Article Text


Air pollution and mortality in a cohort of patients with chronic obstructive pulmonary disease: a time series analysis
  1. Judith Garcia-Aymericha,
  2. Aurelio Tobíasa,
  3. Josep Maria Antóa,b,
  4. Jordi Sunyera
  1. aRespiratory and Environmental Health Unit, IMIM, Barcelona, Spain, bDepartment of Experimental and Health Sciences, Universitat Pompeu Fabra, Barcelona, Spain
  1. Dr J Sunyer, Respiratory and Environmental Health Research Unit, Institut Municipal d'Investigació Mèdica (IMIM), Doctor Aiguader 80, E-08003 Barcelona, Catalonia, Spain.

Statistics from

Many studies have shown an association between current daily levels of air pollution and daily mortality by respiratory and cardiovascular causes in the general population.1 However, the weak associations observed and the ecological nature of the exposure to air pollutants have created some doubts about plausibility of a causal relation. Specificity, which increases the plausibility of causal inference although its lack does not negate it, could be increased using a population more susceptible a priori.2Hence, we assessed the association between daily levels of air pollutants and daily mortality in a cohort of chronic obstructive pulmonary disease (COPD) patients in Barcelona for the years 1985 to 1989. We hypothesised that patients with COPD are more likely to die after increased exposure to urban air pollution than the general population.2


All patients attending emergency room services for either asthma or COPD were recruited during the years 1985 to 1989 in Barcelona. Vital status was obtained through record linkage of the people of the cohort with the Catalonia Mortality Registry for the years 1985 to 1989.3 A total of 9987 people (of the 15 517 in the initial cohort) had a diagnosis of COPD, 3245 of whom died in the period 1985 to 1989 and were used in this analysis. Cardiovascular mortality refers to codes 390 to 459 and respiratory mortality to codes 460 to 519 of the International Classification of Diseases (ICD-9).

Daily information on levels of black smoke, sulphur dioxide (SO2), nitrogen dioxide (NO2), ozone (O3), temperature and relative humidity was collected from the city network.4

Poisson regression time series models were fitted for each pollutant (in a log-linear form) and each different category of mortality following the APHEA methodology5 and adding the natural logarithm of the number of subjects at risk (that is, those still alive) as an offset. The analyses were done using Stata, release 5.0.


The daily mean number of deaths was 1.8 for all causes mortality, ranging from 0 to 9, and 0.7 (0 to 5) and 0.5 (0 to 4) for respiratory and cardiovascular mortality, respectively. Levels of air pollutants have been published elsewhere.4

Table 1 shows the estimated association between air pollutants and mortality in our cohort compared with the estimates for the general population in Barcelona. Associations between one hour maximum of SO2, 24 hours average of NO2 and one hour maximum of NO2, and mortality among COPD patients were stronger than associations obtained with the general population, mainly related to respiratory diseases.

Table 1

Estimated relative risks and 95% confidence limits for total, respiratory and cardiovascular mortality for an increase of 50 μg/m3 in air pollutants, in the general population and a cohort of COPD patients in Barcelona


Daily mortality in COPD patients is related to daily levels of all six pollutants, for all, respiratory and cardiovascular causes. This association is stronger than in the general population only for peaks of SO2 and daily means and peaks of NO2. Particles, measured as black smoke, and daily mean of SO2, the pollutants classically associated with mortality,6showed similar or weaker associations for COPD patients than for the general population. Changes in the profile of urban air pollutants, with a currently more pronounced photochemical component, could explain in part these effects.

Caution is necessary when comparing with the general population given the small range of daily deaths that led to wide confidence intervals. This also precluded an adequate fit using time series models, according to the analysis of residuals. Results did not change using generalised additive models (not shown).

Respiratory mortality showed stronger associations for COPD patients than for the general population for all pollutants excepting averages of SO2, while cardiovascular mortality did so only for peaks of SO2 and the two measures of NO2. This suggests that the susceptibility of COPD patients to air pollution is mainly related to their respiratory condition.


View Abstract


  • Funding: supported in part by grants from the Fondo Investigaciones Sanitarias, Madrid, Spain (FIS, no 96/0042–01), and the Generalitat de Catalunya (CIRIT 1997 SGR 00079). Judith Carcia-Aymerich has a fellowship from Instituto de Salud “Carlos III” (no exp 4365).

  • Conflicts of interest: none.

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.