To the Editor:
I would like to add to the discussion of the article on air pollution health effects by Dominici et al.1 and Morfeld2 concerning distinctions between ecologic (populations) and cohort (individuals) studies. While the flaws pointed out by Morfeld2 are well known even if overlooked on occasion, it has seldom been realized that none of the information on air pollution exposures is specific to individuals. Whether measured at central monitoring stations, inferred from geographic or satellite data, or estimated from emissions, this information relates to places and not necessarily to their inhabitants. It is even less likely that such estimates would pertain to the small fractions of inhabitants that might be susceptible to such exposures. As a result, all health studies based on ambient air quality are “ecological” and as Morfeld points out, some may be more ecological than others.
Variations in these exposures stem from outdoor factors such as hot spots like major roadways or anomalies like green spaces.3 The contributions of indoor air quality are even more important4 especially as outdoor air quality improves. It may be argued that average personal exposures tend to be similar in magnitude to ambient exposures, but it has never been shown that individual personal exposures are equivalent for affected and nonaffected individuals. Moreover, I have shown that average personal exposures across cities are not correlated with the ambient data in each city.4
This situation arises from the two separate constituents of indoor air quality: infiltrated outdoor air and emissions from indoor sources, such as environmental tobacco smoke, gas stoves, fireplaces, candles, and pets. Infiltrated outdoor air is correlated with ambient air quality but at reduced concentrations; thus time-series analyses of health effects may be biased but still valid. However, because of intrinsic variations in indoor pollution sources within a community and across communities, cross-sectional (spatial) analyses are less likely to be valid. Indoor air quality may have been neglected in air pollution epidemiology because it is not regulated (and such studies have not been supported), but this does not diminish the need for realistic risk assessment.
These arguments point to the need to consider physical factors in addition to statistics when evaluating risks. Both population and cohort studies may be informative but both types suffer from uncertain exposures that limit the ability to discern thresholds.
Frederick W. Lipfert
1. Dominici F, Wang Y, Correia AW, Ezzati M, Pope CA 3rd, Dockery DW. Chemical composition of fine particulate matter and life expectancy: in 95 US counties between 2002 and 2007. Epidemiology. 2015;26:556–564.
2. Morfeld P. Re: Chemical composition of fine particulate matter and life expectancy in 95 US counties between 2002 and 2007. Epidemiology. 2016;27:e7–e8.
3. Vos PE, Maiheu B, Vankerkom J, Janssen S. Improving local air quality in cities: to tree or not to tree? Environ Pollut. 2013;183:113–122.
4. Lipfert FW. An assessment of air pollution exposure information for health studies. Atmosphere. 2015 6:1736–1752.