Nearly half of older adults struggle to fall asleep or stay asleep, and the usual suspects are well known: stress, pain, medication, an aging brain. A new analysis from southern China points to a less obvious contributor, one that arrives through the lungs rather than the mind. In more than 2,300 middle-aged and older adults, higher exposure to fine particulate air pollution (PM2.5) over the preceding months tracked with measurably worse sleep, and the study goes a step further than the standard questionnaire: it finds a chemical trail inside blood cells that may explain how invisible particles translate into restless nights.
The work, published in the Journal of Hazardous Materials, draws on GEHAS, a prospective cohort of adults aged 45 and older in the city of Guilin that has been enrolling residents since 2018. The 2023-2024 round of cohort exams was the first to include the Pittsburgh Sleep Quality Index (PSQI), the standard self-report instrument that scores everything from how long it takes to fall asleep to daytime dysfunction on a 0-21 scale. Of the 2,350 participants with complete data, 770 crossed the threshold for poor sleep, which the authors set at a PSQI above 7, a stricter cutoff than the usual 5 to account for the higher baseline scores typical of older adults.
A two-month window matters most
Rather than treating pollution as a single number, the analysis estimated each participant’s average PM2.5 exposure over six different windows, spanning one to six months before the follow-up visit. Daily exposure estimates came from the TAP platform (Tracking Air Pollution in China), a machine-learning system that merges ground monitors with satellite retrievals to produce one-kilometer-resolution maps, matched to each person’s residential address.
The association appeared across every window, but it was strongest for the two-month average. Each additional 10 micrograms per cubic meter of PM2.5 in that window was tied to a 1.33-point rise in PSQI score, sleep duration shrank by 0.79 hours, the odds of poor sleep quality more than doubled (odds ratio 2.17), and abnormal sleep duration became 59% more likely. The short-sleep pattern drove the duration finding: longer exposure meant higher odds of sleeping fewer than seven hours, while no significant link emerged for oversleeping. Sensitivity analyses, including adjustments for income, chronic disease, and anxiety, attenuated the estimates but kept the direction and the two-month peak intact.
The epigenetic bridge
The study’s core novelty sits in the molecular layer. Attention centered on 1,171 CpG sites mapped to 26 circadian rhythm genes among 586 participants with blood DNA methylation profiles. Methylation, a chemical modification that can switch gene activity up or down, is one way the environment leaves lasting marks on the genome.
Associations with sleep quality surfaced at 57 CpG sites, of which 47 survived bootstrap validation and 11 were distilled into a single sleep-related methylation risk score (MRS) using LASSO regression. The score behaved as a coherent biological signal: it correlated with the PSQI and its components more strongly than any single site, and the genes it touched formed a connected functional network that binds regulators including NPAS2, BMAL1 and DEC1 to the clock-related kinases CSNK1E, CSNK2A1 and CSNK2A2.
The mediation analysis is where the pieces come together. Five of those sites, falling within NPAS2, PRKAG2, RORA and CSNK2A2, each accounted for roughly 9% to 16% of the association between PM2.5 and sleep quality. The composite MRS captured more, explaining about a quarter (25.03%) of that association, and the score itself rose with pollution exposure, climbing 8.55% per 10-microgram increase in PM2.5. That pattern suggests the effect of polluted air on sleep runs at least partly through coordinated epigenetic disruption of the circadian network, not through a single gene acting alone.
Why it matters
The finding adds sleep disturbance to the already long ledger of harms linked to fine particles, which the World Health Organization classifies among the most serious environmental health risks. For aging populations, where poor sleep is both common and tied to cognitive decline and faster biological aging, an environmental lever is significant because it is modifiable: unlike genetics, air quality can be improved by policy and individual behavior.
The methylation score may also have predictive value. Participants with higher sleep MRS showed accelerated epigenetic aging, equivalent to about 0.9 to 1.0 extra years on two widely used biological clocks (Hannum and Horvath), and the score tracked aging better than the questionnaire-based PSQI did. The authors are careful to note that the score is not yet a validated biomarker, but it hints at a future where a blood test could flag people whose sleep is most vulnerable to pollution before the sleepless nights accumulate.
Limits
The study is cross-sectional, so it cannot establish that pollution caused the sleep problems, only that the two move together. Exposure was assigned from residential addresses rather than personal monitors, which introduces misclassification, and the single-region sample limits generalizability. Residual confounding, for instance from environmental noise, cannot be fully excluded. Finally, the methylation signatures were validated internally through resampling but not yet in an independent cohort, so their stability across populations remains unproven.
Bottom line
For adults at and beyond middle age, fine particulate pollution is associated with worse sleep, and methylation changes in circadian clock genes appear to be part of the biological bridge. The strongest signal came from the two months before sleep was measured, a window worth remembering for anyone studying, or living with, the health effects of dirty air.
Source
Wang Y, Liu S, Rong J, et al. Ambient PM2.5 exposure and sleep quality: Evidence from circadian DNA methylation signatures in a population-based study (GEHAS cohort). J Hazard Mater. 2026;515:143121. doi:10.1016/j.jhazmat.2026.143121. PMID: 42537304.