Converting negative symptom dimension scores across SANS and PANSS

Area of Science:

• Psychiatry and Clinical Psychology
• Psychometrics
• Statistical Modeling

Background:

• Factor analyses of SANS and PANSS reveal two negative symptom dimensions: expressive (EXP) and motivation/pleasure (MAP).
• Cross-scale comparability of EXP and MAP dimension scores between SANS and PANSS remains underexplored.
• Data harmonization methods are needed to integrate findings from studies using different negative symptom assessment scales.

Purpose of the Study:

• To compare equipercentile linking and linear regression for harmonizing SANS and PANSS negative symptom dimensions (EXP and MAP).
• To evaluate the accuracy and reliability of cross-scale conversions for EXP and MAP dimensions.
• To identify the most reliable conversion methods and specific factor scores for SANS and PANSS.

Main Methods:

• Employed equipercentile linking and linear regression for data harmonization.
• Utilized a Bayesian approach to assess conversion accuracy and reliability (Intraclass Correlation Coefficients - ICCs).
• Compared performance of different factor score conversions across SANS and PANSS for EXP and MAP dimensions.

Main Results:

• Linear regression generally outperformed equipercentile linking for cross-scale conversions.
• Most conversions exhibited moderate to excellent reliability (ICC > 0.6), with EXP dimensions showing higher reliability than MAP.
• Specific factor scores (Ahmed for SANS, Jang/Khan for PANSS) demonstrated superior performance when considering both dimensions.

Conclusions:

• Linear regression is a viable and slightly superior method for harmonizing SANS and PANSS negative symptom dimensions.
• Conversion formulae and reliability metrics provide a foundation for integrating research using SANS and PANSS.
• The study offers practical tools for researchers analyzing EXP and MAP dimensions across different clinical scales.