Methods
Study Population
The study area was the county of Nord–Trøndelag (22 463 km), Norway. The adult population in Nord–Trøndelag (127 000 people in 1995) was invited to participate in three HUNT surveys which have been described elsewhere. The population in Nord–Trøndelag is fairly representative of Norway with respect to age, income, morbidity and mortality. The participants of this study took part in the second (HUNT 2, 1995–97) and third (HUNT 3, 2006–08) surveys of the HUNT, with an average 11 years of follow-up (n = 37 071). We studied 23 599 participants who were 19–55 years of age and free from asthma at baseline. The age limit was set as <65 years in HUNT 3 for two reasons: (i) to reduce misclassification of asthma and chronic obstructive pulmonary disease (COPD) and (ii) obesity increases overall mortality. Senior participants with obesity are more likely to die compared with those without obesity, which can result in an under-estimation of the association of interest. Of the 23 599 participants, we excluded a further 353 (1.5%) with missing information on anxiety, depression or body mass index at baseline and on asthma both at baseline and follow-up, leaving 23 246 participants for the analysis.
Anxiety and Depression Symptoms
Anxiety and depression symptoms were measured with the Hospital Anxiety and Depression Scale (HADS). The HADS consists of 14 questions of which seven measure anxiety symptoms and seven measure depression symptoms, during the past week. Each question is answered on a scale of 0–3, giving one total score for anxiety (HADS-A, range 0–21) and one total score for depression (HADS-D, range 0–21). A total score from 0 to 7 indicates normal state (normal), 8 to 10 indicates borderline anxiety or depression state (borderline) and 11–21 indicates caseness of anxiety or depression (caseness). In the analysis stratified by obesity, we combined the last two categories (score 8–21) to define presence of symptoms (yes, no). Because of the coexistence of anxiety and depression, we defined 'anxiety or depression symptoms' by the participants' highest score on either set of questions. Additional details of HADS are described elsewhere. The HADS was used to describe the presence of symptoms and is not a psychiatric diagnosis.
Obesity
Body mass index (BMI) was used as a measure of obesity. Weight was measured to the nearest half-kilogram (kg) and height was measured to the nearest centimetre. BMI was calculated as weight divided by height squared (kg/m) and obesity was defined as BMI ≥30.
Covariables
Covariables were collected from administrative questionnaires and clinical examinations at baseline. Covariables included age (19–29, 30–39, 40–49, 50–55 years), sex, current smoking (yes, no, unknown), duration of physical activity (<1h, 1–2h, ≥3 h/week, unknown), family history of asthma (yes, no, unknown), years of education (<10, 10–12, ≥13 years, unknown), receiving social benefit (yes, no, unknown) and economic difficulties (yes, no, unknown). Family history of asthma was defined as a close family member (father, mother, brother or sister) with asthma. Social benefit was defined as those that reported receiving sick pay, rehabilitation benefits, retraining benefits, disability pension, old age pension, family income supplement, unemployment benefits, transitional benefits, widow's pension or any other benefit. Participants with economic difficulties were those who reported difficulties meeting the cost of food, transport and/or housing.
Asthma Diagnosis
Asthma was defined by two approaches. In the first approach, asthma was defined by the survey question 'Do you have or have you had asthma?' Participants who reported 'no' to this question at baseline, and 'yes' at follow-up were classified as having incident asthma. In the second approach a stricter definition of asthma was used, i.e. no wheeze or asthma at baseline and reported asthma and use of asthma medication at follow-up, vs the reference group with no wheeze and no asthma at baseline or follow-up.
Statistical Methods
In the analysis cohort of 23 246 participants, we evaluated anxiety or depression symptoms at baseline and cumulative incidence of asthma, in relation to other baseline characteristics. Pearson chi-square tests were used to evaluate differences between categories. To study the association of anxiety or depression symptoms with incident asthma, we estimated odds ratios (ORs) with 95% confidence intervals (CIs) using logistic regression.
In Model I, we adjusted for age, sex, smoking, physical activity, family history of asthma, education, social benefit and economic history at baseline as potential confounding factors. In Model II, we additionally adjusted for BMI as a continuous variable. We found no substantial difference in the associations of anxiety or depression with incident asthma between sexes (data not shown); subsequently we presented data in women and men combined. To test if the associations differed between sexes, we used a multiplicative model. The multiplicative model is commonly used to assess heterogeneity of association measures such as relative risks or odds ratios between subgroups. To assess the joint association of anxiety or depression symptoms and obesity with incident asthma, we generated four subgroups, i.e. non-obese participants without anxiety or depression symptoms (reference group), non-obese participants with anxiety or depression symptoms (ORA), obese participants without anxiety or depression symptoms (ORB) and obese participants with anxiety or depression symptoms (ORAB). The joint association of anxiety or depression symptoms and obesity with incident asthma were estimated by using the relative excess risk due to interaction (RERI) with 95% CIs. The RERI was calculated using an additive model: RERI = ORAB − ORA − ORB + 1. The additive model is used to test biological interaction between two or more risk factors that together assert their influence on disease risk. In brief, RERI >0 and the lower limit of 95% CI >0 suggests a synergistic effect of anxiety or depression and obesity on incident asthma.
To test the robustness of the associations, we repeated the main analysis using the second approach to define incident asthma (n = 20 219). Further to this, we defined anxiety or depression symptoms as above, and additionally included those reporting use of antidepressant medication into the symptom group (n = 20 219). Finally, we excluded participants who reported having or ever having had chronic bronchitis, emphysema or COPD in the follow-up questionnaires (n = 20 048). We used STATA 12.0 for all statistical analyses (StataCorp LP, College Station, TX).
Ethics
The project was approved by the Regional Committee for Ethics in Medical Research. All participants signed informed consent for participation and the use of data in research.