Cross-sectional analysis of racial differences in hydration and neighborhood deprivation in young adults

Abstract

Background

Inadequate hydration is associated with cardiovascular and kidney disease morbidity and all-cause mortality. Compared with White individuals, Black individuals exhibit a higher prevalence of inadequate hydration, which may contribute to racial health disparities. However, the underlying reasons for these differences in hydration remain unclear.

Objective

This cross-sectional study aimed to investigate whether neighborhood deprivation contributes to racial differences in hydration status.

Methods

We assessed 24 Black and 30 White college students, measuring 24-hour urine osmolality, urine flow rate, urine specific gravity, and plasma copeptin concentration. Participants recorded their food and fluid intake for 3 d to assess total water intake from food and beverages. Neighborhood socioeconomic deprivation was measured using a tract-level Area Deprivation Index.

Results

Black participants exhibited higher urine osmolality (640 [314] compared with 440 [283] mOsm/kg H₂O, respectively, P = 0.006) and lower urine flow rate (1.06 [0.65] compared with 1.71 [0.89] ml/min, respectively, P = 0.009) compared with White participants, indicating greater hypohydration among Black participants. Black participants reported lower total water intake from food and beverages than White participants (2.3 ± 0.7 compared with 3.5 ± 1.1 L/day, respectively, P < 0.01). Black participants exhibited higher copeptin than White participants (6.3 [3.1] compared with 4.5 [2.3] pmol/L, P = 0.046), and urine osmolality mediated 67% of the difference (P = 0.027). Black participants reported greater cumulative exposure to neighborhood deprivation during childhood (ages 0–18 y). Furthermore, neighborhood deprivation during childhood was associated with urine specific gravity (P = 0.031) and total water intake from food and beverages (P = 0.042) but did not mediate the racial differences in these measures.

Conclusion

Our data suggest that compared with White young adults, Black young adults are hypohydrated and exhibit higher plasma copeptin concentration, and that greater neighborhood deprivation is associated with chronic underhydration irrespective of race.

This trial was registered at clinicaltrials.gov as NCT04576338.

Introduction

Chronic underhydration (i.e., inadequate hydration status) is associated with all-cause mortality and several risk factors for cardiovascular disease (CVD) and chronic kidney disease (CKD), including obesity, insulin resistance, and high blood pressure (BP) [[1], [2], [3], [4], [5]]. Compared with White adults, Black adults are at an elevated risk of developing hypertension CVD [[6], [7], [8]], and end-stage CKD [[9], [10], [11], [12]]. NHANES data demonstrate Black adults and children are more likely to be hypohydrated compared with White individuals [[13], [14], [15], [16]]. Although the NHANES studies included large-sample sizes and provide valuable information, they measured single urine spot samples, which may not accurately reflect daily hydration status [17,18]. Nonetheless, a recent study that used multiple 24-h urine collections in young adults (18–25 y) also demonstrated Black adults are more likely to be chronically underhydrated compared with White adults [19]. This increased prevalence of chronic underhydration could be important for future CVD and CKD development. Hypohydration is associated with increased production of arginine vasopressin (AVP), a peptide hormone produced in the hypothalamus that influences body water balance via antidiuretic effects [3,20]. Copeptin is an established surrogate marker of circulating AVP concentration [3,21,22]. Importantly, plasma copeptin is associated with several conditions that are characterized by high BP, including incident type 2 diabetes [23,24] and metabolic syndrome [25], and the progression of CKD [26,27] and CVD [28,29]. Further, several [[30], [31], [32], [33]], but not all [31], studies indicate Black adults demonstrated higher circulating AVP/copeptin concentrations compared with White adults, which could be at least partially attributable to differences in hydration.

Prior investigations suggest socioeconomic and neighborhood factors may play a role in hydration practices. For example, the recent Flint, Michigan and Jackson, Mississippi water crises have raised public awareness over safe drinking water availability and access, particularly within less advantaged minority communities [[34], [35], [36]]. Relatedly, increased perceptions of unsafe tap water have been documented among racial and ethnic minorities [15,37,38]. However, the role of neighborhood deprivation in hydration has not been considered despite an increasingly well-documented need to elucidate reasons for racial differences in hydration status. Therefore, the purpose of this analysis was to assess racial differences in hydration status in young adults and to examine the role of neighborhood deprivation. The analyses reported here were exploratory and the primary objective in the parent study was to examine how childhood adversity and contemporaneous stress combine to influence sleep and health disparities in college students [39,40]. Our primary hypotheses for the current investigation were as follows: 1) Black adults would exhibit poorer hydration status compared with White adults as assessed by 24-h urine samples for total volume, osmolality, and specific gravity and 2) greater exposure to cumulative neighborhood deprivation in childhood (ages 0–18 y), as assessed by Area Deprivation Index (ADI) scores, would be associated with poorer hydration. Additionally, we hypothesized that greater exposure to neighborhood deprivation in childhood would partially mediate race differences in hydration markers. Given the association between hydration and copeptin, we also hypothesized Black adults would exhibit higher plasma copeptin concentrations and hydration biomarkers would mediate the difference. Lastly, we sought to determine whether BP, which we expected would be higher among Black adults compared with White adults [39,41], was associated with hydration and neighborhood deprivation.

Methods

Study participants

All participants provided written and verbal consent before engaging in any study activities. Study protocol and procedures were approved by the Institutional Review Board of Auburn University and conformed to the provisions of the Declaration of Helsinki. The data reported in this investigation were collected as part of a study registered on clinicaltrials.gov (NCT04576338). It is important to note that hydration measures were not part of our a priori analytic plan when we conducted this study. Thus, some outcomes are not described on the clinicaltrials.gov listing. We acknowledge that the nature of the analysis presented here was exploratory/hypothesis-generating rather than confirmatory. Participants (n = 54, 50% female, 44% Black, 56% White, mean age = 21.4 ± 0.7 y) were undergraduate students at Auburn University (Auburn, AL, USA) who were rerecruited from a larger study (n = 263) focused on examining social determinants of health and health disparities among young adults [39,40] (See participant flowchart, Supplemental Figure 1). The primary objective in the parent study was to examine how childhood adversity and contemporaneous stress combine to influence sleep among college students. The secondary objective was to determine whether adverse childhood experiences, family socioeconomic risk, and objective indicators of neighborhood and school disadvantage mediate potential racial disparities in objectively measured sleep characteristics. Inclusion criteria included students aged between 18 and 25 y, and race of “Black/African American” or “White” in student records. In agreement with recent American Journal of Clinical Nutrition recommendations [42], we examined race as a social construct and ascribed potential differences to social determinants and health behaviors (i.e., hydration habits and dietary intake). Participants self-reported their race and reported that both of their biological parents were both “Black” or both “White”. Additionally, we quantified skin pigmentation, which can influence physiology (e.g., vitamin D production) [43] and social experiences (e.g., colorism) [44,45]. Skin pigmentation was measured by reflectance spectrophotometry (DSM3 DermaSpectrometer, Cortex Technology) [41,43]. We determined the melanin index (M-index) of the skin on the participant’s inner aspect of the upper arm due to ease of access and its relatively low-sun exposure [46,47]. We took measurements in triplicate and report the average of the 3 readings. Residential information and social determinants of health questionnaires were collected in the parent study (September 2018 to April 2019). We collected the biological data (i.e., anthropometrics, BP, blood, and urine samples) for the present study between January and December 2021. Further, regarding time of the year (which could influence M-index), we coded January as “1” and increased the months in increments of 1 with December as “12.” There was no racial difference in time of year for which data was collected between Black and White adults with April being the median month of data collection for both Black and White adults (4[5] compared with 4(3), P = 0.555, effect size = 0.096). Study completion stopped once all potential participants had been contacted and either participated or declined to participate. An additional description of the original study design is provided elsewhere.[39,40]

Study design

The current study protocol involved a laboratory visit for a resting BP measurement, as well as measurements of blood and urine biomarkers preceded by a 3-d food and fluid record. Primary outcomes included 24-h urine osmolality, urine flow rate, urine specific gravity, and cumulative ADI. Secondary outcomes included plasma copeptin concentrations and resting BP. All of these measures are described in detail below. Prior to the 2021 data collection visit where BP was measured participants were instructed to abstain from alcohol, caffeine, and exercise for the 24 h preceding their study visit, which also coincided with their 24-h urine collection (described below). Exclusion criteria for the current study included a history of hypertension diagnosis, CVD, cancer, diabetes mellitus, or kidney disease. After providing consent, study participants underwent a medical history screening, including a report of habitual physical activity. Participants’ height and body mass were measured for the calculation of BMI. Following ≥15 min of rest we measured supine BP via oscillometry (Suntech CT40, SunTech Medical). We report the average of the triplicate measures after excluding the first measure (i.e., 4 measures captured, final 3 reported).

Venipuncture and biochemical analysis

We obtained venous blood samples after ≥10 min of quiet supine rest. We centrifuged samples at 1500 × g for 10 min at 4°C (Sorvall ST8R, ThermoFisher). Serum was collected from silicone-treated vacutainers and plasma was obtained from vacutainers treated with sodium heparin or dipotassium EDTA (i.e., K₂-EDTA). Serum and plasma samples that were not used for point-of-care measures were aliquoted into cryogenic tubes to be stored at −80°C. We assessed plasma sodium, potassium, and chloride (Smartlyte Easylyte Electrolyte Analyzer), plasma osmolality (Advanced 3D3 Osmometer model 3250, Advanced Instruments), hemoglobin (Hb 201+, HemoCue), and hematocrit concentrations (Sure prepÔ capillary tubes, Clay Adams centrifuged at 1950 × g for 5 min, Legend Micro 17, Thermo Sorvall) in triplicate. The intrasample coefficients of variation were as follows: sodium concentration (0.27%), potassium concentration (0.32%), chloride concentration (0.34%), osmolality (0.47%), hemoglobin concentration (1.13%), and hematocrit (1.30%).

Twenty-four hour urine collection

As previously described [[48], [49], [50]], urine was collected for 24 h before the study visit in a light-protected, sterile 3500-mL container. Briefly, participants were instructed to record the time of their first void into a toilet the day preceding the study visit and their last void into the urine collection container (the day of the study visit). Participants returned the container at their study visits. We measured total urine volume, urine specific gravity (ATC 3-in-1 Scale Clinical Refractometer), electrolyte concentrations (sodium, potassium, and chloride), and osmolality from a mixed aliquot of the 24-h urine collection container. Urine electrolytes and osmolality were measured in triplicate. The intrasample coefficients of variation were as follows: sodium (0.92%), potassium (0.30%), chloride (0.92%), and urine osmolality (0.43%). Urine flow rate was derived from urine volume and the self-reported time the participant used the container. Based on NHANES recommendations, we did not include data that required indexing to 24-h urine flow rate for participants (n = 3) who self-reported a urine collection time of <22 h [51,52]. We stored mixed aliquots from the 24-h urine sample in cryogenic tubes at −80°C.

Assays for kidney function

We assessed plasma blood urea nitrogen (BUN) and urine urea as general measures of kidney function at the University of Alabama at Birmingham-University of California-San Diego O’Brien Center for Acute Kidney Injury Research [53]. Both BUN and urea were assayed in duplicate using a BioAssay Systems (DIUR-100) 96-well assay kit protocol. The kit’s detection range was 0.04–47 mg/dL for BUN and 0.08–100 mg/dL for urea. The average interassay coefficient of variation was 4.6% for BUN and 1.1% for urea. Plasma copeptin concentration was measured at the University of Hartford Department of Health Sciences using the BRAHMS Copeptin proAVP KRYPTOR (ThermoFisher Scientific) [21,22].

We used frozen urine and serum samples to measure creatinine via kinetic modification of the Jaffe procedure and the measurements were performed at the University of Alabama at Birmingham Bioanalytical Core (Clinical Biochemical Genetics Laboratory). We calculated creatinine clearance, both adjusted and unadjusted for body surface area, as an assessment of glomerular filtration rate using urine and serum creatinine concentration and urine flow rate [54]. Body surface area was calculated as follows:

$$\frac{\sqrt{\mathrm{h}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}\left(\mathrm{c}\mathrm{m}\right)\times \mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\left(\mathrm{k}\mathrm{g}\right)}}{3600}$$

We also calculated estimated glomerular filtration rate with serum creatinine using the recent nonrace adjusted Chronic Kidney Disease Epidemiology Collaboration equation [55].

Assessment of water intake from food and beverages

After screening we provided participants instructions for completing a 3-day food and beverage log including ≥2 weekdays and one weekend day. To help with estimating portion sizes participants were provided with a paper printout of 2D food models [56], as recently described by Lobene et al [57]. The food and beverage log was returned at the laboratory testing visit, and a trained researcher reviewed the record with the participant to obtain additional clarification and detail. The food and beverage logs were analyzed for food group and nutrient intakes using the Nutrition Data System for Research (NDSR version 2020, University of Minnesota). We report the average nutrient intake for all reported days (mean = 3.0 ± 0.3 d). Total water in liters was calculated from both food and fluid intake as reported in NDSR. Using approaches informed by prior papers using NHANES data [[13], [14], [15], [16]], we used the food and beverage log to manually estimate water intake from different sources. Using the 3-d food and fluid diaries, we operationalized beverage intake into 8 categories (water, sugar-sweetened beverages [e.g., soda or lemonade], milk or nondairy alternatives, 100% juice, noncaloric diet beverages [e.g., diet soda or Crystal Light], noncaloric coffee or tea, caloric coffee or tea drinks, and alcohol). We calculated the proportion of total water intake from plain water by dividing the average self-reported plain water volumes by total (food and beverage) water content derived from NDSR (see Supplemental Table 1).

Determination of socioeconomic status and neighborhood characteristics income

We used the income-need ratio (income:needs) as a proxy for socioeconomic position [39]. Participants reported each parent’s income in the primary household on a scale from 1 (<$5,000) to 32 (>$500,000) with each interval split by $5000 increments (e.g., 16 = $75,000–$79,999). The incomes were summed where applicable to create an estimated household income. Household income was divided by the corresponding family size specific US Census poverty line to translate a family’s household income into an adjusted measure.

Neighborhood socioeconomic deprivation

Participants identified their residential address, or closest geographic unit (e.g., city and county), for each year of their life. To help identify the correct residential address, research assistants and participants used Google Maps to visually confirm each address, and participants were encouraged to contact family members or friends to assist in recalling each address. Neighborhood characteristics were geocoded by matching the primary period address to census-tract-level data from the corresponding US Census or American Community Survey year (i.e., 2000, 2010, and 2015) [58]. These Census survey years were chosen so that residential data for our sample, originally collected in 2018–2019 with an average age of 19 ± 1 y at the time, closely corresponded to the 3 developmental periods [39,40]. Census-tract boundaries were used to define each residential neighborhood [58].

We used the ADI as a composite measure of neighborhood socioeconomic deprivation [59,60]. The ADI is composed of 17 poverty, education, housing, and employment census-tract indicators and has been previously used to track neighborhood-level disparities [59,61]. Each indicator was independently weighed by factor scores to ensure that poverty, income, and education had the largest relative weights [59]. Similar to prior studies [61], the composite score was standardized to have a mean of 100 and a SD of 20 to assist interpretation. Based on the number of years the participant lived in the neighborhood, primary residential addresses were identified to reflect each of the 3 developmental periods: early childhood (0–5 y), middle childhood (6–12 y), and adolescence (13–18 y). We used the average ADI score from each of the 3 developmental periods to calculate a cumulative ADI score. These age categories reflect the same periods used by the Panel Study of Income Dynamics’ Childhood Retrospective Circumstance Study [62]. If the participant lived in 2 addresses for a similar amount of time within the same period, the participant chose which address they felt was most reflective of that developmental period. The same residential address could be used as the primary residence for multiple developmental periods, if applicable. Importantly, ADI is a validated approach for comprehensively assessing neighborhood deprivation [63].

Statistical analyses

Any data points exceeding 1.5 times the length of the interquartile range (i.e., 1.5 times the range between the first and third quartiles) below the first or above the third quartiles were defined as outliers and removed from the analysis [64,65]. In cases where outliers were removed, only 1 to 2 participants were excluded. Moreover, sample sizes are indicated in each graph and in table legends. We inspected all variables for normality using the Shapiro-Wilk test and quantile-quantile plots to compare the shapes of distributions. We compared baseline measures between Black and White participants with independent samples t-tests for normally distributed data and Mann-Whitney U-tests for non-normally distributed data. We used Cohen’s d to assess the effect size for normally distributed data [66], interpreted as small (d = 0.2–0.49), medium (d = 0.5–0.79), and large (d > 0.8). We used Rank biserial correlation to assess effect size for non-normally distributed data [67], interpreted as small (0–0.19), medium (0.2–0.29), large (0.3–0.39), and very large (0.4–1.0). Spearman’s rho (ρ) correlations were used to evaluate associations between variables of interest. We used multiple regression models to examine variables associated with systolic BP and report standardized estimates and 95% confidence intervals for individual variables and adjusted R² for overall model fit. We used mediation analysis to examine whether racial differences in hydration biomarkers were fully or partially mediated by the cumulative ADI scores, and whether racial differences in copeptin concentrations were fully or partially mediated by hydration status [68]. Statistical significance was defined a priori as P value of ≤0.05. All data are presented as means ± SD or median (interquartile range) in the tables and as individual data points superimposed on the median and first and third quartiles in the figures. All statistical analyses were performed using Jamovi version 2.2.5 (the Jamovi project) and GraphPad Prism version 10.0.0 for Windows (GraphPad Software)

Results

Participant demographics and social determinants of health are presented in Table 1. Black adults exhibited a higher M-index, indicative of darker skin color, and also had higher systolic and diastolic BP. Regarding social determinants of health, Black adults exhibited a lower income:needs ratio and higher early, middle, and adolescent ADI scores. Thus, their cumulative ADI score of childhood exposure to neighborhood deprivation was higher.

TABLE 1.

Participant characteristics and social determinants of health

	Black adults			White adults			P value	Effect size
Sex (Female; Male)	15; 9			12;18			0.414	-
Melanin index	54.5	±	9.8	27.8	±	3.4	<0.001	3.771
Age (y)	21.6	±	0.7	21.3	±	0.8	0.261	0.165
Height (cm)	168	±	7.5	174	±	10.9	0.018	0.667
Mass (kg)	76	±	13	76	±	13	0.930	0.024
BMI (kg/m2)	27.1	±	4.3	25.0	±	3.6	0.058	0.532
Systolic BP (mmHg)	115	±	10	109	±	9	0.017	0.677
Diastolic BP (mmHg)	71	±	8	64	±	7	<0.001	0.981
Glucose (mg/dL)	83		(11)	86		(11)	0.081	0.291
Triglycerides (mg/dL)	65		(28)	73		(38)	0.919	0.003
Total cholesterol (mg/dL)	167		(53)	164		(35)	0.777	0.048
HDL cholesterol (mg/dL)	56		(21)	55		(20)	0.747	0.055
LDL cholesterol (mg/dL)	100		(50)	111		(36)	0.612	0.087
	Social determinants
Income to needs ratio	3.6		(3.0)	4.8		(3.8)	0.031	0.357
Early childhood ADI	109		(16)	97		(19)	0.001	0.517
Middle childhood ADI	107		(10)	96		(21)	0.006	0.437
Adolescence ADI	110		(11)	96		(21)	0.002	0.499
Cumulative ADI	110		(9)	97		(23)	<0.001	0.558

Data are presented as mean ± SD for normally distributed data (Shapiro-Wilk > 0.05) or median (interquartile range) for nonparametric data (Shapiro-Wilk < 0.05). When data were normally distributed, we used independent samples t-test for difference testing and Cohen’s d for effect size. When data were not normally distributed, we used Mann-Whitney U test for difference testing and Rank biserial correlation for effect size.

Regarding sample sizes for M-index n = 24 for Black adults and n = 28 for White adults. For anthropometric measures and BP n = 24 for Black adults and n = 30 for White adults. For blood measures n = 22 for Black adults and n = 28 for White adults. For ADI measures n = 23–24 for Black adults and n = 29–30 for White adults.

Abbreviations: ADI, Area Deprivation Index; BP, blood pressure; HDL, high density lipoprotein; LDL, low density lipoprotein.

Racial differences in biomarkers of hydration status are presented in Figure 1. Although there was not a race difference in urine specific gravity (Black participants: 1.021 [0.008], White participants: 1.014 [0.017), P = 0.065, Figure 1A), Black adults exhibited higher urine osmolality (640 [314] compared with 440 [283] mOsm/kg H₂O, P = 0.006, Figure 1B) and lower urine flow rates (1.06 [0.65] compared with 1.71 [0.89] ml/min, P = 0.009, Figure 1C) compared with White adults. Black adults also exhibited higher plasma copeptin (6.3 [3.1] compared with 4.5 [2.3] pmol/L, P = 0.046, Figure 1D). Additionally, urine specific gravity (Figure 1E) and urine osmolality (Figure 1F) were correlated with higher plasma copeptin concentrations. As presented in Supplemental Table 1, there were no significant group differences in macronutrient intake, but Black participants tended to eat less total calories compared with White participants (1856 [530] compared with 2136 [497] calories, P = 0.054). However, Black participants reported lower total water intake from food and beverages than White participants (2.3 ± 0.7 compared with 3.5 ± 1.1 L/d, P < 0.01). The proportion of total water intake from plain water also did not differ between Black and White participants. Compared with White adults, Black adults also reported lower fiber, calcium, magnesium, iron, and potassium intakes, and a higher dietary sodium:potassium ratio (P/S≤ 0.037).

FIGURE 1.

Racial differences in biomarkers of hydration status. Compared with White adults; Black adults exhibited higher urine specific gravity (A); urine osmolality (B); and lower urine flow rates (C). Black adults also exhibited higher plasma copeptin (D). Additionally; urine specific gravity (E) and urine osmolality (F) were correlated with higher plasma copeptin. Data are presented as individual data points and the dashed lines with violin plots representing the median and interquartile range. Orange circles represent Black adults and blue squares represent White adults. We used the Mann-Whitney U test for difference testing and Rank biserial correlation for effect size. For correlations we used Spearman’s rho for correlations.. Sample sizes are provided for each graph. Abbreviations: Osm, osmolality.

Participants’ biochemical measures and markers of kidney function are presented in Table 2. Black adults exhibited higher plasma chloride, but otherwise, there were no differences in plasma electrolytes, osmolality, BUN, or serum creatinine concentrations. Black adults demonstrated significantly lower urine volumes despite no difference in collection times. There were no racial differences in urine urea concentration, urine creatinine excretion, or creatinine clearance.

TABLE 2.

Biochemical measures of kidney function

	Black adults			White adults			P value	Effect size
Plasma Na+ (mEq/L)	140.5	±	1.7	140.1	±	1.7	0.440	0.223
Plasma K+ (mEq/L)	3.6		(0.4)	3.7		(0.5)	0.634	0.079
Plasma Cl- (mEq/L)	101.1	±	2.2	99.8	±	2.0	0.031	0.628
Plasma Osm (mOsm/kgH2O)	281	±	5	282	±	5	0.558	0.169
Blood urea nitrogen (mg/dL)	18.6		(4.1)	18.9		(3.5)	0.594	0.089
Serum creatinine (mg/dL)	0.91	±	0.29	0.86	±	0.14	0.455	0.214
Estimated GFR (mL/min/1.73m2)	152		(24)	150		(18)	0.663	0.075
Hemoglobin (g/dL)	13.5	±	1.6	14.4	±	1.8	0.073	0.517
Hematocrit (%)	42.6	±	4.6	44.0	±	4.0	0.261	0.322
Urine volume (mL)	1501	±	978	2526	±	1310	0.004	0.463
Collection time (min)	1448	±	143	1489	±	90	0.210	0.361
Urine urea (mg/dL)	1455		(825)	1212		(805)	0.139	0.241
Urine creatinine excretion (mg/min)	1.17		(0.49)	1.20		(0.61)	0.701	0.064
Creatinine clearance (mL/min/1.73m2)	130	±	34	116	±	28	0.135	0.463

Data are presented as mean ± SD for normally distributed data (Shapiro-Wilk > 0.05) or median (interquartile range) for nonparametric data (Shapiro-Wilk < 0.05). When Shapiro-Wilk was > 0.05 we used independent samples t-test for difference testing and Cohen’s d for effect size. When Shapiro-Wilk was < 0.05 we used Mann-Whitney U test for difference testing and Rank biserial correlation for effect size.

Samples sizes for blood and urine measures was n = 21–23 for Black adults and 28–30 for White adults. Because creatinine clearance includes both urine and blood samples the sample size was n = 19 for Black adults and n = 27 for White adults.

Abbreviations: Na⁺, sodium; K⁺, potassium; Cl^-, chloride; Osm, osmolality; GFR, glomerular filtration rate.

Correlations of neighborhood socioeconomic deprivation with BP and hydration biomarkers are presented in Figure 2. Using cumulative ADI scores, we determined there were positive associations between ADI and systolic BP (Figure 2A), diastolic BP (Figure 2B), and urine specific gravity (Figure 2C). There was no association between cumulative ADI and urine osmolality (Figure 2D). However, cumulative ADI was inversely associated with self-reported total water intake from food and beverages (Figure 2E). The associations between early, middle, and adolescent ADI and BP or hydration biomarkers were similar to that of the cumulative ADI score, and are depicted in Supplemental Figures 2–4.

FIGURE 2.

The influence of neighborhood socioeconomic deprivation on blood pressure and hydration biomarkers. Using cumulative Area Deprivation Index (ADI) scores we determined there were associations between ADI and systolic BP (A); diastolic BP (B); and urine specific gravity (C). There was not an association between cumulative ADI and urine osmolality (D); however; cumulative ADI was associated with self-reported water intake from food and beverages (E). For correlations we used Spearman’s rho. Sample sizes provided for each graph. Abbreviations: BP, blood pressure; Osm, osmolality.

In Figure 3, we summarize 3 separate mediation analyses. Above each arrow, we present path estimates with their standard errors in parentheses. Effects with P values of ≤0.05 are demarcated with asterisks. Urine specific gravity (see Figure 3A) mediated 47% of the racial difference in plasma copeptin concentration (indirect effect, P = 0.064). After controlling for the mediating effect of urine specific gravity, there was no direct effect of race (P = 0.213), but there was a total effect (P = 0.032). Urine osmolality (Figure 3B) mediated 67% of the racial difference in plasma copeptin (indirect effect, P = 0.027). There was no direct effect of race after controlling for the mediating effect of urine osmolality (P = 0.450), however, there was a total effect (P = 0.042). Urine flow rate (Figure 3C) mediated 44% of the racial difference in plasma copeptin (indirect effect, P = 0.078). There was no direct effect of race after controlling for the mediating effect of urine specific gravity (P = 0.264), however, there was a total effect (P = 0.032). In contrast to the objective hydration status biomarkers in Figure 3. For additional data on % mediation, see Supplemental Table 2. In contrast, self-reported water intake from food and beverages (i.e., subjective measure) only mediated 16% of the racial difference in plasma copeptin amounts (indirect effect P = 0.603, data not shown).

FIGURE 3.

Objective hydration biomarkers mediate racial differences on plasma copeptin. Path estimates for urine specific gravity and plasma copeptin (A). Path estimates for urine osmolality; a significant mediator of racial differences in plasma copeptin (B). Path estimates for urine flow rate and plasma copeptin (C). ∗ Indicates significance. Data are presented as path estimates with standard errors in parenthetical. The p values and 95% confidence intervals (CI) (p values; 95% confidence intervals) were as follows (A): for race → urine specific gravity: (0.035; 0.0036, 0.0095); for urine specific gravity → plasma copeptin: (<0.001; 74.1, 229.7); for race → plasma copeptin (0.213; −0.490, 2.201). (B): for race → urine osmolality: (0.011; 49.8, 378.1); for urine osmolality → plasma copeptin: (<0.001; 0.0011, 0.0027); for race → plasma copeptin (0.450; −0.822, 1.853). (C): for race → urine flow rate: (<0.001; −1.19, −0.33); for urine flow rate → plasma copeptin: (0.041; −1.831, −0.039]; for race → plasma copeptin (0.264; −0.671, −2.454). Regarding sample sizes, the mediation analyses include n = 50–54. Abbreviations: Osm, osmolality.

Results from multiple regression models examining variables potentially associated with systolic BP are presented in Table 3. Unadjusted multiple linear regressions with race and sex (model 1) indicated race was associated with systolic BP. The sequential addition of BMI (model 2), urine osmolality (model 3), and cumulative ADI (model 4) in additional models revealed that race was not independently associated with systolic BP. However, BMI and cumulative ADI were independently associated with systolic BP. We conducted similar regression models for diastolic BP (Supplemental Table 3). Briefly, race was independently associated with diastolic BP in models 1, 2, and 3. However, after the addition of cumulative ADI (model 4), race was not associated with diastolic BP. In contrast to systolic BP, cumulative ADI was not associated with diastolic BP.

TABLE 3.

Results from multiple regression models examining variables potentially associated with systolic blood pressure

Variables	Model 1			Model 2			Model 3			Model 4
	$\beta$	(95% CI)	P value	$\beta$	(95% CI)	P value	$\beta$	(95% CI)	P value	$\beta$	(95% CI)	P value
	Systolic blood pressure (mmHg)
Race	0.77	(0.19, 1.37)	0.041	0.56	(−0.03, 1.15)	0.159	0.44	(−0.16, 1.03)	0.147	0.12	(−0.54, 0.77)	0.722
Sex	−0.24	(−0.53, 0.06)	0.298	−0.21	(−0.49, 0.07)	0.316	−0.15	(−0.43, 0.14)	0.299	−0.14	(−0.41, 0.14)	0.321
BMI (kg/m2)				0.32	(0.04, 0.60)	0.011	0.38	(0.09, 0.67)	0.011	0.39	(0.11, 0.67)	0.007
Urine Osm (mOsm/kgH2O)							−0.06	(−0.35, 0.23)	0.684	−0.09	(−0.37, 0.20)	0.547
Cumulative ADI										0.31	(0.01, 0.62)	0.041
Adjusted R2(%)	5.2			16.3			14.7			20.9
P value	0.11			0.012			0.026			0.009

Unadjusted multiple linear regressions with race and sex (model 1) indicated race was associated with systolic blood pressure. The addition of BMI (model 2); urine flow rate; specific gravity; and osmolality (model 3); and cumulative ADI (model 4) revealed that race was not independently associated with systolic blood pressure. However; BMI and cumulative ADI were associated with systolic blood pressure. We are reporting standardized estimates and 95% confidence intervals and adjusted R². Bolded text indicates a significant association.

For the final regression n = 53

Abbreviations: ADI, Area Deprivation Index; Osm, osmolality.

Analyses investigating whether neighborhood deprivation mediates the association between racial differences and hydration biomarkers are presented in Supplemental Figure 5. The cumulative ADI score (see Supplemental Figure 5A) mediated 30% of the racial difference in urine specific gravity (indirect effect P = 0.249). After controlling for the mediating effect of cumulative ADI, there was not a direct effect of race (P = 0.188), but there was a total effect (P = 0.034). Cumulative ADI (Supplemental Figure 5B) mediated 6.5% of the racial difference in urine osmolality (indirect effect, P = 0.772). There was a direct effect of race after controlling for urine osmolality (P = 0.042), and there was a significant total effect (P = 0.013). Cumulative ADI (Supplemental Figure 5C) mediated 4.6% of the racial difference in urine flow rate (indirect effect, P = 0.780). There was a direct effect of race after controlling for urine flow rate (P = 0.009), and there was a significant total effect (P = 0.002). Additional details of the mediation analysis are provided in Supplemental Table 4.

Lastly, biomarkers of hydration status were generally not correlated with BP or kidney function other than urine osmolality being correlated with BUN (ρ = 0.294, P = 0.039). Also of interest, plasma copeptin concentration was not correlated with systolic (ρ = 0.121, P = 0.396) or diastolic BP (ρ = 0.083, P = 0.561).

Discussion

The primary findings of this investigation were that Black adults were hypohydrated compared with White adults assessed by 24-h urine samples for urine flow rate and osmolality. Although not statistically significant (P = 0.065), urine specific gravity was also directionally higher in Black adults with a medium effect size (Rank biserial correlation = 0.294, see Figure 1A). Total fluid intake assessed from food and fluid records demonstrated that Black adults consumed less total water from food and fluid compared with White adults. A novel finding of this investigation was that ADI was associated with self-reported total water intake from food and beverages and urine specific gravity. However, ADI was not a mediator of racial differences in hydration status. Plasma copeptin concentration was higher in Black adults compared with White adults, and urine osmolality mediated these differences.

Our findings of racial disparities in hydration status are consistent with multiple studies using NHANES data, which have demonstrated differences in hydration status between Black and White American adults and children using spot urine samples [[13], [14], [15], [16]]. Importantly, urine spot samples are subject to circadian variation and may not accurately reflect daily hydration status [17,18]. In contrast, 24-hour urine osmolality, which indicated greater hypohydration among Black adults compared with White adults in our study, provides an accurate assessment of total fluid intake and is a gold standard measure of hydration status [17,18,69]. Although urine specific gravity was not significantly higher, as we noted above, it was directionally higher in Black adults and the effect sizes bordered on medium to large. Thus, the specific gravity findings are broadly consistent with our urine flow rate and osmolality findings. A larger sample would likely yield a significant difference. Additionally, the median urine specific gravity for Black adults (1.021) was greater than the commonly used value of 1.020 that indicates relative dehydration [70]. Taken together, our findings add to prior studies using 24-h urine collections indicating that Black adults are more likely to be underhydrated compared with White adults [19,71]. Regarding potential reasons for the hydration disparity, the limited literature in this area suggests associations between poorer drinking water quality and social determinants of health, such as lower socioeconomic status and race/ethnicity [35]. Our findings add to prior work by demonstrating neighborhood deprivation in childhood, assessed with the cumulative ADI, was associated with total water intake from food and beverages and urine specific gravity.

Tap water is an ideal hydration strategy as it is low cost, federally mandated to be protected and publicly available [72], and does not contribute to calorie intake. However, low-income and predominately minority communities often face disproportionately high pollutant exposures in their drinking water. Major recent incidents have raised concerns over equitable safe drinking water availability and access [34]. In 2015, there were reports of lead contamination in the drinking water supply of Flint, Michigan leading to elevated blood lead concentration in children [35]. The contamination occured because of aging infrastructure and the use of the Flint River as a new drinking water source in a cost-saving measure by the public municipality [36]. Because ∼55% of Flint’s residents are Black and ∼40% live below the poverty line [73], the crisis sparked a nationwide debate about environmental justice and drinking water quality. More recently, there was a widely publicized water crisis in Jackson, Mississippi where over ∼80% of residents are Black and ∼25% live below the poverty line [73]. The city's largest municipal water treatment plant failed after local flooding, leading to 150,000 citizens having unsafe drinking water [74,75]. Although these incidents represent just 2 highly publicized stories, environmental injustices may contribute to disparities in water intake that ultimately manifest in racial differences in water intake. Indeed, our findings support the role of neighborhood socioeconomic deprivation playing a role in hydration status. Specifically, the cumulative ADI score of childhood exposure to neighborhood deprivation was correlated with higher urine specific gravity and self-reported total water intake from food and beverages (Figure 2).

Negative perceptions of tap water safety are associated with both lower water intake and higher sugar-sweetened beverage intake [15], indicating individuals may replace water with sugar-sweetened beverages to meet hydration needs. Racial minorities exhibit disparities in the distribution of sugar-sweetened beverage intake, and are more likely to report negative perceptions of tap water safety, which is significantly associated with lower water intake [15,76,77]. These perceptions may begin in childhood as 26% of Black compared with 15% of White children (aged 9–19 y) believe that tap water is neither safe nor clean [78], and this apprehension may carry over into adulthood. Although our study was not focused on sugar-sweetened beverages, an interesting future direction to pursue would be to study whether neighborhood deprivation is associated with increased sugar-sweetened beverage intake and mediates racial disparities. A leading hypothesis is that low water intake coupled with sugar-sweetened beverage intake activates both vasopressin and the aldose reductase-fructokinase pathways which contribute to the development of obesity and cardiometabolic disease [20]. However, to our knowledge, there has been a lack of data as to whether racial differences in hydration status may contribute to potential racial differences in copeptin and cardiometabolic disease.

Interestingly, we demonstrated that osmolality was associated with copeptin concentration and mediated the association between race and plasma copeptin concentration (Figure 3). These findings contribute to a growing body of literature on social determinants of health and health behaviors (e.g., hydration status) contributing to health disparities. Applied physiology and clinical research from prior generations has often implied racial disparities derive from inherent biological differences. Additionally, whereas most of the published data on disparities in copeptin concentration have been on middle-aged and older adults, another novel aspect of our investigation is that we measured circulating copeptin concentration in young adults.

In agreement with prior studies [[30], [31], [32], [33]], we also demonstrated racial differences in circulating AVP/copeptin. For example, a prior study demonstrated selective AVP receptor inhibition lowered mean arterial pressure in Black, but not White, adults [30]. These findings suggest AVP played a more important role as a pressor hormone in contributing to hypertensive BP in Black adults. Another prior study demonstrated Black men exhibit lower plasma volume compared with White male adults [79]. Our data indicate that chronic underhydration may play a role in such findings. Future larger investigations in young adult cohorts matched for BP may provide additional insight regarding racial differences in copeptin concentration. The higher AVP concentration in Black individuals may contribute to increased risk of future cardiometabolic disease [[23], [24], [25],28,29], and demonstrates the importance of interventions targeted at addressing racial disparities in hydration status.

Irrespective of race, prior investigations have demonstrated that urine osmolality and specific gravity are associated with plasma copeptin, and that increased water intake reduces urine osmolality and plasma copeptin [21,22]. From a physiologic perspective, having a lower plasma volume as a result of chronic underhydration could contribute to elevated solute concentration, such as copeptin. We did not measure plasma volume and most data suggest fluid regulatory mechanisms function to maintain plasma apart from transient changes elicited by acute perturbations (e.g., intense exercise, heat stress, or diarrhea) [80]. Alternatively, increased AVP and concomitant copeptin release are part of the homeostatic response to maintain plasma volume in the context of underhydration. Thus, we hypothesized that racial differences in hydration biomarkers would mediate differences in copeptin release. Prior data indicates that hydration interventions (using water and low-calorie beverages) improve hydration status, reduce copeptin concentration, and improve cardiometabolic health [22,81]. In the context of the current investigation Black adults exhibited higher BP and a trend for higher BMI. Thus, it is tempting to speculate that hydration interventions could result in disproportionate improvements in these clinically meaningful measures in Black adults, but future trials are needed to test this hypothesis.

Apart from hydration interventions there is also a clear need for actions to address racial inequities in the social determinants and health behaviors that may contribute to BP disparities. Similar to our previous work in young adults, Black adults in the current investigation exhibited elevated resting BP compared with White adults [41]. Our findings build on larger-scale cross-sectional studies demonstrating that Black individuals have a higher BP and develop hypertension at an earlier age than other racial/ethnic groups in America [[6], [7], [8]]. We have previously demonstrated that neighborhood deprivation mediates racial disparities in BP among young Black and White adults [39]. Similarly, in the current investigation, we demonstrated cumulative ADI correlated with BP. Additionally, race was not associated with BP after adjustment for BMI and childhood exposure to neighborhood deprivation. Using multiple regression, we demonstrated that every one unit change in cumulative ADI was associated with a 0.35 mmHg higher systolic BP (unstandardized β coefficient). Based on our reported racial difference of 13 in the cumulative ADI score of childhood exposure to neighborhood deprivation (see Table 1) that would suggest differences in childhood neighborhood deprivation explained ∼4.6 mmHg of the higher systolic BP that Black participants exhibited. Thus, our findings support the critical need for multilevel primary prevention strategies (e.g., community-level policy or environmental enhancements) to address the primary contributors to BP disparities experienced by Black Americans [82,83].

Limitations

Our study is not without limitations. The size of our cohort was modest in part due to the COVID-19 pandemic. Future studies should recruit larger sample sizes that would enable greater statistical power. Another limitation of retrospective data collection pertains to residential addresses, which we used to determine ADI. Prospective studies that assess hydration and social determinants across several time points would provide additional fidelity. Lastly, the results of this study should be interpreted within the context of a college student sample in the southeastern region of America. As we have previously highlighted [39], a community sample would have likely presented greater variability in income, education, and childhood neighborhood socioeconomic conditions. Additionally having a larger sample from various regions of the country would be more generalizable to the larger implications of early exposure to neighborhood deprivation on health.

Summary and future directions

In summary, we have demonstrated racial disparities in 24-h hydration status in young adults. Specifically, Black adults consumed less total water from food and beverages compared with White adults. Further, hydration biomarkers measured in 24-h urine samples indicated that Black adults were hypohydrated compared with White adults. Plasma copeptin concentration was higher in Black adults compared with White adults and 24-h urine osmolality, a gold standard biomarker of hydration status, mediated this difference. Neighborhood deprivation was associated with total water intake from food and beverages and urine specific gravity. Lastly, we demonstrated racial differences in BP that were attenuated after adjusting for neighborhood deprivation, but not hydration status. Future directions from the current investigation include investigations aimed at examining when hydration disparities emerge and targeted interventions aimed at improving hydration status in Black adults to presumably reduce urine osmolality and plasma copeptin concentration. Interestingly, hydration biomarkers are associated with gut microbiota, [84] which is also associated with cardiometabolic health. An interesting future direction would be to examine potential racial differences in the gut microbioata, and how greater fluid intake may influence the gut microbiome.

Acknowledgments

We would like to thank Dr. Colleen Munoz (University of Hartford) for completing copeptin measures. We also thank undergraduate students Sydney Jones (Tuskegee University) and Sarah Nix (Auburn University) for their assistance with data collection and analyses.

Author contributions

The authors’ responsibilities were as follows – ATR, TEFR: designed the research and procured the materials; ATR, BAL, AMB, SJ, OIN, MCM, ZJH, MAT, TEFR: conducted the experiments; ATR, BAL, AMB, SJ, SOS, OIN: analyzed the study; ATR, BAL, SOS, MCM, JCW, OMG, TEFR: interpreted the results; ATR, SOS, TEFR: wrote the paper; ATR, BAL, AMB, SJ, SOS, OIN, MCM, ZJH, MAT, JCW, OMG, TEFR: edited the paper and all authors: read and approved the final manuscript.

Conflict of interest

The authors report no conflicts of interest.

Funding

This study was supported by National Institutes of Health (NIH) Grants K01HL147998 and R15HL165325 (to ATR), R15HL140504 (to TEFR), K01HL160772 (to JCW), K24DK116180, P30DK079337, P50MD017338 and T32GM141739 (to OMG), and UL1TR003096 (to ATR, BAL, and OMG); American Heart Association Career Development Award 23CDA1037938 (J.C.W.); and an Auburn University Presidential Graduate Research Fellowship (to BAL).

Data availability

Data described in the manuscript, code book, and analytic code will be made available upon reasonable request pending a data and material transfer agreement approval.

Appendix A. Supplementary data

The following is the Supplementary data to this article: