The new rasch-based depression screening (DESC): Evaluation and validation in different patient groups and a large German population sample

Abstract

Introduction

Comorbid depression in somatic illness is highly prevalent. However, most depression questionnaires show violation of unidimensionality and hypersensitivity of items.

Objectives

The Depression Screening (DESC) is a new self-rating scale for the measurement of depression severity that we developed from a calibrated Rasch-homogeneous item bank (N = 367) to improve short depression assessment in patients beyond psychiatric settings.

Aim

To present data for both parallel forms (10 questions each) of the DESC, and to examine its Rasch model qualities in different patient groups and a representative German general population sample.

Methods

Patient groups with depression, heart disease, neurological or otolaryngological disease (N = 333) were investigated with the DESC accompanied by diagnostic interviews and established depression scales. The demographic sample (N = 2509) was interviewed face to face, too. Adherence to Rasch model assumptions was determined with analysis of model fit, and further measures. Norm values were calculated.

Results

Both DESC versions feature different but highly correlated item sets. Cut-off scores with good sensitivity (0.82, 0.81) and specificity (0.78, 0.77) for the diagnosis of an affective disorder were developed with ROC analyses. No serious Differential Item Functioning in Rasch analyses for both scales was found. In the German validation sample the Rasch dimension “depression” explained 68.5% and 69.3% respectively of the variance. Validity was determined through sufficient intercorrelations with other scales (e.g. BDI, HADS).

Conclusions

Our studies support the good psychometric RASCH qualities of both DESC forms and offer important advancements in depression screening being of use for time-limited clinical and research applications.