Electronic Health Records for Dementia Care Research in 2026: Who Was Studied and Who Was Not for People With Mild Cognitive Impairment

In 2026, electronic health record research on mild cognitive impairment studied mainly diagnosed patients in large health systems, while missing most early cases and underrepresented groups. Mild cognitive impairment means noticeable memory or thinking changes that do not yet block daily life. Electronic health records are the digital notes, diagnoses, prescriptions, and test results collected during routine care. Researchers now mine those records to find who develops dementia and why, but the records reflect who got care and a diagnosis.

Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.

Why do records miss so many early cases?
Who did large studies actually follow?
Who was not studied well?
What can families do with this evidence?

Why do records miss so many early cases?

About half of people with dementia and about 90% with mild cognitive impairment are undiagnosed, so record-based studies start without most early cases, according to eRADAR trial investigators via PMC in the eRADAR trial summary. A missing diagnosis means no code, no referral, and no follow-up in the data. Studies then track the visible minority. Kaiser Permanente Washington and UCSF tested eRADAR to fix that gap.

eRADAR is a validated structured-record algorithm with C-statistics of 0.79-0.84, used in embedded trials among patients 65 and older without documented dementia to find undiagnosed dementia. Better detection would bring more true early cases into both care and research. Missing cases also skew risk estimates. If healthier, younger, or less-connected patients go undiagnosed longer, their progression looks absent. Targeted screening matters more than larger databases alone.

Who did large studies actually follow?

UCSF Health researchers followed 6,529 patients with new mild cognitive impairment from 2010 to 2024 to test heart and social factors as predictors of Alzheimer's disease and related dementias. That group had a recorded diagnosis, longitudinal care, and enough visits to study change over time. It represents clinic-engaged patients, not all people with early symptoms. A national volunteer registry adds another studied group.

The National Alzheimer's Coordinating Center Uniform Data Set holds over 50,000 Alzheimer's Disease Research Center participants, with over 17,000 actively followed. Comparison of 36,639 center volunteers with 52 million weighted Health and Retirement Study respondents found the volunteers older, more educated, and not representative on sociodemographics and health. The All of Us program supplied a very large adult sample. Among 187,341 adults age 50 and older followed about 10 years, 6.7% with recorded ADHD versus 3.4% without developed mild cognitive impairment or dementia. That finding applies to enrolled adults with sustained records and measured genetic data.

Who was not studied well?

Everyday patients without a diagnosis were largely absent. Undiagnosed people do not appear in mild-cognitive-impairment cohorts, even when symptoms affect work or family life. Pragmatic trials now aim at that hidden group before dementia codes appear. Racial, ethnic, and socially disadvantaged groups also remain thinly captured. In the UCSF mild-cognitive-impairment cohort, racial, ethnic, and social disparities in progression persisted after adjusting for preferred language, indicating structural and sociocultural factors beyond language access affect outcomes, according to Miramontes and colleagues via PMC/NIH in the UCSF progression study.

Language help alone did not erase the gap. Social and structural details are the weakest fields in records. Housing, caregiving, discrimination, income, and neighborhood resources are inconsistently coded across clinics. NIA-designated Alzheimer's Disease Research Centers reported in 2026 that recruiting higher-risk populations introduces selection, prompting expanded outreach cores to enroll underrepresented and high-risk groups. Special outreach helps, but it also creates a selected study sample.

What can families do with this evidence?

Ask for cognitive symptoms to be written down at primary care. A recorded concern creates the visit, test, and referral trail that future care depends on. Bring a short list: examples, dates, medications, sleep, mood, blood pressure, and family observations. Ask whether the clinic uses a record-based risk check.

Kaiser investigator Sascha Dublin presented an EHR risk score at the 2026 Alzheimer's Association International Conference that helps primary-care clinicians spot high-risk impairment and dementia patients for earlier evaluation and treatment, according to Healio reporting on AAIC in the conference report. A high score is a prompt for testing, not a diagnosis. Push for complete social and heart-health information in the visit note. Control of blood pressure, diabetes, smoking, hearing loss, and depression is practical now. If language, transport, cost, or caregiving limits follow-up, say so and ask staff to record it.