1997; 154:609-615) N umerous studies have sought to evaluate the relationship between self-reported memory complaints and memory performance in elderly populations (1-14). The results have been mixed-some studies have demonstrated a weak association (1, 2, 6, 9, 10, 14), while others have not (5, 7, 12, 13)-and the validity of the self-reported memory problems themselves has been questioned. However, many of these studies have been clinic-based or cross-sectional in design and have not included subjects with cognitive impairment or comparison subjects without memory complaints. We investigated the relationship between memory complaints and cognitive performance among elderly participants of a population-based longitudinal study. In our study, in contrast with many others, not all subjects had memory complaints, some had cognitive impairment, and a proportion of subjects were evaluated twice.
METHOD
SubjectsSubjects for the current study comprised a subset of subjects in a registry that had been created for the study of aging and neurological conditions in North Manhattan (15). To establish the registry, nursing homes, home health care agencies, private practitioners, and hospital admission and discharge lists were canvassed to identify individuals with possible cognitive impairment. These individuals were then invited to complete a cognitive screening questionnaire (16) that had been modified from the Comprehensive Assessment and Referral Interview (17). Scores on this screening questionnaire could range from 0 to 15, with scores above 2 indicative of possible impairment.