Table 3.
Results of multilevel linear or Cox models with an interaction between segregation index and race/ethnicity estimating the association of residential segregation with later life cognitive abilities among respondents from different groups. Estimates for which the 95% confidence intervals do not include the null are highlighted in bold.
| Residential Segregation 1 Indices |
Memory 2 | Language 2 | Visuospatial 2 | Incident Dementia 3 |
|---|---|---|---|---|
| β (95% CI) | β (95% CI) | β (95% CI) | HR (95% CI) | |
| NH Black respondents | ||||
| Dissimilarity NH Black—NH White | −0.073 (−0.143; −0.003) | −0.071 (−0.129; −0.012) | −0.008 (−0.067; 0.051) | 1.74 (1.34; 2.27) |
| Dissimilarity Hispanic-NH White | 0.032 (−0.039; 0.103) | −0.040 (−0.101; 0.021) | −0.042 (−0.103; 0.018) | 1.05 (0.78; 1.41) |
| Isolation NH Black | −0.053 (−0.121; 0.015) | −0.074 (−0.132; −0.017) | −0.017 (−0.075; 0.042) | 1.72 (1.32; 2.24) |
| Isolation Hispanic | 0.032 (−0.040; 0.104) | −0.010 (−0.072; 0.052) | −0.040 (−0.100; 0.020) | 1.27 (0.95; 1.69) |
| Isolation NH White 4 | 0.093 (−0.069; 0.255) | 0.140 (0.019; 0.262) | 0.093 (−0.033; 0.219) | 1.01 (0.32; 3.19) |
| Interaction NH Black—NH White | 0.028 (−0.042; 0.098) | 0.044 (−0.012; 0.100) | 0.045 (−0.014; 0.104) | 1.21 (0.91; 1.61) |
| Interaction Hispanic—NH White | 0.087 (−0.007; 0.180) | 0.066 (−0.010; 0.142) | 0.004 (−0.072; 0.080) | 0.82 (0.50; 1.34) |
| Hispanic respondents | ||||
| Dissimilarity NH Black-NH White | −0.015 (−0.107; 0.077) | −0.027 (−0.101; 0.048) | −0.066 (−0.142; 0.010) | 1.05 (0.76; 1.44) |
| Dissimilarity Hispanic-NH White | 0.020 (−0.038; 0.078) | −0.034 (−0.082; 0.014) | −0.050 (−0.098; –0.002) | 1.05 (0.86; 1.28) |
| Isolation NH Black | −0.029 (−0.125; 0.067) | −0.024 (−0.101; 0.054) | −0.086 (−0.166; –0.007) | 1.05 (0.76; 1.46) |
| Isolation Hispanic | −0.009 (−0.068; 0.050) | −0.041 (−0.089; 0.007) | −0.079 (−0.126; –0.031) | 1.13 (0.93; 1.39) |
| Isolation NH White | 0.048 (−0.065; 0.162) | 0.095 (0.006; 0.185) | 0.071 (−0.020; 0.162) | 0.87 (0.54; 1.38) |
| Interaction NH Black-NH White | 0.030 (−0.040; 0.101) | 0.043 (−0.012; 0.099) | 0.030 (−0.029; 0.089) | 1.04 (0.81; 1.33) |
| Interaction Hispanic—NH White | 0.060 (0.002; 0.119) | 0.076 (0.029; 0.123) | 0.080 (0.031; 0.128) | 0.95 (0.78; 1.17) |
| NH White respondents | ||||
| Dissimilarity NH Black-NH White 4 | –0.096 (–0.308; 0.115) | −0.016 (−0.192; 0.159) | −0.164 (−0.330; 0.002) | 0.98 (0.24; 3.98) |
| Dissimilarity Hispanic-NH White | 0.034 (–0.054; 0.121) | −0.114 (−0.187; –0.041) | −0.028 (−0.099; 0.044) | 0.53 (0.34; 0.84) |
| Isolation NH Black 4 | –0.185 (–0.507; 0.137) | 0.068 (−0.182; 0.318) | −0.166 (−0.414; 0.083) | N/A |
| Isolation Hispanic | 0.036 (–0.053; 0.124) | −0.122 (−0.196; –0.049) | −0.018 (−0.089; 0.053) | 0.60 (0.38; 0.94) |
| Isolation NH White | –0.038 (–0.114; 0.038) | 0.126 (0.063; 0.190) | 0.013 (−0.050; 0.076) | 0.52 (0.35; 0.78) |
| Interaction NH Black-NH White | –0.017 (–0.094; 0.059) | 0.099 (0.038; 0.159) | −0.044 (−0.107; 0.019) | 0.50 (0.31; 0.83) |
| Interaction Hispanic-NH White | 0.049 (–0.031; 0.128) | 0.051 (−0.013; 0.116) | 0.029 (−0.036; 0.094) | 0.58 (0.42; 0.80) |
1 Racial/ethnic residential segregation indices are measured at the block group level from 2005–2009 ACS data. Dissimilarity measures the number of people who would have to move to create an equal distribution of a racial/ethnic groups in the geographic area of interest. Isolation and interaction measure the likelihood of interacting with someone in the same racial/ethnic group or in a different racial/ethnic group, respectively. For all indices, we used the mean value within our sample to create binary indicators of segregated census blocks. Higher dissimilarity and isolation indicate a block group with more people from minoritized backgrounds, whereas higher interaction indicates more desegregated areas; 2 multilevel linear models adjusted for age, sex/gender, race/ethnicity, cSES, years of education, occupation, language of test administration, birthplace, and recruitment cohort. All cognitive scores were converted to z-scores; 3 multilevel Cox models with age as the underlying time-scale, adjusted for sex/gender, race/ethnicity, childhood socioeconomic position (cSES), years of education, occupation, language of test administration, birthplace, and recruitment cohort; 4 due to stark segregation between NH Black and NH White adults in the studied area, very few (n = 70) of our NH Black respondents live in a census block where the majority of residents are NH White adults, as indicated by the isolation NH White index. Similarly, very few NH White respondents live in census blocks where the majority of residents are NH Black as indicated by the dissimilarity NH Black-NH White index (n = 43) and isolation NH Black index (n = 14). Subsequently, the concerned estimates have wide confidence intervals; abbreviations: CI, confidence interval; HR, hazard ratio; NH, non-Hispanic.