Long-term PM2.5 linked to worse Covid outcomes

Air pollution was suspected early in the pandemic of worsening COVID-19, but much of the supporting evidence compared regions rather than people. Studies of that kind are vulnerable to ecological bias: an area with dirtier air also differs in density, deprivation, age structure and testing behaviour, and the pollution signal can absorb all of it. Whether the association survives when researchers follow individuals with their own exposure estimates and their own outcomes is the question that decides how seriously to take it. It also matters beyond COVID-19, as a test of whether chronic pollution exposure shapes infectious disease severity.

The authors systematically searched MEDLINE and Scopus in July 2023, updating the search in April 2025, for studies linking PM2.5, nitrogen dioxide and ozone to SARS-CoV-2 infection, COVID-19 hospitalisation and COVID-19 mortality. They deliberately restricted the synthesis to cohort studies carrying individual-level data. Risk of bias was assessed with a modified version of a World Health Organization instrument. All eligible studies were synthesised qualitatively, and random-effects meta-analyses were run where more than three studies existed for an exposure-outcome pair, after removing studies with overlapping populations.

Long-term PM2.5 exposure was associated with a raised risk of every outcome examined: relative risks of 1.04 for infection, 1.11 for hospitalisation and 1.09 for mortality, each per 1 µg/m3 increase. Nitrogen dioxide was associated with COVID-19 hospitalisation and mortality, with relative risks of 1.02 and 1.01 per 1 µg/m3. Ozone exposure showed no association with any of the three outcomes. Univariate meta-regression indicated that the country in which a study was conducted accounted for a substantial share of the heterogeneity between estimates.

By excluding ecological designs, this synthesis addresses the main criticism levelled at earlier pooled analyses, and the associations largely persist. That strengthens the case that long-term particle exposure is relevant to infection risk and severity rather than being a proxy for where people live. It does not establish causation, and the strong country effect in the meta-regression signals that healthcare systems, testing regimes and exposure ranges differ enough to complicate a single global number. The null result for ozone leaves open whether that pollutant matters through other pathways.

The practical reading is that chronic exposure to fine particles appears to sit alongside the more familiar risk factors for severe respiratory infection, with the estimated increases per unit of pollution being small but consistent across outcomes. These are population averages from observational cohorts, not a prediction for any individual, and they do not describe what happens if your own exposure changes. For anyone thinking about where they live or how they ventilate indoor spaces, this adds infection outcomes to the list of things air quality has been linked to.