Early improvement (EI), i.e. a symptom reduction from baseline of at least 20% after 2 weeks, has been proven to be a clinically useful predictor for later treatment outcome. In most studies EI is identified by using the sum score of the Hamilton Depression Rating Scale (HAMD). Several unidimensional subscales of the HAMD exist, which have proven to be an economic measure of treatment change. Their ability to detect onset of improvement in comparison to the full HAMD has not been researched yet. The present study investigated in patients with major depression (MD) (1) whether the HAMD subscales are a valid and economic option to predict antidepressant treatment response in the early course of treatment and, (2) to validate the 20% EI criterion.

Based on data from 210 patients of a 6-week randomised, placebo-controlled trial comparing mirtazapine (MIR) and paroxetine (PAR) in patients with MD, the discriminative and predictive validity of EI for (stable) response/remission at treatment end was evaluated for the existing subscales in comparison to the HAMD17 in the total group as well as in the different treatment arms (MIR vs. PAR). Receiver operating characteristics (ROC) curves were used to validate the 20% EI criterion for the subscales.

Two subscales had similar predictive values than the full HAMD17, but overall, the HAMD17qualifies best for predicting antidepressant treatmentresponse/remission in the early course of treatment. The established 20%threshold of EI of the full HAMD17 scale also seems appropriate for the subscales.