Failure of the French health care services to diagnose and manage problem drinkers has been established on many occasions. This results from a relative lack of knowledge of the basics of alcoholism and the low level of involvement of health care professionals in the management of these patients. In response to this inadequacy, the French Public Health Ministry promoted the development of managed care and coordination of care for these patients. Teams in charge of coordinating care for problem drinkers have been implanted in 96 French hospitals since 1996. These teams aim to provide support to health care providers caring for problem drinkers. For this purpose, they have delivered continuing medical education designed to prepare health care providers to identify and manage these patients in various settings. However, no formal assessment of the impact of these interventions on the knowledge, perceptions, and practices of health care professionals has been planned. The assessment of these interventions can rely on qualitative methods such as observation, focus groups, or individual interviews. However, qualitative methods require specific skills, are time-consuming, and cannot be implemented on a large scale. In contrast, quantitative methods using survey questionnaires or standardized instruments appear to be more appropriate for large scale or repeated evaluations. However, the accuracy of ratings provided by these instruments can be affected by many methodological factors, including the quality of the instrument in terms of validity, reliability, and sensitivity to changes. The aim of this paper is to describe the properties of French and English language questionnaires designed to assess the knowledge, perceptions, and practices of health care professionals with regards to alcoholic patients.
Methods: The instruments were retrieved by searching the Medline, Pascal, and Sudoc computerized bibliographic databases from January 1964 to December 2002. The following medical Subject Headings (MeSH) and text words were used: "*alcoholism/psychology/therapy", "questionnaire", "healthy care surveys", "attitude of health personnel", and "*knowledge/attitude/practice". In addition, the table of contents of the French journals devoted to alcoholism that were not indexed in electronic bibliographic databases were examined (Alcoologie et addictologie, Alcoologie). The bibliographies of relevant articles were also examined for additional citations. Finally, a phone survey of 34 professionals caring for alcoholic patients located in French university hospitals was carried out to retrieve unpublished questionnaires. Two authors independently extracted data from each study using a standard date abstract form. Disagreements were resolved through consensus. They extracted information on the development process and properties of validity, reliability, and sensitivity to changes. Validity is a concept concerned with the extent to which an instrument actually measures what it is supposed to measure. It is assessed through different facets (content, construct, criterion, discriminant, and predictive validity). Reliability reflects the amount of error inherent in any measurement. Sensitivity to change corresponds to the property of an instrument to identify small but clinically significant changes in attitude or practice.
Results: A total of 57 relevant publications involving 39 original instruments were identified. Eighty questionnaires were not available, despite the solicitation of their developers. Overall, the study included 21 instruments. Of these, 20 were English-language questionnaires and one was in French. The conceptual frameword was specified for only two questionnaires: "the Addiction Belief Scale" and "the Alcohol and Alcohol Problems Perception Questionnaire". The number of items ranged from 9 to 122. Items were derived from qualitative surveys in only two cases. In the other cases, they were derived from the literature, expert panels, or form unknown sources. The internal construct validity of eight questionnaires was assessed using principal component analysis, factor analysis, or cluster analysis. We considered that external construct validity was analyzed by testing empirical hypotheses derived from the literature for 15 of the instruments studied. Predictive validity cas evidenced for only one questionnaire. Internal consistency was documented for seven instruments, using Cronbach's coefficient. Test-retest or inter-rater reliability was assessed for none of the instruments. The sensitivity to change of two instruments was analyzed. The response rate to the instruments ranged from 46% to 93%.
Discussion: The main finding of this study is that properties of validity, reliability, and sensitivity of questionnaires designed to assess the knowledge, perceptions, and practices of health care professionals with regards to alcoholic patients are neglected. Moreover, these questionnaires generally lack a theoretical background. Hence, the interpretation of responses to these questionnaires may be misleading. The present study has several limitations. The search strategy could be criticized for searching only a small number of databases, and selecting French and English language questionnaires. However, given the large range of journals included, it was unlikely that the strategy has limited the generality of the results. In addition, we identified no additional relevant articles by contacting professionals working with alcoholic patients in French university hospitals. Articles reporting instruments designed to assess the knowledge, perceptions, and practices of health care professionals with regards to alcoholic patients are incomplete. Journals should require minimum evidence of validity, reliability, and sensitivity when reporting results of surveys, development of questionnaires or standardized instruments.
Conclusion: The psychometric characteristics of French language questionnaires were given very little attention. In order to evaluate the effectiveness of their continuing medical education programs, the teams that train French medical staff caring for alcoholic patients should either develop a standardized survey questionnaire following a strict methodology or translate an English language instrument, which would also require a cross-cultural validation.