Article Text

Download PDFPDF

Investigating the dose-response relation between air pollution and total mortality in the APHEA-2 multicity project
  1. E Samoli1,
  2. G Touloumi1,
  3. A Zanobetti2,
  4. A Le Tertre3,
  5. Chr Schindler4,
  6. R Atkinson5,
  7. J Vonk6,
  8. G Rossi7,
  9. M Saez8,
  10. D Rabczenko9,
  11. J Schwartz2,
  12. K Katsouyanni1
  1. 1Department of Hygiene and Epidemiology, University of Athens, Medical School, Athens 115 27, Greece
  2. 2Harvard School of Public Health, Environmental Epidemiology Program, Boston, USA
  3. 3Institut de Veille Sanitaire, Paris, France
  4. 4Universitat Basel, Institut fur Sozial- und Praventivmedizin, Switzerland
  5. 5Department of Public Health Sciences, St George’s Hospital Medical School, University of London, UK
  6. 6Department of Epidemiology and Biostatistics, Faculty of Medicine, University of Groningen, Groningen, Netherlands
  7. 7Institute of Clinical Physiology, CNR, Department of Epidemiology and Biostatistics, Pisa, Italy
  8. 8Research Group on Statistics, Applied Economics and Health (GRECS), Department of Economics, Universitat de Girona, Spain
  9. 9National Institute of Hygiene, Department of Medical Statistics, Population Studies Laboratory, Warsaw, Poland
  1. Correspondence to:
 Dr Katsouyanni
 Department of Hygiene and Epidemiology, University of Athens, Medical School, 75 Mikras Asias Str, Athens 115 27, Greece;


Background: Several recent studies have reported significant health effects of air pollution even at low levels of air pollutants, but in most of these studies linear non-threshold relations were assumed.

Aims: To investigate the NO2 mortality dose-response association in nine cities participating in the APHEA-2 project using two different methods: the meta-smooth and the cubic spline method.

Methods: The meta-smooth method developed by Schwartz and Zanobetti is based on combining individual city non-parametric smooth curves; the cubic spline method developed within the APHEA-2 project combines individual city estimates of cubic spline shaped dose-response relations. The meta-smooth method is easier and faster to implement, but the cubic spline method is more flexible for further investigation of possible heterogeneity in the dose-response curves among cities.

Results: In the range of the pollutant common to all cities the two methods gave similar and comparable curves. Using the cubic spline method it was found that smoking prevalence acts as an effect modifier with larger NO2 effects on mortality at lower smoking prevalence.

Conclusions: The NO2–mortality association in the cities included in the present analysis, could be adequately estimated using the linear model. However, investigation of the city specific dose-response curves should precede the application of linear models.

  • dose-response
  • total mortality
  • NO2
View Full Text

Statistics from


    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.