What we measure, and why it matters.

In a study of 3,977 US homes with indoor sensors, most of them in California, Washington and Oregon, researchers compared indoor readings with outdoor sensors within 5 km. In the typical home, about half of the indoor fine particles traced to outdoor air and about half to indoor activity. Homes varied a lot: the indoor share ranged from roughly one-sixth to three-quarters. Indoor spikes were most frequent at mealtimes and on Thanksgiving and Christmas, which points to cooking. On average, indoor levels were lower than outdoor levels, and that held in 92% of the homes.

In a small pilot trial funded by an air purifier maker, 30 healthy UK adults each used a bedroom air purifier for 2 weeks with a real HEPA filter and 2 weeks with a placebo filter, without knowing which was which. With the real filter, people reported about 19 more minutes a night in bed. Their diary-reported sleep time was about 12 minutes longer, but that difference fell just short of statistical significance. Wrist-watch data showed more time awake during the night with the real filter, and most other sleep and mood measures did not change. Bedroom particle levels were already low in both conditions.

Short trials of home air cleaners, mostly lasting days to a few weeks, have found small average drops in systolic blood pressure. A 2021 review of 14 trials found an average drop of about 2.3 mmHg and rated the overall certainty of this evidence as very low. These trials measured short-term markers such as blood pressure, not heart attacks, strokes or deaths.

In lab tests of four home range hoods, the share of cooking particles a hood captured varied widely with the cooking setup and fan speed. Frying a hamburger at medium heat on a back burner, the hoods mostly captured more than 80%. Stir-frying green beans in a wok at high heat on a front burner, they captured only 3% to 11% of 0.3 to 2.0 µm particles on low fan settings and 16% to 70% on high settings. The dish, heat and burner all changed between the two tests.

In a lab analysis of 116 healthy adults aged 18 to 30, spending the 8 hours before bed in ordinary room light (under 200 lux) instead of dim light (under 3 lux) made the nightly melatonin rise start later in 99% of people and shortened the melatonin night by about 90 minutes. The study measured melatonin, not sleep or health.

In a study of 552 older adults (ages 63 to 84) in the Chicago area who wore a wrist light sensor for a week, the 54% who had any measurable light during their darkest five hours were more likely to have obesity, diabetes and high blood pressure. The study captured one point in time, so it cannot show which came first; health problems such as nighttime bathroom trips could themselves lead to more light at night.

In about 89,000 UK adults (average age 62) who wore a wrist light sensor for one week, people in the brightest 10% for nighttime light had about a 21% higher risk of dying over roughly eight years than the half of people with the darkest nights. People in the brightest 10% for daytime light had about a 17% lower risk than the half of people with the darkest days. These figures come from the models adjusted for the most social and lifestyle factors. This is a link, not proof that light was the cause.

In a study of 109 US federal office workers who wore a light meter for a week, those who got high levels of body-clock-relevant light in the morning reported falling asleep faster (especially in winter) and sleeping better than those who got low levels. It was an observational study in offices, and sleep was self-reported.

Reviews of mostly small lab studies find that brighter or bluer-white daytime electric light tends to make people feel more alert, but the effect depends on the time of day and how much sleep pressure a person has. Effects on complex thinking are inconsistent.

After Flint switched to a more corrosive water source without adequate corrosion control, the share of tested children under 5 with blood lead at or above 5 µg/dL rose from 2.4% to 4.9%. In the three wards with the highest water lead, it rose from 4.0% to 10.6%. Outside the city, where the water source did not change, the rise was not statistically significant. The study compared neighborhoods and did not match each child to their own tap. A later analysis of 15,817 Flint blood lead tests from 2006 to 2016 found that the 2014 to 2015 rise happened within a long decline in children's blood lead (Gómez et al., Journal of Pediatrics, 2018).

Arsenic can get into well water from natural rock, and long-term exposure has been linked to bladder cancer. A study in Maine, New Hampshire and Vermont compared 1,213 people with bladder cancer and 1,418 without. People in the top 2.5% of estimated lifetime arsenic intake from drinking water had about twice the odds of bladder cancer (odds ratio 2.24) of those in the lowest quarter. This is an association from an observational study, and it does not show what risk, if any, exists at or below EPA's 10 ppb limit.

EPA's nitrate limit (10 mg/L, measured as nitrogen) was set to protect babies under six months from blue-baby syndrome, and other health effects were not considered when it was set. A 2018 review of more than 30 newer studies found the strongest evidence beyond that for colorectal cancer, thyroid disease and neural tube defects, and many studies saw higher risk below regulatory limits. These are associations, not proof of cause.

The nitrate evidence is mixed. A 2022 meta-analysis found no overall link between drinking-water nitrate and colorectal cancer across 10 studies, and found an association with stomach cancer across 4 studies (odds ratio 1.91 per 10 mg/L of nitrate, a different unit from EPA's limit, which is measured as nitrogen). These are associations, not proof of cause.

Trihalomethanes form when the disinfectants used to treat tap water react with natural material in the water. A pooled analysis of six case-control studies (2,806 people with bladder cancer, 5,254 without) looked at 40 years of exposure. Men whose long-term average exposure was above 1 µg/L had 24% higher odds of bladder cancer, and those above 50 µg/L had 44% higher odds. No link was seen in women. These are associations, not proof of cause.