The Patient Health Questionnaire-2 (PHQ-2) is the most commonly used ultra-brief depressive symptomatology pre-screening instrument. However, this analysis demonstrated the PHQ-2 is not superior to other PHQ-9 item pairings in this sample. 12 out of the 36 pairings had an equal or higher cross-validation area under the receiver operating characteristic curve than the PHQ-2. The two pairings of phq2&4 and phq2&8 performed best and suggest low energy or psychomotor retardation or agitation may be more discriminatory symptoms than anhedonia. Pairings were combined with machine learning models which are not restricted to the summation and greater-than-or-equal-to logic of the PHQ-2 thresholds.
Funding
SFI Centre for Research Training in Foundations of Data Science
This work has emanated from research conducted with the financial support of Science Foundation Ireland (SFI)
under Grant Number 18/CRT/6049 (D.G.). The PROACTIVE study was funded by São Paulo Research Foundation
(FAPESP, process number 2017/50094‐2) and the Joint Global Health Trials initiative jointly funded by Medical
Research Council, Welcome Trust, and the UK Department for International Development (MRC, process number
MR/R006229/1). C.A.N. is supported by FAPESP (2018/19343-9) and M.S. by CNPq-Brazil (307579/2-19-0). We
would like to acknowledge the Guarulhos Health Secretariat who has supported the conduction of the PROACTIVE
study.