Comparison of Disability Rates Among Older Adults in Aggregated and Separate Asian American/Pacific Islander Subpopulations

Abstract

Objectives. We assessed the prevalence and adjusted odds of 4 types of disability among 7 groups of older Asian American/Pacific Islander (AAPI) subpopulations, both separately and aggregated, compared with non-Hispanic Whites.

Methods. Data were from the nationally representative 2006 American Community Survey, which included institutionalized and community-dwelling Hawaiian/Pacific Islander (n = 524), Vietnamese (n = 2357), Korean (n = 2082), Japanese (n = 3230), Filipino (n = 5109), Asian Indian (n = 2942), Chinese (n = 6034), and non-Hispanic White (n = 641 177) individuals aged 55 years and older. The weighted prevalence, population estimates, and odds ratios of 4 types of disability (functional limitations, limitations in activities of daily living, cognitive problems, and blindness or deafness) were reported for each group.

Results. Disability rates in older adults varied more among AAPI subpopulations than between non-Hispanic Whites and the aggregated Asian group. Asian older adults had, on average, better disability outcomes than did non-Hispanic Whites.

Conclusions. This study provides the strongest evidence to date that exclusion of institutionalized older adults minimizes disparities in disabilities between Asians and Whites. The aggregation of Asians into one group obscures substantial subgroup variability and fails to identify the most vulnerable groups (e.g., Hawaiian/Pacific Islanders and Vietnamese).

In 2008, more than 1 in 3 immigrants arriving in the United States were born in Asia, and just under 1 in 5 were 50 years of age or older.¹ Whereas the elderly non-Hispanic White population in the United States is projected to increase by 73% between 2000 and 2025, the comparable Asian American/Pacific Islander (AAPI) population is projected to grow by 246%.² Despite the anticipated growth of Asian older adults in the United States, few studies have examined aging-related outcomes among specific AAPI subpopulations.³ Instead, most research has used the common practice of aggregating all Asians in one category, which results in data that lack specificity and obscures important differences in morbidity and mortality rates among distinct subpopulations. To accommodate the imminent growth of the Asian older adult population in the United States, health policy and planning decisions need to be made now that address the full diversity of health needs spanning the various AAPI subpopulations.

The AAPI population comprises an extremely diverse and heterogeneous group that includes Asian Indian, Chinese, Cambodian, Filipino, Hmong, Japanese, Korean, Laotian, Pakistani, Samoan and other Pacific Islander, Thai, and Vietnamese individuals.⁴ US data sources including the US census,⁴ and in turn, many researchers, use the designation Asian American/Pacific Islander to represent all Asian subpopulations within 1 category. This practice of combining data across multiple AAPI subpopulations creates a semblance of homogeneity across groups that is both inaccurate and misleading.⁵

Although comparative studies of health outcomes among AAPI subpopulations are relatively atypical, existing research demonstrates important intergroup variability in rates of several diseases associated with aging. Both cancer incidence rates—including cancer of the lung,⁶ breast,^7–9 and thyroid¹⁰—and cancer survival rates¹¹ have been shown to vary significantly across AAPI groups. Similarly, disparities in rates of hypertension and heart disease^12–15 and type 2 diabetes^16–18 have been documented across AAPI groups. With respect to diabetes, Hawaiian/Pacific Islanders^17–19 and Filipinos²⁰ have been shown to have elevated levels of type 2 diabetes compared with other AAPI subpopulations and Whites.

Risk profiles for various diseases have also been shown to be distinct to particular AAPI subpopulations.^18,21,22 Hawaiian/Pacific Islanders, for example, are observed to have disproportionately high rates of obesity, which is thought to be a result of genetic factors.¹⁹ In contrast, Filipinos have low rates of obesity despite high rates of type 2 diabetes,²⁰ which suggests that the degree to which various factors contribute to the development of diabetes varies by AAPI subtype.¹⁸

The health outcomes of AAPI populations are also known to vary according to whether the individual was born in the United States.²² In general, the newly immigrated are “healthier” than are the US-born.²³ This “healthy migrant effect” or “positive immigrant selectivity” suggests that those able to successfully migrate to the United States are healthier, more physically fit, and have greater drive than those who remain in their country of origin.^3,23–25 Migrant selectivity effects, however, do not always confer good health. Asian immigrants who were predominantly refugees, such as the Vietnamese, have on average poorer health than do other AAPI groups and the US-born population.^3,26

Effects of acculturation are also thought to vary across immigrant groups. One line of argument suggests that, despite advantage at arrival, risk for disease increases as immigrants adapt to the American lifestyle. Immigrants living in the United States for many years, therefore, begin to have health profiles comparable to those of their American-born counterparts. Research among AAPI subpopulations has shown that the health advantage erodes as the duration of time since immigration increases.^3,23 Environmental factors related to lifestyle, such as physical activity level and dietary animal fat intake, may contribute to escalating levels of risk.²⁷ Specifically, risk for obesity and type 2 diabetes has been shown to increase as immigrants become more affluent and urbanized.²⁸ Another line of argument suggests that acculturation confers health benefits. Immigrants who have lived in the United States for several years may be better able to accumulate health resources such as access to medical insurance and health care.²⁹ The disparity of health outcomes observed across AAPI subpopulations, therefore, may be best understood in the context of interactions between migrant selectivity effects, negative versus positive acculturation, and details of immigration history.²⁶

Disability is an important health outcome to assess in older adult populations because it is closely related to the need for both formal and informal³⁰ long-term care. Research on disability is especially important given the projected growth of the Asian older adult population over the next 15 years.² Few population-based studies exist that have focused on the health of older AAPI subpopulations, reflecting the shortage of datasets large enough for meaningful comparisons between subgroups to be made.³ As a result, current studies focusing on disability in older populations rely on an aggregate Asian category,³¹ which obscures potential variation across AAPI subpopulations and fails to identify particularly vulnerable groups.

Recently, pooled datasets from sources such as the National Health Interview Survey^3,32 and the US Census²⁶ have been used by researchers to yield large enough sample sizes to study disability prevalence in older AAPI groups. These datasets, however, only sample from community-dwelling individuals and exclude those living in an institution; as a result, national disability levels of older adults are underestimated.

The purpose of the current study was to assess the prevalence of 4 types of disability in older AAPI subpopulations, both separately and aggregated, compared with non-Hispanic Whites by use of the 2006 American Community Survey, which sampled both community-dwelling and institutionalized Americans. We addressed the following research questions: (1) What are the prevalence and national population estimates of 4 types of disability [functional limitations, limitations in activities of daily living (ADLs), cognitive problems, and blindness or deafness] in Hawaiian/Pacific Islander, Vietnamese, Korean, Japanese, Filipino, Asian Indian, Chinese, and non-Hispanic White US racial groups? (2) What are the prevalence and national population estimates of the 4 types of disability in the aggregated group of Asians compared with non-Hispanic Whites and how do the rates in the aggregated Asian group compare with the rates in each of the distinct 7 AAPI subpopulations (Hawaiian/Pacific Islander, Vietnamese, Korean, Japanese, Filipino, Asian Indian, and Chinese)? (3) How do the rates of the 4 disability types in each of the distinct racial groups vary when the sample includes institutionalized respondents versus the sample restricted to solely community dwellers? (4) How does the odds ratio of each of the 4 types of disability vary among the 7 distinct racial groups when adjusted for the following: age and gender (model 1); age, gender, and education (model 2); or age, gender, education, immigration history, citizenship status, and language spoken at home (model 3).

METHODS

The nationally representative 2006 American Community Survey included both community-dwelling and institutionalized Americans³³ and had a response rate of 97.5%.³⁴ We selected a subsample (unweighted n = 663 455) for study that included all respondents aged 55 years and older who reported that they were non-Hispanic White (n = 641 177) or from one of the following racial groups: Hawaiian/Pacific Islander (n = 524), Vietnamese (n = 2357), Korean (n = 2082), Japanese (n = 3230), Filipino (n = 5109), Asian Indian (n = 2942), or Chinese (n = 6034). To ensure that there was no overlap between AAPI groups, we excluded respondents who reported more than one racial group.

The main outcome measures were 4 types of disability: functional limitations, limitations in ADLs, cognitive problems, and blindness or deafness. Limitations in ADLs and cognitive problems were determined by asking respondents whether, owing to a physical, mental, or emotional condition lasting 6 months or more, they had difficulty doing any of the following activities: “dressing, bathing, or getting around inside the home” (ADL) or “learning, remembering, or concentrating” (cognitive problems), respectively. Respondents were also asked whether they had any of the following long-lasting conditions: (1) a condition that substantially limited one or more basic physical activities such as walking, climbing stairs, reaching, lifting, or carrying (classified as functional limitations) or (2) blindness or deafness or a severe vision or hearing impairment.

The independent variable of race was divided into 8 categories: non-Hispanic White (hereafter called White), Hawaiian/Pacific Islander, Vietnamese, Korean, Japanese, Filipino, Asian Indian, and Chinese (reference category). Control variables included age (50–64, 65–74, 75–84, and 85 years or older), gender, education (highest level), immigration status (yes or no), decade of immigration, citizenship status (yes or no), and whether English was spoken at home (yes or no). Immigration status and decade of immigration were combined into one variable termed immigration history with 8 categories: immigrated before 1950, immigrated in 1950 to 1959, immigrated in 1960 to 1969, immigrated in 1970 to 1979, immigrated in 1980 to 1989, immigrated in 1990 to 1999, immigrated in 2000 or later, and born in the United States. We note that the term Asian American/Pacific Islander implies American citizenship; however, in the context of this research, we used the term AAPI to describe those who were living in the United States at the time of the survey, regardless of their citizenship status. Institutionalization referred to individuals in nursing homes, in-patient hospice facilities, psychiatric hospitals, and adult correction facilities.³³

Eight racial groups were compared according to the 4 disability types. The weighted prevalence, population estimates, and odds ratios of the 4 disability types in the full sample and in the subsample of community-dwellers only are reported. We also conducted a series of logistic regression analyses to investigate the association between race and each of the 4 disability types for the complete sample only, by using 3 nested models that controlled for the following variables: (1) gender and age; (2) gender, age, and education; and (3) gender, age, education, immigration history, citizenship status, and English language spoken at home.

RESULTS

Variation in disability rates was greater among the older adult AAPI subpopulations than between the White group and the aggregated Asian group (Table 1). For example, there was a 0.2% spread between the rate of cognitive problems for Whites (10.4%) and the rate for the aggregated Asian group (10.2%). Yet, Vietnamese respondents had a prevalence rate of cognitive problems of 16.6%, which was more than double the rate for Koreans (7.6%).

TABLE 1.

Prevalence, National Population Estimates, and Odds Ratios of Disability Among Separate and Aggregated Older Asian American/Pacific Islander Subpopulations Compared With Non-Hispanic Whites: American Community Survey, 2006

	Full Sample			Subsample
	Prevalence Rate, %	National Population Estimate	OR (95% CI)	Prevalence Rate, %	National Population Estimate	OR (95% CI)
Functional limitations
Separate comparison
Non-Hispanic White	25.4	13 804 168	1.58 (1.49, 1.68)	23.5	12 340 578	1.52 (1.43, 1.62)
Hawaiian/Pacific Islander	26.3	13 761	2.09 (1.73, 2.54)	25.2	12 940	2.04 (1.68, 2.49)
Vietnamese	21.6	57 809	1.55 (1.39, 1.72)	21.2	56 370	1.56 (1.40, 1.73)
Korean	16.4	37 246	1.08 (0.95, 1.21)	15.1	33 748	1.02 (0.90, 1.15)
Japanese	17.9	50 249	0.86 (0.77, 0.97)	16.5	45 334	0.85 (0.76, 0.95)
Filipino	19.1	97 666	1.23 (1.13, 1.35)	18.8	94 333	1.25 (1.14, 1.37)
Asian Indian	18.3	58 578	1.35 (1.22, 1.50)	18.1	57 626	1.36 (1.23, 1.51)
Chinese (Ref)	17.2	108 599	1.00	16.5	102 387	1.00
Aggregate comparison
Non-Hispanic White	25.4	13 804 168	1.36 (1.32, 1.41)	23.5	12 340 578	1.31 (1.27, 1.36)
Asian, all grouped (Ref)	18.5	423 908	1.00	17.8	402 738	1.00
ADL limitations
Separate comparison
Non-Hispanic White	9.0	4 915 191	1.24 (1.13, 1.36)	6.8	3 595 212	1.09 (0.99, 1.20)
Hawaiian/Pacific Islander	11.4	5 966	2.27 (1.73, 2.97)	9.8	5 005	2.09 (1.56, 2.78)
Vietnamese	8.4	22 460	1.55 (1.32, 1.81)	8.0	21 148	1.60 (1.36, 1.88)
Korean	6.2	14 148	1.09 (0.91, 1.31)	5.0	11 213	0.97 (0.80, 1.19)
Japanese	6.1	17 258	0.69 (0.58, 0.81)	4.8	13 211	0.65 (0.54, 0.78)
Filipino	6.3	32 102	1.00 (0.87, 1.15)	5.8	29 006	1.03 (0.89, 1.19)
Asian Indian	5.8	18 671	1.15 (0.98, 1.36)	5.6	17 686	1.18 (1.00, 1.40)
Chinese (Ref)	6.8	43 218	1.00	6.0	37 166	1.00
Aggregate comparison
Non-Hispanic White	9.0	4 915 191	1.18 (1.12, 1.24)	6.8	3 595 212	1.03 (0.98, 1.09)
Asian, all grouped (Ref)	6.7	153 823	1.00	6.0	134 435	1.00
Cognitive problems
Separate comparison
Non-Hispanic White	10.4	5 663 360	0.90 (0.83, 0.97)	8.6	4 500 143	0.81 (0.75, 0.88)
Hawaiian/Pacific Islander	13.0	6 789	1.51 (1.17, 1.93)	11.6	5 955	1.41 (1.09, 1.83)
Vietnamese	16.6	44 448	1.99 (1.77, 2.25)	16.4	43 423	2.06 (1.82, 2.33)
Korean	7.6	17 364	0.80 (0.68, 0.94)	6.4	14 226	0.71 (0.60, 0.85)
Japanese	9.5	26 770	0.72 (0.63, 0.83)	7.9	21 713	0.67 (0.58, 0.78)
Filipino	8.4	42 893	0.83 (0.74, 0.93)	7.9	39 824	0.84 (0.74, 0.95)
Asian Indian	8.5	27 209	0.98 (0.85, 1.12)	8.3	26 273	0.99 (0.86, 1.14)
Chinese (Ref)	10.7	67 355	1.00	9.8	61 204	1.00
Aggregate comparison
Non-Hispanic White	10.4	5 663 360	0.89 (0.86, 0.93)	8.6	4 500 143	0.81 (0.78, 0.85)
Asian, all grouped (Ref)	10.2	232 828	1.00	9.4	212 618	1.00
Blindness/deafness
Separate comparison
Non-Hispanic White	12.1	6 592 304	1.25 (1.16, 1.36)	11.3	5 950 883	1.25 (1.15, 1.36)
Hawaiian/Pacific Islander	10.5	5 514	1.47 (1.12, 1.94)	10.4	5 330	1.54 (1.17, 2.04)
Vietnamese	9.8	26 163	1.26 (1.09, 1.46)	9.7	25 814	1.31 (1.14, 1.52)
Korean	6.4	14 692	0.82 (0.69, 0.98)	6.3	14 201	0.87 (0.72, 1.03)
Japanese	9.5	26 588	0.82 (0.71, 0.95)	8.9	24 545	0.83 (0.71, 0.96)
Filipino	8.9	45 599	1.09 (0.96, 1.22)	8.8	44 158	1.12 (0.99, 1.27)
Asian Indian	7.1	22 861	0.97 (0.83, 1.12)	6.9	22 028	0.97 (0.84, 1.13)
Chinese (Ref)	9.4	59 478	1.00	9.0	55 769	1.00
Aggregate comparison
Non-Hispanic White	12.1	6 592 304	1.24 (1.19, 1.30)	11.3	5 950 883	1.22 (1.17, 1.27)
Asian, all grouped (Ref)	8.8	200 895	1.00	8.5	191 845	1.00

Note. ADL = activities of daily living; CI = confidence interval; OR = odds ratio. Respondents were aged 55 years and older. The presence of a disability was examined in the full sample of community-dwelling and institutionalized respondents and in a subsample of community-dwelling respondents only. Odds ratios were adjusted for age and gender.

The age- and gender-adjusted odds ratio (OR) of ADL limitations was significantly higher for Whites than for Asians as a group in the full sample [OR = 1.18; 95% confidence interval (CI) = 1.12, 1.24; 9.0% versus 6.7%], but not when institutionalized respondents were excluded (OR = 1.03; 95% CI = 0.98, 1.09; 6.8% versus 6.0%; Table 1)

When the age- and gender-adjusted odds were examined (Table 2, model 1), compared with the Chinese reference group, the Hawaiian/Pacific Islander and Vietnamese subgroups had the highest odds for all 4 disabilities among the AAPI subpopulations. In contrast, Japanese respondents had the best disability profile; they had significantly lower age- and gender-adjusted odds for all 4 disabilities than did the Chinese. Koreans had significantly lower odds of cognitive problems and blindness or deafness, and Filipinos had lower odds of cognitive problems than did the Chinese. Both Filipinos and Asian Indians had higher odds of functional limitations. Compared with the Chinese group, Whites had significantly higher odds of functional limitations, ADL limitations, and blindness or deafness, but had significantly lower odds of cognitive problems.

TABLE 2.

Odds Ratios of Disabilities in Older Asian American/Pacific Islander Subpopulations and Non-Hispanic Whites: American Community Survey, 2006

	Model 1, OR (95% CI)	Model 2, OR (95% CI)	Model 3, OR (95% CI)
Functional limitations
Non-Hispanic White	1.58 (1.49, 1.68)	1.91 (1.79, 2.04)	1.71 (1.59, 1.83)
Hawaiian/Pacific Islander	2.09 (1.73, 2.54)	2.11 (1.74, 2.57)	2.09 (1.71, 2.54)
Vietnamese	1.55 (1.39, 1.72)	1.40 (1.26, 1.56)	1.25 (1.12, 1.40)
Korean	1.08 (0.95, 1.21)	1.21 (1.07, 1.36)	1.26 (1.12, 1.43)
Japanese	0.86 (0.77, 0.97)	1.09 (0.97, 1.22)	1.06 (0.94, 1.19)
Filipino	1.23 (1.13, 1.35)	1.52 (1.39, 1.67)	1.55 (1.41, 1.70)
Asian Indian	1.35 (1.22, 1.50)	1.63 (1.47, 1.82)	1.69 (1.51, 1.88)
Chinese (Ref)	1.00	1.00	1.00
ADL limitations
Non-Hispanic White	1.24 (1.13, 1.36)	1.54 (1.40, 1.69)	2.33 (2.10, 2.59)
Hawaiian/Pacific Islander	2.27 (1.73, 2.97)	2.40 (1.82, 3.15)	3.20 (2.43, 4.22)
Vietnamese	1.55 (1.32, 1.81)	1.42 (1.21, 1.67)	1.18 (1.00, 1.39)
Korean	1.09 (0.91, 1.31)	1.21 (1.00, 1.45)	1.24 (1.03, 1.50)
Japanese	0.69 (0.58, 0.81)	0.87 (0.73, 1.04)	1.29 (1.08, 1.54)
Filipino	1.00 (0.87, 1.15)	1.17 (1.02, 1.35)	1.18 (1.02, 1.36)
Asian Indian	1.15 (0.98, 1.36)	1.32 (1.12, 1.57)	1.29 (1.09, 1.53)
Chinese (Ref)	1.00	1.00	1.00
Cognitive problems
Non-Hispanic White	0.90 (0.83, 0.97)	1.18 (1.10, 1.28)	1.61 (1.48, 1.76)
Hawaiian/Pacific Islander	1.51 (1.17, 1.93)	1.63 (1.26, 2.10)	2.05 (1.58, 2.65)
Vietnamese	1.99 (1.77, 2.25)	1.86 (1.64, 2.10)	1.52 (1.34, 1.73)
Korean	0.80 (0.68, 0.94)	0.92 (0.78, 1.08)	0.98 (0.83, 1.15)
Japanese	0.72 (0.63, 0.83)	1.00 (0.87, 1.16)	1.39 (1.20, 1.61)
Filipino	0.83 (0.74, 0.93)	1.03 (0.91, 1.16)	1.04 (0.92, 1.17)
Asian Indian	0.98 (0.85, 1.12)	1.18 (1.02, 1.36)	1.13 (0.98, 1.30)
Chinese (Ref)	1.00	1.00	1.00
Blindness/deafness
Non-Hispanic White	1.25 (1.16, 1.36)	1.44 (1.33, 1.56)	1.33 (1.21, 1.45)
Hawaiian/Pacific Islander	1.47 (1.12, 1.94)	1.47 (1.11, 1.93)	1.45 (1.10, 1.92)
Vietnamese	1.26 (1.09, 1.46)	1.16 (1.00, 1.34)	1.02 (0.88, 1.18)
Korean	0.82 (0.69, 0.98)	0.90 (0.75, 1.07)	0.96 (0.80, 1.14)
Japanese	0.82 (0.71, 0.95)	0.98 (0.85, 1.13)	0.97 (0.83, 1.13)
Filipino	1.09 (0.96, 1.22)	1.24 (1.10, 1.40)	1.26 (1.11, 1.42)
Asian Indian	0.97 (0.83, 1.12)	1.11 (0.96, 1.29)	1.10 (0.95, 1.28)
Chinese (Ref)	1.00	1.00	1.00

Note. ADL = activities of daily living; CI = confidence interval; OR = odds ratio. Respondents were institutionalized and community-dwelling elderly persons aged 55 years and older. Model 1 was adjusted for age and gender; model 2 was adjusted for age, gender, and education; and model 3 was adjusted for age, gender, education, and immigration variables. Immigration variables included immigration status, decade of immigration, citizenship status, and whether English was spoken at home.

Adjustments for education (Table 2, model 2), immigration history, citizenship status, and language spoken at home (Table 2, model 3) resulted in declines in the odds of all 4 disabilities for Vietnamese respondents and an increase in the odds for Filipinos and Koreans. In particular, the full adjustment (model 3) resulted in the Vietnamese no longer demonstrating significantly different odds of ADL limitations and blindness or deafness compared with the Chinese. Similarly, Koreans no longer had significantly lower odds of cognitive problems and blindness or deafness than did Chinese respondents. Although the age- and gender-adjusted odds of functional limitations and ADL limitations for Koreans were not significantly different from those for Chinese, when full adjustments were made, Korean respondents had significantly elevated odds for these outcomes.

After adjustment for education (model 2), the Japanese respondents no longer had significantly lower odds of all 4 disability types than did the Chinese respondents. When further adjustments were made for immigration history, citizenship status, and language spoken at home (model 3), the Japanese respondents had significantly higher odds of ADL limitations and cognitive problems than did the Chinese respondents.

Full adjustments did not substantially change the elevated odds of functional limitations or blindness or deafness among Hawaiian/Pacific Islanders. However, these adjustments resulted in a large increase in the odds of ADL limitations and cognitive problems. Hawaiian/Pacific Islanders had the highest odds of all 4 disabilities among the AAPI groups and Whites after full adjustments were made. Compared with the Chinese respondents, Hawaiian/Pacific Islanders had at least twice the odds of functional limitations, ADL limitations, and cognitive problems.

For Whites, adjustments for education (model 2) resulted in an increase in the odds of all 4 disabilities. Although Whites had significantly lower odds of cognitive problems when only age and gender were adjusted for, including education in the analysis resulted in this difference becoming significantly higher than for the Chinese. Adjustment for immigration history and language spoken at home resulted in a decline in the odds of functional limitations and of blindness or deafness and an increase in the odds of ADL limitations and cognitive problems. However, Whites had significantly higher odds of all 4 types of disability than did Chinese respondents.

DISCUSSION

We have presented evidence that greater variation exists in rates of disabilities among older AAPI subpopulations than between Asians as an aggregated group and Whites. In fact, aggregation seems to cloak particularly vulnerable groups such as Hawaiian/Pacific Islanders and to a lesser degree the Vietnamese. The finding of large between-group differences among AAPI populations is consistent with previous disability research focusing on community-dwelling older Asian subpopulations.^3,26,32

Hawaiian/Pacific Islanders had higher odds of all 4 disability types than did the Chinese in the fully adjusted analysis (model 3). Previous research has similarly reported greater levels of disability (i.e., activity limitation) for Hawaiian/Pacific Islanders than for other AAPI groups.³ The elevated odds of disability found for Hawaiian/Pacific Islanders may be related to their disproportionately high levels of obesity¹⁹ and type 2 diabetes.^17,19 The propensity for obesity of Hawaiian/Pacific Islanders is speculated to have developed from a “thrifty genotype” that maximizes efficient intake and utilization of food.³⁵ Although efficient intake of food was at one time highly advantageous—allowing the original island settlers to survive the long journeys between the Pacific Islands—within the current US environment of overabundance of inexpensive food with poor nutritional value, it may relate to the high rates of obesity and in turn the elevated odds of disability observed for this group. Moreover, only a small number of Hawaiian/Pacific Islanders are migrants to the United States,³ which suggests that this group may benefit less from the positive selection effects associated with immigration, thus accounting in part for their worse observed outcomes.

Although to a lesser degree than seen in the Hawaiian/Pacific Islander group, the Vietnamese tended to have higher rates of disability than did other AAPI subgroups. This is consistent with previous research demonstrating greater disability for this group.^3,26 Poor health outcomes of Vietnamese may reflect their history as refugees in the United States²⁶ and exposure to war in Vietnam during the 1960s and 1970s.

Despite poorer outcomes for Hawaiian/Pacific Islanders and Vietnamese, AAPI subpopulations, in general, had better disability outcomes than did Whites. This finding supports previous research.^3,32 Many reasons are possible for the better health of Asian groups compared with Whites. First, the findings are consistent with the “healthy migrant” effect that conjectures that individuals who successfully immigrate are able to do so because they are healthier and more robust than are those who stay in their country of birth. Thus, favorable health and mortality outcomes observed in immigrant populations may actually reflect a positive self-selection bias.^3,23–25,36 Second, to immigrate to the United States, individuals must meet certain minimum health standards imposed by the US government, which further contributes to the positive selection effect. Third, protective effects associated with various Asian cultures, such as a firmer bond to community and family and healthier dietary habits^37,38 that include a low intake of saturated fat, red meat, and dairy, as well as a high intake of vegetables,³⁹ may facilitate better health.

Asian Indians may derive comparatively more benefit from “healthy migrant effects” because they are a relatively recent group to arrive in the United States. This speculation is consistent with our finding that Asian Indians had one of the lowest rates of disabilities (excluding functional limitations) and with Coustasse et al., who found that Asian Indians have better disability outcomes than do Chinese and Filipino Americans.³² In contrast, Mutchler et al. found Asian Indians to have a higher probability of disability than other AAPI subpopulations.²⁶

Despite better disability outcomes, on average, for AAPI subpopulations than for Whites, considerable variation existed between AAPI groups. Variation in outcomes may represent differences among groups in the timing of arrival to the United States and country of origin, which in turn reflect differing degrees of migrant selectivity effects and negative (adopting American lifestyles) versus positive (accumulating health resources) acculturation effects. Therefore, to understand the independent effect of AAPI group on disability, variables such as immigration history, socioeconomic status, and other acculturation variables such as language spoken at home must be accounted for. For example, Japanese individuals were found to have better age- and gender-adjusted disability outcomes than the Chinese across all 4 disability types. However, this probably reflected the higher socioeconomic status of the Japanese, many of whom are second- or third-generation immigrants.⁴ The Japanese have a median annual income that is more than $10 000 higher than that of other AAPI subpopulations (excluding Asian Indians, who have a similar income)⁴ and have one of the highest levels of university degree completion among the AAPI groups.³ When model 2 was adjusted for education level, the Japanese no longer differed significantly from the Chinese in any of the 4 disability types. When further adjustments were made for immigration history, citizenship status, and language spoken at home, the Japanese had significantly elevated odds of ADL limitations and cognitive problems. This finding would have remained hidden had SES and immigration-related variables not been controlled for.

In general, education and immigration variables (i.e., immigration history, citizenship status, and English language spoken at home) seemed to have a large effect on disability outcomes. Education has been found to be protective of cognitive problems,⁴⁰ which might explain why Whites had lower age- and gender-adjusted odds than did Chinese respondents for this outcome only. It may also explain the higher age- and gender-adjusted odds of cognitive problems among Vietnamese respondents, who had one of the lowest levels of university degree completion among AAPI groups.⁴

One final, important point to emphasize is that the results based on the subsample (which excluded institutionalized individuals) underestimated disability levels for each of the racial groups compared with the full sample as well as the gap in disability rates between Whites and the AAPI subpopulations. The underestimation was the largest for Whites likely because of the high use of nursing homes in this group (E. Fuller-Thomson and M. Chi, unpublished data, 2009). This provides strong evidence that institutionalized individuals need to be included in population-based research investigating the disability outcomes of older adults.

The limitations of this study must be considered when interpreting the results. It is important to acknowledge that race, ethnicity, and culture can shape how disability is perceived, and in turn, how it is reported in national surveys.³¹ This may be especially true with respect to disability types associated with culturally specific stigma (i.e., cognitive decline). Blindness and deafness, however, are probably less vulnerable to personal interpretation. Future research would benefit from the use of clinical assessments of disability.

Important groups were missing from this analysis, including individuals who reported belonging to multiple races, who will represent a growing group of older American adults in the decades to come, and Laotian and Hmong individuals, for whom there were insufficient sample sizes to allow for meaningful comparisons.

Income and wealth vary greatly among AAPI subpopulations⁴ and are highly correlated with older adults' disability levels. However, we could not adjust for these variables in our analyses. Income information was not gathered for respondents who were institutionalized and wealth was not available in the data set.

Finally, it should be noted that the results of this study apply only to the current time and may not remain accurate for future US-born generations. For these generations, the protective effects of immigration on health may disappear, yet access to resources such as education and health care may improve. Future research is needed to monitor how these changes over time impact disability outcomes.

The positive aspects of this study included the use of a large, representative sample of both community-dwelling and institutionalized older Americans, which allowed us to accurately estimate national disability levels among 7 AAPI subpopulations. Another positive feature of this study was the truly multilingual nature of the data collection. The American Community Survey provides informational brochures in Korean, Vietnamese, and Chinese and has trained multilingual staff to conduct the telephone interviews, thereby improving response rates among non-Anglophones. Finally, the inclusion of disability types, such as cognitive problems and deafness or blindness, is a unique addition to previous studies of disability among AAPI subpopulations.

In conclusion, we showed that greater variation existed in disability rates among the 7 older AAPI subpopulations than between Whites and Asians as an aggregated group. The aggregation of Asians into one group obscures such variability and fails to identify vulnerable groups. The disparity in disability outcomes across AAPI groups likely reflects a complex interplay between migrant selection effects, positive versus negative acculturation effects, and socioeconomic status factors that relate to both the timing of immigration and the country of origin. To best accommodate the imminent growth of the AAPI older adult population, accurate and representative data are required for health policy and planning decisions to be made that take into consideration the full diversity of the health needs of the distinct AAPI subpopulations. Our research may contribute to such planning efforts.