Diagnosing Dementia in 2026: What Just Changed, and What It Means for Families

I spent this week at the American Psychiatric Association's 2026 Annual Meeting in San Francisco, which closed today. The session I keep thinking about — and the reason I'm writing this — was "Diagnosing Dementia: An Updated Roadmap for Clinical Practice," led by Dr. Vimal Aga, a geriatric psychiatrist at OHSU, and Dr. Allan Anderson. Their argument was simple: in the last 18 months, the way doctors diagnose dementia has quietly changed in real ways. Most patients — and many doctors — are still operating on the old model.

This article is my attempt to translate what I heard into plain English: what's different now, what it means, and what to do with it if you or someone you love is being evaluated for memory or thinking problems.

What changed in the last year and a half

Four major medical guidelines have come out since late 2023. Together, they shift dementia diagnosis from "let's check your memory and get an MRI" to a structured, multi-step evaluation that now includes a blood test.

The guidelines, in plain language:

• A new playbook for memory clinics that lays out exactly how to evaluate someone for dementia, step by step (the field calls it DETeCD-ADRD, published in 2025 by Atri and colleagues).
• The first official recommendation that doctors can use a blood test for Alzheimer's disease in their everyday practice (Palmqvist and colleagues, 2025).
• A new definition of Alzheimer's disease itself, which for the first time treats it as a disease that can be identified by biology in the brain — not only by symptoms (Jack and colleagues, 2024).
• Updated rules for reading brain scans in people with possible dementia from blood vessel disease (2023).

You don't need to remember the names. What matters is that the floor has moved. A good evaluation in 2026 is meaningfully different from a good evaluation in 2023.

What normal aging actually looks like

Before talking about dementia, it helps to know what's normal. Cognition is not one thing.

There are two kinds of "smart":

• The kind that stays steady or even improves as we age — your vocabulary, your accumulated knowledge, the skills you've practiced your whole life. Most adults in their 60s and 70s do this just as well as they did at 40, or better.
• The kind that slowly fades — speed at solving unfamiliar problems, mental flexibility, how fast you can react. This peaks in your twenties and quietly slips for the rest of your life.

Most older adults notice the second kind getting slower, while the first kind stays sharp. That pattern is not dementia.

Dementia is something else. It means a clear drop in thinking that is significant enough to interfere with day-to-day independence, that is new compared to how the person used to be, and that isn't explained by something temporary like a hospital stay, a new medication, or depression.

Two different ways into dementia

One of the most useful ideas in Dr. Aga's session: there is more than one path into dementia. Until recently, almost all of our research and screening focused on a single path — the memory path. We are now learning that some people enter through a completely different door.

Path one: the memory path. A person starts feeling like their memory has slipped, even though tests are still normal. Some progress to measurable memory and thinking problems. Some progress further to dementia.

Path two: the behavior path. A person — usually over 50, with no obvious memory complaint — starts acting different. Maybe less motivated. Maybe more suspicious. Maybe less empathetic. Maybe socially inappropriate in small ways. The change is real, persistent, and not explained by something obvious like a death in the family. This is Mild Behavioral Impairment, and we now know it can be the first sign of dementia.

Most families, and most primary care visits, only look for the first path. The second path is the news.

Mild Behavioral Impairment, in plain words

The technical criteria boil down to five things:

1. The person is over 50.
2. There's a new behavior or personality change that wasn't there before.
3. It has lasted at least six months, continuously or off-and-on.
4. It isn't explained by life events, another medical illness, a medication, or a clear psychiatric disorder.
5. They can still manage their own life — pay bills, drive, take medications. (If they can't, we're already past this stage.)

Examples I hear from families:

• "Dad used to be the most generous person we knew. Now he's stingy and accuses my mother of taking his money."
• "Mom doesn't care about her grandkids anymore. She'll sit in the recliner all day. This is a woman who never sat still."
• "He keeps saying small inappropriate things in public — never used to do that."
• "She's lost her ability to read people. She'll laugh at the wrong moments."

These are not just "personality" or "aging." When the change is new, real, and persistent in someone over 50, it deserves a cognitive evaluation — even if memory still seems intact. Doctors, especially psychiatrists, are increasingly trained to ask: is this a new psychiatric problem, or is this the first sign of dementia? They are not the same.

The 2024 update to the Alzheimer's diagnosis criteria now formally treats "a recent-onset change in mood, anxiety, or motivation not explained by life events" as a transitional stage of Alzheimer's disease itself. That's a major shift.

The "worried well" — and when worry deserves attention

A lot of people show up at memory clinics worried about their own memory. Most of them are fine.

Of people who feel their memory has slipped but test normally:

• About 7 in 100 per year go on to develop measurable thinking problems.
• About 2 in 100 per year go on to develop dementia.
• Over four-plus years, fewer than half progress at all.

But some of these worried people do have early Alzheimer's brewing. The clues:

• The complaint is specifically about memory, not about everything in general
• It started in the last five years
• They are 60 or older
• They feel worse than people their own age
• A family member confirms that they've actually changed
• The complaint persists over time
• They are bothered enough to seek medical help

If most of those clues fit, it's worth a thorough evaluation. If they don't, reassurance is usually appropriate.

The reassuring news for everyone: temporary memory blips — losing your keys, forgetting a name — are not typical signs of dementia. People with early dementia usually do not realize the full extent of what's happening to them. The people who are most worried about their memory are often the people who are fine.

What a good dementia evaluation actually looks like

In Dr. Aga's session he teaches his trainees a six-step checklist. Translated for the rest of us, a good evaluation should answer six questions:

1. How bad is this, really? Is it normal aging, a subjective feeling of slipping, measurable but mild changes, or dementia?
2. Is there something reversible going on? Many "dementias" are not.
3. Which thinking skills are affected? Memory, language, attention, problem-solving, visual-spatial skills, social judgment.
4. Are there mood or behavior changes? And did they come before the memory changes?
5. What's most likely causing this? Alzheimer's, blood vessel disease, Lewy body disease, frontotemporal disease, mixed, or reversible.
6. What's at stake? Driving, finances, medication mistakes, vulnerability to scams.

For the first question, a structured cognitive screening — even a simple one — is much more useful than "tell me three words and remember them."

For the second question, every patient should get a basic set of blood tests:

• A complete blood count
• A metabolic panel (kidneys, liver, electrolytes)
• Calcium, magnesium, phosphate
• Thyroid hormone
• Vitamin B12 and a related marker called homocysteine
• Markers of inflammation

And every patient deserves a careful look at their medication list. The single most underappreciated cause of "dementia" in older adults is medication side effects — especially:

• Over-the-counter sleep aids containing diphenhydramine (Benadryl, ZzzQuil, Tylenol PM) or doxylamine (Unisom). These are sold without a prescription, often to older adults, and they directly cause cognitive impairment.
• Bladder medications for urinary urgency. Some block the same brain chemical as the sleep aids above.
• Some prescription antidepressants, antihistamines, and gastrointestinal medications.

A 2025 review identified over a thousand products in these categories. People often improve noticeably when these are reduced or stopped.

Sleep apnea is another commonly missed cause. Snoring loudly, gasping at night, daytime sleepiness — these are worth bringing up. Untreated sleep apnea can absolutely look like cognitive decline.

A blood test for Alzheimer's, finally

This is the headline change. In 2025, the FDA cleared the first blood test for Alzheimer's disease for routine clinical use. Several brands are now available; the medical research community has agreed on which ones meet a strict accuracy threshold.

What the test does, in plain terms: it measures two proteins in your blood that reflect what's happening in your brain. If you have the Alzheimer's protein buildup that causes the disease, those proteins show up at altered levels in the bloodstream.

How accurate is it?

• A negative result is very reliable — about 97% confidence that Alzheimer's-type brain changes are not what's happening.
• A positive result is moderately reliable — usually followed up with a brain scan or spinal fluid test before any diagnosis or treatment decision.

What this changes:

• People who used to need a brain scan or a spinal tap to find out whether they had Alzheimer's can now often get the answer from a blood draw.
• People who are clearly not in the Alzheimer's group can be told so with confidence, and the workup can refocus on what else might be going on.

What this doesn't change:

• It is not a test for healthy people. Doing this in someone with no symptoms creates more confusion than clarity.
• It does not replace the rest of the evaluation. A positive blood test in someone with mild symptoms still requires the full workup to understand what's really happening.

The hard truth: most dementia isn't one thing

For decades we talked about dementia as if each person had one disease — Alzheimer's, or vascular, or Lewy body. Brain autopsies have made it clear that this isn't how it actually works.

A 2023 study looked at 1,647 brains and found up to seven different types of brain pathology in a single person, in 161 different combinations. Pure single-disease dementia is the exception.

A 2026 study of memory-clinic patients using blood biomarkers found:

• About 40% had a pattern consistent with Alzheimer's disease
• About 16% had Lewy body or Parkinson-like disease
• About 8% had blood vessel disease as the main driver
• About 26% had a clear mix of Alzheimer's plus blood vessel disease
• About 9% had brain shrinkage without any of the major culprits accounting for it

For families, this matters in a concrete way: a positive test for one thing does not mean that's the whole story. A "negative" test for Alzheimer's does not mean nothing is wrong. Treatments increasingly target specific diseases, but real people often have several.

The 23andMe trap

A growing number of new patients arrive at memory clinics holding a 23andMe result that says they have "the Alzheimer's gene" — usually meaning one or two copies of a gene variant called APOE ε4. Many ask to start a dementia medication "before it's too late."

This is almost always the wrong move.

• APOE ε4 raises your risk. It does not seal your fate.
• People with two copies have roughly a 30–55% lifetime chance of developing memory problems. That's higher than average, but it's not 100%. Many never develop dementia at all.
• People with one copy have only a modestly increased risk.
• The risk also depends on ancestry — the effect is smaller in some populations.

Most importantly: in someone with no symptoms, the gene result by itself is not a reason to order a brain scan, a biomarker blood test, or a dementia medication. The new guidelines specifically reserve those tools for people who are being evaluated for actual cognitive symptoms.

If you have a worrying APOE result and no symptoms, the best things you can do are the same things that help anyone: blood pressure control, exercise, sleep, hearing aids if you need them, social and cognitive engagement, treatment of any depression. None of these are dramatic. All of them are real.

Safety: a fuller picture

When clinicians talk about safety in dementia, they used to mean: can this person drive, manage money, take their pills. That's still important. But the bigger picture has three parts:

1. What the person themselves might do. Driving when they shouldn't. Mismanaging investments. Wandering. Operating tools or stoves unsafely. Some dementias cause specific behaviors — like compulsive eating, or acting out dreams during REM sleep — that have their own safety implications.

2. What dementia does to the rest of the household. Caregivers get sick. Marriages strain. Adult children quit jobs. Savings drain into care costs. The well-being of the family is part of the safety picture, not separate from it.

3. What other people do to the person with dementia. This one is underweighted. Older adults with cognitive impairment are heavily targeted by phone scams, romance scams, and financial exploitation — sometimes by strangers, sometimes by relatives or paid caregivers. Losses are often catastrophic and rarely recoverable. Protecting against this is part of dementia care, not an afterthought.

What this means if you're worried about a loved one

The practical version:

• Take behavior changes seriously, not just memory changes. New, persistent changes in personality, mood, or motivation after age 50 deserve a cognitive evaluation, even if memory seems intact.
• Bring someone with you to the appointment. People with early cognitive changes often don't recognize the full picture of what's happening. A spouse, adult child, or close friend who can describe what's changed is one of the most useful parts of the visit.
• Write down what you've noticed and when it started — before you go. The list will be more accurate and harder to dismiss than a conversation in the room.
• Ask explicitly about reversible causes. Has the medication list been reviewed for things that affect thinking? Were B12, thyroid, and sleep apnea screened for?
• Ask about the new blood test. If the picture looks like it could be Alzheimer's, the question "should we run the blood biomarker test?" is now a reasonable one.
• Don't act on a 23andMe result alone. It is not, by itself, a reason for further testing in someone without symptoms.
• Push for a real evaluation. If your visit is fifteen minutes long and ends with "it's probably just aging," ask for a referral to a memory specialist or geriatric psychiatrist.

Bottom line

Dementia diagnosis in 2026 is both more demanding and more capable than it was two years ago. Clinicians are being asked to evaluate behavior as well as memory, to recognize that most dementia is mixed, and to use a blood test that didn't exist for clinical practice a year ago. Families are being asked to advocate for that level of evaluation — and to take the early, quiet, behavioral signs as seriously as the memory ones.

The reward for doing it well is real. Earlier and more accurate diagnosis finds the reversible causes that can masquerade as dementia, gets the right patients into modern treatments, and gives families enough time to plan before the harder decisions arrive.

---

Session attended: "Diagnosing Dementia: An Updated Roadmap for Clinical Practice," American Psychiatric Association Annual Meeting, San Francisco, May 17, 2026. Presenters: Vimal M. Aga, MD, DFAPA, FAAGP (Oregon Alzheimer's Disease Research Center, OHSU) and Allan Anderson, MD. Any errors of summary or interpretation are mine.