King's College London

Research portal

What is the impact of systematically missing exposure data on air pollution health effect estimates?

Research output: Contribution to journalArticle

Evangelia Samoli, Roger D. Peng, Tim Ramsay, Giota Touloumi, Francesca Dominici, Richard W. Atkinson, Antonella Zanobetti, Alain Tertre, H. Ross Anderson, Joel Schwartz, Aaron Cohen, Daniel Krewski, Jonathan M. Samet, Klea Katsouyanni

Original languageEnglish
Number of pages6
JournalAir quality atmosphere and health
E-pub ahead of print2014

King's Authors


Time-series studies reporting associations between daily air pollution and health use pollution data from monitoring stations that vary in the frequency of recording. Within the Air Pollution and Health: A European and North American Approach (APHENA) project, we evaluated the impact of systematically missing daily measurements on the estimated effects of PM10 and ozone on daily mortality. For four cities with complete time-series data, we created patterns of systematically missing exposure measurements by deleting observations. Poisson regression-derived city-specific estimates were combined to produce overall effect estimates. Analyses based on incomplete time series gave considerably lower pooled PM10 and ozone health effects compared to those from complete data. City-specific estimates were generally lower although more variable. Systematically missing exposure data for air pollutants appears to lead to underestimation of associated health effects. Our findings indicate that the use of evidence from studies with incomplete exposure data may underestimate the impact of air pollution and highlight the advantage of having complete daily data in time-series studies.

View graph of relations

© 2020 King's College London | Strand | London WC2R 2LS | England | United Kingdom | Tel +44 (0)20 7836 5454