STUDY POPULATION

The Finnish twin cohort is a population based sample of like-sex adult twin pairs selected from the Central Population Registry of Finland as pairs of individuals with the same birth date and sex, the same surname, and local community at birth.3 A health questionnaire was mailed to the cohort members in the autumn of 1975 with a question whether asthma had ever been diagnosed by a physician. The response rate was 89%.

In Finland, National Sickness Insurance (NSI) has covered 80–100% of the costs of drug treatment for chronic bronchial asthma since 1970 (including corticosteroids, corticotrophins, cromoglycate, sympathomimetics, theophylline and its derivatives, hydroxyzine, ipratropium, oxytropium, nedocromil, and some combinations). A prerequisite for being registered is an accurate diagnosis of asthma by a specialist in allergy, pulmonary diseases, internal medicine, or paediatrics, or a diagnosis based on an appropriate hospital examination and acceptance by a consultant physician from the Social Insurance Institution of Finland.4 The National Board of Health maintains a register of all patients discharged from the hospital in Finland since 1969, including their diagnoses.5 Permissions for the use of register data were obtained from the national authorities. Data from the registers of the Social Insurance Institution and National Board of Health were obtained by linking the unique social security numbers given to all residents in Finland.

Respondents reporting asthma or with missing data on asthma or on smoking in the 1975 questionnaire, patients entitled to reimbursement for asthma medication and those treated in hospital for asthma before 1977 were excluded (n=1546). This population, initially free of asthma, was then confined to twins alive at the beginning of 1977 aged 20–50 in 1977 (n=20 472). Twin pairs discordant for the right to reimbursed asthma medication in 1977–93—that is, the other twin had that right while his/her twin sibling did not—were then identified (318 pairs) and three pairs were further excluded because the non-medicated twin had been treated in hospital for asthma during follow up. In addition, 18 twin pairs were concordant for asthma medication.

The studied variables were chosen from available data of the Finnish twin cohort on the grounds of previous findings. A history of atopic diseases was based on whether allergic rhinitis (including hay fever), allergic eczema, or urticaria had ever been diagnosed by a physician. Respondents were classified as non-smokers (including occasional smoking), former or current smokers. The number of cigarettes smoked daily was asked separately for former and current smokers.3 Social class was based on occupation: upper and lower professionals, skilled and unskilled workers, farmers and unclassified (students, conscripts, housewives, etc).3Education was measured by the schooling completed by the respondent (nine alternatives) and divided into lower (at most, primary school for 6–7 years and possibly some vocational training) and higher (at least, junior high school for 9 years) grades. Physical activity was measured by a series of questions about leisure time physical activities.6 They were also asked if there were any domestic animals in the household. Standard questionnaire assessment of chronic bronchitis and dyspnoea was used.3

DATA ANALYSIS AND STATISTICAL METHODS

Within the cohort nested case-control analyses were conducted among twin pairs discordant for incident asthma and with no missing data on study variables (n=262 pairs) to examine whether the risk of asthma in cases in relation to chosen determinants differed from the risk of their twin siblings without that determinant. The significance of the relative risk was assessed by McNemars' two sided test (dichotomous variables) and by Bowkers' test of symmetry (multicategorical variables). Conditional logistic regression for 1–1 matched data was then used to calculate risk ratios and their confidence limits to find out the difference in the asthma risk of the twin with the determinant in question (for example, atopy) compared with that of his/her co-twin without that determinant.7All statistical analyses were computed by SAS (SAS Institute, Cary, NC, USA).