Page 8 - IJPS-5-1

Language and self-assessed health in the U.S

Specifically, this study looks at whether an evaluation of someone’s health status, using a self-assessed health (SAH)
survey question as a health outcome measure, varies if the SAH question is asked in Spanish versus in English. The
topic of this study gains significance given earlier research that has exposed a “Hispanic paradox” in which Hispanic
immigrants experience better health relative to U.S.-born citizens of similar socioeconomic status (Markides and Coreil,
1986). Yet this paradox is increasingly subject to debate on multiple counts (Abraído-Lanza, Dohrenwend, Ng-Mak
et al., 1999; Franzini, Ribble, and Keddie, 2001; Palloni and Arias, 2004; Markides and Eschbach, 2005; Smith and
Bradshaw, 2006; Rubalcava, Teruel, Thomas et al., 2008). There is also evidence of poorer health among Hispanics born
in the U.S. or who have resided in the U.S. at length (Cho, Frisbie, Hummer et al., 2004; Vega, Rodriguez, and Gruskin,
2009). Further, research has indicated that the intersecting characteristics of being an immigrant, becoming racialized as
a U.S. minority, and having limited English proficiency have consequences for health (DuBard and Gizlice, 2008; Vega,
Rodriguez and Gruskin, 2009).
This line of inquiry into health disparities across categories of ethnic and citizenship status involves considerable debate
over how to best measure the health status of immigrants, particularly those who do not speak English or do not speak it well
(Finch, et al., 2002; DuBard and Gizlice, 2008; Lee and Schwartz, 2014; Sanchez and Vargas, 2016). A number of standard
health measures – such as mortality, service utilization, and birth weight – involve statistical and practical challenges among
migrating populations. These include the likelihood of multiple border crossings and return migration, unavailability of
birth and death records, inconsistent measurement of Hispanic populations, misclassification of Hispanic deaths, and
mistranslation of healthy surveys (Abraído-Lanza, Dohrenwend, Ng-Mak et al., 1999; Franzini, Ribble and Keddie, 2001;
Palloni and Arias, 2004; Markides and Eschbach, 2005; Smith and Bradshaw, 2006; Rubalcava, Teruel, Thomas et al., 2008).
SAH, based on a survey question that asks someone to simply rate their health overall on a simple scale from excellent
to poor, has become standard on health surveys (Chirinda, Saito, Gu and Zungu, 2018). This global snapshot survey item
may address some of the challenges involved in measuring migrant population health. SAH is commonly considered to
be an economical, simple, quick to administer, and easily translatable indicator that can be validly and reliably employed
across languages and cultures (Maddox and Douglass, 1973; Mossey and Shapiro, 1982; Kaplan and Camacho, 1983;
Idler and Benyamini, 1997). SAH has been found to accurately predict mortality, even when controlling for demographic,
socioeconomic, other health, and psychological factors (DeSalvo, Bloser, Reynolds, et al., 2005; Idler and Benyamini,
1997). Given that SAH is thought to be a valid global health measure, it may not be subject to the same population-level
measurement challenges as other indicators. Its advantages mean SAH could be a helpful gauge for studying inequalities
across heterogeneous populations that vary by primary language.
Nevertheless, the degree to which SAH can be used in comparative research is unclear. Studies have suggested that
SAH may be differentially perceived across cultures, ethnicities, and languages such that people with similar objective
health measures may subjectively assess their health differently (Angel and Guarnaccia, 1989; Jylhä, Guralnik, and
Ferrucci, 1998; Shetterly, Baxter, Mason et al., 1996; Lee and Schwartz, 2014). Attitudes, perceptions, and interpretations
of health vary across time and individual characteristics such as age, sex, and education (Ren and Amick, 1996; Zajacova
and Dowd, 2011). Thus, it may be difficult to make cross-country and cross-linguistic assessments of health disparities
using SAH (Jylhä, 2009; Zimmer, Natividad, and Lin et al., 2000).
A small number of recent studies has explored the likelihood that immigrants who take a health survey in Spanish
interpret the SAH scale differently, and these suggest reasons to be cautious. Shetterly, Baxter, Mason et al. (1996) analyzed
a small sample of 419 Hispanic and 583 non-Hispanic white respondents of the San Luis Valley Health and Aging Study
and found that when controlling for health conditions, Hispanics were 3.6 times more likely to report fair to poor health
than non-Hispanic white respondents. (Notably, however, their study collapsed the “fair” and “poor” category.) Bzostek,
Goldman, and Pebley (2007) examined data from approximately 3000 households from the Los Angeles Family and
Neighborhood Survey conducted between 2000 and 2002. Their results suggested that the Spanish interview language,
but not Spanish household language, is associated with a worse rating of SAH, suggesting that translational issues
influence comparisons of health status. Viruell-Fuentes, Morenoff, Williams et al. (2011) analyzed data from the Chicago
Community Adult Health Study, collected in 2001-2003, and the Behavior Risk Factor Surveillance System, collected
in 2003. Their results showed that adjusting for interview language reduces a gap in SAH between Latino and white
respondents. Finally, Sanchez and Vargas (2016) used data from the 2011 Latino Decisions survey to conduct a built-in
language experiment comparing multiple translations of SAH in a nationally representative dataset of 1200 Latinos. They
demonstrated that there is a problem with the response “fair” SAH when termed as regular in Spanish, which is how key
agencies, such as the U.S. Centers for Disease Control (CDC) and Prevention, normally translate the word.
Our study builds on this extant literature in several ways. First, earlier studies (such as Shetterly, Baxter, Mason et al.,
1996) commonly dichotomized the measure or considered SAH as an ordered variable. Collapsing categories forces

2 International Journal of Population Studies | 2019, Volume 5, Issue 1