Computerised adaptive testing across the paranoia continuum

Area of Science:

• Psychiatry
• Psychological Measurement
• Health Services Research

Background:

• Assessing psychotic experiences is crucial for tailoring and monitoring treatment.
• The heterogeneity of psychosis makes comprehensive assessment batteries impractical.
• Computerised adaptive testing (CAT) offers a solution by presenting the most informative questions, reducing assessment length.

Purpose of the Study:

• To evaluate the potential of a CAT for paranoia to reduce the number of presented items by half.
• To assess the accuracy and efficiency of CAT in measuring paranoia across diverse populations.

Main Methods:

• Utilized item response theory (IRT) properties of the 10-item Revised Green et al Paranoid Thoughts Scale (Persecution).
• Conducted CAT simulations on four datasets: a UK adult survey (N=10,382), a psychosis clinical trial (N=319), an NHS male psychosis cohort (N=836), and a persecutory delusions trial (N=89).
• Employed the graded response model, concluding the test when the standard error of estimation fell below 0.3 or five items were answered.

Main Results:

• The CAT administered an average of 4.0-4.2 items per person across the datasets.
• Achieved high correlations between CAT scores and full-scale paranoia scores (0.87-0.95).
• Observed minimal systematic error (mean bias scores: -0.01 to -0.10), with least precision at the boundary of normal and elevated paranoia.

Conclusions:

• CAT can provide accurate paranoia estimates using less than half the items of the full scale across the paranoia continuum.
• Tailored testing via CAT appears effective for individuals with psychosis.
• CAT presents a viable method for implementing measurement-based care in psychosis services.