A comparison of the PANSS pentagonal and Van Der Gaag 5-factor models for assessing change over time.

The positive and negative syndrome scale (PANSS) pentagonal (PM) and Van der Gaag (VDG) 5-factor models were compared on a range of change-over-time statistical indicators. PANSS data from a randomized, controlled trial for 108 adults diagnosed with schizophrenia were re-analyzed to calculate five factor scores for each model. Random effects regression was used to determine their relative performance in modeling change-over-time, determining covariance structure, and achieving goodness of fit. Performance of the 10 factors in estimating change-over-time was similar, as were significant covariates for the factor dyads: age, gender, chronicity/acuity, and anticholinergic medication. The PM model factors demonstrated the best "goodness of model fit" to the data on all five of the factor dyads. Correlation analyses indicated significantly high, positive correlations between the five factor dyads: Positive and Negative factors for the PM and VDG models, the PM Activation and VDG Excited factors, the PM Dysphoric Mood and the VDG emotional distress factors, and the PM Autistic Preoccupation and the VDG Disorganized factors, although the PANSS items on each factor differed somewhat. These results extend previous findings and indicate several important performance differences between the PM and VDG models in their parsimonious representation of the syndromes of schizophrenia and estimating change-over-time.