Effects of air pollution on health

Abstract

Quantifying the effect alone is not enough to change policy and improve health

The term “smog”—a combination of smoke and fog—was invented by a British doctor a century ago. In 21st century Europe, air pollution has greatly improved by most measures but is still a substantial health problem, responsible for the early deaths of hundreds of thousands of people each year.¹ Estimates of mortality attributable to long term exposure to fine particles are now widely accepted as a key policy indicator of the effect of air pollution. A draft UK government report, written by the Committee on the Medical Effects of Air Pollution and currently open for public review, re-examines the scientific evidence underpinning these estimates.²

The most directly applicable evidence is provided by follow-up of large populations exposed to different long term average levels of air pollution. Findings of the first cohort studies by the American Cancer Society have been confirmed and extended by additional years of follow-up³ and extensive reanalyses,⁴ and with cohort studies in Canada⁵ and Europe.^{6 7} Studies of cities that have experienced substantial reductions in exposure to air pollution as a result of policy changes (such as Dublin⁸ and Hong Kong⁹) provide convincing evidence that the effects are real and not the result of unmeasured confounding.

Time series studies, although not suited to assessing effects of long term exposure,¹⁰ provide further evidence that day to day changes in air pollution have measurable health effects at population scale, even at very low levels of ambient exposure. Slowly varying factors such as socioeconomic status or smoking history, which might bias the results of cohort studies, cannot explain the short term associations between air pollution and health seen in time series studies.

The best estimate from the Committee on the Medical Effects of Air Pollution, based on the American Cancer Society's follow-up study,³ is a 6% (95% confidence interval 2% to 11%) increase in all cause mortality per 10 µg/m³ increase in fine particle exposure (measured as PM_2.5). The authors also report a probability distribution of a “plausible” dose response (range 0-15%). This was based on a Delphi survey that used the expert judgment of committee members to capture uncertainties in the coefficient. The committee recommends using Monte Carlo analysis to sample from the plausible dose response range when applying their findings to assessment of air pollution policy. In practice, the use of statistical confidence intervals would probably give closely similar results, as the distribution of plausible dose response coefficients is clustered within the range of the statistically derived confidence intervals.

The report by the Committee on the Medical Effects of Air Pollution is well argued and supported by working papers discussing the key uncertainties, although a few areas warrant more discussion. Firstly, the American Cancer Society's study excluded those under age 30 years,³ and so the results do not apply to younger adults and children. Other studies that have linked infant mortality with exposure to air pollution^{11 12} are included in a table but not discussed in the main text. Secondly, most epidemiological studies of air pollution have been conducted at the scale of whole cities. More recent studies, conducted at a smaller geographic scale, have reported substantially larger effects.^{5 6 13} These studies have important implications for estimates of years of life lost.

Finally, evidence that the effects of air pollution are modified by social factors¹⁴ should be reflected in health impact assessments. In the American Cancer Society's study, the effects of air pollution seemed greater in those with lower educational status. The distribution of effects between rich and poor communities is important from a social justice perspective, independently of the overall effect on health.

Quantifying the direct effect of air pollution on mortality is an important step towards effective interventions to improve health. However, policy makers need a wider perspective than that provided by the Committee on the Medical Effects of Air Pollution so that they can take into account the indirect, long term health implications of energy and transport policies.

Global climate change will have major effects on health that are difficult to measure using conventional approaches that look at individual aspects rather than the problem as a whole. Coal fired power stations and vehicles are major emitters of carbon dioxide, the main driver of climate change. Energy efficiency measures can have simultaneous benefits through reductions in local air pollution (particulates) and global carbon dioxide emissions. It is becoming increasingly clear that we need to stop burning fossil fuels and that we cannot afford another decade of scientific enquiry, let alone another century, before we act. Rigorous scientific assessments—exemplified by the Committee on the Medical Effects of Air Pollution's report—have an important place but are only part of the policy equation.