Crosswords and memory decline: what one long running study found

If you solve crosswords, here’s a structured, numbers-first summary of a real study (courtesy: crossword.guru) that looked at whether puzzles are linked to how memory changes right before dementia shows up.

The study in one minute

Study name: Association of Crossword Puzzle Participation with Memory Decline in Persons Who Develop Dementia

Cohort: Bronx Aging Study

Design: Observational, followed people over time (not a randomized trial)

Follow up cadence: Clinical + cognitive assessments every 12–18 months

Who they followed

488 community-dwelling adults were enrolled (initially cognitively intact)
For this analysis, they focused on 101 people who later developed incident dementia
Average baseline age in the 101-person dementia sample: 5 years
Follow up time to dementia diagnosis (mean): 0 years
Total follow up in this dementia sample: 505 person-years across 351 clinic visits

Crossword participation in the sample

17 people reported doing crossword puzzles at baseline (vs 84 non-puzzlers)
Frequency among the 17 puzzlers:
- 9 less than once/week
- 2 once/week
- 1 2–6 days/week
- 5 daily

The key result (the “when does decline speed up?” finding)

The researchers modeled memory decline as having a change point: a time before dementia diagnosis when decline starts accelerating.

Main headline

Crossword participation was associated with a 54-year delay in the onset of accelerated memory decline.

What that means in their model

Non-puzzlers: accelerated decline began 42 years before dementia diagnosis
Puzzlers: accelerated decline began 88 years before dementia diagnosis
Difference: 54 years later for puzzlers

So the sharp drop started closer to diagnosis for puzzlers, meaning they stayed in the “slower decline” phase longer.

What happened after the decline started (important nuance)

Once the accelerated decline began, puzzlers declined faster.

Rate of decline after the change point (Buschke SRT points per year)

Puzzlers: 32 points/year
Non-puzzlers: 01 points/year
Difference: 31 points/year faster for puzzlers after the change point

Rate before the change point

64 points/year (reported for puzzlers; the model treated pre-change slope as the same across groups)

Why this pattern matters

This “later start, faster drop” pattern is consistent with the cognitive reserve idea: you may function well longer, but once the brain can’t compensate anymore, decline can appear steeper.

Baseline differences worth knowing (so we don’t oversell it)

At baseline (within the 101 who later developed dementia), puzzlers tended to look “stronger” on some measures:

Mean SRT score at baseline:
- Puzzlers: 2
- Non-puzzlers: 6
Estimated Verbal IQ:
- Puzzlers: 2
- Non-puzzlers: 6

The authors did additional analyses (including matching on verbal IQ) and still saw a similar “delayed change point” pattern, but it’s still an observational study, so it can’t prove cause.

What you should take from this as a crossword solver

If you already do crosswords, this study is a pretty nice piece of evidence that the habit correlates with a later onset of accelerated memory decline (about 5 years later in this dataset).
It doesn’t prove crosswords prevent dementia.
But it supports the idea that crosswords are one of the most accessible “brain stretch” habits you can actually stick with because it’s fun.

Tiny action plan (keep it realistic)

Aim for consistency, not perfection: even “some crosswords” counted in this study (they used a simple “puzzler vs non-puzzler” baseline history measure).
Mix difficulty: do a few easier ones for flow, and one harder puzzle per week for the real workout.
Make it social: compare answers, trade clue logic, or do a puzzle together—engagement helps you stick with it.