Exposure to passive smoking during adolescence is associated with an increased risk of developing multiple sclerosis

Abstract

Background:

Environmental factors are associated with acquiring multiple sclerosis (MS) particularly in adolescence.

Objective:

To test for association between MS and exposure to passive smoking at the age of 10–19.

Methods:

A total of 919 patients from the Danish MS Registry and Biobank and 3419 healthy blood donors who had not smoked before the age of 19 were targeted. We analyzed separately for each sex and for those never-smokers (cohort 1) and active smokers above the age of 19 (cohort 2). All participants completed standardized questionnaires about smoking and lifestyle. We matched cases and controls in the ratio of 1:2 by propensity scores discarding unmatchable individuals and used logistic regression adjusted for all covariates and interactions.

Results:

After matching, we included 110/213 male cases/controls and 232/377 female case/controls in cohort 1. In cohort 2, the numbers were 160/320 and 417/760, respectively. Among women in cohort 1, the odds ratio (OR) for MS by passive smoking at the age of 10–19 was 1.432 (p = 0.037) but in men it was 1.232 (p = 0.39). Among men in cohort 2, OR was 1.593 (p = 0.022) but among women it was only 1.102 (p = 0.44).

Conclusion:

Among never smokers, female MS cases were more often than female controls reported with passive smoking between the age of 10 and 19, and among smokers above the age of 19, male MS patients were more often than male controls reported with passive smoking.

Keywords

Multiple sclerosis passive smoking environmental risk factor adolescence blood donors epidemiology

Introduction

Multiple sclerosis (MS) is an inflammatory demyelinating disease in the central nervous system (CNS). With an uncertain etiology, interactions between environmental and genetic factors are assumed to play a key role.¹

Meta-analyses have revealed strong evidence for association between smoking and MS with odds ratios (OR) ranging from 1.34 to 1.74.^2,3 Nonetheless, nicotine may have a protective effect on MS, and hence, smoke constituents distinct from nicotine are thought to mediate the increased risk of MS among smokers.^4,5 Albeit not as damaging, passive smoking has also been found to have harmful effects on the lungs and is similarly proven to be associated with the risk of MS.⁶

Studies have suggested late childhood and adolescence as life periods with increased susceptibility to risk factors for MS.^7–9 Migration studies have shown that moving before the age of 12–15 from high to low MS incidence countries obtained a decrease in the expected risk of developing MS supporting that adolescence is associated with increased susceptibility to risk factors for MS.^10–13

The aim of this study was to compare the history of passive smoking between 10 and 19 years in both sexes between MS patients and healthy controls both in never-active smokers and in individuals who reported active smoking beyond the age of 19, to explore if the effect of passive smoking in youth also would penetrate the stronger effect of active smoking later in life.

Materials and methods

Design of the study

Case–control study.

Study population

Data of all MS patients in Denmark were retrieved from The Danish Multiple Sclerosis Biobank at Rigshospitalet, Copenhagen and The Danish Multiple Sclerosis Registry.¹⁴ All patients met the MS diagnostic McDonald 2005 or 2010 criteria.^15,16 During the period from October 2009 to December 2014, all patients were asked to complete a comprehensive questionnaire on lifestyle or environmental factors developed at the Karolinska Institute in Stockholm, Sweden.⁴ At the time of this study, 2062 (76%) patients had completed the questionnaire. The group of controls consisted of healthy Danish blood donors¹⁷ recruited from four major donor locations in the Greater Copenhagen area. The questionnaires, identical to those given to the MS patients, were collected together with blood samples on the same day from all control participants from October 2012 to December 2014. The response rates for the controls could only be calculated at two of the four donor blood sites and were 75% and 90% (average 83%), respectively, and it was assumed that the response rates in the other locations were similar. This study was performed in the period 2013–2019.

Results from the investigated case–control cohorts have previously been published regarding the effects of alcohol consumption and shift work on MS risk.^9,18

Inclusion criteria

In order to secure genetic homogeneity, we included only individuals who were born in Denmark, Norway, Sweden, Iceland, or the Faroe Islands and had parents from a Nordic country as well. To avoid reverse causality, patients with disease onset prior to the age of 20 years were excluded. To harmonize the age distribution between cases and controls, all participants born before 1945 and after 1989 were excluded. Furthermore, participants were excluded if the questionnaire missed data about active or passive exposure to tobacco and the age or time of exposure. We excluded all individuals who had been active smokers within the age of 10–19 years. With the intention to focus solely on the possible association between passive smoking and MS and acknowledging that active smoking could be a mediator, we used only never-smokers in the first analyses (cohort 1). In the second analysis, we used individuals who had been active smokers, but only those who had started smoking above the age of 19 (cohort 2) in order to investigate whether the effect of passive smoking in adolescence would also be detectable in these persons.

Exposure to passive smoking

In order to evaluate the total exposure to passive smoking, two questions regarding passive smoking at home and at work, respectively, were asked in the questionnaire: “Have you ever lived with one or more persons, who smoked indoor on a daily basis?” and “Have you ever been exposed to passive smoking at work on a daily basis?” The questions were followed by the opportunity to state the period of time in total years. The duration of total exposure in the ages between 10 and 19 years were subsequently registered. Data on passive smoking were evaluated in two ways: as a dichotomized (categorical) variable divided into exposed or unexposed to passive smoking and as a continuous variable with exposure to passive smoking measured in total years. Active smoking after age 19 concerned people who at any time had been regular active smokers irrespective of the duration of this exposure.

Alcohol consumption at the age of 15–19 years was dichotomized into either above or within The Danish Health Authorities’ recommended maximum weekly intake levels, which are 7 units (1 unit = 12 g of alcohol) for women and 14 units for men.¹⁹ Level of education was divided into three categories: primary and lower secondary school and unskilled; high school or skilled; and higher education or academic. Because only 3%–5% belonged to the lowest level of education, we merged category 1 and 2. Body mass index (BMI) is weight in kilograms divided by the square of the height in meters. BMI had been dichotomized into overweight (⩾25.0) and not (<25.0).

Confounding factors

Confounders are external factors causing bias by being associated with the dependent variables (case–control status) as well as with exposure (passive smoking at the age of 10–19) in the population. Controlling for confounding based on propensity scores for being exposed is not straight forward in case–control studies²⁰ because of the risk of collider bias.²¹ Instead we calculated propensity scores as the probability of being an MS case from the values of the covariates alcohol consumption at the age of 15–19, BMI, and educational level, and the propensity score was obtained by binary logistic regression and the “nearest neighbor method.” By closeness of the propensity scores, we matched each case with up to two controls. The propensity scores were calculated for each sex and for the active smokers and never-smokers separately. The reasons for this partition was the different sex distribution between cases and controls and that active smoking after the age of 19 could be a mediator between passive smoking in adolescence and MS.

Statistical analyses

We performed binary logistic regression for the binary outcome with MS status as the dependent variable and passive smoking between 10 and 19 years of age as independent variable, either as categorical or as number of years exposed. In addition to the propensity score matching, we adjusted for all other covariates and their interactions. This is because the propensity score is a one-dimensional vector of the effects of all included covariates, and it does not account for the intricate web of possible associations and interactions. Moreover, different combinations of values of the covariates can result in the same propensity score. We examined the data for multiple collinearity between the covariates by calculating the variance inflating fraction (VIF), which we considered as acceptable at values less than 20. The only pairs of variables with VIF >20 were age and year of birth, and we kept the more biological meaningful age at filling in the questionnaire, rather than the year of birth. OR between exposed versus unexposed to passive smoking and MS was investigated, either with passive smoking as a categorical variable regardless of passive smoke intensity or as the continuous variable; and years of exposure, where the OR expresses the increase in odds of MS for each additional year of exposure. We also stratified the analyses based on sex. Statistical significance was set at p-values <0.05. We used IBM SPSS statistics V25 for data management and analyses.

Ethical statement

Written informed consent was obtained from all MS patients and approved by the Committee on Health Research Ethics (KF-01-314009). Informed consent from blood donors were obtained through their participation in “The Danish blood donor study” approved by the Committee on Health Research Ethics (M-20090237).

Results

The numbers of patients and controls primarily targeted from the data sources were 2062 and 5315, respectively. After stepwise exclusion of individuals who did not fulfill the inclusion criteria as shown in Figure 1, we arrived at a data pool of 919 MS patients (649 women and 270 men) and 3419 healthy blood donor controls (1477 women and 1942 men). Missing or incomplete information only eliminated a small number of persons. Most of the exclusions could be attributed to the exclusion criterion of being active smokers before the age of 20, and this fraction was considerably higher among patients than control persons. Figure 2 shows a hypothetical direct acyclic graph (DAG) based on assumptions, as it would look before stratification, and it indicates that active smoking after the age of 19 could be a mediator along the path between passive smoking and MS. This is the reason why we split the cohorts into never-smokers and people who had been active smokers beyond the age of 19 years.

Figure 1.

Exclusion criteria flowchart.

Figure 2.

Direct acyclic graph showing relation between exposure to passive smoking within the age of 10–19 and the outcome, MS or control, and all the covariates before stratification and weighting. Active smoking after the age of 19 could be a mediator between exposure and effect, and to isolate the effect of passive smoking per se, we stratified the analyses by active smoking into two distinct studies.

Men

The cohort of male never-active smokers (cohort 1) consisted of 110 MS cases and 414 controls. Their baseline characteristics are shown in Table 1. Among the MS cases, 66 (60.0%) had been exposed to passive smoking at the age of 10–19, and among the controls it was 228 (55.1%). After propensity score matching and supplementary adjustment for alcohol consumption at the age of 15–19, age at filling the questionnaire, BMI, education, and the interactions between alcohol, BMI, education, and passive smoking, OR was 1.23 (95% confidence interval, CI = 0.77–1.98), p = 0.390 (Table 2). Thus, MS was not associated with passive smoking at the age of 10–19 in never-active smoking men.

Table 1.

Baseline variables.

Cohort 1 (never-active smoking)	Men				Women
	Before matching		After matching		Before matching		After matching
	CasesN = 110	ControlsN = 414	CasesN = 110	Controls N = 213	CasesN = 232	ControlsN = 408	CasesN = 232	ControlsN = 377
Mean age at questionnaire (IQR)	42.22 (34–49)	43.66 (34–53)	42.22 (34–49)	42.31 (34–51)	44.39 (37–51)	43.91 (34–54)	44.39 (37–51)	44.46 (35–54)
Year of birth, mean (IQR)	1969 (1962–1977)	1968 (1959–1978)	1969 (1962–1977)	1970 (1962–1978)	1966 (1960–1974)	1968 (1959–1978)	1966 (1960–1974)	1967 (1958–1977)
Exposed to passive smoking at the age of 10–19	66 (60.0%)	228 (55.1%)	66 (60.0%)	118 (55.4%)	139 (59.9%)	207 (50.7%)	139 (59.9%)	192 (50.9%)
Alcohol consumption at the age of 15–19 above maximum recommended level	14 (12.7%)	71 (17.1%)	14 (12.7%)	26 (12.2%)	34 (14.7%)	84 (20.6%)	34 (14.7%)	56 (14.9%)
BMI ⩾25.0 at the age of 20	20 (18.2%)	65 (15.7%)	20 (18.2%)	36 (16.9%)	35 (15.1%)	50 (12.3%)	35 (15.1%)	47 (12.5%)
Education^a
Level 1–2	40 (36.4%)	183 (44.2)	40 (36.4%)	84 (39.4%)	149 (64.2%)	257 (63.0%)	149 (64.2%)	238 (63.1%)
Level 3	70 (63.6%)	231 (55.8)	70 (63.6%)	129 (60.6%)	83 (35.8%)	151 (37.0%)	83 (35.8%)	139 (36.9%)
Cohort 2 (active smoking beyond the age of 19)	CasesN = 160	ControlsN = 1528	CasesN = 160	ControlsN = 320	CasesN = 417	ControlsN = 1069	CasesN = 417	ControlsN = 760
Age at questionnaire, mean (IQR)	43.4 (35–51)	43.7 (35–52)	43.4 (35–51)	43.3 (35–51)	40.9 (33–48)	43.1 (33–51)	40.9 (33–48)	41.3 (33–49)
Year of birth, mean (IQR)	1968 (1961–1975)	1968 (1960–1977)	1968 (1961–1975)	1969 (1961–1977)	1970 (1963–1978)	1969 (1961–1979)	1970 (1963–1978)	1971 (1963–1979)
Exposed to passive smoking age at the age of 10–19	103 (64.4%)	829 (54.3%)	103 (64.4%)	172 (58.3%)	256 (61.4%)	617 (57.7%)	256 (61.4%)	447 (58.8%)
Alcohol consumption at the age of 15–19 above maximum recommended level	25 (15.6%)	191 (12.5%)	25 (15.6%)	45 (14.1%)	40 (9.6%)	155 (14.5%)	40 (9.6%)	74 (9.7%)
BMI ⩾25.0 at the age of 20	29 (18.1%)	275 (18.0%)	29 (18.1%)	57 (17.8%)	71 (17.0%)	142 (13.3%)	71 (17.0%)	108 (14.2%)
Education^a
Level 1–2	71 (44.4%)	606 (39.7%)	71 (44.4%)	147 (45.9%)	252 (60.4%)	672 (62.9%)	252 (60.4%)	463 (60.9%)
Level 3	89 (55.6%)	922 (60.3%)	89 (55.6%)	173 (54.1%)	165 (39.6%)	397 (37.1%)	165 (39.6%)	297 (39.1%)

IQR: interquartile range; BMI: body mass index.

Education level 1: primary or lower secondary school or unskilled, 2: high school or skilled, and 3: higher education or academic.

Table 2.

Results of the binary logistic regression analyses of passive smoking versus MS in never-smokers (cohort 1).

Sex	Matching and adjustment for covariates	OR for MS with passive smokingAge 10–19 versus no passive smoking (95% CI)	p	OR for MS for each year increase in exposure to passive smoking at the age of 10–19 (95% CI)	p
Men	Unmatched Unadjusted	1.224 (0.978–1.877)	0.335	1.014 (0.968–1.062)	0.568
	Matched Unadjusted	1.208 (0.757–1.927)	0.429	1.013 (0.962–1.066)	0.622
	Matched Fully adjusted^a	1.232 (0.766–1.980)	0.390	1.016 (0.965–1.070)	0.555
Women	Unmatched Unadjusted	1.451 (1.047–2.012)	0.025	1.046 (1.009–1.084)	0.013
	Matched Unadjusted	1.440 (1.034—2.006)	0.031	1.046 (1.009–1.085)	0.015
	Matched Fully adjusted^a	1.432 (1.021–2.008)	0.037	1.046 (1.008–1.085)	0.018

MS: multiple sclerosis; OR: odds ratio; CI: confidence interval.

Figures in bold are the final results.

Adjusted for age, alcohol, BMI, education, and the interactions BMI × alcohol × education × passive smoking.

The results were opposite in men who had been active smokers beyond the age of 19 (cohort 2). Among the 160 MS cases, 103 (64.4%) had been exposed to passive smoking, and among the 1528 controls, 829 (54.3%) had been exposed to passive smoking (OR = 1.524; 95% CI = 1.086–2.138; p = 0.015), and after propensity score matching and adjustments, OR was 1.593 (CI = 1.070–2.372; p = 0.022). There was also a trend of a dose effect as OR increased with 4.4% for each extra year of exposure in the ages 10–19 (OR = 0.044; 95% CI = 1.00–1.09, but this difference did not exactly attain statistical significance: p = 0.051).

Women

Female never-active smokers (cohort 1) consist of 232 MS patients and 408 controls. Their baseline characteristics are shown in Table 1. Among the MS cases, 139 (59.9%) had been exposed to passive smoking at the age of 10–19, and among the controls it was 207 (50.7%). After propensity score matching and supplementary adjustment of the above-mentioned covariates and interactions, OR was 1.43 (95% CI = 1.02–2.01) and p = 0.037 (Table 2). Passive smoking also seemed to have a dose effect in women: for each extra year of exposure to passive smoking within the age of 10–19, the odds for MS increased with 4.6% on average (OR = 1.046 in propensity score matched with adjustment, 95% CI = 1.01–1.09, p = 0.018). Thus, MS was statistically significantly associated with passive smoking between the age of 10 and 19 in never-active smoking women.

However, among women who had been active smokers beyond the age of 19 (cohort 2), passive smoking at the age of 10–19 did not add to the risk of MS (see Table 3).

Table 3.

Results of the binary logistic regression analyses of passive smoking versus MS persons who have been active smokers but only after age 19 (cohort 2).

Sex	Matching and adjustment for covariates	OR for MS with passive smokingAge 10–19 versus no passive smoking(95% CI)	p	OR for MS for each year increase in exposure to passive smoking at the age of 10–19(95% CI)	p
Men	Unmatched Unadjusted	1.524 (1.086–2.138)	0.015	1.032 (0.996–1.070)	0.084
	Matched Unadjusted	1.555 (1.052–2.299)	0.027	1.041 (0.998–1.085)	0.065
	Matched Fully adjusted^a	1.593 (1.070–2.372)	0.022	1.044 (1.000–1.089)	0.051
Women	Unmatched Unadjusted	1.165 (0.924–1.468)	0.196	1.007 (0.982–1.033)	0.604
	Matched Unadjusted	1.113 (0.872–1.422)	0.389	1.004 (0.977–1.085)	0.770
	Matched Fully adjusted^a	1.102 (0.861–1.411)	0.440	1.003 (0.976–1.031)	0.808

MS: multiple sclerosis; OR: odds ratio; CI: confidence interval.

Figures in bold are the final results.

Adjusted for age, alcohol, BMI, education, and the interactions BMI × alcohol × education × passive smoking.

When combining cohorts 1 and 2 and adjusting for active smoking after the age of 20, the OR for MS with passive smoking became significant in men (OR = 1.43, p = 0.021) but not in women (OR = 1.207; p = 0.064). The term active smoking after the age of 19 was not a significant predictor of MS, neither in males nor in females. Exposure to passive smoking at the age of 10–19 did not predict post-adult active smoking in men in the combined cohort as 59.9% of the exposed and 59.6% of the unexposed became active smokers later in life (p = 0.942). In women, 68.0% of the exposed versus 63.0% of the unexposed became active smokers (p = 0.30).

Discussion

We found exposure to passive smoking in adolescence to be a strong risk factor for developing MS later in the life in women, increasing the odds by 43% in never-active smokers and with a statistically significant dose effect, as the odds for MS increased by 4.6% for each additional year of exposure within the age of 10–19. In male never-active smokers, the association did not attain statistical significance. For persons who had been active smokers after the age of 19, the sex difference was opposite: the association between passive smoking between 10 and 19 years was only significant in men, in whom the odds were increased with 59%, but the dose effect for additional years of exposure per year was not significant. The confounding effects of the covariates were weak, as propensity score matching and adjusting for the covariates and their interactions changed little. These findings are in accordance with the results from a Swedish population-based cohort study applying the same lifestyle questionnaire, in which the OR was 1.3 (95% CI = 1.1–1.6).⁶ However, the Swedish study used different data management and statistical analyses, as they registered exposure to passive smoking until the year of MS onset (named index year) for every case and subsequently applied the index year on sex and age-matched controls. They looked at the entire period from birth to MS onset, whereas our study explored the possible association between passive smoking from the age of 10–19 and MS. Furthermore, the Swedish study adjusted for numerous MS risk factors, regardless of the association to passive smoking. Despite these differences, the almost identical results between our study and the Swedish evidently support the concept of an association between passive smoking and MS.

Results have differed between some of the other studies^22–24 with ORs ranging from 0.87 to 2.5, but they have not been inconsistent with this study. These varying findings are possibly explained not only by alternative selections of control groups but also by inconsistency in defining and quantifying exposure to passive smoking. This complicates a meaningful comparison between the study outcomes. The non-neurological autoimmune disease, psoriasis, has been related to passive smoking.²⁵ Most studies of other autoimmune diseases mainly investigated the effect of active smoking.^26,27

Our study has several strengths including a large study population, homogeneity of the study population since all had a similar background in terms of ethnicity, culture, and residential climate and finally, the relatively high response rates of 74% and 83% among cases and controls, respectively.

Furthermore, classification of study subjects into those who had never been active smokers and those who had been active smokers, but only above the age of 19, allowed us to elucidate the direct effect of passive smoking in the period of interest (age of 10–19), even in the presence of active smoking later in life.

A crucial weakness in our study is the use of blood donors as controls. Several studies have detected general lower mortality amid blood donors compared to the general population—the healthy donor effect.²⁸ This may be explained by the fact that blood donors are a selected group of healthier individuals.²⁹ Using blood donors as controls is generally not recommendable for studies looking at environmental factors, since they—as a control group—tend to differentiate from non-donors with regard to education and health-related behavior.^29–31 In line with this consideration, we stratified for active smoking and were able to make statistical adjustments for education, BMI, and alcohol and thus account for differences in central environmental factors between cases and controls. Another weakness in our study is the self-registration of passive smoking, which was measured as a qualitative parameter in the questionnaire, only quantified according to number of years of exposure. No detailed information on hours of exposure per day was registered, thus making the results suggesting a cumulative effect per year of exposure questionable. Retrospective questionnaires are dependent on memory and willingness to admit less optimal lifestyles, and there may be some degree of misclassification in individuals who declare themselves as never-smokers. MS patients are inclined to find an explanation for their disease in the past and may better remember exposures in retrospective questionnaire studies or may even over report the exposure. Recall bias may also differ between the sexes. Regarding response rate of blood donors, we have relatively high response rates of 75% and 90%, respectively. These rates were lower than the general 95% response rate in The Danish Blood Donor Study¹⁷ probably because the questionnaire was longer. Unfortunately, we did not record the response rate at all sites, but all the questionnaires were collected in the Capital Region of Copenhagen, which is a relatively small area, and varying degrees of non-responders must be acknowledged as a potential bias. There may be unknown factors which could confound the associations between the known covariates or even confound the results.

Nonetheless, exposure to passive smoking on a daily basis in the household during childhood should be feasible to recall in most cases, as well as the year of leaving home. To that end, a higher mean age when completing the questionnaire among controls may indicate a potential increased risk of recall bias. However, MS patients often suffer from cognitive impairment, including some degree of reduced memory functions, which may compensate in the other direction.³² Certain genetic variabilities and their interactions with passive smoking have been investigated in other studies. The strongest genetic association to MS is found in human leukocyte antigens (HLA) and include the major risk allele HLA-DRB1*15:01 and the protective allele HLA-A*02:01. Interactions between these genotypes and both smoking and passive smoking have been found in Swedish studies.^33,34

We found that MS patients more often than healthy controls reported exposure to passive smoking, which supports the findings of previous studies looking at adults. Elaborated knowledge about the pivotal effects of the environmental factors are crucial in the efforts for reducing the risk of MS, since environmental factors in many cases can be altered, and it may call attention to hitherto unclear pathogenic mechanisms.

Footnotes

Acknowledgements

We thank the Karolinska Institute in Stockholm for permission to translate and use the Genetic and Environment in Multiple Sclerosis (GEMS) questionnaire.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: P.S.S. has received personal compensation for serving on scientific advisory boards, steering committees, independent data monitoring committees, or have received speaker honoraria from Merck, Novartis, TEVA, GlaxoSmithKline, MedDay Pharmaceuticals, Sanofi-Aventis/Genzyme, and Celgene. F.S. has served on scientific advisory boards, been on the steering committees of clinical trials, served as a consultant, received support for congress participation, received speaker honoraria, or received research support for his laboratory from Biogen, Merck, Novartis, Roche, Sanofi Genzyme, and Teva. A.B.O. has served on scientific advisory boards for Biogen Idec, Novartis, and Sanofi Genzyme; has received research support from Novartis and Biogen Idec; has received speaker honoraria from Biogen Idec, Novartis, and TEVA; and has received support for congress participation from Merck, TEVA, Biogen, Roche, Novartis, and Sanofi Genzyme. M.M. has served on scientific advisory board for Biogen, Sanofi, Teva, Roche, Novartis, and Merck; has received honoraria for lecturing from Biogen, Merck, Novartis, Sanofi, and Genzyme; has received research support and support for congress participation from Biogen, Genzyme, Teva, Roche, Merck, and Novartis. N.K.-H. has received support for participation in congresses and symposia by Biogen, Merck, Novartis, Sanofi Genzyme, and Teva. H.U. received research support for his laboratory from Novartis. J.H.L. has received honoraria for lecturing from Teva, Almirall, and Merck Serono; funding for travel; and Biogen—honoraria for writing theme-booklet. S.G. has received support for congress participation from Merck. H.B.S., L.W.T., J.T.K., C.A., and D.B.O. declared that there is no conflict of interest.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by grants from the Danish Multiple Sclerosis Society (A-19376), the Danish Council for Strategic Research (Grant No. 2142-08-0039), Novartis, Biogen (Denmark), the Sofus Carl Emil Friis og Hustru Olga Doris Friis foundation, the Foundation for Research in Neurology, and the Director Einar Jonasson (Johnsen) and Wife foundation.

References

Olsson

Barcellos

Alfredsson

Interactions between genetic, lifestyle and environmental risk factors for multiple sclerosis. Nat Rev Neurol 2016; 13: 26–36.

Poorolajal

Bahrami

Karami

, et al. Effect of smoking on multiple sclerosis: A meta-analysis. J Public Health 2017; 39(2): 312–320.

Zhang

Wang

, et al. The risk of smoking on multiple sclerosis: A meta-analysis based on 20,626 cases from case-control and cohort studies. PeerJ 2016; 4: e1797.

Hedström

Bäärnhielm

Olsson

, et al. Tobacco smoking, but not Swedish snuff use, increases the risk of multiple sclerosis. Neurology 2009; 73: 696–701.

Hedström

Hillert

Olsson

, et al. Nicotine might have a protective effect in the etiology of multiple sclerosis. Mult Scler 2013; 19(8): 1009–1013.

Hedström

Baarnhielm

Olsson

, et al. Exposure to environmental tobacco smoke is associated with increased risk for multiple sclerosis. Mult Scler 2011; 17(7): 788–793.

Hedström

Olsson

Alfredsson

High body mass index before age 20 is associated with increased risk for multiple sclerosis in both men and women. Mult Scler 2012; 18(9): 1334–1336.

Hedström

Akerstedt

Hillert

, et al. Shift work at young age is associated with increased risk for multiple sclerosis. Ann Neurol 2011; 70: 733–741.

Gustavsen

Søndergaard

Oturai

, et al. Shift work at young age is associated with increased risk of multiple sclerosis in a Danish population. Mult Scler Relat Disord 2016; 9: 104–109.

10.

Cabre

Migration and multiple sclerosis: The French West Indies experience. J Neurol Sci 2007; 262(1–2): 117–121.

11.

Gale

Martyn

CN.

Migrant studies in multiple sclerosis. Prog Neurobiol 1995; 47: 425–448.

12.

Kurtzke

JF.

Multiple sclerosis in time and space—geographic clues to cause. J Neurovirol 2000; 6(Suppl. 2): S134–S140.

13.

Munk Nielsen

Corn

Frisch

, et al. Multiple sclerosis among first- and second-generation immigrants in Denmark: A population-based cohort study. Brain 2019; 142(6): 1587–1597.

14.

Magyari

Koch-Henriksen

Sørensen

PS.

The Danish multiple sclerosis treatment register. Clin Epidemiol 2016; 8: 549–552.

15.

Polman

Reingold

Banwell

, et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol 2011; 69(2): 292–302.

16.

Polman

Wolinsky

Reingold

SC.

Multiple sclerosis diagnostic criteria: Three years later. Mult Scler 2005; 11(1): 5–12.

17.

Pedersen

Erikstrup

Kotze

, et al. The Danish Blood Donor Study: A large, prospective cohort and biobank for medical research. Vox Sang 2012; 102(3): 271.

18.

Andersen

Søndergaard

Bang Oturai

, et al. Alcohol consumption in adolescence is associated with a lower risk of multiple sclerosis in a Danish cohort. Mult Scler J 2019; 25: 1572–1579.

19.

The Danish Health Authorities. https://publications.parliament.uk/pa/cm201012/cmselect/cmsctech/1536/153605.htm (accessed 5 March 2020).

20.

Månsson

Joffe

Sun

, et al. On the estimation and use of propensity scores in case-control and case-cohort studies. Am J Epidemiol 2007; 166(3): 332–339.

21.

Cole

Platt

Schisterman

, et al. Illustrating bias due to conditioning on a collider. Int J Epidemiol 2010; 39(2): 417–420.

22.

Ramagopalan

Lee

Yee

, et al. Association of smoking with risk of multiple sclerosis: A population-based study. J Neurol 2013; 260(7): 1778–1781.

23.

Mikaeloff

Caridade

Tardieu

, et al. Parental smoking at home and the risk of childhood-onset multiple sclerosis in children. Brain 2007; 130(Pt 10): 2589–2595.

24.

Abdollahpour

Nedjat

Sahraian

, et al. Waterpipe smoking associated with multiple sclerosis: A population-based incident case–control study. Mult Scler 2017; 23(10): 1328–1335.

25.

Lønnberg

Skov

Skytthe

, et al. Smoking and risk for psoriasis: A population-based twin study. Int J Dermatol 2016; 55(2): e72–e78.

26.

Sugiyama

Nishimura

Tamaki

, et al. Impact of smoking as a risk factor for developing rheumatoid arthritis: A meta-analysis of observational studies. Ann Rheum Dis 2010; 69(1): 70–81.

27.

Harel-Meir

Sherer

Shoenfeld

Tobacco smoking and autoimmune rheumatic diseases. Nat Clin Pract Rheumatol 2007; 3(12): 707–715.

28.

Atsma

Veldhuizen

Verbeek

, et al. Healthy donor effect: Its magnitude in health research among blood donors. Transfusion 2011; 51(8): 1820–1828.

29.

Golding

Northstone

Miller

, et al. Differences between blood donors and a population sample: Implications for case-control studies. Int J Epidemiol 2013; 42(4): 1145–1156.

30.

Burnett

JJ.

Examining the profiles of the donor and nondonor through a multiple discriminant approach. Transfusion 1982; 22(2): 138–142.

31.

Volken

Weidmann

Bart

, et al. Individual characteristics associated with blood donation: A cross-national comparison of the German and Swiss population between 1994 and 2010. Transfus Med Hemother 2013; 40(2): 133–138.

32.

Amato

Zipoli

Portaccio

Cognitive changes in multiple sclerosis. Expert Review of Neurotherapeutics 2008; 8:1585–1596.

33.

Hedström

Bomfim

Barcellos

, et al. Interaction between passive smoking and two HLA genes with regard to multiple sclerosis risk. Int J Epidemiol 2014; 43(6): 1791–1798.

34.

Hedström

Sundqvist

Bäärnhielm

, et al. Smoking and two human leukocyte antigen genes interact to increase the risk for multiple sclerosis. Brain 2011; 134(Pt 3):653–664.