Race, Ethnicity, and Language Data: Standardization for Health Care Quality Improvement
4. Defining Language Need and Categories for Collection
Data on a person's language and communication needs should be part of any minimum data set related to health care delivery and quality improvement. The subcommittee recommends identifying spoken language need by determining first how well an individual believes he/she speaks English and then what language he/she needs for a health-related encounter. The subcommittee defines limited English proficiency (LEP) in the health care context as speaking English less than very well. To simplify the collection of language data, most entities should develop a list of common languages used by their service population, accompanied by an open-ended response option for those whose language does not appear on the list. When an entity has the capacity to collect additional information, the language preferred for written materials and the language spoken at home are also valuable. Locally relevant lists of language categories should be derived from a national standard list, with coding to facilitate information flow across entities.
The collection of data on the language needs of patients is important to improving health and health care. Collection of these data is necessary to meet legal obligations based on federal funding aimed at ensuring equitable access to health services and preventing discrimination based on national origin or limited ability to speak English. More important in the present context, however, knowledge of which patients have limited English proficiency (LEP) and of what their language needs are allows medical services and related interventions (e.g., provision of language assistance services, outreach, educational activities, translation of documents) to be targeted with the aim of improving the quality of care and reducing disparities. Not all persons with LEP are foreign born; more than one in four people aged 5 and over with LEP are born in the United States, and many more are naturalized citizens or documented immigrants (U.S. Census Bureau, 2003d; Youdelman, 2008).
Evidence on variations in health outcomes, medical errors, and receipt of quality health care as a function of English-language ability is persuasive that disparities exist, as reviewed in this chapter. Lack of English proficiency is a barrier not just to effective communication with individual health care providers, but also to accessing care in the first place. A review of the evidence base in this area convinced the subcommittee that the collection of data on language and communication needs is essential to safe, accessible, and effective, quality health care.
The subcommittee reviews various approaches to the collection of language data for health care improvement purposes. These approaches include practices of the U.S. Census Bureau because its data can easily be accessed proficiency with the English language. Additional guidance from the experiences of physicians, hospitals, health plans, states, and advocacy groups informed the subcommittee's deliberations. Issues surrounding the collection of language data include understanding whether there is demand for language services in the health care sector (e.g., among hospitals, physicians) and across states, whether data should be collected for both spoken and written language needs, what languages would make up a national standard set of categories, and how those languages should be coded for sharing of data beyond a single service site. It should be noted that the subcommittee's definition of language is one that is inclusive of communication needs such as sign language.
This chapter begins by reviewing what is known about the role of language in the provision of quality health care and health outcomes. It then summarizes estimates of populations needing language assistance and applicable legislation and regulatory requirements. Next is a discussion of various approaches used to question patients about their language needs. Language categories to be used by health care entities to collect these data and possible code sets are then considered.
The Role of Language in the Health and Health Care
Interactions with Patients Needing Language Assistance
The number of people nationwide needing language assistance is growing rapidly (Shin and Bruno, 2003), and individuals with these needs interact with the health care system daily. The extent of this interaction is revealed by recent surveys on encounters with LEP patients in hospitals, physician offices, and community health centers:
- Eighty percent of hospitals provide services to LEP patients regularly, and 63 percent of hospitals encounter these patients daily or weekly (Hasnain-Wynia et al., 2006).
- Eighty-one percent of general internal medicine physicians commonly treat LEP patients (54 percent at least once a day or a few times a week; 27 percent a few times a month) (American College of Physicians, 2007).1
- Eighty-four percent of federally qualified health centers provide clinical services each day to LEP patients—45 percent see more than 10 LEP patients per day; 39 percent see from one to 10 per day (National Association of Community Health Centers, 2008).
One study of hospitals indicated that 80 percent of hospitals have a health information technology (Health IT) field dedicated to collection of language names, primarily to identify the languages needed for interpreter services (Regenstein and Sickler, 2006). Health care entities use a number of different approaches to collect this information: some limit the response categories to English, Spanish, and an "other language" category, while others offer respondents 200-300 languages from which to choose (Regenstein and Sickler, 2006; Tang, 2009). A study of the practice of internal medicine physicians found that only 28 percent kept detailed records of primary language needs, and about two-thirds of those who did record this information did so on paper rather than in a data system (American College of Physicians, 2007). By contrast, Kaiser Permanente, a health plan and a service provider covering eight states, began collecting data in 2009 in its electronic health record (EHR) system using a list of 131 spoken languages and 119 written languages (Tang, 2009).
Health care entities may serve LEP patients by using bilingual health care providers; other bilingual staff trained in medical terminology; or, frequently, ad hoc interpreters, such as family members or bilingual staff with no knowledge of medical terminology. The latter approach is particularly prone to error (Flores, 2005, 2006b). Telephone interpretation services are also available from numerous sources; more than 200 different languages are offered by some interpretation and translation services (ASIST Translation Services, 2009; Language Line Services, 2009). Depending on the diversity of the population served, an entity may encounter persons with language assistance needs in just a few or many languages. New York Presbyterian Hospital, for example, reports providing interpretation in 95 languages (NQF, 2009). Reimbursement for the provision of interpretation and translation services is not always available (e.g., not under Medicare, or when states elect not to provide a match for Medicaid funding for such services) (Bagchi and Stevens, 2006; Chen et al., 2007; Ku and Flores, 2005; Minnesota Department of Health Office of Rural Health Primary Care, 2008; Ponce et al., 2006b; Youdelman, 2007). Reauthorization of the Children's Health Insurance Program (CHIP) in 2009 increased federal matching for language services from 50 to 75 percent.2 An analysis of the adequacy of different means of providing language services and the funding of such services is beyond the scope of this report, but the issues have been examined by others (Gany and Ngo-Metzger, 2007; Karliner et al., 2007; Saha and Fernandez, 2007).
Too often, either ad hoc or no interpretation services are available when LEP individuals seek health care services. There are no good estimates of how many LEP patients who need interpretation services fail to receive them, but a figure of nearly 50 percent was found in one emergency room study (Baker et al., 1996a).
Uses of Language Data
Entities that collect language data may use the data in various ways. The most obvious ways are to provide direct language assistance during a clinical encounter and information for follow-up care, such as chronic disease management education or discharge instructions in a patient's language. Categorical data on demand for language assistance can inform hiring of bilingual staff or arrangements for interpretation services. An entity also might want to make appointments for patients with providers who are language concordant (Box 4-1). A hospital might want to track whether patients who receive language assistance have better outcomes on quality metrics compared with those who do not receive those services, or it might want to track whether those services are timely (Box 4-2). Language data also are useful for determining the need for translated materials; for example, Kaiser Permanente translated its Health Care Glossary into six languages to communicate commonly used terms and explanatory information about tests or conditions more effectively (NCQA, 2007). The California Healthy Families program indicates in which languages health plans have written materials on coverage, medical care reminders, member handbooks, and newsletters (California Healthy Families, 2008b).
Additionally, the profile of patients being served can be compared with the population statistics of the service area to identify populations not being served. Yet while evidence shows that some health care entities collect language data on their patients, most entities fail to use these data to assess how language barriers impact the quality of care and ultimately patients' health status (Regenstein and Sickler, 2006). It should be noted that recommendations on which specific quality improvement actions should be undertaken by entities is beyond the subcommittee's charge.
Effect of Language on Health Status, Access to Care, Health Outcomes, and Patient Safety
It is well established that LEP patients encounter significant disparities in access to health care (Hu and Covell, 1986; Weinick and Krauss, 2000), decreased likelihood of having a usual source of care (Kirkman-Liff and Mondragon, 1991; Weinick and Krauss, 2000), increased probability of receiving unnecessary diagnostic tests (Hampers et al., 1999), and more serious adverse outcomes from medical errors (Divi et al., 2007) and drug complications (Gandhi et al., 2000).3 The evidence also suggests that language barriers can increase the likelihood that patients will miss appointments, make less use of preventive care (Brach et al., 2005; Ku and Flores, 2005), or feel dissatisfied with health care services (Carrasquillo et al., 1999; Weech-Maldonado et al., 2003). On the basis of the findings detailed below, the subcommittee concludes that assessing language needs for each individual is an essential first step toward ensuring effective health care communication, and that provision of language assistance services is an actionable quality improvement option.
Effect of Language on Health Status
A growing literature documents a link between language barriers and poor quality health care (Pippins et al., 2007; Woloshin et al., 1995) that can lead to lower health status (DuBard and Gizlice, 2008). Research also indicates that this link can be broken by the use of interpreters. For example, use of interpreters is associated with improvements in the rate of follow-up visits after a visit to the emergency department (Karliner et al., 2007), in prescriptions written and filled (Flores et al., 2005), and in the need for obstetrical interventions.
Many studies that do not directly evaluate how language barriers impact health status examine how language incompatibility or LEP leads to different medical management than that received by patients who do not have these limitations or are provided with interpreters (Bard et al., 2004; Bernstein et al., 2004; Sarver and Baker, 2000; Waxman and Levitt, 2000). For example, LEP patients who needed but did not receive interpreter services experienced less satisfaction with their health care interactions, including less friendliness, time spent, and perceived concern for the patient, than those provided with or not needing interpreters (Baker, 1998). Recent studies have also revealed how language barriers can result in delays in prehospital care (Grow et al., 2008), less social interaction between dental staff and LEP patients (Hammersmith and Lee, 2008), and more negative clinical experiences in health care settings (Hampers et al., 1999) relative to non-LEP patients. One survey of care provided mainly in safety net hospitals found that the experiences of uninsured patients with access to an interpreter were comparable to or better than those of insured patients with no need for an interpreter (Andrulis et al., 2002). Moreover, families of non-English-speaking patients receive less information relevant to high-quality end-of life-care (Thornton et al., 2009). Such evidence suggests that language is a central factor in being able to achieve optimal health status and that bridging language gaps is essential to ensure quality care.
Language barriers prevent providers from obtaining accurate patient histories, impair the ability to engage patients in joint patient-provider decision-making on treatment, and limit patients' ability to obtain sufficient information for self-care (Wisnivesky et al., 2009). Poor patient-provider communication has been linked, for example, to poor asthma management practices in children (Chan et al., 2005) and in adults (Wisnivesky et al., 2009). It has also been associated with poor adherence to medication regimens (David and Rhee, 1998; Derose and Baker, 2000; Orrell et al., 2003) and concerns about unequal power dynamics between patients and providers (Schlemmer and Mash, 2006).
Effect of Language on Access to Care
Language barriers are closely linked to limitations in access to care (Wu et al., 2004) and to underuse of primary and preventive services (Woloshin et al., 1997), such as preventive cancer screenings (Jacobs et al., 2005; Ponce et al., 2006a), immunizations (De Alba and Sweningson, 2006; Sun et al., 1998), and routine check-ups (Pearson et al., 2008). In their examination of language-concordant and language-discordant patient-provider interactions in emergency services, Sarver and Barker (2000) discovered that the latter patients were less likely to receive follow-up appointments.
LEP is also associated with lower rates of prescription medication use, ambulatory visits, dental visits, and appropriate mental health treatment (Brach et al., 2005; Derose and Baker, 2000; DuBard and Gizlice, 2008; Sentell et al., 2007). This lower use of services may be associated with feelings of being discriminated against, as well as lower levels of trust and less confidence in medical visits, for those with language barriers, noted in particular among older Latinos (Mutchler et al., 2007). Language barriers have also been closely related to patient-perceived quality of care; for example, 81 percent of English-speaking patients with colorectal cancer reported receiving high-quality care, compared with only 52 percent of non-English-speaking patients (Ayanian et al., 2005). The lower use of services by non-English speaking persons may also reflect the patient's inability to articulate medical or mental health concerns to health care providers that are less easily measured by objective laboratory tests so that appropriate diagnoses can be made (Sentell et al., 2007).
The inability to communicate with insurance personnel can also create difficulties in enrolling in a health plan (Feinberg et al., 2002), as well as in securing a usual source of care (Kirkman-Liff and Mondragon, 1991). One study, for example, found reduced enrollment of Medicaid-eligible children in publicly funded health insurance programs because of parental difficulties in understanding enrollment forms (Feinberg et al., 2002). Similarly, a study of LEP Medicare beneficiaries demonstrated poorer access to a usual source of care as compared with those who were not LEP (Ponce et al., 2006a).
Research suggests that Spanish-speaking patients, as well as Spanish-speaking parents of pediatric patients, experience worse communication with their provider as compared with their English-speaking counterparts and less overall satisfaction with care (Jacobs et al., 2006). In one study, 89 percent of LEP Latinos who reported having a usual source of care cited the presence of interpreters or bilingual providers (Brach and Chevarley, 2008), implying that language capacity may be required to provide continuity of care for LEP patients. In addition, LEP patients who are seen by language-concordant providers demonstrate decreased likelihood of omitting medications and visiting the emergency department relative to those seen by language-discordant providers (Carter-Pokras et al., 2004; Manson, 1988). Tocher and Larson found no differences in meeting quality-of-care guidelines for LEP patients with diabetes (1998) or in the amount of time physicians spent with primary care patients (1999) in a setting with certified interpretation services available.
Effect of Language on Health Outcomes
Research has documented that poor health outcomes are more likely when language and cultural barriers exist between patients and providers (Anderson et al., 2003). Communication breakdowns occur when patients and providers are language discordant (Baker et al., 1998; Karliner et al., 2004). Results from several studies (Marcos et al., 1973; Price and Cuellar, 1981) suggest that LEP patients provide more elaborate replies with greater disclosure when interviewed in their primary language and that conducting the assessment in the patient's primary language may be particularly relevant for accurate diagnosis. The absence of language concordance between patient and provider and consequent reliance on ad hoc interpreters may impede disclosure of sensitive information (Marcos, 1979). It also negatively impacts comprehension of instructions and other treatment information necessary for adherence to and continuity of treatment (Wilson et al., 2005). Flores and colleagues (2005), for example, found that LEP patients who need but lack access to an interpreter have a poorer understanding of their medical diagnosis and treatment (Flores et al., 2005).
Language also appears to impact health outcomes by influencing the quality of the patient-provider relationship, including the development of trust, adherence to treatment, and follow-up (Rivadeneyra et al., 2000). LEP patients are more likely than those with good English-language proficiency to report inaccurate diagnoses, inadequate treatments, or negative health outcomes (Phelan and Parkman, 1995). Thus, if language barriers exist, diagnostic assessments, symptom disclosure, confidentiality, and treatment adequacy may be compromised (Baker et al., 1996b; Carrasquillo et al., 1999; Perez-Stable et al., 1997) and health outcomes suffer as a result.
Effect of Language on Patient Safety
Systematic literature reviews find that use of ad hoc interpreters is related to higher rates of communication errors and increased likelihood of clinical errors (Flores et al., 2005; Karliner et al., 2007). Family members and friends who act as ad hoc interpreters and do not understand the medical terminology involved or lack sufficient fluency in both languages are likely to interpret with errors (Flores et al., 2003). Typical errors include omissions, additions, condensations or abbreviations, substitutions, editorialization (interpreter adds or substitutes words that change the message), and false fluency (use of inaccurate words or phrases) (Flores et al., 2003). Flores and colleagues (2003) found that Spanish-speaking patients in an outpatient pediatric clinic experienced an average of 31 errors in medical interpretation by ad hoc interpreters and providers without sufficient language fluency, and more than half of these patients could have experienced negative adverse effects as a consequence of those errors. (It should be noted that, although research has documented a variety of interpretation errors during assessments, the clinical significance of such errors has not been well characterized.) Elderkin-Thompson and colleagues also found interpretation errors in more than 50 percent of videotaped encounters with nine Spanish-speaking nurses untrained in medical topics they were regularly called upon to interpret for LEP patients (Elderkin-Thompson et al., 2001).
Linguistic discordance can encompass differences in the concepts behind words and in the contexts giving meaning to those words (Flores et al., 2005). As a result, some researchers recommend that providers partner with trained interpreters who can bridge not only linguistic gaps but also cultural gaps that may challenge patient-provider communication (Dohan and Levintova, 2007).
Also of concern is that LEP patients without interpreters (compared with English-speaking patients and LEP patients with professional interpreters) receive fewer tests and procedures, which could lead to an increased risk for problems in the emergency department (Bernstein et al., 2004). Likewise, the lack of English language proficiency among the parents of pediatric patients has been correlated with a doubling of the risk of adverse medical events during pediatric hospitalizations (Cohen et al., 2005). In addition, LEP patients evidenced increased risk of misunderstanding prescription labels when seeing language-discordant providers compared with English-fluent patients (Wilson et al., 2005). Similarly, in one study, 27 percent of patients who needed but failed to receive interpreter services did not understand their medication instructions, compared with 2 percent who received such services (Andrulis et al., 2002).4