Responsiveness or sensitivity to change of a questionnaire is an important criterion in the assessment of the effect of a medical intervention. In the literature there are several criteria for responsiveness, based on changes in the population mean. Here we present an alternative approach for the measurement of change (change reliability), applying the criteria of internal consistency to change scores. Using the Hospital Anxiety and Depression Scale HADS, these approaches are demonstrated and compared for 2 subject samples, comprising 901 oncologic patients and 2 695 cardiac patients. The effect size d is well suited to assess the effect of an intervention, based on changes of the sample mean. The analyses of the 2 samples yielded effect sizes of about 0.25. The newly created change reliability yielded coefficients of about .70 which indicate the reliability of the individual changes in anxiety and depression.
Georg Thieme Verlag KG Stuttgart.New York.