Smoking and asthma in adults

Study design

This study was a population-based incident case-control study. The source population consisted of adults aged 21–63 yrs, living in the Pirkanmaa Hospital District in southern Finland, with a population of 440,913 inhabitants in 1997. Our goal was to recruit all of the new cases of asthma in the source population during the study period. Controls were selected randomly from the source population based on the 1997 population census. The study was approved by the ethics committees of the Finnish Institute of Occupational Health, Helsinki, and the Tampere University Hospital, Tampere, Finland.

Definition and selection of cases

All of the new cases of asthma were systematically recruited, first, from the city of Tampere from September 15 1997, and, then, from the whole Pirkanmaa Hospital District from March 10 1998–March 31 2000. Patients were recruited by physicians at all healthcare facilities diagnosing asthma, including the Dept of Pulmonary Medicine at the Tampere University Hospital, offices of privately practicing physicians and public healthcare centres. As an additional route of case selection, the National Social Insurance Institution of Finland invited all patients whose reimbursement rights for asthma medication began during the period September 1 1997–May 1 1999 and who had not yet participated.

The diagnostic criteria of asthma included the presence of at least one asthmatic symptom and the demonstration of reversible airway obstruction in lung function investigations, in accordance with the national consensus on the criteria for the diagnosis of asthma (table 1⇓) 20. All of the confirmed cases of asthma that fulfilled the general eligibility criteria were selected as cases. A total of 521 cases participated: 362 through the healthcare system (response rate 90%) and 159 through the National Social Insurance Institution (response rate 78%).

Selection of controls

The controls were randomly selected from the source population using the national population registry, which has full coverage of the population. The general eligibility criteria were applied for controls (aged 21–63 yrs, living in the Pirkanmaa Hospital District). After up to three invitation letters and telephone calls, 1,016 controls participated in the study (response rate 80% of those who had a telephone number in the Pirkanmaa area). After excluding persons reporting previous or current asthma (n=76), persons aged >63 yrs (n=6) and those with an incomplete questionnaire (n=2), the current study population included 932 controls.

Smoking information

Information on smoking was collected with a self-administered questionnaire inquiring about current and past smoking, smoking rate, and duration and quitting of smoking. The following questions were posed: Do you smoke currently? (yes, regularly; yes, occasionally; no, I quit <12 months ago; no, I quit >12 months ago; no, I never smoked regularly); How many years have you smoked?; How much do you/did you smoke on average? (cigarettes, cigars and pipefuls per day, or per week for occasional smokers).

Data collection

At the Tampere University Hospital, cases were recruited at their first visit because of suspected asthma, and the diagnosis was then verified in clinical examinations. At the other healthcare facilities, cases were recruited immediately after theasthma diagnosis was verified. The same protocol for diagnosing asthma was applied at all healthcare facilities. TheNational Social Insurance Institution invited the cases toparticipate 0.5–2 yrs after their diagnosis was established. For these patients, the criteria of asthma diagnosis and its date were confirmed from their medical records, and the samediagnostic criteria were applied as for the other study patients. All subjects with a previous diagnosis of asthma or long-term use of asthma medication earlier in life were excluded. Eligible subjects were invited to participate in the study and their informed consent was requested by their physician or by letter sent via the National Social Insurance Institution. The recruitment of controls took place at regular intervals throughout the study period. A form for stating their informed consent was sent in the invitation letter and returned in a prepaid envelope to the research nurse of the study project.

Measurement methods

Statistical methods

Exposure odds ratio (OR) was used to quantify the relationship between exposure and outcome. Adjusted OR was estimated in logistic regression analysis. Sex, age, parental atopy or asthma, education (as an indicator of socioeconomic status), dampness and mould problems in the home or at work, exposure to ETS, any history of pets in the home, and self-reported occupational exposure to sensitisers, dusts or fumes were used as covariates to adjust for potential confounding.

Three parameters of smoking were used, as follows: current smoking status (never-, ex- and current smoker); smoking rate (occasional, 1–14 cigarettes·day⁻¹, ≥15 cigarettes·day⁻¹) and cumulative smoking expressed as cigarette-yrs (average smoking rate×duration of smoking). The reference category was always never-smoking.

The independent and joint effects of parental atopic diseases and smoking were also studied by comparing the risk of asthma in four exposure categories, as follows: 1) no parental atopy and never-smoking (reference category); 2) parental atopy and never-smoking; 3) no parental atopy and smoking (current, quit smoking >12 months ago); and 4) parental atopy and smoking (current, quit smoking >12 months ago). ORs were calculated contrasting each of the three exposure categories to the reference category. Estimates for the independent effects of parental atopy, smoking and their joint effect were derived from the same logistic regression model adjusting for the covariates. An analogous approach was applied for studying the effects of sex and smoking, with never-smoking males forming the reference category.

Characteristics and smoking among cases and controls

There were more females among the cases than the controls, and the cases were younger, had more allergic relatives, and were less educated than the controls (table 2⇓). The distribution of smoking habits, smoking rate and cumulative cigarette-yrs among cases and controls are presented in table 3⇓. There were more current and ex-smokers and fewer never-smokers among cases than controls.

Risk of asthma in relation to smoking

The risk of asthma was significantly related to both former and current smoking, the risk being stronger in ex-smokers (table 3⇓). Considering the smoking rate, the risk of asthma increased from occasional smoking to smoking 1–14 cigarettes·day⁻¹, but was lower in the highest smoking category (≥15 cigarettes·day⁻¹). An increased risk of asthma was also observed among those with lifetime smoking of 1–199 cigarette-yrs, but a slightly lower risk was observed in the highest cumulative smoking category (≥200 cigarette-yrs). Similar trends according to the amount of smoking were detected in both males and females.

Parental atopy and sex as potential determinants of susceptibility

Table 4⇓ presents the estimated independent and joint effects of parental atopy and smoking (both current and ex-smoking). The adjusted ORs in population-based incident case-control studies represent estimates of incidence rate ratios 23. Using a common reference category, the magnitude of independent and joint effects can be compared. In this study, the common reference category was never-smokers without a history of parental atopy. The independent excess risk (95% confidence interval (CI)) of asthma related to parental atopy was estimated to be 115% (52–206), i.e. 2.15 incidence rate ratio between parental atopy/never-smoking and the reference category. Similarly, the independent effects of current and former smoking were 43% (2–99) and 60% (14–124), respectively. The joint effect of parental atopy and current smoking was estimated to be 143% (52–290), which was close to what was expected on the basis of independent effects on an additive scale (115+43=158%). Also, the estimated joint effect of parental atopy and former smoking, 170%, corresponded to the sum of independent effects (115+60=165%).

Table 5⇓ illustrates the independent and joint effects of sex and smoking, using never-smoking males as the reference category. In never-smoking subjects, the female sex was related to a significantly increased risk of asthma, with an adjusted OR (95% CI) of 1.57 (1.09–2.27), corresponding to a 57% excess risk. The association between current smoking and asthma was not significant in males (1.04 (0.66–1.65)). The risk of asthma increased slightly in relation to ex-smoking in males (1.34 (0.86–2.09)). In a separate analysis including females only, the risk of asthma increased significantly in relation to both current smoking (1.62 (1.13–2.34)) and ex-smoking (1.52 (1.04–2.23); data not shown). The joint effect of female sex and current smoking (table 5⇓) resulted in a 143% increase in the risk of asthma (OR 2.43) compared to never-smoking males. The combined effect of sex and smoking would be 61% on the basis of their independent effects (57% for female sex + 4% for current smoking); this means that there was an interaction of 82% on an additive scale (143–61=82%). There was also an interaction of 47% for the effects of female sex and ex-smoking (joint effect 138% – (expected effects 57% for female sex + 34% for ex-smoking)=47%).

Validity issues

Our aim was to reach all newly diagnosed asthma cases in a geographically defined area. With the thorough recruitment system, via the healthcare facilities and the help of the National Social Insurance Institution, it is believed that this objective was well achieved. In Finland, the National Social Insurance Institution offers a significant reimbursement for asthma medicines to every asthmatic fulfilling our diagnostic criteria, which means that most, if not all, new asthma cases are reported to the Institution by physicians. As the response rate among cases and controls was good, there is no reason to believe that any major selection bias would exist.

Newly diagnosed asthmatics may either underestimate or overestimate their smoking history because of fears regarding the new disease, but as the study was introduced to the participants as a study on environmental factors in general, and smoking-related questions were only one part of the questionnaire, major recall bias is not likely to have occurred. All asthmatics, apart from those recruited through the National Social Insurance Institution, answered the questions on smoking habits before or at the time when their diagnosis was established. To accurately assess the exposure, both ex- and current smokers were asked about the number of cigarettes smoked per day and the time since quitting, making it possible to quantify lifetime cumulative exposure and to study the dose–response relationship between smoking and asthma.

The difference between COPD and asthma may be difficult to detect when using questionnaire-based diagnosis only, as the symptoms of these two conditions overlap. To improve the validity of diagnosis of asthma in the current study, asthma diagnosis was confirmed by lung function measurements showing reversible airway obstruction in spirometry and/or PEF surveillance. As a consequence of this, clear COPD cases were excluded from this study population. Inclusion of COPD cases would lead to overestimation of the risk, since smoking is known to be a strong determinant of COPD 25. Conversely, self-reporting of asthma might lead to underestimation of the risk, given that smokers might interpret their respiratory symptoms as asthmatic, even though in reality they could be signs of nonobstructive processes. Since all the physicians diagnosing asthma in the current study region agreed to apply a standardised diagnosing protocol, based on the national consensus criteria for the asthma diagnosis 20, good validity in the assessment of asthma was achieved. Cases with an earlier diagnosis of asthma or long-term use of asthma medication ever during lifetime were excluded to confirm that only new cases were included. Previously diagnosed asthma would have diluted the risk estimate, since physicians usually advise asthmatics to quit smoking.

Adult-onset asthma has a multifactorial aetiology; for example, indoor mould problems 26, pets 27, and ETS 28 all increase the risk of asthma. In the multivariate logistic regression analyses, the effects of many important potential confounders, e.g. age, pets, dampness and mould problems, and occupational exposures, were controlled. In no earlier study on the association between smoking and asthma have all these confounders been simultaneously controlled.

Synthesis with previous studies

This large incident case-control study provides evidence of an increased risk of asthma in current and former smokers, especially in females. Previous studies have provided conflicting results. In two earlier longitudinal studies 17, 18, an increased risk of asthma in smokers compared with never or nonsmokers (including both never- and ex-smokers) was found, in line with the current results. In contrast, in the study by Troisi et al. 16, the risk of asthma among current smokers was approximately half that of never- and ex-smokers. The risk of asthma was increased among ex-smokers for the first 5 yrs after quitting. In the study by Godtfredsen et al. 19, ex-smokers had a higher incidence of self-reported asthma than never-smokers. In the 6-yr follow-up study from Finland 15, asthma risk was not increased in either current or ex-smokers compared with never-smokers. Two earlier cross-sectional studies have found a greater susceptibility to asthma in females in relation to smoking 29, 30, whereas no sex difference was observed in the Finnish 6-yr follow-up study 15.

In all of these studies, the asthma diagnosis was based on the self-reporting of symptoms, doctor-diagnosed asthma or the use of asthma medication, or hospital admissions due toasthma, making some misclassification of asthma cases likely. In only one previous study were persons with doctor-diagnosed chronic bronchitis excluded 16. Occupational exposures or exposure to ETS were not taken into account, and assessment of smoking history did not include lifetime exposure. Compared with the earlier studies, the strength of the current study lies in the accuracy of the assessment of both exposure and asthma, the critical issues in terms of validity of the results. Our efforts aimed to include only new cases of asthma to reduce the potential effects of selection and recall bias that often affect studies with prevalent cases of asthma.

Conclusions

The current results are consistent with the hypothesis that smoking causes asthma in adulthood. The stronger effect of former smoking and the nonlinear relationship with the quantity of smoking seem to point to a behavioural change asa response to beginning symptoms. The results suggest that smokers who are prone to develop asthma either stop smoking or continue to smoke at a moderate rate because of their respiratory symptoms. Females seem to be more susceptible to the adverse effects of smoking than males. Whether this is explained by different biological susceptibility to the harmful effects of tobacco smoke, or by females being more addicted to tobacco, needs to be studied further.