Table 3.

Stepwise logistic regression model estimating for the association between segregation measures and metabolic syndrome

	Model 1	Model 2
	With individual-level characteristics	+ Neighborhood-level characteristics
	Odds ratio [95% confidence interval]	+ Neighborhood-level characteristics
Formal measures of segregation
Evenness
Q1, high evenness
Q2	1.13 [0.81–1.59]	1.11 [0.79–1.55]
Q3	1.27 [0.90–1.79]	1.24 [0.88–1.74]
Q4,	1.02 [0.73–1.43]	0.98 [0.71–1.36]
Q5: low evenness	1.30 [0.96–1.78]	1.22 [0.89–1.69]
P for trend	.211	.438
continuous (0.10-unit change)	1.02 [0.93–1.12]	1.00 [0.91–1.10]
Exposure
Q1, low isolation
Q2	1.33 [0.94–1.90]	1.31 [0.92–1.87]
Q3	1.26 [0.92–1.73]	1.22 [0.88–1.67]
Q4,	1.54 [1.09–2.17]	1.46[1.03–2.07]
Q5: high isolation	2.13 [1.39–3.26]	2.00 [1.30–3.07]
continuous (0.10-unit change)	1.11 [1.02–1.21]	1.09 [1.00–1.19]
P for trend	.003	.008
Evenness, after controlling for exposure
Q1: high evenness
Q2	1.07 [0.78–1.47]	1.06 [0.77–1.47]
Q3	1.29 [0.92–1.80]	1.28 [0.91–1.80]
Q4,	0.91 [0.66–1.26]	0.91 [0.66–1.25]
Q5: low evenness	1.19 [0.88–1.62]	1.18 [0.86–1.63]
Continuous (0.10-unit change)	0.99 [0.90–1.09]	0.99 [0.90–1.09]
P for trend	.538	.616
Exposure, after controlling for evenness
Q1: high exposure
Q2	1.34 [0.95–1.89]	1.34 [0.95–1.89]
Q3	1.29 [0.95–1.76]	1.28 [0.94–1.75]
Q4,	1.55 [1.10–2.18]	1.53^∗ [1.08–2.17]
Q5: low exposure	2.33 [1.52–3.57]	2.29 [1.49–3.52]
P for trend	.006	.011
Continuous (0.10-unit change	1.11 [1.02–1.21]	1.11 [1.02–1.21]
Proxy measures of segregation
ICE for income
Q1, highest extreme concentration of disprivilege
Q2	0.99 [0.75–1.31]	0.99 [0.75–1.31]
Q3	0.93 [0.69–1.26]	0.95 [0.7–1.29]
Q4	0.92 [0.62–1.35]	0.96 [0.65–1.41]
Q5: highest extreme concentration of privilege	0.82 [0.57–1.18]	0.87 [0.61–1.25]
P for trend	.292	.481
Continuous	0.87 [0.77–0.98]	0.88 [0.78–1.00]
ICE for race/ethnicity
Q1, highest extreme concentration of disprivilege
Q2	0.74 [0.53–1.03]	0.75 [0.54–1.04]
Q3	0.66 [0.42–1.02]	0.68 [0.44–1.05]
Q4	0.65 [0.44–0.97]	0.68 [0.45–1.01]
Q5: highest extreme concentration of privilege	0.58 [0.38–0.89]	0.64 [0.42–0.98]
P for trend	.050	.154
Continuous	0.89 [0.79–1.00]	0.93 [0.82–1.04]
ICE combined
Q1, highest extreme concentration of disprivilege
Q2	0.76 [0.55–1.05]
Q3	0.87 [0.63–1.21]
Q4	0.81 [0.55–1.18]
Q5: highest extreme concentration of privilege	0.70 [0.47–1.04]
P for trend	.091
Continuous	0.86 [0.76–0.98]

ICE = Index of Concentration at the Extremes.

Note. The evenness dimension of segregation was measured using the GINI coefficient for race/ethnicity. The exposure dimension of segregation was measured using the Isolation index. The ICE for income was measured as a continuous variable. All model two, three, and effects modification analysis controlled for sex, education (≤high school or less, >high school), employment status (employed, other), language preference (English, Spanish), years in the United States combined with nativity (U.S. born, years U.S. ≥ 10, years U.S. < 10), Hispanic/Latino heritage (Dominican, Cuban, Mexican, Puerto Rican, other/ >1 background group), marital status (married, other), health insurance (private, public, uninsured) and income (less than $10,000, $10,001–$20,000, $40,001–$75,000, more than $75,000). Neighborhood deprivation was included in model 2 of evenness and exposure (separate and together) and of ICE for race/ethnicity. Neighborhood racial/ethnic composition was included in model 2 and interaction models for evenness and ICE for income. Interaction models were tested separately fully adjusted for individual and neighborhood level variables. Exploratory interactions P-values used a Bonferroni adjusted P < .001 based on an alpha of 0.1 for 70 tests. Segregation dimensions and ICE Models N = 6710; interaction models: N = 6706. Exposure variables were analyzed as continuous in models of interactions.