NORC Launches County-Level Dementia Data Portal
This article is from our NORC Now newsletter. Subscribe today.
September 2024
The Dementia DataHub is a first-of-its-kind surveillance system for Alzheimer’s Disease and related dementias.
NORC’s groundbreaking Dementia DataHub provides unprecedented county-level data on Alzheimer's Disease and related dementias (ADRD) across the United States. This innovative tool comes at a critical time in the U.S. Advancing age is the greatest dementia risk factor, and adults over 65 are projected to outnumber children by 2030, according to the U.S. Census Bureau.
ADRD are brain conditions that cause progressive deterioration in memory, language, and bodily function and result in death. They are the sixth leading cause of reported death in the United States and cost the U.S. health care system over $305 billion in 2020. As the nation’s population ages, the need to track ADRD has never been greater. NORC’s new Dementia DataHub meets this need.
“Our Dementia DataHub is a first-of-its-kind ADRD surveillance system,” said David Rein, a senior fellow in NORC’s Public Health department and program area director of NORC’s Public Health Analytics Program Area. “It reports geographic counts of diagnosed ADRD prevalence, incidence, payments, mortality, and COVID-19 infection from the county level on up.”
NORC overcame data limitations by developing evidence-based case definitions.
Built in collaboration with George Washington University’s Milken Institute School of Public Health and KPMG LLP., with support from the National Institute on Aging, our hub draws data from administrative health care data systems, including Medicare fee-for-service claims and Medicare Advantage encounter data.
Although Medicare data provide highly accurate counts of people who have been diagnosed with ADRD, as many as 61 percent of people with ADRD likely remain undiagnosed because of various challenges, such as dementia symptoms going unrecognized in clinical settings or a lack of diagnosis codes indicating ADRD in someone’s Medicare record. To overcome these limitations, our team spent a full year systematically reviewing published diagnosis code case definitions and consulting with ADRD experts to develop these more evidence-based definitions:
- Highly Likely Dementia: A patient having a dementia-specific diagnosis code on at least two separate dates over a three-year period
- Likely Dementia: A patient having a dementia-specific diagnosis code indicating dementia on only one date over a three-year period
- Possible Dementia: A person having diagnosis codes that may indicate dementia over a three-year period or a prescription drug indicated for dementia without a corresponding dementia diagnosis code
- Any Dementia: The sum of highly likely dementia, likely dementia, and possible dementia
“We intend to refine these definitions over time using validation studies and eventually identify how many undiagnosed cases there are for each diagnosed one and how this rate varies by demographic and region,” Rein said.
“By providing researchers and others with epidemiological data in multiple formats and offering downloadable public-use files, the Dementia DataHub enables them to pursue their own hypotheses and conduct potentially life-altering investigations into ADRD.”
Our new definitions revealed that ADRD disproportionately affects minority populations and women.
Over 8.1 million Medicare beneficiaries were diagnosed with ADRD in 2020. Of these, 4.4 million were highly likely to have dementia, 1.1 million were likely to have it, and 2.6 million possibly had it. Even after age standardization, these individuals were older, frailer, more likely to be female, and used long-term care more than other Medicare beneficiaries across the board. We also found that:
- More Hispanics (8.2 percent) and Blacks (7.8 percent) than non-Hispanic whites (7.1 percent) fell into the “Highly Likely Dementia” category. [Some of these differences are explained by the different age distributions of different race and ethnicity groups within Medicare.]
- The locations with the highest “Highly Likely” rates were Puerto Rico (10.6 percent), Florida (8.2 percent), Texas (8.1 percent), Connecticut (8.0 percent), and New York (7.9 percent).
- Regarding COVID-19 infection, 16.5 percent of those classified as “Highly Likely” became infected during 2020 compared to 9.8 percent and 7.6 percent of those in “Likely” and “Possible” categories, respectively.
We created a multifunctional website to make the data easier to explore.
To make this granular data easier to access and analyze, we engaged experts in data visualization and privacy, geographic information systems, communications, and web design from across NORC. Our expanded team developed a visually appealing site that summarizes research and offers interactive dashboards and maps of results at the national, state, and county levels.
“By providing researchers and others with epidemiological data in multiple formats and offering downloadable public-use files, the Dementia DataHub enables them to pursue their own hypotheses and conduct potentially life-altering investigations into ADRD,” Rein said.
Future site updates will include additional years, data sources, and outcome measures.
The JAMA Network Open journal published our article on our case definition development methods and results in its September 2024 edition.
This article is from our flagship newsletter, NORC Now. NORC Now keeps you informed of the full breadth of NORC’s work, the questions we help our clients answer, and the issues we help them address.
