A number of patient-reported outcomes (PROs) have been developed but insufficient attention has been devoted to the assessment of their clinimetric properties. Clinimetrics, the science of clinical measurements, has been considered an emerging approach for evaluating reliability and validity of PROs. This is the first study using clinimetric principles to compare the construct and criterion validity of the Major Depression Inventory (MDI), the Beck Depression Inventory-II (BDI-II), the World Health Organization Well-Being Index (WHO-5), three of the most widely used PROs for the assessment of depression.
Construct validity was evaluated via Item Response Theory (IRT) models (i.e., combining Rasch and Mokken analyses). Using the ICD-10 diagnostic algorithm for any depression as the gold standard, Receiver Operating Characteristic (ROC) curves were performed to examine the criterion validity of PROs.
One hundred healthy subjects (73% females, 32.6 ± 10.5 years) participated in the study, giving a response rate of 90.1%. When using IRT analyses, MDI and WHO-5 were found to be reliable and unidimensional, while BDI-II showed lack of unidimensionality. ROC analyses supported the diagnostic accuracy of MDI and the screening properties of WHO-5.
The main limitations of the present study are that healthy subjects were assessed only via only self-reported measures and a cross-sectional design was used.
WHO-5 and MDI outperformed BDI-II in terms of construct and criterion validity. WHO-5 should be considered when screening for depression, while MDI should be used as a valid diagnostic instrument and as a unidimensional measure to assess depression severity.