Adjustierung für Patientenmerkmale bei der Auswertung von Befragungsdaten

Abstract

Patient questionnaires are a frequently used instrument within the framework of quality management in in- and outpatient care. Often such questionnaires enable a comparison of care suppliers with the consequence that one turns out to be visibly better or poorer than another. This process, in turn, makes it necessary to check whether differences found upon evaluation of the questionnaires are not merely the result of different compositions of the questioned populations. Although frequently demanded, such adjustments are not usually made. The present article describes the choice of variables for adjustments and the statistical procedures for a relatively homogeneous sample of breast cancer patients. In addition, the utility and limitations of adjustments are discussed. On the basis of questionnaire data from 3 840 breast cancer patients of 52 breast cancer centres in North Rhine-Westphalia collected during 2010, we examined which patient characteristics can be employed for the adjustment of satisfaction ratings and to what extent the observed values for the centres differed from the expected results. Independent variables taken into consideration were age, educational level, native language, stage, grading, ASA classification, afffected breast, type of operation, insurance status, partnership status as well as time between operation and receipt of the filled out questionnaire. The variance revealed by the independent variables is low. The expected values showed minimal differences which can be attributed to the high homogeneity of the patients collectives and the centres. The use of adjustments remains limited in the study population. The variance of the independent variables revealed by the adjustors is small. Finally, in our opinion, no clear recommmendation for or against case-mix adjustments can be made in patient populations such as the one examined here. Thus, even when small, effects for a more correct reporting of patient questionnaires are faced with unresolvable methodological challenges. Also of importance but an as yet only rarely discussed factor is the factual intepretation of the association of patient characteristics with a better or poorer evaluation of questionnaires. An adjustment for the respective characteristics would eliminate these findings and not make any contribution to an improvement in health care.