Wildfire particulate carried more risk per microgram than ordinary air pollution

Design

retrospective registry-based cohort with time-varying Cox proportional hazards modelling

Population

414,016 patients aged ≥65 with lung cancer in SEER-Medicare, 2008–2019, 1,211,490 person-years

Primary outcome

all-cause mortality by 3-year moving average PM2.5 exposure

Effect

per 1 μg/m³: wildfire PM2.5 HR 1.0782 (95% CI 1.0641–1.0924); non-wildfire HR 1.0171 (1.0149–1.0192)

This retrospective cohort linked 414,016 patients aged 65 or over with histologically confirmed lung cancer in the SEER-Medicare database between 2008 and 2019 to modelled residential estimates of wildfire-derived and non-wildfire fine particulate, aggregated to postcode and analysed as three-year moving averages in a time-varying Cox model. Follow-up totalled 1,211,490 person-years, median 2 years.

Median concentrations were 0.28 μg/m³ for wildfire-related PM2.5 and 8.11 μg/m³ for the rest. Per microgram, wildfire particulate carried substantially more risk: hazard ratio 1.0782 (95% CI 1.0641 to 1.0924) against 1.0171 (1.0149 to 1.0192) for non-wildfire particulate. Because exposure to wildfire smoke is so much lower, the totals reverse — an estimated 696 average annual excess deaths (576 to 814) were attributed to wildfire particulate against 3,465 (3,061 to 3,864) to the rest.

That pair of findings is the whole story. Wildfire smoke is the more toxic material per unit and the smaller contributor at current levels, and it is the fastest-growing source. For oncologists outside fire-prone regions the transferable point is the second hazard ratio: ordinary ambient particulate, at the levels found in the contiguous United States, was associated with shorter survival after a lung cancer diagnosis. Indian urban PM2.5 routinely runs several times the 8.11 μg/m³ median here, which makes this a question about the air our patients go home to rather than a North American one.

• Ask about home and occupational air exposure as part of a lung cancer history
• Advise practical mitigation during high-pollution periods: indoor filtration, avoiding outdoor exertion
• Do not present air quality as a modifiable prognostic factor for an individual — the evidence is population level
• Read the non-wildfire hazard ratio as the one relevant to most Indian patients

Why it matters

It puts the air a patient returns to after treatment into the same frame as the treatment itself.

Don't overread it

Observational and exposure-modelled at postcode level — it shows association, not that cleaner air would have extended these patients' lives.

The statistics, in plain English

A hazard ratio of 1.0782 per microgram is small per unit but compounds across the range of exposure — and the intervals are narrow because the cohort is enormous, not because the effect is large. This is observational: postcode-level modelled exposure cannot capture where an individual actually spent their time, and residual confounding by the socioeconomic factors that determine where people live is the main alternative explanation. The excess-death figures are model-derived attributions, not counted deaths.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.