Last Data Update: September 2, 2026
Last Site Functionality Update: September 11, 2026
Public Use Files (PUF) Available: 2020, 2021, 2022, Pooled 2020-2022 For access, use the data request form.
New Data: New 2022 Medicare data include all individuals alive and enrolled as of January 1, 2022, with dementia prevalence determined using diagnoses from 2020-2022 (see Case Definitions). Users can select the desired data year using the "Year" dropdown menu.
New Secondary Outcome: Available beginning with the 2022 data, we have added emergency department visits per 1,000 beneficiaries diagnosed with dementia. This measure counts all-cause emergency department use, including visits that result in a hospital admission. Please see Outcome Measure Definitions for more information.
New Data:
Notes on the pooled data: The pooled data file combines three years of Medicare data with the intent of reducing suppression needed to protect privacy of small cells.
In the Pooled data, and cell count is the rounded (see changes to disclosure protection in the next column) count of the sum of annual counts for 2020, 2021, and 2022. A beneficiary who was alive and enrolled in all three years is counted three times. Because of this, pooled counts are larger than annual counts and do not represent a count of unique people.
The primary benefit of the pooled data is to increase the visibility of cells that need to be suppressed to protect beneficiary privacy in the annual data files.
The pooled file covers beneficiaries ages 65 and older and the core dementia measures (beneficiaries, prevalence, incidence, mortality, and COVID-19).
Rates from these files should be created by dividing the full numerator by the full denominator to reflect the three year counts.
New Disclosure Protection Method: Beginning with the 2022 and pooled files, we have replaced our legacy data suppression approach with a certified disclosure limitation system which displays more data and provides stronger privacy protection.
In the new approach, counts of 11 or larger are displayed as rounded values. The prior approach did not use rounding but required more cells to be suppressed. Like in the legacy system, counts <11 are suppressed. The new disclosure protection method allows us to present more data cells while still assuring that no cell with a count of less than 11 can be recreated from other cells in the data.
A small number of cells are interval coded using the form <X (where X is a number value, for example 100). This means that the true count for these cells is below X. This is necessary because certain cells pose a higher risk in exposing cells with a count less than 11 than other cells and need greater rounding.
Every released file is checked by an automated audit that verifies that no suppressed value can be recovered from published values. Each file is accompanied by a certificate documenting the protection standard.
Notes on using the data.
Last Data Update: November 19, 2025
Last Site Functionality Update: November 19, 2025
Public Use Files (PUF) Available: 2020, 2021 For access, please email dementiadatahub@norc.org.
New Data: Updated 2021 Medicare data include all individuals alive and enrolled as of January 1, 2021, with dementia prevalence determined using diagnoses from 2019–2021 (see Case Definitions). Users can select the desired data year using the “Year” dropdown menu.
New Case Definition: Highly Likely + Likely Dementia combines all individuals meeting either of these two case definitions. This category is available for the 2020 and 2021 data. Users can select the desired case definition using the “Dementia Category” dropdown menu.
New Secondary Outcomes Among Prevalent Cases: Available for 2021 data only, we have added the following measures among individuals with prevalent dementia:
Please see Outcome Measure Definitions for more information.
Increased Availability of Age 65+ Data, Reduced Availability of Age <65 Data: We have changed our system’s data suppression implementation strategy to increase the availability of data for the combined age category of all people ages 65 and older. The system uses data suppression of any cells with a count below 11, or any cell with a count below 11 that can be inferred from a combination of other cells. Because of this, data for all people ages 65+ must be suppressed whenever the cell size of the Age <65 group is suppressed. This led to a very large amount of data suppression for the age 65+ category. We adjusted our suppression strategy to enable us to show more data for the ages 65+ group, a primary outcome of interest. This has resulted in the following differences for 2021 data as compared to 2020:
