Article Text


Is there an association between impaired pulmonary function and mortality in never smokers?
  1. H Schünemann
  1. Assistant Professor of Medicine, Social & Preventive Medicine and Clinical Epidemiology & Biostatistics, University at Buffalo, State University of New York, USA;

Statistics from

I read with great interest the article by Mannino et al1 on the association between impaired pulmonary function and mortality, and wish to comment on two statements in the paper.

Firstly, the authors report that “an interesting finding in (their) analysis was that, in never smokers, moderate or severe COPD did not have a significantly increased mortality risk”. In never smokers with severe COPD the point estimate for the hazard ratio forming the basis of this statement is 1.3 with 95% confidence intervals ranging from 0.7 to 3.1. However, these confidence intervals overlap with the point estimate of the hazard ratios in current smokers with severe COPD. There were only 92 participants with severe COPD in the entire sample of the population, hence the wide confidence intervals. The estimates are similar in those with moderate COPD. Furthermore, this trend was also evident in patients with mild COPD who had never smoked. In fact, in the latter group the point estimates were identical in current smokers and never smokers. The authors should therefore be cautious in concluding that never smokers with COPD do not have an increased risk of mortality. An analysis of continuous pulmonary function data in relation to mortality in never smokers independent of the GOLD classification or in all patients with COPD may result in statistically significant results. How would one interpret such a finding? The analysis in never smokers should be seen in the context of other studies reporting increased mortality risks in never smokers,2 as they may be due to small sample size in spite of the overall large sample size in NHANES I. The authors’ statement could be misinterpreted to suggest that never smokers would not require screening, a question that is not yet resolved.

Secondly, our study did report both FEV1% in quintiles as well as continuous variables.3


View Abstract

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.