Black youth with type 1 diabetes (T1D) are at heightened risk for suboptimal glycemic control. Studies of neighborhood effects on the health of youth with T1D are limited. The current study investigated the effects of racial residential segregation on the diabetes health of young Black adolescents with T1D.
A total of 148 participants were recruited from 7 pediatric diabetes clinics in 2 US cities. Racial residential segregation (RRS) was calculated at the census block group level based on US Census data. Diabetes management was measured via self-report questionnaire. Hemoglobin A1c (HbA1c) information was gathered from participants during home-based data collection. Hierarchical linear regression was used to test the effects of RRS while controlling for family income, youth age, insulin delivery method (insulin pump versus syringe therapy), and neighborhood adversity.
HbA1c was significantly associated with RRS in bivariate analyses, whereas youth-reported diabetes management was not. In hierarchical regression analyses, whereas family income, age, and insulin delivery method were all significantly associated with HbA1c in model 1, only RRS, age, and insulin delivery method were significantly associated with HbA1c in model 2. Model 2 explained 25% of the variance in HbA1c (P = .001).
RRS was associated with glycemic control in a sample of Black youth with T1D and accounted for variance in HbA1c even after controlling for adverse neighborhood conditions. Policies to reduce residential segregation, along with improved screening for neighborhood-level risk, hold the potential to improve the health of a vulnerable population of youth.
Living in neighborhoods characterized by adversity increases risk of suboptimal glycemic control in youth with type 1 diabetes. However, the specific effects of racial residential segregation, a root social determinant of inequity, have not been investigated in this population.
Black youth with type 1 diabetes living in 2 Midwestern cities had poorer glycemic control when residing in more segregated neighborhoods. Racial residential segregation accounted for variance in glycemic control even after controlling for effects of neighborhood adversity.
Black youth with type 1 diabetes (T1D) are at heightened risk for suboptimal glycemic control,1 higher rates of hospital admissions for diabetic ketoacidosis,2 and early mortality.3 Although risks and assets at multiple levels of influence within the broader social ecology are likely to affect the health of youth with T1D,4 the majority of published studies have focused on risk and protective factors either at the level of the individual child, such as youth psychosocial adjustment5 or the use of diabetes management technologies,6 or at the level of the family, such as diabetes-related family functioning7 or family resources.8 Recent calls for an increased focus on health equity in pediatric diabetes have highlighted the influence of community-level risk factors in creating disparities9 and the need to better understand the specific mechanisms by which extrafamilial risk and protective factors affect the health of youth of color with T1D. Consistent with such calls, a few studies have begun to investigate the effects of neighborhood factors such as neighborhood-level income,10 disorder,11 and disadvantage12 on glycemic control in youth with T1D. Two of these studies found that living in neighborhoods characterized by higher rates of poverty, public assistance, unemployment, and/or crime increased risk for suboptimal control. However, both studies enrolled predominantly white samples. A study by our group extended these findings to a sample of young, urban Black adolescents. Results demonstrated that within a sample of low-income Black youth living in neighborhoods with relatively high levels of adversity, higher neighborhood adversity scores were associated with poorer youth glycemic control.13
Given the growing literature on neighborhoods’ impact on the health of youth with T1D, understanding whether specific characteristics of neighborhoods are more predictive of outcomes than others could be critical to identifying those youth at highest risk. Racial residential segregation (RRS) refers to the geographic separation of Blacks or other persons of color from other racial/ethnic groups14 and is a form of structural racism that has previously been identified as a critical determinant of inequities in health15 and health care.16 RRS is theorized to affect health by decreasing access to opportunities and increasing disadvantage, as well as lowering socioeconomic status.17 However, because of the US history of disinvestment of economic resources in predominantly Black neighborhoods, highly segregated neighborhoods are also characterized by a number of factors that interfere with diabetes management, including inadequate access to resources for physical activity, such as green spaces and sidewalks, as well as access to healthy foods.18 Although such findings suggest that RRS might indirectly affect glycemic control through its adverse effects on diabetes management, RRS might also directly affect the health of Black youth with T1D through stress-related pathways that influence glycemic control. These pathways could include increased exposure to psychosocial stressors such as racism. For example, a study in adults showed that Black women living in more segregated neighborhoods reported experiencing more episodes of discrimination in their everyday lives than those living in more integrated neighborhoods.19 Conversely, some studies have suggested the potential for segregation to have protective effects on Black residents because residing in areas that reinforce racial identity or traditional forms of community support or empowerment could be beneficial for health.16 In summary, although measures of the concentration of racial or ethnic groups in neighborhoods are an important way to understand segregation’s effects on health outcomes, studies are needed to differentiate its effects from those related neighborhood predictors of poor health such as poverty.20 The inclusion of variables such as composite measures of neighborhood adversity or disadvantage in models predicting health outcomes could help to clarify the specific effects of segregation on the health of Black youth with T1D.
The primary aim of the current study was to test whether RRS was associated with diabetes management and glycemic control among Black youth with T1D. A secondary aim was to determine whether RRS accounted for these health outcomes beyond the effects of neighborhood adversity, because such information could help to inform decision-making regarding those neighborhood indicators that hold highest utility for identifying at-risk youth and how to use these indicators to improve health outcomes.
Methods
Procedure
Participants in the current study were recruited from 3 pediatric diabetes clinics located in the greater metropolitan Detroit area and 4 in Chicago as part of a multicenter clinical trial. To be eligible for the clinical trial, adolescent participants had to be between 10 and 14 years of age, be diagnosed with T1D for at least 6 months, self-identify as Black, and be residing with a caregiver who was willing to participate in the study. No child psychiatric diagnoses were exclusionary, except for moderate or severe cognitive impairment, suicidal ideation, and psychosis. Families were also excluded if they were not English speaking, could not complete study measures in English, or if the child had a medical diagnosis leading to atypical diabetes management (eg, cystic fibrosis). Potential participants were either (1) recruited through letters describing the study followed by phone calls from research staff to assess interest or (2) approached in person at the time of a regularly scheduled clinic visit. The clinical trial investigated the effectiveness of an eHealth intervention to promote optimal glycemic control. The trial was registered at clinicaltrials.gov (no. NCT03168867). The eHealth intervention was delivered at the time of regularly scheduled diabetes clinic visits and provided parenting advice to the adolescent’s primary caregiver regarding ways to support adolescent diabetes management.
The research was approved by the institutional review board of the first author’s university using a single institutional review board agreement. All participants provided informed consent. Parents provided parental permission and youth provided either written or verbal assent to participate based on their age. Data for the current study were drawn from the baseline period before study randomization or receipt of the study intervention and collected by trained research assistants in the participant’s home. Families were provided $50 as compensation for participating in the data collection session.
To calculate scores for neighborhood characteristics of interest for each participant, the MMQGIS add-on package was used within QGIS software21 to geocode the participant’s address to latitude and longitude coordinates and to then link the geocodes to their respective census block group. In 2 cases, youths resided within the same census block group. To avoid any confounding associated with potential lack of independence of the youth-level data and the neighborhood-level data, 1 of these participants was randomly selected and then excluded. Five additional participants resided in areas outside the Detroit or Chicago metropolitan area and were also excluded. The final analyzed sample was 148.
Measures
Sociodemographic and Medical Variables
A self-report questionnaire was used to obtain information from the adolescent’s primary caregiver on demographic variables, including caregiver age and annual family income from all sources. Caregivers were provided with a list of 10 income categories and chose the one that reflected their annual family income. For each category, the midpoint was used in the analyses to represent the income range. The adolescent’s medical chart was reviewed to obtain clinical information such as duration of diabetes and insulin management approach.
Residential Racial Segregation
RRS was calculated at the level of the census block group, a more precise geographic unit of analysis than the census tract, because census tract level measures may obscure neighborhood variability and are also less reliable.22 RRS was assessed based on the concentration-evenness dimension of segregation using the location quotient (LQ) following the methodology established by Brown and Chung.23 The LQ was derived for each census block group within core counties in the greater Detroit and Chicago metrpolitan areas where families in the current study resided. LQs are calculated as LQi = (bi/ti)/ (B/T) where bi and ti are the Black and total population in the block group i; B and T are the Black and total population in the metro area and range from 0 upwards. If the percentage of Blacks in a block group matches its percentage for the urban area overall, LQ = 1; if the percentage in a block group is greater than that for the urban area overall, LQ > 1; if the percentage in a block group is less than that for the urban area overall, LQ < 1. Approaches that compare the racial composition of a neighborhood to that of the larger metropolitan area in which it is embedded have been found to have increased validity in predicting health outcomes over those that do not.24 To infer significance, an LQ of 1.2 or greater indicates a significant concentration of Black residents (high segregation), whereas an LQ of 0.85 or less indicates underrepresentation of Black residents (low segregation).23 Data on racial/ethnic composition of census block groups used to calculate LQs were obtained from the US Census Bureau’s American Community Survey 5-year estimates for 2018, released in 2019.25
Neighborhood Adversity
As described in Ellis et al.,13 the Neighborhood Adversity Index (NAI) comprises 9 neighborhood-level indicators of adversity representing domains of education, employment, housing, occupation, poverty, and sociodemographics available in the US Census Bureau’s American Community Survey. The domains included median household income; percent persons in poverty; percent of households with no vehicle available; percent of persons with less than a 12th grade education, no diploma; percent of households renter occupied; percent females in management occupations; percent males in management occupations; percent of housing units vacant; and percent female headed households. The NAI calculation was informed by an approach validated by Messer and colleagues.26 Five-year estimates for 2018 census block groups (released in 2019)25 were used to calculate census block group NAI scores for the Detroit and Chicago metropolitan areas. Higher (positive) scores represented higher neighborhood adversity. The NAI was included to control for aspects of impoverished neighborhoods that often covary with RRS and to allow the independent effects of RRS to be assessed.
Diabetes Management
Self-management was measured using the Diabetes Management Scale (DMS),27 a self-report questionnaire designed to measure a broad range of diabetes management behaviors, such as insulin management, dietary management, blood glucose monitoring, and symptom response. Each item asks, “What percent of the time do you (take your insulin)?” The response scale is 0% to 100%. A total score is obtained by calculating the mean response to all items to reflect overall management behavior; higher scores indicate higher levels of diabetes management. The measure has previously been shown to be reliable in predominantly Black samples of adolescents with T1D.28
Glycemic Control
Hemoglobin A1c (HbA1c) was used to evaluate glycemic control. Values were obtained during study data collection visits using the Accubase test kit, which is approved by the US Food and Drug Administration. The Accubase test uses a capillary tube blood collection method instead of venipuncture. It is therefore suitable for home-based data collection because it is intended for use by consumers in their home. High-performance liquid chromatography was used to analyze the blood sample.
Analytic Plan
Pearson correlations and hierarchical linear regression were used to test the associations between diabetes health measures and RRS. Insulin delivery method was recoded to a dichotomous variable (syringe therapy = 0, insulin pump = 1). Potential covariates (youth age, insulin delivery method, and family income) were entered in step 1. RRS and NAI were entered in step 2. Rates of missing data were very low overall. Data were missing only for the family income variable (4 values in total), which were estimated with the sample mean. Analyses were conducted using SPSS, version 28.0.
Results
Sample demographics are shown in Table 1. Mean yearly family income was $34 236, corresponding to approximately 125% of the US 2020 poverty line for a family of 4; median family income was $25 000. Mean HbA1c was 11.5% (102 mmol/mol), suggesting that the sample was in suboptimal glycemic control because American Diabetes Association guidelines recommend that for HbA1c to be maintained at or below 7.0% in adolescents.29 Mean LQ was 3.04 (SD = 1.49), indicating residence in highly segregated neighborhoods. More than 80% of families lived in neighborhoods with an LQ greater than or equal to 1.20, indicating high concentrations of Black families (highly segregated) in comparison with the rest of the metro area.
Sample Demographic and Neighborhood Descriptive Variables
Child age, y | 13.3 ± 1.7 |
Child sex | |
Male | 64 (43) |
Female | 84 (57) |
Duration of diabetes, y | 5.8 ± 3.9 |
HbA1c | |
% | 11.5 ± 2.7 |
mmol/mol | 102 ± 30 |
Insulin regimen | |
Basal bolus therapy: injection | 106 (72) |
Basal bolus therapy: pump | 34 (23) |
Other/missing | 8 (5) |
Recruitment site | |
Detroit | 91 (62) |
Chicago | 57 (38) |
Caregiver age, y | 42.2 ± 8.7 |
Annual mean income in US dollars | 34 236 ± 25 477 |
Location quotient | |
≥1.20 (high concentration of Black residents) | 81 (120) |
0.86-1.10 (moderate concentration of Black residents) | 6 (9) |
≤0.85 (low concentration of Black residents) | 13 (19) |
Child age, y | 13.3 ± 1.7 |
Child sex | |
Male | 64 (43) |
Female | 84 (57) |
Duration of diabetes, y | 5.8 ± 3.9 |
HbA1c | |
% | 11.5 ± 2.7 |
mmol/mol | 102 ± 30 |
Insulin regimen | |
Basal bolus therapy: injection | 106 (72) |
Basal bolus therapy: pump | 34 (23) |
Other/missing | 8 (5) |
Recruitment site | |
Detroit | 91 (62) |
Chicago | 57 (38) |
Caregiver age, y | 42.2 ± 8.7 |
Annual mean income in US dollars | 34 236 ± 25 477 |
Location quotient | |
≥1.20 (high concentration of Black residents) | 81 (120) |
0.86-1.10 (moderate concentration of Black residents) | 6 (9) |
≤0.85 (low concentration of Black residents) | 13 (19) |
Data are mean ± SD or n (%).
In bivariate analyses, RRS was significantly associated with HbA1c (r = 0.28, P = .001) but not DMS (r = –0.08, P = .311). Therefore, DMS was not included in the multivariate analyses. NAI and RRS were significantly associated (r = 0.55, P = .001) but the association was moderate, demonstrating that these variables measured different aspects of neighborhoods as expected.
Results of the regression analyses are shown in Table 2. In model 1, family income, age, and insulin management approach were each associated with HbA1c. The model accounted for 19% of the variance in HbA1c (P = 001). The F change associated with model 2 was significant (P = .006), suggesting that adding NAI and RRS to the model accounted for additional variance. The final model accounted for 25% of the variance in HbA1c (P = .001). However, in the final model, only age (β = 0.23, P = .003), insulin delivery method (β = –0.20, P = .01) and RRS (β = 0.18, P = .04) had significant effects on HbA1c. The effects of NAI and family income were not significant.
Hierarchical Multiple Regression Models of Neighborhood Predictors of Glycemic Control With Selected Covariates
. | B . | β . | t . | 95% CI for B . | R . | R2 . | F Change . |
---|---|---|---|---|---|---|---|
Model 1 | 0.44 | 0.19 | 11.31* | ||||
Variable | |||||||
Insulin delivery | −1.06 | –0.17 | −2.13** | −2.05 to –0.07 | |||
Youth age | 0.36 | 0.22 | 2.95* | 0.13 to 0.60 | |||
Family income | –0.02 | –0.27 | −3.51* | −0.02 to –0.02 | |||
Model 2 | 0.50 | 0.25 | 5.35* | ||||
Variable | |||||||
Insulin delivery | −1.00 | –0.16 | −2.04** | −1.97 to –0.03 | |||
Youth age | 0.39 | 0.24 | 3.23* | 0.15 to 0.63 | |||
Family income | –0.01 | –0.15 | −1.86 | −0.01 to –0.08 | |||
NAI | 0.40 | 0.12 | 1.25 | −0.23 to 1.03 | |||
RRS | 0.32 | 0.18 | 2.03b | 0.01 to 0.63 |
. | B . | β . | t . | 95% CI for B . | R . | R2 . | F Change . |
---|---|---|---|---|---|---|---|
Model 1 | 0.44 | 0.19 | 11.31* | ||||
Variable | |||||||
Insulin delivery | −1.06 | –0.17 | −2.13** | −2.05 to –0.07 | |||
Youth age | 0.36 | 0.22 | 2.95* | 0.13 to 0.60 | |||
Family income | –0.02 | –0.27 | −3.51* | −0.02 to –0.02 | |||
Model 2 | 0.50 | 0.25 | 5.35* | ||||
Variable | |||||||
Insulin delivery | −1.00 | –0.16 | −2.04** | −1.97 to –0.03 | |||
Youth age | 0.39 | 0.24 | 3.23* | 0.15 to 0.63 | |||
Family income | –0.01 | –0.15 | −1.86 | −0.01 to –0.08 | |||
NAI | 0.40 | 0.12 | 1.25 | −0.23 to 1.03 | |||
RRS | 0.32 | 0.18 | 2.03b | 0.01 to 0.63 |
P < .01.
P < .05
Discussion
The current study is one of the first to demonstrate the effects of RRS on the glycemic control of youth with T1D. Moreover, the findings suggest that in a sample of urban Black youth, amount of segregation in the youth’s neighborhood was a better predictor of glycemic control than either family income or neighborhood conditions, such as high rates of neighborhood poverty, unemployment, vacant homes, or renter-occupied homes. Results are supportive of previous work contending that residential segregation, a form of structural racism, is the more proximal, or fundamental, cause of health inequity than neighborhood socioeconomic status.20 In samples of adults with type 2 diabetes, mixed evidence has been found to demonstrate that RRS affects glycemic control in the postdiagnostic period.24 However, the methods used in some of these studies, including retrospective approaches that resulted in high rates of missing data, may have reduced the likelihood of detecting effects. In addition, the current study evaluated RRS at the level of the census block group rather than the census tract, which may have improved the precision of the measure.
No association was found between RRS and youth-reported diabetes management; therefore, the study did not find evidence that RRS influenced glycemic control through its potential negative effects on ability to engage in diabetes care tasks such as eating healthy food or engaging in physical activity. Although not measured in the current study, RRS could affect glycemic control via pathways such as exposure to psychosocial stressors such as racism, as such stressors can have a direct effect on glycemic control via dysregulated cortisol.30 Other neighborhood-level influences that were not measured in this study but have previously been shown to covary with RRS include access to health care. For example, Black adults residing in more segregated neighborhoods travel longer distances to obtain health care.31 Fewer visits to pediatric diabetes specialty clinics has previously been demonstrated to be associated with suboptimal glycemic control.32,33 Additional research is needed to determine which aspects of neighborhood are most influential in affecting health outcomes of Black youth with T1D.
Study limitations include the urban nature of the sample; results may not be applicable to youth residing in rural areas. Although segregation in rural Black communities is rising at a faster rate than in urban areas, factors that influence health in segregated rural areas may differ from those found in urban areas.34 Because the sample included only Black youth, findings are not generalizable to youth with T1D from other minoritized backgrounds, such as Latinx youth. Other limitations include the cross-sectional nature of the study and inability to characterize the length of time that youth and their families had resided within the neighborhood and therefore the amount of “exposure” to the effects of segregation.
Conclusions
Findings from the current study add to the growing evidence on aspects of neighborhoods that are linked to health outcomes among youth with T1D and the ways that adverse neighborhood conditions create disparities for Black youth. Although the results of the current study suggest a critical need for policies that address neighborhood inequity as 1 means of improving pediatric diabetes population health (eg, policies to improve access to affordable housing, to decrease redlining),35 they also help to inform strategies for screening youth with T1D to help identify those at highest risk for health disparities. The incorporation of information about community determinants such as neighborhood characteristics into electronic medical records, has been previously recommended by the Institute of Medicine.36 More comprehensive screening of youth with T1D could, in turn, increase the likelihood that factors associated with RRS, such as inadequate access to care, are identified and can be addressed by health care providers.
COMPANION PAPER: A companion to this article can be found online at http://www.pediatrics.org/cgi/doi/10.1542/peds.2022-060800.
Dr Ellis conceptualized the study design and methods, conducted analyses, and wrote the manuscript. Dr Cutchin contributed substantially to the study methods, conducted analyses, and revised the manuscript critically for important intellectual content. Dr Carcone and Ms Worley made a substantial contribution to the acquisition and analyses of data and revised the manuscript critically for important intellectual content. Drs Buggs-Saxton, Boucher-Berry, Dekelbab, Drossos, Evans, Miller, and Weissberg-Benchell made a substantial contribution to the acquisition of data and revised the manuscript critically for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: Funding for the study was provided in part by R01DK110075 from the National Institute of Diabetes and Digestive Kidney Diseases. The NIDDK had no role in the design and conduct of the study.
CONFLICT OF INTEREST DISCLOSURES: Dr Miller’s spouse is a majority owner of Element Bars, Inc, a snack bar company. The other authors have no conflicts of interest to disclose.
Comments