From the authors:
We thank B. Sposato and M. Scalese for their comments on our paper on smoking and bronchial hyperresponsiveness (BHR) [1], and we appreciate their interest for discussing the important, and complex, topic about determinants of BHR. Our paper was focused on the effects of smoking on BHR, a topic that has been under discussion for decades. Already in the very early, large scale, population studies, such as the Dutch Vlagtwedde-Vlaardingen Study [2], evidence was indicated concerning the effects of smoking on BHR. Our contribution was that, in a general population study setting, we could demonstrate a dose-dependent association of smoking on BHR. This effect remained after the adjustment of effects concerning the level of lung function variables and a number of possible confounders.
B. Sposato and M. Scalese interestingly point out the effects of an increased body mass index (BMI) and obesity on BHR and suggest that our data could verify the interaction between smoking and BMI on BHR found in their study. Several studies have addressed the relationship between obesity and asthma, and less studies the association between obesity and BHR. Population-based studies using clinical methods, including BHR testing, have found an association between obesity and incident asthma among both males and females [3], while studies based on asthma cohorts have mostly found this association to be attributed to females [4]. In our current study setting, the association between BMI and BHR was addressed [5]. However, no statistically significant association was found in univariate analyses, why the odds ratio of increased BMI in multivariate settings was not reported in the paper (BMI ≥30 yielded nonsignificant odds ratios from 1.45 to 1.60 for a provocative dose of histamine inducing a 15% fall in forced expiratory volume in 1 s ≤1.6 mg histamine in multiple logistic regression analyses). A probable explanation is lack of statistical power due to the limited sample size of the subjects who underwent BHR testing in our study, and thus interaction analyses of smoking and BMI were not meaningful. Interestingly, the magnitude of the odds ratio for increased BMI is similar to what B. Spolato and M. Scalese have found.
Differences in outcomes can be explained by the large differences in sample sizes of subjects who underwent BHR testing in the study by B. Sposato and M. Scalese and in our study, and further by the different study populations. The study by B. Sposato and M. Scalese was based on almost 4000 asthma patients (approximately 2000 were hyperreactive) from their outpatient department, while our study reflected the adult general population of Helsinki, Finland.
Investigating selected patient populations may cause bias when studying associations between, for instance, obesity and BHR. In the 1980s Dodge et al. [6] had already pointed out problems with selection bias, i.e. that smoking subjects with respiratory symptoms were mostly classified as having bronchitis, while nonsmokers with similar symptoms were diagnosed as having asthma. It would be an advantage if the interesting results by B. Sposato and M. Scalese could be verified in studies based on the general population.
Footnotes
Conflict of interest: Disclosures can be found alongside the online version of this article at www.erj.ersjournals.com
- Received September 16, 2013.
- Accepted September 30, 2013.
- ©ERS 2014