New HEI Report on Air Pollution Easy to Misinterpret

Published August 1, 2004

A study released June 16 by the Health Effects Institute (HEI) aims to improve our understanding of the relationship between very small particulate matter (PM) in our air and cardiovascular-respiratory human health. Unfortunately, the study may end up confusing more people than it enlightens.

A statement accompanying the report said, “the investigators found that for total mortality and cardiovascular-respiratory mortality there was no evidence of a threshold down to daily ambient concentrations of PM10 as low as 10 F:g/m3. Thus, they concluded that their findings indicate that linear models without a threshold are appropriate for assessing the effect of air pollution on daily mortality.”

Easy to Misrepresent

Unfortunately, environmental and health activists will find it easy to use this study to cast doubt on the belief–widely held among scientists and health experts and supported by ample research–that some threshold must exist below which pollution has no health impact. That belief is often summarized as, “The dose makes the poison.”

Saying there is no threshold below which exposure to PM10 is safe allows environmental activists to claim that even tiny amounts of PM, which is emitted by cars, trucks, factories, and power plants, can be assumed to “kill thousands of people” simply by extrapolating from animal experiments involving very high levels of exposure.

Journalists, policymakers, and concerned citizens should realize this is a misinterpretation of the HEI study. Worse still, characterizing the study this way ignores some of the study’s more important findings.

The study’s authors admit that the shape of the response curve at extremely low levels of exposure cannot be identified with certainty because of exposure measurement error, variation in city-specific concentration-response curves, and city-specific features of pollutant mixture and differences in health status of the populations.

The HEI panel that received the report correctly observed that the “lack of evidence against should not be confused with evidence in favor of the log-linear model. If the relation between PM10 and other cause mortality [i.e., causes of death other than cardiovascular-respiratory] was not log linear (as the investigators and Panel agree), then it is unlikely that the relation between PM10 and total mortality could be linear.”

Uncontrolled Biases

One year ago, HEI released a report titled, “Revised Analyses of Time-Series Studies of Air Pollution and Health.” That report pointed to serious shortcomings in data used to link air pollution with health effects. “Compared with randomized experimental studies in which the investigator controls the intervention, findings from observational studies (such as time series) are always susceptible to uncontrolled biases and must therefore be interpreted cautiously,” the HEI statement accompanying the 2003 study reads. “Observational air pollution and health studies are no exception.”

The 2003 study revealed a major error in previous time-series studies, an “inappropriate default convergence criteria in the GAM function.” When corrected, “the revised effect estimates for mortality typically decreased and standard errors increased. Across the 90 cities, the revised mean effect on mortality decreased substantially from 0.41% (increase per 10 F:g/m3 increase in PM10 concentration at lag 1) to 0.27% when using GAM with stricter criteria and to 0.21% when using GLM with natural cubic splines: an overall decrease of nearly 50%.”

In other words, the association between air pollution and health effects is only half as strong as researchers previously thought. The new research found smaller but still significant decreases in effect estimates for hospitalizations for cardiovascular diseases and for chronic obstructive pulmonary disease. The effect estimate for pneumonia hospitalizations was “substantially reduced.”

One would expect that a study finding the association between air pollution and human illness to be just half what we previously thought it was, and that the association at current levels is so weak it is probably impossible to determine whether a threshold exists below which there are no effects, would be newsworthy. Sorry, but I missed the headlines.

In its latest statement, the HEI panel “advises caution in drawing conclusions from the apparent absence of a threshold” and recommended additional studies in settings where the particulate matter concentrations are low. To which we can only say, “Amen!”

The primary problem with this excellent analysis is that it allows biased reporting of individual statements not supported by the totality of the study. This is precisely how activist groups fool the media, who in turn fool the public.

Jay Lehr, Ph.D., is science director of The Heartland Institute and editor of Wiley’s Remediation Technologies Handbook and other major science reference books. His email address is [email protected].