Cardiovascular and respiratory effects of air pollution : application of different observational study designs and analysis approaches

Abstract

The effect of air pollution exposure on human health has been a topic of an increasing volume of research over the past half century. Due to the ubiquitous nature of ambient air pollution exposure and the comparatively weak effects that can be expected, one of the most viable and most utilized avenues of research has been to conduct large scale observational studies. This thesis presents four such studies covering several health outcomes in a variety of populations using a number of different observational study designs and analysis methods. The overarching aim of the thesis was to elucidate challenges that come with observational research on health effects of air pollution and to assess how the studies and their designs address those challenges. In addition, the individual works were described, interpreted, and assessed on the basis of how they met their respective objectives.

BAMSE is an ongoing birth cohort in Stockholm of over 4,000 children who were recruited between 1994 and 1996 and are followed with respect to their respiratory health and development of allergies, as well as their environmental exposures. An association was seen between exposure to traffic generated air pollution during their first year of life and early onset persistent wheezing during their first four years of life, as well as lung function (measured by peak expiratory flow) and allergic sensitization at age 4. HEAPSS was a European five-center register-based study of more than 25,000 first-time acute myocardial infarction (MI) survivors who were followed with respect to re-infarction and mortality, linking the health events to the daily air pollution levels. Daily all-cause mortality among the MI survivors was linked to increased levels of both ultrafine and larger particles during the days preceding the deaths.

ALVA was a study of 211 patients in Stockholm and Gothenburg with implantable cardioverter defibrillators, devices implanted in the chest of the patients and designed to recognize and treat ventricular arrhythmias. Since the devices record the exact time when they have treated an arrhythmia it was possible to very precisely determine the time of an event and link it to the air pollution levels immediately before that time. An association was indicated between particulate air pollution levels in the two hours immediately preceding the event and the onset of ventricular arrhythmia. The Swedish Onset study comprised 660 first-time MI survivors in Stockholm and was designed to identify triggers for the onset of MI through very detailed interviews conducted shortly after the event about the time leading up to the MI. This allowed for very accurate determination of the exact time of the events and that information could be linked to the air pollution levels preceding that time. We did not observe any association between air pollution exposure and the onset of first-time non-fatal MI.

Major challenges included study design, exposure and outcome assessment, potential confounding and selection bias, as well as study power and analytical strategies. All four studies used state-of-the-art designs and were conducted with great care for data quality and effort was made to rule out that the observed results were due to misrepresentation of exposures or outcomes, uncontrolled confounding, selection bias, or other unintended effects. All four studies had sample sizes that were comparatively small, which yielded rather imprecise estimates of the weak air pollution effects, and each study have some results that cannot easily be explained. In summary, all four studies met their respective objectives to assess how air pollution exposure affect the respective health outcomes and were designed, conducted, analyzed, presented, and interpreted in a manner that allow them to effectively contribute to the collective scientific evidence of health effects of air pollution.