Background and Purpose--Poststroke memory dysfunction is a prerequisite for the diagnosis of poststroke dementia. This diagnosis is made within months after a stroke, apparently assuming a relatively stable course of the poststroke memory function. Clinical experience added to anecdotal evidence from the literature suggests that poststroke memory function may be reversible. The aim of the present study was to systematically review the available data on the time course of poststroke memory function in nondemented stroke survivors. In addition, we wanted to investigate the role of (pre-)stroke characteristics on poststroke memory function.

Methods--We performed systematic literature search of PubMed with the following medical subject heading terms: memory and stroke. The search strategy yielded 798 articles of which 65 fulfilled our inclusion criteria and went on to the data extraction stage.

Results--Five studies reported the prevalence of poststroke memory dysfunction at different poststroke intervals. The prevalence of poststroke memory dysfunction varied from 23% to 55% 3 months poststroke, which declined from 11% to 31% 1 year poststroke. Larger stroke volume, prestroke medial temporal lobe atrophy, and white matter lesions were related with decreased poststroke memory function.

Conclusions--Not all patients with poststroke memory dysfunction 3 months after a stroke had memory dysfunction 1 year poststroke. Consequently, not all criteria for the dementia diagnosis were fulfilled any more. This may indicate that poststroke dementia may be reversible in a substantial proportion of patients with stroke. Preferably, standardized reassessment of cognitive function should be performed in each patient diagnosed with poststroke dementia.