Letters to the Editor
Do Time-Series Studies Contain Residual Confounding by Risk Factors for Acute Health Events?
WW Consulting; Dayton, OH; firstname.lastname@example.org
To the Editor:
Recent papers in Epidemiology have reported on time-series studies comparing daily pollution levels and daily health events among a city or group of cities.1,2 However, such designs have a potential for uncontrolled/residual confounding that may not have been adequately considered.
Time-series studies are ecological in nature. As Morgenstern3 suggests, “ecological bias may be severe in practice” and “adjustment for extraneous risk factors may not reduce the ecologic bias” and may even increase it. The weather variables included in time-series models are themselves crude ecological surrogates for heat stress, and provide no information on important individual-level factors (eg, access to air conditioning). Such heat-related variables have a strong and nonlinear effect on health,4 which may not be adequately accounted for in linear time-series models.
Perhaps most importantly, it is not clear that time-series studies fully control for traditional confounding variables, such as smoking and stress. It is true that chronic risk factors such as smoking status and high blood pressure are not associated with daily pollution levels, and are therefore not important confounders in time-series studies. However, these chronic variables are also not pertinent to the daily risks (eg, death or hospitalization) of interest in time-series studies. That is to say, smoking status may predict someone's long-term probability of a heart attack, but the risk on any given day should be more dependent on the pattern of smoking that day.5,6 Similarly, stress and high blood pressure may increase overall risk of heart attack and stroke, but the risk on any given day would be predicated more on emotional stress during that day.7–10
These acute predictors of daily risk are plausibly associated with pollution through variables correlated with pollution. High-risk people might very well increase risky behaviors (eg, smoking, overeating, and anger) because of stress associated with high temperature, heavy traffic, increased occupational workload, or other variables known to be associated with air pollution.
In conclusion, time-series studies suggest very weak associations between air pollution and acute health events. For example, the largest US study reported an average relative risk for nonexternal all-cause mortality of 1.002 per 10 μg PM10,11 which implies less than 1% excess mortality over typical levels of pollution. By comparison, daily triggers of cardiovascular disease (which makes up much of all-cause mortality) are associated with relative risk as high as 6.0–15.0.9,10 Therefore, the potential for residual bias would seem plausible and worth investigating further.
1. Jerrett M, Burnett RT, Ma R, et al. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology. 2005;16:727–736.
2. Analitis A, Katsouyanni K, Dimakopoulou K, et al. Short-term effects of ambient particles on cardiovascular and respiratory mortality. Epidemiology. 2006;17:230–233.
3. Morgenstern H. Ecological studies. In: Rothman KJ, Greenland S, eds. Modern Epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven; 1998.
4. Armstrong B. Models from the relationship between ambient temperature and daily mortality. Epidemiology. 2006;17:624–631.
5. Gabbay FH, Krantz DS, Kop WJ, et al. Triggers of myocardial ischemia during daily life in patients with coronary artery disease: physical and mental activities, anger and smoking. J Am Coll Cardiol. 1996;27:585–592.
6. Giri S, Mitchel J, Kiernan FJ, et al. Cigarette smoking increases the magnitude of intracoronary thrombus observed during acute myocardial infarction [abstract]. Circulation. 2001;103:1362–e.
7. Strike PC, Steptoe A. Behavioral and emotional triggers of acute coronary syndromes: a systematic review and critique. Psychosom Med. 2005;67:179–186.
8. Strike PC, Perkins-Porras L, Whitehead DL, et al. Triggering of acute coronary syndromes by physical exertion and anger: clinical and sociodemographic characteristics. Heart. 2006;92:1035–1040.
9. Möller J, Hallqvist J, Diderichsen F, et al. Do episodes of anger trigger myocardial infarction? A case-crossover analysis in the Stockholm Heart Epidemiology Program (SHEEP). Psychosom Med. 1999;61:842–849.
10. Möller J, Theorell T, de Faire U, et al. Work related stressful life events and the risk of myocardial infarction. Case-control and case-crossover analyses within the Stockholm heart epidemiology programme (SHEEP). J Epidemiol Community Health. 2005;59:23–30.
11. Dominici F, McDermott A, Daniels M, et al. Revised analyses of the National Morbidity, Mortality, and Air Pollution study: mortality among residents of 90 cities. J Toxicol Environ Health A. 2005;68:1071–1092.
This article has been cited 1 time(s).
Regulatory Toxicology and PharmacologyCounterpoint: Time-series studies of acute health events and environmental conditions are not confounded by personal risk factorsRegulatory Toxicology and Pharmacology
© 2007 Lippincott Williams & Wilkins, Inc.