We had predicted based on prior research that changes in objective cognitive function in our Memory Disorders Clinic sample would correlate strongly with increased medical co-morbidity, depression, low SES and low education. Surprisingly, decreased objective memory function correlated strongly with only medical co-morbidity, partially with education and not at all with residential SES. Subjective memory complaints did correlate with depression but not at all with medical co-morbidity and inversely with residential SES.
Prior research has suggested that SES correlates strongly with cognitive functioning [8
]. It is possible in our study, given that we used residential SES as opposed to income levels as our measure of SES, that our sample size was not sufficiently large to detect differences. The weak link with education, however, is even more surprising given that low levels of education have been shown to be significantly associated with poor cognitive functioning. Education did correlate mildly with impaired objective memory performance as measured by the MMSE but did not correlate with a more specific measure of cognitive function, the BNA. Given that the sample looked at is quite elderly, it is possible that years of education is not a good proxy for level of intellectual function. The inverse relationship between subjective memory complaints and SES does suggest that patients with low SES tend to complain more about their memory function, the opposite trend from what would be observed using the theory of cognitive reserve.
Our study suggested a strong correlation between medical co-morbidity and objective memory function but a poor correlation with subjective memory function in our Memory Clinic sample. Thus, patients evaluated in our clinic with significant medical issues are clearly at risk for cognitive impairment. Subjective memory complaints are a measure of awareness of brain dysfunction. Awareness of such brain dysfunction is important as it may lead to the adoption of compensatory strategies, including use of calendars, aids, etc, which may enhance brain function. It may also make patients more likely to seek help or perhaps go on medications (such as cognitive enhancers) that may preserve their brain function. Previous studies have examined subjective memory complaints and have found a relatively poor correlation between complaints and actual cognitive performance [16
]. This study contributes further to our understanding of the link between subjective memory complaints and objective memory function by demonstrating that this lack of association persists even in the context of significant medical illness.
It is interesting to speculate as to why subjective memory complaints would be low in patients with elevated medical co-morbidity, especially given the strong correlation with reduced cognitive performance. One potential explanation is that such patients may be so preoccupied with issues regarding their physical health that they have little awareness of cognitive issues. In addition, it is possible that it may have something to do with the theory of cognitive reserve [26
]. According to this theory, it is possible that patients with high medical co-morbidity have low cognitive reserve to begin with and therefore experience cognitive changes in a more gradual way, resulting in poor awareness. Finally, it is possible that the relationship between medical co-morbidity and subjective memory complaints may be mediated by some other correlate, such as low SES or limited education. Our study in fact showed only a partial correlation with education and a negative correlation with SES, suggesting this is not likely the mechanism.
Interestingly approximately half of the patients in the clinic studied had a diagnosis of dementia while half did not. There is some evidence to suggest that patients with more severe cognitive impairment may be more inclined to underreport their cognitive impairment while the opposite is true of patients with less severe cognitive impairment. Thus, it is possible that diagnosis alone may explain to some degree the observed associations. Whatever the mechanism, the implications of these research findings are apparent. Patients evaluated in our Memory Disorders Clinic who are in a sense more impaired medically are less likely to be aware of their cognitive deficits, making them more vulnerable. Such patients may be less likely to seek help or develop compensatory strategies, resulting in elevated risk. At a clinical level, it may mean that such patient with high levels of medical co-morbidity need to be assessed differently, perhaps screened more rigorously for the presence of cognitive deficits, or the physician may need to make a greater effort to contact collateral sources. Medication compliance is an area where physicians need to be especially vigilant, as poor cognition may lead to poor compliance, resulting in a greater burden of medical illness.
This study has a number of limitations which should be discussed. First, all of the measurement scales used have some limitations. Specifically, the PAOF has not been validated in the elderly and the BNA is a compilation of tests that has not been correlated with other neuropsychological tests. However, the PAOF has been used in other cognitively impaired populations and the BNA has shown to be superior to the MMSE in detecting dementia. The CIRS is a good attempt to quantify medical co-morbidity but as with many scales may not capture the full impact of medical illness. Also, in our sample we looked at all patients and did not stratify based on diagnosis (cognitively normal, MCI, dementia) as we were more interested in looking at the effects of medical co-morbidity on overall subjective/objective memory function as opposed to diagnosis. Furthermore, the design of the study was cross sectional as opposed to longitudinal, the sample size was relatively small and patients were recruited from a Memory Disorders clinic as opposed to the community. Thus, the findings are not generalizable to other clinical settings such as the community, family practice setting, etc. Finally, while more sophisticated statistical methodology could be used to analyze the relationship between mood, medical co-morbidity and cognition we feel such analyses goes beyond the current scope of our paper given our sample size.