Antisaccade performance in biological relatives of schizophrenia patients: a meta-analysis

Poor performance on the antisaccade (AS) task has been interpreted as a potential indicator of genetic liability that may enhance the power of linkage studies of a multidimensional phenotype for schizophrenia. Every study has replicated the finding of significantly worse performance in schizophrenia patients regardless of which specific antisaccade paradigm was employed. In some studies involving a standard version of the antisaccade task, relatives of schizophrenia patients made an increased number of errors, but in other studies that used this same paradigm, relatives of schizophrenia patients did not differ from controls. In this paper, we report the results of a meta-analysis on studies that used the standard antisaccade paradigm. The meta-analysis shows that those studies that reported large effect sizes and statistically significant differences between relatives of schizophrenia patients and controls used inclusion/exclusion criteria that were not symmetrical between the two groups, whereas those studies that reported small and nonsignificant differences between relatives of schizophrenia patients and controls used symmetrical inclusion/exclusion criteria. Specifically, studies that applied stricter psychopathology exclusion criteria to controls than to relatives of schizophrenia patients had larger effect sizes than studies that applied comparable exclusion criteria to both groups, suggesting that antisaccade performance is compromised by psychopathology in general rather than by schizophrenia per se. Since symmetrical inclusion/exclusion criteria between relatives of schizophrenia patients and controls are essential for a genetic analysis, and those studies that did apply symmetrical criteria had small effect sizes, the available data suggest that poor antisaccade performance is unlikely to be useful in identifying clinically unaffected carriers of genes for schizophrenia.