Accurate Assessments of Healthcare Use Along the Course of Cognitive Decline

Abstract

The emotional and economic toll of Alzheimer's disease and related dementias on older adults and their families is substantial.1, 2 Individuals with dementia experience some of the largest out-of-pocket spending on medical care during their final 5 years of life;3 have high use of care facilities; undergo many transitions between hospitals, nursing homes, and home health care;4 and are at risk of using inappropriate over-the-counter and prescription medications.5 Studies of individuals with dementia have demonstrated a clear relationship between the severity of cognitive impairment and the use and costs of medical care.6, 7 For example, they have shown that a 1-point decrease in Mini-Mental State Examination (MMSE) score is associated with approximately $2,000 greater medical costs and that the difference in costs of an individual with an MMSE score of 18 to 23 and a score of more than 24 is $53,506 per year.6 It is unclear whether these results can be extrapolated to older adults with mild cognitive impairment (MCI). In this issue of the Journal of the American Geriatrics Society, Zhu and colleagues present data concerning the direct medical care costs and the formal and informal nonmedical care costs of older adults with MCI.8 Their work contributes to our understanding of how the presence of MCI, arguably the earliest stage of dementia for a substantial proportion of older adults, affects the use of medical and nonmedical care. In their analysis of data derived from 259 older adults with MCI and 107 older adults with normal cognition—all of them from 68 Alzheimer's Disease Cooperative Study (ADCS) sites—they found that self- and proxy-reported use of health-related resources were higher at baseline and over 3 years for older adults with MCI measured at baseline than for those with normal cognition measured at baseline. With the Resource Use Inventory (RUI) tool, the authors captured the use of medical care (e.g., hospitalizations and prescription drug use) and nonmedical care, primarily provided by unpaid and paid caregivers who assist with activities of daily living. Their findings add to the scant literature and to our understanding of the possible differences in medical and nonmedical needs of older adults as their cognition declines. Three important features of their study should be taken into account for future research in this area. First, the results of this study may not be generalizable to community-dwelling older adults with MCI. The study participants were recruited, in parallel, from clinical trials being conducted at ADCS sites across the country and had specific inclusion and exclusion criteria. The comorbidity information concerning participants was collected by asking them or their proxies about the presence of major problems in 18 general areas (e.g., cancer, neurological conditions, endocrinologic and metabolic conditions). No information about specific comorbid diagnoses that may lead to greater use of care was collected. In general, community-dwelling older adults with MCI tend to be older,9, 10 have more medical problems,11, 12 and rarely have their cognitive impairment identified.13 Second, Zhu and colleagues used the RUI to collect data retrospectively for the outcome variable of their study: health-related resource use. This required older adults or their proxies to recall and report any use of medical services and nonmedical care. The RUI was developed for use in the ADCS14 and has been validated in the ADCS population but does not capture known individual-level differences that may affect use in other populations, such as comorbid diagnoses, variations in use of care in particular geographic locations, access to services, and insurance status (Medicare supplements). An alternative approach to measuring healthcare use would be to identify individuals with MCI, or with a particular cognitive trajectory, and measure their medical care use according to Medicare claims data. Claims analysis would enable adjustment for known regional variation in the services that Medicare beneficiaries receive15 and for differences in comorbid conditions.12 Third, the criterion used in the study for diagnosing MCI is considered the gold standard and included data from informant reports and neuropsychological tests. Based on the results presented, it is unclear what effect, if any, the stability of cognitive diagnoses over the 3-year study period had on the results. The instability of MCI in clinical and epidemiological samples is well documented.16-19 For instance, a study that followed a clinical sample of older adults with MCI found that as many as 50% converted to dementia in 3 years.19 Conversely, studies of MCI in population-based samples have found that similar rates of older adults with MCI revert to normal cognition.17, 18 In the study of Zhu and colleagues, the finding that individuals with normal cognition and MCI had differences in resource use at baseline was not surprising given the uniform criteria to determine an MCI diagnosis. What is unclear is the effect that changes in cognitive status over time had on resource use. Individuals diagnosed with MCI at baseline may have worsened to dementia or may have improved to normal over the 3-year study period. Additionally, participants classified as cognitively normal at baseline may have declined to MCI or possibly even to dementia. Future studies of longitudinal resource use of older adults with normal cognition, MCI, and dementia should be careful to report information about cognitive status over time to ensure accurate calculations of health-related resource use in distinct or mutually exclusive cognitive groups. These future studies could be greatly enhanced by using the ADCS infrastructure to accurately identify a population with detailed longitudinal data on cognitive status and by matching and merging the information with Medicare claims data. The use of claims data could reduce the risk of recall bias, control for regional differences in patterns of healthcare utilization, and identify profiles of comorbidity within samples of individuals with MCI to measure use. Understanding these usage patterns as they relate to cognitive changes over time will help determine whether use is related to cognitive impairment or to comorbid conditions that the presence of cognitive impairment may or may not exacerbate; determine whether care is appropriate or inappropriate because of the lack of care coordination or poor transitions from one care facility to another; strengthen the ability of researchers to identify predictors of cognitive decline; and identify appropriate points along the disease course for interventions aimed at improving the quality and continuity of care, reducing the burden of caregivers, and reducing the level of unnecessary and unwanted care provided. Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the author and has determined that the author has no financial or any other kind of personal conflicts with this paper. Author Contributions: NF is responsible for the entire content of this editorial. Sponsor's Role: None.