The distribution of BOMC total error scores in our population of veterans aged 60 and older is roughly equivalent to other studies (19
). It is noteworthy that 11.7% of this population of older adults referred for MH/SA assessment or started on a new antidepressant actually had BOMC scores consistent with dementia. Further, an additional quarter of the population (25.8%) made errors on the BOMC at a rate higher than 1 standard deviation above the average of their peers. While further evaluation would be needed to determine if these screening results were consistent with MCI, early dementia, or cognitive symptoms of their MH/SA conditions, the findings strongly suggest that cognitive functioning should be routinely evaluated when older adult mental health issues are evaluated in primary care. For example, it is possible that a significant portion of the PCI group are currently experiencing reversible causes of cognitive impairment resulting from treatable conditions such as substance abuse, polypharmacy, or the variety of medical conditions that can have an impact on cognition in older adults.
Our results suggest that approximately 11% of this group of older adults referred for screening of MH/SA symptoms is likely to have unrecognized dementia. This rate is lower than might be expected in a general primary care population where 50 to 75 percent of patients with dementia may not be recognized (28
). The lower rate may reflect some degree of screening related to the severity of the patient’s functional status. When considering referral to BHL, primary care providers may not opt to refer patients who seem less capable of managing responses to the telephone interview. As a result, the older patients who were not referred to BHL may include those with more severe cognitive impairments. Nevertheless, those patients screened by BHL with cognitive impairment would still likely benefit from family support and education, additional dementia-specific services such as day care and aide service, and possibly from cholinesterase inhibiting medications. The fact that this segment of the population likely has dementia and is being referred or treated for MH concerns in primary care strongly suggests the need to also provide non-pharmacological and pharmacological treatments for behavioral symptoms of dementia. Implementation of dementia-specific behavioral screening measures may improve the identification of this sub-group of patients. Without more routine and careful evaluation, multiple opportunities to improve patient care and quality of life are being lost.
In our study, increased age, non-Caucasian ethnicity and self-perceived inadequate finances were associated with increased scores on the indicator of cognitive impairment. Age is generally accepted as the single greatest risk factor for MCI and dementia across multiple studies (31
). Further, common risk factors typically include ethnicity and SES level (33
). Unfortunately, as our study is solely a naturalistic examination of clinical findings, it is not possible to determine whether our findings of increased BOMC scores in non-Caucasian and less financially secure veterans in primary care is a reflection of true impairment or a reflection of the too-common finding that cognitive assessment tools are subject to bias when used in minority populations or with those who have lower education levels as is common in lower SES populations. Further work is necessary to delineate whether the BOMC is subject to this nearly universal failing of all cognitive screening tools (35
One intriguing finding of our study was the association between the initiation of a new antidepressant and higher BOMC scores. This finding begs the question of causality. Do subtle symptoms of early cognitive impairment and dementia present a picture in the primary care clinic suggesting depression to the primary care provider? Alternatively, is our study picking up the well-documented association between cognitive function and depression in older adults (36
)? Depression has been noted to be both a risk factor for cognitive decline (37
) and an early manifestation of both Alzheimer’s disease (38
) and Vascular Dementia (39
). Further, Butters and colleagues have demonstrated that the relationships between treated and untreated late life depression and cognitive impairment are complex (40
) and that the diagnosis of depression in late life can be used to identify patients at high risk for dementia (11
). Similarly, this study suggests that the initiation of antidepressant treatments for older primary care patients can be a useful marker for the prediction of dementia. Therefore, it is important when considering the initiation of an antidepressant in older adult primary care patients, that cognitive screening is performed. More thorough evaluation for dementia should be considered whenever family members report functional decline out of proportion to symptoms of depression (29
Our study also found an association between the outcome assessment and BOMC scores. Veterans whose assessments indicated diagnoses of Major Depressive Disorder or symptoms of psychosis were more likely to have higher BOMC scores. Depression and psychosis are known to have an impact on cognitive function irrespective of age (41
), but cross-sectional studies of older adults frequently have found correlations between high levels of depressive symptoms and cognitive impairment in older adults (43
). Interestingly, Dufouil and colleagues found that high levels of depressive symptoms were not predictive of later cognitive decline when they used a longitudinal approach (47
). Conversely, depressive symptoms and psychosis are common in Alzheimer’s patients (48
) and both can be present in early stages (49
). Our results did not reveal the association between PTSD and dementia recently reported by others (13
). Nor did we find an association between Generalized Anxiety Disorder and cognitive impairment, a relationship which has been documented with increasing consistency in recent years (51
). The lack of association between anxiety and cognitive impairment in our study may be due to the use of only the BOMC screening in BHL procedures. The impact of late-life anxiety is more likely to be seen on neuropsychological tests than on tests of general cognitive functioning such as MMSE and BOMC. Further, many studies of anxiety and cognitive impairment use longitudinal approaches. Because our study is cross-sectional, some associations that become apparent over time could not have been detected. We also cannot determine whether the associations seen are the result of increased risk for onset of dementia or are the result of poorer cognitive function often seen in MDD and psychotic disorders. Nevertheless, evaluation of cognitive status should be conducted whenever a new diagnosis of these disorders is made. For example, even if dementia is ruled out and apparent cognitive decline represents the impact of depression on the older adult’s cognitive function and not an early stage of a degenerative dementia process, it is important to evaluate the impact of the patient’s cognitive functioning on his or her ability to adhere to treatments for depression and other illnesses.
Our study has several limitations. First, as mentioned, this study is a naturalistic, cross-sectional design. We are therefore unable to determine the temporal relationship of the symptoms being measured. It would be valuable to have follow-up information on the patients in order to evaluate the change in symptoms with treatment and whether depression symptoms, initiation of a new antidepressant medication or new onset psychotic symptoms are markers for the onset of dementia. Also because of the naturalistic nature of this study, we are unable to include Veterans with the highest level of cognitive impairment as they were excluded from BHL services. Second, the findings of our study cannot be applied to non-patient veterans or to non-veteran populations without consideration of how the average veteran primary care patient differs from patients in other settings. VA patients tend to be male, older, and more medically complex than non-VA primary care settings (14
). While our findings should not be uniformly applied to all general populations of primary care, it may be helpful, nevertheless to consider these findings when assessing individual primary care patients who are older and more medically complex. Finally, despite the robust relationship between cognitive status and education level (52
), we unfortunately did not have access to data regarding years of education. Therefore we could not include it as a covariate in our analyses.
Overall, our findings and the literature point to the importance of evaluating cognitive status in older adults when a new MH/SA diagnosis is made. This practice would improve recognition of cognitive impairment and dementia, both reversible and progressive, and allow for improved quality of care and life for many older adults.