Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Dec 1.
Published in final edited form as: Res Nurs Health. 2021 Sep 10;44(6):920–930. doi: 10.1002/nur.22182

Income-Driven Socioeconomic Status and Presenting Illness Severity in Children with Acute Respiratory Failure

Alicia G Kachmar 1,2, David Wypij 3,4,5, Mallory A Perry 6, Martha AQ Curley 7,8,9; RESTORE Study Investigators
PMCID: PMC8559130  NIHMSID: NIHMS1736735  PMID: 34505720

Abstract

Children living in low socioeconomic communities are vulnerable to poor health outcomes, especially when critically ill. The purpose of this study was to investigate the association between socioeconomic status and illness severity upon pediatric intensive care unit admission in children with acute respiratory failure. This secondary analysis of the multicenter Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) clinical trial includes children, 2 weeks - 17 years old, mechanically ventilated for acute respiratory failure; specifically, subjects who had parental consent for follow-up and residential addresses that could be matched with census tracts (n = 2006). Subjects were categorized into quartiles based on income, with a median income of $54,036 for the census tracts represented in the sample. Subjects in the highest income quartile were more likely to be older, non-Hispanic White, and hospitalized for pneumonia. Subjects in the lowest income quartile were more likely to be Black, younger, and hospitalized for asthma or bronchiolitis, to have age-appropriate baseline functional status and history of prematurity and asthma. After controlling for age group, gender, race, and primary diagnosis, there were no associations between income quartile and either Pediatric Risk of Mortality (PRISM) scores or pediatric acute respiratory distress syndrome (PARDS). As measured, income-based socioeconomic status was not associated with illness severity upon PICU admission in this cohort of patients. More robust and reliable methods for measuring SES may help to better explain the mechanisms by which socioeconomic affects critical illness.

Keywords: critical illness, severity of illness index, social class, socioeconomic factors

Children living in low socioeconomic communities may be at higher risk for poor health outcomes due to a complex interplay of individual, family, and neighborhood characteristics. Socioeconomic status (SES) is a multifaceted and complex construct that stratifies members of a society based on factors such as income, education, and occupation but there is a lack of consensus surrounding its conceptualization and measurement (Braveman et al, 2005; Cheng et al., 2015; Kachmar et al., 2019; Oakes & Rossi, 2003). Recent definitions define and conceptualize SES using three distinct approaches: 1. Material and structural factors, such as access to information/resources and attainment of services, 2. gradient approaches where SES is a continuous variable and comparisons among individuals and/or groups are relative to inequality, and 3. class models, which include hierarchies of power and privilege based on social class as they pertain to social inequalities in policies and institutions (APA Task Force on Socioeconomic Status, 2007). For the purposes of this manuscript, we focus on gradient approaches to understanding SES.

Socioeconomic status, as measured by proxy variables (e.g. education, income) and health outcomes, can exhibit a gradient effect across the lifespan: as SES increases, health status improves (Adler et al., 1994; Chen, 2004; Nepomnyaschy, 2009; Williams, Priest, & Anderson, 2016). In the United States (U.S.), low SES is correlated with many health- and development-related factors including low birth weight, inadequate nutrition, low maternal education level, decreased health literacy, and uninsured or underinsured status (Dahl & Lochner, 2012; Duncan, Magnuson, & Votruba-Drzal, 2017; Ronfani et al., 2015; Shonkoff & Garner, 2012). Health inequities such as these are a part of a larger inequity across society – such as, race/ethnicity, insurance status, and unique family characteristics not attributed to biology or genetics (Flores & Committee On Pediatric Research, 2010; Intrator, Tannen, & Massey, 2016; Lopez et al., 2006). Low SES in the context of disparate health outcomes is conceptualized as inequitable access and/or delivery of healthcare services due to social determinants of health (SDOH), many of which are not in the individual’s locus of control. These characteristics have the potential to negatively impact initial cause, severity, and subsequent progression of illness. Due to socioeconomic constraints, such as limited healthcare access, individuals of lower SES are more likely to obtain treatment at the peak of their illnesses and/or symptoms, requiring hospitalization, compared to those of higher SES (Kielb, Rhyan, & Lee, 2017). As a result, children of low SES may be at risk for greater illness severity upon pediatric intensive care unit (PICU) admission (Lopez et al., 2006).

Illness severity represents the extent to which physiological variables deviate from clinically normal ranges. According to Naclerio, Gardner, and Pollack (1999, p 382), “ICU admission is at the end of the spectrum of illness severity and cost.” Studies conducted in two major U.S. metropolitan areas, with data collection spanning 1997 through 2011, found that lower SES children arrived at the PICU sicker than their higher SES counterparts, as measured by illness severity scores (Epstein et al., 2014; Naclerio, Gardner, & Pollack, 1999). Research regarding SES and pediatric critical illness has certainly developed in rigor and methodology, but is not new. One of the earliest published studies examining the impact of SES on PICU admission, found that low SES children were disproportionately admitted to the PICU at higher rates, positing that preventative interventions for this group may have been lacking due to failures in the healthcare system at large (Naclerio et al., 1999).

Income-driven SES and Illness Severity

In contrast with adults, children have little ability to alter the socioeconomic group in which they were born and live. This privilege, or lack thereof, is in itself a determinant of other SDOHs, which thereby dictates individual choices and overall behavior. Regarding the social context of health, neighborhood socioeconomic characteristics such as median income or the unemployment rate can be used as a proxy for SES (APA Task Force on Socioeconomic Status, 2007). Specific SES-related factors such as poverty, literacy, and race/ethnicity are known to interfere with healthcare access and illness prevention, leading to worsening health conditions that can only be treated emergently (Epstein et al., 2014; Shipman, Lan, Chang, & Goodman, 2011). Specific to income-driven SES, Chetty et al. (2016) found an association between income and life expectancy in adults aged 40 to 76 years, as influenced by healthcare behaviors. For example, those with higher incomes may have increased life expectancy due to increased capital to spend on health-promoting goods and services including, but not limited to, nutritious foods and housing in safe and accessible neighborhoods (Chetty et al., 2016). However, a gap remains in the pediatric critical care literature. Critically ill children are collectively characterized by physiological dysfunction to the extent that there are disturbances in the body’s homeostasis (Lacroix, Cotting, & Pediatric Acute Lung Injury and Sepsis Investigators Network, 2005; Pollack et al., 2016). As such, we aim to explore the association between income-driven SES and illness severity upon PICU admission, which has not been studied in a large, cohort of children admitted to U.S. hospitals – especially in those with acute respiratory failure. Using available baseline demographic and health characteristics, we examined the association between income-driven SES—operationalized using census tract median income—and severity of illness within the first day of PICU admission for acute respiratory failure in a geographically rich and socioeconomically diverse evaluation of thirty-one PICUs across the U.S. We hypothesized that lower median income, as a proxy for SES, would be associated with higher illness severity.

Methods

Study Design and Population

This study is a secondary analysis of the Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) study, a cluster randomized controlled trial that enrolled 2449 mechanically ventilated subjects across 31 U.S. pediatric intensive care units (PICUs) from 2009 – 2013 (Curley et al., 2015; Curley et al., 2018). The 31 PICUs were located across the U.S. and varied in size, organizational structure, and affiliation (e.g. teaching vs non-teaching hospitals). Its primary aim was to investigate the impact of a nurse-led sedation management protocol (versus usual care) on the duration of mechanical ventilation. While the intervention did not significantly reduce the duration of mechanical ventilation, analyses of secondary outcomes found that the intervention arm subjects were safely managed in a more awake and calm state and had less exposure to opioids but without a significant increase in inadequate pain and sedation management. Details of the study’s methodology and primary results have been reported elsewhere (Curley et al., 2015; Curley et al., 2018).

Patients were eligible for the acute phase of RESTORE if they were mechanically ventilated for acute respiratory failure and were between 2 weeks and 17 years of age. Of the total RESTORE population, parents/guardians of 2138 subjects also consented to post-PICU follow-up that included a parent survey and interview. Our sample of analysis consists of all subjects whose families consented to follow-up and provided residential addresses that could be matched with census tracts for income-driven SES approximation, which is described in detail below.

Data Collection

Baseline data included demographic variables (age, race, ethnicity), primary cause of acute respiratory failure, and functional status at baseline as measured by the Pediatric Cerebral Performance Category (PCPC) and Pediatric Overall Performance Category (POPC) (Fiser, 1992).

Illness severity was evaluated with two measures. The Pediatric Risk of Mortality (PRISM III-12) score, a prognostic composite measure of criticality consisting of seventeen variables, and representing physiological dysfunction (Pollack, Patel, & Ruttimann, 1996), was assessed within the first twelve hours of PICU admission, with higher scores indicating increased risk of mortality (Pollack et al., 1996). The highest oxygenation index (OI) or the oxygenation saturation index (OSI) for each subject on day 0/1 was used to compute the severity of pediatric respiratory acute distress syndrome (PARDS) using the 2015 Pediatric Acute Lung Injury Consensus Conference (PALICC, 2015) criteria:

  1. At-risk: OI <4.0 or OSI <5.0,

  2. Mild: OI 4.0 to <8.0 or OSI 5.0 to <7.5,

  3. Moderate: OI 8.0 to < 16.0 or OSI 7.5 to < 12.3, or

  4. Severe: OI ≥16.0 or OSI ≥12.3.

The OI is calculated using an arterial blood gas while the OSI uses a non-invasive pulse oximeter to acquire the peripheral capillary oxygen saturation (SpO2). Each patient had at least one OI or OSI value, and for patients who had multiple values, the worst value was used to calculate PARDS severity.

While follow-up procedures were robust, 63% of subjects’ consenting parents/guardians declined to provide their annual income and 29% did not provide highest level of education (Watson et al., 2018). However, evidence shows that measurements of SES at the neighborhood level are both feasible to collect and can be representative of individual level SES, which is prone to underreporting as well as misreporting (Braveman et al., 2005; Braveman et al., 2010; Kachmar et al., 2019; Sampson, Morenoff, & Gannon-Rowley, 2002). As a result of limited data, we derived an approximation of a subject’s SES using the U.S. census tract-derived variable, “median family income by presence of own children under 18 years” from the year 2011 (the midpoint of the RESTORE study) to represent SES. Income in particular has been linked to health outcomes because of the resources it procures. For example a job with increased income may provide better healthcare and workplace benefits, as well as a tendency to live in desirable, non-polluted areas and have higher social capital (Braveman et al., 2005). The residential address for each RESTORE subject was matched to a unique 11-digit code—2-digit state code, 3-digit county code, and 6-digit tract code—using the Census Geocoder (US Census Bureau, 2019a). We subsequently linked each 11-digit code to tract-level median income using the Census Bureau’s American Fact Finder (US Census Bureau, 2019b). See Appendix A for details regarding the census as a data source and our methods of connecting census-derived data to the RESTORE dataset.

Statistical Analysis

Descriptive statistics were calculated for baseline characteristics and illness severity variables. Following U.S. census-level research recommendations, median income level was categorized into quartiles, which we have designated as Low Income, Low Middle Income, High Middle Income, and High Income (AHRQ, 2019). Because they reflect an ordinal progression, baseline characteristics and illness severity differences between quartiles were compared using the nonparametric Jonckheere-Terpstra test for ordinal and continuous variables and the Cochrane-Armitage trend test for binary variables (Bewick, Cheek, & Ball, 2004). Pearson’s chi-square test was used to compare nominal variables (race and primary diagnosis). The Wilcoxon rank sum test was used to compare continuous variables between survivors and non-survivors. The association between PRISM III-12 and PARDS was assessed using the Spearman’s rank correlation coefficient.

Adjusting for PICU as a cluster variable, linear regression was used to model the effects of independent variables on PRISM III-12 scores using an exchangeable working assumption, and cumulative logistic regression was used to model the effects of independent variables on PARDS severity using an independence working assumption. First, univariable models were created based on hypothesized associations between independent variables and each of the outcome variables. Next, multivariable models were run including income quartiles as well as sociodemographic factors and primary diagnosis. Race and ethnicity were retained in final models to address potential confounding with income quartiles. Statistical analyses were performed using SAS (version 9.4; SAS Institute).

Results

Of the 2138 subjects who consented to RESTORE follow-up, 94% were matched to a census tract median income (n = 2006). Subjects were excluded from this secondary data analysis if residential addresses were: unavailable (n = 93), incomplete (n = 28), did not sufficiently match a U.S. census tract (n = 4), were not U.S.-based (n = 3), or were connected to a medical facility (n = 1). Three additional subjects originally consented for follow-up withdrew from follow-up after PICU discharge, and thus were not included in this study. The median income for the census tracts represented in the sample was $54,036 and the income distribution was positively skewed. Non-survivors were equally distributed across the income quartiles. Median income did not differ between the 43 subjects who died before PICU discharge and the 1963 subjects who survived ($55,837 vs. $54,028; p = 0.76).

Differences between income quartiles were significant for all demographic characteristics and baseline health variables except gender, PRISM III-12 score, and risk of mortality (Table 1). Patients in the High Income quartile were more likely to be older, non-Hispanic White, and admitted to the PICU for pneumonia. Patients in the Low Income quartile were more likely to be Black, of younger age, and have age-appropriate functional status (PCPC/POPC score = 1). They were also more likely to have a history of prematurity and asthma, and to be admitted for asthma or bronchiolitis. As income increased across the quartiles the proportion of subjects identifying as White also increased, while the proportion of subjects identifying as Black decreased.

Table 1.

Baseline Characteristics and Illness Severity Measures According to Income Quartiles

Characteristics Low Incomea (n = 501) Low Middle Incomea (n = 502) High Middle Incomea (n = 501) High Incomea (n = 502) pb
Median Age (IQR), yr 1.4 (0.3–6.1) 1.5 (0.3–7.0) 1.8 (0.4–7.9) 3.5 (0.6–10.3) <0.001
 Age, n (%) <0.001
 2 wk to < 1 yr 231 (46) 226 (45) 190 (38) 158 (31)
 1 yr to < 3 yr 98 (20) 80 (16) 100 (20) 80 (16)
 3 to < 6 yr 46 (9) 61 (12) 58 (12) 66 (13)
 6 to < 18 yr 126 (25) 135 (27) 153 (31) 198 (39)
Male, n (%) 297 (59) 277 (55) 256 (51) 274 (55) 0.07
Race, n (%) <0.001
 White 238 (48) 347 (70) 380 (76) 411 (82)
 Black/African American 201 (40) 105 (21) 74 (15) 45 (9)
 Multiracial 41 (8) 25 (5) 17 (3) 18 (4)
 Otherc 20 (4) 22 (4) 29 (6) 27 (5)
Hispanic ethnicity, n (%) 164 (33) 146 (29) 73 (15) 53 (11) <0.001
Age-appropriate functional status at baseline, n (%)d 397 (79) 363 (72) 366 (73) 340 (68) <0.001
History of prematurity (< 36 wk postmenstrual age), n (%) 85 (17) 79 (16) 74 (15) 62 (12) 0.04
History of asthma (prescribed bronchodilators or steroids), n (%) 97 (19) 73 (15) 74 (15) 63 (13) 0.005
PRISM III-12, median (IQR) 7 (3–11) 8 (3–13) 7 (3–12) 8 (3–13) 0.57
Risk of mortality, median (IQR) 3 (1–8) 4 (1–14) 3 (1–10) 4 (1–12) 0.78
Died before discharge, n (%) 10 (2) 11 (2) 12 (2) 10 (2) 0.95
Primary diagnosis, n (%) <0.001
 Pneumonia 145 (29) 161 (24) 165 (33) 211 (42)
 Bronchiolitis 146 (29) 155 (31) 150 (30) 102 (20)
 Acute respiratory failure related to sepsis 62 (12) 63 (13) 61 (12) 78 (16)
 Asthma or reactive airway disease 68 (14) 39 (8) 43 (9) 36 (7)
 Aspiration pneumonia 28 (6) 28 (6) 32 (6) 33 (7)
 Othere 52 (10) 56 (11) 50 (10) 42 (8)
PARDSf, n (%) 0.01
 At risk 87 (17) 80 (16) 76 (15) 67 (13)
 Mild 125 (25) 101 (20) 122 (24) 106 (21)
 Moderate 136 (27) 166 (33) 129 (26) 147 (29)
 Severe 153 (31) 155 (31) 174 (35) 182 (36)
Open in a new tab

IQR = interquartile range, PRISM III-12 = Pediatric Risk of Mortality score from first twelve hours in the PICU, PARDS = Pediatric Acute Respiratory Distress Syndrome.

Not all column percentages sum to 100% due to rounding. Data are complete except for missing race (n = 14) and Hispanic ethnicity (n = 7).

a

Low Income < $35,878; Low Middle Income = $35,878 - $54,036; High Middle Income = $54,037 - $80,357; High Income > $80,357.

b

p values for comparison between the income quartiles were calculated using the Jonckheere-Terpstra test for ordinal and continuous variables, the Cochrane-Armitage trend test for binary variables, and Pearson’s chi square test for nominal variables.

c

Asian, Native Hawaiian or Other Pacific Islander, American Indian or Alaskan Native.

d

Age-appropriate functional status at baseline was defined as Pediatric Cerebral Performance Category (PCPC) = 1 and Pediatric Overall Performance Category (POPC) = 1 (Fiser, 1992).

e

Other diagnoses include: acute chest syndrome/sickle cell disease, acute respiratory failure post bone marrow transplant, chronic lung disease (cystic fibrosis or bronchopulmonary dysplasia), laryngotracheobronchitis (croup/tracheitis), pertussis, pulmonary edema, pulmonary hemorrhage, thoracic trauma (pulmonary contusion or inhalation burns).

f

PARDS severity was defined using the 2015 Pediatric Acute Lung Injury Consensus Conference (PALICC) criteria - At-risk: OI < 4.0 or OSI < 5.0; Mild: OI 4.0 to < 8.0 or OSI 5.0 to < 7.5; Moderate: OI 8.0 to < 16.0 or OSI 7.5 to < 12.3, Severe: OI ≥ 16.0 or OSI ≥ 12.3 (2015).

Illness Severity

PRISM III-12 scores were similar across income quartiles (p = 0.57; Table 1). Subjects who died before PICU discharge had higher PRISM III-12 scores (median = 13; interquartile range [IQR] = 8 – 21) compared with those who survived (median = 7; IQR = 3 – 12; p < 0.001). PRISM III-12 scores were independently associated with age group and primary diagnosis (Table 2). Compared with the reference age group (2 wk to < 1 yr), scores were 1.9 points higher for subjects 1 yr to < 3 yr, 2.1 points higher for subjects 3 yr to < 6 yr and 4.4 points higher for subjects 6 yr to < 18 yr. In addition, the Low Middle Income quartile had PRISM III-12 scores 1.0 points higher (95% confidence interval, 0.0 to 2.0; p = 0.05) compared to the Low Income quartile, though this is a small difference and overall there were no statistically significant differences among the four quartiles (p = 0.29). Race and ethnicity were not associated with PRISM III-12 but were retained in the multivariable analysis due to potential confounding with income. In a multivariable model controlling for age group, gender, race, ethnicity, and primary diagnosis, income quartile was not significantly associated with PRISM III-12 score (p = 0.31). Additional adjustment for functional status at baseline, history of prematurity, or history of asthma did not appreciably affect these results.

Table 2.

Association of Illness Severity as Measured by PRISM III-12

Variables Unadjusted Multivariable
β (95% CI) p a β (95% CI) p
Age (ref = 2wk to <1yr) 0.002 0.008
 1yr to <3yr 1.9 (1.1, 2.7) <0.001 1.1 (0.3, 1.8) 0.004
 3yr to <6yr 2.1 (0.9, 3.4) <0.001 1.0 (−0.4, 2.3) 0.15
 6yr to <18yr 4.4 (3.3, 5.6) <0.001 2.8 (1.6, 4.0) <0.001
Female gender (ref = male) 0.7 (0.0, 1.3) 0.07 0.2 (−0.5, 0.9) 0.57
Raceb (ref = White) 0.37 0.42
 Black/African American 0.1 (−0.9, 1.0) 0.85 −0.1 (−0.9, 0.7) 0.79
 Multiracial 0.4 (−1.4, 2.1) 0.67 0.6 (−1.1, 2.3) 0.47
 Other 1.0 (0.1, 2.1) 0.07 0.8 (−0.1, 1.6) 0.07
Ethnicity (ref = non-Hispanic) −0.8 (−1.7, 0.1) 0.10 −0.3 (−1.1, 0.5) 0.49
Primary diagnosis (ref = bronchiolitis) 0.002 0.007
 Pneumonia 2.4 (1.4, 3.4) <0.001 1.1 (0.0, 2.3) 0.04
 Acute respiratory failure related to sepsis 8.5 (7.5, 9.6) <0.001 7.1 (6.0, 8.2) <0.001
 Asthma or reactive airway disease 4.4 (3.1, 5.7) <0.001 2.8 (1.4, 4.1) <0.001
 Aspiration pneumonia 4.3 (3.4, 5.3) <0.001 2.9 (1.5, 4.3) <0.001
 Other 3.8 (2.6, 5.1) <0.001 2.6 (1.2, 4.1) <0.001
Income quartilec (ref = Low) 0.29 0.31
 Low Middle 1.0 (0.0, 2.0) 0.05 0.9 (−0.1, 1.9) 0.08
 High Middle 0.6 (−0.4, 1.6) 0.26 0.3 (−0.7, 1.3) 0.55
 High 0.8 (0.0, 1.6) 0.06 0.1 (−0.5, 0.8) 0.69
Open in a new tab

PRISM III-12 = Pediatric Risk of Mortality score from first twelve hours in the PICU; CI = Confidence Interval.

Higher PRISM scores are associated with greater illness severity.

a

Variables significant at p <0.20 in unadjusted analyses were included in the multivariable analysis.

b

Race and ethnicity were included in the multivariable analysis to control for potential confounding with income.

c

Income quartile was included in the multivariable analysis because it is the predictor variable of interest.

Subjects in the two oldest age groups (3 yr to < 6 yr and 6 yr to < 18 yr) had more than twice the odds of having severe PARDS than those in the reference age group (2 wk to < 1 yr). PARDS severity was significantly different across income quartiles (p = 0.01, Table 3). More subjects from the Low Income quartile were at risk for PARDS while more subjects from the High Income quartile had severe PARDS. The odds of dying before PICU discharge were more than five times as likely for subjects with severe PARDS as compared with subjects who were at risk for PARDS (odds ratio = 5.39; 95% confidence interval, 2.51 to 11.57; p < 0.001). PARDS severity was independently associated with age group, gender, primary diagnosis, and the High Income quartile (Table 3). However, after controlling for age group, gender, race, ethnicity, and primary diagnosis, income quartile was not significantly associated with severe PARDS (p = 0.96). Additional adjustment for functional status at baseline, history of prematurity, or history of asthma did not appreciably affect these results.

Table 3.

Association of Illness Severity as Measured by PARDS

Variables Unadjusted Multivariable
OR (95% CI)a p b OR (95% CI)a p
Age (ref = 2wk to <1yr) <0.001 <0.001
 1yr to <3yr 1.45 (1.13, 1.85) 0.003 1.45 (1.13, 1.86) 0.003
 3yr to <6yr 2.70 (1.96, 3.72) <0.001 2.74 (2.02, 3.73) <0.001
 6yr to <18yr 2.56 (2.02, 3.24) <0.001 2.54 (1.90, 3.38) <0.001
Female gender (ref = male) 1.25 (1.06, 1.47) 0.01 1.18 (0.97, 1.42) 0.09
Racec (ref = White) 0.83 0.90
 Black/African American 0.90 (0.73, 1.12) 0.35 0.96 (0.76, 1.22) 0.74
 Multiracial 0.93 (0.61, 1.40) 0.72 1.07 (0.71, 1.61) 0.76
 Other 0.90 (0.63, 1.29) 0.58 0.86 (0.57, 1.31) 0.48
Ethnicity (ref = non-Hispanic) 0.86 (0.71, 1.04) 0.13 0.96 (0.78, 1.19) 0.74
Primary diagnosis (ref = bronchiolitis) 0.02 0.07
 Pneumonia 2.11 (1.74, 2.55) <0.001 1.31 (1.05, 1.64) 0.02
 Acute respiratory failure related to sepsis 1.98 (1.46, 2.69) <0.001 1.17 (0.79, 1.73) 0.44
 Asthma or reactive airway disease 1.23 (0.93, 1.61) 0.13 0.64 (0.46, 0.88) 0.006
 Aspiration pneumonia 1.72 (1.19, 2.47) 0.004 1.00 (0.70, 1.43) 0.99
 Other 1.85 (1.28, 2.67) 0.001 1.13 (0.77, 1.64) 0.53
Income quartile (ref = Low) 0.10 0.96
 Low Middle 1.14 (0.92, 1.41) 0.23 1.05 (0.84, 1.32) 0.67
 High Middle 1.17 (0.93, 1.47) 0.17 1.04 (0.82, 1.31) 0.78
 High 1.35 (1.07, 1.70) 0.01 1.07 (0.81, 1.41) 0.64
Open in a new tab

PARDS = Pediatric Acute Respiratory Distress Syndrome, OR = Odds Ratio, CI = Confidence Interval.

PARDS severity was defined using the 2015 Pediatric Acute Lung Injury Consensus Conference (PALICC) criteria (2015). Higher scores are associated with greater illness severity.

a

Odds ratio > 1 indicates greater risk of having a higher level of PARDS.

b

Variables significant at p <0.20 in unadjusted analyses were included in the multivariable analysis.

c

Race and ethnicity were included in the multivariable analysis to control for potential confounding with income.

Age was significantly associated with illness severity: younger children had less severe diagnoses (e.g. bronchiolitis), lower PRISM III-12 scores, and lower PARDS severity. Due to the strong association between age and illness severity, as well as the finding that there were more older children in the higher income quartiles, regression models were created for each age group separately. No statistically significant associations were found between income quartiles and either PRISM III-12 or PARDS for any of the age groups.

Discussion

In this geographically diverse cohort of children with acute respiratory failure, income was not associated with illness severity upon PICU admission in multivariable models. While PARDS severity was independently associated with high income in the unadjusted model, this did not hold when controlled for age group, gender, race, ethnicity, and primary diagnosis. These findings indicate that utilizing income as a proxy for SES is not associated with mortality risk on PICU admission or PARDS severity. While this is true, it further illustrates the need for robust clinical research, whose design includes prospectively collected SES and SDOH variables such as insurance status and other social and community contextual factors as primary outcomes. In doing so, researchers may be better suited to draw conclusions, provide comprehensive recommendations and create interventions to best support critically ill children of varying socioeconomic backgrounds with unique SDOHs.

Interestingly, the fourth iteration of PRISM (PRISM IV) which aims to predict pediatric risk of mortality, was introduced in 2016, post-Affordable Care Act. PRISM IV briefly addressed socioeconomic factors in its development and validation phase but excluded them from the overall mortality risk algorithm (van Keulen, Polderman, & Gemke, 2005). Addressing socioeconomic factors, such as insurance, PRISM over-predicted mortality of commercially insured children and under-predicted the mortality of those with Medicaid or in the Children’s Health Insurance Program (CHIP), which are both federally and state funded programs primarily based on household income and associated with lower SES.

Unlike the PRISM III-12 score, PARDS focuses on illness severity within a single body system (respiratory). The weak correlation suggests that these two parameters are measuring different aspects of pediatric criticality and are independently relevant. Two retrospective studies using chart review, as opposed to prospectively collected data from recruited and consenting families, found an association between median income and overall illness severity (Epstein et al., 2014; Naclerio et al., 1999). One of these studies took place in a predominantly Latino city (Epstein et al., 2014) and included race in regression models; the other did not report on race and was conducted in the Washington D.C./Baltimore area, which has a predominantly Black population as compared with most other U.S. cities (Naclerio et al., 1999). These two homogenous and centralized populations differ from our geographically diverse population. Additionally, both Epstein et al. (2014) and Naclerio et al. (1999), assessed overall severity of illness, utilizing the Pediatric Index of Mortality 2 (PIM2) and PRISM-III respectively. The present study assessed both overall illness severity (PRISM-III 12) and PARDS severity (OI/OSI), which shows differences in measurement and assessment.

The present study included the following proportions by race compared with U.S. population proportions in 2011: White: 68.6% (study) vs 74.2% (U.S.); Black: 21.2% (study) vs 12.6% (U.S.); multiracial: 5.0% (study) vs 2.7% (U.S.) (US Census Bureau, 2019b). In a secondary analysis of the RESTORE study, there were racial disparities in parental consent. Compared with non -Hispanic White families, fewer non-Hispanic Black families were approached for consent and subsequently fewer Black and Hispanic families provided consent (Natale et al., 2017). This is due in part to limited availability of non-Hispanic Black and Hispanic families during the time in which approach for consent was scheduled by the various PICUs. While definite conclusions cannot be drawn, this inconsistency in consent and approach for consent may be attributed to a complex interplay of work commitments, additional family responsibilities, and/or childcare. While there were identified disparities in consent rates, the sample of children included in the RESTORE clinical trial adequately reflects the general population of United States’ PICUs. A large secondary analysis (n = 80,739) using the Virtual PICU database (all U.S. PICUs) depicted a population racially disproportionate to that of the U.S.: 55.1% non-Hispanic White, 17.5% Black, and 4.3% mixed race (Epstein et al., 2011). The remaining categories and proportions were: 16.8% Hispanic, 2.9% Asian, and 3.4% unspecified. This may suggest that the RESTORE study’s population was on par with the racial makeup of PICU populations, which tend to be overrepresented by minorities.

The income distribution of our study may not accurately represent the U.S. income spectrum: the sample’s median income was lower compared to that of the entire U.S. in 2011 ($61,619) but it included very few children from impoverished tracts (US Census Bureau, 2019b). In 2011, the overall poverty rate for the U.S. was 15.0% and specifically for children living in poverty, 21.9% (US Census Bureau, 2019b).The U.S. guidelines for living below the poverty threshold are calculated based on family size and while the family sizes for our study sample subjects are unknown, if each subject were to belong to a four-person household, only 5.9% of our sample would fall below the poverty threshold.

Income is a commonly used SES proxy, and at the census level can be analyzed at the state, city, tract, or block level, with census tracts comparable to neighborhoods and averaging 4,000 people, depending on population density (Grimes, 2011; Oakes & Rossi, 2003). Epidemiological studies comparing the use of tracts vs. block groups when analyzing socioeconomic factors did not find significant advantages to examining populations at the block group level; however, tract- and block group-derived SES did not match ZIP Code-derived SES (Diez-Roux et al., 2001; Krieger et al., 2002).

Strengths and Limitations

While census tract data as a measure of SES is common, this study presents the use of this methodology in a geographically and socioeconomically diverse population of critically ill children. Additionally our analysis includes an exploration into the association of not only overall illness severity, but PARDS severity, a single system measure.

We approximate SES in a secondary analysis of data from a rigorously conducted randomized controlled trial. Of importance, neither the parent study (RESTORE) nor this study was designed to test for causal effects of family income on illness severity measures. Due to the primary aims of the parent study (RESTORE), limited data regarding SES and/or SDOH was collected. The strength of this secondary data analysis is an innovative means of approximating SES via census-derived income. RESTORE did not include census-derived median income as a predetermined variable of interest, and a substantial amount of its parent-reported SES data was missing. The census-based methods described in this study were executed with accuracy and we believe they provide reasonable approximations to individual-level income data. However, we acknowledge that the same dollar value for median income in a Midwestern small town versus a Northeastern large city may not represent the same value and access to goods. Our data reflected several trends that provide support for the use of this census-derived income method: children in the lowest income quartile were more likely to be Black or Hispanic and have a history of asthma or prematurity; children in the highest income quartile were more likely to be White, with markedly lower rates of asthma or prematurity. These demographic and health-related trends are well documented in U.S. health disparity literature (Flores & Committee On Pediatric Research, 2010; Gern, 2010). It is possible that our census tract-derived median income variable did not reflect the substantial differences in income when stratified by racial identity (DeNavas-Walt, Proctor, & Smith, 2012). Furthermore, minorities may experience more neighborhood-based poverty and less access to societal resources impacting HRQOL, regardless of individual income level (Intrator, Tannen, & Massey, 2016).

In addition, our study only included residential addresses from about a third of PICU non-survivors and while they were equally distributed across the four income quartiles, we cannot be certain this pattern would have continued. Self-reported income can be intentionally misreported, due to its association with the receiving of government aid and resources, taxes, and social status (Oakes, n.d.). It is typically “missing not at random,” with high-income and low-income individuals withholding this information and skewing data (Dong & Peng, 2013). Nonetheless, there is a margin of error of varying magnitude for each census-collected median income value related to missing data. An alternative to an income-driven SES measure includes robust SDOH measures, such as self-reported education level or insurance type, which may have more substantially impacted illness severity because of associations with delay of care, healthcare access, lack of consistent primary care, and differences in recognizing and reacting to a developing illness course. Our results highlight the need for a more comprehensive approach to assessing SES, with an emphasis on SDOH as well.

Social determinants of health are known contributors to health outcomes, including but not limited to health disparities (Healthy People 2030, 2021). Employing a SDOH framework when conducting health outcomes research may allow for the identification of factors that place someone in a particular SES category and could provide a more nuanced depiction of how and why health is affected by these factors. Berman, Patel, and Belamarich (2018) demonstrated that poverty-related SDOH could be collected and utilized in real time during the pediatric clinical encounter. Screenings could help connect pediatric patients and their families with the resources and care that they need.

Lastly, we acknowledge that the RESTORE clinical trial took place from 2009–2013, whereas our median reported incomes are from 2011 (RESTORE’s midpoint). We acknowledge that much change occurred in the American healthcare landscape post-RESTORE in 2014, including provisions instated by the Affordable Care Act. Even though Medicaid expansion was part of the ACA, states had varying eligibility criteria for individuals, including qualifying income, and not all states opted to expand. While overall disparities still persist, the enactment of the ACA has reduced disparities in insurance coverage in racial and ethnic minorities, providing increased access particularly in states that expanded Medicaid (Buchmueller et al., 2016). This is even more reason for researchers to consider SDOH variables in their research to account for this variability to access and equitable care.

Conclusions

To our knowledge, this is the first paper that uses census tract data to explore the association between severity of illness and SES in a large, geographically diverse cohort of critically ill children. However, as we operationalized SES, it did not appear to have an association with presenting illness severity in children with acute respiratory failure. As more robust and reliable methods for measuring SES are developed, we may be able to better explain the mechanisms by which SES affects critical illness, particularly those which address not only SES, but overall SDOHs that may impact a critical ill child’s illness severity.

Acknowledgement:

Research reported in this publication was supported by the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health under Award Number U01HL086622 and Hillman Scholars Program. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors thank the RESTORE Investigators (see Acknowledgements Supplement) for their hard work in carrying out this research. The authors have no conflicts of interest to report.

Acknowledgements Supplement

The authors gratefully acknowledge the work of the RESTORE study investigators,: Martha A.Q. Curley (Principal Investigator; School of Nursing and the Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; Critical Care and Cardiovascular Program, Boston Children’s Hospital, Boston, MA); David Wypij, (Principal Investigator – Data Coordinating Center; Department of Biostatistics, Harvard T.H. Chan School of Public Health; Department of Pediatrics, Harvard Medical School; Department of Cardiology, Boston Children’s Hospital, Boston, MA); Geoffrey L. Allen (Children’s Mercy Hospital, Kansas City, MO); Derek C. Angus (Clinical Research, Investigation, and Systems Modeling of Acute Illness Center, Pittsburgh, PA); Lisa A. Asaro (Department of Cardiology, Boston Children’s Hospital, Boston, MA); Judy A. Ascenzi (The Johns Hopkins Hospital, Baltimore, MD); Scot T. Bateman (University of Massachusetts Memorial Children’s Medical Center, Worcester, MA); Santiago Borasino (Children’s Hospital of Alabama, Birmingham, AL); Cindy Darnell Bowens (Children’s Medical Center of Dallas, Dallas, TX); G. Kris Bysani (Medical City Children’s Hospital, Dallas, TX); Ira M. Cheifetz (Duke Children’s Hospital, Durham, NC); Allison S. Cowl (Connecticut Children’s Medical Center, Hartford, CT); Brenda L. Dodson (Department of Pharmacy, Boston Children’s Hospital, Boston, MA); E. Vincent S. Faustino (Yale-New Haven Children’s Hospital, New Haven, CT); Lori D. Fineman (University of California San Francisco Benioff Children’s Hospital at San Francisco, San Francisco, CA); Heidi R. Flori (University of California at San Francisco Benioff Children’s Hospital at Oakland, Oakland, CA); Linda S. Franck (University of California at San Francisco School of Nursing, San Francisco, CA); Rainer G. Gedeit (Department of Pediatrics, Medical College of Wisconsin, Milwaukee, WI); Mary Jo C. Grant (Primary Children’s Hospital, Salt Lake City, UT); Andrea L. Harabin (National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD); Catherine Haskins-Kiefer (Florida Hospital for Children, Orlando, FL); James H. Hertzog (Nemours/Alfred I. duPont Hospital for Children, Wilmington, DE); Larissa Hutchins (The Children’s Hospital of Philadelphia, Philadelphia, PA); Aileen L. Kirby (Oregon Health & Science University Doernbecher Children’s Hospital, Portland, OR); Ruth M. Lebet (School of Nursing, University of Pennsylvania, Philadelphia, PA); Michael A. Matthay (University of California at San Francisco School of Medicine, San Francisco, CA); Gwenn E. McLaughlin (Holtz Children’s Hospital, Jackson Health System, Miami, FL); JoAnne E. Natale (University of California Davis Children’s Hospital, Sacramento, CA); Phineas P. Oren (St. Louis Children’s Hospital, St. Louis, MO); Nagendra Polavarapu (Advocate Children’s Hospital-Oak Lawn, Oak Lawn, IL); James B. Schneider (Cohen Children’s Medical Center of New York, Hyde Park, NY); Adam J. Schwarz (Children’s Hospital of Orange County, Orange, CA); Thomas P. Shanley (C. S. Mott Children’s Hospital at the University of Michigan, Ann Arbor, MI); Shari Simone (University of Maryland Medical Center, Baltimore, MD); Lewis P. Singer (The Children’s Hospital at Montefiore, Bronx, NY); Lauren R. Sorce (Ann & Robert H. Lurie Children’s Hospital of Chicago, Chicago, IL); Edward J. Truemper (Children’s Hospital and Medical Center, Omaha, NE); Michele A. Vander Heyden (Children’s Hospital at Dartmouth, Dartmouth, NH); R. Scott Watson (Center for Child Health, Behavior, and Development, Seattle Children’s Research Institute, Seattle, WA); Claire R. Wells (University of Arizona Medical Center, Tucson, AZ).

Appendix

Appendix A: The United States census as a data source and methods for connecting census and subject data

The mission of the United States (U.S.) Census Bureau is to “serve as the leading source of quality data about the nation’s people and economy” (n.d). As mandated by Article I, Section 2 of the U.S. Constitution, the census counts each U.S. resident every ten years; in addition to achieving the most accurate population count, the census asks residents a series of demographic questions, including number of people living in the household and self-identified race. The census survey has existed in both short and long form, but in 2005 the American Community Survey (ACS) was created to replace the long form; the ACS is administered annually to 3.5 million American residents, roughly 1% of the total U.S. population. The ACS is the primary source for socioeconomic status-related information on both population and neighborhood levels. Based on this demographic and socioeconomic collected information, federal funds are then allocated to schools, hospitals, public works, and roads (US Census Bureau, 2020). Data for both the decennial census survey and the ACS are obtained via internet, mail, phone, and in-person, with response rates for both near 95%.

The census divides the U.S. into geographically smaller meaningful units, allowing for a variety of data analyses at the ZIP Code, tract, and block group levels. The United States Postal Service (USPS) created ZIP Codes—Zone Improvement Plan Codes— in order to make mail delivery more efficient, but they can span large areas comprised of socioeconomically heterogeneous populations. Census tracts, on the other hand, are “small, relatively permanent statistical subdivisions of a county” and are “designed to be homogeneous with respect to population characteristics, economic status, and living conditions” (Grimes, 2011).

Each RESTORE subject is represented by a unique identification number, which is linked to a residential address, including street number and name, city, state, and ZIP Code. Residential addresses were initially reviewed for misspellings (e.g. Walnut Streat) and edited as needed. We created “Dataset A,” which included the street number and name, city, state, and ZIP code for each RESTORE subject. Dataset A was loaded into the Census Geocoder where each residential address was matched with measures of latitude and longitude as well as a 2-digit state code, 3-digit county code, and 6-digit tract code (US Census Bureau, 2019b). These three numerical strings were concatenated to form a unique 11-digit code. Residential addresses that returned either a “no match” or a “non-exact match” were checked against the RESTORE study’s original case report forms (CRFs). Online searches often revealed misspellings or transposed numbers originating from the CRFs. Dataset A was continuously cleaned and loaded into the Geocoder until all residential addresses were linked to an 11-digit code.

“Dataset B” was created by downloading median income for every census tract in the United States from the Census Bureau’s “American Fact Finder” (US Census Bureau, 2019a). This dataset also included the 11-digit code for each tract. Microsoft Excel was used to link the two datasets by merging queries on the 11-digit code. Fifty subjects chosen at random were manually geocoded and individually matched to median income via “American Fact Finder” in order to perform an accuracy check.

Footnotes

Ethical Conduct of Research: This research is a secondary data analysis of the RESTORE clinical trial and follow-up. Institutional Review Board approval and parental consent were obtained prior to data analysis.

Clinical Trial Registration: The Sedation Management in Pediatric Patients With Acute Respiratory Failure (RESTORE) clinical trial PI: Curley; U01HL086622) was funded by NHLBI and registered in the US National Library of Medicine’s ClinicalTrials.gov on December 2008 (NCT00814099). The trial began in January 2009 and was completed in December 2013. A link to RESTORE’s registry is as follows, https://clinicaltrials.gov/ct2/show/NCT00814099

Contributor Information

Alicia G. Kachmar, Department of Family and Community Health, School of Nursing, University of Pennsylvania, Philadelphia, Pennsylvania; Hillman Scholars Program in Nursing Innovation.

David Wypij, Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts; Department of Cardiology, Boston Children’s Hospital, Boston, Massachusetts; Department of Pediatrics, Harvard Medical School, Boston, Massachusetts.

Mallory A. Perry, The Children’s Hospital of Philadelphia Research Institute, Philadelphia, Pennsylvania.

Martha AQ Curley, Department of Family and Community Health, School of Nursing, University of Pennsylvania, Philadelphia, Pennsylvania; Department of Anesthesia and Critical Care, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; The Children’s Hospital of Philadelphia Research Institute, Philadelphia, Pennsylvania.

References

  1. Adler NE, Boyce T, Chesney MA, Cohen S, Folkman S, Kahn RL, & Syme SL (1994). Socioeconomic status and health. The challenge of the gradient. American Psychologist, 49(1), 15–24. doi: 10.1037//0003-066x.49.1.15 [DOI] [PubMed] [Google Scholar]
  2. Agency for Healthcare Research and Quality (AHRQ). (2019). National healthcare quality and disparities Report: Introduction and methods. Rockville, MD: AHRQ. [Google Scholar]
  3. American Psychological Association (APA) Task Force on Socioeconmic Status (2007). Report on the APA Task Force on Socioeconomic Status. Washington, DC: American Psychological Association. [Google Scholar]
  4. Berman RS, Patel MR, Belamarich PF, & Gross RS (2018). Screening for poverty and poverty-related social determinants of health. Pediatrics in Review (39), 235–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bewick J, Cheek L, & Ball J. (2004). Statistics review 10: Further nonparametic methods. Pediatric Critical Care Medicine, 8, 196–199. doi: 10.1186/cc2857 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Braveman P, Cubbin C, Egerter S, Chideya S, Marchi K, Metzler M, & Posner S. (2005). Socioeconomic status in health research: one size does not fit all. JAMA, 294(22), 2879–2888. doi: 10.1001/jama.294.22.2879 [DOI] [PubMed] [Google Scholar]
  7. Braveman PA, Cubbin C, Egerter S, Williams DR, & Pamuk E. (2010). Socioeconomic disparities in health in the United States: What the patterns tell us. American Journal of Public Health, 100(Suppl 1), S186–S196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Buchmueller TC, Levinson ZM, Levy HG, & Wolfe BL (2016). Effect of the Affordable Care Act on Racial and Ethnic Disparities in Health Insurance Coverage. American journal of public health, 106(8), 1416–1421. 10.2105/AJPH.2016.303155 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen E. (2004). Why socioeconomic status affects the health of children: a psychosocial perspective. Current Directions in Psychological Science, 12, 112–115. doi: 10.1111/j.0963-7214.2004.00286.x [DOI] [Google Scholar]
  10. Cheng TL, Goodman E, & Committee on Pediatric Research (2015). Race, ethnicity, and socioeconomic status in research on child health. Pediatrics, 135(1), e225–e237. 10.1542/peds.2014-3109 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Curley MAQ, Wypij D, Watson RS, Grant MJ, Asaro LA, Cheifetz IM, Dodson BL, Franck LS, Gedeit RG, Angus DC, Matthay MA, & RESTORE Study Investigators and the Pediatric Acute Lung Injury and Sepsis Investigators Network (2015). Protocolized sedation vs usual care in pediatric patients mechanically ventilated for acute respiratory failure: a randomized clinical trial. JAMA, 313(4), 379–389. doi: 10.1001/jama.2014.18399 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Curley MAQ, Gedeit RG, Dodson BL, Amling JK, Soetenga DJ, Corriveau CO, Asaro LA, Wypij D, RESTORE Investigative Team (2018). Methods in the design and implementation of the Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) clinical trial. Trials, 19(1), 687. doi: 10.1186/s13063-018-3075-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chetty R, Stepner M,A, Abraham S, Lin S, Scuderi B, Turner N, Bergeron A, Cutler D. (2016). The association between income and life expectancy in the United States, 2001–2014. JAMA, 315(16):1750–1766. doi: 10.1001/jama.2016.4226 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dahl GB & Lochner L. The impact of family income on child achievement: evidence from the earned income tax credit. (2012). American Economic Review, 1025, 1927–1956. [Google Scholar]
  15. DeNavas-Walt C, Proctor B, & Smith J. (2012). Income, poverty, and health insurance coverage in the United States: 2011. Retrieved from https://www.census.gov/prod/2012pubs/p60-243.pdf [Google Scholar]
  16. Diez-Roux AV, Kiefe CI, Jacobs DR, Haan M, Jackson SA, Nieto FJ, Paton C, Schulz R. (2001). Area characteristics and individual-level socioeconomic position indicators in three population-based epidemiologic studies. Annals of Epidemiology, 11(6), 395–405. doi: 10.1016/s1047-2797(01)00221-6. [DOI] [PubMed] [Google Scholar]
  17. Dong Y, & Peng CY (2013). Principled missing data methods for researchers. Springerplus, 2(1), 222. doi: 10.1186/2193-1801-2-222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Duncan GJ, Magnuson K, & Votruba-Drzal E. (2017). Moving beyond correlations in assessing the consequences of poverty. Annual Review of Psychology, 3(68), 413–434. Doi: 10.1146/annurev-psych-010416-044224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Epstein D, Reibel M, Unger JB, Cockburn M, Escobedo LA, Kale DC, Chang JC, Gold JI (2014). The effect of neighborhood and individual characteristics on pediatric critical illness. Journal of Community Health, 39(4), 753–759. doi: 10.1007/s10900-014-9823-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Epstein D, Wong C, Khemani R, Moromisato D, Waters K, Kipke M, & Markovitz B. (2011). Race/Ethnicity is not associated with mortality in the PICU. Pediatrics, 127, e588–597. doi: 10.1542/peds.2010-0394 [DOI] [PubMed] [Google Scholar]
  21. Fiser DH (1992). Assessing the outcome of pediatric intensive care. The Journal of Pediatrics, 121(1), 68–74. doi: 10.1016/S0022-3476(05)82544-2 [DOI] [PubMed] [Google Scholar]
  22. Flores G, & Committee on Pediatric Research. (2010). Technical report--racial and ethnic disparities in the health and health care of children. Pediatrics, 125(4), e979–e1020. doi: 10.1542/peds.2010-0188 [DOI] [PubMed] [Google Scholar]
  23. Gern JE (2010). The urban environment and childhood asthma study. Journal of Allergy and Clinical Immunology, 125, 545–549. doi: 10.1016/j.jaci.2010.01.037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Grimes Z. (2011). Census Tracts vs. Census Block Groups. Retrieved from https://current360.com/research-101-census-tracts-vs-census-block-groups/ [Google Scholar]
  25. Healthy People 2030 (2021). Social Determinants of Health. Retrieved from https://health.gov/healthypeople/objectives-and-data/social-determinants-health [Google Scholar]
  26. Intrator J, Tannen J, & Massey DS (2016). Segregation by race and income in the United States 1970–2010. Social Science Research, 60, 45–60. doi: 10.1016/j.ssresearch.2016.08.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kachmar AG, Connolly CA, Wolf S, & Curley MAQ (2019). Socioeconomic status in pediatric health research: A scoping review. Journal of Pediatrics, 213, 163–170. doi: 10.1016/j.jpeds.2019.06.005 [DOI] [PubMed] [Google Scholar]
  28. Kielb ES, Rhyan CN, & Lee JA (2017). Comparing health care financial burden with an alternative measure of unaffordability. Inquiry, 54, 46958017732960. doi: 10.1177/0046958017732960 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, & Carson R. (2002). Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: does the choice of area-based measure and geographic level matter?: the Public Health Disparities Geocoding Project. American Journal of Epidemiology, 156(5), 471–482. doi: 10.1093/aje/kwf068 [DOI] [PubMed] [Google Scholar]
  30. Krieger N, Chen JT, Waterman PD, Renkopf DH, & Subramanian SV (2005). Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: The public health disparities geocoding project. American Journal of Public Health, 95(2), 312–323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lacroix J, Cotting J, & Pediatric Acute Lung Injury and Sepsis Investigators (2005). Severity of illness and organ dysfunction scoring in children. Pediatric Critical Care Medicine, 6(3 Suppl), S126–134. doi: 10.1097/01.PCC.0000161287.61028.D4 [DOI] [PubMed] [Google Scholar]
  32. Lopez AM, Tilford JM, Anand KJ, Jo CH, Green JW, Aitken ME, Fiser DH Variation in pediatric intensive care therapies and outcomes by race, gender, and insurance status. Pediatric Criical Care Medicine 2006; 7: 2–6. doi: 10.1097/01.pcc.0000192319.55850.81 [DOI] [PubMed] [Google Scholar]
  33. Naclerio AL. Gardner JW, & Pollack MM (1999). Socioeconomic factors and emergency pediatric ICU admissions. Annals of the New York Academy of Science, 896, 379–382. doi: 10.1111/j.1749-6632.1999.tb08147.x [DOI] [PubMed] [Google Scholar]
  34. Natale JE, Lebet R, Joseph JG, Ulysse C, Ascenzi J, Wypij D, Curley MAQ, Randomized Evaluation of Sedation Titration for Respiratory Failure (RESTORE) Study Investigators. (2017). Racial and ethnic disparities in parental refusal of consent in a large, multisite pediatric critical care clinical trial. Journal of Pediatrics, 184, 204–208.e201. doi: 10.1016/j.jpeds.2017.02.006 [DOI] [PubMed] [Google Scholar]
  35. Nepomnyaschy L. (2009). Socioeconomic gradients in infant health across race and ethnicity. Maternal and Child Health Journal, 13(6), 720–731. doi: 10.1007/s10995-009-0490-1 [DOI] [PubMed] [Google Scholar]
  36. Oakes J. (n.d). Measuring Socioeconomic Status. Retrieved from http://www.esourceresearch.org/eSourceBook/TheMeasurementofSocioeconomicStatus/1LearningObjectives/tabid/767/Default.aspx [Google Scholar]
  37. Oakes J, & Rossi P. (2003). The measurement of SES in health research: Current practices and steps towards a new approach. Social Science and Medicine, 56, 769–784. doi: 10.1016/s0277-9536(02)00073-4 [DOI] [PubMed] [Google Scholar]
  38. Pollack MM, Holubkov R, Funai T, Dean JM, Berger JT, Wessel DL, Meert K, Berg RA, Newth CJL, Harrison RE, Carciollo J, Dalton H, Shanley T, Jenkins TL, Tamburro R, & for the Eunice Kennedy Shriver National Institute of Child Health and Human Development Collaborative Pediatric Critical Care Research Network. (2016). The Pediatric Risk of Mortality Score: Update 2015. Pediatric Critical Care Medicine, 17(1), 2–9. doi: 10.1097/PCC.0000000000000558 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Pollack MM, Patel KM, & Ruttimann UE (1996). PRISM III: an updated Pediatric Risk of Mortality score. Critical Care Medicine, 24(5), 743–752. doi: 10.1097/00003246-199605000-00004 [DOI] [PubMed] [Google Scholar]
  40. Ronfani L, Vecchi Brumatti L, Mariuz M, Tognin V, Bin M, Ferluga V, Knowles A, Montico M, Barbone F. (2015). The complex interaction between home environment, socioeconomic status, maternal iq and early child neurocognitive development: a multivariate analysis of data collected in a newborn cohort study. PLoS One, 10(5), e0127052. doi: 10.1371/journal.pone.0127052 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sampson R, Morenoff J, & Gannon-Rowley T. (2002). Assessing “neighborhood effects”: social processes and new directions in research. Annual Review of Sociology, 28, 443–478. doi: 10.1146/annurev.soc.28.110601.141114 [DOI] [Google Scholar]
  42. Shipman SA, Lan J, Chang CH, & Goodman DC (2011). Geographic maldistribution of primary care for children. Pediatrics, 127(1), 19–27. doi: 10.1542/peds.2010-0150 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Shonkoff JP & Garner AS (2012). The lifelong effects of childhood adversity and toxic stress. American Academy of Pediatrics, 129, e232–e246. doi: 10.1542/peds.2011-2663 [DOI] [PubMed] [Google Scholar]
  44. The Pediatric Acute Lung Injury Consensus Group. (2015). Pediatric acute respiratory distress syndrome: consensus recommendations from the Pediatric Acute Lung Injury Consensus Conference. Pediatric Critical Care Medicine, 16(5), 428–439. doi: 10.1097/PCC.0000000000000350 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. U.S. Bureau of Congress (2019a). Census Geocoder. Retrieved from https://geocoding.geo.census.gov/
  46. U.S. Bureau of Congress (2019b). American Fact Finder. Retrieved from https://data.census.gov/cedsci/
  47. van Keulen JG, Polderman KH, & Gemke RJ (2005). Reliability of PRISM and PIM scores in paediatric intensive care. Archives of Disease in Childhood, 90(2), 211–214. doi: 10.1136/adc.2003.046722 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Watson RS, Asaro LA, Hertzog JH, Sorce LR, Kachmar AG, Dervan LA, Angus DC, Wypij D, Curley MAQ, RESTORE Study Investigators and the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network (2018). Long-Term Outcomes after Protocolized Sedation versus Usual Care in Ventilated Pediatric Patients. American Journal of Respiratory and Critical Care Medicine, 197(11), 1457–1467. doi: 10.1164/rccm.201708-1768OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Williams DR, Priest N, & Anderson NB (2016). Understanding associations among race, socioeconomic status, and health: Patterns and prospects. Health Psychology, 35(4), 407–411. Doi: 10.1037/hea0000242 [DOI] [PMC free article] [PubMed] [Google Scholar]

Appendix A references

  1. Grimes Z. (2011). Census Tracts vs. Census Block Groups. Retrieved from https://current360.com/research-101-census-tracts-vs-census-block-groups/
  2. U.S. Census Bureau. (n.d). What We Do. Retrieved from https://www.census.gov/about/what.html#par_textimage
  3. U.S. Census Bureau. (2020). Retrieved from https://www.census.gov/programs-surveys/decennial-census/2020-census/about/why.html
  4. U.S. Bureau of Congress (2019a). American Fact Finder. Retrieved from https://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml
  5. U.S. Bureau of Congress (2019b). Census Geocoder. Retrieved from https://geocoding.geo.census.gov/

RESOURCES