Study design
The first study (EIMS; Epidemiological Investigation of Multiple Sclerosis) was designed as a population-based case–control study using incident cases of MS, with a study group comprising the population aged 16–70 years in geographically defined areas of Sweden. Cases were recruited via hospital-based neurology units as well as privately run neurology units in Sweden. All university hospitals participated in the study, and in total, 40 study centers reported cases of MS to the study. All cases were examined and diagnosed by a neurologist located at the unit in which the case was entered. Cases that did not fulfill the McDonalds criteria [2] at the time of this report were excluded (116 cases with clinically isolated syndrome). For each potential case, two controls were randomly selected from the national population register, taking into consideration the case’s sex, age (5-year age group) and residential area (county). If information could not be obtained from the control selected, then another control was chosen using the same principles. If a third control agreed to participate before the originally chosen controls had answered the questionnaire, all controls were included in the study. Completed questionnaires were obtained from 1,798 cases and 3,907 controls, the response proportion being 91 % for the case group and 69 % for the controls. The study period for this report was April 2005–March 2012.
In the second study (GEMS; Genes and Environment in Multiple Sclerosis), prevalent cases fulfilling the McDonald criteria were identified from the Swedish National MS-registry [3]. For each case, a control was randomly selected from the national population register matched for age, gender, and residential area at the time of the disease onset. With a response rate of 82 % for the cases and 66 % for the controls, the study comprised 6,085 cases with MS, distinct from those in EIMS, and 5,357 matched controls recruited between November 2009 and November 2011. Ethical approval for both EIMS and GEMS were obtained from the relevant ethics committees. The general structure of both studies have been reported previously [4, 5].
Data collection and definition of smoking
In both studies, information on exposures and other circumstances was collected using a standardized questionnaire containing questions about demographic and reproductive factors, heredity, previous health, body weight and height, lifestyle factors, occupational exposures and socioeconomic circumstances. Information on smoking was obtained by asking about current and previous smoking including duration of smoking, average number of cigarettes smoked per day, and type of cigarettes. Information on exposure to passive smoking was obtained by asking if the subjects had been daily exposed to environmental tobacco smoke at home or at work (supplementary table 1). The questions on smoking habits and exposure to passive smoking were identical in both questionnaires.
For each case, the time at the initial appearance of symptoms indicative of MS was used as an estimate of the disease onset, and the year in which this occurred was defined as the index year. Tobacco smoking was considered prior to the index year in the cases and during the same period of time in the corresponding controls. Subjects who had smoked during the index year were defined as current smokers, those who had stopped smoking prior to the index year were defined as ex-smokers, and people who had never smoked before or during the index year were defined as never-smokers.
In order to analyze whether age at smoking debut affects the association between MS and smoking, we categorized the smokers into groups based on when they started and stopped smoking (age). We further categorized the smokers into groups based on the amount of cigarettes smoked (pack-years) prior to index. One pack year is defined as 20 cigarettes smoked per day for 1 year. Those who had smoked more than 10 pack years prior to index were defined as heavy smokers. Subjects who reported they had been exposed to passive smoking before or during the index year were defined as exposed to passive smoking.
Statistical analysis
Using logistic regression, the occurrence of MS in subjects who had started and stopped smoking in different life periods was compared with that in never-smokers by calculating odds ratios with 95 % confidence intervals. Trend test for a dose response relationship regarding cumulative dose of smoking and risk of MS was performed by using a continuous variable for cumulative dose of smoking, expressed as pack-years, in a logistic regression model. In order to determine whether intensity or duration of smoking contributes most to the risk of MS, we separately examined the components comprising pack-years.
We performed matched analyses based on all available case–control sets, as well as unmatched analyses of the data based on all available cases and controls. In the matched analyses, we lost a significant number of cases and controls (2 % of the cases and 13 % of the controls in EIMS, and 42 % of the cases and 34 % of the controls in GEMS). In the matched design, each case in GEMS had one matched control (3,541 cases and 3,541 controls) whereas each case in EIMS on average had 1.92 matched controls (1,762 cases and 3,389 controls). However, the results were similar to those in the unmatched analyses and all reported results that were statistically significant in the unmatched analysis were significant in the matched analysis as well. Therefore, only the results from the unmatched analyses are presented in this report since these were in close agreement with those from the matched analyses but had a somewhat higher degree of precision.
All analyses were adjusted for age, gender, residential area (according to study design), ancestry, passive smoking, and study. In the analysis, age was categorized into the following 8 strata: 16–19, 20–24, 25–29, 30–34, 35–39, 40–45, 45–49, and 50–70 years of age. Assessment of ancestry was based on whether the subject was born in Sweden or not, and whether either of the subject’s parents had immigrated to Sweden. A subject who was born in Sweden, whose parents had not immigrated, was classified as Swedish. Passive smoking was dichotomized into those who had and had not been exposed to environmental tobacco smoke before the index year.
Other potential factors taken into consideration were heredity (having a first or second degree relative with MS or not), educational level (university degree or not), snuff use (yes/no), adolescent body mass index (≤27 versus >27 kg/m2), sun exposure habits during the last 5 years (high/low), and a history of infectious mononucleosis (yes/no), but these factors had minor influence on the results of the study and were not adjusted for in the final analyses. All analyses were conducted using statistical analysis system (SAS) version 9.