Psychometric properties of most commonly used screening and case finding tools for major depressive disorder in non-psychiatric settings

Background:
Major depressive disorder (MDD) is the most common mental disorder in medical population, but its identification remains poor. Various screening strategies, involving the use of brief questionnaires to identify patients with unrecognized MDD, have been proposed as a way to improve this. However, the recommended use of specific questionnaires was based on limited or indeed, inaccurate information.

Aim:
To investigate how the most commonly recommended instruments to screen or case find for depression perform in non-psychiatric settings.

Methods:
This thesis is based on six published papers that used rigorous systematic review and meta-analytic research methods to assess the diagnostic accuracy of depression identification instruments recommended by national guidelines.

Main findings:
This work has highlighted significant shortcomings in the existing evidence on depression screening instruments’ accuracy. The methodological quality of primary validation studies was generally poor. The included reviews identified that most validation studies have been conducted in samples too small to precisely estimate accuracy and may have selectively published accuracy results from high-performing cut-offs. For the standard cut-off points, the performance of examined instruments was generally poorer than that reported in the original validation studies. Moreover, the diagnostic performance varied by healthcare setting. Presented evidence suggests that the same single threshold might not be appropriate in all settings.

Conclusions:
The research included in this thesis has been used in various national guidelines on identifying depression, and may further change how common depression identification instruments are used. The quality of these reviews has been recognised in subsequently published research. This work has also highlighted important methodological issues that have not been previously identified in diagnostic test accuracy research - the potential allegiance effect in studies co-authored by the original developers of an instrument. Important recommendations are made that are hoped to improve research and clinical practice in this area.