Related Experiment Video
Updated: Apr 11, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Shortening the PHQ-9: a proof-of-principle study of utilizing Stochastic Curtailment as a method for constructing
Niels Smits1, Matthew D Finkelman2
1Department of Clinical Psychology and Department of Methods, Faculty of Psychology and Education, VU University, Amsterdam, the Netherlands.
Objective:
In primary care, screening instruments for mental health should be ultrashort to allow for routine usage. In this paper, Stochastic Curtailment is introduced as a method for constructing ultrashort screeners.
Method:
In a post hoc diagnostic accuracy study using the item scores on the Patient Health Questionnaire depression module (PHQ-9) of a large sample (N=20, 685), Stochastic Curtailment was compared with two existing ultrashort versions of PHQ-9. The first was PHQ-2 (which includes the first two items of PHQ-9), and the second was a 'two-step' method (only if the PHQ-2 screened positive were all nine PHQ items administered). For PHQ-2 and two-step, both cut scores 2 and 3 were evaluated.
Results:
PHQ-2 showed the lowest and Stochastic Curtailment the highest diagnostic accuracy with reference to the classifications based on the full PHQ-9. To do so, Stochastic Curtailment used 3.08 items on average (S.D.=1.98), which was slightly less than two-step (M=3.18, S.D.=2.62) under its most accurate cut score (≥2).
Conclusions:
It was concluded that Stochastic Curtailment is a promising method for constructing ultrashort screeners.
Related Concept Videos
01:09Censoring Survival Data
01:02Quantifying and Rejecting Outliers: The Grubbs Test