The value of informant versus individual’s complaints of memory impairment in early dementia

Subjects and methods.

The investigation was a prospective, blinded observational study. All procedures and measures for obtaining informed written consent for the participants were approved by the Human Studies Committee of Washington University School of Medicine. Written consent also was obtained from the informant. The recruitment and assessment procedures in the Washington University AD Research Center (ADRC), which include criteria for DAT and the psychometric tests, have been described.5 Dementia severity is staged with the Washington University Clinical Dementia Rating (CDR).6

Specific questions about memory complaints in the assessment protocol are as follows: for the participant, “Do you have any problems with your memory and thinking?” For the informant, “Do you believe the subject has problems with memory or thinking?” Participants answering “yes” to this question during the initial clinical assessment were classified for this study as self-reporting memory complaints; similarly, a “yes” response by informants was considered as reporting memory complaints for participants. To allow longitudinal analysis, only participants with at least 2 years of visits were included in the sample. At entry, 158 CDR 0, 165 CDR 0.5, and 159 CDR 1 subjects were available.

Participants with active major depression requiring medical attention were excluded from entry in the ADRC before 1996. Measures of “depression” used in this analysis, therefore, assess depressive features rather than frank depression, although occasionally a major depression may develop after enrollment. Two scaled measures of depressive manifestations were used: the Geriatric Depression Scale7 and Diagnostic and Statistical Manual of Mental Disorders (DSM) III criteria for a major depressive episode.8 The total number of depressive features were reported by both the subject and the informant, and were defined by the DSM.

Results.

Baseline features. Characteristics of the three CDR groups along with the percentage of memory complaints and the number of DSM-III depressive features as noted both by subject self-report and by informant are shown in the table. Informants were able to identify 86% of subjects who were assessed as cognitively intact at baseline, and 92% of those subjects judged by the clinician to have dementia.

Little or no correlation was seen between memory complaints reported by either the participant or the informant with scores on demographic, psychometric, or clinical variables other than depression for any CDR group. A positive correlation (r = 0.23, p < 0.05) was found between the self-report of memory problems and the clinician’s judgment whether the CDR 0 participants had insight into those problems, and negative correlations were seen for CDR = 0.5 (r = −0.39, p < 0.05) and CDR = 1.0 (r = −0.59, p < 0.05) participants. A negative correlation was seen with age and self-reported memory problems for CDR = 1 (r = −0.24, p < 0.05) and informant-reported memory problems for CDR = 0.5 (r = −0.32, p < 0.05). A significant correlation was found between informant-reported problems with memory and thinking in CDR = 0.5 subjects and the Sum of the Boxes score (r = 0.32, p < 0.05).

Self-reported memory complaints correlated with the total number of DSM-III depressive features reported by the subjects for CDR = 0 (r = 0.26, p < 0.05) and CDR = 1 (r = 0.22, p < 0.05) groups, and the Geriatric Depression scale for CDR = 1 subjects (r = 0.43, p < 0.05). Little or weak correlations were seen of memory complaints with informant-assessed depressive features in any CDR group.

To determine the reliability or stability of the memory question, the consistency of the response was examined by comparing answers from the informant at the next annual assessment. The kappa value was found to be 0.75, implying that the response was not only stable but also likely to represent a high level of test-retest reliability.

Predictive features.

Survival analysis curves for the progression of dementia were constructed for all CDR groups. The median time to progression for CDR = 0 participants, CDR = 0.5, and CDR = 1.0 participants occurred on average at 8.2 years, 5.0 years, and 2.1 years after baseline. Participants self-report of memory problems did not correlate with onset (CDR = 0) or progression (CDR 0.5 and 1) of dementia.

In contrast, informant-reported memory complaints at baseline predicted future onset of dementia for CDR 0 participants (p = 0.02; see figure). By 5 years after the initial assessment, 45% of normal subjects whose informant reported the subject as having memory problems at baseline, had progressed to a non-CDR 0 status. Education, socioeconomic scale, and gender were also entered as covariates; none were significant. No difference in age was noted between those CDR 0 subjects who were thought to have memory problems by their informants and those that did not, indicating that advancing age with its higher rate of onset of a new dementing illness was not a factor.

• Download figure
• Open in new tab
• Download powerpoint

Figure. Progression of nondemented subjects to a Clinical Dementia Rating (CDR) >0 based on the presence or absence of memory complaint.

Discussion.

Of nondemented elderly subjects, 44% self-reported problems with their memory. These self-reported problems did not correlate with poor psychometric performance, but did correlate mildly with the presence of depressive features. Self-reported memory problems did not predict the future onset of dementia, which is consistent with a recently reported study.9 Informants were successful in identifying normal subjects and those with dementia as judged by the clinician. Thus, the informants’ opinion whether the subject had memory and thinking problems was a powerful predictor of the subjects’ current cognitive status. In addition, an informant’s report of memory problems in a nondemented person was associated with an increased risk of future progression to dementia. This finding is consistent with results from a recent study, which also performed longitudinal assessments of cognition on nondemented subjects.10

Although the current study actively excluded depressed individuals, self-reported memory complaints in controls correlated with self-reported measures of depression. The correlation between memory complaints and depressive manifestations in nondemented individuals has been demonstrated. Depressive features such as “having had more trouble concentrating or less ability in thinking than is normal” or “marked difficulty making decisions” may overlap with the a subject’s report of problems with memory and thinking.

Informants have been shown to be a reliable and a valid source of information in determining the presence of DAT, even in the mild stages. The current study adds to a growing literature that informants may be identifying a population at risk for the development of DAT, even before clinically evident dementia is detected by standard assessment. These findings also indicate that assessment methods for control subjects in studies of cognitive aging cannot rely on the participants’ self-report of cognitive normality or impairment, because self-reported memory problems did not correlate with psychometric impairment. On the other hand, an informant’s report is a reliable guide to the participants’ current cognitive status.