There are two major strengths of the present study. First, it estimated the prevalence and risk factors of airflow obstruction in a random sample of young adults. Second, the study protocol (8
) allows for comparison of results with other studies (16
There was no evidence that the participation rate of 38.4% in stage 2 of the study biased the findings of the study. Adjustment for variables that were significantly different between participants and nonparticipants (smoking and symptoms) did not change the prevalence of airflow obstruction; nonadjusted rates are reported to facilitate comparison with results from the international sites of the ECRHS (16
). Bronchial responsiveness was not determined for approximately 20% of subjects. This did not bias the results either, because the relationship of variables in model 1 (obtained from the questionnaire and available for all subjects) to airflow obstruction did not differ for subjects with and without bronchial responsiveness.
Airway obstruction was defined as an FEV1
/FVC less than the LLN of Hankinson et al (11
). While any such definition is arbitrary, this may be more appropriate than the GOLD definition of less than 0.70, which is lower than Hankinson’s criterion in young adults. The Hankinson’s LLN reflects the lower bounds for 95% of a normal (nonsmoking) population; therefore, one may expect 5% of a normal population to fall below these limits. Only 6.6% of study subjects had airflow obstruction as defined. This is not much more than was expected in a normal population. On the other hand, the study subjects were not a normal population, including smokers and asthmatics, and more importantly, the individuals identified using the arbitrary definition of airflow obstruction were distinctly different from those without airflow obstruction. Even nonsmoking nonasthmatics who had an FEV1
/FVC below the LLN demonstrated significantly higher rates of bronchial hyper-responsiveness than nonsmoking nonasthmatics without airflow obstruction. The arbitrary definition of airflow obstruction that we used had objective meaning.
The prevalence of airflow obstruction defined by Hankinson’s prediction equations (11
) was 6.6%, and was not related to sex and age. The lower prevalence of airflow obstruction using the definition by GOLD was due to Hankinson’s LLN of FEV1
/FVC exceeding 0.70 in some subjects. Using the definition by GOLD, the prevalence of airflow obstruction in the present study was slightly higher (4.2%) than the median (3.6%) for 16 ECRHS countries (16
). Our data are consistent with both ECRHS (16
) and surveys in the United States (17
), which found that in the developed world, there is a substantial prevalence of airflow obstruction in young adults. Data in suggest that it is not a trivial disorder.
The relationship between risk factors and airflow obstruction in the present study must be interpreted carefully because it was assessed using a cross-sectional design. While this design identifies variables associated with current cases of airflow obstruction, it can only suggest, because of a lack of temporal relationship, that these variables were responsible for the development of the disorder.
Smoking was a major independent risk factor for airflow obstruction. The small difference in smoking exposure between smokers with and without airflow obstruction (3.5 pack-years) suggested that airflow obstruction at this age chiefly reflected unusual sensitivity to smoking (18
). In never smokers, contrary to the ECRHS study (16
), there was no effect of environmental exposure to tobacco. The relationship between environmental tobacco smoke and decreased lung function is not consistent and the effect on lung function is relatively small (19
The effect of self-reported exposure to dust, fumes and gases was not independent of smoking, although the literature supports the role of occupational exposures in the development of airflow obstruction (20
). These findings may be the result of the healthy worker effect (23
), or due to the lack of prolonged exposure or inaccurate reporting.
The finding that elevated levels of IgE were not associated with airflow obstruction independent of asthma and positive skin reactivity is consistent with the literature (24
); allergies were, however, significantly correlated with asthma (both ever and current) and with lung trouble before 16 years of age. Because both allergy and asthma related independently to airflow obstruction, presumably reflected that some asthma is not associated with allergy.
Bronchial hyper-responsiveness was strongly (OR=5.9) associated with airflow obstruction. This effect was not substantially diminished when other variables were in the model (OR=5.3) (). The effects of current asthma, lung trouble before 16 years of age and allergies were not significant in the presence of bronchial hyper-responsiveness. The effects of having asthma previously were retained and had an effect on airflow obstruction; apparently some previous asthmatics had normal methacholine reactivity.
It is conceivable that the relationship between bronchial hyper-responsiveness and airflow obstruction was due to increased responsiveness of people with small airway calibre, ie, low FEV1
). In the present study, there was a significant relationship between FEV1
expressed as per cent of predicted value and bronchial responsiveness expressed as logarithm of slope, but the correlation was poor (r=0.24) because of scatter. When the analysis was restricted to subjects with airflow obstruction, there was little improvement in the correlation; FEV1
accounted for less than 10% of the variation in bronchial responsiveness. We believe that bronchial hyper-responsiveness was, to some extent, independent of FEV1
and that it conveyed useful information. This is compatible with several longitudinal studies (27
) showing that bronchial hyper-responsiveness increased the risk of airflow obstruction as well as being its consequence.
The combination of smoking and asthma accounted for 21% of subjects with airflow obstruction, smoking alone for 50% of subjects with airflow obstruction, and asthma alone for 12% of subjects with airflow obstruction. However, 17% of those with airway obstruction never smoked and had no history of asthma. Bronchial hyper-responsiveness increased the risk of airflow obstruction in all four categories including the latter. It was not possible to identify any significant risk factors other than bronchial hyper-responsiveness in this group.
In the present study, airway obstruction in young adults was associated with two well-known risk factors: smoking and asthma. Most individuals could be characterized as either having early COPD or asthma; however, such a dichotomous analysis may be an oversimplification for several reasons. First, a substantial fraction of subjects with airway obstruction both had smoked and had a history of asthma; whether these subjects will subsequently develop clinical COPD or remain ‘asthmatic’ is unknown. It should be noted that this is the group with the highest prevalence of airflow obstruction (14.5%) () and, when coupled with a high level of hyper-responsiveness, may indicate a poor prognosis. Second, bronchial hyper-responsiveness was an important risk factor of airway obstruction in both asthmatics and smokers. Methacholine reactivity is both a characteristic of asthma and a predictor of lung function decline in smokers. It is not clear, however, that the mechanism of bronchial hyper-responsiveness is the same in smokers and asthmatics; the clinical characteristics of nonsmoking asthmatics and smokers with the same level of methacholine reactivity are thought to be quite different. Finally, a significant fraction of our young adults with airway obstruction had bronchial hyper-responsiveness but were neither smokers nor known asthmatics; the subsequent course of the disease in these people is also unknown, but may be important to our understanding of clinical obstructive lung disease.