Not one enclave: Disaggregation and cardiometabolic health in Asian ethnic enclaves

Highlights

• •East and South Asian enclaves in NYC are geographically distinct.
• •East Asian enclave residence is linked to lower obesity risk.
• •South Asian enclave residence is linked to higher cardiometabolic risk.
• •Enclave–health associations appear only among foreign-born individuals.
• •Neighborhood socioeconomic factors explain little of these associations.

Abstract

Many Asian American (AA) subgroups experience disproportionate rates of cardiometabolic (CMB) conditions, yet the contextual drivers of these disparities remain unclear. Little is known about the role of Asian residential segregation, often conceptualized as Asian enclaves, with limited prior work largely ignoring region of origin and nativity. Using six years of cross-sectional survey data from New York City (N = 6686 AAs) linked with multiple sources of community data, we examine how residence in ethnicity-specific enclaves relates to CMB risks (obesity, hypertension, and diabetes), whether these associations differ by nativity, and the extent to which neighborhood socioeconomic conditions, the built environment, social cohesion, and institutional support account for observed associations. Our combined concentration-based and spatial clustering analysis identified five East Asian enclaves and six South Asian enclaves, with no geographic overlap between the two. Multilevel logistic regression analyses show that residence in an East Asian enclave was associated with lower odds of obesity (OR=0.60; 95%CI=0.38, 0.95), while residence in a South Asian enclave was linked to higher odds of diabetes (OR=1.36; 95%CI=1.02, 1.82) and hypertension (OR=1.44; 95%CI=1.07, 1.95). These associations were present only among foreign-born individuals. After adjusting for neighborhood characteristics, the association between East Asian enclaves and obesity persisted, while those of South Asian enclaves with diabetes and hypertension were partly attenuated. Both suggest a role for unmeasured enclave factors, including cultural and food environments. Our findings challenge the view that Asian enclaves are monolithically health-promoting and redirect scholarly attention toward disaggregated approaches to investigating AA health disparities.

1. Introduction

Many subgroups within the Asian American (AA) community experience disproportionate rates of cardiometabolic (CMB) risk (Shah et al., 2022). However, the drivers of these disparities remain unclear. Research has primarily focused on individual-level factors (Min et al., 2022), leaving a significant knowledge gap regarding the role of context, particularly neighborhood environment. With a growing influx of immigrants since the 1960s, Asian residential segregation, often conceptualized as Asian ethnic enclaves, has been increasing in the US (Anderson and Simburger, 2022; Iceland, 2004). In New York City, approximately 30% of AAs reside in ethnic enclaves (Lim et al., 2017; Kandula et al., 2009). These percentages are likely to increase as Asian immigrants are projected to grow and be the largest group of all US immigrants by 2055 (Hanna and Batalova, 2021).

While ethnic enclaves can be a critical upstream factor in the CMB health, their health implications remain unclear. Ethnic enclaves are often seen as protective for health, as they can offer social support networks and protection against racism (Bécares et al., 2012; Charles, 2003). This is in contrast to the fact that racial residential segregation can also function as a fundamental cause of negative health disparities (Williams and Collins, 2001), particularly for Black Americans (Kershaw et al., 2015). It has been argued that the causes underlying Asian segregation differ from those for black communities (Yang et al., 2017), and may reflect personal preferences to a larger degree than structural disadvantages (Anderson and Simburger, 2022). Even so, the “healthful enclave” perspective cannot be applied uniformly across Asian subgroups, as mechanisms and historical drivers of segregation vary widely based on their migration histories, cultural backgrounds, and racialized experiences in the US (Yi et al., 2022). Empirical findings are mixed. Asian enclave residence is associated with poorer self-rated mental health (Hong et al., 2014) and quality of life (Guo et al., 2022), healthier dietary patterns (Osypuk et al., 2009; Yi et al., 2014), lower smoking rates (Kandula et al., 2009; Yang et al., 2014), and reduced adverse birth outcomes (Walton, 2009), as well as null association for self-reported health (Anderson and Simburger, 2022), physical activity (Osypuk et al., 2009), diabetes and hypertension (Lim et al., 2017), and obesity (Yu et al., 2018).

Several key limitations contribute to this inconsistency. First, prior work has used a pan-national measure for enclaves without differentiating by Asian subgroup (Anderson and Simburger, 2022; Shariff-Marco et al., 2021). This may not accurately capture the influence of enclaves, because many Asian groups have few to no linguistic or cultural similarities. Second, most studies fail to consider nativity (Lim et al., 2017; Guo et al., 2022; Yi et al., 2014; Yang et al., 2014). This is a significant limitation given that approximately 70% of AA adults are foreign-born (Budiman and Ruiz, 2025). Third, few studies have attempted to understand the multiple mechanisms through which ethnic enclaves relate to CMB outcomes. This makes it difficult to decouple the potential health benefits of co-ethnic social networks from the negative effects of socioeconomic deprivation and poor built environments.

The current study examines how the characteristics and health correlates of Asian enclaves vary across ethnic-specific enclaves and by nativity in New York City. New York City provides a particularly informative setting for examining these processes because of its large and heterogeneous Asian immigrant population and its long-established ethnic communities. At the same time, the composition, spatial organization, and institutional structure of Asian enclaves in New York City may differ from those in other US metropolitan and suburban settings. Accordingly, we approach New York City as a case study rather than as representative of Asian enclave contexts nationally.

2. Background

2.1. Ethnic-specific Asian enclaves and health

Most existing research relies on pan-ethnic measures of Asian enclaves, which aggregate diverse national-origin Asian groups into a single construct. This oversimplification likely obscures important differences in the socioeconomic, cultural, and institutional characteristics of enclaves. Several studies have noted this as a key limitation (Anderson and Simburger, 2022; Lim et al., 2017), stemming from systematic limitations in data collection as well as conventional practices that collapse heterogeneous Asian populations into a singular category (Yi et al., 2022). Empirical evidence shows that Asian ethnic subgroups likely form very different neighborhood environments, which in turn shape how ethnic concentration influences health outcomes. Walton (Walton, 2015) demonstrates that Korean and Chinese neighborhoods in California are more likely to be affluent, resurgent communities associated with better self-reported health, while Filipino and Vietnamese enclaves are disproportionately low-income communities tied to poor health. Similarly, Janevic et al. (Janevic et al., 2014) document that South Asian enclaves in New York are associated with an elevated risk for gestational diabetes, whereas Chinese enclaves show no such association. These patterns raise our first set of research questions:

Q1a.How do the location and characteristics of ethnic-specific Asian enclaves in New York City differ by ethnicity?

Q1b. How is residence in an ethnic-specific Asian enclave (vs. non-enclave areas) associated with CMB outcomes (i.e., obesity, hypertension, and diabetes)?

2.2. The role of nativity in shaping enclave impacts

Nativity further complicates these dynamics. While enclaves may serve as a robust environment for new arrivals, they may be less salubrious for native-born AAs and established immigrants. Enclaves can reinforce social and economic isolation from mainstream society, and, when coupled with neighborhood-level deprivation, restrict access to healthcare, quality housing, and safe environments (Osypuk et al., 2009). Thus, enclaves may be less beneficial, or even detrimental, for US-born AAs, who may experience constrained upward mobility or increased psychosocial stress within these environments. Morey et al. (Morey et al., 2020) show that living in an Asian enclave is protective against discrimination-related stress for recent immigrants but is associated with higher stress for US-born Asians. This challenges the notion that enclaves are monolithically beneficial for health, as socioeconomic disadvantage may outweigh advantages from social capital especially for later-generation residents. These findings motivate our second research question:

Q2. Does the association between residence in ethnic-specific Asian enclaves and health outcomes differ by nativity status?

2.3. Neighborhood pathways linking enclaves to cardiometabolic health

Understanding the pathways through which ethnic enclave residence shapes health is essential to reconciling previous mixed findings. Prior research has shown that many inner-city Asian enclaves in New York and Los Angeles have higher poverty and lower educational and occupational attainment (Logan et al., 2002; Zhang et al., 2022), and that residence in economically disadvantaged enclaves is associated with poor health outcomes (Walton, 2009; Walton, 2015; Zhang et al., 2022). Robust co-ethnic social capital has also been framed as a potential health resource (Yang et al., 2020), as it can buffer residents from racial discrimination (Zhou and Kim, 2006), while encouraging healthier behaviors and improving access to culturally appropriate health information and services. However, these economic and social conditions capture only part of the broader structural and environmental context of ethnic enclaves. The built environment and institutional support for immigrants represent two important, yet underexamined, dimensions. Studies find that immigrant-dense Chinese neighborhoods suffer from poorer physical environments characterized by reduced access to walkable infrastructure and fewer exercise facilities (Guo et al., 2022; Osypuk et al., 2009), and even affluent Asian-majority neighborhoods experience systematic disinvestment in their physical environments (García et al., 2012; Pincetl et al., 2003). In a similar vein, institutional support for immigrants, including local budget allocations for language-access services and immigrant-serving organizations, may be relevant to health but remains rarely addressed in studies. Considering these dimensions may provide a more complete understanding of how ethnic enclaves serve as sources of both risks and resilience. This leads to our third research question:

Q3. To what extent are the associations between ethnic-specific Asian enclave residence and health outcomes explained by neighborhood socioeconomic status, the built environment, social cohesion, and institutional support for immigrants?

3. Methods

3.1. Data and sample

Individual-level data are drawn from the New York City Community Health Survey (NYCCHS), an annual cross-sectional telephone survey of NYC residents aged 18 years or older (New York City Department of Health and Mental Hygiene, 2023). Data were collected through computer-assisted telephone interviews using telephone numbers obtained from a commercial sampling vendor. Starting in 2009, the survey included both landline and cellular telephone samples, with weights provided to account for landline-only, cell-phone-only, and dual-phone users. Further methodological details are available at: https://www.nyc.gov/site/doh/data/data-sets/community-health-survey-methodology.page. To maximize the sample size of AAs, we pooled three 2-year datasets spanning 2015 to 2020, the years for which geographic identifiers at the Public Use Microdata Area (PUMA) level are available, via a data use agreement with the NYC Department of Health. The pooled sample included 6825 AAs. After excluding 139 respondents with missing data on any covariate using listwise deletion, the final analytic sample includes 6686 AAs. Because missingness varies by outcome, sample sizes differ slightly across models: 6576 for obesity, 6664 for hypertension, and 6664 for diabetes. In stratified analyses, we disaggregate the sample by ethnicity (East Asian: n = 4467; South Asian: n = 1603; Southeast Asian: n = 405; 211 others excluded due to small samples) and by nativity (US-born: n = 988; foreign-born: n = 5698).

Community-level data are drawn from multiple sources. To construct measures of ethnic enclaves, we use population counts by racial/ethnic group from the 2015–2019 American Community Survey (ACS) 5-year estimates. Additional neighborhood characteristics were downloaded from NYC Planning’s Community District Profiles, which provide neighborhood socioeconomic indicators and built environment data. Litter basket coverage was retrieved from the NYC Department of Sanitation, via the Environment & Health Data Portal. Finally, data on public funding for immigrant services were obtained from Data2Go.NYC, reflecting allocations by the Department of Youth and Community Development during the 2019–2020 fiscal year.

3.2. Measures

3.2.1. Cardiometabolic (CMB) risk outcomes

CMB risk outcomes include obesity, hypertension, and diabetes. Obesity is defined based on self-reported height and weight, with a calculated body mass index of 30 kg/m² or higher. Hypertension and diabetes are assessed through self-reports of having ever been diagnosed with these conditions by a health professional.

3.2.2. Ethnic enclaves

We defined ethnic-specific Asian enclaves at the PUMA level, which was selected as the geographic unit of analysis because it represents the most granular level available in our individual-level dataset. PUMAs are US Census-defined statistical areas constructed from contiguous census tracts with a minimum population of 100,000. In New York City, PUMAs closely approximate Community Districts (CDs), administratively recognized neighborhood units established in the City Charter and widely used for local planning and service delivery. Because PUMAs generally align with CDs, they capture socially and institutionally meaningful neighborhood contexts, making them a reasonable proxy for enclave boundaries in NYC.

For ethnic subgroups, we used the American Community Survey “Asian alone” categories, regardless of Hispanic origin, and defined East Asians as Chinese, Japanese, Korean, Okinawan, and Taiwanese individuals, South Asians as Asian Indian, Bangladeshi, Bhutanese, Nepalese, Pakistani, and Sri Lankan individuals, and Southeast Asians as Burmese, Cambodian, Filipino, Hmong, Indonesian, Laotian, Malaysian, Thai, and Vietnamese individuals. The enclave classification focused on East Asian and South Asian populations because preliminary analyses indicated that these groups show sufficiently concentrated and spatially distinct residential patterns to support robust enclave characterization, whereas Southeast Asian residential patterns did not meet criteria for defining distinct enclaves. Southeast Asian individuals were nevertheless included in the analytic sample.

We then classified East Asian and South Asian ethnic enclaves using a two-step approach. First, we applied a concentration-based definition to identify PUMAs with high proportions of East Asian and South Asian residents. For each PUMA, we calculated the proportion of the total population identifying as East Asian and South Asian. Because the population shares of these groups differ substantially, we adopted a group-specific threshold using the “double share criterion” (Alba et al., 1997; Edin et al., 2003). Under this approach, a PUMA was classified as an ethnic enclave if the proportion of a specific ethnic group was at least twice its share in the overall New York City population. This yielded threshold values of 16.4 percent for East Asians (8.2% citywide) and 8.6 percent for South Asians (4.3%).

Second, we refined this classification by further incorporating a second criterion: spatial clustering of underlying census tracts, which are subunits within PUMAs. This approach reflects the idea that ethnic enclaves involve not just demographic concentration, but also geographic coherence, consistent with established definitions of ethnic neighborhoods (Logan et al., 2002; Anselin, 1995). We measured local spatial clustering using local Moran’s I statistic (Anselin, 1995), expressed as: ${\mathrm{I}}_i$ = ${\mathrm{z}}_i\sum _k{w}_{ik}{z}_k$, where ${\mathrm{z}}_i$ and ${\mathrm{z}}_k$ denote the standardized proportions of an ethnic group in tract i and neighboring tract k, respectively, and $w_{ik}$ represents the row-standardized spatial weight between the two tracts. Spatial relationships were defined using a first-order queen contiguity matrix, in which census tracts were considered neighbors if they shared a boundary or a vertex. A tract was defined as a high-high cluster when it had an above-average proportion of that ethnic group and the average proportion among its neighboring tracts was also above-average, suggesting a spatially cohesive pattern of the group. Statistical significance was assessed using 999 conditional permutations with a two-sided significance level of .05.

We then returned to the PUMAs identified as enclaves by the double share criterion above and further required that at least 40% or at least 15 of their census tracts be classified as high-high clusters, whichever threshold was lower. The absolute threshold of 15 tracts approximated 40% of an average-sized PUMA (38.6 tracts) and was adopted to prevent PUMAs with larger number of tracts from being excluded solely because of their tract counts, despite containing substantial numbers of high-high clusters. As no established rule exists for aggregating tract-level spatial clustering into a PUMA-level classification, we conducted sensitivity analyses using thresholds of 30%, 50%, and 60%, with corresponding absolute thresholds of 12, 19, and 23 high-high tracts. The results of these analyses are described in the Sensitivity Analysis section.

3.2.3. Neighborhood environment

To assess neighborhood SES and built environment features at the PUMA-level, we examined a broad set of indicators and refined them using exploratory factor analysis (details in S Appendix 1). The number of factors was determined based on eigenvalues greater than 1 and visual inspection of the scree plot. These criteria supported a two-factor solution. The first factor represented socioeconomic disadvantage with strong loadings for poverty rate, rent burden, no bachelor’s degree, unemployment (eigenvalue = 3.12; common variance explained = 58.7%). The second factor represented the built environment, with strong loadings for park access, number of hospitals, public facilities, and litter basket coverage (eigenvalue = 2.25; common variance explained = 42.4%). Factor scores for each construct were generated and used as continuous predictors in regression models. To assess social cohesion, we aggregated five NYCCHS survey items (perceived close-knit neighborhood, neighbors’ willingness to help, trust among residents, shared values, and whether people in the neighborhood get along) at the PUMA level, capturing perceptions of neighborhood connectedness. Factor analysis supported a single-factor solution (eigenvalue = 3.40; common variance explained = 93.8%), and the resulting factor scores were used to represent area-level cohesion. Full factor loadings and related statistics are reported in S Table 1. We also included PUMA-level immigrant-targeted public funding, measured in U.S. dollars and defined as the annual amount of public funding allocated to programs supporting immigrant families, including civic education classes, English for Speakers of Other Languages, and immigration application assistance.

3.2.4. Covariates

We control for individual-level sociodemographic characteristics that may confound the association between enclave residence and health status. These include age group (18–24; 25–64; 65+), sex (female vs. male), nativity (native-born vs. foreign-born), Asian ethnic subgroup membership (East Asian; South Asian; Southeast Asian; Other), marital/partner status (married or cohabiting with a partner vs. divorced, widowed, separated, or never married), education (less than high school vs. high school or above), and household income (below 200% of the federal poverty guideline vs. higher).

3.3. Analyses

We first estimated age-standardized prevalence of health outcomes by Asian ethnic subgroup using NYCCHS data from 2015 to 2020. Estimates were weighted using CHS-provided survey weights and standardized to the age distribution of NYC's adult population. We then compared characteristics of adults residing inside and outside each type of Asian enclave. Next, we estimated two-level mixed-effects logistic regression models, with individuals at level 1 nested within PUMAs at level 2, to assess the association of ethnic enclaves with each health outcome, adjusting for covariates. A random intercept for PUMA was included to allow the baseline odds of each outcome to vary across PUMAs and to account for residual correlation among individuals residing within the same PUMA. Separate models were estimated for obesity, diabetes, and hypertension, with enclave residence categorized as non-enclave, East Asian enclave, or South Asian enclave. Analyses were conducted for the full Asian sample and stratified by ethnic subgroup and nativity. For each outcome, the model was specified as follows:

$$\text{logit}\left[P\left(\text{CM}{\mathrm{B}}_{\text{ij}}=1\right)\right]={\beta }_0+{\beta }_1\mathrm{E}{\mathrm{A}}_{\mathrm{j}}+{\beta }_2\mathrm{S}{\mathrm{A}}_{\mathrm{j}}+{{\beta }_x}^{\prime }{\mathrm{X}}_{\text{ij}}+{\beta }_{\mathrm{z}}{}^{\prime }{\mathrm{Z}}_{\mathrm{j}}+{\mathrm{u}}_{\mathrm{j}},\text{where}{\mathrm{u}}_{\mathrm{j}}\sim \mathrm{N}\left(0,{\sigma }_u^2\right)$$

The subscripts $\mathrm{i}$ and j represent individuals and $\mathrm{j}$ PUMAs, respectively. CMBᵢⱼ represents the binary CMB outcome for individual i in PUMA j. EAⱼ and SAⱼ indicate whether PUMA j is an East Asian or South Asian enclave, respectively, with non-enclave areas as the reference; Xᵢⱼ is a vector of individual-level covariates (e.g., age), and Zⱼ is a vector of PUMA-level factors (e.g., neighborhood SES); and uⱼ represents the PUMA-level random intercept.

To explore explanatory pathways, we compared neighborhood characteristics across enclave types and assessed whether adjustment for these factors attenuated enclave-health associations. CHS-provided survey weights were applied to account for differential probabilities of selection and improve representativeness of estimates for the NYC adult population. Because 6-year pooled data and weights were not provided, we combined three consecutive 2-year datasets and divided each dataset’s 2-year weights by three, a common rescaling practice for pooled repeated cross-sectional data with similar sample sizes (Centers for Disease Control and Prevention, 2021). For the multilevel models, PUMA-level sampling weights were unavailable; therefore, we rescaled the individual-level survey weights to have a mean of 1 within each PUMA, with no specification for weights at the PUMA-level. This normalization preserved the relative values of the individual weights within each PUMA and resolved the convergence difficulties observed with the unscaled weights in preliminary analyses. We also examined an alternative global rescaling of the weights, an unweighted multilevel model, and a single-level logistic regression model with standard errors adjusted for clustering within PUMAs. The findings were substantively similar across these specifications. All analyses were conducted using Stata 18.0.

4. Results

4.1. Health status among Asian ethnic subgroups

Fig. 1 shows clear disparities in health status across AA ethnic subgroups in New York City. Across the survey years, East Asians reported the lowest obesity prevalence (4–6%), substantially lower than that among non-Hispanic White adults (19–20%), non-Hispanic Black adults (34%), and Hispanic adults (30–33%). However, their prevalences of diabetes (9%) and hypertension (19–21%) were comparable to those of non-Hispanic White adults. South Asians had the highest diabetes prevalence (20–26%) compared with the other groups (5–17%) and their hypertension prevalence (28–33%) was similar to that among non-Hispanic Black adults (33–36%) and Hispanic adults (30–32%) and higher than that among the remaining groups. Their obesity rate (16–19%) was the highest among the Asian subgroups and similar to that among non-Hispanic White adults (19–20%). Southeast Asians also showed a high prevalence of hypertension (21–28%) but relatively lower obesity (10–11%) and diabetes rates (5–11%).

Fig. 1.

Health Status among Asians in New York City, 2015–2020 NYCCHS

Note. NH = non-Hispanic. All categories refer to adults. Estimates were weighted using sample weights and age-standardized using standard age adjustment weights derived from the age distribution in New York City.

4.2. Geographic distribution of ethnic-specific Asian enclaves

Using the double share concentration threshold combined with the spatial clustering criterion, we identified five East Asian enclaves and six South Asian enclaves. Fig. 2 presents the geographical locations of each Asian enclave in New York City. It shows that there is no geographic overlap between East Asian and South Asian enclaves. East Asian enclaves are primarily located in northeastern Queens and southwestern Brooklyn, including Flushing and parts of Sunset Park. South Asian enclaves are concentrated in southeastern and south-central Queens, in neighborhoods such as Jamaica and Richmond Hill.

Fig. 2.

Geographic Locations of Ethnic-Specific Asian Enclaves in New York City

Note. Light gray areas denote Joint Interest Areas, including parks and airports. Ethnic-specific Asian enclaves were defined as PUMAs in which the proportion of the corresponding ethnic group was at least twice that group’s share of the overall New York City population (16.4% for East Asians and 8.6% for South Asians). In addition, a PUMA was required to have either at least 40% of its census tracts or at least 15 tracts classified as significant high–high spatial clusters based on the group-specific population proportion.

4.3. Racial/Ethnic composition by enclave classifications

S Table 2 presents the racial and ethnic composition of PUMAs across NYC, stratified by enclave classification. As expected, the ethnic composition of enclave-designated PUMAs shows a substantially higher proportion of their respective subgroup populations. East Asian enclaves had the highest proportion of East Asian residents (33.9%), compared to 5.1% in non-enclave PUMAs. Similarly, South Asian enclaves had 14.9% South Asian residents, compared to 2.8% in non-enclaves. East Asian and South Asian enclaves also differ in their broader racial and ethnic makeup. East Asian enclaves include a high proportion of non-Hispanic White residents (36.0%), whereas South Asian enclaves have high shares of non-Hispanic Black (25.0%) and Hispanic (24.7%) residents. Southeast Asians constituted a slightly larger share of residents in South Asian enclaves (2.8%) than in East Asian enclaves (1.6%).

4.4. Sample characteristics by enclave residency

After linking enclave data to our AA sample, we compared the sociodemographic and health characteristics across enclave residence types. Differences between the reference group and each East Asian and South Asian enclave resident group were assessed with Pearson's chi-squared tests. As shown in Table 1, compared to non-enclave residents, individuals living in East Asian enclaves had significantly lower obesity prevalence (4.3%), while those in South Asian enclaves had notably higher rates of obesity (14.3%), diabetes (15.5%), and hypertension (24.8%). Compared to non-enclave residents, individuals in East Asian enclaves were more likely to be foreign-born. Both East Asian and South Asian enclave residents were more likely to be married or living with a partner. East Asian enclave residents had higher rates of limited education (29.8%vs. 21.6%) and poverty (69.1%vs. 48.1%) than those among non-enclave residents, whereas South Asian enclave residents had the lowest rates of limited education (16.4%).

Table 1.

Sample Characteristics by Ethnic-Specific Asian Enclave Residency, All Asians.

	Non-enclave Resident (Ref)	East Asian Enclave Resident	South Asian Enclave Resident
N	3585	1984	1117
Obesity	8.8%	4.3%***	14.3%***
Diabetes	8.4%	8.2%	15.5%***
Hypertension	17.4%	18.8%	24.8%***
Female	52.7%	50.7%	49.2%
Foreign born	83.3%	90%***	83.8%
Age group
Ages 18–24	16.4%	11.6%***	16.3%
Ages 25–64	72.3%	76.2%*	72.0%
Ages 65+	11.3%	12.3%	11.6%
Married/living with a partner	55.5%	66.1%***	61.4%*
Less than high school education	21.6%	29.8%***	16.4%*
Households in poverty	48.1%	69.1%***	53.5%

Note. Difference between the reference group and each East Asian and South Asian enclave resident group was assessed with Pearson's chi-squared test. *** p < .001; ** p < .01; * p < .05.

4.5. Ethnic-specific Asian enclaves and health outcomes

Fig. 3 presents the adjusted odds ratios (ORs) with 95% confidence intervals from two-level mixed-effects logistic regression models estimating the associations between residence in ethnic-specific Asian enclaves and three CMB health outcomes. Separate logistic regression models predicting obesity, diabetes, or hypertension were estimated by Asian enclave residence status (non-enclaves [reference], East Asian enclaves, and South Asian enclaves). All models adjusted for sociodemographic covariates, including Asian ethnic subgroup membership, which accounts for underlying ethnic differences in individual health status. Living in East Asian enclaves was significantly associated with lower odds of obesity (OR=0.60; 95%CI=0.38, 0.95, p < .05), but not with hypertension or diabetes. In contrast, residence in South Asian enclaves was significantly associated with higher odds of diabetes (OR=1.36; 95%CI=1.02, 1.82, p < .05) and hypertension (OR=1.44; 95%CI=1.07, 1.95, p < .05).

Fig. 3.

Associations between Asian Enclave Residency and Cardiometabolic Health Outcome

Note. Estimates represent adjusted odds ratios (ORs) and 95% confidence intervals from separate two-level mixed-effects logistic regression models predicting each obesity, diabetes, or hypertension by Asian enclave residence status (non-enclaves [reference], East Asian enclaves, and South Asian enclaves). All models were adjusted for age group, sex, nativity, Asian ethnic subgroup, marital or partnership status, educational attainment, and household poverty status. * p < .05.

Subgroup analyses by Asian ethnic subgroup (S Table 3) generally mirrored the patterns observed in the full Asian sample. Enclave-health associations were directionally consistent with the pooled results: East Asian enclave residence tended to be associated with lower odds of obesity, while South Asian enclave residence was associated with higher odds of adverse health outcomes. Although some associations did not reach statistical significance within subgroups, likely due to reduced sample sizes, the general protective pattern of East Asian enclaves and the adverse associations of South Asian enclaves remained evident, regardless of the resident’s own ethnic subgroup.

4.6. Asian enclaves and health outcomes by nativity

In nativity-stratified models (Table 2), enclave associations were pronounced only among foreign-born Asians. Among them, living in East Asian enclaves was significantly associated with lower odds of obesity (OR=0.50, p < .01), and living in South Asian enclaves was associated with higher odds of obesity (OR=1.39, p < .05), diabetes (OR=1.37, p < .05), and hypertension (OR=1.49, p < .01). In contrast, none of the enclave associations reached statistical significance among US-born Asians.

Table 2.

Associations between Ethnic-Specific Asian Enclave and Cardiometabolic Health Outcome by Nativity.

A. US-born
		Obesity				Diabetes				Hypertension
	N (%)	OR		95% CI		OR		95% CI		OR		95% CI
Non-Enclaves (ref)	624 (63%)
East Asian Enclaves	194 (20%)	1.26		0.59	2.69	0.84		0.24	2.99	1.33		0.54	3.29
South Asian Enclaves	170 (17%)	1.01		0.40	2.53	1.33		0.26	6.86	1.03		0.28	3.84

B. Foreign-born
		Obesity				Diabetes				Hypertension
	N (%)	OR		95% CI		OR		95% CI		OR		95% CI
Non-Enclaves (ref)	2961 (52%)
East Asian Enclaves	1790 (31%)	0.50	⁎⁎	0.32	0.78	1.03		0.71	1.50	1.04		0.73	1.49
South Asian Enclaves	947 (17%)	1.39	*	1.03	1.86	1.37	*	1.04	1.79	1.49	⁎⁎	1.12	1.98

Note. Estimates represent adjusted odds ratios (ORs) and 95% confidence intervals from separate two-level mixed-effects logistic regression models predicting each obesity, diabetes, or hypertension by Asian enclave residence status (non-enclaves [reference], East Asian enclaves, and South Asian enclaves). All models were adjusted for age group, sex, Asian ethnic subgroup, marital or partnership status, educational attainment, and household poverty status.

^⁎⁎p < .01;

^⁎p < .05. 

4.7. Variation in neighborhood conditions by enclave status

Descriptive comparisons of PUMA-level characteristics (Table 3) revealed some differences in neighborhood profiles by Asian enclave type. Both enclaves had higher proportions of residents who were not U.S. citizens at birth than non-enclave areas (47–48% vs. 34%). East Asian enclaves had higher shares of residents with limited English proficiency (41%vs. 23% in South Asian enclaves and 21% in non-enclaves) and adults aged 65 or older (16%vs. 14% in both South Asian enclaves and non-enclaves). Socioeconomic profiles were mixed, although both enclaves had higher rates of rent burden (46–48% vs. 44% in non-enclaves) and larger proportions of residents without a bachelor’s degree (66–70% vs. 63% in non-enclaves). Social cohesion indicators were relatively similar across enclave types and non-enclaves, with only modest differences. Both enclaves had fewer built environment amenities, shown as substantially limited access to a park (57–67% vs. 77% in non-enclaves), fewer urgent care facilities (13–17 vs. 24 in non-enclaves), a smaller share of land used for public facilities/institutions (6–7% vs. 11% in non-enclaves), lower litter basket density per square mile (47–61 vs. 171 in non-enclaves), than non-enclave areas. Notable differences emerged between two enclaves in immigrant-targeted funding, with East Asian enclaves receiving the most ($418,021) and South Asian enclaves receiving the least ($70,430).

Table 3.

Differences in PUMA-level Profiles by Asian Enclave Status.

	Non-Enclave	East Asian Enclave	South Asian Enclave
	N = 44	N = 5	N = 6
Demographic Characteristics
Residents age 65 and over, unit: %	13.8	16.2	13.9
Female residents, unit: %	52.7	51.1	51.5
Residents not U.S. citizens at birth, unit: %	34.3	47.8	47.1
Residents self-identifying as limited English proficiency, unit: %	20.7	41.3	22.6
Socioeconomic Status
Residents with incomes below NYC poverty threshold, unit: %	20.0	19.9	16.4
HHs spend >35% on rent, unit: %	44.4	48.1	46.4
Residents without a bachelor's degree+, unit: %	62.6	65.5	70.4
Civilian labor force unemployed, unit: %	4.6	3.4	4.7
Built and Amenity Environment
Residents within walking distance of a park or open space, unit: %	76.8	67.0	56.8
Urgent care hospitals, centers, and health facilities, unit: count	24.2	16.6	13.2
Area public facilities/institutions as the primary land use, unit: %	10.7	6.9	6.1
Litter baskets, unit: density per square mile	171.4	61.3	47.4
Social Environment
Live in a close-knit neighborhood, unit: % agree	66.1	67.6	69.0
Neighbors are willing to help, unit: % agree	77.0	76.0	78.4
Neighbors get along with each other, unit: % agree	70.4	69.3	74.0
Neighbors share the same values, unit: % agree	54.3	55.9	58.1
Neighbors can be trusted, unit: % agree	62.6	68.6	67.6
Budget for Immigrant Family
Funding for literacy and immigrant services, unit: $	$ 245,373	$ 418,021	$ 70,430

Note. Public Facilities and Institutions include hospitals, churches, Asylums and homes, educational structures, community center, etc.

4.8. Community-level pathways linking Asian ethnic-specific enclaves to health

Given that enclave–health associations were observed only among foreign-born AAs, we estimated models for this subgroup that simultaneously incorporated all neighborhood socioeconomic, social, built, and institutional factors to examine whether the enclave associations could be explained by area-level conditions. The findings are presented in Table 4. After adjusting for neighborhood characteristics, the lower odds of obesity observed among foreign-born residents of East Asian enclaves remained significant (OR=0.48, p < .01), suggesting that this association is not explained by the measured community-level factors. In contrast, the higher odds of diabetes and hypertension previously observed in South Asian enclaves were attenuated and no longer statistically significant after accounting for community-level factors. For diabetes, the OR decreased from 1.37 to 1.26 (an approximate 8% reduction), and for hypertension, the OR decreased from 1.49 to 1.29 (about a 13% reduction).

Table 4.

Associations between Asian Enclave Residency and Cardiometabolic Outcomes Adjusting for Community-level Factors, Among Foreign-born.

	Obesity				Diabetes				Hypertension
	OR		95% CI		OR		95% CI		OR		95% CI
Enclave residency
Non-Enclaves (ref)
East Asian Enclaves	0.48	**	0.30	0.79	0.99		0.67	1.48	1.01		0.71	1.44
South Asian Enclaves	1.41		0.97	2.04	1.26		0.94	1.70	1.29		0.90	1.86
Area characteristics
Socioeconomic deprivation	1.15		0.78	1.70	1.47	**	1.14	1.89	1.07		0.82	1.40
Social cohesion	1.10		0.74	1.64	1.13		0.85	1.51	0.91		0.67	1.24
Built environment	1.04		0.77	1.40	0.99		0.78	1.24	0.90		0.73	1.11
Institutional support for immigrants	1.06		0.81	1.39	0.96		0.81	1.14	0.89		0.76	1.03

Note. Estimates represent adjusted odds ratios (ORs) and 95% confidence intervals from separate two-level mixed-effects logistic regression models predicting each obesity, diabetes, or hypertension by Asian enclave residence status (non-enclaves [reference], East Asian enclaves, and South Asian enclaves). All models were adjusted for age group, sex, Asian ethnic subgroup, marital or partnership status, educational attainment, and household poverty status.

** p < .01; * p < .05; + p < .10.

4.9. Sensitivity analysis

Since there are no universally established criteria for defining Asian enclaves (Osypuk et al., 2009; Janevic et al., 2014; Logan et al., 2002; Kye, 2018), we tested several alternative enclave operationalizations based on concentration, spatial clustering, and hybrid approaches, which yielded similar results (S Figures 1–3; S Table 4). East Asian enclaves remained associated with lower odds of obesity only, whereas South Asian enclaves remained associated with higher odds of diabetes and hypertension. Under the more restricted enclave definitions, South Asian enclave residence was also associated with higher odds of obesity. To assess whether differential access to healthcare and opportunities for diagnosis associated with health insurance coverage influenced the observed findings, we stratified the analyses by health insurance status (S Table 5). Although statistical significance slightly changed, the direction of the primary findings was generally consistent. In both insured and uninsured groups, we found associations between East Asian enclave residence and lower odds of obesity, and between South Asian enclave residence and higher odds of diabetes and hypertension.

5. Discussion

Using data from New York City, this study provides insight into how ethnic-specific neighborhood contexts are related to CMB health among AAs in this setting, with findings suggesting that ethnic enclaves may not be uniformly health-protective. By disaggregating Asian populations and operationalizing enclaves at the ethnic subgroup level, we observe contrasting geographic and health patterns that may be obscured by aggregated, pan-ethnic measures. Our findings also show that the relationship between enclave residence and health differs by nativity and that measured neighborhood conditions capture a modest portion of these relationships. In the sections that follow, we discuss these findings in relation to each of the three guiding research questions.

First, East Asian and South Asian communities occupy distinct residential spaces in New York City, consistent with prior work documenting spatial separation across Asian subgroups (Kim and White, 2010; Logan and Zhang, 2013). These geographic differences suggest that pan-national Asian enclave measures may obscure meaningful variation across ethnic-specific communities, which may help explain the largely mixed findings in prior studies that used pan-national Asian enclaves, which ranged from null effects on health (Anderson and Simburger, 2022; Yu et al., 2018), to protective effects (Lim et al., 2017; Yang et al., 2014), and detrimental effects (Hong et al., 2014; Guo et al., 2022).

More importantly, our findings reveal that the health associations of enclave residence vary substantially across ethnic-specific Asian enclaves, with East Asian enclave residence showing an inverse association with obesity and South Asian enclave residence showing positive associations with diabetes and hypertension. These patterns are consistent with prior research highlighting the heterogeneity of enclave effects across ethnic-specific enclaves. Janevic et al. (Janevic et al., 2014) reported elevated risk of gestational diabetes among women residing in South Asian enclaves in New York City, but found no such association in Chinese neighborhoods. These findings underscore that the relationship between enclave residence and health may not be uniform across AA populations in New York City, and the need to move beyond broad racial categories and toward ethnic-specific measures.

We also find meaningful differences by nativity in how ethnic enclave residence is related to CMB outcomes. Protective associations between East Asian enclave residence and obesity, as well as the adverse associations between South Asian enclaves and CMB outcomes, were observed only among foreign-born Asians. Among US-born Asians, these associations were not significant. This pattern contrasts with prior research suggesting that enclaves may offer psychosocial benefits for immigrants while posing risks for US-born Asians (Morey et al., 2020). One possible explanation is that foreign-born and US-born Asians might differ in the extent to which they rely on their immediate residential environments. Kalmijn (Kalmijn, 2022) found that immigrants tend to be more embedded in their neighborhoods than natives, in part due to greater demand for practical and social support arising from limited access to mainstream institutional resources and experiences of social exclusion in wider society. As a result, both the health-supportive and health-constraining features of enclave settings may be more relevant to the daily lives of immigrants as they navigate unfamiliar institutions and fulfill everyday needs. In contrast, US-born Asians may have broader institutional familiarity and greater individual-level resources that can reduce their dependence on the immediate neighborhood.

Our findings reveal important similarities and differences in neighborhood conditions across ethnic-specific Asian enclaves. Both East Asian and South Asian enclaves had slightly higher levels of socioeconomic disadvantage than non-enclave areas, a pattern consistent with prior studies of urban Asian enclaves, which often contrast with more affluent suburban counterparts (Logan et al., 2002; Zhang et al., 2022; Kye, 2018). On the other hand, neighborhood social cohesion indicators were relatively similar between Asian enclaves and non-enclave neighborhoods. This diverges from longstanding theoretical perspectives, which posit that enclaves foster strong ethnic networks and institutions that offer emotional and instrumental support (Zhou and Kim, 2006), which can promote health through reduced exposure to discrimination and reinforcement of healthy behaviors stemming from the culture of origin (Bécares et al., 2012; Charles, 2003). However, several empirical studies have reported that AA enclave residents often perceive lower levels of social cohesion (Hong et al., 2014; Osypuk et al., 2009) and social capital (Guo et al., 2022). One possible explanation is that the ethnic density of AAs is generally lower than that of other racial or ethnic groups, limiting the formation of cohesive networks large enough to generate collective benefits (Hong et al., 2014). It is also important to note that our cohesion measures reflect general perceptions of neighborhood cohesion rather than cohesion specifically among co-ethnic neighbors, thus potentially underestimating the types of social support most relevant within ethnic enclaves.

In addition, both Asian enclaves had less favorable built environment conditions than non-enclave areas, pointing to broader infrastructural underinvestment. These disparities are consistent with prior work. In the City of Los Angeles, Asian-majority neighborhoods receive significantly less park space per capita, just 1.2 acres per 1000 residents, compared with 17.4 acres in white-majority areas (García et al., 2012). These differences are not a result of a general lack of green space in the county but reflect how public resources are unevenly distributed across racial and ethnic lines (Pincetl et al., 2003). The place stratification perspective (Alba and Logan, 1991) offers a useful lens for understanding these patterns, highlighting how discriminatory practices in the housing market and unequal political representation contribute to sustained racial segregation and unequal neighborhood conditions, regardless of length of residency in the US and individual socioeconomic resources (Charles, 2003; Turner et al., 2002). Within this framework, communities that lack the political or institutional capacity to mobilize resources are more likely to experience poor infrastructure and limited access to health-promoting amenities.

Enclaves differed in aggregated resident characteristics and the degree of institutional investment. East Asian enclave residents had higher rates of limited education and poverty, features aligned with early-stage immigrant communities (Guo et al., 2022; White et al., 1993). Yet, East Asian enclaves received the highest levels of immigrant-targeted funding. In contrast, South Asian enclaves showed a disconnect between individual-level advantage and neighborhood-level disinvestment. Despite residents’ relatively high educational attainment, these enclaves had the most under-resourced built environments and the lowest levels of immigrant-related funding. This mismatch aligns most closely with the place stratification framework, which emphasizes how racialized structures can prevent even socioeconomically mobile immigrant groups from translating individual gains into improved neighborhood conditions (Charles, 2003).

Our neighborhood-level factors were associated with attenuation in some, but not all, of the observed associations between ethnic enclave residence and CMB outcomes among foreign-born AAs. For East Asian enclaves, the association with obesity was largely unchanged even after adjusting for neighborhood socioeconomic disadvantage, social cohesion, built environment, and institutional support. This persistence suggests the presence of other unmeasured, enclave-specific characteristics that are beneficial for metabolic health but not captured by our neighborhood-level measures. We elaborate on these potential mechanisms later in this section. For South Asian enclaves, the estimates for diabetes and hypertension were attenuated by 8% and 13%, respectively, once neighborhood factors were adjusted for. These shifts indicate that neighborhood conditions considered in our study may partly help to characterize the observed patterns. However, they capture only a small portion of the association and the loss of significance may also reflect, to some extent, reduced statistical precision after covariate adjustment. The remaining differences point to additional mechanisms beyond those represented by the available neighborhood indicators.

One plausible explanation that warrants future study involves cultural differences in health behaviors. It has been reported that ethnic enclaves facilitate access to culturally preferred foods and traditional diets, as greater availability of ethnic groceries and restaurants may support the retention of healthier eating patterns (Lim et al., 2017). Studies suggest that East Asian diets, characterized by low intake of red meat and high intakes of fish and plant foods (Tsugane, 2021), share some characteristics with the Mediterranean diet (Sakamoto, 2001) that are favorable to cardiovascular and metabolic health. East Asian enclaves may therefore promote healthier diets through restaurants and food options aligned with these traditional, low-fat, vegetable-rich meals, contributing to lower cardiometabolic risks. In contrast, some dietary patterns reported among South Asian populations have high intakes of carbohydrates, saturated fats, and omega-6 polyunsaturated fatty acids, with lower intake of monounsaturated fats and fiber, which are linked to obesity, insulin resistance, diabetes, and cardiovascular disease (Martyn-Nemeth et al., 2017; Misra et al., 2008). Compounding this, high levels of physical inactivity have been reported among South Asian immigrants in countries such as Australia (Nisar et al., 2024), Canada (Mirza et al., 2022), and the UK (Jepson et al., 2012), along with gendered barriers to exercise (Misra et al., 2008), including norms related to religious modesty, avoidance of mixed-sex activities, and concerns about going out alone (Sriskantharajah and Kai, 2007). These cultural expectations often place family and household responsibilities above physical activity, limiting opportunities for women to engage in exercise (Ranasinghe et al., 2013). These country-of-origin norms, when reinforced within enclaves, may reduce opportunities for health-promoting behaviors and be associated with less favorable CMB outcomes. However, because of data limitations, we were unable to empirically test whether individual dietary behaviors, the local food environment, physical activity, or cultural norms contributed to the observed patterns. The extent to which these factors may underlie the observed differences remains an important question for future research.

Another area for future investigation is variation in ethnic-specific cultural institutions. Prior studies suggest that enclave contexts may differ not only in the co-ethnic population density but also in the availability of robust co-ethnic businesses and services, including clinics, cultural centers, religious institutions, and schools, which may reduce experiences of discrimination and facilitate access to culturally congruent resources (Morey et al., 2020). These institutions may contribute to a strong "ethnic feel," marked by visible cultural cues such as signage, language, and retail offerings (Wen et al., 2009). For example, Chinese enclaves in California’s San Gabriel Valley have been described as institutionally complete and culturally vibrant (Li, 1998; Li, 2005). Zhou and Kim (Zhou and Kim, 2006) similarly describe how Chinese and Korean neighborhoods often feature well-developed educational infrastructures, including ethnic language schools and after-school academies, that both reinforce cultural identity and support academic success. Such institutional resources could facilitate access to information, social networks, and opportunities for upward social mobility, and may help explain the protective associations we observed in East Asian enclaves. In contrast, fewer studies have documented similar levels of institutional development in South Asian enclaves. These differences in institutional capacity may therefore provide another possible reason for the contrasting CMB patterns between East and South Asian enclaves in our study. Because we were unable to measure these institutional differences directly, this interpretation should be viewed as a testable hypothesis for future research.

5.1. Limitations

Several limitations warrant consideration. First, our pooled cross-sectional design limits the ability to draw causal inferences. Observed associations may reflect residential self-selection rather than enclave effects per se. Although we adjusted for a range of available sociodemographic characteristics, unobserved factors, including housing affordability, family and co-ethnic networks, migration history, and pre-existing health status, may influence residential location and also be associated with cardiometabolic health. Thus, our findings should be interpreted as associational. Longitudinal data with residential histories can help disentangle whether ethnic enclave residence contributes to changes in health over time or whether individuals with particular health profiles are more likely to reside in certain neighborhoods. Second, although our enclave definitions are more refined than pan-ethnic classifications, they are constrained by the geographic resolution at the PUMA-level, which may obscure important variation at finer spatial scales. Future research should leverage more granular geographic units, such as census tracts, to better capture neighborhood-level dynamics and heterogeneity within enclaves. Third, our neighborhood-level indicators may not fully capture residents’ experiences. Future studies should incorporate additional indicators that reflect features central to the health effects of ethnic enclaves, such as the density of ethnic businesses, cultural institutions, and community organizations, as well as food-environment measures (e.g., availability and accessibility of culturally specific groceries and restaurants). Fourth, all our health outcomes were based on self-reported measures and may therefore be subject to reporting error or underdiagnosis. Although sensitivity analyses stratified by health insurance status produced estimates that were directionally consistent with the primary findings, insurance coverage does not fully capture access to care, healthcare utilization, or opportunities for diagnosis. For example, if East Asian enclave residents were less likely to receive or report a diabetes diagnosis, diabetes prevalence would be underestimated in this group, biasing the corresponding estimate downward and potentially obscuring a positive association. Finally, the generalizability of our findings is limited to New York City, a metropolitan area with distinctive patterns of Asian immigration and settlement. New York City should be viewed as an informative case that illustrates the importance of examining ethnic-specific enclaves within their local historical and institutional contexts. Comparative research across geographic settings is needed to assess the extent to which these patterns extend beyond New York City.

5.2. Conclusion

Overall, our results suggest that residence in East Asian enclaves may be protective against obesity, while residence in South Asian enclaves may be related to higher risks of diabetes and hypertension among foreign-born AAs. These findings challenge the conventional view that Asian ethnic enclaves are universally protective for health. Our results underscore the need for health-specific, ethnicity-specific, and nativity-disaggregated approaches when examining the impact of Asian enclaves on health outcomes. By exploring distinct explanatory pathways, we highlight the importance of moving beyond traditional neighborhood factors, such as socioeconomic status and social cohesion, to incorporate dimensions like the built environment and the investment for immigrants. These multidimensional factors are critical for understanding the mechanisms through which enclaves shape health and for identifying pathways that may confer risk or resilience across diverse Asian subgroups.

Financial disclosure

The authors have no financial relationships to disclose.

Data sharing

This study draws on both publicly available data sources and restricted data product from the New York City Community Health Survey, which is conducted by the New York City Department of Health and Mental Hygiene. Public-use files are freely downloadable from the website at https://www.nyc.gov/site/doh/data/data-sets/community-health-survey.page. Data on participants’ geographic location are available via a data use agreement with the NYC Department of Health. Community-level data used in this study are publicly available from the American Community Survey, NYC Planning’s Community District Profiles, the NYC Department of Sanitation via the Environment & Health Data Portal, and Data2Go.NYC. Analytic code used in this study shall be made available upon request from the corresponding author.

Funding source

This work was supported by the Rutgers-NYU Center for Asian Health Promotion and Excellence under the National Institutes of Health/The National Institute on Minority Health and Health Disparities [grant numbers 1P50MD017356– 01].

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work, the authors used ChatGPT to improve the clarity and readability of the language and to assist with identifying background literature search. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the New York University Institutional Review Board. (Approval No IRB-FY-2023– 7228).

CRediT authorship contribution statement

Eun Young Choi: Writing – review & editing, Writing – original draft, Visualization, Software, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Virginia W. Chang: Writing – review & editing, Validation, Supervision, Resources, Methodology, Investigation, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors have no conflicts of interest to disclose.