The PAISA Study
By DENIS BARD
Published: March 11, 2011
Dr. Denis Bard, a professor of Epidemiology within the Department of Epidemiology and Clinical Research at EHESP, is a physician by training. Dr. Bard found himself avidly interested in public health after a mission in Afghanistan with Médecins sans Frontières in 1981. Following a second mission in Chad, Dr. Bard was convinced that his professional future lay in this area. He then went on to obtain masters degrees in nutritional epidemiology and Public Health. Before joining EHESP in 1998, he worked for 6 years as a researcher in risk assessment at the French utility company where he contributed to the implementation of this type of public health approach in France. He also worked at the French Institute for Radioprotection where he founded its Epidemiology Laboratory. Dr. Bard is based in Rennes, Britanny.
In 2003, I realized that some diseases associated with exposure to environmental stressors also featured a socio-economic gradient. For the same diseases, environmental and social epidemiologists studied these two aspects independently. I felt that studying the interplay between socioeconomic position and exposure to environmental stressors might be of interest. I had to find the appropriate model, that is, diseases for which a socioeconomic gradient was convincingly established and deemed causally linked to an environmental stressor. In addition, the time to onset from exposure needed to be short: it was very tempting to study cancers, but the time to onset is several decades, which means that not only do the study subjects life course need to be traced back for at least 20 years- a difficult and costly task- but also there are essentially no data currently available to document environmental exposures for the past 20 years. So I decided to study asthma exacerbation in relation to ambient air pollution, which met the above criteria. The study design and the setting were shown to be very productive and I eventually launched another study with myocardial infarction as health outcome on the same basis.
This study had a case-crossover design, that is, the subjects are their own controls. A subject’s exposure on the hours preceding the health event is compared to exposure for the same days during the preceding weeks. In other words, if the case occurs on a Tuesday, the level of exposure (here ambient air pollutants concentration) on this day is compared to the levels of exposure for the 4 preceding Tuesdays. Exposure was assessed on an ecological basis, using an air dispersion modeling. Our studies had the particularity to be carried out at a very small geographical scale, with the aim of limiting exposure misclassification as much as possible, since a subject’s exposure was that of his/her area of residence. This area was comparable to the US Census Block, in average 2,000 persons. The socio-economic status was also that of the area of residence, assessed from an index specifically built for this study. Another interesting point is that we studied not only the asthma exacerbation episodes as retrieved from the emergency mobile services, but also the delivery of asthma specific drugs. The 2 health events studied yielded consistent results, with no apparent effect of socioeconomic position on asthma exacerbation risk for a given increase in ambient air pollution.
As a general comment, this work is still to be considered as exploratory. In addition, our results are at odds with the only study truly comparable to ours as regards to air pollution levels, exposure assignment, carried out in Vancouver (Canada) but with 4 times more subjects. These discrepant results might be due to the effect of an insufficient statistical power in our studies, but I think local specificities would give the most likely explanation. It remains that to make sense, the same kind of studies should be replicated elsewhere.
Although the field work and analyses ran perfectly, even above our reasonable expectations, this work and a growing literature point to the relevance of charactering the subjects living status (neighborhood context) to fully account for influences beyond simple social and mostly economic data such as those taken from census data, as we did. This work is currently in progress, and will test also the effect of a prolonged study period, with twice as many cases, increasing our statistical power. In addition, with the help of geographers, we designed a new, neighborhood data driven, aerial unit as an alternative to using a predefined spatial unit taken from census data. We expect this work to be completed by the end of the year.
For the above studies, I gained the help of Professor Severine Deguen, whose statistical skills were essential. I also worked with geographers, a discipline so far not represented within the EHESP. Currently, I’m in the process of opening a new, very different perspective, for which I plan to seek the involvement of global health colleagues, sociologists and of course statisticians. The overall idea is to develop disaster epidemiology.
The editors of PHIN Newsletter would like to thank Dr. Bard for his contribution. His recently published article, on which the study mentioned in this issue is based, is titled “Ambient air pollution, social inequalities and asthma exacerbation in Greater Strasbourg (France) metropolitan area: The PAISA study.” The full article can be accessed for free at www.intechopen.com.