Understanding the Impact of Treatment on the Dimensions of Childhood Depression

Background: Childhood depression is often chronic and recurrent, with only modest improvements after optimal treatments. Novel approaches are needed for accurate diagnosis, timely selection of the best treatment for childhood depression, and improvement of remission rates. An important part of diagnosis is to determine how many consistent dimensions underlie childhood depression and what impact treatment may have on these dimensions. Objective: The goal of the current study is evaluate the consistency of the number of depressive symptom dimensions and examine the effect of depression treatment on the change in dimensions by seeing if this dimensional structure is the same for groups that were given a placebo and groups that were given an active drug, fluoxetine. Methods: We examined a combined sample of three clinical trials data sets with a total of 269 placebo patients and 261 fluoxetine patients at baseline, and 253 placebo patients and 249 fluoxetine patients at exit. Principal components analysis with associated parallel analysis followed by a ProMax oblique rotation was conducted on each week's scores. The similarity of the rotated structures between treatments and between weeks was then assessed using Wrigley and Neuhaus' (1955) degree of factorial similarity measure, which is a correlation-like statistic that takes two solutions based upon the same measures and assesses how similar they are. Results: Preliminary analysis indicated that mean scores of both treatment groups declined from baseline to exit, but the decline was greater for the fluoxetine group. There were four dimensions at baseline and two dimensions at exit in each treatment group. Hence the numbers of dimensions were the same in the two groups, whereas the content of the components was the same at baseline for the two groups, but was different at exit. Basically, the factorial similarities of the rotated factors were very high at baseline between the two groups for both factors (Ψ = 0.97 for factor I and 0.94 for factor II) before they were treated differently. However, at exit, the respective similarities dropped to 0.58 and -0.04 for factors I and II. Therefore, time had a large effect in changing both groups' factor structure, but it was especially large when there was an active treatment. Conclusions: In comparing the baseline and exit dimensions in a forced two factor solution of two treatment groups, it was revealed that treatment did have an impact on the change in dimensions. These results further the generality of the simplification of the factor structure of the Childhood Depression Rating Scale- Revised (CDRS-R) over time. In addition, these results show the effect of time and therapy on the factor structures. The fluoxetine group followed the trend of the content of the dimensions revealed in a previous study, whereas the placebo group had a less distinct change over time.