heart disease among infants whose mothers lived farther away from a top 50 pediatric cardiac center (Kaltman et al., 2020). Other studies have found no difference in mortality based on birth location for infants with HLHS and other forms of critical congenital heart disease (Bennett et al., 2010; Fixler et al., 2012). In practice, in the setting of a prenatal diagnosis of HLHS, expectant mothers are often relocated as the delivery date approaches so that the infant can be born into a pediatric cardiac center with quaternary neonatal intensive care services (Donofrio, 2018; Donofrio et al., 2015). This practice is supported by the American Heart Association guidelines (Donofrio et al., 2014), but it is not clear if the practice of relocating mothers prior to delivery results in better outcomes for their infants. Data on where mothers of infants with HLHS give birth are confounded by a number of factors, including geographic region, availability and quality of cardiac surgical care, type of insurance, and where these mothers are referred for delivery by their providers. There is a gap in knowledge of how maternal birth patterns vary by sociodemographic or infant factors.

Our group recently published data on transfer patterns in neonates with HLHS in California from 2006 to 2011. We found that late transfer was more likely in low birthweight infants and infants born to US-born Hispanic and Black mothers (Purkey et al., 2021). However, this analysis did not address whether or not infants were born close to home or the distance between the birth and repair hospitals. We therefore sought to determine if infant characteristics or sociodemographic factors were associated with the distance from the maternal home to delivery hospital and the need for transfer to another hospital for repair among infants with HLHS. We further investigated whether distance or transfer were predictive of mortality in infants with HLHS.

Methods

Study population

We identified our study population from linked data from two statewide databases on births in California, the California Perinatal Quality Care Collaborative (CPQCC) and the California Office of Statewide Health Planning and Development (OSHPD). The CPQCC prospectively collects data on premature infants and infants meeting specific criteria who are admitted to participating neonatal intensive care units (NICUs), including more than 90% of the care provided by NICUs in California (cpqcc.org). OSHPD captures all live births in the state of California and includes linked data on hospital admissions for infants from birth through one year of life, birth certificates, infant death certificates, and maternal hospitalizations during the pregnancy (oshpd.ca.gov). 99.4% of live birth delivery hospitalization records from 2006 to 2011 were successfully linked to the vital statistics birth records. Linked CPQCC and OSHPD data were available from January 1, 2006, to December 31, 2011.

We used CPQCC data to identify potential cases for our study, combined with data review as described below. Data abstractors for the CPQCC manually review each patient’s chart to collect data on delivery and maternal history, post-delivery diagnoses and interventions, and hospital disposition for each infant enrolled in the database (cpqcc.org). Data definitions in CPQCC are aligned with those specified by the Vermont Oxford Network. Prior studies have evaluated the validity of ICD-9-CM coding for congenital heart disease and adult congenital heart disease and found that less than 50% of ICD-9-CM coding accurately reflects the cardiac diagnosis (Frohnert et al., 2005; Khan et al., 2018; Steiner et al., 2018). We believe CPQCC reporting of HLHS diagnoses to be more accurate than ICD-9-CM codes found in OSHPD due to the prospective nature of data collection and the need for chart review prior to recording a diagnosis. We therefore used CPQCC coding to identify the cases in our study.

We initially identified 773 cases with a CPQCC code for HLHS during the defined study period. We excluded 67 cases that were not linked to OSHPD data. Data in CPQCC and OSHPD were queried for a surgical or catheter-based intervention in the first 60 days of life. ICD-9 codes were manually reviewed by the first author to confirm that patients classified as “no intervention” had no evidence of a surgical or catheter-based intervention. Fifty-five infants survived to 60 days of life without a surgical or catheter-based intervention documented in CPQCC or OSHPD; data errors (either in diagnosis, intervention or discharge information) were considered likely in this group and they were excluded. One infant underwent a fetal surgery and was also excluded, leaving 650 cases (see Supplemental Figure 1).

Distance Variables

For the 650 cases, data in CPQCC and OSHPD were queried for maternal addresses at delivery (from birth certificates). These addresses were geocoded and used to derive neighborhood-level (i.e., census tract) poverty based on information from the U.S. census (using the 2011 American Community Survey 5-year estimates reflecting 2007 to 2011). The Environmental Health Investigations Branch (EHIB) of the California Department of Public Health conducted the geocoding, using methods described previously (Roberts et al., 2007). Geocoded maternal addresses were available for 629 out of 650 cases (96.7%).

Covariates

Demographic data about mother-infant dyads were obtained from infants’ birth certificates. Maternal data included age, race/ethnicity, education, payer status, timing of initiation of prenatal care, and geocoded residential addresses. Education level was divided into less than high school, high school graduate, or some advanced education. Payer status was defined by the maternal payer status listed on the patient discharge data, defined as private insurance, government-based insurance, or missing (the final study sample had no missing values). Infant data included sex, birthweight, gestational age, and 5-minute Apgar score. CPQCC data were used to identify the presence of other major non-cardiac congenital anomalies (see Supplemental Table 2 for further information), and cases were divided based on the presence or absence of any additional non-cardiac anomaly. We assigned hospital level of neonatal care for birth and intervention hospitals based on American Academy of Pediatrics guidelines from 2012; level I centers are well newborn nurseries, level II centers are special care nurseries with the ability to care for infants born after 32 weeks’ gestation and weighing more than 1500g at birth, level III NICUs provide more comprehensive care, and level IV NICUs are regional, quaternary care centers with access to complex cardiac surgery (American Academy of Pediatrics Committee on & Newborn, 2012). Level I and II centers were combined for the analysis due to small numbers. Census tract poverty level was assigned using the 2007–2011 American Community Survey Files.

Transferred infants were those who were transferred from the birth hospital within 30 days of life. In California, some delivery hospitals are co-located with a cardiac center, and a mother gives birth at one hospital with an immediate plan for hospitalization of the infant at the co-located children’s hospital. There were 129 infants born at three birth hospitals known to be co-located with cardiac centers; we did not consider these infants to have been “transferred” after birth.

Analyses

We divided cases based on their geographic distance from mother’s home to birth hospital. The median distance from maternal residence to birth hospital, 11 miles, was chosen as the cut-off between “lived far” and “lived close.” Small sample sizes at each end and lack of clarity on “far” vs “close” in the literature made the median distance the most logical cut-off. Cases were then further divided based on whether or not they were transferred from the delivery hospital after birth, creating four groups: “lived far/not transferred,” “lived close/not transferred,” “lived close/transferred,” and “lived far/transferred” (see Supplemental Table 1). Of note, the median distance from maternal residence to birth hospital were 30.4 miles in the lived far/not transferred group, 6.1 miles in the lived close/not transferred group, and 4.3 miles in the lived close/transferred group.

Lived far/not transferred was theorized to likely include cases who were chosen for prenatal relocation of the mother prior to delivery. Lived close/not transferred was thought to likely represent a combination of suboptimal prenatal planning or coincidence (living close to a cardiac center). Lived close/transferred was thought to likely represent a lack of optimal prenatal planning, where an infant was born at a local hospital and then transferred for intervention. The lived far/transferred group was also thought to describe a challenge to optimal prenatal planning (eg, prenatal relocation to an incorrect center) or other rare event. Our primary interest was comparing patients theoretically chosen for prenatal relocation vs. those who were not (ie, the lived far/not transferred group compared to the lived close/transferred and lived close/not transferred groups); the 77 cases in the lived far/transferred group were therefore excluded from further analyses (in addition, this group was considered too small for informative multivariable logistic regression models). After excluding this group of 77 infants, there were 552 cases.

We first describe covariates among the 552 cases included in analyses. Log-binomial regression models were used to generate unadjusted and adjusted estimates of relative risk (RR) comparing infant, care-related and sociodemographic characteristics among the three groups, with the lived far/not transferred group as reference. Adjusted models only included the primary variable of interest plus the (remaining) sociodemographic variables (maternal age, race/ethnicity, education, payer, and census tract poverty level), to avoid over-adjustment for related or intermediate variables.

To account for infants who were determined to be inoperable or whose families opted to pursue comfort care (ie, repair was arguably not an option for these infants), a sensitivity analysis was performed excluding the 95 infants who died before 7 days of life.

We conducted Cox proportional hazards analyses to determine the association of distance to birth hospital and transfer with neonatal (i.e., up to 28 days) and post-neonatal (i.e., 29–365 days) mortality. These models were also adjusted for the sociodemographic variables listed above. We also ran the neonatal mortality model excluding deaths in the first 7 days in alignment with reasons stated above.

This study was approved by the California Committee for the Protection of Human Subjects and Stanford University’s Institutional Review Board. Analyses were performed using SAS 9.4 (SAS Institute, Inc. Cary, NC).

Results

Of 629 available cases, 248 (39.4%) cases were in the lived far/not transferred group; 117 (18.6%) in the lived close/not transferred group; 187 (29.7%) in the lived close/transferred group; and 77 (12.2%) in the lived far/transferred group. The last group was excluded from further analysis (see Supplemental Table 3). Among the 552 remaining infants, 337 (61.1%) were male, 55 (6.3%) had other major non-cardiac congenital anomalies, 302 (56.3%) were born to mothers who self-identified as Hispanic, and the majority were of normal birthweight or greater (426, 77.2%) and born at term (427, 79.8%) (Table 1).

Table 1:

Infant, care-related and sociodemographic factors among infants with hypoplastic left heart syndrome, California, 2006–2011.

Characteristics	Total population n = 552
	n	%
Lived far/not transferred	248	44.9
Lived close/not transferred	117	21.2
Lived close/transferred	187	33.9
Infant and care-related factors
Sex
Male	337	61.1
Female	215	38.9
Birth weight (g)
<2500	126	22.8
≥2500	426	77.2
Gestational age (wk)
<37	108	20.2
≥37	427	79.8
5-Minute APGAR score
<7	77	14.0
≥7	475	86.1
Co-occurring major birth defects
None	517	93.7
Any	35	6.3
Prenatal care initiation
Trimester 1	456	84.4
Trimester 2 or later or none	84	15.6
Level of care of birth hospital
Level I/II	96	17.4
Level III	272	49.3
Level IV	184	33.3
Sociodemographic factors
Age (yrs)
<20	41	7.4
20–34	393	71.2
≥35	117	21.2
Race/ethnicity
White Non-Hispanic	150	28.0
Hispanic – US born	110	20.5
Hispanic – Foreign born	192	35.8
Black Non-Hispanic	35	6.5
Asian/Pacific Islander	26	4.9
Other	23	4.3
Education
Less than high school	172	33.3
High school graduate	160	31.0
Some college or higher	185	35.8
Medical Insurance
Private	212	38.6
Medi-Cal/Other Governmental	337	61.4
Census tract poverty level
<25 percentile	110	20.2
25–75 percentile	277	50.7
>75 percentile	159	29.1

In adjusted multivariable analyses, compared to infants in the lived far/not transferred group, infants in the lived close/not transferred group were less likely to be male (RR 0.70, 95% CI 0.51–0.95), or to be born into level IV centers (RR 0.41, 95% CI 0.27–0.64) and more likely to have low 5-minute Apgar scores (RR 1.48, 95% CI 1.04–2.10), mothers with some college education (RR 1.65, 95% CI 1.07–2.55), and to have lived in census tracts above the 75^th percentile for poverty (versus tracts in the 25–75 percentile of poverty, RR 1.94, 95% CI 1.41–2.68) (Table 2). Compared to infants in the lived far/not transferred group, those in the lived close/transferred group were less likely to be born into a level IV center (RR 0.62, 95% CI 0.53–0.72) and more likely to have lived in census tracts above the 75^th percentile for poverty (RR 1.21, 95% CI 1.05–1.40). Sensitivity analyses excluding the 95 infants who died before 7 days of life revealed similar results (see Supplemental Table 4).

Table 2:

Associations of infant, care-related and sociodemographic factors with distance from home to birth hospital and transfer status for repair, among all infants with hypoplastic left heart syndrome (including those who died within the first 7 days of life), California, 2006–2011.

	Lived Close/Not Transferred† (n=117)		Lived Close/Transferred† (n=187)
	Unadjusted risk ratio (95% CI)	Adjusted risk ratio (95% CI)‡	Unadjusted risk ratio (95% CI)	Adjusted risk ratio (95% CI)‡
Infant and care-related factors
Sex
Male	0.86 (0.64–1.16)	0.70 (0.51–0.95)	1.10 (0.87–1.38)	1.01 (0.87–1.16)
Female	Ref	Ref	Ref	Ref
Birth weight (g)
<2500	1.29 (0.94–1.76)	1.25 (0.88–1.78)	0.87 (0.65–1.16)	0.96 (0.81–1.13)
≥2500	Ref	Ref	Ref	Ref
Gestational age (wk)
<37	1.06 (0.73–1.55)	1.00 (0.67–1.49)	1.06 (0.81–1.38)	1.01 (0.81–1.26)
≥37	Ref	Ref	Ref	Ref
5-Minute APGAR score
<7	1.57 (1.13–2.17)	1.48 (1.04–2.10)	0.74 (0.49–1.12)	0.88 (0.70–1.11)
≥7	Ref	Ref	Ref	Ref
Co-occurring major birth defects
None	Ref	Ref	Ref	Ref
Any	1.78 (1.19–2.65)	1.46 (0.94–2.27)	1.17 (0.76–1.81)	1.12 (0.75–1.67)
Prenatal care initiation
Trimester 1	0.73 (0.51–1.05)	0.78 (0.54–1.11)	0.91 (0.68–1.23)	0.98 (0.77–1.26)
Trimester 2, later or none/unknown	Ref	Ref	Ref	Ref
Level of care of birth hospital
Level I/II	Ref	Ref	Ref	Ref
Level III	0.62 (0.46–0.83)	0.77 (0.51–1.16)	0.63 (0.54–0.73)	0.94 (0.84–1.05)
Level IV	0.22 (0.15–0.34)	0.41 (0.27–0.64)	0.07 (0.03–0.12)	0.62 (0.53–0.72)
Sociodemographic factors
Age (yrs)
<20	1.11 (0.62–1.99)	1.83 (0.97–3.48)	1.31 (0.95–1.80)	1.19 (0.94–1.49)
20–34	Ref	Ref	Ref	Ref
≥35	0.86 (0.59–1.27)	0.80 (0.53–1.19)	0.78 (0.57–1.06)	0.93 (0.77–1.11)
Race/ethnicity
White Non-Hispanic	Ref	Ref	Ref	Ref
Hispanic – US born	1.29 (0.77–2.16)	1.12 (0.64–1.95)	1.21 (0.89–1.65)	0.99 (0.81–1.21)
Hispanic – Foreign born	1.54 (1.01–2.36)	1.56 (0.92–2.65)	1.04 (0.77–1.39)	0.92 (0.75–1.12)
Black Non-Hispanic	2.20 (1.22–3.94)	1.82 (0.98–3.39)	1.69 (1.19–2.42)	1.10 (0.85–1.43)
Asian/Pacific Islander	2.36 (1.27–4.39)	2.12 (0.94–4.82)	1.77 (1.22–2.58)	1.26 (0.93–1.71)
Other	1.92 (0.99–3.72)	1.46 (0.82–2.61)	1.14 (0.62–2.06)	1.02 (0.72–1.45)
Education
Less than high school	1.24 (0.82–1.87)	0.99 (0.67–1.47)	1.11 (0.85–1.44)	1.04 (0.88–1.22)
High school graduate	Ref	Ref	Ref	Ref
Some college or higher	1.22 (0.81–1.82)	1.65 (1.07–2.55)	1.02 (0.78–1.34)	1.06 (0.88–1.27)
Medical Insurance
Private	Ref	Ref	Ref	Ref
Medi-Cal/Other Governmental	1.20 (0.87–1.64)	1.01 (0.67–1.53)	1.20 (0.95–1.52)	1.05 (0.88–1.24)
Census tract poverty level
<25 percentile	0.76 (0.47–1.23)	0.62 (0.35–1.09)	0.75 (0.53–1.06)	0.89 (0.73–1.08)
25–75 percentile	Ref	Ref	Ref	Ref
>75 percentile	1.96 (1.45–2.65)	1.94 (1.41–2.68)	1.54 (1.24–1.91)	1.21 (1.05–1.40)

^†Lived close was defined as <11 miles from home to the birth hospital and lived far as ≥11miles, based on the median. The Reference Group was 248 infants who lived far and were not transferred.

^‡Each risk ratio was adjusted only for the sociodemographic factors (age, race/ethnicity, education, insurance, and neighborhood poverty)

CI = confidence interval

Of note, the 12 level IV hospitals included in our study were not more likely to be located in high poverty areas: 4 were located in census tracts in the lowest quartile of poverty, 2 in the second quartile, 4 in the third quartile, and 1 in the highest quartile.

Among the entire cohort of 629 cases, 99 (15.7%) died by 7 days of life, 172 (27.3%) died by 28 days of life and 288 (45.8%) died by one year of age. Table 3 shows mortality risks across the three groups studied. Compared to the lived far/not transferred group, neonatal mortality was higher among the lived close/not transferred group (HR 1.77, 95% CI 1.17–2.67) but similar for the lived close/transferred group (HR 0.92, 95% CI 0.62–1.37). Risk ratios for post-neonatal mortality did not show statistical differences by group (95% CIs included 1.0). Sensitivity analysis excluding those infants who died before 7 days of life showed no statistically significant differences in mortality by group (95% CIs included 1.0, data not shown).

Table 3.

Risk of death based on distance from home to birth hospital and transfer status for repair, among 522 infants with hypoplastic left heart syndrome, California 2006–2011.

	Neonatal mortality (died at <=28 days of life)			Postneonatal mortality (died at 29–365 days of life)
	n (%)	Adjusted hazard ratio (95% CI)‡	Adjusted hazard ratio (restricted to deaths at 8–28 days) (95% CI)‡	n (%)	Adjusted hazard ratio (95% CI)‡
Lived Far, Not Transferred† (n=248)	64 (25.8)	Ref	Ref	45 (18.2)	Ref
Lived Close, Not Transferred† (n=117)	47 (40.2)	1.77 (1.17–2.67)	0.97 (0.45–2.07)	16 (13.7)	0.71 (0.37–1.37)
Lived Close, Transferred† (n=187)	49 (26.2)	0.92 (0.62–1.37)	1.12 (0.63–2.00)	38 (20.3)	0.95 (0.60–1.50)

^†Lived close was defined as <11 miles from home to the birth hospital and lived far as >= 11 miles, based on the median. The Reference Group was 248 infants who lived far and were not transferred.

^‡Each hazard ratio was adjusted for sociodemographic factors (age, race/ethnicity, education, insurance, and neighborhood poverty), infant sex, transfer status and day of life of arrival at the cardiac center.

CI = confidence interval

Discussion

This study describes where infants with HLHS are born in California in relationship to their mothers’ homes, and the extent to which they are transferred for repair. In this cohort, 39% of the total cohort of 629 infants were born far from the mother’s home and not transferred, which may suggest prenatal planning of delivery location for these infants. Relative to this group, infants who were born close to home and then transferred (which may suggest a lack of planning) were two-fold more likely to live in the highest-poverty neighborhoods at the time of birth. These findings suggest that infants born to mothers experiencing poverty may be less likely to have appropriate prenatal planning of their delivery location. Additionally, infants who are born close to home and then transferred have a higher risk of neonatal mortality than infants who are born far from home and not transferred.

Birth Location and Maternal Place of Residence

Infants who were born far from home and not transferred (lived far/not transferred) were considered most likely to include infants who had been diagnosed prenatally and then planned for delivery far from home in a cardiac center. Infants whose mothers lived close to the birth hospital and were not transferred (lived close/not transferred) were thought to represent a heterogenous group: some infants were likely born close to home despite a prenatal diagnosis of cardiac disease because families chose to pursue comfort care only (either prenatally or postnatally), whereas some infants coincidentally lived close to a cardiac center (regardless of prenatal diagnosis and planning). When comparing these two groups, infants in the lived far/not transferred group were more likely to be born into a level IV center, supporting our hypothesis that these were patients with a prenatal plan for delivery in a cardiac center. Sicker infants (with low 5-minute Apgar scores) were more likely to live close and not be transferred, again supporting our theory that these infants may have been chosen for comfort care or had more severe forms of HLHS. There is also a possibility that these sicker infants died because they could not be adequately resuscitated at a lower-level center.

The sociodemographic differences between the lived far/not transferred and lived close/not transferred groups were noteworthy. Infants born to mothers with more education and infants born to poorer mothers were both more likely to live close and not be transferred. There were no differences based on race/ethnicity. It is possible that these findings reflect the heterogeneity of the lived close/not transferred group. In California, large cardiac centers are located in both urban and suburban communities, so education and poverty level may not correlate with distance from the cardiac center. In a sensitivity analysis done after excluding those infants who died before 7 days of life, the difference by maternal education was no longer significant, but the higher rate of poverty in the lived close/not transferred group persisted.

We also found a higher risk of neonatal mortality among infants who were born close to home and not transferred versus those who were born far from home and not transferred. This is not surprising, as we would expect that neonates with HLHS who were born into a lower-level center and not transferred to a cardiac center (and therefore did not receive cardiac intervention) would not survive. After excluding infants who died before 7 days of life, there was no longer a difference in mortality between the two groups. In combination with the demographic findings above, these data suggest a possible correlation between higher maternal education level and a higher likelihood of early neonatal mortality. We speculate that this association may reflect a higher likelihood of electing comfort care among more educated mothers. One prior case series supports this association between higher maternal education and the choice to pursue comfort care (Vandvik & Forde, 2000).

Our data do not have the granularity to clearly link poverty to failure of prenatal planning and to mortality. It is unknown if the association between mortality and birth closer to home reflects infants who were elected prenatally for comfort care, sicker patients who were born in extremis unexpectedly, infants without a prenatal diagnosis who were chosen for comfort care after a postnatal diagnosis, infants who died prior to transfer, or infants who would have survived had they been born into a cardiac center.

Prenatal Planning for Delivery of an Infant with HLHS

Infants with HLHS who were born close to home and then transferred (lived close/transferred) were hypothesized to reflect suboptimal prenatal planning, either a failure of prenatal diagnosis or a barrier to prenatal assignment of delivery location at a cardiac center and relocation of a mother prior to delivery. Again, infants in the lived close/transferred group were less likely to be born into level IV centers, supporting the hypothesis that infants with planned delivery into cardiac centers were more common in the lived far/not transferred group.

Infants born into >75^th percentile for census tract poverty level were more likely to live close to the birth hospital and be transferred. Prior studies, including a recently published review using a national Medicaid dataset, have linked poverty with a lower likelihood of prenatal diagnosis (Campbell et al., 2021; Hill et al., 2015; Krishnan et al., 2021). Our study further supports these findings. Additionally, based on the distribution of census tract poverty level in which the level IV hospitals were located, it does not seem that location of cardiac centers in higher-poverty census tracts would be responsible for our finding.

Bucholz, et al. previously reported low socioeconomic status was associated with mortality and lower functional status at 6 years of age in children with HLHS in the Single Ventricle Reconstruction trial (Bucholz EM, 2020; Bucholz et al., 2020). Our study further adds to this discussion by linking neighborhood poverty with transfer after delivery.

In our study, it is unknown if the link between poverty and transfer after delivery is related to a failure of prenatal diagnosis, challenges in planning for birth hospitalization or relocation, or if there are other patient-related barriers preventing relocation of mothers in this situation. A recent multicenter analysis from the Fetal Heart Society Research Collaborative, including 1171 patients with HLHS from 2012 to 2016, found prenatal diagnosis of the defect to be less likely among mothers experiencing poverty, Hispanic mothers, and mothers living in rural locations (Krishnan et al., 2021). Future research should attempt to confirm if prenatal relocation is also less likely in poorer populations, and if this difference is attributable to physician error, systemic bias related to socioeconomic status, insurance status, or patient choice due to other stressors such as the need for childcare for other children, unstable housing, or job insecurity. In addition to research efforts, pediatric cardiologists can work to challenge their own implicit biases related to poverty and other sociodemographic factors, advocate for broader access to health coverage for all patients, and engage in outreach to address some of the social stressors that may impact a family’s ability to choose prenatal relocation prior to delivery of a fetus with congenital cardiac disease.

Despite this link between poverty and birth location for infants with HLHS, there were no differences in mortality for infants born close to home and transferred versus those born far from home and not transferred. Two prior studies have linked birth far from a cardiac center with mortality in infants with critical congenital heart disease (Kaltman et al., 2020; Morris et al., 2014). However, many smaller studies, including one by our own team, have not found transfer after birth to be associated with a higher risk of mortality (Bennett et al., 2010; Fixler et al., 2012; Purkey et al., 2021). Future studies should investigate if other outcomes, including long-term morbidities, are associated with birth location and transfer.

Strengths and Limitations

Our study includes a large number of neonates with HLHS in the most populous state in the country using a linkage of robustly created and managed databases. However, we are limited by the utility of ICD-9 codes to fully capture a patient’s hospital course. While we are confident that the prospective nature of CPQCC data collection combined with the double check of OSHPD coding allows us to identify all cases of HLHS, there is always the possibility of misclassification of the diagnosis in registry data. Prenatal diagnosis status, severity of illness, and the specific decision making around transfer are not available from the datasets we used, forcing us to make some assumptions about the cases. We have assumed that infants in the “lived far/not transferred” category represent infants who were chosen to be born far from their home due to prenatal planning of delivery in a cardiac center. However, this is still an assumption, and cannot account for families whose closest delivery hospital is more than 11 miles from their residence or infants who could not be transferred after delivery due to critical illness or insurance issues.

Additionally, our definition of transfer may not have accounted for all infants transferred from a women’s birth center to a “co-located” children’s hospital; infants in this category may have had extensive prenatal planning, but may have inadvertently been counted as “transferred” in our study. However, as noted above, we did account for three hospitals known to be co-located with cardiac centers and did not count these infants as “transferred.” We also divided infants into “level of care of birth hospital” by their true birth hospital (so an infant born at a level II hospital and transferred to the co-located level IV hospital would be counted as “level II”). If the infants born into the three noted co-located hospitals had been counted as born at a level IV hospital, this would have increased the percentage of cases born at a level IV hospital to 313 (56.7% of the total population analyzed).

Furthermore, the cutoff between “lived far” and “lived close” of 11 miles may seem insignificant. However, we did confirm that the median distances from home to delivery hospital were substantially different for each of the groups analyzed. An additional limitation is that distance does not equate to driving time (particularly within urban centers). However, given changes in traffic patterns over the course of a day or week and given the varying routes possible, we chose to use median distance instead of driving time. Minimal prior research has been done to clarify the definitions of “close” versus “far” transfer distances as they relate to birth defects; this paper represents a beginning attempt to answer these types of questions.

We also excluded infants in the lived far/transferred category from our analyses. The lived far/transferred cases were thought to represent a heterogeneous group with a variety of possible challenges to ideal prenatal planning. Even with 77 cases, the numbers in each category were small. We cannot account for the specific barriers to prenatal planning in this group.

Given the constraints of our databases, our data reflect an earlier era in the management of HLHS (2006 to 2011). Prenatal relocation of mothers prior to delivery was still in its infancy nationwide. However, a study from Northern California from 2004 to 2005 reported a prenatal detection rate of HLHS of 61% and further found that 72% of infants with a prenatal diagnosis of congenital heart disease were born into a referral center compared to 9% of infants with a postnatal diagnosis (Friedberg et al., 2009). Recent studies still associate poverty with poor outcomes in infants with HLHS (Hirsch et al., 2011; Kucik et al., 2014; Siffel et al., 2015). Our suggestion of an association between poverty and poor prenatal planning remains a relevant target for action.

Conclusions

In this study, infants with HLHS who were delivered close to home and not transferred to a cardiac center were more likely to live in high-poverty neighborhoods and they had a higher risk of neonatal mortality than infants who were born far from home and not transferred after delivery. Although we believe prenatal planning was most likely among infants who were born far from home and not transferred, our study was limited by not having direct information regarding prenatal diagnosis and delivery planning. Future studies should attempt to investigate the barriers to prenatal relocation of families experiencing poverty in order to identify targets for change.

Data Availability Statement

The datasets analyzed in the current study are available from the California government’s Office of Statewide Health Planning and Development (OSHPD) and the California Perinatal Quality Care Collaborative (cqpcc.org).