Air purifiers sped a brain test only after researchers sorted by age

Across 119 people living beside highways in Somerville, Massachusetts, a month of HEPA filtration made no overall difference on a timed connect-the-dots test. When the researchers split the group at age 40 after the fact, the 50 older participants finished the harder half about 7 seconds faster. That's about the gain from one retake of the test, though the design balances practice across months. Brain function in daily life wasn't measured.

5 October 2026 · How we rate evidence

For two months of a cool season, 119 people in Somerville homes lived with a pair of boxy air purifiers humming in the living room and bedroom. For one of those months the purifiers were real. For the other, the filter had been quietly taken out, leaving a fan that sounded and blew almost the same. Participants weren't told which month was which, and most couldn't tell.

The claim going around. "HEPA air purifiers may boost brain function after just one month," ScienceDaily announced on 30 August. The text is a reprint of an article three of the study's authors wrote for The Conversation in April, which opens: "Using an in-home HEPA purifier for one month spurs a small but significant improvement in brain function in adults age 40 and older." European Medical Journal ran "HEPA Air Purifiers Improve Cognition in Older Adults" in September, though its story did report the null overall result. The paper itself is more careful than either headline: it reports the null overall result and labels the age split post hoc.

What they did. The trial, called HAFTRAP and run by Doug Brugge's team at the University of Connecticut, was built to test blood pressure. Somerville sits against Interstate 93, and the homes were all within 200 metres of a major highway. In the 19 homes where air was monitored, the real purifiers roughly halved indoor fine particles, from 5.2 to 2.5 micrograms per cubic metre.

At the start and end of each month, the trial manager also gave each participant the Trail Making Test: connect numbered circles in order as fast as you can (Part A), then alternate numbers and letters, 1-A-2-B and so on (Part B). Part B is the harder one, and is used as a measure of mental flexibility. The paper gives 75 seconds as average, from norms for people 55 and older; over 273 seconds counts as deficient. These participants averaged a little over a minute at the start.

What they found. Across all 119 people, the purifiers made no clear difference. Part A times were about 20 seconds either way. Part B averaged 53.6 seconds after the real month and 55.8 after the sham month, a gap small enough to be chance.

The researchers then split the group at age 40. For the 50 people aged 40 or older, Part B took 54.0 seconds after a real month and 61.4 after a sham month. That's 7.4 seconds faster, a 12% relative difference, and the statistical test said the older and younger groups responded differently. The 69 people under 40 were, if anything, slightly slower after the real month (53.7 against 51.4 seconds, not a clear difference), and Part A showed nothing in either age group.

Two things here are done well: the sham kept participants blind, and the real purifiers roughly halved indoor fine particles.

What it doesn't show. The first limit is the rank of the finding. Cognition wasn't the trial's question. It isn't among the outcomes listed in the trial's public registration or its published methods paper. The age split came later still: in the paper's own words, the age analysis was "post hoc." The authors give reasons for picking 40, including a review that found the clearest evidence of pollution's effects on thinking in studies of adults over 40. But the choice was made with the data in hand, and it produced the only clear result out of six comparisons (two parts of the test, each overall and in two age groups).

The second limit is the size of the effect next to the test's own quirks. People get faster at the Trail Making Test simply by doing it again. In this study, participants who had the sham month first went from 70.0 to 62.4 to 57.8 seconds on Part B over their first three attempts, with no real filtration involved. The 7.4-second purifier effect is about the size of that first retake. The analysis adjusts for each person's starting time and for which month came first, and because every participant had both months, practice is balanced across real and sham. What the paper doesn't report is the first month alone, the check that sidesteps practice and carryover entirely.

Third, the person who ran and timed the test knew which purifier was in the home. The authors argue the test is standardized and the participants were blind. That helps, and a tester's bias would be expected to show in both age groups, yet only the over-40s moved. Still, a few seconds of stopwatch timing or encouragement is exactly the scale of the effect.

And 37 of the 156 people enrolled were left out, mostly because forms went missing. The 11 who declined some tests were older and slower, and 9 of them were over 40. One of the trial's two registered primary outcomes, blood pressure, showed no overall difference either: 0.5 mmHg, reported in JACC last year, with a subgroup benefit in people whose pressure started at 120 or above.

What it could mean for people. Could cleaner air at home help people think more clearly, especially as they age? It's a fair question. Air pollution has been tied to worse cognitive test scores in many observational studies, and HEPA purifiers clearly cut indoor particles. This trial shows the test is feasible at home and gives a number to aim at. Neither it nor the blood-pressure paper reports on harms or side effects. What it can't do is tell a 45-year-old that a purifier will sharpen their thinking. In The Conversation, the authors compared the gain to the benefit of more daily exercise; the paper doesn't test that comparison. They also noted that fewer than 10 participants were over 60, the group most worried about decline.

Why it's still interesting. Most evidence linking air pollution and thinking comes from comparing people who live in dirtier and cleaner places, who differ in many other ways. This is one of a handful of trials that change the air and hide which months are real, in ordinary homes where people went about their lives. The design is the right one. It now needs a trial built for the question, with the age group and the test named in advance.

Open questions.

Was the over-40 result real or a chance find? Run a trial with cognition as the main outcome, the age groups fixed beforehand, and enough people to detect a 7-second difference. If it holds, a cheap home device would have a measurable effect on thinking; if not, this becomes one of many subgroup leads that faded.

How much of the gap is practice or the tester? Repeat it with alternate versions of the test, which blunt practice gains, and testers or computer timing that don't know which purifier is running. The answer tells us whether a few seconds on a stopwatch can carry claims about brain function.

Would a longer or older trial show more, or nothing? Follow people over 60, or with mild memory problems, for months rather than weeks, with fuller cognitive testing. That's the group that would buy purifiers for their brains, and the one this trial barely included.

A note on timing. The paper came out on 16 April 2026, and the authors' Conversation article followed within a week. ScienceDaily's reprint on 30 August, four and a half months later, set off a second round of coverage. The study hadn't changed.