Mild Cognitive Impairment — What It Is, and Why It Isn't Early Dementia

October 1, 2026

We have fourteen posts on this blog about lowering your risk of dementia, and until now not one explaining what the diagnoses actually are. That is a strange gap for a brain health company, and this is the first of six posts fixing it.

Mild cognitive impairment is the right place to start, because it is the most consequential diagnosis almost nobody understands — and because the single most important thing about it is the thing least often said out loud: MCI is not early dementia, and between a fifth and a third of people who receive the label go back to normal.

We are not doctors. This explains what the research shows about a diagnosis, and it is not a substitute for the clinician who made it.

What MCI actually is

Mild cognitive impairment describes measurable cognitive change — usually memory, but it can be language, attention, planning or spatial judgement — that is greater than expected for someone's age and education, while day-to-day function remains broadly intact.

That last clause is the whole boundary. Someone with MCI forgets conversations, repeats questions, loses the thread of a book. Someone with dementia can no longer manage the activities that keep an independent life running — finances, medication, cooking, navigation. The cognitive test score matters less than that functional line, which is also why the diagnosis has a soft edge and why two clinicians can disagree about the same person.

It is also worth saying plainly: being worried about your memory is not MCI. Subjective memory complaints and measurable impairment are different things, and the overlap between them is poorer than you would expect. We went through that distinction in memory worries: normal ageing or not.

A lot of people go back to normal

This is the finding that should be the headline of every MCI explainer and almost never is. Four separate meta-analyses have pooled the reversion rate — the proportion of people diagnosed with MCI who, on later assessment, test as cognitively normal:

18% (95% CI 14–22) across 25 longitudinal studies, follow-up two years or more, evidence quality rated moderate.¹
~24% across 25 studies in a separate 2016 analysis.²
27.6% across 17 studies and 6,829 participants.³
31% across 48 studies and 31,876 participants, the largest pooled estimate to date.⁴

So somewhere between roughly one in five and one in three. That is not a rounding error on the way to dementia. It is a substantial share of everyone who gets told they have MCI.

Why those four numbers disagree — and why it matters to you specifically

An 18%-to-31% spread looks like the usual mess of meta-analysis. It isn't. Two of those analyses dug into the heterogeneity and found the same explanation: it depends enormously on where you were diagnosed.

The 2016 review reporting 18% overall found 8% (4–11) reversion in clinic-based studies versus 25% (19–30) in population-based studies.¹
The other 2016 analysis found the same pattern independently: 14% in clinic-based studies versus 31% in community-based ones.² It also found significantly lower reversion in Asian cohorts than North American or European ones.

The practical translation: if you were diagnosed at a memory clinic, having gone there because you or your family were worried, your chance of reverting is nearer 8–14%. If an MCI label came out of a population screening study or a routine check where nobody had raised a concern, it is nearer 25–31%. Those are very different situations wearing the same name, and the averaged figure describes neither.

And a caveat that cuts against over-reading reversion as good news. Some of it is genuine recovery — a treated depression, resolved sleep apnoea, a medication changed, a thyroid problem corrected, a vitamin deficiency fixed. Some of it is measurement: cognitive tests have poor day-to-day reliability, and someone who scored badly once because they were anxious, exhausted or unwell will score better next time without anything in their brain having changed. Reversion is partly real and partly regression to the mean, and the published rates do not separate the two.

What makes reversion more likely

The 17-study analysis identified the factors associated with going back to normal.³ Positive predictors were younger age (standardised mean difference −0.345, 95% CI −0.501 to −0.189), higher education (0.337, 0.117–0.558), absence of an APOE ε4 allele (OR 0.728, 0.575–0.922), absence of hypertension (0.826, 0.692–0.987), absence of stroke (0.696, 0.507–0.953), and a higher baseline MMSE score (0.707, 0.461–0.953).

Look at that list against the one we have already written about fourteen times. Age, education and APOE status are fixed. Hypertension and stroke are not. They are two of the biggest items on the modifiable dementia risk list, and here they turn up again as predictors of whether someone walks back from an MCI diagnosis. That overlap is the most actionable thing on this page, and it is covered properly in modifiable dementia risk and the fourteen factors.

The larger 2025 analysis added further associated factors, including depression, living alone, and doing housework daily or exercising weekly.⁴ Worth reading carefully: those are factors associated with reversion in observational data, not interventions shown to cause it.

Can anyone tell you what will happen to you? No.

This is the question everyone actually asks, and there is now a clear evidence-based answer to it.

A 2025 systematic review in Alzheimer's & Dementia assessed every multi-source model built to predict which MCI cases convert to dementia — 62 studies, comprising 41 machine learning models, 11 regression models and 5 disease state indexes, using between 2 and 60 predictors apiece, most commonly MRI and cognitive scores.⁵ On paper they perform reasonably: area under the curve from 0.58 to 0.98, accuracy from 66.1% to 96.3%.

Then the problems. Most studies were at high risk of bias, and 47 of the 62 had no external validation at all — meaning they were never tested on a population other than the one they were built from, which is where prediction models routinely fall apart. The authors' conclusion is unambiguous: there is currently no prediction model with low risk of bias that has been externally validated and can accurately predict conversion risk.

So if someone offers you a number for your personal odds — a clinic, an app, a supplement company — the honest state of the field is that the tools to produce that number reliably do not yet exist.

What biomarkers do and don't tell you

Amyloid and tau testing, via PET imaging or spinal fluid, is the closest thing to a real prognostic signal. A 2024 meta-analysis pooled 36 cohorts covering 7,793 people.⁶

In people with MCI who were amyloid-positive, the odds of converting to dementia were 5.18 times higher (95% CI 3.93–6.81) than in those who were not.
In people who were cognitively unimpaired but amyloid-positive, the odds of converting to MCI or dementia were 5.79 times higher (2.88–11.64).
Spinal fluid and PET imaging gave consistent results.

That is a large and real effect. It is also an odds ratio, not a prophecy: it tells you how the odds shift across a group, not what happens to one person. Plenty of amyloid-positive people do not convert within a study's follow-up, and the biomarker does not come with a timeline.

Two things that may be modifiable, reported honestly

Two recent meta-analyses looked at whether psychiatric symptoms predict progression from MCI to dementia, and they differ sharply in how much weight they can bear.

Anxiety. A review of 15 cohort studies, 14 of which entered the meta-analyses, found anxiety symptoms associated with increased risk of all-cause dementia (HR 1.25, 95% CI 1.11–1.41) and of Alzheimer's disease specifically (HR 1.32, 1.16–1.49).⁷ Notably, the pooled odds ratios across the same literature did not reach significance (OR 1.15, 0.84–1.57), which the authors report rather than bury, and they call for larger prospective studies before anyone leans on the finding.

Depression. Across 17 longitudinal studies, baseline depression was associated with progression overall at HR 1.66 (1.22–2.26) in unadjusted analysis — but with heterogeneity of I² = 99.1%.⁸ That figure means the studies disagree with each other almost completely, and the authors state plainly that it limits confidence in the pooled estimate. An I² of 99% is a signal that a single summary number should not be quoted as if it means something precise. We are printing it because leaving it out would be the less honest choice, not because it settles anything.

If anxiety or low mood is part of your picture, that is a reason to treat it on its own merits — it is miserable, and it is treatable. Whether treating it changes a dementia trajectory is not something this evidence establishes.

Where we come in, which is barely at all

We will be direct, because an MCI diagnosis is one of the moments in life when a person is most exposed to supplement marketing, and we are not going to be part of that.

We sell nothing for mild cognitive impairment. No product in our range has been tested in an MCI population, no ingredient we sell has been shown to affect reversion or progression, and anyone telling you a capsule changes this trajectory is selling you something the evidence does not support. The one ingredient in this territory with a real older-adult literature is phosphatidylserine, and we went through what it does and does not show in phosphatidylserine and memory — which is not a Proco product.

What does have evidence behind it is the unglamorous list: blood pressure, hearing, vision, activity, alcohol, sleep, social contact. Two of those — hypertension and stroke — showed up above as predictors of reverting to normal cognition. If you want the version of this with the actual effect sizes, modifiable dementia risk is the place, and exercise versus supplements is the honest comparison between the thing that works and the thing we sell.

The short version

MCI is measurable cognitive change with day-to-day function broadly intact. That functional line, not the test score, is the boundary with dementia.
Pooled reversion to normal cognition runs from 18% to 31% across four meta-analyses. Between a fifth and a third.
The spread is explained by setting: 8–14% reversion if you were diagnosed in a memory clinic, 25–31% if the label came from population screening.
Some reversion is real recovery of a treatable cause. Some is measurement noise. The published rates don't separate them.
Reversion is more likely without APOE ε4, hypertension or stroke — and two of those three are modifiable.
Across 62 prediction models, 47 were never externally validated and most carry high risk of bias. No validated model exists to tell an individual their odds.
Amyloid positivity raises conversion odds about fivefold. That is a real signal and still not a forecast for one person.
We sell nothing that affects any of this, and will not pretend otherwise.

The rest of this series

Six diagnoses, each written the same way: what it is, what the evidence actually shows, and what we do and do not sell for it.

This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease. We are not doctors, and mild cognitive impairment is a clinical diagnosis that requires proper assessment — including investigation of reversible causes such as thyroid dysfunction, vitamin deficiency, depression, sleep disorders and medication effects. If you are worried about your memory or someone else's, that is a conversation for a doctor, not a supplement. Nothing here is a reason to start, stop or change any prescribed medication.

References

Canevelli M, Grande G, Lacorte E, Quarchioni E, Cesari M, Mariani C, Bruno G, Vanacore N. "Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis." Journal of the American Medical Directors Association, 2016. doi:10.1016/j.jamda.2016.06.020
Malek-Ahmadi M. "Reversion From Mild Cognitive Impairment to Normal Cognition: A Meta-Analysis." Alzheimer Disease and Associated Disorders, 2016. doi:10.1097/WAD.0000000000000145
Xue H, Hou P, Li Y, Mao X, Wu L, Liu Y. "Factors for predicting reversion from mild cognitive impairment to normal cognition: A meta-analysis." International Journal of Geriatric Psychiatry, 2019. doi:10.1002/gps.5159
Zhao Y, Wei S, Liu Y, He X, Li J, Gao T, Wang X, Li Y, Nan J, Wang Y, Ma Y. "The prevalence and influencing factors of reversion from mild cognitive impairment to normal cognition: A systemic review and meta-analysis." Geriatric Nursing, 2025. doi:10.1016/j.gerinurse.2025.03.013
Vermeulen RJ, Andersson V, Banken J, Hannink G, Govers TM, Rovers MM, Olde Rikkert MGM, et al. "Limited generalizability and high risk of bias in multivariable models predicting conversion risk from mild cognitive impairment to dementia: A systematic review." Alzheimer's & Dementia, 2025. doi:10.1002/alz.70069
Huszár Z, Engh MA, Pavlekovics M, Sato T, Steenkamp Y, Hanseeuw B, Terebessy T, et al. "Risk of conversion to mild cognitive impairment or dementia among subjects with amyloid and tau pathology: a systematic review and meta-analysis." Alzheimer's Research & Therapy, 2024, 16. doi:10.1186/s13195-024-01455-2
Fresnais D, Fure B. "Anxiety Symptoms and Risk of Dementia Progression in Mild Cognitive Impairment: A Systematic Review and Meta-Analysis." Dementia and Geriatric Cognitive Disorders, 2026. doi:10.1159/dem/aciag007
Kassymzhanova M, Shonbay K, Shikhova OM, Raspopova N, Karibayeva I. "Depression as a prognostic factor for conversion from mild cognitive impairment to all-cause dementia and Alzheimer's disease: a systematic review and meta-analysis of longitudinal studies." Frontiers in Public Health, 2026. doi:10.3389/fpubh.2026.1831762