HEPA air filter in the bedroom reduced pre-diabetic sugar spike 17% at breakfast - RCT

New study suggests bedroom air purifier may support healthy blood sugar

Claude AI summary on the topic - Nov 11, 2026

Yes. It's an NYU Langone trial, presented November 8 at AHA Scientific Sessions 2026. A full paper doesn't seem to be out yet, so the details below come from the press release and the trial registration.

Who ran it: The lead author is cardiologist Sharine L. Wittkopp, MD, PhD, and the senior author is Jonathan Newman, MD, MPH. Another co-investigator was Robert Brook, MD, at Wayne State University, who has studied air pollution and metabolism for a long time. The Doris Duke Foundation funded it. The registration number is NCT05994937. NYU release

Design:

More than 150 men and women with prediabetes and no known atherosclerosis. They were randomized to real or sham portable air cleaners in their bedrooms in New York City for a month, and indoor air pollution was measured throughout. EurekAlert
It was double-blind. The sham was the same model, identical in look and sound, with the HEPA filter taken out. Units ran 24 hours a day for 4 weeks. Registry
Participants wore continuous glucose monitors for one week before and again after the four weeks of filtration. An AI tool picked out mealtimes from the glucose patterns so the team could focus on breakfast. The models adjusted for age, sex and BMI. News-Medical

Results:

Measure	Change with real HEPA	Ref
Fine particles	−36%	NYU
Daily glucose swings	About −8%	NYU
Total extra glucose after breakfast (iAUC)	About −17%	EurekAlert
Peak glucose after breakfast	−13%	EurekAlert

Sources: 36% particle cut and nearly 8% drop in daily swings; about 17% lower post-breakfast glucose and 13% lower peak.

What's still missing:

The registered primary outcome is the percent change in the CGM coefficient of variation (CV), so the "8%" is probably CV. The release doesn't name the metric, though, and gives no time-in-range data. There are also no confidence intervals or p-values.
Inflammation was a stated aim: the primary aims were the effect on glycemic variability and on circulating inflammation biomarkers. Yet the release reports no inflammation results. That's worth watching for in the full paper.
It doesn't give absolute PM2.5 levels, only the 36% reduction.
The authors themselves say the findings may not apply to people who already have diabetes, or to regions with different types or levels of pollution. It also lasted only a month, and the breakfast-only analysis looks like a secondary one.
Next, the team plans longer studies to see whether filtration can slow progression from prediabetes to diabetes and lower the risk of heart attack and stroke.

Why the morning effect makes sense: An earlier trial suggests a stress-hormone mechanism. In Shanghai, 55 healthy college students got real and sham purifiers in their dorms in random order for 9 days each, with a 12-day washout. Personal PM2.5 exposure averaged 24.3 µg/m³ with real filtration versus 53.1 µg/m³ with the sham. More PM2.5 raised cortisol, cortisone, epinephrine and norepinephrine, as well as insulin resistance and markers of oxidative stress and inflammation. Li 2017 Circulation

In an editorial on that trial, Brook and Rajagopalan pointed to earlier animal work showing that hypothalamic inflammation, through NF-κB signaling, plays a central role in the metabolic and blood-pressure effects of PM2.5. Editorial Breathing cleaner air all night could plausibly calm the stress axis that drives the morning cortisol rise, which would fit with the blunted breakfast spike. That link is my inference, not something the NYU team has said.

Sources:

Perhaps 10% of indoor PM2.5 air poollution is due to mcioroplastics

Claude AI The honest answer is that nobody knows yet. Published estimates run from under 0.01% to over 50% of indoor PM2.5 mass, and the spread comes mostly from how plastic was measured. In most mass-based studies plastic is well under 1% to a few percent of PM2.5. The one study claiming more than half is an outlier.

Study	Setting	Method	Plastic in PM2.5	Share of PM2.5	Ref
Hashemihedeshi 2024	1 home, 1 lab	Py‑GC×cIMS (heats the sample and identifies plastics by their breakdown products), 4 polymers	Home 16 µg/m³; lab 27 µg/m³	~50–60% (home)	JASMS 2024 Preprint
Rindelaub & Miskelly 2025	2 chemistry lab rooms	Py‑GC/MS	0.51 µg/m³	Not reported; ~3–10% if room PM2.5 was 5–15 µg/m³ (my arithmetic)	J Expo Sci 2025
Yakovenko 2025	French apartments, cars	Raman counting, ≥1 µm	528 particles/m³ (homes)	~0.1% or less (my rough conversion of particle counts to mass)	PLOS One 2025
Chen 2020; Liu 2019	Outdoor, urban/suburban China	Mass-based	—	0.01–0.3%	Vilnius review
Materić 2020	Outdoor, Netherlands (nanoplastics)	Mass-based	—	Likely <0.01%	Vilnius review
Leipzig 2024	Outdoor, urban Germany	Py‑GC/MS	0.33 µg/m³ (~70% tire wear)	Not reported	TROPOS abstract

The high outlier. Hashemihedeshi et al. sampled particles from <56 nm up to 18 µm, including nanoplastics. Their preprint said that 50-60% of the total particulate matter in the private residence was plastic, with polystyrene the most abundant. The peer-reviewed abstract keeps the 16 µg/m³ figure but drops the "over half of PM2.5" wording. That is one home measured with a new instrument. Plastic alone at 16 µg/m³ would be three times the WHO annual PM2.5 guideline of 5 µg/m³.

Why the numbers disagree so much:

Counting methods can't see small particles. Raman and FTIR (light-based identification) only detect particles down to about 1 µm, so nanoplastics are missed. Converting particle counts to mass gives tiny percentages.
The heating method can overcount. The Ma et al. 2025 method paper notes that standard Py‑GC/MS is often inaccurate because of false positive signals caused by similar polymers and organic compounds. A 2026 technical note found that ammonium sulfate can generate matrix interference in the quantification of AMNPs in PM2.5; ammonium sulfate is a major component of airborne particles.
Whether tire wear counts as plastic. Some estimates put 0.1–10% of PM10 and 3–7% of PM2.5 as tire and road abrasion, which reviewers consider an upper limit. That is outdoors; indoors it only matters through outdoor air leaking in.
The rest of indoor PM2.5 varies hugely. Cooking, candles, smoking and outdoor air dominate indoor PM2.5 mass. In a clean, tightly sealed home full of synthetic carpet and upholstery, plastic's share would rise.

No study has yet measured both total PM2.5 mass and plastic mass across a representative set of homes. The Vilnius school study measured indoor 4.3 ± 2.5 µg m⁻³ PM2.5, but it only counted plastic particles and didn't weigh them. Rindelaub measured plastic mass but not total PM. Until someone does both, my rough guess for a typical home is about 1% (range 0.1–10%). The >50% figure needs replication before it's cited as established.

Sources:

Related in VitaminDWiki

Fine particle pollution (PM2.5) worsened all 4 measures of Rheumatoid Arthritis activity - cohort
Alzheimer's risk increased 2X if high PM2.5 air pollution
Higher air pollution associated with more fractures and lower levels of vitamin D
Household air pollution increases the risk of pregnancy disorders, Vitamin D reduces the risk
The Evidence is strong: Air pollution seems to cause dementia (perhaps low Vitamin D)
Prediabetes 1.5 X more likely to go away if take Vitamin D – meta-analysis
Take Vitamin D to prevent prediabetes from progressing into diabetes – American Diabetes Association – 2024
Founder of vitaminDWiki has had 4 HEPA air filters running continuously in his house since ~2016. Together, they cost less than $100 per year, including replacing the filters or the units and the cost of electricity. He also has been running a robot vaccum during that time

Microplastics

More microplastics in 5 types of cancerous tissue than in non-cancerous tissue
Nanoplastics might result in more health problems than microplastics
Both are substantially captured by HEPA filters
Prostate Cancers have 2.5 X more microplastics than non-tumor sections
Microplastics and Nanoplastics associated with 8 Cancers