- Top of page
- MATERIAL AND METHODS
Quality-of-life assessment is being used increasingly in clinical research. This is true particularly in the case of survivors of cancer in childhood, where improving survival rates have raised concern regarding the long-term effects of medical cure. Health-status assessment and quality-of-life instruments have been developed for the most part in the English language, thus necessitating their translation and cultural adaptation for use in non-English-speaking countries. Our purpose was to develop a set of Spanish-language questionnaires for application with a population of children with cancer in a tertiary-care center in Buenos Aires, Argentina. The Health Utilities Index (HUI), a conceptual framework for assessing health status, was chosen for this study. Three distinct questionnaires, based on the HUI, were used: a self-completed one for health professionals and teachers (15Q) to report assessments of children and 2 interviewer-administered ones, for child survivors (42Q) to report assessments about their own health status and parents (45Q) to report assessments about their children's health status. The original translations and reviews were accomplished with direct oversight by members of the HUI Group, to ensure conceptual equivalence. The instruments were then tested in Buenos Aires by application to staff of the hematology-oncology service, childhood cancer patients and the parents of childhood cancer patients. Several modifications were made based on these tests. We concluded that the translation and cultural adaptation of these instruments was adequate for use with the groups tested in a pilot survey of survivors of childhood cancer in Argentina. Int. J. Cancer Suppl. 12:119–124, 1999. ©1999 Wiley-Liss, Inc.
The World Health Organization (1952) defined health as being not only the absence of disease but also the presence of physical, mental and social well-being. Since then, quality-of-life issues have become increasingly more important in health-care practice and research. The use of health-related quality of life (HRQL) evaluation techniques has increased exponentially in clinical research since 1973 (Testa and Simonson, 1996).
The majority of adult patients who are afflicted by advanced forms of the most common malignant diseases are incurable. In such cases, treatment is administered to maximize the prolongation of life. Thus, not only the duration of survival but also the quality of life associated with the effects of modern medicine should be considered when judging the benefit of such treatments.
However, many common malignant diseases in childhood are considered curable, and improved survival rates have raised concerns regarding the long-term effects associated with medical “cure”. Because the potential life expectancy is long among survivors of cancer in childhood, this group of patients in particular could benefit from health-status assessments to evaluate morbidity. Although these types of measure are becoming increasingly available, they have been developed for the most part in the English language, thus requiring that they be translated and culturally adapted before use in non-English-speaking countries (alternatively, instruments could be developed locally).
There is little or no published information about the comprehensive health status and HRQL of childhood cancer survivors treated in developing, non-English-speaking countries. The Health Utilities Index (HUI) was chosen as the basic tool to assess health status and HRQL for this pilot study, for 2 reasons. Firstly, it has been shown to provide reliable, valid, responsive and readily interpretable measures of health status when applied in a wide variety of pediatric population groups (Feeny et al.,1993; Boyle et al.,1995; Furlong et al.,1996; Gemke and Bonsel, 1996; Barr et al.,1997; Grootendorst et al.,1997; Costet et al.,1998). Secondly, there is accumulating published evidence about the health status and HRQL, assessed within the HUI framework, for survivors of various childhood cancers (Feeny et al.,1992, 1993; Barr et al.,1993, 1994, 1995, 1997; Feeny et al.,1992, 1993; Billson and Walker, 1994; Kanabar et al.,1995; Kiltie and Gattamaneni, 1995; Glaser et al.,1997a,b), and for other groups of pediatric patients treated for serious or life-threatening health problems (Saigal et al.,1994a,b, 1996; Gemke et al.,1995; Gemke and Bonsel, 1996). Thus, there is a substantial body of evidence with which results from application of the HUI in Argentina could be compared.
This report describes the process used to develop a set of questionnaires for surveying the comprehensive health status of childhood cancer patients at a large pediatric tertiary-care center in Buenos Aires, Argentina. This pilot survey, designed to be completed by one of the authors (NS) during a medical school elective term spent in Argentina between September 1996 and February 1997, involved 3 major tasks. Firstly, it was necessary to translate and culturally adapt a set of 3 existing English-language questionnaires into Spanish. Secondly, the questionnaires were tested to assess the feasibility of using HUI-based assessment tools in Argentina. Thirdly, the translated instruments were used in conducting 2 surveys that collected preliminary data about the comprehensive health status of a group of patients treated for a number of childhood cancers in Argentina (retinoblastoma, Hodgkin's and non-Hodgkin's lymphoma, acute lymphoblastic leukemia, Ewing's sarcoma, osteosarcoma, Wilms' tumor and brain tumors) and a group of control subjects (children in the general population).
The health status surveys were conducted in 2 settings: the hematology-oncology service of the Juan P. Garrahan Pediatric Hospital in Buenos Aires and 2 public schools, one in the city and another in an adjacent county in the greater metropolitan area of Buenos Aires. The Juan P. Garrahan Pediatric Hospital is a publicly funded, 12-year-old facility situated in the capital city of Argentina. It is an academic center linked to the University of Buenos Aires. For the first few years of operation it was strictly a referral center that dealt only with the more complex pediatric cases in the country. Now, although the hospital continues to be an important referral center for rare and serious pediatric illnesses, it also provides primary care. The hematology-oncology service sees approximately 300 new cancer patients every year, 40% of whom are referred from other provinces in the country. The patient population is mostly from low socio-economic class families. Many middle and higher socio-economic class families seek pediatric oncology services at private institutions, with the expectation of obtaining better care. Some middle and upper socio-economic class families, however, seek treatment of the Garrahan Hospital because they recognize the international caliber of the services offered by this center. Despite their ability to pay more for medical services, these families choose to use the Garrahan Hospital for the care of their sick children.
- Top of page
- MATERIAL AND METHODS
The initial validation activities described in the Methods section led to various observations about the questionnaires and their reception by the test respondents. Concerns about the clarity, appropriateness or relevance of certain questions arose during this process, and modifications were made to the instruments based on these results. This section will describe the observations of the validation process and the modifications made.
The review of the translated version of the 15Q by the hematology-oncology staff led to no modifications in the questionnaire. Several suggested changes were indicated for syntax, and a few comments were made about lack of relevance of certain attributes. After careful consideration, the translation committee (NS and GM) decided that the syntax suggestions reflected the personal style of the respondent and not the general appropriateness of the wording of the questions.
Most of the problems encountered and subsequent modifications made in the interviewer-administered questionnaires stemmed from pre-testing the Spanish 42Q. Although the 45Q and the 42Q were pre-tested separately, many of the revisions made to the 42Q were also extended to the 45Q, to maintain conceptual equivalence between the 2 questionnaires.
Children had difficulty with a few questions in the pre-test of the 42Q. For example, when evaluating the attribute of hearing, the English questionnaire asks “Can you hear, without a hearing aid, what is said when at least three other people are talking together?” In the Spanish version, this was simplified by taking out “at least” from the question. The pre-test revealed that the children understood that all 3 people were talking simultaneously; to “talk together” in Spanish implies this meaning. They would answer therefore that they were not able to “hear” in this situation. In fact, they meant that they were not able to follow the conversation when more than 1 person was talking at the same time. Consequently, we changed this question to “Can you hear what is said in a conversation with three other people, without using a hearing aid?” This appeared to elicit responses more consistent with the intent of the original version of the questionnaire during the actual data collection.
Some children were so eager to answer the questions that they did not wait until the whole question was read. Additionally, they frequently considered the set of questions as an evaluative exercise, even though it had been explained to them that the objective was to record their opinions. They would answer, therefore, as they thought they should, as though there was a “right” answer. For example, when asking about vision, the questionnaire seeks to distinguish between the ability to see far and the ability to see near, not as different levels of the same ability but rather as distinct abilities. The questionnaire asks about near vision first and then far vision. Some children indicated that they had difficulty seeing far, when in fact they had no problems with their eyesight, presumably thinking that the 2 questions were asked for comparative purposes and that we were testing their logic. When the question was finished, or repeated to them if necessary, they would often “clue in” to its true intent.
Emphasizing a readily understandable and concrete ability was helpful in eliciting a sensible response. For example, it is better to ask about the ability to recognize a friend across the street than the ability “to see far”. A further example of this arose when evaluating speech. The Spanish 42Q asks the subject to report how well others can understand them. The respondents would often report no difficulty in being understood by those who knew him/her well but great difficulty with those who did not. Again, this may have been related to the order in which the questions were asked, namely, about acquaintances first and then about strangers. They felt it was “right” to be understood by close acquaintances and, subsequently, not by strangers, a reasonable progression. However, the interviewer, who had never met the respondents before, reported no difficulty in understanding the child's speech. Furthermore, to be “understood” by others can be interpreted as relating to thoughts and emotional needs. It is not difficult to see how a child might feel about his/her parents in this respect and even more so about strangers, all of this leading to answers which do not reflect the child's true perception about his/her speech and articulation. The investigator switched the order of the questions relating to speech for actual data collection with the test group and found mostly positive responses for both levels in children who had no problems with this attribute.
When evaluating cognition, the 42Q asks about the ability to learn and remember school work and if subjects believe they can do it “like their classmates”. Many children would answer “No” to this question initially, implying to us that they had difficulties with learning and that they were not up to par with the rest of their class. This response led the interviewer to ask whether they thought their learning was a little slower or much slower than their classmates. This second question often produced hesitation and caused the children to say that they were the same as their classmates or to change their answer to the first question. The first question is stated in a general way, allowing for flexibility of interpretation. Having difficulties with learning could be interpreted as finding homework difficult and tedious, a prevalent feeling among most school children. The second question, however, focuses on disability and forces the respondent to select 1 of 2 distinct levels of functioning (a little slower vs. much slower). To reduce the potential ambiguity of the first question, it was modified to emphasize a comparison with their classmates from the very beginning of the set of questions about cognitive ability.
The revised questionnaires were used subsequently with a group of cancer patients from the hospital as well as with a group of school children in the general population for the purpose of comparison. The demographic characteristics of these subjects (both patients and controls) are summarized in Tables I through IV. The data in Tables V and VI summarize the results from a group of retinoblastoma patients who have disabilities in the expected attributes, as reported (in this instance) by their physicians.
Table I. DISTRIBUTION OF DISEASES IN THE PATIENT SAMPLE
|Acute lymphoblastic leukemia||12||15.0|
Table II. AGE DISTRIBUTION OF PATIENTS AND CONTROLS
Table III. SEX DISTRIBUTION OF PATIENTS AND CONTROLS
Table IV. DISTRIBUTION OF HOME LOCATIONS FOR PATIENTS AND CONTROLS
|Greater metropolitan area||39||48.8||50||62.5|
Table V. RETINOBLASTOMA
|A: Distribution of patients by HUI 2 functional levels, all attributes (Physician assessments, n = 13)|
|Functional level||Sensation n (%)||Mobility n (%)||Emotion n (%)||Cognition n (%)||Self-care n (%)||Pain n (%)|
|1||10 (76.9)||11 (84.6)||11 (84.6)||12 (92.3)||13 (100)||13 (100)|
|2||(0)||(0)||1 (7.7)||1 (7.7)||(0)||(0)|
|3||1 (7.7)||(0)||1 (7.7)||(0)||(0)||(0)|
|4||2 (15.4)||2 (15.4)||(0)||(0)||(0)||(0)|
|B: Distribution of patients by HUI 3 functional levels, all attributes (Physician assessments, n = 13)|
|Functional level||Vision n (%)||Hearing n (%)||Speech n (%)||Ambulation n (%)||Dexterity n (%)||Emotion n (%)||Cognition n (%)||Pain n (%)|
|1||10 (76.9)||13 (100)||13 (100)||11 (84.6)||12 (92.3)||11 (84.6)||12 (92.3)||13 (100)|
|2||(0)||(0)||(0)||(0)||1 (7.7)||1 (7.7)||1 (7.7)||(0)|
|6||2 (15.4)||—||—||1 (7.7)||(0)||—||—||—|
Table VI. RETINOBLASTOMA: SINGLE-ATTRIBUTE AND MULTI-ATTRIBUTE HUI SCORES (PHYSICIAN ASSESSMENTS, n = 13)
- Top of page
- MATERIAL AND METHODS
We have described our experience translating and adapting 3 HUI 2/3 questionnaires into Argentine Spanish. The translations were reviewed, pre-tested and reviewed again, generating modifications at each step. A major strength of this experience has been that the translations were produced with direct oversight by the instrument developers. Based on the preliminary results, it appears that one can translate instruments culturally for use with children, parents and health-care professionals. This process may have sufficed for the present pilot study. However, the translations can be viewed only as provisional. More steps are required to render the Spanish instruments as culturally equivalent as possible to each other and the original English source documents.
There are extensive publications providing guidance for translation and adaptation of questionnaires (Guillemin et al.,1993; Mathias et al.,1994; Acquadro et al.,1996; Anderson et al.,1996a,b; Bullinger et al.,1996; Cella et al.,1996). For example, Guillemin et al. (1993) describe several steps. These include the execution of several forward translations, use of qualified translators (as defined by several conditions, only some of which were met in the present pilot study), back-translation using appropriate back-translators, review by a multi-disciplinary committee, formalized modification protocols and pre-testing using a probe technique or bilingual appraisal.
The HUI is a generic preference-based system for assessing health status and HRQL. The health-status classification systems are comprised of attributes which are applicable to a very broad range of subjects (Feeny et al.,1995), the levels focus on the “within the skin” functional capacity of individuals with minimal reference to environmental context (Feeny et al.,1996) and the utility scoring formula is based on preferences of a multi-cultural general population (Torrance et al.,1996). It is possible that socio-economic, demographic and cultural factors could affect health-status assessment as measured by the HUI instruments. In general, there is considerable evidence about heterogeneity of preferences for health states among individuals (Furlong, 1996). However, there is little evidence to support the notion that there are systematic differences in preference scores of general population respondents for comprehensive health states by socio-economic factors (Froberg and Kane, 1989; Stouthard and Essink-Bot, 1991; Sackett and Torrance, 1992; Kaplan, 1994; Furlong, 1996). Furthermore, as noted at the beginning of this report, appropriate HRQL instruments are becoming available only now in developed countries, and there is increasing demand for HRQL information about subjects living in less developed areas. For these reasons and with acknowledgement that the HUI was developed originally in Canada and additional evidence is required for a definitive answer about this issue, it is suggested that the HUI be considered a provisional measurement tool for assessing health status and HRQL of childhood cancer patients in Argentina.
Given that the order of some questions was reversed and other modest revisions were made to adapt existing questionnaires for use in Argentina, it will be important that the standard coding algorithms for the English-language source questionnaire be modified to ensure that attribute levels are determined based on the content of information collected using the Spanish-language questionnaire structure.
We will be able later to make an exploratory assessment of construct validity by comparing health status between groups to see if we observe the expected results. For example, patients with bone tumors who needed amputation of a lower limb should have reported deficits in mobility and ambulation, whereas those who are unable to communicate, as a result of brain surgery, should have reported deficits in speech or hearing. Furthermore, the results from this population should be similar to results from populations assessed previously, after taking into account social, economic, demographic and cultural factors.