Race, Ethnicity, and Language Data: Standardization for Health Care Quality Improvement

Chapter 4: Defining Language Need and Categories for Collection (cont.)

Defining Language Need and Categories for Collection (continued)

Estimates of Populations Needing Language Assistance and Applicable Requirements

This section examines national estimates of the numbers of people in the United States whose primary language at home is not English and the portion who is not proficient in English who therefore may need language assistance during health care encounters. It also reviews applicable national legislative and regulatory requirements that may guide the collection of language-related data.

Estimates of Populations Needing Language Services

Census questions provide a starting point for determining the language needs of individuals in different geographic areas through a comparable data set (Shin and Bruno, 2003). Since 1980, the Census has asked whether each person aged 5 years and older speaks a language other than English at home. This population doubled in absolute numbers from 1980 to 2000, and its percentage of the population over age 5 grew from 11 percent (23.1 million) in 1980 and 14 percent (31.8 million) in 1990 to 18 percent (47 million) in 2000 (Shin and Bruno, 2003). Respondents who speak a language other than English at home are also asked to enter the language they speak on an open-format response line and to rate their facility with spoken English (Figure 4-1). The same questions will be posed in Census 2010 and on the American Community Surveys. The Census asks no questions about reading or writing ability in English.

Assessment of Limited English-Speaking Ability

One simplified approach to assessing English-speaking ability is to ask people to rate themselves. The Census asks people to rate their ability to speak English on a scale from "very well" to "not at all" (Figure 4-1). These ratings are based on self-defined and -perceived ability and not any specific test. Of the 47 million people aged 5 and older who reported speaking a language other than English at home on Census 2000, 55 percent reported speaking English "very well," 22 percent "well," 16 percent "not well," and 7 percent "not at all" (Shin and Bruno, 2003). The proportion who spoke English very well was similar in 1980, when it was at 56 percent (Kominski, 1989).

The criteria chosen to define LEP significantly affect the size of the LEP population. If LEP is defined as those who speak English less than "very well," the Census 2000 LEP population numbers 21.3 million people over the age of 5 (more recent American Community Survey [ACS] LEP data estimate the total population at 23 million) (Youdelman, 2008). If it is defined as those who fall into the categories of "not well" and "not at all," the LEP population numbers 10.9 million. The Census employs another measure called "linguistic isolation," meaning that no one ages 14 or older in the household speaks English. This population of 11.9 million is similar in size to that resulting from the more constrained LEP definition (Shin and Bruno, 2003). LEP individuals may have someone in their family that they can call upon when they need help with interpretation, but those in linguistically isolated households must look elsewhere for language assistance.

Through schooling, children of immigrants eventually achieve a high degree of linguistic integration, and only a minority of immigrants' grandchildren retains bilingualism (Alba, 2005). A larger proportion of young people (aged 5-17) than of those who are older, who live in homes where a non-English language is spoken, speak English "very well" (U.S. Census Bureau, 2003b, 2003c). Even among first-generation immigrants to the United States, most children develop English-speaking ability; for example, 79 percent of Mexican and 88 percent of Chinese first-generation children speak English "well" or "very well," even while they continue to speak a language other than English at home (Alba, 2005). Thus, it is not surprising that children are often called upon to interpret for their parents and grandparents. As discussed above, however, the appropriateness of this arrangement for health care purposes has been questioned for several reasons, including the high frequency of errors with clinical consequences and the tendency to avoid sensitive and embarrassing subjects, such as those pertaining to sexual issues, domestic violence, abuse of drugs or alcohol, and the possibility of death (Flores, 2006a; McQuillan and Tse, 1995). Reflecting this concern, the California state assembly passed a bill in 2005 prohibiting the use of children under age 15 as medical interpreters; the bill was ultimately not enacted, however (EXODUS On-line, 2009).

Effect of Being Foreign Born

Being foreign born is not itself a marker for poor English skills: 39 percent of the 30.7 million foreign-born people aged 5 and over now living in the United States speak English "very well" and indeed may come from a country where English is spoken (e.g., Jamaica) (Grieco, 2003; Larsen, 2004; U.S. Census Bureau, 2003d). However, about three-fourths of the 21.3 million people identified in Census 2000, who are LEP by a definition of speaking English "less than very well," are foreign born; this accounts for 15.6 million people (U.S. Census Bureau, 2003d). More recent ACS data that estimate the LEP population at 23 million reveal that about 10.5 million are native born or naturalized citizens, and approximately 4 million more are documented immigrants (Youdelman, 2008). The proportion of the immigrant population that is proficient in English increases with time in the United States; for example, 36 percent of those in the country five years or less speak English very well, compared with more than 70 percent in the country for more than 30 years (Siegel et al., 2001).

Proficiency is lower among low-wage workers and those with less than a high school diploma-population groups that might be more likely to access public programs (Capps, 2003). High school graduation rates among the foreign-born populations from Europe and Asia now living in the United States are comparable to those among persons born in the United States—around 85-87 percent. However, the rate is much lower for immigrants from Central America, at 37.7 percent (Larsen, 2004). This is important because more than one-third of the U.S. foreign-born population comes from this region, particularly Mexico (Malone et al., 2003). Low literacy can compound the effect of a lack of English proficiency on understanding health-related information (Downey and Zun, 2007; Sudore et al., 2009; Zun et al., 2006).

Applicable Legal Requirements

Civil Rights Act Requirements to Identify the Service Population

Title VI of the Civil Rights Act of 1964 prohibits discrimination on the basis of race or national origin by those who receive federal funds:

No person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance." {42 U.S.C § 2000d}

Language needs have been considered a factor in deciding discrimination cases based on national origin under Title VI5 (Chen et al., 2007) and in determining whether there have been violations of equal access for language minorities under the Voting Rights Act.6 Lau v Nichols, 414 U.S. 563 (1974). Department of Justice. 42 U.S.C. Chapter 20 § 1973aa-1a. The Public Health and Welfare Act, Elective Franchise. Settlements have resulted in requirements to collect localized and granular data directly from those receiving services or indirectly through data descriptive of the service area (HHS, 2009c).

HHS' Office for Civil Rights (OCR) states that HHS is "committed to enhancing access to HHS services by LEP persons and closing the health care gap" (HHS, 2009b). Language assistance is to be made available at all points of contact with federally funded programs-enrollment, registration, and direct medical services. HHS describes LEP persons more broadly than the Census questions, which focus on spoken English. For HHS, LEP includes persons:

  • Who "are unable to communicate effectively in English because their primary language is not English and they have not developed fluency in the English language."
  • Who "may have difficulty speaking or reading English."
  • Who "will benefit from an interpreter who will translate to and from the person's primary language."
  • Who "may also need documents written in English translated into his or her primary language so the person can understand important documents related to health and human services" (HHS, 2009a).

Executive Order 13166, Improving Access to Services for Persons with Limited English Proficiency, requires each federal agency to review its services and develop and implement reasonable steps by which LEP persons can have "meaningful access" to programs or activities without charge for language services (Executive Office of the President, 2000). The guidance seeks to clarify the obligations of recipients of federal funds to provide language assistance services. Additionally, LEP persons are to be notified that free interpretation services are available so that they can make an informed choice about whether to use a friend or family member as an interpreter instead. HHS Title VI Civil Rights guidance allows patients to choose whether to use a language service. But interpreter services still must be provided if good medical practice might be compromised, the competence of the family interpreter is in question, or issues of confidentiality or conflicting interests arise. The emphasis is on voluntary compliance with these provisions.

The Department of Justice issued four Title VI "balancing factors" to be applied across all federal agency-funded programs: the number or proportion of LEP persons in the service population, the frequency of contacts, the importance of the services to the persons' lives, and the resources available to support services (U.S. Department of Justice, 2002). HHS subsequently revised its guidance accordingly (HHS, 2009b). Yet lack of knowledge of the requirements by both providers and patients or of willingness of LEP patients to pursue complaints when faced with language barriers leaves many persons without meaningful access to health care, and few states have comprehensive laws mirroring the federal requirements (Chen et al., 2007; Perkins and Youdelman, 2008).

Requirements of the Americans with Disabilities Act

Communication needs extend beyond spoken language capability to include barriers imposed by disabilities affecting hearing, speech, and vision. The Americans with Disabilities Act (ADA) of 1990 and Section 504 of the Rehabilitation Act of 1973 address nondiscrimination on the basis of such disabilities. Resolution of legal cases has resulted in requiring the availability of qualified sign language interpreters within a certain time frame (e.g., 2 hours) and the use of other auxiliary aids, such as TTY or TDD,7 in venues such as hospitals (HHS, 2009c; U.S. Department of Justice, 2003). Further examples of the types of auxiliary aids or services that might be required to ensure accommodation of a person with a disability are outlined in regulations.8

There are an estimated 1 million functionally deaf persons in the United States (Mitchell, 2005), and up to 36 million people have some degree of hearing loss (National Institute on Deafness and Other Communication Disorders, 2009). Only rough estimates—of 360,000 to 517,000 persons—exist of the number of deaf individuals who use sign language (Mitchell, 2005). Of note, immigrants who are deaf may have learned a different sign language from that taught in the United States (Gordon, 2005).

State Laws

States have instituted a number of additional laws to address language access. These are not reviewed in detail in this report. However, the status of laws nationally was recently reviewed by Perkins and Youdelman (Perkins and Youdelman, 2008), and Au and colleagues focused on activities in three states—California, Massachusetts, and New York (Au et al., 2009). These laws address the provision of direct language assistance, the setting of thresholds for applicable languages, continuing medical education requirements for physicians, the availability of interpreters for specific services (e.g., admissions to mental health facilities), facility licensure, and certification of interpreters.

Approaches to Eliciting Language Needs

The subcommittee considered different approaches to questions to elicit language needs. Assessment of English-language ability is widely used in studies evaluating the effects of language proficiency on disparities in the quality of health and health care (Jacobs et al., 2001). Table 4-1 lists approaches to questioning about patients' language needs that are employed by some health care entities. Questions address the individual's English proficiency, primary or preferred spoken language, language spoken at home, and preferred written language.

English Proficiency

An advantage of using a question to assess English proficiency, such as that used on the Census (Figure 4-1), is the ability to determine quickly whether a patient is likely to have language barriers that will limit his/her ability to navigate the health care system and communicate effectively with health care providers. Proficiency level data can be obtained for the entire population or matched to different languages (for example, among persons who speak a language other than English at home, 66 percent of Vietnamese speak English less than very well, compared with 23 percent of Hindi-speaking Asian Indians) (Kagawa-Singer, 2009; U.S. Census Bureau, 2003e). When an entity is considering which languages to list on its data collection instruments, knowing not just how many people speak a language but also their level of English proficiency and thereby their need for services will be helpful.

Since the response is based on self-report, it is important to understand the question's reliability in determining proficiency. The Census Bureau does not define which level of ability represents LEP (Griffin and Shin, 2007). However, the Census Bureau field tested the question to assess the validity of responses. Respondents who indicated that they spoke English "less than very well" had difficulty with the tests administered in the English Language Proficiency Survey (ELPS), and researchers found a strong correlation between self-assessment of speaking ability and understanding of tested concepts. The ELPS is a test of English-understanding ability and was administered in people's homes by the Census Bureau for the Department of Education. Those who rated their English-speaking proficiency as "very well" scored similarly on the test to those who spoke English as their first language, lending validity to the self-assessed ratings. Further analyses found that those who answered "not at all" and "not well" represented a distinct population that would definitely need English assistance because they rarely, if ever, spoke English and had limited reading skills as well (Kominski, 1989). Additionally, when setting threshold languages under the Voting Right Act, it was determined that people who spoke English less than very well were LEP (Kominski, 1985). Persons who fall into the category of speaking English "well" are assimilated to varying degrees but still speak English less frequently than those who rate their ability as "very well" (Kominski, 1989). The Census Bureau has done no recent analyses on the association between the LEP question and English-language abilities (Griffin and Shin, 2007).

One could argue that a person may have to have greater proficiency in English for health care encounters than for other daily tasks because of the unfamiliarity of health concepts and the complexity of medical terminology; such situational factors can affect people's assessment of their capability (Siegel et al., 2001). The association between the Census English proficiency question and accurate and effective communication in English in the health care setting remains undetermined. However, a recent article by Karliner and colleagues (2008) evaluated the accuracy of the Census English proficiency question in predicting the ability of 302 patients from a cardiology clinic to communicate effectively in English (Figure 4-2) (Karliner et al., 2008). The authors reported that in evaluating the sensitivity and specificity of four different questions in predicting outcomes of patient-reported ability to discuss symptoms and to understand physician recommendations in English, "the Census-LEP item using the high-threshold of less than 'very well' was the most sensitive for predicting both of the effective communication outcomes". Because the Census LEP question also had the lowest specificity, the authors recommend using a combination of that question and preferred language for medical care as a way to increase specificity with a marginal decrease in sensitivity. Different language groups may over- or underreport their competence; for example, Asians tend to underreport and Hispanics to overreport (McArthur, 1991; Zun et al., 2006). Therefore, health care entities may need to be mindful of their own population's response patterns.

Primary or Preferred Spoken Language

OCR has used the term "primary language" to mean the language that an LEP individual identifies as the one that he or she uses to communicate effectively and would prefer to use to communicate with service providers (HHS, 2008). The American Recovery and Reinvestment Act of 2009 (ARRA) similarly directs the inclusion of primary language in electronic health records.9 American Recovery and Reinvestment Act of 2009, Public Law 115-5 § 3002(b)(2)(B)(vii), 111th Cong., 1st Sess. (February 17, 2009). The NQF cultural competency framework uses the following definition:

Primary written and spoken language-the self-selected language the patient wishes to use to communicate with his or her health care provider. ( NQF, 2009)

Alternative phrasings of questions can elicit the name of a specific language (go to examples in Table 4-1). The Health Research & Education Trust (HRET) Toolkit suggests, "What language do you feel most comfortable speaking with your doctor or nurse?" California regulations suggest, "What is your preferred spoken language?" A Toolkit for Physicians developed for the California Academy of Family Physicians endorses a similarly phrased question as best practice: "In what language do you (or the person for whom you are making the appointment) prefer to receive your health care?" (Roat, 2005). It goes on to say, "Asking the question this way will provide you information on the language the patient feels he or she needs to speak in a health-related conversation. If the answer is a language other than English, you can plan to have language assistance available for the patient, and you can add this information to the record" (Roat, 2005, p. 5).

A concern with using a preference question alone is that it may not always capture a person's language need. For example, respondents may answer English if they believe that not doing so might limit their access to good medical providers. Similarly, respondents may state a preference for English because they know their providers are not fluent in their primary language. These examples are based on anecdotal report, and there are no research findings with which to assess the frequency of such occurrences. In practice, it is assumed that most people respond with their primary language so they can access the services of an interpreter or language-concordant provider.

The HRET Toolkit, endorsed by NQF, asks both the Census LEP question and a preference question. The subcommittee believes language need for effective communication with health care providers is defined by these two questions, and encompasses those with English proficiency of less than "very well." The subcommittee also believes the LEP question should be used to screen patients before they are asked about preference.

Language Spoken at Home

The Census asks whether a person speaks a language other than English at home and then asks what that language is (Figure 4-1). Detailed and comparable response data are available for states and localities on the languages spoken at home, so a health care entity can easily track what percentage of the population in its practice area reports using a language other than English in the home environment. Other data collectors, including National Health and Nutrition Examination (NHANES), ask about both language spoken at home and English proficiency. Even when people speak English well, the language spoken at home is generally an indicator of one's cultural background, and that cultural knowledge may provide a window into beliefs about health care.

However, there are disadvantages in using solely a language spoken at home to evaluate individual needs and to plan for language assistance capacity. More than half of people who state they speak a language other than English at home also report speaking English very well (Glimpse, 2009; Shin and Bruno, 2003). This suggests that using only this question in the assessment of language capacity could result in overestimating the need for language assistance; this was a problem encountered in earlier national Censuses that helped lead to adding the question on language proficiency (Kominski, 1989). Also of concern is that this question does not allow respondents to indicate language dominance when they are bilingual/multilingual.

Preferred Written Language

The approach to asking about written language has been to ask people their preference or some variation thereof. For example, "In which language would you feel most comfortable reading medical or health care instructions?" (HRET Toolkit see Hasnain-Wynia, 2007) or "What language should we write to you in?" (California Healthy Families, 2008a). The phrasing of a preferred-language question may need to be tailored to particular circumstances (Table 4-1). The phrasing of the first question would apply particularly within a health care delivery setting, while that of the latter might be sufficient for health plan communications, such as for enrollment or benefits information.

There is some evidence that the response to a written-language question will be the same as the response to a spoken-language question. To determine whether English-language proficiency in speaking varies significantly from that in writing and reading, the subcommittee conducted analyses using data on English-language proficiency for reading, speaking, and writing from the National Latino and Asian American Study (NLAAS) (Alegría et al., 2004a, 2004b). The NLAAS is a nationally representative household survey of Latinos and Asians aged 18 and older residing in the coterminous United States, where these data were collected. The findings show high-weighted Pearson correlation coefficients for English-language proficiency among speaking, reading, and writing ability. For example, for the full sample (both Asians and Latinos), the correlation between speaking and reading was 0.93, between speaking and writing was 0.90, and between reading and writing was 0.94 (Table 4-2). These results appear to indicate that English-language speaking proficiency can be extrapolated to English-language proficiency in reading and writing.

The Census Bureau does not routinely ask a question about a person's facility with written language. But two studies assessed how well people's ability to read a newspaper or fill out a form (e.g., driver's license, job application) in English conformed to their reported speaking ability (Table 4-3) (Kominski, 1989; Siegel et al., 2001). Those who answered with the two lowest ratings clearly had diminished capability for reading, but the results were equivocal for the "well" category. Another study testing language ability and comprehension in an emergency room setting found that a person's ranking on verbal and written competence was similar (Downey and Zun, 2007).

Because of the overlap between speaking, reading, and writing ability, an additional question about written language may not be essential when an entity needs to limit the number of questions asked. At the same time, a person who is relatively fluent in speaking English and answers "very well" on English proficiency may read English "less well" or "not at all." Knowledge of the education level of the population served can help illuminate the risk of lower or higher reading comprehension. One cannot assume language ability from ethnicity; for example, Contra Costa Health Plan found that less than 2 percent of Hispanic commercial members wanted written materials in Spanish.10

Reading many health-related materials with comprehension requires education at the high school level as most materials are written at a 10th grade reading level or higher (D'Alessandro et al., 2001; Downey and Zun, 2007; IOM, 2004), and even when low-literacy health-related materials are available at the fifth-grade level or below, medical terminology can be mystifying (Health Literacy Innovations, 2007; RTI International-University of North Carolina Evidence-based Practice Center, 2009). Further it is noted that about 40 million people in the United States read below the fifth-grade level, and this cannot always be attributed to a lack of spoken English proficiency. To ensure effective communication, patients may need to discuss written materials with an interpreter or bilingual provider even if the materials are translated into the patients' primary language.

Medical information can be quite complex to understand even without the added barrier of having a primary language other than English. Health literacy has been defined as:

The degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions. ( Ratzan and Parker, 2000, p. vi)

Half of LEP adults have a ninth-grade education or less (GCIR, 2008; Wrigley, 2003), making health-related materials less accessible to those who are less literate even in their native tongue. Twenty-two percent of non-English speakers indicate that they can read or write only in their own language, and 35 percent can be classified as functionally illiterate (IHA, 2009). Additionally, similar words can be confused. For example, someone who reads only Spanish might misread the English word "once" as meaning eleven times, creating the danger of taking a medication an inappropriate number of times (ISMP, 1997).

The subcommittee concludes that a patient's language preference for written materials is useful information, but if a health care entity must limit the number of questions it asks because of either administrative burden or Health IT capacity, asking about written language is a lower priority than asking about spoken language since written-language needs can generally be inferred from responses about spoken language. Additionally, the subcommittee believes more effective communication occurs when LEP patients have the opportunity to discuss translated documents with an interpreter or bilingual provider.

Assessment of Language Need

The subcommittee concludes that collection of data on language need is fundamental to improving service delivery to LEP populations and to conducting research aimed at identifying disparities in access and outcomes. The subcommittee explored various ways to determine patient spoken and written language needs so that steps can be taken to best enhance effective communication between patients and providers. Patients' proficiency with English and the language needed for effective communication should be taken into account to gauge their ability to understand their options for health services and to follow through on care plans and self-management. The subcommittee concludes that two questions define language need: one that determines whether English-language proficiency is less than "very well" and a second that determines the preferred language needed for a health-related encounter. The subcommittee sets a hierarchy among four possible types of language questions in widespread use and based on the previous discussion, recommends:

Recommendation 4-1: To assess patient/consumer language and communication needs, all entities collecting data from individuals for purposes related to health and health care should:

  • At a minimum, collect data on an individual's assessment of his/her level of English proficiency and on the preferred spoken language needed for effective communication with health care providers. For health care purposes, a rating of spoken English-language proficiency of less than very well is considered limited English proficiency.
  • Where possible and applicable, additionally collect data on the language spoken by the individual at home and the language in which he/she prefers to receive written materials.

When the individual is a child, the language need of the parent/guardian must be determined. Similarly, if an adult has a guardian/conservator, that person's language information must be assessed.

Return to Contents
Proceed to Next Section

Page last reviewed October 2014
Page originally created September 2012
Internet Citation: Chapter 4: Defining Language Need and Categories for Collection (cont.). Content last reviewed October 2014. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/research/findings/final-reports/iomracereport/reldata4a.html