Cognitive Aging After COVID: New Risk Scores, Real-World Trends

The CAIDE score was built to look forward from middle age. You add up points for the familiar variables — older age, fewer years of formal schooling, higher systolic blood pressure, higher cholesterol, higher BMI, physical inactivity, being male — and the total gives a rough probability of dementia twenty years on. Useful at 50. Less obviously useful at 70, when the twenty-year horizon is a different proposition and many of the inputs have already done their work.

So a group of Hungarian researchers, writing in GeroScience, took 101 healthy adults over 55, sorted them into lower- and higher-CAIDE groups, and ran them through neuropsychological testing and MRI. The higher-risk group performed measurably worse on the Trail-Making Test, a standard probe of executive function and processing speed. They also had smaller global brain volumes and smaller regional volumes, including in the nucleus accumbens, with a trend toward reduced functional connectivity in the default mode, salience, and central attention networks.

None of that is a diagnosis. It is a signal — and a modest-sized one, in a single cohort of healthy volunteers — that the same arithmetic clinicians have used to flag midlife risk still tracks something real in the brains of older adults. The authors put it carefully, and so should we: the score might help identify cognitively higher-risk individuals later in life. That is the appropriate temperature for this finding.

The second study is a different animal: bigger, broader, and observational. Researchers drew on the Chinese Longitudinal Healthy Longevity Survey — 64,872 older adults across multiple waves from 2002 to 2022 — and measured cognitive impairment with a Chinese version of the Mini-Mental State Examination. Through 2018, across four survey waves, prevalence sat at an average of 4.3%. In the 2022 wave, post-COVID, it rose to 6.8%, and the trend held independently of gender, age band, and rural-versus-urban residence.

That is a meaningful step up, and it is the first published look at this question extending through the pandemic. It is also worth reading the same paper's quieter findings. The 2022 wave recorded a decrease in mean calf circumference — a rough proxy for muscle mass, and one worth its own column — alongside an increase in the proportion of overweight participants and, oddly, increases in daily fruit and vegetable intake and regular physical activity. The lifestyle inputs moved in a mixed direction. The cognitive output moved in one.

What does that suggest? Not, on its own, that COVID infection caused the rise. An observational study of this kind cannot make that case, and the authors do not. What it does suggest is that the pandemic era — illness, isolation, disrupted care, deconditioning, the whole package — landed on older brains in a way that shows up at the population level. Whether that bend in the curve straightens out in the next wave of data is, frankly, the question worth watching.

Read together, these two studies do not announce a breakthrough. They do something more useful: they sharpen the focus. The CAIDE work suggests the same modifiable risks that mattered at 50 are still leaving fingerprints on brain structure and function at 60-plus. The CLHLS work suggests the population-level picture got worse in the pandemic years, and that recovery is not automatic.

The practical implications are unglamorous, and they should be. Blood pressure remains the single most consequential number on the CAIDE sheet that you and a clinician can actually move. Lipids, weight, and regular movement are the other three. Education is not retroactively adjustable, but cognitive engagement in the years you have is — reading, conversation, problem-solving, novelty. None of that is a cure, and nobody serious is selling it as one. It is risk reduction, which is the only game on the table.

One caution worth naming: a risk score is a probability tool, not a diagnosis or a prognosis. A high CAIDE number is a reason to have a careful conversation with your doctor about the inputs you can change. A low one is not a permission slip. And neither study tells you anything about what to do about a specific symptom you are noticing this week — that is what an appointment is for.

The evidence here is moderate. One cohort is small. The other is large but observational. Both are pointing, with reasonable confidence, in the same general direction: the levers we already knew about still work, and the last few years have made pulling them more important, not less.

The headlines will keep moving — new biomarkers, new drugs, new scoring systems. The unchanged part is that the brain rewards the same boring habits the heart does, and that the years immediately around a major public-health shock appear to have cost something that the long-running data is now starting to measure. Worth knowing. Worth acting on, at whatever age you are reading this. And worth, as ever, a conversation with someone who knows your chart.