Is semantic memory impaired in schizophrenia? A systematic review and meta-analysis of 91 studies

Abstract

Introduction. Semantic memory impairments in schizophrenia have been reported across a wide range of neuropsychological tests. Set against a backdrop of fairly widespread cognitive impairments, it is difficult to know whether there is a primary, or secondary, impairment of semantic memory in schizophrenia. Also, whether there is a profile of differential impairment across the range of neuropsychological tests.

Methods. Employing a systematic search strategy, 91 papers were identified which have assessed participants with schizophrenia on a measure of semantic memory. A series of meta-analyses were then conducted which provided combined weighted means for performance on tasks of naming, word–picture matching, verbal fluency, priming, and categorisation.

Results. An uneven profile of impairment is reported with large effect sizes for tests of naming and verbal fluency, medium effect sizes for word–picture matching and association and small effect sizes for categorisation and priming tests.

Conclusions. This uneven profile supports the claim that a degradation of semantic knowledge may not be adequate in explaining the semantic memory impairment in schizophrenia. This conclusion is supported by the data which report a relationship between an executive dysfunction and poor priming and fluency performance particularly. The data support a link between Formal Thought Disorder and semantic memory impairments on tests of naming and verbal fluency but on other tests evidence is equivocal.

Keywords:

• Meta-analysis
• Schizophrenia
• Semantic memory
• Systematic review

Acknowledgements

The authors would like to thank Joerg Schultz, David Wellsted, and Sam Norton for their statistical support and advice.

Supplementary data (Appendix) published online alongside this article at http://www.psypress.com/cogneuropsychiatry

Notes

¹Negative values of Cohen's d occur when the correct score for the patients exceeds that of the controls, and vice versa for positive values. So the sign of d is arbitrary.

²Many studies reported Formal Thought Disorder based on scores of some formal rating scale (e.g., PANSS, BPRS). However, not all papers used these criteria, but since all studies had adopted DSM-IV criteria we have assumed that Formal Thought Disorder was one of the characteristic symptoms.