CROSS-CULTURALADAPTATION OF A HEALTH STATUS CLASSIFICATION SYSTEM IN CHILDREN WITH CANCER. FIRST RESULTS OF THE FRENCH ADAPTATION OF THE HEALTH UTILITIES INDEX MARKS 2 AND 3 Catherine LE GALE ` S 1 *, Nathalie COSTET 1 , Jean-Claude GENTET 2 , Chantal KALIFA 3 , Didier FRAPPAZ 4 , Christine EDAN 5 , Eric SARIBAN 6 , Dominique PLANTAZ 7 and Franc ¸ois DOZ 8 1 INSERM U 357, Ho ˆ pital de Bice ˆtre, Le Kremlin-Bice ˆtre, France 2 CHU La Timone, Marseille, France 3 Institut Gustave Roussy, Villejuif, France 4 Centre Le ´on Be ´rard, Lyon, France 5 CHR Sud, Rennes, France 6 Children’s Hospital Reine Fabiola, Brussels, Belgium 7 CHU de Grenoble, Grenoble, France 8 Institut Curie, Paris, France O ur objective was to adapt and validate the H ealth U tilities Index Mark 2 (HUI 2) and HUI 3 health status classification systems self-report questionnaire in a population of children with cancer, a group of 42 children already included in a multi-centre database designed by the Group on Brain Tu- mors in Children of the French Society for Pediatric Oncol- ogy. Children were recruited during a routine consultation. Most of them had completed treatment. The version of the questionnaire for French adults was adapted linguistically for children. Open-ended queries by children about the compre- hensiveness of the questions and very low non-response rates showed a good acceptability of the questionnaire. The main psychometric properties of the H U I 2 and H U I 3 classification systems were assessed in 3 groups of raters (child, parent, physician): construct validity was tested against the rating of the child’shealth state on a Likert scale and through compari- son with clinical data, and internal consistency was deter- mined through multi-trait analysis. W eighted and un- weighted kappa values were used to measure the inter-rater agreement between the child’s, parent’s and physician’s assessment of the child’shealth state. T he convergent validity was satisfactory, with better results when the physician’s assessment was used. The most affected attributes were the expected ones ( i.e., cognition, pain and emotion). Disagree- ment was observed between the 3 raters, more often in the same direction: taking the child’s assessment as the refer- ence, the parents tended to under-estimate the health status while physicians tended to over-estimate it. Int. J. Cancer Suppl. 12:112–118, 1999.  1999 Wiley-Liss, Inc. The evolution of the survival rates in pediatric oncology during the recent decades raises 2 very different issues. On the one hand, for many diagnostic subgroups of cancer, 5-year survival rates from diagnosis of 60% to 70% can be achieved as a consequence of effective treatments (Bleyer, 1990). These results have led to the current increasing concern to minimize short-term toxicity, as well as the long-term consequences of treatment, without compromising survival results. On the other hand, the survival rates for a variety of tumors have remained very low for many years, despite active clinical therapeutic research. In both situations, traditional end points to measure the effectiveness of treatment, such as disease- free survival, need to be supplemented by indicators reflecting the overall impact that aggressive treatments have on the patient and sometimes on his/her family. Thus, any new therapeutic regimen should be evaluated simultaneously in terms of duration and quality of life. As a consequence, there is a general and growing consensus among pediatric oncologists to consider that health- related quality of life (HRQL) is a pertinent issue for children with cancer (Reaman and Haase, 1996). Assessing HRQL is therefore relevant to improving clinical management, to assessing the status of survivors, and to aiding societal decision making about treat- ment alternatives and resource allocation. To address this obvious need, the existing HRQL tools are scarce and their performance characteristics sometimes incompletely documented, especially in the pediatric population (Bullinger and Ravens-Sieberer, 1995). In France, there is no specific or generic instrument available to assess HRQL for children with cancer. The lack of such an instrument is not limited to cancer but includes the very large majority of children’s health conditions. We report here on the continuing French experience in adapting and validating the multi-attribute Health Utilities Index Mark 2 (HUI 2) and HUI 3 in the French population of children with brain cancer. We focus on the overall strategy developed to make an HRQL instrument available, thus responding to clinicians’ expecta- tions and objectives. We shall address successively the rationale of our approach and describe the methods and first results we have obtained. The dearth of a French quality-of-life instrument in pediatrics led the French Group on Brain Tumors in Children to discuss 2 approaches: whether to develop an original French instrument or to adapt an existing adult tool to pediatric patients and to assess the psychometric properties of such a version. From the clinician’s point of view, the instrument has to provide an overall description of the health status of children with brain cancer which explicitly integrates the multi-dimensionality of the HRQL concept. In addition, several constraints have to be met, including briefness of administration of the questionnaire, ease of understanding by children and their families and the possibility of different modes of administration. HUI 2 and HUI 3 are generic, multi-attribute health status classification systems, for which cross-cultural adaptation and psychometric assessment of the self-administered questionnaire have been conducted by our research team in France (Costet et al., 1998). The self-administered questionnaire was shown to be very easy and quick to complete in different French adult patient populations. Other important elements in favor of developing a French pediatric version of the HUI were the original development of the HUI 1 for the purposes of measuring of health status in children (Feeny et al., 1995), the use of HUI 2 and HUI 3 in a number of clinical studies of children with cancer (Feeny et al., 1992, 1993; Barr et al., 1993, 1994, 1995; Kanabar et al., 1995) and the growing experience in national adaptations of HUI in Europe within a pediatric context (Billson and Walker, 1994; Gemke et al., 1995; Gemke and Bousel, 1996). Two major objectives were addressed when designing the research protocol for children: (i) to obtain evidence on the performance characteristics of the HUI 2 and HUI 3 with a special *Correspondence to: INSERM U357 ‘‘Health Economic Research’’, Ho ˆ pital de Bice ˆtre, Pavillon de la Force, 80, rue du Ge ´ne ´ral Leclerc, 94275 Le Kremlin-Bice ˆtre Cedex, France. Fax: +33-1-46-71-32-70. Int. J. Cancer: Supplement 12, 112–118 (1999)  1999 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer