Why Your Annual Physical Won't Catch Early Cognitive Decline

There is a comforting belief that most people carry without examining: if something were really wrong with my brain, my doctor would catch it. It seems reasonable. You go for your annual checkup. Your doctor runs bloodwork, checks your blood pressure, listens to your heart. Surely, somewhere in that process, they would notice if your cognition were slipping.

They almost certainly would not. And this is not because your doctor is bad at their job. It is because the system is not designed to detect the thing you are most afraid of. Early cognitive decline falls through the gaps of routine medical care so consistently that by the time most people receive a cognitive diagnosis, the disease has been progressing for years — sometimes more than a decade.

Understanding why this happens is not about losing faith in medicine. It is about understanding that cognitive health monitoring is your responsibility, not your doctor's — at least for now, and at least at the early stages that matter most.

The average primary care visit lasts 15 to 20 minutes. In that time, your doctor needs to review your medications, address your acute complaints, conduct a physical examination, order any necessary labs, and document everything. Cognitive assessment is rarely on the agenda unless you specifically bring it up.

Even when cognitive screening is recommended — the Medicare Annual Wellness Visit, for example, includes an “assessment of cognitive function” — it is often conducted perfunctorily. A study in the Journal of General Internal Medicine found that many primary care physicians fulfill this requirement simply by asking the patient if they have noticed any memory problems. If the patient says no, the box is checked. If the patient has anosognosia (lack of awareness of their own deficits), they will say no regardless of their actual cognitive state.

There is also a social dynamic at play. In a brief office visit, the doctor is having a scripted, familiar conversation with a patient who is trying to present their best self. The patient is alert, dressed, oriented, and responsive. They answer questions appropriately. They laugh at the right moments. They seem fine. And they probably are fine, in the context of a 15-minute structured interaction. But that interaction tells you nothing about whether they can manage their finances, follow a complex recipe, or navigate an unfamiliar route — the kinds of tasks where early cognitive decline actually shows up.

When doctors do perform a cognitive screening, the most commonly used tool is the Mini-Mental State Examination, or MMSE. It has been the default screening instrument for decades. It is also woefully inadequate for detecting early cognitive decline.

The MMSE is a 30-point test that covers orientation (what year is it, where are we), registration (repeat three words), attention and calculation (count backwards from 100 by 7), recall (remember those three words from earlier), and language (name objects, follow instructions, write a sentence). It takes about 10 minutes to administer.

The problem is that the MMSE was designed to identify moderate to severe cognitive impairment. A score of 24 or above (out of 30) is considered “normal.” This means you can lose up to 20 percent of your cognitive function on this test and still pass. A person with mild cognitive impairment or very early Alzheimer's will routinely score 26, 27, 28, or even 29. The test is simply not sensitive enough to detect the subtle changes that matter most for early intervention.

Additionally, the MMSE has significant ceiling effects for educated individuals. If you went to college and have been intellectually active throughout your life, your cognitive reserve may allow you to ace the MMSE even with meaningful underlying pathology. You might score a perfect 30 while your hippocampus is already showing significant atrophy on an MRI. The test just is not hard enough to challenge the compensatory abilities of a well-educated brain.

There are better screening tools. The Montreal Cognitive Assessment (MoCA) is more sensitive than the MMSE and is increasingly used in clinical practice. But even the MoCA is a one-time snapshot. It tells you how you performed on one occasion. It does not tell you whether that performance is different from how you performed six months ago.

Even if your doctor administered the most sensitive cognitive screening tool available, doing it once a year has a fundamental limitation: it gives you a snapshot, not a trend.

Cognitive performance varies significantly from day to day. Sleep, stress, medications, caffeine, mood, time of day, recent physical activity, and dozens of other factors influence how well you perform on any given cognitive assessment. A single test score is a composite of your actual cognitive ability plus all of these confounding variables, and you have no way to separate the two from a single measurement.

To detect early cognitive decline, you need to see change over time. Specifically, you need enough data points that the day-to-day variation averages out and the underlying trajectory becomes visible. This is basic signal processing. If the signal (real cognitive change) is small relative to the noise (day-to-day variation), you need many measurements to detect it.

An annual cognitive test gives you, at most, two data points per comparison: this year and last year. If this year's score is lower, it could be decline. Or it could be that you slept poorly, were anxious about the appointment, or had a cold. You literally cannot tell from two data points. You need dozens, ideally hundreds, of measurements before the noise subsides and the signal, if it exists, becomes clear.

There is an uncomfortable structural reason why early cognitive detection is not a priority in primary care: there has historically been limited treatment to offer even if decline is detected early. For decades, the clinical logic was: why screen for something if you cannot do anything about the result?

This reasoning is becoming outdated. New treatments targeting amyloid plaques are showing modest but real effects in slowing early-stage Alzheimer's. Lifestyle interventions — exercise, sleep optimization, cardiovascular risk management, cognitive engagement, hearing loss treatment — are increasingly supported by evidence as neuroprotective, particularly when started early. And a 2024 Lancet Commission identified 14 modifiable risk factors that collectively account for roughly 45 percent of dementia cases worldwide.

But the clinical infrastructure has not caught up with the science. Primary care is still organized around acute problems and established disease management, not presymptomatic detection. Insurance reimbursement for cognitive screening is limited. Time constraints make comprehensive cognitive assessment impractical. And the liability calculus creates reluctance: telling a patient “your cognition seems slightly off today compared to some abstract norm” on the basis of a single brief screening is not something many doctors are comfortable doing.

None of this is your doctor's fault. They are working within a system that was not designed for this problem. But the practical implication is clear: if you are waiting for your healthcare provider to proactively detect early cognitive changes, you are likely waiting too long.

The consequences of these structural gaps are measurable. The Alzheimer's Association estimates that more than half of people with Alzheimer's disease have never received a formal diagnosis. Among those who are diagnosed, the average delay between the onset of symptoms noticed by family members and formal diagnosis is approximately two to three years.

But “onset of symptoms noticed by family” is itself a lagging indicator. The neurobiological changes of Alzheimer's begin 15 to 20 years before symptoms are obvious enough for family members to notice. So the true delay — from the start of disease to diagnosis — is measured in decades, not years.

During this delay, the brain is losing neurons and synapses that cannot be replaced. Cognitive reserve is being drawn down. Treatment options that work best in the early stages are not being used because the early stages are not being detected. By the time a clinical diagnosis is made, the disease has often progressed to a point where the most effective interventions are already past their window of maximum benefit.

This is not a system that is failing at the margins. It is a system that is structurally mismatched to the problem. The problem requires continuous monitoring; the system offers intermittent screening. The problem requires sensitivity to subtle change; the system offers blunt instruments. The problem requires years of longitudinal data; the system offers annual snapshots.

If the healthcare system is not going to catch this early, what will? The answer is the same approach that works for any gradually changing system where single measurements are noisy and the signal is small: frequent, standardized, multi-domain measurement over time.

This is not a radical idea. It is how we already handle blood pressure, blood sugar, heart rate, and body weight. Nobody expects a single blood pressure reading at an annual physical to catch developing hypertension. We know that blood pressure varies throughout the day and that you need multiple readings over time to establish a reliable baseline and detect meaningful change. Cognitive health is no different in principle. It has just been treated differently because we lacked practical tools for daily cognitive measurement outside of clinical settings.

That is changing. Keel provides five standardized cognitive tests — processing speed, reaction time, working memory, executive function, and verbal fluency — in four minutes. You do them daily. You log contextual factors that affect performance. And the data builds into a longitudinal profile that is orders of magnitude more informative than an annual screening.

After 30 days, you have a baseline. After 90 days, you have a trend. After six months, you have a dataset that can detect changes that no annual screening and no subjective self- assessment could identify. And crucially, you have this data before symptoms become obvious — during the window when intervention matters most.

Cognitive tracking does not replace medical care. It makes medical care more effective. Here is how.

It changes the conversation. Walking into your doctor's office and saying “I am worried about my memory” is subjective and vague. Walking in with six months of daily cognitive data and saying “my processing speed has declined by one standard deviation over the past three months while my other domains are stable” is specific and actionable. The first statement might get you a reassuring pat on the back. The second will get you a referral.

It provides the longitudinal view that clinical care lacks. Your doctor sees you for 15 minutes once or twice a year. They do not have a longitudinal cognitive profile for you. Your data fills that gap. It gives them the trajectory information they need to make better decisions about whether to investigate further.

It controls for confounders. Because you log context (sleep quality, illness) along with your cognitive scores, you can demonstrate whether a decline is independent of known confounders or explained by them. “My scores dropped, but it coincided with three weeks of poor sleep after starting a new medication” is a very different story than “my scores dropped and I have been sleeping fine.”

It establishes a pre-symptomatic baseline. If you do eventually need formal neuropsychological testing, having a documented baseline from when you were cognitively healthy is enormously valuable. It gives the neuropsychologist a reference point for what “normal” looks like for you specifically, not just for your age and education group.

The healthcare system will eventually get better at early cognitive detection. Blood-based biomarkers for Alzheimer's are advancing rapidly. AI-assisted screening tools are in development. Clinical guidelines are slowly beginning to emphasize earlier assessment. But “eventually” does not help you now.

The gap between what is possible for early cognitive detection and what the medical system routinely provides is one of the largest unaddressed gaps in healthcare. You can wait for the system to close it, or you can close it yourself with four minutes a day.

Your annual physical is valuable. Your doctor is valuable. But for the specific task of detecting subtle, early cognitive changes over time, the current healthcare model is a hammer being asked to do the work of a seismograph. It is the wrong tool for this particular job.

Build your own seismograph. Track daily. Let the data accumulate. If it shows stability, you have something more reliable than your doctor's 15-minute impression. If it shows change, you have something more specific than a vague concern. Either way, you have what the system currently cannot give you: a continuous, objective, longitudinal picture of your cognitive health that catches changes early enough for them to matter.