Depression factors in depressed and in heterogeneous in-patient samples

Meta-analysis of the factor structures of four depression questionnaires: Beck, CES-D, Hamilton, and Zung

Four separate metaanalyses of factor analyses were conducted for the Beck Depression Inventory (BDI), the Center for Epidemiological Studies Depression Scale (CES-D), the Hamilton Rating Scale for Depression (HRSD), and the Zung Self-Rating Depression Scale (SDS). The total number of participants (N) and studies (k) included in each of the metaanalyses were the following: BDI (N = 13,643, k = 33), CES-D (N = 22,340, k = 28), HRSD (N = 2,606, k = 17), and SDS (N = 12,621, k = 13). Metaanalysis results suggest that the specific depression symptom factors within each test appear to be relatively robust and well established and match fairly closely previously hypothesized factor structures. A general Depression Severity factor and a small Somatic Symptoms factor are found in all four tests and two tests had a small Positive Affect factor. There were fewer common specific depression symptom factors across tests than expected.

Factorial structure and factor reliability of the Hamilton Rating Scale for Depression

Identical principal components factor analyses of total Hamilton Rating Scale for Depression scores were conducted for two separate sub-samples (N = 183 and N = 182) drawn from the same clinic population of moderately depressed outpatients. A comparison of the two factor analyses revealed substantial agreement for four factors across the two sub-samples (i.e., four factors from the first sub-sample correlated at least 0.80 with a homologous dimension in the second sub-sample). The four factors were labelled Somatic Complaints, Anorexia, Sleep Disturbance, and Agitation/Retardation. Some additional factors emerged in the analysis of one sub-sample but failed to appear in the other. Overall, these results suggest that the HRS exhibited a relatively stable factorial structure based on a large sample of outpatients with unipolar depressive disorders. Methodological problems with earlier research are discussed in light of the current findings.

[Dimensions of the Hamilton Depression Scale. Factor analysis studies]

The results of factor analysis of HAMD are not unitary. Studying the stability in course and the independence of selection criteria for the population we rated two groups (n = 107, n = 98) of inpatients with HAMD; these groups differ in nosological selection criteria. We furthermore rated one group at different timepoints (before treatment and 3 weeks after). The one-factor-solution is the only stable one and the only one which is independent from selection criteria. In all groups the one-factor-solutions show a high similarity with other published solutions of the German version of HAMD. But there is no way to extract a general factor. Thus the ability of HAMD in judging the severity of depression is doubtful. Ways are discussed to circumvent this problem.

The Hamilton depression scale. Evaluation of objectivity using logistic models

The consistency of the Hamilton Depression Scale (HDS) as a measure of the severity of depressive states has been examined when the scale was used weekly during a trial when imipramine. By use of logistic models (Rasch) the consistency of the HDS has been considered across patient-variables as age, sex, plasma levels of imipramine, and diagnosis. The results showed that the original 17-item HDS was without adequate consistency, i.e. the total score of the sample of items was no one-dimensional measure of depressive states. However, a melancholia subscale of the HDS contained items the total of which can be used to compare patients quantitatively, although in some part of the analysis one of these items showed ceiling effect. It was concluded that the melancholia subscale (containing the items depressed mood, guilt, work and interests, retardation, psychic anxiety, and general somatic symptoms) can form the basis for further improvements in the field of quantitative rating scales for depressive states.