Social Determinants of Physical Self-Rated Health among Asian Americans ; Comparison of Six Ethnic Groups

Background: A growing literature has revealed ethnic group differences in determinants and meanings of their self-rated health (SRH). Aim: To explore ethnic variations in the effects of socioeconomic determinants on poor physical SRH of Asians in the United States. Methods: Data came from the National Asian American Survey (NAAS), 2008, with 4977 non-U.S. born Asian Americans, including Asian Indian (n = 1150), Chinese (n = 1350), Filipino (n = 603), Japanese (n = 541), Korean (n = 614), and Vietnamese (n = 719) Americans. Demographic factors (age and gender), socioeconomic status (SES; education, employment, income, and marital status), and physical SRH were measured. Ethnic-specific logistic regressions were applied for data analysis where physical SRH was the outcome and demographic and social determinants were predictors. Results: According to logistic regressions, no social determinant was consistently associated with physical SRH across all ethnic groups. Being married was associated with better physical SRH in Asian Indians and worse SRH in the Filipino group. Education was associated with better SRH in Asian Indian, Chinese, Korean, and Vietnamese Americans. High income was associated with better SRH in Chinese, Filipino, and Vietnamese Americans. Employment was associated with better SRH in Filipino Americans. Conclusion: Social determinants of physical SRH vary across ethnic groups of Asian Americans. Different ethnic groups are differently vulnerable to various social determinants of health. Application of single item SRH measures may be a source of bias in studies of health with ethnically diverse populations. Policy makers should be aware that the same change in social determinants may not result in similar change in the health of ethnic groups.

A growing body of research has shown that demographic, SES, and health determinants impact the SRH of diverse ethnic groups differently [12,26,28].The very same factors show different patterns of association with SRH across ethnic groups [30][31][32][33][34][35][36].Overall, it is believed that poor SRH better reflects the real health need for White, rather than non-White, populations, including Asians, African Americans, and Hispanics [12].For instance, poor SRH better predicts premature risk of mortality for Whites than non-Whites [30].Even within a racial group, such as Asians, ethnic variations exist in correlates of SRH [28,31].However, very few studies have compared ethnic groups of Asians in the United States for social determinants of physical SRH.

Aims
This study was conducted to compare six ethnic groups of Asian Americans for social determinants of poor physical SRH.

Design and Setting
Using a cross-sectional design, this study was a secondary analysis of the National Asian American Survey (NAAS), 2008.The study is on the six largest national-origin Asian groups in the United States.The study was funded by the James Irvine Foundation, Eagleton Institute of Politics, Rutgers University, Carnegie Corporation, and Russell Sage Foundation [37].

Interviews
Data collection was conducted using telephone interviews.Interviews were conducted between 12 August and 29 October, 2008.Survey interviews were performed in eight languages, which were chosen according to the interviewee's preference.Interview languages were English, Cantonese, Mandarin, Korean, Vietnamese, Tagalog, Japanese, and Hindi.Overall, 40 percent of the interviews were conducted in English, as it was the preferred language by the participant.The mode of data collection was computer-assisted telephone interview (CATI).Forty-seven percent of respondents (12% of all valid numbers dialed) agreed to take the survey [37].

Data Collection
The study measured demographics, SES, political behaviors, as well as experiences related to immigration in the United States.Some of the constructs measured included discrimination, religious attendance, and social networks.The interviews took about 29 min on average [37].

Participants
The NAAS included 5159 individuals who all self-identified as Asian/Asian American residents of the United States.Asians/Asian Americans were defined as adults in the United States who had a family background from an Asian country.The study excluded Middle Eastern countries.The study sample composed of 4977 non-U.S.born Asian Americans including Asian Indian (n = 1150), Chinese (n = 1350), Filipino (n = 603), Japanese (n = 541), Korean (n = 614), and Vietnamese (n = 719).We did not include 182 additional respondents who identified as multi-ethnic or were from other Asian countries [37].

Ethics
The NAAS study protocol was approved by the University of California (Irvine and Riverside) Institutional Review Board (IRB).All participants provided consent.

Measures
Physical Self-Rated Health.We asked participants "How would you rate your overall physical health?"Response items included the following five categories: (1) excellent; (2) very good; (3) good; (4) fair; and (5) poor.Single-item SRH measures strongly correlate with multi-item health measures [9].Single-item SRH measures also predict risk of mortality, net of confounders [15].Reliability of single-item SRH measures is shown to be high [38].These single-item measures also strongly correlate with standard well-being scales [38].
Demographic Factors.Demographic factors included gender (dichotomous variable, males [reference category] vs. females) and age (continuous measure).
Ethnicity.In this study, ethnicity was self-identified and included Asian Indian, Chinese, Filipino, Japanese, Korean, and Vietnamese.The items to measure ethnicity were by these questions: (1) What race or ethnicity do you consider yourself?(2) Are there any other racial or ethnic groups that describe you? (3) What part of Asia is that part of your family from?(4) What country were you born in?

Weights
To accommodate the National Asian American Survey (NAAS) multi-stage sampling design, we applied sampling weights to all our data analysis.This approach enabled us to generate nationally representative statistics.Taylor series linearization was used to estimate design-based standard errors and variances.To perform our subsample analyses, we applied sub-pop survey commands.

Analysis Plan
Stata 13.0 (Stata Corp., College Station, TX, USA) was used to conduct the analyses.For descriptive statistics, we reported mean (SE) and proportions (SE).For multivariable analysis, we ran logistic regressions in the pooled sample, as well as in each ethnic group.In all models, poor SRH was the outcome.First, we ran a model in the pooled sample.Then we ran ethnic-specific models.Odds Ratio (OR) with 95% Confidence Intervals (CI) were reported.A p-value less than 0.05 was considered significant.

Descriptive Statistics
Table 1 provides a summary of descriptive statistics for each ethnic group.Best SRH was reported by Asian Indians, followed by Japanese.Worst SRH was reported by Koreans, followed by Vietnamese.Asian Indian and Other Asians were the youngest group, while Filipinos and Koreans were the oldest group.Asian Indian and Japanese had the highest education attainment, and Vietnamese had the lowest education level.Asian Indian and Japanese had the highest income, and Chinese and Vietnamese had the lowest income (Table 1).

Logistic Regression in the Pooled Sample
Table 2 summarizes a logistic regression model, with poor physical SRH as the outcome in the pooled sample.Based on this model, age, education, income, employment, and marital status were associated with SRH (Table 2).

Logistic Regression across Ethnic Groups
Table 3 also shows the results of logistic regressions specific to each ethnic group.Ages higher than 50 were associated with worse physical SRH in Asian Indian, Chinese, and Korean Americans.Being married was associated with better physical SRH in Asian Indians, and worse SRH in Filipinos.Female gender was associated with better SRH among Japanese Americans.Education was associated with better SRH in Asian Indian, Chinese, Korean, and Vietnamese Americans.High income was associated with better SRH in Chinese, Filipino, and Vietnamese Americans.Employment was associated with better SRH in Filipino Americans (Table 3).

Discussion
This study explored ethnic variation in social determinants of physical SRH in Asian Americans.The results suggested that social determinants of physical SRH vary across ethnic groups of Asian Americans.That is, different ethnic groups are differently vulnerable to various social determinants of health.
Asian Indians, and then Japanese, had the highest SES (education and income), so ethnic groups of Asian Americans vary in their class.In line with SES, Asian Indians reported the highest level of physical SRH.The main effects of ethnicity on SRH were significantly above and beyond demographic and social determinants of health.This finding suggests that not all of the associations between ethnicity and SRH are due to social determinants.
In the pooled sample, education, employment, marital status, and income were associated with high SRH among Asian Americans.However, none of these associations were consistent across ethnic groups.Age is a risk factor for poorer SRH in the literature.In a recent study using the Collaborative Psychiatric Epidemiology Surveys (CPES) 2001-2003 data, high age was associated with poor mental SRH in Vietnamese, Filipino, and Chinese, but not other Asians [39].
For Asian Indians and Koreans, income was not associated with SRH.For other groups, income was at least marginally protective.Income was similarly protective for ethnic groups in the United States [40].Income may be a more salient determinant of health in U.S. than some other countries [41].Education and income have shown diminished returns for minorities in the United States [40,42].Minority groups with high education and income may even be at risk of worse mental health [43][44][45][46][47].
With Japanese being the only exception, gender did not independently correlate with physical SRH in any of the ethnic groups, when SES indicators were controlled.Female gender was also not associated with poor SRH in the pooled sample, again when SES was controlled.Although an association between gender and SRH is reported, ethnic groups differ in these effects [41,48].SRH may differently reflect health of men and women [49].Based on the sponge hypothesis [50], women are more aware of their physical symptoms [49], so their reports may be more accurate than men's reports.For men, poor SRH is more likely to represent life-threatening conditions than mild health problems.As a result, poor SRH better predicts the risk of mortality in men, rather than older female Americans [30].
The associations between ethnicity, gender, SES, and SRH are complex.In a study in Costa Rica, Uruguay, Argentina, Barbados, and Cuba, chronic disease explained gender disparities in subjective health.Such mediation was not found in the other countries [41].In another study, in Chinese and Cubans, but not other ethnic groups (i.e., Vietnamese, Filipino, other Asian, Puerto Rican, Mexican, other Hispanic, African American, and non-Latino Whites), female gender was a risk factor for poor physical and mental SRH [39].
The direction of the association between marital status and physical SRH was reversed for Asian Indians and Filipinos.Married individuals report better SRH due to causation or selection mechanisms [51].Married individuals have significantly lower mortality rates than unmarried persons [52].The health gain of marital status may be smaller for minorities than Whites [53].Future research should explore diminished health return of education, employment, marital status, and even income for Asian Americans in the United States.Similar diminished return of marital status [53] and other SES indicators [54,55] have been reported for other minority populations, such as African Americans.

Limitations
The study had a few limitations.First, the outcome was a single-item SRH.Validity of SRH may vary across ethnic groups of Asian Americans.A second limitation of the study was that only 47% of respondents (12% of all valid numbers dialed) agreed to take the survey.A third limitation was the possibility of measurement bias due to social desirability and self-serving bias, which could differ for populations by gender, age, and ethnicity.Finally, the current study did not collect data on medical conditions and comorbidities.

Conclusions
To conclude, demographic and social determinants of physical SRH vary across different ethnic groups of Asian Americans.Different ethnic groups may be differently vulnerable to various demographic and SES indicators on SRH.These ethnic differences may cause bias in cross-ethnic comparison of self-rated health.

Table 2 .
Factors associated with poor physical self-rated health (SRH) in the pooled sample.

Table 3 .
Factors associated with poor physical self-rated health (SRH) across ethnic groups.