Study Design and Population
The present study is part of the FinEsS (Finland-Estonia-Sweden) study, which is a postal questionnaire study on respiratory epidemiology conducted in collaboration in these three Northern European countries. Similar postal surveys were conducted in 1996, 2006 and 2016. The present study sample is part of the latest survey conducted in Finland in February 2016 and is formed from a random sample of 8000 subjects aged 20–69 years from the population in western Finland (Hospital Districts of Vaasa and Seinäjoki). The study sample was obtained from the Finnish Population Register and it was matched to the age and gender distribution of the population in the geographical area of our study. Finland is a bilingual country and the registered native language of a subject determined whether questionnaire in Finnish or Swedish language was used. The questionnaire was sent to a random sample of 7986 subjects after exclusion of subjects with unknown address. Two reminders were sent to those not responding. The sample size was 7942 subjects after further exclusion of subjects with non-analyzable data as shown in Figure 1. In total, 4173 subjects responded yielding to a response rate of 52.3%. Of the responders, 206 were excluded because of missing data regarding smoking habits and thus, the actual sample size was 3967 responders included in the study. The study protocol was approved by the Ethical Committee of Helsinki University Hospital (approval number 200/13/03/00/15).
The study area is mainly rural with two major towns (Seinäjoki and Vaasa, about 62,000 and 68,000 inhabitants, respectively). It has a subarctic climate and the yearly average temperature is 4 °C (from −7 °C in the winter to 17 °C in the summer). The most common allergic sensitizations in Finland are against dogs, cats and pollens whereas sensitization to house dust mites and molds is less common.
Questionnaire and Definitions
The FinEsS questionnaire was developed from the Obstructive Lung Disease in Northern Sweden (OLIN) questionnaire, which is modified from the Swedish translation of the British Medical Research Council (BMRC) questionnaire. The questionnaire includes questions on respiratory diseases, symptoms, medication and comorbidities, risk factors and occupational factors considered relevant in respiratory epidemiology.
A physician-diagnosed asthma was defined by an answer "yes" to the question "Have you been diagnosed by a doctor as having asthma?". Age at asthma diagnosis was determined by an answer to the question" What age were you when asthma was diagnosed?". Allergic rhinitis was defined by an answer "yes" to either of the questions "Have you been diagnosed by a doctor as having allergic rhinitis caused by pollens (caused by e.g. birch, grass, mugwort)?" or "Have you been diagnosed by a doctor as having other allergic rhinitis (caused by e.g. cat or dog, but not pollen)?". Allergic conjunctivitis was defined by an answer "yes" to the question "Have you been diagnosed by a doctor as having symptoms of allergy in your eyes?". Age at diagnosis of allergy was not asked for. We used the presence and absence of allergic rhinitis as an indication of asthma being allergic or non-allergic, respectively. A sensitivity analysis was made by using the presence of either allergic rhinitis, allergic conjunctivitis or both as an indication of allergic asthma. Current smokers were considered those who reported smoking currently or during the 12 months preceding the survey. Ex-smokers reported previous smoking but had quit smoking at least 1 year prior to the survey. Never smokers did not report current or previous smoking.
Reconstructing Age-specific Incidence of Asthma From Cross-sectional Data. Incidence of asthma in different age groups was estimated based on cross-sectional data on current age of the responders and age at diagnosis of asthma.[22,23] Longitudinal data were retrospectively reconstructed from the questionnaire data as if the 3967 subjects were a cohort of newborns recruited between 69 and 20 years ago. A "time-to-event" (age at diagnosis of asthma) was recorded for each individual, and the population at risk at each age was updated by subtracting both events (subjects reporting asthma diagnosed at younger age) and censorships (asthma-naïve responders younger than the age for which population at risk was calculated) from the total sample.
In brief, subjects were divided into 10-year age groups based on their current age, and annual incidence of asthma per 1000 person-years (new asthma diagnoses/1000/year) was calculated in each age group by dividing the number of incident asthmas in each group by age-group-specific population at risk, dividing the result by 10 and further multiplying with 1000. The 10-year age group -specific population at risk was a mean value of annually calculated respective 10-year risks. With respect to age 0, population at risk was all responders. For ages 1–20 years, subjects reporting asthma diagnosed at younger age than the age for which the population at risk was calculated were subtracted to form the 1-year population at risk. The youngest responders were 20 years of age at the time of the study. After age 20, the responders who did not report physician-diagnosed asthma (i.e. asthma-naïve responders) and who were younger than the age for which the population at risk was calculated, were also subtracted from all responders to calculate populations at risk for ages 21–69 years. Subjects reporting physician-diagnosed asthma but not the age at diagnosis were excluded from calculations.
Controlling for Differences Between Older and Younger Age Groups. When calculating incidence based on cross-sectional data, incidence rates at younger age represent means from several different age cohorts while incidence rates at older age represent those from older age cohorts only. Since different age cohorts may also have different overall incidence of atopy, relative proportion of allergic and non-allergic asthma may vary between different age cohorts and thus, might affect our estimates of early-onset and late-onset asthma. Therefore, we calculated separately in different age groups the proportion of allergic asthma among subjects with asthma diagnosed before the age of 40 years.
Statistical Comparisons. Statistical analyses were performed using SPSS software version 23 (IBM Corporation, Armonk, NY) and 95% confidence intervals (CI) were calculated with EpiTools by using the Wilson method. Mann-Whitney U test was used for continuous variables and Pearson chi-square test for categorical variables in comparisons between two groups. A p-value <0.05 was considered significant. Results are presented as percentages (95% CI) or medians (Interquartile range [IQR]).
BMC Pulm Med. 2020;20(9) © 2020 BioMed Central, Ltd.