Introduction

The United States (US) has one of the highest preterm birth rates among developed countries, with disproportionately higher rate of preterm births among racial and ethnic minorities [1,2,3,4,5]. In 2022, while the overall US preterm birth rate was 10.38%, there was a higher rate observed among Black mothers (14.58%) and Hispanic mothers (10.06%) when compared to White mothers (9.44%) [6]. Similar disparities are observed in common complications of preterm birth, such as respiratory distress syndrome (RDS), which further contribute to higher rates of morbidity and mortality among minority patients [7,8,9]. While these differences are multifactorial, limitations in access to healthcare and other resources due to structural barriers experienced by minorities play an important role in impacting health outcomes in these populations [10].

Healthcare resource utilization (HCRU) among preterm infants varies widely by gestational age. For example, Mowitz et al. reported that comorbidities, hospital stays, and healthcare costs increase with decreasing gestational age, with extremely preterm infants experiencing the greatest HCRU burden [11]. While gestational age is a driver of HCRU, studies suggest race and ethnicity differences in healthcare utilization. Disparities in HCRU have been well-documented in extremely ( < 28 weeks) and very preterm (28–31 weeks) infants, with Black and Hispanic infants continuing to experience higher rates of morbidity despite overall improvements in clinical outcomes including mortality, hypothermia, and late-onset sepsis [11, 12]. In contrast, moderate and late preterm infants (32–36 weeks), who represent a substantial proportion of preterm births, remain understudied regarding their HCRU patterns. Focusing on this population may help uncover persistent disparities and guide targeted interventions to improve outcomes among more mature preterm infants who occupy the majority of the NICU beds in the US.

To address gaps in research on disparities in healthcare utilization, this study examined racial and ethnic disparities in HCRU among moderate to late preterm infants with RDS using data from Kaiser Permanente Northern California (2019–2023). We evaluated differences in outcomes during birth hospitalization and one year post discharge.

Methods

Study design and data source

This retrospective cohort study utilized comprehensive, real-world data from Kaiser Permanente Northern California (KPNC) healthcare system serving nearly 4.5 million members across 15 birth facilities. KPNC’s integrated electronic medical record (EMR) database provides detailed, longitudinal data from birth throughout hospitalization and one year post-discharge, allowing tracking of infant care across multiple time points and insight into treatment patterns over time. The database includes information on enrollment, demographics, census-level data, procedures and diagnoses, pharmacy data, laboratory reports, hospitalizations, emergency department visits and mortality.

Study population

The study population included liveborn moderate (32 0/7 – 33 6/7 weeks) and late (34 0/7 – 36 6/7 weeks) preterm infants delivered or treated at KPNC facilities between January 1, 2019, and December 31, 2023, with a diagnosis of RDS, defined as ICD-10-CM code P22.0 and requiring >12 h of respiratory support (including low flow nasal canula, high flow nasal canula, continuous positive airway pressure, nasal intermittent positive pressure ventilation, conventional mechanical ventilation, and high-frequency ventilation) after birth. Exclusion criteria include infants with congenital anomalies, those receiving respiratory support for non-RDS indications, infants with early onset sepsis, those transferred from KPNC facilities and those who died during birth hospitalization. Infants were followed from discharge up to 1 year or until they were no longer enrolled within KPNC, with all outcomes reflecting utilization during the available follow up period.

Assessment of key variables

Exposure and outcome data were extracted from KPNC’s EMR. Infant race/ethnicity was determined from maternal self-reported information. Infants were classified into mutually exclusive racial/ethnic groups (White, Black, Asian, or Hispanic), in which individuals identified as Hispanic were not additionally classified as White, Black, or Asian. Short-term outcomes were evaluated during birth hospitalization and included respiratory support duration, length of hospital stay (LOS), and NICU LOS. Long-term outcomes were evaluated up to one-year post-discharge, including emergency department (ED) visits and hospitalizations overall and for respiratory and infectious respiratory conditions, and outpatient medications for respiratory conditions (diuretics, systemic corticosteroid, inhaled bronchodilator, inhaled steroids, and leukotriene receptor antagonists). Other infant characteristics collected included gestational age, infant sex, birth weight, Apgar score, level of NICU at birth, and neighborhood deprivation index (NDI). The NDI is a composite measure of neighborhood socioeconomic disadvantage derived from maternal ZIP-code during pregnancy and subsequently categorized into quartiles, with higher values indicating greater deprivation [13].

Statistical analysis

We used univariate analyses to examine differences in HCRU by race/ethnicity. Appropriate statistical tests were chosen based on data type and distribution including Chi-square test and Fisher’s exact test (cell counts <5), which were used for categorical variables and Mood’s test, which was used to compare medians of continuous variables. Assumptions for regression models were evaluated where applicable and robust standard errors were used to account for any violations in model assumptions. Multivariate analyses were conducted, adjusting for established predictors of infant outcomes including gestational age, infant sex, and NDI. Generalized linear models were used to evaluate associations for continuous outcomes and logistic regression was used for categorical outcomes. These models determined differences in outcomes by race/ethnicity categories (Black, Hispanic, and Asian) when compared with White infants as the reference category. Results are reported as adjusted mean differences or odds ratios (aORs), including 95% confidence intervals (CIs). Statistical significance of outcomes across race/ethnicity was determined using a p-value < 0.05. To evaluate the potential impacts of the COVID-19 pandemic and the subsequent ‘tripledemic’ when COVID-19, RSV and influenza circulated concurrently, we conducted sensitivity analyses stratifying infants discharged pre-pandemic (January 1, 2019-February 29, 2020) and during the pandemic (March 1, 2020-December 31, 2023), and before compared to during the tripledemic period (November 1, 2022-December 31, 2023) [14, 15]. These analyses evaluated differences in healthcare utilization across these distinct time periods. Analyses were conducted using SAS 9.4 (Cary, NC) and figures were created using R 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics

There were 1674 moderate to late preterm infants with RDS included in this study, of which 37.5% were White, 22.4% Hispanic, 17.4% Asian, and 7.4% Black (Supplemental Fig. 1; Table 1). Infant characteristics such as sex, gestational age, and 5 min Apgar score were generally similar across race/ethnicity (p > 0.05). Notable differences were observed across other infant characteristics. White infants were more likely to be born >2500 grams while Asian and Black infants were more likely to be born between 1500 and 2499 grams (p < 0.001). The distribution of race/ethnicity varied across NDI quartiles. NDI for White infants were more evenly distributed (19.1% in Q1 and 16.6% in Q4). In contrast, among Black infants, only 4.0% were among the least deprived quartile compared to 43.6% in the most deprived quartile. Similar distribution was observed in Hispanic infants, in which the proportion were 8.8% and 37.1% in the two extreme quartiles, respectively (Table 1). There were no significant variations across race/ethnicity in neonatal care transitions, surfactant administrations or discharge status during hospitalization (p > 0.05) (Supplemental Table 1; Table 1).

Table 1 Infant Demographic and Clinical Characteristics by Race/Ethnicity, Kaiser Permanente Northern California (2019–2023).
Full size table

Racial/ethnic HCRU disparities during birth hospitalizations

Short-term outcomes among moderate to late preterm infants with RDS during hospitalization did not differ by race/ethnicity (Table 1). There were no significant racial or ethnic disparities in treatment including respiratory support (duration, and type) during hospitalization (p > 0.05). Median duration of respiratory support (including invasive and non-invasive support) was overall higher among White infants (59.5 h) but did not differ significantly across race/ethnicity (p = 0.20). Black and Asian infants had the longest median NICU stays, while White infants had the shortest, though this did not differ significantly across race/ethnicity (p = 0.58; Fig. 1A, B). Similar findings were observed for hospital LOS (Supplemental Fig. 2). After adjustment, disparities in HCRU were not observed for outcomes during birth hospitalization (Supplemental Figs. 3, 4). Mean differences in hospital and NICU LOS were elevated among Asian and Black infants while a greater mean difference in respiratory support was observed among Black infants only, but these differences were not statistically significant.

Fig. 1: Post menstrual age at discharge and length of NICU stay by race/ethnicity among preterm infants, Kaiser Permanente Northern California (2019–2023).
figure 1

A Distribution of post menstrual age at discharge among infants <35 weeks gestation by race/ethnicity. B Distribution of infant length of NICU stay among infants ≥35 weeks gestation by race/ethnicity. Boxes represent the interquartile range with mean and median values indicated. Whiskers represent the range and points indicate outliers. Mood’s test was used to determine p-values across medians of continuous variables. Pairwise comparisons with White infants for Panel A: Black vs. White = 0.6207; Hispanic vs. White = 0.9052; Asian vs. White = 0.8058. Pairwise comparisons with White infants for Panel B: Black vs. White = 0.1428; Hispanic vs. White = 0.7211; Asian vs. White = 0.7862.

Full size image

Racial/ethnic HCRU disparities post birth hospitalization

Racial and ethnic disparities in HCRU emerged during the period after discharge to one-year post discharge (Table 2). Black (41.9%) and Hispanic (38.7%) infants had the highest rates of emergency department (ED) visits in the first year, compared to White infants (27.5%; p < 0.001). Similarly, differences were observed for respiratory-related ED visits (Black: 16.1%, Hispanic: 14.7%, White: 6.8%, p < 0.001), infectious respiratory ED visits (Hispanic: 20.5%, Black: 12.1%, White: 9.7%, p < 0.001), and respiratory/infectious respiratory ED visits (Hispanic: 26.1%, Black: 22.6%, White: 12.9%, p < 0.001). Regarding post-birth hospitalizations, overall rates were not significantly different across groups (p = 0.19). However, Black (3.2%) and Hispanic (3.5%) infants had the highest rates of respiratory-related hospitalizations compared to 1.3% in White infants (p = 0.01). Similar, differences were observed for respiratory/infectious respiratory hospitalizations (Hispanic: 4.3%, Black: 3.2%, White: 1.6%, p < 0.03). However, there were no significant racial or ethnic disparities in treatment outpatient medication use during the first year (p > 0.05).

Table 2 Distribution of Long Term Clinical and HCRU Outcomes by Race/Ethnicity, Kaiser Permanente Northern California (2019–2023).
Full size table

There were also differences observed for post-discharge outcomes after adjusting for confounders, particularly among ED visits (Fig. 2). Black and Hispanic infants had significantly higher odds of ED visits at one-year post discharge compared to White infants (Black: aOR=1.79, 95% CI: 1.19–2.68; Hispanic: aOR=1.59, 95% CI: 1.20–2.10). These associations were even stronger for respiratory-related ED visits, where Black infants had over twice the odds of having respiratory-related ED visits compared to White infants (aOR=2.18, 95% CI: 1.22–3.91), and Hispanic infants had similar increased odds (aOR=2.07, 95% CI: 1.34–3.19) (Fig. 2). For infectious respiratory ED visits, Hispanic infants had significantly higher odds (2.16, 95% CI: 1.49, 3.15), while differences for Black infants were not statistically significant (1.10, 95% CI: 0.59, 2.02). Additional outcomes for hospitalizations did not vary significantly between Black and White infants, although Hispanic infants had 2.53 times the odds (95% CI: 1.11, 5.78) of respiratory/infectious respiratory hospitalizations compared to White infants (Fig. 2).

Fig. 2
figure 2

Adjusted Association Between Race/Ethnicity and Long-Term Healthcare Resource Utilization (Any, Respiratory and Infectious Respiratory), Kaiser Permanente Northern California (2019–2023). Forest plots present adjusted odds ratios (ORs) and 95% confidence intervals for emergency department visits and hospitalizations (any, respiratory and infectious respiratory) stratified by race/ethnicity (Asian, Black, Hispanic) with White as the referent group. Estimates were obtained from logistic regression models adjusted for gestational age, neighborhood deprivation and infant sex. Points represent adjusted ORs and horizontal lines represent 95% confidence intervals. The vertical dashed line indicates the null value (OR=1).

Full size image

Sensitivity analyses comparing pre- and post- COVID-19 pandemic periods showed generally consistent associations between race/ethnicity and healthcare utilization, with overlapping confidence intervals for most outcomes (Supplemental Table 2). Results were similarly consistent when comparing infants discharged before compared to during the 2022–2023 tripledemic period, with no significant changes in observed disparities across time periods.

Discussion

In this study, we evaluated racial and ethnic disparities in HCRU among moderate to late preterm infants with RDS. While HCRU during birth hospitalization was largely similar across racial and ethnic groups, disparities emerged in the first year of life. Black and Hispanic infants had significantly higher rates of ED visits within the first year of life when compared with White infants. While hospitalization rates were similar across Black and White infants, Hispanic infants had significantly higher odds of respiratory-related hospitalizations.

We did not observe racial/ethnic disparities among outcomes during birth hospitalization including LOS and respiratory support, potentially due to our homogenous clinical population including infants with RDS and using a single integrated healthcare system implementing standardized protocols for early respiratory management. A large nationally representative study of over 80,000 late preterm infants (33-36 weeks’ gestation) examined demographic and clinical predictors of hospital LOS. This study found gestational age, sex, birthweight and clinical complications including RDS, sepsis and necrotizing enterocolitis were stronger predictors of longer LOS than race [16]. These findings suggest that clinical factors may play a greater role than sociodemographic factors in short-term outcomes during birth hospitalization. Other studies have reported racial and ethnic disparities in pediatric HCRU, including one study suggesting that Black, Hispanic, and other minorities experience longer LOS for pediatric conditions, even after accounting for clinical and demographic factors [17]. However, this study focused on a broader pediatric population (children <21 years), including common diagnoses like pneumonia and bronchiolitis rather than newborn-specific conditions such as RDS. The study also encompassed a wide range of hospital settings including rural and urban teaching and non-teaching hospitals, which may contribute to variability in care delivery and a lack of standardized treatment protocols, potentially leading to the observed differences. Additionally, in a large national cohort of over 219,000 infants born at 22-29 weeks’ gestation, Boghossian et al. [2019] observed racial and ethnic disparities in hospital outcomes including mortality and morbidity[12]. Although disparities narrowed over time for some outcomes, rates of hypothermia, LOS, and mortality remained elevated, particularly for Black infants. These differences likely reflect variability in care across NICUs, especially in minority-serving hospitals and the greater complexity and duration of care for this less mature patient population. Similarly, racial disparities in short-term outcomes such as LOS and mortality have been observed among infants with bronchopulmonary dysplasia (BPD), even after accounting for severity [18]. However, unlike our study, which focused on a single, acute condition with standardized treatment protocols, BPD is a chronic condition with variation in management. These differences in clinical course and management protocols in BPD likely contributed to greater variability in care and outcomes. Overall, previous studies show conflicting results related to short-term outcomes. In contrast to prior studies involving chronic conditions and varied care settings, our study focused on RDS, an acutely managed condition with standardized treatment, within a single healthcare system, potentially representing more consistent care standards and therefore fewer disparities observed among short-term outcomes.

While racial and ethnic disparities in HCRU were not observed during birth hospitalization, significant differences across ED visits and hospitalizations were observed post-discharge. These findings are consistent with prior research on ED visit recidivism and hospital readmission rates in pediatric populations, likely occurring because of access to care and overall systemic inequities that disproportionately affect Black and Hispanic infants [19,20,21]. LeDuc et al. (2005) reported increased odds of ED return visits among Black and Hispanic infants, primarily due to limited availability of primary care providers [20]. Within a California cohort, Karvonen et al. (2021) found Black and Hispanic MLPT infants to experience higher post-discharge mortality and hospital readmission rates compared to White infants despite having lower mortality while in hospital [21]. These differences were likely attributed to reduced access to both primary and subspecialty care post-discharge, financial limitations, and lower adherence to follow-up appointments [21, 22].

Although both Black and Hispanic infants had greater risk of ED visits post-discharge, only Hispanic infants showed greater risk of hospitalizations. In a similar study, Hispanic infants had the highest adjusted risk of rehospitalization during the first year of life [23]. These disparities may reflect differences in the utilization of preventive or follow-up care and the management of respiratory symptoms after discharge, rather than differences in access alone. For example, Hispanic preterm infants were found to be significantly less likely than White infants to receive treatment for respiratory symptoms after discharge whereas Black infants were more likely to receive these treatments, regardless of severity of symptoms [24]. This potential underutilization of respiratory treatment may contribute to a higher risk of complications requiring hospitalization. Additionally, Hispanic individuals have been found to experience greater challenges in accessing healthcare, including difficulty finding providers and greater transportation barriers, compared to both White and Black individuals [25, 26]. Despite all infants in this study having medical insurance through KPNC, racial and ethnic disparities in outcomes persisted.

Structural racism may play an important role in contributing to these disparities. Black preterm infants living in areas of greater racial and economic disparities tend to have increased risks of acute care visits, rehospitalization, and post-discharge mortality even after adjusting for clinical and sociodemographic factors [27]. Although we account for neighborhood deprivation in our study, the NDI may not fully capture residential segregation and structural factors such as limitations in the built environment and access to healthcare. Similar to our findings, previous studies suggest an unlikely trend among Black infants who are at a reduced risk of having respiratory complications in the NICU but experience higher rates of these complications and reduced lung function after hospital discharge [28]. This shift post-discharge likely results from impacts of experiences beyond the NICU, where minority infants face greater exposure to environmental hazards, reduced access to preventive respiratory care, impacts by insurance coverage, and other adverse conditions that may worsen these chronic respiratory conditions [21, 23, 29].

This study has several strengths, including the use of a large, diverse population from an integrated healthcare system in California, allowing for detailed assessment of racial and ethnic disparities in HCRU. Additionally, the inclusion of both short-term and long-term outcomes provides a comprehensive look at disparities beyond the neonatal period. However, there are limitations to consider. First, our study was conducted within a single healthcare system, which may limit generalizability to other settings. Second, while we adjusted for gestational age, NDI, and infant sex, unmeasured confounders, such as individual socioeconomic position, may contribute to the disparities observed. Third, the study period (January 2019 to December 2023) included both the COVID-19 pandemic and the 2022-2023 “tripledemic” of COVID-19, RSV and influenza, which may have influenced healthcare seeking behavior (e.g., vaccination, telehealth) or differential viral exposure risks by race/ethnicity. To evaluate its impact on the findings, we conducted a sensitivity analyses evaluating the racial/ethnic disparities in outcomes by both pre- and post- periods for both the pandemic and tripledemic. We found that the direction and magnitude of generally remained consistent, suggesting these events did not impact observed patterns in this population. Additionally, three infants died between discharge and one year and did not contribute data for outcomes beyond their time of death. Given the small number of deaths, the impact on overall estimates is likely minimal. Finally, because our analysis includes follow-up of infants after discharge, there is potential for selection bias due to differences in loss to follow-up by race/ethnicity. In supplemental analyses, we observed that Black infants had shorter follow-up time compared to White infants. These differences may potentially bias findings towards the null. Overall, it is important to note that race and ethnicity are social constructs used to capture the differences in exposures likely attributed to structural racism. Because these categories include diverse populations with different risk factors and healthcare experiences, this study may not fully capture individual experiences.

Conclusions

While short-term HCRU during hospitalization was comparable across racial and ethnic groups, significant disparities emerged in post-discharge healthcare use, especially for respiratory-related hospitalizations among Hispanic infants and ED visits among both Black and Hispanic infants. These disparities may reflect structural inequities, which disproportionately impact minority populations through adverse environmental exposures and barriers to timely care beyond birth hospitalization. All infants in this study received care within the same integrated health system, offering the opportunity to examine patterns in follow-up care and better understand the factors contributing to disparities. These results highlight the need for targeted interventions to improve disparities in post-discharge care and understand the range of respiratory-related morbidities among more mature preterm infants with RDS.