| dc.description.abstract |
Self-reporting of psychotic symptoms varies significantly between cultures and ethnic groups. Yet, limited validated screening instruments are available to capture such differences in the African continent.
Methodology
Among 9,059 individuals participating as controls in a multi-country case–control study of the genetic causes of psychosis, we evaluated the psychometric properties of the Psychosis Screening Questionnaire (PSQ). We applied multi-group confirmatory factor analysis and item response theory to assess item parameters.
Results
The overall positive endorsement of at least one item assessing psychotic symptoms on the PSQ was 9.7%, with variability among countries (Uganda 13.7%, South Africa 11%, Kenya 10.2%, and Ethiopia 2.8%). A unidimensional model demonstrated good fit for the PSQ (root mean square error of approximation = 0.009; comparative fit index = 0.997; and Tucker-Lewis Index = 0.995). Hypomania had the weakest association with single latent factor (standardized factor loading 0.62). Sequential multi-group confirmatory factor analysis demonstrated that PSQ items were measured in equivalent ways across the four countries. PSQ items gave more information at higher levels of psychosis, with hypomania giving the least discriminating information.
Limitations
Participants were recruited from general medical facilities, so findings may not be generalizable to the general population.
Conclusion
The PSQ demonstrated a unidimensional factor structure in these samples. Items were measured equivalently across all study settings, suggesting that differences in prevalence of psychotic symptoms between countries were less likely to represent measurement artifact. The PSQ is more reliable in screening for psychosis in individuals with higher degrees of psychotic experiences—hypomania excluded—and might decrease the false-positive rate from mild nonspecific psychotic experiences. |
en_US |