Determination of dose-response
functions in CEEHJakob H. Bønløkke, MD, PhD
CEEH and Dept. of Environmental and Occupational Medicine, Inst. of Public Health, Aarhus University, DK
Aarhus University in Århus
Use of D-R-functions
•Dose/Exposure/Concentration - response relationships are crucial if health risks are to be trusted - and evaluated
•Increased responses with increased doses, e.g. number of asthma attacks with increased ozone concentrations
•or mortality rates ...
Dockery et al NEJM 1993
TexteTexte
Adjusted Mortality-Rate Ratios and Pollution Levels in the Six Cities
Cohort studies
•Six Cities Study a prospective cohort study of 8000+ Americans with 15 years of follow-up using annual means of pollutants
•Important basis of many D-R-functions used in models since together with studies by Pope and colleagues on 500000+ Americans
Current estimates in CEEH
•Primarily based on cohort studies
•Example:
•Relative risk for all-cause mortality: 1.05 for a 10 microg/m3 increase of PM2.5 .
•“chronic death”
Time series studies•Estimates of short-term effects
•Correspondingly the D-R-function could be increase in mortality the day after an episode of high levels of a pollutant
•“acute deaths”
•By nature included in the long-term functions! (however poorly)
Where are people exposed?
• D-R-functions estimated outdoors at home
• People spent most time indoors
• and get extreme exposures during transport
• and at work
• By nature this too is reflected in the long-term function (however poorly because of “noise”)
• - insufficient evidence to make good separate estimates for separate locations
Other issues
•Harvesting?
•Thresholds?
•Combined effects of several pollutants
Does harvesting take place: Is an increase in mortality followed by
a decrease shortly after?
Schwartz Am J Epidemiol 2000
Is there a threshold belowwhich no effect of particles
onhealth has been observed?
Daniels NMMAPS Am J Epidemiol 2000
CEEH pollutants
•No indications of thresholds for any pollutant
•Ozone is poorly documented at low levels so currently it is being used as if there was a threshold at 70 microg/m3
Are associations with particles
sensitive to adjustment forco-pollutants?
Samet NMMAPS NEJM 2001
•D-R-function between particulate matter and mortality is robust to correction for other pollutants
•Not the case for all functions
•And there are separate effects of different pollutants
•Problem in estimating D-R-functions is that several pollutants are highly correlated
Not only mortality ...
•Respiratory symptoms
•Medication
•Absence from work
•Cancer
•Heart attacks, strokes,
•Loss of IQ, etc.
A matter of dose ...•Dose determines the response (?)
•Dose a function of
• concentration
• time
• uptake in body (usually via lungs)
• clearing from body
• uptake in target organ
Suggested estimates in CEEH
• RR for all-cause mortality: 1.05 for a 10 microg/m3 increase of PM2.5 .
• RR among survivors of myocardial infarcts: 1.30 for a 10 microg/m3 increase of PM10. (Zanobetti and Schwartz EHP 2007)