Abstract
Background:
Dementia is often underdiagnosed and this problem is more common among some ethnoracial groups.
Objective:
To examine racial and ethnic disparities in the timeliness of receiving a clinical diagnosis of dementia.
Research design:
Prospective cohort study.
Subjects:
3,966 participants age ≥70 with probable dementia in the Health and Retirement Study, linked with their Medicare and Medicaid claims.
Measures:
We performed logistic regression to compare the likelihood of having a missed or delayed dementia diagnosis in claims by race/ethnicity. We analyzed dementia severity, measured by cognition and daily function, at the time of a dementia diagnosis documented in claims, and estimated average dementia diagnosis delay, by race/ethnicity.
Results:
A higher proportion of non-Hispanic Blacks and Hispanics had a missed/delayed clinical dementia diagnosis compared to non-Hispanic whites (46% and 54% vs. 41%, p<0.001). Fully adjusted logistic regression results suggested more frequent missed/delayed dementia diagnoses among non-Hispanic Blacks (OR=1.12; 95% CI: 0.91-1.38) and Hispanics (OR=1.58; 95% CI: 1.20-2.07). Non-Hispanic Blacks and Hispanics had poorer cognitive function and more functional limitations than non-Hispanic whites around the time of receiving a claims-based dementia diagnosis. The estimated mean diagnosis delay was 34.6 months for non-Hispanic Blacks and 43.8 months for Hispanics, compared to 31.2 months for non-Hispanic whites.
Conclusions:
Non-Hispanic Blacks and Hispanics may experience a missed or delayed diagnosis of dementia more often and have longer diagnosis delays. When diagnosed, non-Hispanic Blacks and Hispanics may have more advanced dementia. Public health efforts should prioritize racial and ethnic underrepresented communities when promoting early diagnosis of dementia.
Keywords: dementia, health disparities, diagnosis, cognitive health
INTRODUCTION
Although underdiagnosed dementia is a problem across all populations,1-6 it is more common among racial and ethnic underrepresented groups.7-12 For example, survey data from the Health and Retirements Study showed that fewer than half of older adults with dementia reported being told by a physician about their condition, and that more non-Hispanic Blacks and Hispanics with dementia were unaware of their diagnosis, despite higher dementia prevalence in these groups.13 Racial/ethnic underrepresented populations also have a higher risk of missed or delayed diagnosis of dementia, according to analyses comparing agreement between dementia case definitions classified with cognitive test results in survey data and dementia diagnosis codes in Medicare claims.14-16 Moreover, some evidence suggested these populations have more cognitive impairment at the time of their diagnosis.17,18
An assessment of missed or late diagnosis of dementia is crucial because these delays can exacerbate disability, reduce treatment effectiveness, impede timely care, and increase health care costs.19,20 The resulting negative health consequences among the affected individuals can in turn impose caregiver burden and reduce productivity.21 In addition, older adults with undiagnosed dementia may be more likely than those who receive a formal diagnosis to participate in activities that could cause them harm (e.g., driving).22 Because most of existing studies have small or unrepresentative samples,1,5,6,17,18 however, they cannot support conclusions about ethnoracial disparities in dementia diagnosis in the broader population. Further, prior research that have reported trends do not establish whether outcome differences across groups reflect differences in dementia diagnosis rates, later diagnoses in some groups compared to others, or both. Finally, the extent of diagnosis delay remains largely unquantified.
Extensive evidence has suggested ethnoracial differences in the incidence and prevalence of dementia,10,23 but these data do not address how dementia diagnoses differ by race and ethnicity in real-world clinical settings. For example, data from epidemiology cohorts indicate that non-Hispanic Blacks are about twice24 and Hispanics about 1.5 times25 as likely to have Alzheimer’s disease and related dementias, compared to non-Hispanic whites who are the same age. Estimates based on dementia diagnosis codes recorded in Medicare claims data suggest that the condition is more prevalent among non-Hispanic Blacks and Hispanics.26 These data, while useful, indicate disease burden disparities, not differences in clinical identification of dementia once symptoms arise. Moreover, claims-based estimates alone can understate dementia burden in general,1,2,5,6 and even more so among racial/ethnic underrepresented populations because these groups may be less likely than non-Hispanic whites to receive a clinical diagnosis.14-16
This study investigated whether there are ethnoracial differences in the timeliness of receiving a clinical diagnosis among older adults with cognitive and functional declines consistent with dementia. We hypothesized that non-Hispanic Blacks and Hispanics were more likely than non-Hispanic whites to experience a missed or delayed diagnosis of dementia. Our analysis extends the existing literature14-16 by quantifying this delay among different ethnoracial groups. Specifically, we examined dementia symptom severity at diagnosis and the delay between development of symptoms and receipt of a clinical diagnosis in administrative claims records. Our analyses leveraged nationally representative survey data with unique cognitive function measures, linked with Medicare and Medicaid claims, making our findings generalizable to the U.S. population.
METHODS
Data source and sample selection
Our analyses used U.S. national surveys from the 2000-2014 Health and Retirement Study (HRS), linked with corresponding Medicare and Medicaid claims. The HRS surveys more than 20,000 Americans over age 50 every two years, collecting a broad range of economic, health, and psychosocial information. The longitudinal nature of the HRS has made it possible for us to examine diagnosis delays (described in the next section). The dataset links claims data to a population-based survey with unique cognitive and functional measures, allowing us to assess dementia severity at diagnosis. The HRS oversamples non-white populations, making it well-suited for our investigation of racial/ethnic health inequalities.
We used a statistical model developed by Hurd and colleagues27 to identify participants with probable dementia. The model estimated the probability of having dementia among HRS respondents aged ≥70, based on their age, gender, education, cognition, and daily functioning. Our analysis used a model to estimate dementia status because the HRS dataset does not contain a direct measure of dementia. Hurd’s dementia prediction model has been well validated, with details described elsewhere.27,28 Briefly, an ordered probit model predicts the likelihood of “dementia,” “cognitive impairment no dementia (CIND),” or “normal.” Following Hurd’s approach, we classified an HRS participant as having “dementia” if their predicted probability of dementia exceeded their summed probability of CIND or normal. The predicted dementia status then served as the “gold standard” (i.e., probable dementia vs. no dementia). Tests for within sample fit of a subset of HRS respondents with known cognitive status, namely the Aging, Demographics and Memory Study,29,30 showed that our recreated Hurd model demonstrated good predictive power to discriminate dementia cases (sensitivity: 237/304=78.0%; specificity: 469/540=86.9%); overall, the model correctly classified 83.6% (706/844) of cases.13 These performance metrics track closely with those reported by Hurd et al.27 Details of our implementation and source code of the dementia prediction model are published elsewhere.13 The model identified 4,065 unique HRS participants aged 70 years or older with probable dementia (Appendix Figure 1, Box 2).
We linked HRS survey information for participants with probable dementia with their corresponding Medicare and Medicaid claims using unique identifiers provided by the HRS (n=3,966, Appendix Figure 1, Box 3). Study participants without linked Medicare or Medicaid identifiers (n=98) or actual claims records (n=1) were excluded. The Centers for Medicare and Medicaid (CMS) claims recorded information such as ICD-9/ICD-10 diagnosis codes, date of service, use of health care services and prescription drugs, and reimbursement paid by Medicare and Medicaid, among other details. We used the most recent CMS claims data available at the time of our analysis. Specifically, we used 2000-2015 Medicare claims, including Part A (i.e., inpatient, skilled nursing facility, hospice, and home health) and Part B (i.e., physician visits, outpatient care, and durable medical equipment) files. The HRS-linked Medicare claims were available for fee-for-service beneficiaries who provided consent for the linkage or who are now deceased (linkage rate 98% in our sample). For HRS participants covered by Medicaid, which disproportionately consist of non-Hispanic Blacks and Hispanics, we linked corresponding Medicaid claims, namely the Medicaid Analytic Extracts and Summary Files (MAX), for years 2000-2012. Medicaid claims record services for inpatient, outpatient, long term care, home health, physician visits, and prescription drugs of the beneficiaries.
Outcome measures
This study measured three outcomes. First, we examined missed or delayed dementia diagnosis in Medicare and Medicaid claims. We calculated the proportion of study participants not receiving a coded diagnosis of dementia in their CMS claims (Appendix Table 1)15,31,32 by the time the statistical model classified them as having dementia. Per Hurd’s model specification, predicted dementia status referred to the time period one year after the HRS interview.27 For example, for a 2002 HRS respondent, the model would use the person’s responses to the 2000 and 2002 HRS interviews to estimate whether they had dementia in 2003. We designated an individual as having a “missed or delayed diagnosis” if, for example, our implementation of the Hurd algorithm classified them as having dementia in 2003, but their claims showed no dementia diagnosis on or before December 31, 2003. In other words, we allowed for a lag until the start of the next HRS wave for a dementia diagnosis code to appear in an individual’s claims records before assigning a missed/delayed diagnosis status.
Second, we assessed dementia severity at diagnosis, measured by 1) cognitive function, i.e., the Telephone Interview for Cognitive Status (TICS) scores for self-respondents and the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) scores for respondents represented by a proxy, and 2) functional status, i.e., number of activities of daily living (ADL) and instrumental activities of daily living (IADL] limitations. In the HRS surveys, ADLs include getting dressed, walking across the room, bathing oneself, eating, getting in and out of bed, and using the toilet; IADLs include preparing meals, shopping for groceries, using the telephone, taking medication, and managing one’s money. We measured symptom severity using data from the closest HRS wave prior to or during the year of the person’s first claims-based dementia diagnosis.
Our third outcome measured the delay between our estimate of when dementia symptoms developed and receipt of a claims-based diagnosis of dementia. Because the predicted dementia status only indicated the year, we assigned each participant a random date within that year for the purpose of calculating the number of days between exhibiting symptoms and receiving a coded diagnosis in claims.
Analysis
Missed or delayed dementia diagnosis:
The sample for this analysis included all study participants with probable dementia and linked CMS claims (Appendix Figure 1, Box 3). The descriptive analysis calculated the proportion of dementia participants with a missed or delayed diagnosis in claims records. A logistic regression compared the likelihood of a missed or delayed dementia diagnosis in claims by race and ethnicity. Covariates in the parsimonious model included age, gender, and year of dementia prediction; the expanded model further adjusted for cognition, functional limitations, comorbidities, nursing home status, Medicare-Medicaid dual eligibility, and education. This analysis adjusted for HRS sampling weights from the year of dementia prediction. We conducted a sensitivity analysis that excluded nursing residents (n=896) because their likelihood of getting a dementia diagnosis may differ from that of community-dwelling individuals.
Severity of dementia at diagnosis:
The sample for this analysis included all study participants who received a claims-based diagnosis of dementia (Appendix Figure 1, Box 4). We calculated mean cognitive scores (i.e., TICS for self-respondents and IQCODE for proxy-respondents) and average numbers of ADL and IADL limitations around the time the person received a first coded dementia diagnosis in claims, by race/ethnicity. This analysis adjusted for HRS sampling weights from the year of dementia diagnosis.
Dementia diagnosis delay:
The sample for this analysis included all study participants who had no coded diagnosis of dementia in claims by the end of the year during which the prediction model first classified them as having dementia, but who received a claims-based diagnosis in a later year (Appendix Figure 1, Box 6). We estimated average diagnosis delay using a generalized linear model with a negative binomial distribution and log link function, adjusting for the same set of patient characteristics described above. This analysis adjusted for HRS sampling weights from the year of dementia prediction.
We used Stata Statistical Software Release 16 (College Station, TX) for all data analyses. The Tufts Medical Center/Tufts University Health Sciences Institutional Review Board approved this study.
RESULTS
Sample characteristics
Our sample included 3,966 HRS participants age ≥70 with probable dementia who had linked CMS claims. Mean age was 84 years and 67% were female, 39% used a proxy respondent, 23% resided in a nursing home, and 24% were dually eligible for Medicare and Medicaid (Table 1; all statistics are weighted). The sample comprised 81% non-Hispanic whites, 12% non-Hispanic Blacks, and 7% Hispanics. Non-Hispanic whites were older and more likely to reside in a nursing home. The proportion of individuals with a proxy respondent, the comorbidity burden, and level of functional impairment were similar across ethnoracial groups. Among self-respondents, non-Hispanic whites had higher cognitive scores on average, whereas among participants represented by a proxy, cognitive scores did not differ by race/ethnicity.
Table 1.
Sample characteristics of the HRS weighted sample with dementia, by race and ethnicity
All (n=3,966) |
Non-Hispanic white (n=2,908) |
Non-Hispanic black (n=687) |
Hispanic (n=371) |
||
---|---|---|---|---|---|
Weighted % | 100.0% | 80.8% | 11.9% | 7.3% | p value |
Mean age (SE) | 84.4 (0.1) | 84.9 (0.1) | 82.5 (0.3) | 82.3 (0.3) | <0.01 |
Age category | <0.01 | ||||
70-74 | 7.1% | 5.8% | 12.6% | 12.1% | |
75-79 | 14.0% | 12.9% | 19.1% | 18.6% | |
80-84 | 28.5% | 27.5% | 30.9% | 35.4% | |
85+ | 50.4% | 53.8% | 37.4% | 33.9% | |
Female | 66.8% | 67.1% | 67.1% | 63.0% | 0.34 |
Proxy respondent | 39.0% | 39.4% | 36.2% | 38.6% | 0.38 |
Education | <0.01 | ||||
Less than high school | 42.6% | 35.3% | 66.7% | 84.6% | |
High school | 44.2% | 49.6% | 26.6% | 13.1% | |
More than high school | 13.2% | 15.1% | 6.6% | 2.3% | |
Mean TICS score (SE)* | 9.84 (0.08) | 10.16 (0.09) | 8.35 (0.18) | 8.93 (0.19) | <0.01 |
Mean IQCODE score (SE) † | 3.48 (0.03) | 3.49 (0.03) | 3.47 (0.08) | 3.43 (0.09) | 0.52 |
Mean number of ADL limitations (SE) ‡ | 1.68 (0.03) | 1.66 (0.04) | 1.72 (0.08) | 1.79 (0.12) | 0.24 |
Mean number of IADL limitations (SE) § | 2.01 (0.03) | 2.01 (0.03) | 1.94 (0.07) | 2.04 (0.10) | 0.83 |
Mean number of comorbidities (SE)# | 2.90 (0.03) | 2.90 (0.03) | 2.96 (0.07) | 2.84 (0.09) | 0.85 |
Nursing home resident | 22.5% | 24.6% | 15.7% | 10.7% | <0.01 |
Medicare-Medicaid dually eligible Status | 24.1% | 20.4% | 35.6% | 46.5% | <0.01 |
TICS: Telephone Interview for Cognitive Status. Only for self-reported respondents. Scale from 0-33; higher scores indicate higher cognitive function.
IQCODE: Informant Questionnaire on Cognitive Decline in the Elderly. Only for participants who used a proxy respondent. Scale from 0-5; lower scores indicate higher cognitive function.
ADL: Activities of Daily Living. Numbers are the reported number of activities (6 total) participants have difficulty performing; lower scores indicate higher functional ability.
IADL: Instrumental Activities of Daily Living. Numbers are the reported number of activities (5 total) participants have difficulty performing; lower scores indicate higher functional ability.
Comorbidity count ranges from 0 to 8, including the following conditions based on HRS survey reports: high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, psychiatric problems, and arthritis.
Disparities in missed or delayed dementia diagnosis
Overall, 42% of older adults with probable dementia had a missed or delayed diagnosis in their CMS claims. This proportion was higher among non-Hispanic Blacks and Hispanics, compared to non-Hispanic whites (46% and 54% vs. 41%; p<0.01). Appendix Table 2 summarizes characteristics of subjects with and without a missed/delayed diagnosis. We found overall that 24% of older adults with probable dementia never received a claims-based diagnosis throughout the study period (Appendix Table 3). This proportion was substantially higher among Hispanics (32%), even though the mortality was similar across groups.
In adjusted analyses, the parsimonious model suggested a higher likelihood of missed/delayed dementia diagnosis for non-Hispanic Blacks (OR=1.27; 95% CI: 1.05-1.54) and Hispanics (OR=1.84; 95% CI: 1.43-2.36) (Model 1 in Table 2). The fully adjusted model showed similar trends (non-Hispanic Blacks: OR=1.12; 95% CI: 0.91-1.38; Hispanics: OR=1.58; 95% CI: 1.20-2.07) (Model 2 in Table 2). Female sex, older age, worse cognitive function, more IADL limitations, residing in a nursing home, Medicare-Medicaid dual eligibility and higher education levels were associated with lower odds of having a missed or delayed diagnosis of dementia in claims. Sensitivity analysis that excluded nursing home residents yielded substantively similar results (and thus not presented).
Table 2.
Odds ratios of having a missed or delayed diagnosis of dementia among individuals with probable dementia
Model 1 | Model 2 | |||
---|---|---|---|---|
Odds ratio | 95% CI | Odds ratio | 95% CI | |
Race/Ethnicity | ||||
Non-Hispanic white | Reference | - | Reference | - |
Non-Hispanic black | 1.27 | (1.05, 1.54) | 1.12 | (0.91, 1.38) |
Hispanic | 1.84 | (1.43, 2.36) | 1.58 | (1.20, 2.07) |
Female vs. male | 0.73 | (0.63, 0.85) | 0.77 | (0.66, 0.90) |
Age category | ||||
70-74 | 0.72 | (0.55, 0.95) | 1.10 | (0.82, 1.49) |
75-79 | 0.66 | (0.53, 0.82) | 0.88 | (0.70, 1.11) |
80-84 | 0.81 | (0.68, 0.95) | 0.89 | (0.74, 1.06) |
85+ | Reference | - | Reference | - |
Dementia prediction year | 0.93 | (0.90, 0.96) | 0.88 | (0.85, 0.91) |
Cognitive impairment* | - | - | 0.80 | (0.76, 0.84) |
Number of ADL limitations† | - | - | 0.99 | (0.95, 1.04) |
Number of IADL limitations‡ | - | - | 0.80 | (0.76, 0.85) |
Number of comorbidities# | - | - | 0.98 | (0.93, 1.03) |
Nursing home residents | - | - | 0.39 | (0.32, 0.49) |
Medicare-Medicaid dually eligible | - | - | 0.75 | (0.63, 0.90) |
Education | ||||
Less than high school | Reference | - | ||
High school | 0.74 | (0.62, 0.87) | ||
More than high school | 0.56 | (0.43, 0.73) |
Cognitive impairment is on a 0-10 scale by using normalized TICS scores and IQCODE scores. 0: No impairment; 10: High impairment.
Activities of Daily Living. Numbers are the reported number of activities (6 total) participants have difficulty performing; Lower scores indicate higher functional ability.
Instrumental Activities of Daily Living. Numbers are the reported number of activities (5 total) participants have difficulty performing; Lower scores indicate higher functional ability.
Comorbidity count ranges from 0 to 8, including the following conditions based on HRS survey reports: high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, psychiatric problems, and arthritis.
Disparities in dementia severity at diagnosis
Over the study period, 76% of study participants had a dementia diagnosis documented in their CMS claims at some point (Appendix Figure 1, Box 4). At the time of diagnosis, non-Hispanic Black and Hispanic self-respondents had poorer cognitive function, compared to their non-Hispanic white peers, whereas respondents represented by a proxy had similar cognitive function (Table 3). Non-Hispanic Blacks and Hispanics had modestly more functional limitations at diagnosis.
Table 3.
Dementia severity at the time of dementia diagnosis in claims data, by race and ethnicity
Mean scores (SD) | ||||||||
---|---|---|---|---|---|---|---|---|
TICS* (n=2,010) |
P value |
IQCODE† (n=1,001) |
P value |
ADL‡ (n=3,011) |
P value |
IADL§ (n=3,011) |
P value |
|
Race/Ethnicity | <0.01 | 0.08 | <0.01 | 0.02 | ||||
Non-Hispanic white | 13.04 (0.15) | 3.52 (0.05) | 1.36 (0.04) | 1.64 (0.04) | ||||
Non-Hispanic black | 10.16 (0.31) | 3.41 (0.11) | 1.50 (0.09) | 1.82 (0.09) | ||||
Hispanic | 11.45 (0.41) | 3.31 (0.11) | 1.79 (0.14) | 1.87 (0.13) |
TICS: Telephone Interview for Cognitive Status. Only for self respondents. Scale from 0-33; higher scores indicate higher cognitive function.
IQCODE: Informant Questionnaire on Cognitive Decline in the Elderly. Only for participants who had a proxy respondent. Scale from 0-5; Lower scores indicate higher cognitive function.
ADL: Activities of Daily Living. Numbers are the reported number of activities (6 total) participants have difficulty performing; lower scores indicate higher functional ability.
IADL: Instrumental Activities of Daily Living. Numbers are the reported number of activities (5 total) participants have difficulty performing; lower scores indicate higher functional ability.
Disparities in dementia diagnosis delay
We found that 58% of participants with probable dementia received a dementia diagnosis in their claims without delay, whereas 18% experienced a delayed diagnosis (Appendix Table 3). Among these subjects, the diagnosis delay averaged 32.8 months (median: 23.6 months) (Appendix Figure 2). The fully adjusted model indicated that the dementia diagnosis delay was 11% longer for non-Hispanic Blacks (IRR=1.11; 95% CI: 0.96-1.28) and 40% longer for Hispanics (IRR=1.40; 95% CI: 1.19-1.66), compared to non-Hispanic whites (Table 4, Model 2). Having more IADL limitations and residing in a nursing home were associated with shorter diagnosis delays. The diagnosis delay when experiencing a delay averaged 34.6 months for non-Hispanic Blacks and 43.8 months for Hispanics, compared to 31.2 months for non-Hispanic whites (Figure 1).
Table 4.
Incidence rate ratios of dementia diagnosis delay
Model 1 | Model 2 | |||
---|---|---|---|---|
IRR | 95% CI | IRR | 95% CI | |
Race/Ethnicity | ||||
Non-Hispanic white | Reference | - | Reference | - |
Non-Hispanic black | 1.09 | (0.95, 1.25) | 1.11 | (0.96, 1.28) |
Hispanic | 1.39 | (1.18, 1.63) | 1.40 | (1.19, 1.66) |
Female vs. Male | 1.21 | (1.08, 1.36) | 1.21 | (1.08, 1.36) |
Age Group | ||||
70-74 | 1.44 | (1.15, 1.82) | 1.50 | (1.19, 1.89) |
75-79 | 1.11 | (0.91, 1.36) | 1.17 | (0.95, 1.44) |
80-84 | 1.09 | (0.95, 1.25) | 1.09 | (0.96, 1.24) |
85+ | Reference | - | Reference | - |
Dementia prediction year | 0.92 | (0.90, 0.95) | 0.92 | (0.89, 0.95) |
Cognitive impairment* | 0.97 | (0.93, 1.01) | ||
Number of ADL limitations† | 1.02 | (0.98, 1.07) | ||
Number of IADL limitations‡ | 0.95 | (0.91, 1.00) | ||
Number of comorbidities# | 0.97 | (0.93, 1.01) | ||
Nursing home residents | 0.71 | (0.57, 0.90) | ||
Medicare-Medicaid dually eligible | 0.91 | (0.80, 1.03) | ||
Education | ||||
<High school | Reference | - | ||
High school | 1.01 | (0.89, 1.15) | ||
>High school | 0.91 | (0.71, 1.15) |
Cognitive impairment is on a 0-10 scale by using normalized TICS scores and IQCODE scores. 0: No impairment; 10: High impairment.
Activities of Daily Living. Numbers are the reported number of activities (6 total) participants have difficulty performing; Lower scores indicate higher functional ability.
Instrumental Activities of Daily Living. Numbers are the reported number of activities (5 total) participants have difficulty performing; Lower scores indicate higher functional ability.
Comorbidity count ranges from 0 to 8, including the following conditions based on HRS survey reports: high blood pressure, diabetes, cancer, lung disease, heart disease, stroke, psychiatric problems, and arthritis.
Figure 1. Estimated means of delay in receiving a claims-based diagnosis of dementia, by race and ethnicity.
*Average diagnosis delay was estimated using a generalized linear model with a negative binomial distribution and log link function. Covariates in the parsimonious model included age, gender, and year of dementia prediction; the expanded model further adjusted for cognition, functional limitations, comorbidities, nursing home status, Medicare-Medicaid dual eligibility and education.
Non-Hispanic White
Non-Hispanic Black
Hispanic
DISCUSSION
This study is the first to our knowledge to characterize the extent of dementia diagnosis delay by race and ethnicity in a nationally representative cohort. Our study found that, among older adults with cognitive and functional declines consistent with dementia, more than 40% had a late diagnosis or no diagnosis at all in their CMS claims. Non-Hispanic Blacks and Hispanics experienced a missed or delayed diagnosis of dementia more often than non-Hispanic whites, even after adjusting for differences in symptom severity. Although all ethnoracial groups experienced late diagnosis of dementia, the delay was more than three months longer for non-Hispanic Blacks and more than a year longer among Hispanics, compared to non-Hispanic whites. These findings suggest that non-Hispanic Blacks and Hispanics access dementia-focused care later in the disease process.
Numerous factors contribute to racial and ethnic health disparities in the U.S., such as health insurance coverage, proximity to a medical facility, mistrust of the system, racism, and a lack of diversity in the health care workforce.11,12,33 Although our data cannot pinpoint the exact cause of the observed dementia diagnosis disparities, our findings reflect several health inequality issues faced by this population. First, access to dementia care services differs by race and ethnicity. Studies have identified race and ethnicity as a social risk marker for inferior care quality in general.34 Although most seniors with probable dementia in our sample (>92%) had seen a doctor in the last two years (with an average of 13 visits), one analysis suggested that non-Hispanic Blacks and Hispanics had lower utilization of Medicare annual wellness visits,35 which include a short medical evaluation for cognitive impairment performed by a primary care practitioner. Underrepresented racial/ethnic groups, particularly Hispanics, are also less likely to receive follow-up care by dementia specialists (e.g., neurologists and psychiatrists) after receiving a clinical diagnosis.32
Second, the pervasive effects of lower socioeconomic status, which is more common among underrepresented racial and ethnic populations, can impede timely diagnosis. Individuals with lower education levels, for example, have poorer access to health care services in general.36 Indeed, a lower level of education was associated with worse cognitive and functional performance at diagnosis in our sample.
Third, survey data have shown that Blacks and Hispanics are more likely to perceive cognitive decline and Alzheimer’s disease as a normal experience of aging.37 However, a lack of knowledge about early signs of dementia, rather than culturally influenced beliefs, can deter cognitive assessment among some underrepresented racial/ethnic groups.38
Understanding which groups are most likely to experience dementia diagnosis delay or no diagnosis at all can assist policy interventions such as the Brain Health Initiative (a partnership between the Alzheimer’s Association and the Centers for Disease Control and Prevention) in targeting their efforts. Our findings suggest that dementia outreach programs should prioritize underrepresented racial/ethnic communities when promoting disease awareness and early detection. Moreover, although Medicare annual wellness visits provide an opportunity for early identification of cognitive decline in the primary care setting, use of these services is generally low.39 Survey data further show that fewer than one third of Medicare beneficiaries are aware that their annual wellness visits include a brief cognitive assessment.39 We fully recognize that early detection and diagnosis is challenging given the subtlety of typical signs of early cognitive or functional impairment. Still, much remains to be done to increase knowledge of the services available to older adults and to implement cognitive assessments during these visits. Early diagnosis is an essential first step to managing dementia because it helps optimize treatment and allows individuals, care partners, and families to plan.40-42 A timely diagnosis of dementia can also confer economic benefits for society, such as reducing excess health care expenditures incurred during the long periods leading up to a clinical diagnosis.43,44
Dementia policy interventions also should target health care providers serving underrepresented racial and ethnic communities. Although clinicians generally agree with the importance of diagnosing dementia early in the disease process,39,45 a number of diagnostic barriers have been reported, such as a lack of familiarity with diagnostic criteria, insufficient time to evaluate symptoms and provide follow-up care, belief that there is no treatment, and cultural resistance.46,47 The new Alzheimer’s Association Clinical Practice Guidelines,48 which provide specific recommendations and tools for the clinical evaluation of suspected Alzheimer’s disease and related disorders, can address some of these barriers. The medical community should implement provider training on culturally competent dementia care, recognizing each individual’s cultural diversity. Moreover, clinicians’ training should improve documentation of dementia diagnostic findings in health insurance claims. For the subjects in our dataset, it is possible that patient charts may have recognized, discussed, and recorded a dementia diagnosis, but nonetheless that diagnosis never made it into administrative claims records. Obtaining a specialist referral may depend on the provision of a coded diagnosis in claims records, depending on patient’s insurance.
Several limitations merit discussion. First, we used a prediction algorithm to identify our study sample, i.e., individuals with probable dementia. We used this approach because HRS surveys do not include a direct measure of dementia status and because there is no uniformly accepted case definition for dementia in observational studies. The Hurd model has been well-validated and outperforms several dementia prediction algorithms, such as those cutoff-based approaches that classify dementia based solely on composite cognitive and/or functional scores.49 Our recreated Hurd model has demonstrated good predictive performance in the ADAMS subsample, not only among non-Hispanic whites (sensitivity=75.1%; specificity=91.0%), but also among non-Hispanic Blacks (sensitivity=84.8%; specificity=76.7%) and Hispanics (sensitivity=91.3%; specificity=75.4%).13 Although misclassification probably does not bias against underrepresented racial/ethnic populations, the model’s predictive performance varies for the different groups. Future methodological work is needed to further enhance model performance comparability across subgroups.49
Second, because we do not know the specific date of an individual’s dementia symptom onset, we assigned a random prediction date. This approach likely introduced a similar degree of imprecision for all three ethnoracial groups that we investigated. We found no differences in the distribution of assigned dates across groups and HRS waves. In addition, we re-analyzed the data by replacing the prediction dates with another set of randomly generated dates. This exercise yielded substantively similar results. Third, HRS-linked Medicare claims were available for fee-for-service beneficiaries only. Future studies should examine dementia diagnosis disparities among individuals enrolled in Medicare Advantage plans, which may deliver health care more efficiently than traditional Medicare, especially for beneficiaries with Alzheimer’s disease and related dementias.50 Fourth, our data do not include HRS-linked claims before 2000, although this limitation does not introduce differential bias for non-Hispanic whites, non-Hispanic Blacks and Hispanics and thus is unlikely to alter our conclusion. The 2014 HRS was the last data wave with linked Medicare claims available to us at the time of our analysis. Future research should evaluate these trends when more current data are released. Finally, our sample had too few Asian American, Pacific Islander, American Indian, Alaska Native, or other race/ethnicity subjects for us to examine dementia diagnosis disparities in these groups. Nor was our sample size sufficient for us to analyze subcultures or various countries/regions of origin for non-Hispanic Blacks (e.g., American-born, Caribbean, or African-born) and Hispanics (e.g., Mexican, Puerto Rican, or central American in origin).
Our analysis has generated new, empirical, population-based evidence of racial/ethnic health inequalities in dementia diagnosis. Our findings emphasize the need to reduce dementia diagnosis delays, especially among non-Hispanic Black and Hispanic older adults. Public health efforts to promote early diagnosis of dementia should prioritize underrepresented racial/ethnic individuals and their health care providers. Further research must identify the individual, cultural, and health care system factors causing the delays our study has documented so that clinicians and researchers can develop measures to address the barriers.
Supplementary Material
Acknowledgments:
We would like to thank Norma Terrin, PhD, at the Biostatistics, Epidemiology, and Research Design Center at Tufts Medical Center for assistance with statistical design.
Funding source:
National Institutes of Health (R01AG060165) and Alzheimer’s Association
Footnotes
COI disclosure: The authors report no potential conflicts of interest.
REFERENCES
- 1.Taylor DH Jr., Ostbye T, Langa KM, Weir D, Plassman BL. The accuracy of medicare claims as an epidemiological tool: The case of dementia revisited. Journal of Alzheimer's disease : JAD 2009;17:807–815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lin PJ, Kaufer DI, Maciejewski ML, Ganguly R, Paul JE, Biddle AK. An examination of alzheimer's disease case definitions using medicare claims and survey data. Alzheimers Dement 2010;6:334–341. [DOI] [PubMed] [Google Scholar]
- 3.Savva GM, Arthur A. Who has undiagnosed dementia? A cross-sectional analysis of participants of the aging, demographics and memory study. Age and ageing 2015;44:642–647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lang L, Clifford A, Wei L, et al. Prevalence and determinants of undetected dementia in the community: A systematic literature review and a meta-analysis. BMJ open 2017;7:e011146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zhu CW, Ornstein KA, Cosentino S, Gu Y, Andrews H, Stern Y. Misidentification of dementia in medicare claims and related costs. J Am Geriatr Soc 2019;67:269–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lee E, Gatz M, Tseng C, et al. Evaluation of medicare claims data as a tool to identify dementia. Journal of Alzheimer's disease : JAD 2019;67:769–778. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fitten LJ, Ortiz F, Ponton M. Frequency of alzheimer's disease and other dementias in a community outreach sample of hispanics. J Am Geriatr Soc 2001;49:1301–1308. [DOI] [PubMed] [Google Scholar]
- 8.Clark PC, Kutner NG, Goldstein FC, et al. Impediments to timely diagnosis of alzheimer's disease in african americans. J Am Geriatr Soc 2005;53:2012–2017. [DOI] [PubMed] [Google Scholar]
- 9.Chin AL, Negash S, Hamilton R. Diversity and disparity in dementia: The impact of ethnoracial differences in alzheimer disease. Alzheimer Dis Assoc Disord 2011;25:187–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lines LM, Sherif NA, Wiener JM. Racial and ethnic disparities among individuals with alzheimer's disease in the united states: A literature review. RTI Press, Research Triangle Park, NC, 2014. [Google Scholar]
- 11.Babulal GM, Quiroz YT, Albensi BC, et al. Perspectives on ethnic and racial disparities in alzheimer's disease and related dementias: Update and areas of immediate need. Alzheimers Dement 2019;15:292–312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Brewster P, Barnes L, Haan M, et al. Progress and future challenges in aging and diversity research in the united states. Alzheimers Dement 2019;15:995–1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Lin PJ, Emerson J, Faul JD, et al. Racial and ethnic differences in knowledge about one's dementia status. J Am Geriatr Soc 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Amjad H, Roth DL, Sheehan OC, Lyketsos CG, Wolff JL, Samus QM. Underdiagnosis of dementia: An observational study of patterns in diagnosis and awareness in us older adults. Journal of general internal medicine 2018;33:1131–1138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Chen Y, Tysinger B, Crimmins E, Zissimopoulos JM. Analysis of dementia in the us population using medicare claims: Insights from linked survey and administrative claims data. Alzheimer's & dementia (New York, N Y) 2019;5:197–207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gianattasio KZ, Prather C, Glymour MM, Ciarleglio A, Power MC. Racial disparities and temporal trends in dementia misdiagnosis risk in the united states. Alzheimer's & dementia (New York, N Y) 2019;5:891–898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cooper C, Tandy AR, Balamurali TB, Livingston G. A systematic review and meta-analysis of ethnic differences in use of dementia treatment, care, and research. Am J Geriatr Psychiatry 2010;18:193–203. [DOI] [PubMed] [Google Scholar]
- 18.Livney MG, Clark CM, Karlawish JH, et al. Ethnoracial differences in the clinical characteristics of alzheimer's disease at initial presentation at an urban alzheimer's disease center. Am J Geriatr Psychiatry 2011;19:430–439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gilligan AM, Malone DC, Warholak TL, Armstrong EP. Racial and ethnic disparities in alzheimer's disease pharmacotherapy exposure: An analysis across four state medicaid populations. Am J Geriatr Pharmacother 2012;10:303–312. [DOI] [PubMed] [Google Scholar]
- 20.Gilligan AM, Malone DC, Warholak TL, Armstrong EP. Health disparities in cost of care in patients with alzheimer's disease: An analysis across 4 state medicaid populations. American journal of Alzheimer's disease and other dementias 2013;28:84–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Chiao CY, Wu HS, Hsiao CY. Caregiver burden for informal caregivers of patients with dementia: A systematic review. International nursing review 2015;62:340–350. [DOI] [PubMed] [Google Scholar]
- 22.Amjad H, Roth DL, Samus QM, Yasar S, Wolff JL. Potentially unsafe activities and living conditions of older adults with dementia. J Am Geriatr Soc 2016;64:1223–1232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chen C, Zissimopoulos JM. Racial and ethnic differences in trends in dementia prevalence and risk factors in the united states. Alzheimer's & dementia (New York, N Y) 2018;4:510–520. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Potter GG, Plassman BL, Burke JR, et al. Cognitive performance and informant reports in the diagnosis of cognitive impairment and dementia in african americans and whites. Alzheimers Dement 2009;5:445–453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Samper-Ternent R, Kuo YF, Ray LA, Ottenbacher KJ, Markides KS, Al Snih S. Prevalence of health conditions and predictors of mortality in oldest old mexican americans and non-hispanic whites. Journal of the American Medical Directors Association 2012;13:254–259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Matthews KA, Xu W, Gaglioti AH, et al. Racial and ethnic estimates of alzheimer's disease and related dementias in the united states (2015-2060) in adults aged >/=65 years. Alzheimers Dement 2019;15:17–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hurd MD, Martorell P, Delavande A, Mullen KJ, Langa KM. Monetary costs of dementia in the united states. New England Journal of Medicine 2013;368:1326–1334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hurd MD, Martorell P, Langa K. Future monetary costs of dementia in the united states under alternative dementia prevalence scenarios. J Popul Ageing 2015;8:101–112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Langa KM, Plassman BL, Wallace RB, et al. The aging, demographics, and memory study: Study design and methods. Neuroepidemiology 2005;25:181–191. [DOI] [PubMed] [Google Scholar]
- 30.Plassman BL, Langa KM, Fisher GG, et al. Prevalence of cognitive impairment without dementia in the united states. Ann Intern Med 2008;148:427–434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Centers for Medicare & Medicaid Services. Chronic conditions data warehouse condition categories. Volume 2020, 2020. [Google Scholar]
- 32.Drabo EF, Barthold D, Joyce G, Ferido P, Chang Chui H, Zissimopoulos J. Longitudinal analysis of dementia diagnosis and specialty care among racially diverse medicare beneficiaries. Alzheimers Dement 2019;15:1402–1411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Williams DR. Race, socioeconomic status, and health. The added effects of racism and discrimination. Annals of the New York Academy of Sciences 1999;896:173–188. [DOI] [PubMed] [Google Scholar]
- 34.Disparities in health care quality among racial and ethnic minority groups: Selected findings from the ahrq 2010 nhqr and nhdr. Agency for healthcare research and quality, rockville, md. Available at: Https://www.Ahrq.Gov/sites/default/files/wysiwyg/research/findings/nhqrdr/nhqrdr10/minority.Pdf. [Google Scholar]
- 35.Lind KE, Hildreth K, Lindrooth R, Crane LA, Morrato E, Perraillon MC. Ethnoracial disparities in medicare annual wellness visit utilization: Evidence from a nationally representative database. Med Care 2018;56:761–766. [DOI] [PubMed] [Google Scholar]
- 36.Adler NE, Newman K. Socioeconomic disparities in health: Pathways and policies. Health Aff (Millwood) 2002;21:60–76. [DOI] [PubMed] [Google Scholar]
- 37.Connell CM, Scott Roberts J, McLaughlin SJ. Public opinion about alzheimer disease among blacks, hispanics, and whites: Results from a national survey. Alzheimer Dis Assoc Disord 2007;21:232–240. [DOI] [PubMed] [Google Scholar]
- 38.Mahoney DF, Cloutterbuck J, Neary S, Zhan L. African american, chinese, and latino family caregivers' impressions of the onset and diagnosis of dementia: Cross-cultural similarities and differences. Gerontologist 2005;45:783–792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Alzheimer's Association. Alzheimer's disease facts and figures. Special report — alzheimer’s detection in the primary care setting: Connecting patients and physicians 2019.
- 40.Roberts JS, Karlawish JH, Uhlmann WR, Petersen RC, Green RC. Mild cognitive impairment in clinical care: A survey of american academy of neurology members. Neurology 2010;75:425–431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Morley JE, Morris JC, Berg-Weger M, et al. Brain health: The importance of recognizing cognitive impairment: An iagg consensus conference. Journal of the American Medical Directors Association 2015;16:731–739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Wimo A, Winblad B. Brain health: A primary health care viewpoint. Journal of the American Medical Directors Association 2015;16:720–721. [DOI] [PubMed] [Google Scholar]
- 43.Lin PJ, Zhong Y, Fillit HM, Chen E, Neumann PJ. Medicare expenditures of individuals with alzheimer's disease and related dementias or mild cognitive impairment before and after diagnosis. J Am Geriatr Soc 2016;64:1549–1557. [DOI] [PubMed] [Google Scholar]
- 44.Rand hrs 2000 fat file (v1c). In: RAND Center for the Study of Aging UoM, ed., V1C edn, 2008. [Google Scholar]
- 45.Dubois B, Padovani A, Scheltens P, Rossi A, Dell'Agnello G. Timely diagnosis for alzheimer's disease: A literature review on benefits and challenges. Journal of Alzheimer's disease : JAD 2016;49:617–631. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Bernstein A, Rogers KM, Possin KL, et al. Dementia assessment and management in primary care settings: A survey of current provider practices in the united states. BMC health services research 2019;19:919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Boise L, Camicioli R, Morgan DL, Rose JH, Congleton L. Diagnosing dementia: Perspectives of primary care physicians. Gerontologist 1999;39:457–464. [DOI] [PubMed] [Google Scholar]
- 48.Atri A, Knopman DS, Norman M, et al. Details of the new alzheimer's association best clinical practice guidelines for the evaluation of neurodegenerative cognitive behavioral syndromes, alzheimer's disease and dementias in the united states. Alzheimer's & Dementia: The Journal of the Alzheimer's Association 2018;14:P1559. [Google Scholar]
- 49.Gianattasio KZ, Wu Q, Glymour MM, Power MC. Comparison of methods for algorithmic classification of dementia status in the health and retirement study. Epidemiology (Cambridge, Mass) 2019;30:291–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Park S, White L, Fishman P, Larson EB, Coe NB. Health care utilization, care satisfaction, and health status for medicare advantage and traditional medicare beneficiaries with and without alzheimer disease and related dementias. JAMA network open 2020;3:e201809. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.