Time-dependent structural validation of the Hamilton depression rating scale (HAM-D) in patients with moderate depression – Revaluation of two controlled trials

Introduction

There is a long lasting discussion to use short forms of the HAM-D instead of the common used 17-item form to safe time and get more accurate results. In more theoretical terms this means looking for responsive factors which account for the variation between visits. We therefore assessed the latent structure and its stability over time of the HAM-D in 2 controlled trials with different time courses.

Methods

We investigated a 6 week 3-arm study comparing Hypericum extract STW3-VI, citalopram and placebo and a 24 week study comparing Hypericum extract STW3 and sertraline. To assess the latent structure of the HAM-D 17 item form over time we conducted structural equation models at several time intervals (using AMOS 7.0). An analysis of variance (ANOVA) for repeated measures (using SPSS 15.0) was used to assess the improvement over time of different total scores if appropriate.

Results

In the course of the short term study no clear factorial structure of the HAM-D 17 item form was to be found; in the long term study however a significant 1 factor solution resulted in the later visits which was then stable over time. This factor can be addressed as a core-depressive factor. A recalculation with a so defined short form of the HAM-D revealed the same treatment course of the study as in the original calculation.

Conclusion

The latent structure of the HAM-D is not stable across time. With longer treatment duration a single factor solution which is treatment responsive can be found. Therefore short forms of HAM-D should be used carefully because in short term treatment other variables may represent valid treatment success than in later phases.