Memory tests in first-degree adult relatives of schizophrenic patients: a meta-analysis

Memory deficits have been clearly demonstrated in schizophrenic patients. However, studies of memory performances in their relatives compared to normal controls provide conflicting results. A meta-analysis was carried out to synthesize all the published data. Unlike previous meta-analyses, which were based on composite scores, we analyzed each memory test separately. This prevents theoretically questionable choices in grouping variables, leads to results with clearer implications for applied research (e.g. the best choice of a test according to its sensitivity) and is more productive in suggesting explanatory hypotheses.

We initially selected 77 potentially relevant articles, but only 19 met our inclusion criteria. These articles provided data on eight different tasks, from five different memory tests: four tests from the Wechsler Memory Scale (WMS) and the California Verbal Learning Test (CVLT). For each task, we assessed data homogeneity, identified the outliers if any and then estimated effect sizes and tested publication bias using funnel plots.

Adult relatives of schizophrenic patients were significantly impaired on most, but not all, tasks. The largest deficits were observed for the verbal paired associates test, the logical stories the digit span forward test and the digit span backward test. We found no significant differences in tasks of delayed recall, when deficits in immediate conditions (reflecting encoding) were taken into account.

Adult relatives of schizophrenic patients have wide but not severe memory impairments. The size of estimated effects suggests that encoding processes are impaired, whereas storage and retrieval processes are relatively unaffected.