Assessment of symptom change from improvement curves on the Hamilton depression scale in trials with antidepressants

A total of 97 patients, who participated in two studies on the relationship between the clinical effect and plasma levels of imipramine and clomipramine, were examined for improvement curves by use of weekly ratings on the Hamilton Depression Scale (HDS). Although we confirmed that our six-item HDS subscale, in contrast to the total 17-item HDS, was a one-dimensional measure of depression, the Rasch analysis showed that the weekly improvement in subscale scores only applied to the individual patient, i.e. an average improvement curve for a group of depressed patients is an abstraction to which the individual curves cannot be transferred. Our results indicate, however, that when the subscale scores are transformed into three clinical categories of depression: no, mild (minor), moderate/-severe (major) they could be described by a common improvement curve for all patients. This is illustrated by the percentage of patients who, week to week, changed from major to minor or no depression, or from minor to no depression. We found no specific improvement pattern for imipramine or clomipramine which could be used diagnostically. There is reason to assume that patients completing a controlled trial necessarily will follow a monotonic improvement curve, and the improvement pattern of all patients fulfilling the entry criteria should, therefore, always be reported. The present study thus indicates that calculation of average improvement curves is neither clinically nor statistically meaningful, and should be replaced by measures of changes in number of patients in different main severity categories, or by the final rating score. No difference in outcome between imipramine and clomipramine was shown neither on the subscale nor on the 17-item HDS.