For questions or to access public use data, contact us at:
The Dementia DataHub analyzes Medicare Fee-for-Service claims and Medicare Advantage encounter data to estimate and report the prevalence, incidence, mortality, rate of COVID-19 infections, and payments related to diagnosed dementia (including Alzheimer’s Disease and related dementias) at the national, state, and county level. In the future, the Dementia DataHub will expand to include additional data sources and outcome measures.
Among Medicare beneficiaries enrolled in January 2020, over 8.1 million had diagnostic or drug code evidence of either dementia or a less certain diagnosis that can sometimes be used as a diagnosis code for dementia. Of these beneficiaries, 4.3 million (7.2 percent of all Medicare beneficiaries) were classified by our system's case definitions as highly likely dementia cases, another 1.1 million (1.9 percent of all Medicare Beneficiaries) were classified by our system's case definitions to as likely to have dementia, and an additional 2.6 million (4.3 percent of all Medicare beneficiaries) were categorized by our system’s cases definitions as possibly having dementia. It’s important to note that beneficiaries in the possible category may have dementia or they may not, and additional research is necessary to understand all our system’s case definitions better. Beneficiaries in each of our system’s case definition categories were older, frailer, more likely to be female, use long-term care, and die than those beneficiaries who did not meet a case definition. These differences persisted after age-standardization.*
The Dementia DataHub provides data visualizations and public use files with information about the number and percentage of beneficiaries in each category that met each case definition by demographic and geographic strata. This system does not capture people without diagnostic or drug code evidence recorded by Medicare including people who may have been diagnosed outside of Medicare and people who have not received a diagnosis.
Research regarding differences by demographics and geography is ongoing. Research regarding the associations of area-level variables such as health system factors and social determinants of health with differences in dementia outcomes measured across states and counties may provide insights into why diagnosis rates of outcomes vary by place.
* Gianattasio, K. Z., Wachsmuth, J., Murphy, R., Hartzman, A., Montazer, J., Cutroneo, E., Wittenborn, J., Power, M. C., & Rein, D. B. (2024). Diagnostic code case definitions for Alzheimer’s disease and related dementias: A systematic review and application using Medicare data. JAMA Network Open. https://doi.org/10.1001/jamanetworkopen.2024.27610
The Dementia DataHub provides national, state, and county data measuring diagnosed dementia and dementia outcomes in Medicare data. Future updates to the system may include additional data sources. The project created and will validate and update case definitions to identify dementia using diagnosis and drug codes in billing claims and medical encounter data. The DataHub applies these case definitions to Medicare data to analyze and report national, state, and county measures of diagnosed dementia prevalence and incidence, and mortality, COVID-19 infections, and all-cause payments among people with prevalent dementia.
The Dementia DataHub is a joint effort led by NORC at the University of Chicago, with technical support from George Washington University’s Milken Institute School of Public Health and KPMG LLP. The DataHub is funded by the National Institutes on Aging through Grant R01-AG-075730. The contents are those of the author(s) and do not necessarily represent the official views of, nor an endorsement, by NIA/NIH, or the U.S. Government.
Dementia DataHub estimates should be considered provisional, pending revisions and adjudications of administrative data records, and possible future revisions or enhancements to the system’s dementia case definitions.
The Dementia DataHub developed case definitions based on diagnosis and prescription drug codes to identify people diagnosed with dementia. We reviewed existing case definitions and categorized them based on their frequency of inclusion in prior case definitions. Expert input was used to remove certain codes from the case definitions. We are conducting ongoing validation studies to assess the accuracy of these case definitions. Therefore, case definitions may be revised over time as new information becomes available.
We developed our case definitions by aggregating codes used by 20 different previously published algorithms and the CMS Chronic Condition Warehouse definition, sorting by frequency of inclusion in prior case definitions, reviewing individual codes with clinical experts, and organizing codes into those that are directly related to dementia, Alzheimer's disease and other related dementias, and others that may indicate dementia but do not specifically state dementia or Alzheimer's disease. We used other Medicare evidence to provide support for these definitions. We published the methods for developing our case definition in JAMA Network Open on September 3, 2024.
We used 100 percent of the 2018, 2019 and 2020 Medicare fee-for-service (FFS) inpatient, outpatient, carrier, Skilled Nursing Facility (SNF), Home Health Agency (HHA), and hospice claims; Medicare Advantage (MA) inpatient, outpatient, carrier, SNF, and HHA encounter data; and Medicare Part D Prescription Drug Event (PDE) data.
Our estimates include any Medicare beneficiaries with at least Part A (the premium-free Medicare benefit) enrollment who was alive and enrolled as of January 1, 2020. We excluded people with missing gender, and invalid US state or territory codes.
We applied the case definitions to included datasets using ICD-10 diagnosis codes and NDC drug codes in any position on the claim. We analyze and report outcomes by calendar year.
To ensure the privacy of individuals in Medicare, we suppressed any data cell value with a numerator or denominator count that was less than 11. We additionally suppressed related cells that could be used in combination to construct a cell count less than 11. Suppression is most extreme for results for the 0 to 64, and 65+ summary categories because wherever the 0-64 category is suppressed, the 65+ category must also be suppressed.
All data used in this site are available for independent analysis as a public use file. If you are interested in accessing the public use data, please email the project team at Dementiadatahub@norc.org.
When using the “Add relationship” feature in the map, the map produces the Spearman rank correlation coefficient which measures the strength and direction of association between the rank order of the unsuppressed values of the outcome selected and the rank order of the relationship variable. This provides a rough measure of an obvious association between two variables. Users should be cautioned that correlations are not the same as causality. Further seemingly uncorrelated variables may be associated with each other using methods that do not assume a linear relationship between the two variables, and/or methods that control for the confounding effect of other variables.
Prevalence indicated by both percentage and (count)
Average payments indicated by dollar amount and (number of applicable beneficiaries)
