Can We Please Stop Asking if Crossword Puzzles Prevent Dementia?

In the pantheon of lay press health articles that are sure to make me crazy are ones about the supposed benefits of vitamin D, the risks (or benefits depending on the month) of coffee, and the most effective ways to exercise. Lording over this pantheon, however, are articles about how brain exercise can ward off dementia. These articles are guaranteed churnalism and reliable for proving Betteridge's law of headlines.

Before diving into the genre, let’s look at one recent example from the Washington Post, “Can a daily crossword puzzle slow cognitive decline?”[i] Now I have to say, this isn’t a bad article about a study that, although essentially meaningless, is better than most. The study in question was actually a randomized trial. Patients with mild cognitive impairment, generally referred to as an intermediate clinical state between normal cognition and dementia, were randomized to 12 weeks of intensive, home-based, computerized training with cognitive games or to computerized crossword puzzles. The primary outcome in the study was a change from baseline on the Alzheimer’s Disease Assessment Scale-Cognitive (ADAS-Cog) score.

The results of the study were underwhelming. On the 70-point ADAS-Cog (on which higher scores are associated with worse cognition), scores worsened for people assigned to the games (9.53 to 9.93) and improved for those assigned to crosswords (9.59 to 8.61[ii]). The WaPo article does not overstate the results article but it does commit many of churnalistic sins:

“While the study didn’t investigate whether crosswords benefit younger adults who are not dealing with cognitive decline, it suggests that keeping your mind active as you age may benefit your brain. And the research offers hope to those diagnosed with mild cognitive impairment that they may be able to stave off further declines in the memory, language problems and decision-making that are the hallmark of the condition.”

We can certainly accuse this paragraph of Sin #2 Extrapolation and generalization, Sin #4 Neglecting plausibility, and Sin #5 The Disclaim and Pivot maneuver, as well as mistaking statistical significance for clinical significance. But, as I said, this is one of the good articles covering one of the better studies on the topic. Most articles published in the lay press cover observational trials which show that people who are keeping their minds active – reading, playing cards, doing puzzles – have less dementia than those who don’t. Let’s enumerate a few of the reasons we should be suspicious of these articles.

1.       Little bio-plausibility. As we learn more and more about dementia in general, and Alzheimer's in particular, we realize the complexity of the disease. In Alzheimer’s there is an overproduction and/or decreased clearance of amyloid beta peptides. The pathogenesis also involves the hyperphosphorylation and aggregation of the tau protein. The more we learn, the less likely it seems plausible that puzzle-doing would prevent disease.

2.       Confounding. Doing puzzles or other challenging mental activities is unpleasant for people with cognitive impairment. A person with “premorbid” symptoms of dementia, even extremely mild ones, will shy away from these activities. In an observational study this will cause confounding. People with dementia will not have been puzzlers, not because puzzling prevents dementia but because premorbid dementia prevents puzzling.

3.       Terrible endpoints. If we are attempting to prevent dementia, we would like to know that we actually are. This means showing that an intervention actually preserves people’s independence. Showing that they can do better on a test, one that the intervention group is essentially practicing for, is not an adequate endpoint.

4.       Publication bias. Remember that the entire medical literature is probably biased toward positive findings. This is especially true for small studies. Large studies that have looked at this topic have been negative or profoundly unimpressive (1, 2 and 3).

5.       Sin #6  Keep testing; report just once. I have no proof that this sin applies to this literature but I would be surprised if it did not. In a literature filled with observational studies of clinically insignificant and barely statistically significant results, there is a high likelihood that multiple testing or p-hacking is happening.

Dementia frightens us all. Many of us have experienced a friend or loved one fading away (sometimes quietly and sometimes excruciatingly. The prevalence of the disease is frighteningly high and increasing. We also still have little to offer patients in the way of medical treatment. We all wish there was something simple we could do to prevent dementia[iii]: a crossword puzzle a day; a couple of sudoku, a word finding game. Unfortunately, all signs point to this not being the case and another lazy piece of churnalism reporting on a small, confounded, observational study won’t change that.

Although I would not recommend wasting money on this study, if we wanted to design a good study we could. Randomize people in their 50’s to a number of different cognitively stimulating activities or a non-intervention control. Enrich the population with people with a first degree relative with dementia. Follow the patients for two decades with endpoints being a clinical diagnosis of dementia or institutionalization. I’m pretty sure we would find no benefit.

I hope only hope we good therapy by the time the study ends.

---

[i] You’ll never guess the answer to that question.

[ii] I actually had to review the article a few times to convince myself that the benefit was, if fact, as small as it seemed.

[iii] We do know that somethings prevent dementia. Mostly, over the course of a lifetime, things that help the heart help the brain. This is because a a sizable minority of people with dementia have vascular dementia. Don’t smoke, eat well, exercise regularly.