Development and initial validation of a dual‐language english ...

1153

BRIEF REPORT

DEVELOPMENT AND INITIAL VALIDATION OF A DUAL-LANGUAGE ENGLISH-SPANISH FORMAT FOR THE ARTHRITIS IMPACT MEASUREMENT SCALES WILLIAM D. HENDRICSON, I. JON RUSSELL, THOMAS J. PRIHODA, JAMES M. JACOBSON, ALICE ROGAN, GEORGE D. BISHOP, and RICHARD CASTILLO

Language, cultural, and educational barriers complicate efforts to validate health status questionnaires that have been translated into Spanish. To overcome these problems, a prototype dual-language format was developed for the Arthritis Impact Measurement Scales. Validity testing with 72 patients diagnosed as having rheumatoid arthritis indicated high levels of test-retest reliability, item-to-scale internal consistency, and construct validity for both Anglo and Hispanic subjects. A technique for developing and pilot-testing a questionnaire written in a regional Spanish dialect is described. Linguistic considerations, questionnaire deFrom the Division of Instructional Development and the Departments of Medicine, Pathology, and Psychiatry, The University of Texas Health Science Center at San Antonio: the Endocrinology Service, Wilford Hall Medical Center, Lackland Air Force Base, Texas; and the Division of Cultural and Behavioral Sciences, The University of Texas at San Antonio. Supported by NIH Multipurpose Arthritis Center grant AM-35605. William D. Hendricson. MS: lnstructional Development Specialist, Division of Instructional Development, The University ofTexas Health Science Center at San Antonio: 1. Jon Russell. MD. PhD: Associate Professor, Department of Medicine. The University of Texas Health Science Center at San Antonio: Thomas J . Prihoda, PhD: Assistant Professor, Department of Pathology, The University of Texas Health Science Center at San Antonio: James M. Jacobson, MD: Endocrinology Service. Wilford Hall Medical Center, Lackland AFB. Texas; Alice Rogan. PhD: Department of Psychiatry, The University of Texas Health Science Center at San Antonio: George D. Bishop, PhD: Associate Professor. Division of Cultural and Behavioral Sciences, The University of Texas at San Antonio; Richard Castillo, BS: Senior Research Assistant, Department of Pathology, The University of Texas Health Science Center at San Antonio. Address reprint requests to William D. Hendricson. MS. Division of Instructional Development, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78284. Submitted for publication April 28, 1988: accepted in revised form March 13, 1989. Arthritis and Rheumatism, Vol. 32, No. Y (September 1YSY)

sign, and other applications are discussed in light of the results obtained.

Census Bureau statistics indicate that the number of Hispanics in the United States has increased by 34% since 1980, from 15 million to 20 million-a rate of growth that is 5 times faster than that of the general population. As a result, sociologists, economists, educators, and the medical community have focused considerable attention upon various aspects of Hispanic culture, including cross-cultural perspectives on health care issues. In this report, we review methodologic issues that confront health professionals who are using questionnaires to study health care issues among subjects with limited literacy in English, present a new approach for designing a Spanishlanguage version of the Arthritis Impact Measurement Scales (AIMS) ( I ) , describe the initial validation of this instrument, and offer guidelines, based on our experiences, for the translation of health status questionnaires from English to Spanish. A patient’s impressions of what constitutes disease or illness may depend on his or her age, schooling, cultural traditions, language, financial status. and peer group mores (2). Meaningful comparisons between ethnic groups cannot be made until cultural variables are examined using reliable and valid instruments. A number of the health status questionnaires, including the AIMS, have been translated from English into Spanish. These Spanish-language instruments have been used to evaluate how language differences between medical care providers and Hispanic patients affect delivery of services (3), to assess the influence of medical “folk beliefs” (4), and to investigate the ways that cultural values influence the pa-

1154

tient’s sense of personal responsibility for health maintenance (5). However, there have been difficulties with translations of health status questionnaires. In some cases, the problem appeared to result from an inadequate translation process (6,7). Yet, in other studies where a rigorous translation and back-translation process was followed, reliability problems were still reported for Spanish-speaking subjects (8,9). Several theories to explain these difficulties have been proposed, including low English literacy among Hispanics, cultural values that influence perceptions of impairment, language diversity among Hispanic subgroups (there are numerous regional dialects or variawritten nature of these tions), and the oral dialects (10). There are a number of regional Hispanic subcultures in North America, and even though they share a common heritage, - they exhibit different traditions, Values, and language variations. Although there is an undeniable Spanish language foundation, regional variations or dialects of Spanish have their own linguistic flavor. The Spanish heard in South Florida is influenced by the Cuban and Caribbean background of many Hispanics in that area. It is different from the west coast “Chicano,” the so-called “Tex-Mex” found in South and West Texas, or the Spanish spoken in northern cities by individuals of Puerto Rican heritage. The regional Spanish dialects are technically defined by linguists as “second languages.” In years past, these languages were used by immigrants for communication among themselves until members of the group became fluent in the “target” or primary language of their adopted country (11). Like other second languages, the Spanish dialectic variation often heard in South Texas uses a nonstandard and simplified grammatical structure and relies on a continuous infusion of “loan words” from the target language of that geographic area. Because the South Texas Spanish dialect is commonly spoken in the home (12), many Hispanic children learn to speak it before they are exposed to English. Some individuals within the Hispanic community also view regional Spanish dialects as symbols of cultural identity. These factors have contributed to the persistence of these dialects, which may in turn explain the subjects’ low level of literacy in both Spanish and English. Many language specialists now discourage the use of formal Spanish vocabulary and grammar in questionnaires because they are poorly understood. Instead, they recommend customizing translations to the dialect most commonly spoken in the survey area

BRIEF REPORTS

(13). Linguists who favor this approach believe that translators should strive for conceptual rather than literal equivalence between the source language (in this case, English) and the target language (Spanish) (14). Although a number of investigators have reported on translation and validation of other health status instruments from English to Spanish (7,9,13,15,16),we could not locate previous reports of such work with the AIMS. PATIENTS AND METHODS

Assessment of patient health status via the AIMS is one of the outcome measures being used in an office-based continuing education program that is being conducted by the South Texas Multipurpose Arthritis Center. This program iS aimed at family prxtitiOnerS and internists in Small, rural COmmUnitieS in South and West Texas, where the population is primarily Hispanic. To perform this assessment, there was a need for a reliable and valid version of the AIMS that could be administered by mail to the rheumatoid arthritis patients of participating physicians. Study instrument. The AIMS has been extensively tested in its English-language version and has been shown to have good reliability and validity (17). It contains 66 health status questions. The first 45 questions are grouped into 9 scales: Mobility, Physical Activity, Dexterity, Household Activities, Activities of Daily Living, Anxiety, Depression, Social Activity, and Pain. There are 4-7 questions per scale. Responses to the items within each scale are summed to produce an individual score for each component scale and then a total score. The first 5 component scales are classified as “physical function,” while the Anxiety and Depression scales compose “psychosocial function”; overall scores are calculated for these 2 functions as well. Translation process. It was anticipated that the language skills of the subjects would vary considerably in both English and Spanish. The decision was therefore made to present corresponding English and Spanish versions of all material in the AIMS, to enhance comprehension. This format is depicted in Table 1 . Its development involved 5 steps, which are outlined below. Development of Spanish translations. After reviewing 3 Spanish-language translations of the AIMS that were obtained from other institutions on the east coast and west coast, it became apparent that a local South Texas version of the AIMS was needed. To

BRIEF REPORTS

1155

Table 1. Format of dual-language English-Spanish Arthritis Impact Measurement Scales questionnaire (circle one number for each question) (circule un numero par cudu pregunrci) 31.

During the past month, how often have you had severe pain from your arthritis? En el mes pasado, con qrre ,fiecrtenciu trri’o dolor-fiierte ccrirsudo por la artritis? Always (Siempre) Very often (Cusi siernpre) Fairly often (Bastante seguido) Sometimes (Algunu I J E C F S ) Almost never (Casi nuncu) Never (Nunca)

I 2 3 4 5 6

32. During the past month, how long has your morning stiffness usually lasted from the time you wake u p ? En el mes pasado, cirLinio riernpo le duro lu rigidez ul despertar? Over four hours (MUSde ciiutro lioras) 1 Two to four hours (Dos a cuatro liorus) 2 One to two hours (Una a dos liorus) 3 Thirty minutes to an hour (Trientu minrrios ci una hor-u) 4 Less than thirty minutes (Memos de trienta minutos) 5 Do not have morning stiffness ( N o tengo rixidez ul despertur-me) 6 33.

During the past month, how often have you had pain in two or more joints at the same time? el mes pasado, cuuntus I~CCOSt i i v o dolor en icnu o mas coyunturus (hiresos) a1 rnismo tiempo? Always (Siempre) I Very often (Casi siempre) 2 Fairly often (Bastante segrrido) 3 Sometimes (AIgunus rwt.s) 4 Almost never (Casi niinc.o) 5 Never INimcci) 6

Elz

34. During the past month, how much of the time have you enjoyed the things you do? En el rnes pasado, ciiantus i w e s sintio sciti.s.jiicc~io/ien l t i s c o s u qite hizo? All of the time (Todo el tiempo) Most of the time (Casi todo el riempo) A good bit of the time (La muyoriu del tiempo) Some of the time (Parre del fiempo) A little of the time (Poco tiempo) None of the time (Nrrnca)

accomplish this, 2 teams of experienced translators ( 3 per team) worked independently to produce Spanishlanguage versions of the questionnaire. One of the teams produced a “literal” translation involving a direct, word-for-word English-to-Spanish conversion. Equivalency in grammar and syntax was maintained. Back-translations were produced by other bilingual individuals to verify accuracy. The second team produced a conceptual or “idiomatic” translation. (The term “idiom” refers to the unique style of language used by a people in a particular community or region.) This team did not try to achieve “pure” wordfor-word equivalence, but instead tried to capture the essential meaning of each item in the questionnaire. The Spanish-language translation was then written in the most concise form, using vocabulary commonly used in South Texas. Another individual versed in the

1 2 3

4 5 6

South Texas dialect then back-translated the questionnaire. Establishment o f f a c e validity. To establish face validity, a 15-member review panel was set up to critique the 5 Spanish-language instruments (the 3 instruments previously translated elsewhere plus the 2 instruments developed specifically for this project). By allowing the review committee to compare and contrast instruments, it was hoped that the translation with the “best fit” for South Texas would be identified. Criteria for panel membership included Hispanic ancestry, lifetime residence in South Texas, birth and school attendance in a rural community within the study area, no Spanish courses in school, currently speaking Spanish at home with family members, and the ability to read aloud and interpret a paragraph written in Spanish that includes characteristic idioms

1156

and syntax used in this region. The reviewers completed each of the Spanish-language AIMS questionnaires independently. They were also asked to share the instruments with family members and friends to elicit additional opinions about suitability for South Texas. When the questionnaires were returned, each reviewer was asked to indicate which of the questionnaires was the “one best translation” for this region. Eighty-six percent (13 of 15) selected the idiomatic translation, prepared specifically for South Texas, as being the most suitable. Back-trunslation. Two additional bilingual individuals, both of whom met the aforementioned criteria, independently and then jointly “fine-tuned” the idiomatic version, selected by the larger group of reviewers, via translation back and forth between English and Spanish. Pilot-resting. The resultant AIMS translation was pilot-tested with 12 more bilingual individuals to assess comprehension and readability. This group identified a number of English words that could be translated into Spanish in different, but equally acceptable ways. In these cases, alternative Spanish wordings were placed in parentheses within the text. Formnt design. The Spanish text (South Texas idiomatic translation) was typed beneath the corresponding English phrases. Some of the original review panel members were then recruited to proofread the final document. Validation. The validation study was conducted with patients who were attending the rheumatology clinic of the University of Texas Health Science Center at San Antonio. All patients had been previously diagnosed as having rheumatoid arthritis. The dual-language AIMS was mailed to the homes of 177 patients in January 1987. A mail-out approach was used to simulate the conditions planned for the continuing education program. Sixty-one percent of the subjects returned the questionnaire. To establish testretest reliability, a second mailing of the instrument was sent to these 107 responders approximately 6 weeks later. Seventy-two questionnaires in the second mailing were returned (67.3%). During the same period of time, the subjects also completed 2 administrations of the Sickness Impact Profile (SIP) (18), designed in a similar dual-language format, so it could be used as an independent measure of disease severity in these patients. The ethnic composition of the subject group was ascertained by self-report. Fifty-eight percent were Hispanic, 31% were Anglo (non-Hispanic Caucasian), and

BRIEF REPORTS

the remainder were either black (7%) or failed to indicate ethnicity (4%). Fifty-two percent reported that they spoke primarily Spanish at home, while the rest spoke English or used the 2 languages interchangeably. This information was not certified independently, but is consistent with other reports (12). Seventy-six percent of the subjects were female. The mean age was 56 years. Approximately 70% had not completed more than 10 years of schooling, and 80% reported an annual income of