Bidirectionality in the Relationship Between Asthma and Smoking in Adolescents: A Population-Based Cohort Study
Article Outline
Abstract
Purpose
Several cross-sectional studies have showed higher smoking rates among adolescents with asthma, but hardly any study has investigated this relation longitudinally. This study examines whether these cross-sectional results are caused by higher smoking onset among adolescents with asthma, or by the development of asthma after smoking onset.
Methods
This was a longitudinal study (22 months) among 7,426 Dutch adolescents (mean age at baseline = 12.9 years). Asthma was assessed with self-report questionnaires of the International Study of Asthma and Allergies in Childhood (ISAAC) and the student questionnaire of the American College of Allergy, Asthma and Immunology. Adolescents’ smoking and parental smoking were also assessed by adolescent-report. All analyses were controlled for age, gender, education, ethnicity, and parental smoking.
Results
In baseline non-smokers, adolescents with current diagnosed asthma and with more severe asthma had an increased risk to become regular smokers. Among girls and adolescents with a smoking mother, having asthma symptoms was a risk factor for starting experimental smoking. In contrast, among boys and adolescents with a non-smoking mother, having asthma symptoms was protective for experimental smoking. With regard to the effect of smoking on asthma, adolescent smoking predicted a higher incidence of asthma symptoms. In addition, smoking predicted increased symptom severity scores, and this effect was stronger in girls. Adolescent smoking was a stronger predictor for the development of asthma symptoms when the mother smoked.
Conclusions
The relationship between asthma and smoking in adolescence seems to be bidirectional, with relations in both directions being stronger in females and when the mother smokes.
Keywords: Asthma, Smoking, Adolescent, Cohort studies, Incidence
Several cross-sectional studies have investigated the prevalence of smoking among adolescents with asthma. Although some studies reported no difference in smoking between asthmatic and non-asthmatic adolescents [1], other studies showed that smoking is more prevalent among adolescents with asthma [2], [3], [4]: asthmatic adolescents were 1.15 [3] to 2.55 [4] more likely to be daily smokers than non-asthmatic adolescents. Although these results might be interpreted as asthma being a risk factor for smoking onset, because of the cross-sectional design of these studies they do not allow for conclusions on the directionality of the relationship. Cohort studies are needed to disentangle whether active smoking predicts a higher asthma incidence and/or whether having asthma predicts a higher smoking onset.
Surprisingly, there are hardly any cohort studies investigating differences in smoking onset between asthmatic and non-asthmatic adolescents. A study that retrospectively asked for age of smoking initiation and age of asthma onset concluded that asthma onset preceded smoking in 90% of the cases [5], implying that smoking onset is higher among asthmatic individuals. However, the only prospective study on the effect of asthma on smoking demonstrated equal smoking transitions between asthmatic and non-asthmatic adolescents [6]. Unfortunately, this study did not differentiate between smoking onset and smoking continuation.
Parental smoking can modify the relation between smoking predictors and active smoking among the general population of adolescents. For instance, non-using parents have a buffering effect on the influence of friends’ substance use in adolescence [7]. Likewise, several studies found gender differences in the effect of smoking predictors on smoking onset in adolescents [8]. Among asthmatic adolescents, it has not been studied yet whether parental smoking and gender can also moderate the effect of asthma on smoking onset.
With regard to the role of active smoking in the development of asthma, cohort studies have provided contradictory results [9], [10]. Few of these studies focused on adolescents [11], [12], [13], [14], and these studies provide conflicting evidence as well. Furthermore, very little is known about whether the effects of active smoking are different for female or male adolescents. Gender differences in the development of asthma during adolescence have been frequently reported [15], with increased risk of asthma in girls. In males, sex steroids could be protective for the development of asthma, whereas in females, the sex steroids are proinflammatory, increasing the risk of atopic disorders [16]. Because of these gender differences in the susceptibility to development of asthma, we want to examine whether smoking predicts asthma development differentially in boys and girls.
Moreover, there is still conflicting evidence about the role of parental smoking on asthma development among adolescents [17], and to our knowledge, only one study has investigated whether parental smoking modulates the effect of smoking on asthma development in adolescents [14], i.e., whether an adolescent who actively smokes develops asthma depends on whether or not parents smoke.
Summarized, results of cohort studies are inconclusive about the temporal relation between asthma and smoking among adolescents. To test the direction of the relation between asthma and smoking we conducted a large population-based cohort study among 7,426 Dutch adolescents to address the following questions: (a) are the incidence and symptoms of asthma higher among active smokers, and (b) do non-smoking asthmatic adolescents differ from non-smoking non-asthmatic adolescents with regard to the uptake of smoking? Furthermore, the main effects of gender and parental smoking on smoking and asthma development, as well as moderating effects of gender and parental smoking on the relation between asthma and smoking onset and the relation between smoking and asthma development were investigated as well. Since mothers in the Netherlands are more often the primary caretakers than fathers and spend more time with their adolescent children, we have analyzed the smoking behavior of fathers and mothers separately.
Methods
Sampling and data collection
The present longitudinal study was approved by the medical ethics committee (CMO Arnhem-Nijmegen). Data collection of the first wave took place in January 2003 as part of the International Study of Asthma and Allergies in Childhood (ISAAC) phase III [18]. All first-year and second-year students in 33 secondary schools in four regions of the Netherlands were asked to fill out a written questionnaire during school hours, under supervision of a teacher. Teachers received instructions about the procedure and how to handle questions. They were asked to remind the students of the confidentiality and to note which students were absent and why. CD gift vouchers were raffled among students who filled out the questionnaires sincerely. Of the 11,124 students in the participating schools, 10,087 filled out the questionnaire completely and consistently. Non-response was primarily caused by absence on the measurement day. Only 15 students refused to fill out the questionnaire; 67 students had left school between contacting the schools and data collection. A more detailed description of the baseline data collection can be found elsewhere [19]. The follow-up questionnaire was administered in November 2004, with a similar procedure.
Asthma measures
WheezeLifetime wheeze (“have you ever had wheezing or whistling in the chest at any time in the past?” yes/no) and current wheeze (“have you had wheezing or whistling in the chest in the past 12 months?” yes/no) were assessed with the validated ISAAC written self-report questionnaire for 13–14-year-olds [18], translated into Dutch [20].
Indication of asthmaRespondents filled out a Dutch translation of the student questionnaire of the American College of Allergy, Asthma and Immunology (ACAAI) [21]. Respondents were asked to indicate on a 3-point scale (never, sometimes, a lot) how often they suffered from seven symptoms (e.g., “I wake up at night because I have trouble breathing”). For each symptom, the responses “sometimes” and “a lot” were recoded as “1,” the response “never” was recoded as “0,” and scores were added. A score of three or more was an indication of asthma [21].
Symptom severitySeverity of asthma symptoms was also assessed with the ACAAI student questionnaire [21]. Responses on the seven items with a 3-point scale (never, sometimes, a lot) were averaged to form a symptom severity score (Cronbach’s α = .72).
Current diagnosed asthmaThose with current diagnosed asthma fulfilled two criteria: (1) a positive response on either “did you have asthma in the past 12 months?” or “did you use asthma medication in the past 12 months?” and (2) a positive response to “did a doctor confirm you have asthma?” [20].
Smoking measures
Parental smokingParental smoking was assessed by asking the adolescents to report on their parents’ smoking behavior: “Does your father/mother smoke (yes/no).”
Active smokingAdolescents’ smoking behavior was assessed by a single item on a 9-point scale ranging from (1) I never smoked, not even a puff to (9) I smoke at least once a day [22]. This variable was recoded in three categories: Those who responded “I never smoked, not even a puff” were defined as never smokers. Those who responded in the categories 2–7 (“I tried smoking but I don’t smoke anymore” to “I smoke at least once a month”) were defined as experimental smokers. The adolescents who reported they smoked “at least once a week” or “at least once a day” were defined as regular smokers. This instrument has been used in various health studies in the Netherlands [22], [23], [24].
Demographic information
Age, gender, education level (lower vocational training, intermediate vocational training, and high school/pre-university education), and ethnicity (Dutch or non-Dutch) were assessed at baseline and were included in all analyses as potential confounders.
Statistical analyses
Data were analyzed using the statistical package for the social sciences (SPSS) for Windows, version 14.0. Four sets of analyses were conducted, all controlled for age, education level and ethnicity. In the first step of the analyses, we used multinomial logistic regression analyses to predict smoking at follow-up among never-smoking adolescents at baseline (N = 4,762). Multinomial logistic regression analysis is an extension of binary logistic regression, allowing for dependent variables with more than two categories. Using this technique, odds ratios (OR) are calculated to simultaneously estimate the odds of being in one of the dependent variable groups compared to being in the reference group. Four multinomial logistic regression analyses were conducted to predict smoking onset by current wheeze, by indication of asthma (IoA), by severity, and by current diagnosed asthma; all measured at baseline. Each of the regression analyses consisted of three models: model 1 included age, education level, and ethnicity, as well as gender and parental smoking behavior at baseline. In model 2, one of the asthma variables (wheeze, IoA, severity or diagnosed asthma) was added in addition to the variables of model 1. Model 3 consisted of the three interaction terms, namely between the asthma variable on the one hand and gender, smoking mother and smoking father on the other, as well as the main effects variables of model 1 and 2.
Second, three hierarchical binary logistic regression analyses were conducted to predict current wheeze at follow-up among adolescents who reported no lifetime wheeze at baseline (N = 5,686); to predict IoA at follow-up among adolescents without IoA at baseline (N = 5,000); and to predict current diagnosis of asthma at follow-up among adolescents without current diagnosis of asthma at baseline (N = 6,723). Besides age, education level, and ethnicity, gender and smoking behavior of parents at baseline were entered at step 1, and adolescents’ smoking behavior at baseline was added to the model at step 2. The final step consisted of all main effects and the three interaction terms, namely between adolescent smoking on the one hand and gender, smoking mother and smoking father on the other.
Finally, univariate analyses of variance were conducted to predict symptom severity at follow-up, controlled for baseline severity. In the first model, the effects of age, education level, ethnicity, gender, and smoking behavior of parents at baseline were included. In the second model, adolescents’ smoking behavior at baseline was added to the model, and in the last step we added the interaction terms.
Results
Characteristics of the study population
Of the 10,087 students who participated in the first wave, 7,426 students (73.6%) filled out both questionnaires consistently and completely, and these students were included in the present study. Table 1 describes the characteristics of the participants at baseline and follow-up.
Table 1. Characteristics of the study population at baseline and follow-up (N = 7,426)
| Baseline | Follow-up | |
|---|---|---|
| Age | 12.9 (0.78)a | 14.8(0.84)a |
| Gender | ||
| Female | 3822(51.6)b | |
| Male | 3585(48.4)b | |
| Lifetime wheeze | ||
| No | 5686(77.8)b | 5992(81.5)b |
| Yes | 1623(22.2)b | 1362(18.5)b |
| Current wheeze | ||
| No | 6437(88.2)b | 6640(90.5)b |
| Yes | 862(11.8)b | 695(9.5)b |
| IoA | ||
| No | 5000(70.1)b | 5327(73.5)b |
| Yes | 2128(29.9)b | 1916(26.5)b |
| Symptom severity | 1.30(0.28)a | 1.26(0.28)a |
| Current diagnosed asthma | ||
| No | 6723(92.9)b | 6833(94.3)b |
| Yes | 513(7.1)b | 415(5.7)b |
| Smoking mother | ||
| No | 5103(70.6)b | 5342(72.7)b |
| Yes | 2126(29.4)b | 2006(27.3)b |
| Smoking father | ||
| No | 4685(64.9)b | 4904(67.2)b |
| Yes | 2535(35.1)b | 2390(32.8)b |
| Smoking adolescent | ||
| Never | 4762(65.7)b | 3587(49.0)b |
| Experimenting | 2085(28.8)b | 2691(36.8)b |
| Regular | 399(5.5)b | 1035(14.2)b |
aValues represent mean (SD). |
bValues represent n (%). |
Predicting adolescent smoking behavior by asthma
Of the baseline non-smoking adolescents, 1,189 (25.0%) were experimental smokers and 208 (4.4%) were regular smokers at follow-up. Smoking at follow-up was predicted by maternal smoking (Table 2). Higher age predicted more regular smoking, and a lower education level was associated with more frequent smoking at follow-up; no differences were found for ethnicity. When the mother smoked, adolescents were more likely to become experimental smokers or regular smokers instead of staying non-smokers, and more likely to be regular smokers than experimental smokers. Adolescent smoking onset was not predicted by paternal smoking or gender.
Table 2. Prediction of never, experimental, and regular smoking at follow-up among non-smokers at baseline (N = 4,762) by indicators of asthma at baseline: current wheeze, indication of asthma, severity, and current diagnosed asthma: Results of multinomial logistic regression analyses
| Experimenting vs never | Regular vs never | Regular vs experimenting | |
|---|---|---|---|
| OR (95% CI) | OR (95% CI) | OR (95% CI) | |
| Current wheeze | |||
| Model 1 | |||
| Gendera | 1.03(0.90–1.18) | 1.08(0.80–1.45) | 1.05(0.76–1.54) |
| Smoking motherb | 1.33(1.12–1.57) | 1.99(1.42–2.78) | 1.50(1.05–2.13) |
| Smoking fatherc | 1.15(0.98–1.35) | 1.20(0.86–1.68) | 1.05(0.74–1.48) |
| Model 2 | |||
| Current wheezed | 0.83(0.66–1.05) | 0.94(0.57–1.53) | 1.13(0.67–1.89) |
| Model 3 | |||
| Current wheeze*Gender | 1.34(0.83–2.15) | 2.87(1.00–8.25) | 2.15(0.72–6.46) |
| Current wheeze*Smoking mother | 1.90(1.10–3.29) | 1.87(0.63–5.54) | 0.99(0.32–3.04) |
| Current wheeze*Smoking father | 0.66(0.38–1.14) | 0.29(0.08–1.04) | 0.45(0.12–1.65) |
| IoA | |||
| Model 1 | |||
| Gender a | 1.03(0.90–1.18) | 1.08(0.80–1.45) | 1.05(0.76–1.54) |
| Smoking motherb | 1.33(1.12–1.57) | 1.99(1.42–2.78) | 1.50(1.05–2.13) |
| Smoking fatherc | 1.15(0.98–1.35) | 1.20(0.86–1.68) | 1.05(0.74–1.48) |
| Model 2 | |||
| IoAe | 1.06(0.90–1.24) | 1.18(0.84–1.65) | 1.11(0.78–1.58) |
| Model 3 | |||
| IoA*Gender | 0.72(0.52–1.00) | 1.01(0.51–1.99) | 1.40(0.69–2.84) |
| IoA*Smoking mother | 1.70(1.16–2.48) | 1.56(0.74–3.28) | 0.92(0.43–1.99) |
| IoA*Smoking father | 0.71(0.49–1.02) | 0.66(0.31–1.40) | 0.94(0.43–2.04) |
| Symptom severity | |||
| Model 1 | |||
| Gender a | 1.03(0.90–1.18) | 1.08(0.80–1.45) | 1.05(0.76–1.54) |
| Smoking motherb | 1.33(1.12–1.57) | 1.99(1.42–2.78) | 1.50(1.05–2.13) |
| Smoking fatherc | 1.15(0.98–1.35) | 1.20(0.86–1.68) | 1.05(0.74–1.48) |
| Model 2 | |||
| Severity | 1.07(0.83–1.39) | 1.74(1.05–2.88) | 1.62(0.95–2.76) |
| Model 3 | |||
| Severity*Gender | 0.57(0.33–0.96) | 1.05(0.38–2.90) | 1.86(0.64–5.37) |
| Severity*Smoking mother | 2.34(1.26–4.34) | 2.32(0.76–7.09) | 0.99(0.31–3.16) |
| Severity*Smoking father | 0.61(0.34–1.10) | 0.32(0.10–1.01) | 0.53(0.16–1.73) |
| Current diagnosed asthma | |||
| Model 1 | |||
| Gender a | 1.03(0.90–1.18) | 1.08(0.80–1.45) | 1.05(0.76–1.54) |
| Smoking motherb | 1.33(1.12–1.57) | 1.99(1.42–2.78) | 1.50(1.05–2.13) |
| Smoking fatherc | 1.15(0.98–1.35) | 1.20(0.86–1.68) | 1.05(0.74–1.48) |
| Model 2 | |||
| Asthmaf | 0.69(0.51–0.93) | 1.32(0.79–2.21) | 1.91(1.10–3.34) |
| Model 3 | |||
| Asthma*Gender | 1.12(0.61–2.04) | 1.65(0.55–4.97) | 1.48(0.46–4.80) |
| Asthma*Smoking mother | 1.24(0.61–2.51) | 3.89(1.25–12.07) | 3.15(0.93–10.70) |
| Asthma*Smoking father | 0.63(0.32–1.23) | 0.38(0.12–1.23) | 0.60(0.17–2.14) |
aReference category is females. |
bReference category is non-smoking mothers at baseline. |
cReference category is non-smoking fathers at baseline. |
dReference category is no current wheeze at baseline. |
eReference category is no indication of asthma at baseline. |
fReference category is no current diagnosed asthma at baseline. |
Smoking was also predicted by asthma and symptoms of asthma, though only for symptom severity and current diagnosed asthma (Table 2). When symptom severity was higher, adolescents were more likely to become regular smokers instead of staying non-smokers (OR = 1.74). Adolescents who reported current diagnosed asthma were less likely to experiment with smoking than adolescents without current diagnosed asthma (OR = .69). However, if the adolescents with asthma started smoking between baseline and follow-up, they were more often regular smokers than experimenters (OR = 1.91).
The effects of asthma on experimenting versus never-smoking were different for girls and boys, as shown by the significant interaction between severity and gender, and between IoA and gender. Additional separate analyses (Table 3) for boys and girls show that more severe asthma has a protective effect on starting to experiment with smoking among boys, but is a risk factor for experimenting among girls. Similar results are found for IoA: baseline IoA reduces the likelihood of experimenting among boys but increases the likelihood among girls.
Table 3. Results of separate analyses for the significant interaction effects
| Interaction | Predictor | OR (95% CI) |
|---|---|---|
| Prediction of experimental versus never-smoking by symptom severity*gender | ||
| Boys | Symptom severity | 0.78(0.52–1.16) |
| Girls | Symptom severity | 1.36(0.97–1.51) |
| Prediction of experimental versus never-smoking by IoA*gender | ||
| Boys | IoAa | 0.88(0.68–1.13) |
| Girls | IoAa | 1.20(0.97–1.47) |
| Prediction of experimental versus never-smoking by current wheeze*smoking mother | ||
| Non-smoking mothers | Current wheezeb | 0.72(0.54–0.95) |
| Smoking mothers | Current wheezeb | 1.16(0.76–1.78) |
| Prediction of experimental versus never-smoking by IoA*smoking mother | ||
| Non-smoking mothers | IoAa | 0.96(0.79–1.15) |
| Smoking mothers | IoAa | 1.37(1.01–1.85) |
| Prediction of experimental versus never-smoking by symptom severity*smoking mother | ||
| Non-smoking mothers | Symptom severity | 0.90(0.67–1.23) |
| Smoking mothers | Symptom severity | 1.66(1.01–2.73) |
| Prediction of regular versus never-smoking by current asthma diagnosis*smoking mother | ||
| Non-smoking mothers | Current diagnosisc | 0.80(0.37–1.77) |
| Smoking mothers | Current diagnosisc | 2.27(1.10–4.68) |
| Prediction of onset current wheeze by adolescent smoking*smoking mother | ||
| Non-smoking mothers | Experimental smokingd | 1.11(0.73–1.69) |
| Regular smokingd | 1.40(0.54–3.61) | |
| Smoking mothers | Experimental smokingd | 2.34(1.39–3.93) |
| Regular smokingd | 5.15(2.57–10.30) | |
| Prediction of onset IoA by adolescent smoking*smoking mother | ||
| Non-smoking mothers | Experimental smokingd | 1.27(1.00–1.60) |
| Regular smokingd | 3.16(1.98–5.06) | |
| Smoking mothers | Experimental. smokingd | 1.83(1.32–2.52) |
| Regular smokingd | 3.57(2.18–5.85) |
aReference category is no indication of asthma at baseline. |
bReference category is no current wheeze at baseline. |
cReference category is no current diagnosed asthma at baseline. |
dReference category is never smoking at baseline. |
Besides a main effect, maternal smoking also moderated the association between asthma and adolescent smoking onset. For all four asthma measures, significant interactions with maternal smoking showed that the strength of the effect of asthma on smoking depended on whether or not the mother was a smoker. Additional analyses (Table 3) showed that when the mother was a non-smoker, having asthma or symptoms of asthma was a protective factor, whereas having asthma or symptoms of asthma is a risk factor for smoking onset when the mother was a smoker.
Predicting the development of asthma symptoms
All measures of asthma were moderately to strongly related (Pearson’s correlations between .31 and .81; all p values <.001) at baseline and follow-up. Incidence of current wheeze, IoA, and current diagnosis of asthma was 4.1% (n = 233), 15.2% (n = 762) and 1.2% (n = 82), respectively. Table 4, Table 5 describe the results for the prediction of the incidence of asthma and symptoms of asthma. Except for increased severity scores with increasing age, no significant differences were found for the effect of the confounding variables of age, ethnicity, and education level on severity, IoA, current wheeze and current diagnosis of asthma. Male adolescents were less likely to have developed current wheeze and IoA between the waves, and had lower symptom severity scores at follow-up. Whereas paternal smoking predicted only higher symptom severity at follow-up, maternal smoking predicted not only higher symptom severity but also a higher incidence of current wheeze and IoA.
Table 4. Prediction of current wheeze, indication of asthma and current diagnosed asthma by parental and adolescent smoking at baseline: Results of binary logistic regression analyses
| Current wheeze† | IoA†† | Current diagnosis of asthma††† | |
|---|---|---|---|
| OR (95% CI) | OR (95% CI) | OR (95% CI) | |
| Block 1 | |||
| Gendera | 0.47(0.35–0.63) | 0.43(0.36–0.51) | 0.67(0.41–1.08) |
| Smoking motherb | 1.73(1.26–2.37) | 1.29(1.06–1.56) | 0.90(0.51–1.57) |
| Smoking fatherc | 0.99(0.72–1.35) | 1.18(0.98–1.41) | 1.13(0.67–1.91) |
| Block 2 | |||
| Experimental smoking at baselined | 1.46(1.07–2.00) | 1.43(1.19–1.73) | 0.96(0.55–1.66) |
| Regular smoking at baselined | 2.86(1.72–4.75) | 3.19(2.28–4.46) | 2.08(0.88–4.90) |
| Block 3 | |||
| Smoking adolescent*Gender | |||
| Male, adolescent experimental smokinge | 1.69(0.89–3.22) | 0.74(0.51–1.07) | 0.46(0.14–1.52) |
| Male, adolescent regular smokinge | 1.15(0.41–3.27) | 0.96(0.51–1.83) | 2.26(0.44–11.69) |
| Smoking adolescent*Smoking mother | |||
| Smoking mother, adolescent experimental smokingf | 2.55(1.26–5.14) | 1.62(1.06–2.48) | 1.74(0.51–5.96) |
| Smoking mother, adolescent regular smokingf | 2.98(0.93–9.56) | 1.22(0.62–2.41) | 0.62(0.10–4.02) |
| Smoking adolescent* Smoking father | |||
| Smoking father, adolescent experimental smokingg | 0.58(0.29–1.14) | 0.73(0.49–1.09) | 0.47(0.14–1.52) |
| Smoking father, adolescent regular smokingg | 2.32(0.74–7.33) | 0.77(0.39–1.51) | 0.23(0.04–11.69) |
†Prediction of current wheeze at follow-up among adolescent with never wheeze at baseline (N = 5,686). |
††Prediction of IoA at follow-up among adolescents without IoA at baseline (N = 5,000). |
†††Prediction of current diagnosis of asthma at follow-up among adolescents without current diagnosis of asthma at baseline (N = 6,723) |
aReference category is females. |
bReference category is non-smoking mothers at baseline. |
cReference category is non-smoking fathers at baseline. |
dReference category is never smoking at baseline. |
eReference category is female never smoking adolescent |
fReference category is non-smoking mother, never smoking adolescent |
gReference category is non-smoking father, never smoking adolescent |
Table 5. Prediction of symptom severity at follow-up, by parental and adolescent smoking at baseline (N = 7,426): Results of analyses of variance
| Mean (SE) | F | p | |
|---|---|---|---|
| Model 1 | |||
| Gender | 219.54 | 0.000 | |
| Female | 1.32(0.005) | ||
| Male | 1.23(0.005) | ||
| Smoking behavior mother at baseline | 6.33 | 0.012 | |
| Non-smoking mother | 1.26(0.004) | ||
| Smoking mother | 1.28(0.006) | ||
| Smoking behavior father at baseline | 8.61 | 0.003 | |
| Non-smoking father | 1.26(0.005) | ||
| Smoking father | 1.28(0.005) | ||
| Model 2 | |||
| Smoking behavior adolescent at baseline | 25.18 | 0.000 | |
| Never smoker | 1.26(0.005) | ||
| Experimental smoker | 1.29(0.006) | ||
| Regular smoker | 1.33(0.013) | ||
| Model 3 | |||
| Smoking adolescent*Gender | 5.09 | 0.006 | |
| Never smoker & male | 1.22(0.006) | ||
| Experimental smoker & male | 1.24(0.008) | ||
| Regular smoker & male | 1.27(0.019) | ||
| Never smoker & female | 1.29(0.006) | ||
| Experimental smoker & female | 1.35(0.008) | ||
| Regular smoker & female | 1.39(0.017) | ||
| Smoking adolescent*Smoking mother | 4.11 | 0.016 | |
| Never smoker & non-smoking mother | 1.25(0.006) | ||
| Experimental smoker & non-smoking mother | 1.28(0.007) | ||
| Regular smoker & non-smoking mother | 1.32(0.018) | ||
| Never smoker & smoking mother | 1.25(0.008) | ||
| Experimental smoker & smoking mother | 1.31(0.009) | ||
| Regular smoker & smoking mother | 1.34(0.018) | ||
| Smoking adolescent* Smoking father | 0.64 | 0.527 | |
| Never smoker & non-smoking father | 1.25(0.006) | ||
| Experimental smoker & non-smoking father | 1.29(0.008) | ||
| Regular smoker & non-smoking father | 1.31(0.019) | ||
| Never smoker & smoking father | 1.26(0.007) | ||
| Experimental smoker & smoking father | 1.31(0.009) | ||
| Regular smoker & smoking father | 1.35(0.017) |
Adolescent smoking at baseline predicted a higher incidence of current wheeze (OR = 1.46 for experimental versus never smoking; OR = 2.86 for regular smoking versus never smoking) and IoA (OR = 1.43 for experimental versus never smoking; OR = 3.19 for regular smoking versus never smoking) and increased symptom severity at follow-up (mean = 1.26, 1.29, and 1.33; respectively for never, experimental, and regular smokers). No significant effects of baseline smoking on the incidence of current diagnosis of asthma were found.
A significant interaction between adolescent smoking and gender was found for symptom severity. As indicated by the mean scores in Table 5, the increase in severity scores because of adolescent smoking was twice as high among females (increase of .10 between never and regular smoking) compared with males (increase of .05 between never and regular smoking).
Further, significant interactions between adolescent smoking at baseline and maternal smoking were found for the incidence of current wheeze and IoA and for the development of symptom severity. The results of severity in Table 5 and the results of the separate analyses (Table 3) for current wheeze and IoA show that the effect of adolescent smoking on asthma and symptoms of asthma depended on whether or not the mother smoked. Adolescent smoking was a stronger predictor of the incidence of current wheeze and IoA when the mother was a smoker. Likewise, when adolescent smoking level increased, the increase in symptom severity scores was higher when the mother was a smoker.
Discussion
The purpose of this longitudinal study was to investigate whether smoking is predictive of asthma development or whether asthma predicts smoking onset among adolescents. The results of this study suggest bidirectionality: on the one hand, having (symptoms of) asthma predicted smoking onset, and on the other hand, baseline smoking was found to increase the risk of developing asthma and symptoms of asthma. Gender and maternal smoking moderated these effects.
The results provide support for the predictive value of asthma on the development of smoking behavior. Although adolescents with current diagnosed asthma were less likely to start experimenting with smoking, if they started smoking during the course of the study, they were more often regular smokers than experimenters, indicating an accelerated development in smoking behavior among adolescents with asthma. So even though the diagnosis of asthma seems to make adolescents reluctant to start smoking, once they have tried it, they seem to move faster into more persistent and regular smoking habits. A similar result was found for symptom severity: adolescents experiencing more symptoms at baseline had a higher risk of becoming regular smokers. This could be explained by short-term beneficial effects of smoking. Experimental mice studies in which asthma was induced before smoking exposure have shown short-term anti-inflammatory effects of smoking on allergic inflammation [25], [26]. It has been speculated that these short-term beneficial effects of smoking might be caused by one or several of the known components in cigarette smoke. Specifically carbon monoxide (CO), nitric oxide (NO), and nicotine have been suggested [25]. A cross-sectional study on humans suggested likewise [27]. Another explanation could be that adolescents with asthma are more susceptible to develop nicotine dependence. This is in line with research on COPD, where nicotine dependence is higher among patients with COPD compared to healthy smokers [28].
This study also showed gender differences in the association between asthma and smoking onset. Although higher severity and IoA have a protective effect on experimenting with smoking among boys, it is a risk factor for smoking among girls. One explanation for this gender difference could be that girls have lower self-esteem than boys [29], and especially for girls with asthma, smoking may be a way to increase self-esteem. However, the results could also be explained by earlier onset of adolescence in girls. In adolescence, rebellion is a way to strive for independence. Given the age of the cohort, girls would be more rebellious as a result of earlier maturation. For girls with asthma, smoking would even be a stronger symbol of rebellion than for girls without asthma.
Concerning the effect of maternal smoking, the findings of this study replicate earlier studies showing an increase in wheezing and non-diagnosed asthma due to ETS [30]. However, to our knowledge, this is the first study to show that maternal smoking moderates the effect of asthma and respiratory symptoms on smoking onset. (Only parental baseline smoking is included in the models for smoking onset. The models were not corrected for changes in parental smoking during the follow-up period, but these changes could influence adolescent smoking onset as well, although few parents changed smoking status in the 22-month period [e.g., 96% of the non-smoking mothers and 94% of the non-smoking fathers at baseline were still non-smokers at follow-up]). Additional analyses on smoking status were also conducted with parental smoking status at follow-up added to the first step of the model (including the confounding variables, gender, and parental smoking at baseline). None of the significant interactions between asthma and maternal smoking changed. They all remained significant with similar odds ratios. Having asthma or symptoms of asthma is only a risk factor of adolescent smoking onset when the mother smokes, and this was shown consistently across all measures of asthma. In families where the mother was a non-smoker, having asthma or symptoms of asthma was a protective factor for starting experimenting with smoking, whereas in families with maternal smoking, adolescents with symptoms of asthma or with diagnosed asthma were up to 2.3 times more likely to start experimenting with smoking than those without (symptoms of) asthma. This might be explained by an increased modeling effect. According to the social learning theory [31], parents who smoke act as role-models for their adolescents and this modeling might be stronger among asthmatic adolescents. If they see their parents smoking even though they have asthma, they might interpret this as smoking not being really harmful for people with asthma; otherwise their parents would not smoke [24]. To test this modeling hypothesis, we tested three-way interactions between parental smoking (father and mother separately), adolescent asthma and adolescent gender on smoking onset, to see whether girls are more influenced by mothers, and boys by fathers, especially among the asthmatic adolescents. However, only two of the 24 tested interactions were significant, and no consistent pattern was detected. So no support was found that the interaction between maternal smoking and asthma status is different for boys and girls. Or interpreted in another way: the possible two-way interaction between maternal or paternal smoking and adolescent gender is not different for adolescents with asthma and adolescents without asthma. So it is still possible that girls are more influenced by mothers, and boys more by fathers, but if this is the case, the three-way interactions do not show that this was different for asthmatic adolescents.
With regard to the prediction of asthma by smoking, the results of previous studies on adolescents have provided contradictory results [11], [12], [13], [14]. Our findings show that even early in adolescence, when smoking behavior has just begun, it already increases the change of symptoms of asthma, with a clear dose-response effect. This is in line with recent studies [11], [14], showing increasing risk for onset of asthma with increasing levels of active smoking among adolescents.
Baseline smoking predicted higher symptom severity among both genders, but the increase in severity due to smoking was twice as strong among females. (To rule out that this is caused by a larger number of cigarettes smoked by girls we conducted a t test. Among the smoking adolescents, girls and boys did not differ in their mean number of cigarettes smoked per week: mean number of cigarettes per week was 29.00 [SD 1.65] for boys and 30.17 [SD 1.61] for girls, p > .5). This might be explained by the same process identified as possible explanation for changes in the gender ratio in asthma (in childhood and early adolescence males are more likely to have asthma than females whereas this changes during adolescence, when males are less likely to have asthma) [15]. Testosterone is an immunosuppressant, and this could protect males from developing asthma, whereas female sex steroids will increase the risk of atopic diseases because these steroids are proinflammatory [16]. Therefore, females may be more susceptible to the effects of smoking. No gender differences in the effect of smoking were found for onset of wheeze, IoA, or current diagnosed asthma, which is in line with a recent study showing that the association of regular smoking and adolescent onset of asthma did not differ by gender [14].
Another important finding is that maternal smoking also moderated the effect of adolescents smoking, after controlling for main effects of maternal and adolescent smoking. When the mother is a smoker, adolescents who smoke have an increased risk for developing asthma symptoms. Among those whose mothers did not smoke, there is only a small increase in development of symptoms due to active smoking. In this study, no measures of maternal smoking during pregnancy were included, only smoking at baseline. Since it is likely that a proportion of the smoking mothers were also smoking during pregnancy, the results could be explained by in utero exposure to smoke. A recent study showed that regularly smoking adolescents exposed to maternal smoking in utero were at higher risk than non-exposed regularly smoking adolescents, both compared with non-smoking non-exposed adolescents [14]. So even though the results indicate an effect of maternal smoking, the results of this study cannot discern to what extent this is caused by maternal smoking during pregnancy. In addition, maternal smoking is a likely source of exposure to environmental tobacco smoke (ETS) during childhood and adolescence. The cumulative exposure of in utero exposure and ETS could increase the risk of bronchial hyperreactivity, which in combination with adolescent smoking may increase the risk of developing asthma.
This is the first study investigating the directionality of the relationship between smoking and asthma symptoms. This was tested among a large community sample with validated questionnaires to assess asthma and symptoms of asthma. Because we used multiple indicators of asthma, we were able to replicate the main effects in this study over different measures. However, this study suffers from some limitations. First, the follow-up period was brief, only 22 months. For studies on the onset of smoking among adolescents, this period is in line with many studies on smoking onset [23], [32]. The design of our study allows us to follow adolescents from approximately age 13–15 years of age. In the Netherlands, the largest increase in smoking behavior can be found between teens 13–16 years of age [33]. Because we measured smoking onset in a developmental period in which many adolescents start smoking, our study period of 22 months is enough to detect predictors of the onset of smoking (primarily experimentation, not highly advanced stages like daily smoking or nicotine dependence). Nevertheless, a 22 month follow-up period was short for studying the development of asthma. Because of this shorter period, less asthma and symptoms of asthma might have developed, lowering the statistical power of the analyses. However, if predictive factors of asthma development are found, even with fewer new cases, this does certainly shed light on predictors of asthma development during adolescence.
Second, asthma and symptoms of asthma were assessed by self-report, which may have influenced odds ratios. However, the ISAAC questionnaire was designed for population-based research [34] and proved to be a valid way to measure the prevalence of atopic diseases [18], [35]. A comparison of the ISAAC questionnaire with a physician’s assessment of asthma demonstrated sensitivity and specificity [36]. A validation study of the ACAAI questionnaire showed that the cut-off score of three or more affirmative responses had an estimated sensitivity of 80% and a specificity of 70% for an asthma clinical consensus designation according to a validation study [21]. However, some of the effects of smoking on asthma seem inconsistent. Significant effects were found for current wheeze, IoA, and symptom severity, but not for current diagnosed asthma. This is probably due to the relatively small number of new cases of asthma (n = 82). However, the effect of adolescent smoking is in the same direction (OR = 2.08; p < .1 for regular smoking), suggesting that with a larger incidence, this effect would be significant as well. Similar inconsistencies were found for the effect of the different asthma measures on smoking. Although the correlations between the four asthma measures were moderate to strong, they measure different aspects of asthma, and could have been differentially biased by the use of self-reports (e.g., memory bias is more likely for reporting “wheeze during the last 12 months” than for “asthma diagnosed by a doctor”).
Adolescent smoking was assessed by a single item on a nine-point scale, and this variable was recoded in three categories (never smokers, experimental smokers, and regular smokers). In recent publications of longitudinal studies using the same item, six categories of smoking initiation were identified (never smoker, trier, experimenter, regular smoker, nonsmoking decider, and quitter) [37], [38]. In the present paper, the same categorization is used for never smokers and regular smokers. However, triers, experimenters, nonsmoking deciders, and quitters are combined into one experimental smokers category, consisting of all adolescents who have initiated smoking since baseline, but were not regular smokers at follow-up. Using six categories of smoking initiation in the non-smokers at baseline would have resulted in a too small number of observations in some subgroups of the analyses, especially in the interaction analyses for the subgroups of adolescents with current wheeze and current diagnosed asthma. Therefore, we used a very broad definition of experimental smoking, which could have an impact on the odds ratios in this study.
Furthermore, smoking was assessed by self-reports. However, confidentiality was guaranteed, and studies showed that self-reports on smoking are reliable and valid in the case of guaranteed anonymity, and estimates of smoking rates were similar for self-reports and biological markers such as cotinine [39]. Information about parental smoking was also collected by adolescent-reports. Studies have shown that adolescents are a very reliable source to report on current parental smoking status [40]. However, parental smoking was assessed with a single question only, and we did not assess history and quantity of smoking and whether the parents smoke in front of the child.
Like all longitudinal studies, this study suffered from a drop-out in the 22 months between the waves (26.4%). A logistic regression analysis including all variables of this study showed differences with regard to ethnicity, age, education level, baseline smoking status and baseline symptom severity between non-respondents and participants at follow-up. No significant differences in gender, IoA, current wheeze, and current diagnosed asthma were found. All variables together only explained 2.3% of the variance in drop-out suggesting that potential selective attrition is rather limited.
In conclusion, the relationship between asthma and smoking in adolescence seems to be bidirectional. The prediction of smoking by asthma, as well as the prediction of asthma by smoking is stronger in females and in families where the mother is a smoker. Because of the higher smoking levels of adolescents with symptoms of asthma, it is important for health professionals to discuss smoking with adolescents with respiratory problems. Furthermore, mothers should be addressed in particular, because maternal smoking is a stronger predictor of smoking onset among adolescents with asthma.
This study also extends earlier results showing increased incidence of symptoms of asthma caused by active smoking. Female smokers and smokers exposed to maternal smoking were most likely to develop asthma or have increased symptoms of asthma. With regard to the development of asthma, tobacco prevention programs should pay more attention to this increased risk of respiratory symptoms. In particular in families of adolescents with respiratory conditions and symptoms effort should be made to inform mothers about the risk of their own smoking behavior for their children.
Acknowledgments
This research was funded by a grant from the Dutch Asthma Foundation. The contribution of Rutger Engels was supported by the Netherlands Organisation for Scientific Research. The authors are grateful to all schools and adolescents who participated in the project, and to the students who assisted with data collection.
References
- . Asthma prevalence and management in Australian adolescents: results from three community surveys. J Adolesc Health. 1992;13:707–712
- . Risk factors for smoking among adolescents with asthma. J Adolesc Health. 2002;30:279–287
- . Smoking patterns among adolescents with asthma attending upper secondary schools: a community-based study. Pediatrics. 2003;111:e562–e568
- . Asthma, health behaviors, social adjustment, and psychosomatic symptoms in adolescence. J Asthma. 1996;33:157–164
- . Adolescent smoking and early respiratory disease: a longitudinal study. Ann Allergy. 1987;59:135–140
- . Brief report: social risk factors predict cigarette smoking progression among adolescents with asthma. J Pediatr Psychol. 2006;31:246–251
- . Parental substance use as a modifier of adolescent substance use risk. Addiction. 2002;97:1537–1550
- . Predictors of smoking development in a population-based sample of adolescents: a prospective study. J Adolesc Health. 2004;35:172–181
- . Prospective study of asthma in relation to smoking habits among 14,729 adults. Thorax. 1988;43:534–539
- . Incidence of asthma in adults—report from the Obstructive Lung Disease in Northern Sweden Study. Allergy. 1997;52:1071–1078
- Smoking and the incidence of asthma during adolescence: results of a large cohort study in Germany. Thorax. 2006;61:572–578
- . Incidence of asthma in Swedish teenagers: relation to sex and smoking habits. Thorax. 1995;50:260–264
- . Risk factors for development of asthma in children and adolescents: findings from a longitudinal population study. Respir Med. 1996;90:623–630
- Regular smoking and asthma incidence in adolescents. Am J Respir Crit Care Med. 2006;174:1094–1100
- . Gender differences in airway behaviour over the human life span. Thorax. 1999;54:1119–1138
- . Therapeutic implications of sex differences in asthma and atopy. Arch Dis Child. 2003;88:587–590
- . Health effects of passive smoking. 6 (Parental smoking and childhood asthma: longitudinal and case-control studies). Thorax. 1998;53:204–212
- International Study of Asthma and Allergies in Childhood (ISAAC): rationale and methods. Eur Respir J. 1995;8:483–491
- . Atopic diseases and related risk factors among Dutch adolescents. Eur J Public Health. 2006;16:549–558
- . Gender differences in respiratory, nasal and skin symptoms: 6–7 versus 13–14–year-old children. Acta Paediatr. 1999;88:147–149
- Development and validation of school-based asthma and allergy screening questionnaires in a 4-city study. Ann Allergy Asthma Immunol. 2004;93:36–48
- . “Kicking the initiation”: do adolescent ex- smokers differ from other groups within the initiation continuum?. Prev Med. 2001;33:392–401
- . Influence and selection processes in friendships and adolescent smoking behaviour: the role of parental smoking. J Adolesc. 2004;27:531–544
- . Parental smoking and smoking behavior in asthmatic and nonasthmatic adolescents. J Asthma. 2005;42:349–355
- Short-term smoke exposure attenuates ovalbumin-induced airway inflammation in allergic mice. Am J Respir Cell Mol Biol. 2004;30:880–885
- Mainstream cigarette smoke exposure attenuates airway immune inflammatory responses to surrogate and common environmental allergens in mice, despite evidence of increased systemic sensitization. J Immunol. 2005;175:2834–2842
- . Smoking and airway inflammation in patients with mild asthma. Chest. 2001;120:1917–1922
- Smoking characteristics: differences in attitudes and dependence between healthy smokers and smokers with COPD. Chest. 2001;119:1365–1370
- . Gender differences in self-esteem: a meta-analysis. Psychol Bull. 1999;125:470–500
- . Effects of maternal smoking during pregnancy and environmental tobacco smoke on asthma and wheezing in children. Am J Respir Crit Care Med. 2001;163:429–436
- . Social Learning Theory. Englewood Cliffs, NJ: Prentice-Hall; 1977;
- . Self-comparison processes, prototypes, and smoking onset among early adolescents. Prev Med. 2005;40:785–794
- Stivoro. Roken, de harde feiten: Jeugd 2005. Den Haag: Stivoro-voor een rookvrije toekomst; 2006.
- Worldwide time trends in the prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and eczema in childhood: ISAAC Phases One and Three repeat multicountry cross-sectional surveys. Lancet. 2006;368:733–743
- Measuring the prevalence of bronchial hyper-responsiveness in children. Int J Epidemiol. 1995;24:597–602
- Validation of questionnaire and bronchial hyperresponsiveness against respiratory physician assessment in the diagnosis of asthma. Int J Epidemiol. 1996;25:609–616
- The European Smoking Prevention Framework Approach (EFSA): an example of integral prevention. Health Educ Res. 2003;18:611–626
- . Development and longitudinal test of an instrument to measure behavioral stages of smoking initiation. Subst Use Misuse. 2004;39:225–252
- . An assessment of the validity of adolescent self-reported smoking using three biological indicators. Nicotine Tob Res. 2003;5:473–483
- . Correspondence between proxy and self-reports on smoking in a full family study. Drug Alcohol Depend. 2005;84:40–47
PII: S1054-139X(07)00234-0
doi:10.1016/j.jadohealth.2007.05.015
© 2007 Society for Adolescent Medicine. Published by Elsevier Inc. All rights reserved.
