Sunny Lin
Research Scientist
NORC at the University of Chicago
October 2026
The Missouri River divides Omaha, Nebraska from Council Bluffs, Iowa, but the two cities form a single metropolitan area, sharing a regional economy, transportation network, and health care systems. Exploring patterns from the 2022 Dementia DataHub (DDH) public use file, we found that the combined dementia prevalence across the system's three levels of case definitions (Highly Likely, Likely, and Possible) was 38 percent higher on the Iowa side (18.3 percent) compared to the Nebraska side (13.2 percent).
This five-percentage-point gap across state lines drew our attention because a gap of this magnitude was surprising in a single community with shared resources, raising the question: were older adults on one side of the state line truly more likely to have dementia than adults on the other side? To answer this question, we explored publicly available data from the DDH and other sources. Here is what we found.
First, we examined whether rates varied across the DDH's case definitions. The DDH reports county-level estimated dementia prevalence by diagnostic certainty: Highly Likely, Likely, Possible, and No Evidence. We compared eight counties: three on the Iowa side (Pottawattamie, Harrison, and Mills Counties) and five on the Nebraska side (Douglas, Sarpy, Cass, Saunders, and Washington Counties).
Source: Dementia DataHub public use file, 2022.
The Highly Likely and Likely tiers were nearly identical across the state line. The entire prevalence gap was accounted for by differential rates in the Possible tier, where Iowa's rate was more than twice Nebraska's.
This mattered for how we interpreted the dementia prevalence gap. In our validation study, Gianattasio et al. (2026) found that beneficiaries in the Possible tier had substantially higher false-positive rates than those in the Highly Likely or Likely tiers, driven by non-specific diagnoses such as general aging and frailty. We suspect the prevalence gap across the state line does not reflect a true difference in dementia prevalence but instead measures a difference in how aging-related conditions were documented on each side of the Missouri River.
Figure 1. Map of dementia prevalence by tier, Iowa and Nebraska counties in the Omaha-Council Bluffs metropolitan area, 2022.
To investigate further, we examined three sets of descriptive comparisons. First, beneficiary characteristics: if the populations on both sides of the state line were different in age, sex, race and overall health, it could suggest that beneficiary-level health was driving the prevalence differences. Second, payment incentives: if people in different states had different access to and enrollment in Medicare Advantage (MA) plans, it could suggest that MA coding incentives were driving elevated diagnoses on the Iowa side. Third, health care services: if health care infrastructure, including access to nursing homes and geriatric services, were different on each side of the river, it could suggest that differential access to health care was driving differences in coding practices.
If the Iowa side simply had a frailer or older population, we would have expected to see elevated dementia rates across all tiers, not just the Possible tier. We did not. Highly Likely and Likely rates were nearly identical on both sides of the state line (7.9 percent versus 8.4 percent, Table 1), which argued against a population that was meaningfully sicker or more cognitively impaired overall.
We examined this more directly using subgroup comparisons and age-standardized rates. Among non-Hispanic White beneficiaries, the only racial group with sufficient sample size on the Iowa side for comparison, the prevalence gap persisted (10.3 percent versus 5.1 percent, Table 2a). Similarly, the prevalence gap held across state lines for both males and females (Table 2a), and after standardizing to the 2022 national Medicare age distribution (Table 2b).
Source: Dementia DataHub public use file, 2022. ASR=Age-standardized Rate to 2022 national Medicare age distribution.
We then shifted our attention to proxy measures of health status. The DDH public use file reports mortality, hospitalization, short- and long-stay nursing facility (NF) use, and emergency department (ED) visit rates for each diagnosed tier. We used these to assess whether Possible-tier beneficiaries on the Iowa side were sicker than those on the Nebraska side. If they were, we would have expected to see higher mortality and greater health care utilization on the Iowa side. We found the opposite. Among Possible-tier beneficiaries, mortality, hospitalizations, short- and long-stay nursing facility episodes, and ED utilization were all lower on the Iowa side than the Nebraska side (Table 3).
Taken together, these results suggest that the prevalence gap was not driven by a sicker or older population. If anything, Possible-tier beneficiaries on the Iowa side appeared healthier than those on the Nebraska side.
Medicare Advantage (MA) plans are risk-adjusted to allow MA payers to receive higher payments for sicker patients. This creates a financial incentive for MA to increase documentation of conditions such as dementia, mild cognitive impairment, and frailty. We wanted to test the hypothesis that MA enrollment was higher in Iowa, leading to greater documentation of Possible-tier diagnoses. In addition, we examined market concentration. If a single plan dominates either MA market, that plan's coding practices could be responsible for driving the observed prevalence gap.
To explore whether the two sides differed in MA enrollment and market characteristics, we examined publicly available data on MA enrollment and plan information from CMS. We found that MA enrollment was similar (34.26 percent in Iowa and 36.22 percent in Nebraska), and MA markets on both sides were similar in terms of market dynamics: the Herfindahl-Hirschman Index (HHI) was 1,259 on the Iowa side and 1,569 on the Nebraska side. The HHI, a standard measure of market concentration, suggests that MA is marginally more concentrated in Nebraska than Iowa, but not enough so to support the explanation that a small number of MA plans are driving significant coding differences across state lines (Federal Trade Commission, 2023). Further, the same national payers administered the three largest plans on both sides (Table 4b). None of these signals suggested that differences in MA market characteristics were driving the observed difference in prevalence gap, though we cannot completely rule out the possibility that differences in plan practices contribute.
Source: Dementia DataHub public use file and CMS State/County/Contract (SCC) enrollment file, January 2023. MA = Medicare Advantage. HHI= Herfindahl-Hirschman Index, Values above 1,800 indicate a market is highly concentrated. Normalized HHI takes into account the relative number of plans in each market and is calculated as (HHI – 1/n)/(1-1/n) where n is the number of plans in the market.
Source: CMS State/County/Contract (SCC) enrollment file, January 2023. MA = Medicare Advantage.
Finally, to see if access to health care services might explain the prevalence gap, we examined county-level data on health care access from the Area Health Resources Files (AHRF), CMS Nursing Home Compare, and LTCFocus to see if there were differences in physician supply and nursing facility dementia care across the state line.
We found that the Nebraska side had 645 primary care physicians per 100,000 adults 65 and older, compared to 241 on the Iowa side. General internal medicine physicians followed the same pattern: 285 per 100,000 on the Nebraska side versus 83 on the Iowa side. The Nebraska side had five hospitals with dedicated geriatric services; the Iowa side had none. And yet the Iowa side had more nursing facility capacity. There were 48.3 certified nursing facility (NF) beds per 1,000 adults 65 and older on the Iowa side, compared to 34.2 on the Nebraska side, a 41 percent higher rate (Table 5).
*Physician supply: primary care and general internal medicine physicians across CBSA counties divided by total population 65+ across those counties, expressed per 100,000. Source: AHRF 2022-2023, AMA Physician Master File, 2021.
**Nursing facility beds: total certified beds across CBSA counties divided by total population 65+ across those counties, expressed per 1,000. Source: CMS Nursing Home Compare, January 2023; population denominator from AHRF Census population estimate, 2021.
Per capita, Iowa had more NF beds than Nebraska. This difference may be meaningful because NFs must assess all residents regularly, and this systematic documentation may increase the capture of dementia-related diagnosis codes in Medicare claims. If greater NF supply increased the share of beneficiaries who receive regular assessment, it could raise measured Possible dementia prevalence in Iowa.
However, two additional findings do not fit this explanation. First, in our analysis of Hypothesis 1 using the DDH data, Possible-tier beneficiaries in Iowa used short-stay and long-stay NF care less often than those in Nebraska (Table 3). Second, in LTCFocus public use data, longterm care residents in the two states had similar rates of dementia diagnoses (42.08 percent in Iowa and 42.17 percent in Nebraska, county averages weighted by certified beds). These results do not support the hypothesis that differential NF access drives the prevalence gap. The lower NF use among Possible-tier beneficiaries in Iowa indicates that the gap lies among beneficiaries living in the community, not NFs, and comparable dementia diagnosis rates among NF residents in both states suggest minimal differences in documentation patterns.
The analyses above were all based on 2022 public use data. As a final attempt to understand this phenomenon, we recreated Table 1 using data from the 2020 and 2021 DDH public use files to see whether the Possible-tier prevalence gap between the Iowa and Nebraska sides of the Omaha-Council Bluffs metropolitan area reflected a longstanding pattern or a more recent change. We found that Possible-tier prevalence rose on both sides of the state line from 2020 to 2022, and it rose faster in Iowa than in Nebraska (Table 6). Iowa’s prevalence was 1.33 times its 2020 level in 2021 and 2.00 times its 2021 level in 2022. Nebraska’s prevalence was 1.05 times its 2020 level in 2021 and 1.14 times its 2021 level in 2022. As a result, the ratio of Iowa to Nebraska prevalence grew by a factor of 1.27 from 2020 to 2021 and by a factor of 1.76 from 2021 to 2022.
Source: Dementia DataHub public use files, 2020, 2021, and 2022.
To examine whether population influx could have been a driver of this increase in Possible-tier prevalence, we examined the numerator and denominator of prevalence rates by year (Table 7). We found that the denominator (total Medicare beneficiaries 65 and older) grew at a similar pace on both sides of the state line between 2020 and 2022 (Iowa +5.4 percent, Nebraska +6.3 percent), making population influx an unlikely explanation.
Source: Dementia DataHub public use files, 2020, 2021, and 2022. Δ = percent change 2020 to 2022.
Temporal increases in the prevalence of possible dementia may not be unique to counties in the Omaha metro area. We compared Possible-tier prevalence among Medicare beneficiaries 65 and older in every county with a reportable value in both the 2021 and 2022 DDH public use files (3,100 counties, Figure 2). Prevalence rose in 65.1 percent of these counties, with a median change of +0.20 percentage points (interquartile range −0.15 to +0.58), and the pooled rate across these counties moved from 4.53 percent to 4.86 percent. Mills and Pottawattamie Counties, on the Iowa side of the Omaha MSA, had the two largest absolute increases in the country (8.20 and 5.47 percentage points) and ranked first and sixth in relative change (+146.5 percent and +101.5 percent). These results suggest a modest increase in possible dementia prevalence nationwide, one worth monitoring in 2023 and 2024 files (release forthcoming). If this trend continues, it could be a sign of larger changes in documentation patterns, changes in the prevalence of aging-related conditions, or both.
Figure 2. Map of absolute change in Possible dementia prevalence from 2021–2022, Omaha-Council Bluffs metropolitan area highlighted.
In this blog post we used public use files to explore an interesting case where dementia prevalence on one side of a river was substantially higher than that of the other within a single metropolitan area. We determined that the prevalence gap was entirely accounted for by the Possible tier, most likely among community-dwelling residents, without accompanying elevated rates of mortality or acute care use, suggesting that the increase was more likely a story about how dementia is documented than actual dementia rates.
We examined three candidate explanations: beneficiary characteristics, payment incentives, and health care access, and found limited evidence for each of them. Moreover, we found that Possible dementia prevalence has increased since 2020 on both sides of the river, but at a higher rate in Iowa than Nebraska: Iowa’s prevalence growth relative to Nebraska’s was 1.27 from 2020 to 2021 and 1.76 from 2021 to 2022. We also found evidence that the use of possible dementia codes may be part of a larger temporal and national trend worth monitoring in the future. This analysis was enabled by the DDH public use file and other publicly available data at the county level: the Area Health Resources Files, CMS Nursing Home Compare, CMS MA enrollment files, and LTCFocus Public Use Data. If you have local knowledge that could shed light on what we explored in Omaha, email us at Dementiadatahub@norc.org. We would love to hear from you. Or if you want to use the data yourself, fill out our data request form to download the public use files.
Gianattasio K.Z., Steffan M., Reist B., Power M.C. & Rein D. Performance of ICD-10 code-based dementia case definition in the Health and Retirement Study. Alzheimer's & Dementia. 2026;22(5):e71481.
Federal Trade Commission. Merger Guidelines [Internet]. Washington (DC): FTC; 2023 Dec 18. (Accessed 2026 October 6.)
LTCFocus Public Use Data sponsored by the National Institute on Aging (P01 AG027296) through a cooperative agreement with the Brown University School of Public Health. Available at www.ltcfocus.org. (Accessed 2026 October 6.)
Centers for Medicare & Medicaid Services. Provider Information (Nursing Home Care Compare), file NH_ProviderInfo_Jan2023.csv. Provider Data Catalog. (Accessed 2026 September 24.)
Health Resources and Services Administration, Bureau of Health Workforce, National Center for Health Workforce Analysis. Area Health Resources Files (AHRF), 2022 to 2023 release. (Accessed 2026 September 24.)
Centers for Medicare & Medicaid Services. Medicare Advantage/Part D Contract and Enrollment Data, monthly enrollment file for January 2023. (Accessed 2026 September 24.)
