Associations of Obesity and Neighborhood Factors With Urinary Stone Parameters

Abstract

Introduction:

Obesity is associated with kidney stone disease (KSD), but it is unknown whether this association differs by SES. This study assessed the extent to which obesity and neighborhood characteristics jointly contribute to urinary risk factors for KSD.

Methods:

This was a retrospective analysis of adult patients with KSD evaluated with 24-hour urine collection (2001-2020). Neighborhood-level socioeconomic data were obtained for a principal component analysis (PCA), which identified 3 linearly independent factors. Associations between these factors and 24-hour urine measurements were assessed using linear regression, as well as groupings of 24-hour urine results using multivariable logistic regression. Finally, multiplicative interactions were assessed testing effect modification by obesity, and analyses stratified by obesity were performed. Analyses were performed in 2021.

Results:

In total, 1,264 patients met study criteria. Factors retained on PCA represented SES, family structure, and housing characteristics. On linear regression, there was a significant inverse correlation between SES and 24-hour urine sodium (p=0.0002). On multivariable logistic regression, obesity was associated with increased odds of multiple stone risk factors (OR: 1.61; 95% CI: 1.15-2.26) and multiple dietary factors (OR: 1.33; 95% CI: 1.06-1.67). No significant and consistent multiplicative interactions were observed between obesity and quartiles of neighborhood SES, family structure, or housing characteristics.

Conclusions:

Obesity was associated with the presence of multiple stone risk factors and multiple dietary factors; however, the strength and magnitude of these associations did not vary significantly by neighborhood SES, family structure, and housing characteristics.

INTRODUCTION

Kidney stone disease (KSD) affects nearly 1 in 11 adults.¹ At least 50% of adults with a stone-related event will experience another one within 10 years.² Thus, KSD is best viewed as a chronic disease. Clinical guidelines recommend metabolic testing consisting of 24-hour urine collections for recurrent stone formers and high-risk or interested first-time stone formers.³ Differences in contributors to stone formation, including 24-hour urine findings, can be due to nutrition, lifestyle, or physiologic responses; many of which can be modified through secondary prevention.

To date, there is limited research evaluating the influence of SES on urinary risk factors for KSD. A recent scoping review evaluating disparities in KSD in the U.S.⁴ identified only 3 studies focused on associations between socioeconomic factors and urine chemistry.^5–7 The relationship between SES and urine chemistry among stone formers is further complicated by risk factors that may vary by sociodemographic group. Obesity is one such risk factor for KSD.⁸ The relationship between obesity and SES is complex,⁹ and it is unclear whether an interaction between obesity and SES influences urinary risk factors for KSD.

To better understand the joint influence of obesity and SES on urinary risk factors for KSD, an analysis of 24-hour urine data from adult kidney stone formers with available BMI and neighborhood-level socioeconomic data was performed. Due to the multidimensional nature of neighborhood SES, a principal component analysis (PCA) was performed. Associations with individual 24-hour urine testing results, as well as 2 groupings of these results were evaluated. Finally, interactions between neighborhood characteristics and obesity were tested to determine whether they influenced the aforementioned outcomes.

METHODS

Study Sample

This was a retrospective evaluation of prospectively collected data among patients with KSD evaluated with 24-hour urine collection at an academic medical center. Patients aged 18 years or older who completed at least one collection between 2001 and 2020 met inclusion criteria. The steps in study sample development are shown in Appendix Figure 1. Patients without at least one BMI measurement, patients with diagnoses or procedures associated with gastrointestinal malabsorption (codes listed in Appendix Table 1), patients with missing Census tract information, patients not residing in Alabama or a neighboring state, and patients without at least one adequate 24-hour urine creatinine (Cr) as defined by Taylor and Curhan (>600 mg for a female patient and >800 mg for a male patient)¹⁰ were excluded. The IRB at the University of Alabama at Birmingham approved this study (protocol # 300006901).

Measures

Patient age at the time of the first 24-hour urine collection, sex, marital status, race, and ethnicity were reported. The BMI reported with the patient’s 24-hour urine result was used. If one was not reported, the most recent BMI in the electronic medical record on or before the collection date was used. The diagnosis codes used to define chronic kidney disease, diabetes, and hypertension are available through the Chronic Conditions Data Warehouse.¹¹ For patients who underwent a stone removal procedure (ureteroscopic stone removal or percutaneous nephrolithotomy) with an available stone analysis, a cutoff of >50% to define predominant stone composition was applied.

To obtain neighborhood data, each patient’s charted residential address at the time of the analysis was geocoded and linked to the 2017 American Community Survey (ACS) 5-year estimates aggregated to Census tracts.¹² Fifteen variables in the domains of education, income, disability, healthcare access, family structure, and housing and living conditions were obtained to examine the relationship between neighborhood disadvantage and KSD. Variables are listed in Appendix Table 2; all measures were expressed as percentages.

Litholink (Laboratory Corporation of America, Burlington, NC, USA) chemically analyzed all 24-hour urine collections.¹³ The measurements upon which stone risk factors were defined included urine volume, calcium, oxalate, citrate, pH, and uric acid. Relative supersaturation indices were not included. The measurements upon which dietary factors were defined included urine sodium, potassium, magnesium, phosphorus, ammonium, sulfate, and urea nitrogen. Appendix Table 3 defines how measurements were classified as abnormal. The 2 outcomes of interest in this study were the presence of (1) multiple (>1) stone risk factors and (2) multiple (>1) dietary factors, as defined in Litholink reports. These outcomes were determined using each patient’s first adequate 24-hour urine collection based on urine Cr.

Statistical Analysis

A PCA identified neighborhood characteristics explaining the variance in the study sample. This allowed for a multidimensional evaluation of neighborhood characteristics, which was preferred over a single index of neighborhood disadvantage. The PCA resulted in 3 factors based on eigenvalues >1, accounting for 72% of the variance. Factor eigenvalues and loadings are listed in Table 1. For all subsequent analyses, neighborhood factors were assessed as standardized scores (mean=0, SD=1), with higher scores representing higher disadvantage.

Table 1.

Factor Eigenvalues and Loadings of Neighborhood Variables Derived Through a Principal Component Analysis

Variable	Factor 1: SES, Eigenvalue=7.89	Factor 2: Family structure, Eigenvalue=1.49	Factor 3: Housing characteristics, Eigenvalue=1.38
Adults aged ≥25 years without high school diploma	0.83	−0.19	−0.18
Adults aged ≥25 years with bachelor’s degree	−0.82	0.22	0.21
Households in poverty	0.83	0.38	0.06
Households with SNAP/food stamp benefits	0.90	0.25	0.00
Civilian unemployment rate	0.73	0.19	−0.02
Disabled population	0.76	−0.38	−0.10
Civilians without health insurance	0.78	0.14	−0.02
Civilians with private health insurance	−0.95	−0.11	0.03
Households with computer access	−0.87	0.30	0.02
Households with broadband access	−0.90	0.24	0.08
Households without vehicles	0.65	0.16	0.30
Single-parent households with children	0.51	0.69	0.05
Households with seniors (aged ≥65 years) living alone	0.38	−0.56	0.21
Households lacking complete kitchen facilities	0.10	−0.03	0.81
Households lacking complete plumbing facilities	0.16	−0.15	0.70

Note: Boldface indicates the loading with the greatest absolute value corresponding to each variable.

SNAP, Supplemental Nutrition Assistance Program.

Differences in patient and neighborhood characteristics stratified by the 2 outcomes of interest (multiple stone risk factors and multiple dietary factors) were evaluated using chi-square tests to assess differences in categorical variables, 2-sample t-tests to assess differences in continuous variables, and appropriate nonparametric tests when assumptions of parametric tests were not met. An exploratory analysis assessed individual 24-hour urine measurements and the 3 neighborhood factors as continuous variables using linear regression. Multivariable logistic regression models evaluated associations between the 3 neighborhood factors and the 2 outcomes of interest. Effect modification by obesity (BMI ≥30 kg/m²) was assessed by testing the statistical significance of multiplicative interaction terms, and building models stratified by obesity. Finally, a secondary analysis was performed by excluding patients without at least one appropriate 24-hour urine Cr as defined by the stricter Litholink cutoffs (18–24 mg/kg for a male patient and 15–20 mg/kg for a female patient).¹³

Statistical analyses and plot generation were performed in 2021 using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA), MATLAB version R2021a (MathWorks, Inc., Natick, MA, USA), and Prism version 9.2.0 (GraphPad Software, LLC, San Diego, CA, USA). All statistical tests were 2-sided, and p-values <0.05 were considered statistically significant, except for the linear regressions in which a Bonferroni correction was applied to correct for multiple testing (statistical significance was p≤0.001 for these 39 analyses).

RESULTS

The analytic sample included 1,264 adult patients: mean age 51.1 years (SD 15.1), 649 (51.3%) male, 482 (38.1%) obese (Appendix Figure 1). The median time between the recorded BMI and 24-hour urine result was 25 days (IQR: 12 to 82 days); this difference exceeded 1 year for 205/1,264 (16.2%) patients. As detailed in Table 1, the PCA generated 3 neighborhood-level factors, representing SES (Factor 1), family structure (Factor 2), and housing characteristics (Factor 3).

Individual- and neighborhood-level characteristics of the analytic sample are listed in Table 2. Compared to patients without multiple stone risk factors, those with multiple stone risk factors resided in neighborhoods with higher rates of poverty and single-parent households with children, and lower rates of private health insurance coverage. Compared to patients without multiple dietary factors, those with multiple dietary factors resided in neighborhoods with higher rates of adults aged ≥25 years without a high school diploma and disability, and lower rates of adults aged ≥25 years with a bachelor’s degree. There were also statistically significant differences in marital status and BMI (Table 2).

Table 2.

Individual and Neighborhood Characteristics of the Study Sample

Characteristic	Overall (n=1,264)	One or fewer stone risk factors (n=195)	Multiple stone risk factors (n=1,069)	p-value	One or fewer dietary factors (n=648)	Multiple dietary factors (n=616)	p-value
Individual-level
Age, years, mean (SD)	51.1 (15.1)	52.3 (13.5)	50.9 (15.3)	0.2	51.6 (15.3)	50.6 (14.8)	0.3
Sex, n (%)				0.3			0.3
Female	615 (48.7)	88 (45.1)	527 (49.3)		325 (50.2)	290 (47.1)
Male	649 (51.3)	107 (54.9)	542 (50.7)		323 (49.9)	326 (52.9)
Married, n (%)	840 (66.5)	141 (72.3)	699 (65.4)	0.15	462 (71.3)	378 (61.4)	<0.001
Race, n (%)				0.5			0.7
White	985 (77.9)	154 (79.0)	831 (77.7)		508 (78.4)	477 (77.4)
Black	116 (9.2)	23 (11.8)	93 (8.7)		64 (9.9)	52 (8.4)
Asian	18 (1.4)	1 (0.5)	17 (1.6)		11 (1.7)	7 (1.1)
American Indian/Alaska	4 (0.3)	0 (0.0)	4 (0.4)		2 (0.3)	2 (0.3)
Native
Other	111 (8.8)	14 (7.2)	97 (9.1)		49 (7.6)	62 (10.1)
Multiple	5 (0.4)	0 (0.0)	5 (0.5)		2 (0.3)	3 (0.5)
Unknown	25 (2.0)	3 (1.5)	22 (2.1)		12 (1.9)	13 (2.1)
Ethnicity, n (%)				0.6			0.09
Non-Hispanic	1,033 (81.7)	162 (83.1)	871 (81.5)		544 (84.0)	489 (79.4)
Hispanic	13 (1.0)	3 (1.5)	10 (0.9)		7 (1.1)	6 (1.0)
Unknown	218 (17.2)	30 (15.4)	188 (17.6)		97 (15.0)	121 (19.6)
BMI, kg/m2 mean (SD)	29.5 (7.4)	28.0 (6.3)	29.8 (7.5)	<0.001	28.7 (6.7)	30.4 (8.0)	<0.001
Chronic kidney disease, n (%)	394 (31.2)	66 (33.9)	328 (30.7)	0.4	213 (32.9)	181 (29.4)	0.2
Diabetes, n (%)	165 (13.1)	26 (13.3)	139 (13.0)	0.9	84 (13.0)	81 (13.2)	0.9
Hypertension, n (%)	401 (31.7)	59 (30.3)	342 (32.0)	0.6	221 (34.1)	180 (29.2)	0.06
Predominant stone composition, n (%)				0.6			0.2
Calcium oxalate	287 (22.7)	51 (26.2)	236 (22.1)		159 (24.5)	128 (20.8)
Calcium phosphate	58 (4.6)	8 (4.1)	50 (4.7)		31 (4.8)	27 (4.4)
Uric acid	34 (2.7)	3 (1.5)	31 (2.9)		14 (2.2)	20 (3.3)
Other	10 (0.8)	1 (0.5)	9 (0.8)		3 (0.5)	7 (1.1)
Unknown	875 (69.2)	132 (67.7)	743 (69.5)		441 (68.1)	434 (70.5)
Neighborhood-level
SES, Factor 1, mean (SD)				0.10			0.08
Adults aged ≥25 years without high school diploma	8.1 (5.4)	8.0 (5.5)	8.2 (5.3)	0.6	7.8 (5.4)	8.5 (5.3)	0.02
Adults aged ≥25 years with bachelor’s degree	18.9 (11.8)	19.7 (12.2)	18.7 (11.7)	0.3	19.9 (12.2)	17.8 (11.4)	0.002
Households in poverty	10.7 (8.4)	9.4 (7.8)	11.0 (8.5)	0.007	10.7 (8.5)	10.8 (8.4)	0.6
Households with SNAP/food stamp benefits	11.2 (8.7)	10.4 (8.8)	11.4 (8.6)	0.06	10.9 (8.8)	11.5 (8.6)	0.11
Civilian unemployment rate	6.3 (4.2)	6.1 (3.9)	6.3 (4.2)	0.6	6.1 (4.0)	6.5 (4.3)	0.2
Disabled population	15.3 (6.1)	15.2 (6.5)	15.3 (6.1)	0.7	14.9 (6.1)	15.7 (6.1)	0.009
Civilians without health insurance	9.2 (5.1)	8.7 (5.3)	9.2 (5.1)	0.10	8.9 (5.0)	9.4 (5.2)	0.06
Civilians with private health insurance	72.2 (13.8)	73.7 (14.8)	71.9 (13.6)	0.04	72.9 (13.7)	71.4 (13.8)	0.06
Households with computer access	84.9 (9.2)	85.7 (9.2)	84.8 (9.2)	0.2	85.3 (9.0)	84.6 (9.4)	0.3
Households with broadband access	74.6 (12.6)	75.5 (13.3)	74.4 (12.5)	0.2	75.1 (12.5)	74.1 (12.7)	0.2
Households without vehicles	4.9 (4.9)	4.7 (4.4)	5.0 (4.9)	0.2	5.1 (5.2)	4.7 (4.5)	0.4
Family structure, Factor 2, mean (SD)				0.14			0.3
Single-parent households with children	8.1 (5.2)	7.3 (4.7)	8.2 (5.3)	0.03	8.1 (5.3)	8.1 (5.0)	0.6
Households with seniors (aged ≥65 years) living alone	10.9 (4.5)	10.7 (4.5)	10.9 (4.4)	0.5	10.9 (4.5)	10.8 (4.4)	0.8
Housing characteristics, Factor 3, mean (SD)				1.0			0.02
Households lacking complete kitchen facilities	0.7 (1.1)	0.6 (1.0)	0.7 (1.1)	0.7	0.7 (1.1)	0.6 (1.0)	0.5
Households lacking complete plumbing facilities	0.3 (0.7)	0.3 (0.6)	0.3 (0.7)	0.8	0.3 (0.7)	0.3 (0.7)	0.7

Note: Boldface indicates statistical significance (p<0.05).

SNAP, Supplemental Nutrition Assistance Program.

Results of an exploratory analysis correlating 24-hour urine measurements with neighborhood SES, family structure, and housing characteristics through linear regression are shown in Appendix Figure 2 for stone risk factors and Appendix Figure 3 for dietary factors. Of the 39 regressions performed, only one showed a statistically significant correlation following correction for multiple testing: 24-hour urine sodium was positively correlated with neighborhood SES (Factor 1) (p=0.0002). This implies an inverse relation between 24-hour urine sodium and SES, because a higher Factor 1 standardized score represents higher socioeconomic disadvantage.

Results from multivariable logistic regression models of associations of neighborhood SES, family structure, and housing characteristics with the 2 outcomes of interest (multiple stone risk factors and multiple dietary factors) are listed in Table 3. Patients aged 35-49 years had lower odds of multiple stone risk factors compared to those aged 18-34 years; Black patients had lower odds of multiple stone risk factors compared to White patients; and patients with obesity had higher odds of multiple stone risk factors compared to those without obesity. No statistically significant differences in odds of multiple stone risk factors by neighborhood characteristics (Factors 1–3 quartiles) were detected. For the second outcome, multiple dietary factors, patients who were married had lower odds compared to those who were not married and patients with obesity had higher odds compared to those without obesity. Again, no statistically significant differences in odds of multiple dietary factors by neighborhood characteristics (Factors 1–3 quartiles) were detected.

Table 3.

Odds of Multiple Stone Risk Factors and Multiple Dietary Factors for the Study Sample (n=1,264)

Characteristic	Multiple stone risk factors OR (95% CI)	Multiple dietary factors OR (95% CI)
Age, years
18–34	ref	ref
35–49	0.49 (0.27, 0.86)	1.04 (0.72, 1.51)
50–64	0.58 (0.33, 1.02)	1.16 (0.81, 1.66)
≥65	0.58 (0.32, 1.07)	0.97 (0.65, 1.44)
Male	0.90 (0.66, 1.24)	1.21 (0.96, 1.52)
Married	0.81 (0.58, 1.14)	0.64 (0.50, 0.81)
Race
White	ref	ref
Black	0.53 (0.31, 0.92)	0.77 (0.50, 1.18)
Other or unknown	1.48 (0.83, 2.65)	1.14 (0.79, 1.66)
Obese	1.61 (1.15, 2.26)	1.33 (1.06, 1.67)
Chronic kidney disease	0.82 (0.57, 1.18)	0.89 (0.68, 1.16)
Diabetes	0.90 (0.54, 1.52)	1.14 (0.78, 1.66)
Hypertension	1.28 (0.85, 1.91)	0.82 (0.61, 1.09)
SES (Factor 1)
1st quartile (least disadvantaged)	ref	ref
2nd quartile	1.14 (0.75, 1.72)	1.06 (0.77, 1.45)
3rd quartile	1.43 (0.93, 2.21)	1.33 (0.97, 1.83)
4th quartile (most disadvantaged)	1.32 (0.86, 2.03)	1.16 (0.85, 1.59)
Family structure (Factor 2)
1st quartile (least disadvantaged)	ref	ref
2nd quartile	1.09 (0.72, 1.66)	0.88 (0.64, 1.20)
3rd quartile	1.15 (0.75, 1.75)	1.00 (0.73, 1.37)
4th quartile (most disadvantaged)	1.39 (0.89, 2.16)	0.78 (0.57, 1.07)
Housing characteristics (Factor 3)
1st quartile (least disadvantaged)	ref	ref
2nd quartile	1.06 (0.68, 1.64)	0.94 (0.69, 1.28)
3rd quartile	0.89 (0.58, 1.36)	0.79 (0.57, 1.07)
4th quartile (most disadvantaged)	1.01 (0.65, 1.56)	0.76 (0.55, 1.03)

Notes: Ethnicity not included as a covariate due to limited number of Hispanic patients in the dataset (1.0%). Predominant stone composition not included as a covariate due to majority of patients with unknown predominant stone composition (69.2%).

Tests of multiplicative interactions between obesity and neighborhood SES, family structure, and housing characteristics with respect to odds of multiple stone risk factors and multiple dietary factors did not demonstrate statistical significance, with the exception of one: the interaction between obesity and the most disadvantaged (quartile 4) neighborhood housing characteristics was associated with lower odds of multiple dietary factors, relative to the interaction between obesity and the least disadvantaged (quartile 1) neighborhood housing characteristics (Appendix Table 4).

Figure 1 reports the odds of multiple stone risk factors and multiple dietary factors stratified by obesity and neighborhood SES, family structure, and housing characteristics. Relative to patients without obesity, those with obesity frequently had higher odds of multiple stone risk factors and multiple dietary factors. However, for patients with and without obesity, a consistent increase or decrease in odds of multiple stone risk factors and multiple dietary factors with increasing neighborhood disadvantage was not observed.

Figure 1.

Odds of multiple stone risk factors and multiple dietary factors by obesity and quartiles of neighborhood factors for the study sample (n=1,264).

Notes: Error bars indicate 95% CIs. Models adjusted for age, sex, marital status, race, chronic kidney disease, diabetes, and hypertension. The first quartile corresponds to the least disadvantaged group and the fourth quartile corresponds to the most disadvantaged group.

In total, 626/1,264 (49.5%) patients met criteria for a secondary analysis in which patients without at least one appropriate 24-hour urine Cr were excluded, as defined by Litholink cutoffs. Individual and neighborhood characteristics of the secondary analytic sample are listed in Appendix Table 5. Multivariable logistic regression models assessing associations of neighborhood SES, family structure, and housing characteristics with the 2 outcomes of interest are listed in Appendix Table 6. Black patients had lower odds of multiple stone risk factors compared to White patients; patients with obesity had higher odds of multiple stone risk factors compared to those without obesity; and patients with the most disadvantaged neighborhood SES (quartiles 3 and 4) had higher odds of multiple stone risk factors compared to those with the least disadvantaged neighborhood SES (quartile 1). For the second outcome, patients who were married had lower odds of multiple dietary factors and patients with obesity had higher odds of multiple dietary factors. Two tests of multiplicative interactions between obesity and neighborhood SES, family structure, and housing characteristics with respect to odds of the 2 outcomes of interest demonstrated statistical significance: the interaction between obesity and the second quartile of neighborhood family structure (Factor 2) was associated with lower odds of multiple stone risk factors, relative to the interaction between obesity and the least disadvantaged (first) quartile (Appendix Table 7); and the interaction between obesity and the second quartile of neighborhood housing characteristics (Factor 3) was associated with lower odds of multiple stone risk factors, relative to the interaction between obesity and the least disadvantaged (first) quartile (Appendix Table 7). Appendix Figures 4 and 5 report the odds of multiple stone risk factors and multiple dietary factors, respectively, stratified by obesity and neighborhood SES, family structure, and housing characteristics, in quartiles. Relative to patients without obesity in the least disadvantaged (first) quartile, patients with obesity frequently had higher odds of multiple stone risk factors and multiple dietary factors; however, for patients with and without obesity, a consistent increase or decrease in odds with increasing neighborhood disadvantage was not observed.

DISCUSSION

This study evaluated the extent to which obesity and neighborhood characteristics contribute to urinary risk factors for KSD. In a sample of 1,264 adult patients with KSD who completed a 24-hour urine collection, 3 linearly independent neighborhood factors were identified: SES, family structure, and housing characteristics. Patients with obesity had increased odds of both multiple stone risk factors and multiple dietary factors. These associations with obesity persisted on a secondary analysis using more stringent urine Cr cutoffs to define an adequate 24-hour urine collection. While no differences were observed in odds of these outcomes across levels of neighborhood SES, family structure, or housing characteristics in the primary analysis, there were increased odds of multiple stone risk factors in the third and fourth quartiles of SES (i.e., the most disadvantaged half of the sample) in the secondary analysis. Additionally, consistent significant multiplicative interactions between obesity and levels of neighborhood SES, family structure, or housing characteristics with respect to the outcomes of interest were not observed. In analyses stratified by obesity, patients with obesity frequently had higher odds of multiple stone risk factors and dietary factors, which is expected based on the aforementioned findings of the unstratified analysis. Nonetheless, these odds did not consistently increase or decrease with increasing disadvantage in neighborhood SES, family structure, or housing characteristics. These results suggest that obesity is associated with urinary risk factors for stone disease, but the magnitude and strength of this association does not necessarily vary by levels of neighborhood disadvantage.

The rising prevalence of KSD parallels the rising prevalence of obesity. An analysis of the National Health and Nutrition Examination Survey (NHANES) showed that individuals with obesity had increased odds of self-reported KSD compared to individuals with normal BMI.¹ Obesity is also associated with abnormal urinary stone risk parameters in stone formers. For example, Eisner et al. performed an analysis of 880 patients evaluated at a metabolic stone clinic and found that higher BMI was associated with higher urine sodium and lower urine pH in men, as well as higher urine uric acid, sodium, and lower urine citrate in women.¹⁴ The results of this study support prior work demonstrating increased risk of urinary abnormalities among stone formers with obesity.

Few studies have focused on relationships between sociodemographic factors and urinary abnormalities among stone formers. A recent assessment of associations between the Distressed Communities Index (DCI) and 24-hour urine results found that higher DCI (i.e., lower SES) correlated with lower urine citrate and potassium, suggesting lower intake of fruits and vegetables.⁵ In an analysis of 435 patients from 2 stone clinics using neighborhood data, increasing poverty level and decreasing education level were both associated with significant increases in urine calcium excretion.⁶ A subsequent evaluation of patients from the same clinics demonstrated that those with state-assisted insurance had significantly higher urine sodium and pH compared to those with private insurance.⁷ Consistent with these findings, the present study identified an inverse correlation between neighborhood SES and urine sodium, which may reflect differences in sodium intake, a modifiable dietary factor.

The relationship between SES and obesity is multifactorial.⁹ For example, a recent analysis of NHANES found that neighborhood SES was positively associated with healthy body weight in women, but not in men.¹⁵ Other factors such as psychosocial stress may also play a significant role.¹⁶ To the authors’ knowledge, this is the first study to examine the joint influence of obesity and neighborhood characteristics on urinary stone risk parameters among stone formers. Due to the numerous neighborhood characteristics relevant to the outcomes of interest in this analysis, a PCA was performed to reduce dimensionality. The development and evaluation of multifactorial neighborhood variables specific to the patient sample, rather than the use of a single index of neighborhood disadvantage, is a strength of this study. Although 2 dichotomous outcome variables were defined, 24-hour urine results should also be assessed as continuous variables; thus, an exploratory analysis was performed using linear regression with a conservative correction for multiple testing. Finally, analyses were repeated in a secondary sample of patients meeting more stringent 24-hour urine Cr cutoffs. The results of the primary and secondary analyses were comparable, notwithstanding 2 new statistically significant interaction terms for obesity with family structure and housing characteristics which were present for only the second quartile compared to the first quartile (least disadvantaged), but not the more disadvantaged third and fourth quartiles.

This study has several important implications for secondary prevention of KSD. First, urinary risk factors for KSD are present throughout all neighborhood strata, emphasizing the need for enhanced access to guideline-based³ evaluation and management, especially for disadvantaged groups. Second, obesity is likely associated with urinary risk factors across all sociodemographic groups; thus, weight loss could be a worthwhile preventive measure for many stone formers, though further research is needed to confirm this. Third, contributors to urinary risk factors may still differ between neighborhood strata due to variables not accounted for in this study; examples include household food insecurity¹⁷ and neighborhood walkability.¹⁸ Studies assessing whether community-level improvements in these environmental factors alter biological responses, such as urinary stone risk parameters, are needed.

Limitations

This study has several limitations. The patients included in the analytic sample resided in the southeast U.S. and were evaluated at an academic medical center. Furthermore, prior studies have demonstrated that patients with KSD and low SES are less likely to complete 24-hour urine evaluation, suggesting a selection bias.¹⁹ Therefore, the results of this study may not be generalizable to other geographical regions, other clinical settings, or the population as a whole. Neighborhood characteristics do not always accurately reflect individual characteristics;²⁰ however, relevant individual-level sociodemographic variables were also included in analyses including age, sex, marital status, race, and ethnicity. Residual confounding is possible given the observational nature of this study, though numerous covariates relevant to KSD were accounted for and patients with conditions associated with gastrointestinal malabsorption were excluded. In the study sample, 16.2% of patients had BMI measurements and 24-hour urine results separated by >1 year; the impact of temporal changes in diet and body weight is unclear for these patients. Medication use was not accounted for in the analyses, but each patient’s first adequate 24-hour urine collection was used, so pharmacotherapy for KSD is less likely to have been introduced. Covariates such as chronic kidney disease, diabetes, and hypertension were captured using ICD 9/10 codes, which are not always appropriately included in the electronic medical record. While residential address and the corresponding Census tract identifier were available for the majority of patients, the charted address at the time of this analysis (2021) and ACS data from 2017 were used, either of which could be discordant with the patient’s address at the time of the 24-hour urine collection. Finally, a larger study sample would have resulted in greater statistical power to detect associations between the predictors and outcomes of interest in this study, particularly in the secondary analysis. Further evaluation of obesity, SES, and urinary stone risk factors within multi-institutional datasets or large prospective cohort studies may be practical approaches to achieve a larger sample size.

CONCLUSIONS

Among patients with KSD completing a 24-hour urine collection, obesity was independently associated with the presence of multiple stone risk factors and multiple dietary factors; however, the strength and magnitude of these associations did not vary significantly across sociodemographic groups defined by neighborhood variables.