Multiple Sclerosis Resource Centre

Welcome to the Multiple Sclerosis Resource Centre. This website is intended for international healthcare professionals with an interest in Multiple Sclerosis. By clicking the link below you are declaring and confirming that you are a healthcare professional

You are here

Physical and social environment and the risk of multiple sclerosis

Multiple Sclerosis and Related Disorders, 5, 3, pages 600 - 606



The incidence of multiple sclerosis (MS) in Denmark has doubled in women since 1970, whereas it has been almost unchanged in men. The rapid epidemiological changes suggest that environmental factors may modify the risk of MS.


To investigate whether occupational, physical, or social environmental influence the risk of MS differently in women than in men.


The cohort consists of all 1403 patients (939 women, 464 men) identified through Danish Multiple Sclerosis Registry aged 1–55 of years at clinical onset between 2000 and 2004, and up to 25 control persons for each case, matched by sex, year of birth and residential municipality. The same cohort was previously used to investigate the influence of the reproductive factors on the risk of MS.


By linkage to Danish population registers we found a slight albeit statistically significant excess for 6 female MS patients who had been employed in agriculture: OR 3.52; 95% CI 1.38–9.00,p=0.008 (0.046 when corrected for multiple significance) and a trend for exposure to outdoor work in 12 : OR 1.94, 95% CI 1.06–3.55,p=0.03 (0.09 when corrected for multiple significance), but the numbers of cases were small, and the effects were not found in men. Educational level, housing conditions in youth, or the presence of children unrelated by blood in the household did not influence the risk of MS.


Our study did not reveal any additional factors beyond the previously published childbirths which could explain the extent of the MS incidence increase in women.



  • The distribution of the educational level was the same in cases and controls before clinical onset for both sexes.
  • The “hygiene hypothesis” could not be confirmed by our study.
  • Working in agriculture may show association with a higher MS risk in women.
  • MS was not associated with a greater exposure to infective agents due to cohabitation with children.

Keywords: Multiple sclerosis, Gender, Risk factors, Education, Occupation, Hygiene theory.

1. Introduction

It is generally accepted that the risk of multiple sclerosis (MS) may be modified by environmental factors ( Koch et al., 2013 ) in genetically susceptible individuals ( Ascherio and Munger, 2007 ). The incidence of MS affecting women is rising in many countries ( Koch-Henriksen and Sorensen, 2010 ) and this increase may be related to the changing western life style over the last half-century ( Orton et al., 2010 ). Over the past century the lives of women underwent major changes regarding lifestyle, education, work life, and later and fewer childbirths. Women today are employed in different industries, earlier prerogatives of men. The increased MS incidence in women may reflect influence of new environmental factors only affecting women or affecting both sexes but with women being more susceptible.

In a previous study using the same study subjects, established to determine the effect of the various factors on the risk of MS, we found that women with MS had fewer pregnancies up to five years before clinical onset ( Magyari et al., 2013 ). This suggests the role of childbirth/pregnancy as a lifestyle factor linked to a reduced risk on MS, also reported in an Australian study ( Ponsonby et al., 2012 ), and the decreasing birth rate of Danish women may have its share in the increasing incidence of MS in women, but it is not the full explanation.

One of the interesting ideas is the hygiene hypothesis proposed in 1966 by Leibowitz et al., suggesting that the risk of MS may be higher in individuals with a high level of sanitation during their childhood ( Leibowitz et al., 1966 ). Danish families have experienced improvements in household amenities, decreasing family size, and higher standards of personal cleanliness, hygiene and sanitation, all of which have reduced the risk of infections including the likelihood of cross-infections among family members. According to the hygiene hypothesis, infections in childhood could to some degree protect against autoimmune diseases ( Fleming and Cook, 2006 ), and the reduced incidence of infectious diseases may have resulted in an increased incidence of allergic diseases ( Bach, 2002 ). Results from different studies regarding the effects of the exposures to younger siblings and thereby to infections, and of birth order from different studies are conflicting. Some studies have reported a reduced risk of MS after exposure to younger siblings during childhood, (Ponsonby et al, 2005, Levin et al, 2005, and Hughes et al, 2013) whereas others found no correlation between birth order and the risk of developing MS (Sadovnick et al, 2005 and Koch-Henriksen, 1989a).

Increased urbanisation has resulted in major changes to women׳s lifestyle and this has proven to be a significant co-factor for the increase of MS in women in Crete ( Kotzamani et al., 2012 ). However, growing up in an urban centre was shown to be associated with a lower risk for developing MS in a recent case-control study conducted in Berlin ( Conradi et al., 2011 ).

There are differing evidences regarding the role of occupational exposures in the risk of MS. An association between the risk of MS and exposure to organic solvents has been found in some studies, (Amaducci et al, 1982 and Landtblom et al, 1996) but no association was found in a population based prospective study ( Stenager et al., 2003 ).

In this study we aimed to determine whether environmental factors including occupational physical/chemical exposures, social status expressed as the level of education, or housing conditions in youth influence differently the risk of developing MS in women and men using a case-control approach. We also aimed to distinguish the physical environmental factors of having children in the household from the known effect of childbirths ( Magyari et al. 2013 ).

2. Material and methods

We established a study database for the purpose of investigating different factors available from population based registries that may contribute to the increasing incidence of MS in women. All patients were identified through the nationwide Danish MS registry that has a high validity (94%) and completeness (> 90%) ( Bentzen et al., 2010 ).

For the case-control study we included all patients (1403) with clinical onset in the five-year period 2000–2004 at age 15–55 years and who fulfilled the McDonald 2001 criteria of MS ( McDonald et al., 2001 ). The cohort was limited to this age at onset interval, because there were very few patients outside. The time interval 2000–2004 was chosen to ensure sufficient retrospective data, as most of the populations registers to be linked with the Danish MS Registry was established in the late nineteen seventies and in the beginning of eighties. For each case, 25 control persons were drawn at random from the Danish Civil Registration System ( Pedersen, 2011 ), matched by sex, year of birth and residential municipality by January 1st in the year of onset of the first demyelinating symptom of the matched MS case. Subjects not born in Denmark were excluded. The Civil Registration System has to each Danish citizen assigned a unique personal identification number that permits linkage at the individual level between the nationwide registries and other data sources. Social data for all cases and their individually matched controls and their families were collected from the various registers, mentioned below by Statistics Denmark after encryption of the person identifications after linkage. The same cohorts of subjects had been used for two other studies (Magyari et al, 2013 and Magyari et al, 2014); but with inclusion of different data.

Information on educational level, occupation, family status, and housing condition was retrieved for each year as far back in time as possible.

The Education Registry ( Jensen and Rasmussen, 2011 ) provided data about the highest obtained educational level at each year. In the Education Registry, information about a person׳s highest achieved educational level was specified in more than 1300 different categories. This was simplified to four categories defined as basic school; secondary school, short-term on-the-job training or postsecondary school; bachelor׳s degree or moderate-term on-the-job training; and long-term education.

By linkage to The Danish Industrial Classification Database maintained by Statistics Denmark, information about the index-persons׳ occupation in different registered industries was obtained.

Based on the original 303-item classification of occupations, five main categories were defined for the analysis: agriculture; craftsman work; health sector; workers in chemical industries; and work without specific physical exposure. An index person was considered as being unexposed if he/she worked in retail, finance sector, insurance, education, social services, etc, and this group was chosen as reference category. The period in which the occupational exposure was determined ranged from 1982 to 1995, because in 1995 the database was reorganised and new categories were defined, making grouping of different exposures into the same categories uncertain. In this period the frequency of missing values amounted to 6.4% on average. Persons without employment in this period were excluded. Applying this criteria 7422 index persons, were recorded to have an occupation in the Danish Industrial database between 1982 and 1995. To be classified as exposed, the subject had to be employed in the specific occupational category for at least three years. An individual who has been exposed to more than one occupational category will contribute to more of the OR estimates, and the sum of exposures add up to a higher number than the number of index persons. The Building and Housing Registry provided information about dwelling size, sanitary installations, size and type of household. This database contains systematic registration of housing conditions for all Danish citizens dating back to 1977.

We defined a coefficient as living space/person during the subjects 10–15 years age interval and calculated it by summing up the areas of the housing spaces in square metres over the respective five years, and divided by the sum of persons living in the households over these years. The logarithmic transformed values were well fitted to a normal distribution for both men and women. An independent samplet-test was conducted to compare the average living space/person for cases and controls.

The Household and Family Statistics provided information about the number of persons and their status in the family. In the Household- and Family database, households are defined as address households (i.e. each address at which persons are registered). Children are defined as persons younger than 18 years. Data was incomplete for 9.9% of cases and 10.2% of controls. An index person was categorised as in a relationship if he/she was married or was living in a relationship for more than eight years in the observation period. In Denmark more than one-third of marriages or relationships end in divorce or broken partnership and if the couple has children, they often share parental custody or the one of the divorced parents get the custody for the child. For this reason even childless persons may through a relationship be physically exposed to children in the household before the reference year, enabling us distinguish between the biological effect of giving birth to a child and physical exposure to a child (infections etc.).

We investigated whether childless female MS cases more often than childless female controls had been exposed to children (at age lower than 18 years) for at least three years before the reference year.

2.1. Statistical analysis

Conditional logistic regression was used to calculate the odds ratios (OR) with 95% confidence intervals using the SPSS Coxreg procedure, with inclusion of a number of possible confounders as age of onset, and educational level in some of the analysis. The mean living-area per person was compared using Student׳st-test. Statistical significance was defined atp<0.05. For groups of variables defined a priori, correction for the false discovery rate was employed using the Bonferroni method ( Bland and Altman, 1995 ).

The analysis was performed with the statistical package SPSS version 19.

3. Results

The study population consisted of all 1403MS cases, 939 women and 464 men from the Danish Multiple Sclerosis Registry with clinical onset from 2000–2004 and 35045 matched controls. The female: male ratio was 2.02:1.

The mean age at clinical onset of MS was 35.3 years, (36.1 years for men and 34.9 years for women). The distribution regarding disease onset was relapsing-remitting in 83.7% (1242) and primary progressive in 16.3% (2 4 1) of the patients.

3.1. Educational level

One hundred and twenty-six (20%) cases and 2945 (19.3) controls for whom information on educational level were missing in the Education Registry, were omitted from the analyses. The distribution of the educational level, divided into four main categories, was the same in cases and controls before the reference year for both sexes ( Table 1 ). Thus the educational level did not have an effect on the risk of MS in neither women nor men. Including onset age as a covariate did not change the results. The ORs are presented in Table 1 with basic school as reference category.

Table 1 Odds ratios for developing multiple sclerosis for different educational levels before clinical onset with basic school as the reference category.

Education Gender N (%) cases/ controls Odds ratio (95% CI) p -Value
Basic school Women 406(43.2)/ 9783(41.7) 1.00 (reference)  
  Men 242(52.2)/ 5809(50,1) 1.00 (reference)  
Secondary-postsecondary school/ short-term on-the-job training Women 272(29,0)/ 6856(29,2) 0.86 (0.64–1.16) 0.34
Men 93(20)/ 2445(21,1) 1.09 (0.77–1.53) 0.64
Bachelor׳s degree/ moderate-term on-the-job training Women 147(15.7)/ 3862(16.5) 0.82 (0.60–1.12) 0.21
Men 28(6.0)/ 904(7.8) 0.99 (0.68–1.43) 0.93
Long-term education Women 49(5.2)/ 1544(6.6) 0.79 (0.58–1.09) 0.15
Men 40(8.6)/ 897(7.7) 0.80 (0.50–1.23) 0.36

3.2. Occupational exposures

The numbers of cases and controls exposed to any of the five main categories of occupational exposure are shown in Table 2 .

Table 2 Case/control-odds ratios for each of the five main occupational categories. The same person may appear in more than one category.

    Cases (%) Controls (%) OR 95% CI p -Value
Agriculture Men 11(9.5) 176(6.2) 1.42 (0.71–2.83) 0.32
  Women 6(3.3) 50(1.2) 3.52 (1.38–9.00) 0.008
Craftsman work Men 29(25) 894(31.3) 0.88 (0.56–1.38) 0.58
  Women 8(4.3) 149(3.5) 1.08 (0.51–2.28) 0.85
Health sector Men 5(4.3) 48(1.7) 2.30 (0.88–6.02) 0.87
  Women 30(16.3) 777(18.2) 0.84 (0.55–1.26) 0.40
Chemical industries Men 3(2.6) 54(1.9) 0.80 (0.31–2.04) 0.73
  Women 5(2.7) 138(3.2) 1.23 (0.37–4.13) 0.64
No physical exposure Men 68(58.6) 1682(58.9) Reference    
  Women 135(73.4) 135(73.4) Reference    

Case/control odds ratios, obtained with conditional logistic regression is shown in Table 2 for both sexes. The only deviation from unity in any of the five occupational categories was a slight albeit statistically significant excess of female MS patients who had been employed in agriculture (OR 3.52; 95% CI 1.38–9.00,p=0.008) but this was based on only 6 cases ( Table 2 ). The result remained significant after correction with the Bonferroni methodp=0.04. The frequency of craftsman work, employment in health sector or in chemical industry did not differ between female cases and controls. The distribution of any of the five occupational categories prior to reference year was the same in male cases and controls independent of the classification into the five occupational categories.

We also divided the 303 categories the Danish Industrial Classification Database into three independent dichotomous variables of physical exposure so all index persons could be classified as exposed or non-exposed to each of the following three factors: chemical substances; organic solvents; and outdoor work (e.g. other than workers in chemical industry may be exposed to chemicals). Again, the same individual could appear in more than one of the categories. Analysis of these three variables showed that only outdoor work was correlated with an increased risk of MS in women (OR 1.94, 95% CI 1.06–3.55,p=0.048), but based on just 12 female MS-cases, however after correction for multiple significance, the result was not significantp=0.09 ( Table 3 ). There was an overlap between these and the six female MS patients employed in agriculture. There was no difference between male MS cases and controls for any of the three variables. We found no evidence of correlation between solvent or chemical exposures and the risk of MS in neither women nor men. The ORs are presented in Table 3 .

Table 3 The odds ratios for developing MS among those exposed to organic solvents, chemicals and outdoor work.

    Cases (%) Controls (%) OR 95% CI p
Organic solvents Men 28(5.7) 683(5.6) 1.03 (0.67–1.58) 0.89
  Women 5(0.5) 226(0.9) 0.54 (0.22–1.33) 0.18
Chemical exposure Men 33(6.7) 888(7.2) 0.91 (0.61–1.35) 0.62
  Women 9(0.9) 330(1.4) 0.67 (0.34–1.31) 0.24
Outdoor work Men 17(3.5) 493(4.0) 0.84 (0.50–1.41) 0.52
  Women 12(1.2) 159(0.7) 1.94 (1.06–3.55) 0.03

In the analysis of the dichotomous explanatory variables, the reference was ‘unexposed’.

3.3. Housing conditions

The housing conditions in Denmark have generally been of high standards over the last fifty years, and very few index persons were living in a household with deficient sanitary conditions. Only 0.4% female cases vs. 0.4% female controls and 0.4% of male cases vs. 0.3% of male controls lived under insufficient sanitary conditions between age 10 to 15 (no difference between cases and controls (p=0.48 for men andp=0.97 for women)).

The mean space per person in the household over these years was 39.42 m2 for male cases and 40.47 m2 for male controls and 37.79 m2 for female cases and 40.15 m2 for female controls (t values of logarithmic transformed data were; for men:t=0.031, df=3813,p=0.97; for women:t=−1.44, df=9328,p=0.15). Thus there was no association between the living space in the youth and the risk of MS.

3.4. Cohabiting with children after the age of 25 years

In women after the age of 25 years and up to the age at clinical onset of MS, 86.1% (N=310) of nulliparous female cases and 86.9% (N=7193) of nulliparous female controls had been exposed to children for more than 3 years in their household during any period (p=0.66). Regarding males 82.0% (N=159) of male cases and 86.0% (N=4044) of male controls were sharing their home with children (p=0.12). The lower age limit of 25 years was chosen, because giving birth before age 25 is uncommon, whereas there is no such lower age limit to exposure to partner׳s children.

There was no difference between the frequency of cases and controls older than 25 years living in a relationship. Including educational level and onset age as covariates in the conditional logistic regression analysis did not change the effect of living with living with only non-consanguineous children in the household in nulli-para women (OR 0.70, 95% CI 0.27–1.80,p=0.45), or men (OR 1.49, 95% CI 0.45–4.90,p=0.52). As it was previously shown ( Magyari et al., 2013 ) that only numbers of childbirths within the five year period before clinical onset distinguished female MS-cases from controls, the analysis was repeated for this period only. In the five years preceding the occurrence of first symptom 70.9% (N=558) of childless cases and 71.9% of (N=13064) childless controls have shared their household with children, so neither in this period were there any statistical significant differences between cases and controls; OR 1.03 (95% CI 0.73–1.45,p=0.88). A similar analysis was conducted for men, and again the presence of children in the household did not influence the risk of MS, the OR being 0.70 (95% CI 0.47–1.05,p=0.08).

4. Discussion

In a previous study performed using the same database, we already found one important lifestyle factor that may partly explain the increased temporal MS incidence particularly among women: less childbirth up to five years before MS presentation. In the present study we did not find any additional environmental factors that could contribute to the increasing incidence of MS. The risk of MS was not associated with the educational level or occupation in men or women, except for a higher frequency of outdoor work or employment in agriculture, but as this only involved 6 or 12 MS cases, respectively, the aetiologic fraction is negligible although the association was statistical significant.

A number of older studies have found higher risk of MS with higher education (Beebe et al, 1967, Miller et al, 1960, Russell, 1971, and Visscher et al, 1981). Conversely, other more recent studies have reported that higher education was associated with reduced risk of MS (Ghadirian et al, 2001 and Riise et al, 2011). However, other studies failed to find any association between social level and the risk of MS (Koch-Henriksen, 1989b and Lauer and Firnhaber, 1985).

Our findings with a lack of correlation between physical household conditions and the risk of MS do not support the hygiene hypothesis of protection against MS by exposure to infections in early life.

The importance of what happens in the first two decades for risk of MS is supported by migration studies ( Gale and Martyn, 1995 ), and by the associations with early menarche and childhood obesity and the risk of MS (Munger et al, 2009, Munger et al, 2013, and Ramagopalan et al, 2009) Previous Danish studies reported that common childhood infections are not associated with increased risk of multiple sclerosis later in life ( Bager et al., 2004 ), and birth order is not associated with the risk of MS ( Bager et al., 2006 ). When investigating the correlation between cohabitation with children and the risk of MS, we suppose that a person living with children have a greater exposure to infective agents, but there were no associations with MS. When we defined cohabitation with children, we did not distinguish between a partner׳s children or younger siblings, which can explain the high percentage of persons sharing home with children for more than 3 years, however cohabiting with parents and siblings over the age of 25 years is unusual in Denmark.

One of the limitations of the present study was the need to reclassify the industry codes when investigating the occupational exposure. As there were 303 occupational categories, misclassification, when reducing the categories into the five groups, cannot be excluded, and a certain exposure may be difficult to ascertain; for example, a person working in the administrative department of a chemical industry, is probably not physically exposed to chemicals, but will nevertheless be classified as so. Due to redefinitions of categories in the population database in 1995, only occupational exposure before 1995 was taken in consideration. This means that a part of the cohort had not yet reached their professional age, whereas for the others it was possible to investigate a possible occupational exposure in the young ages. Due to the limited size of the cohort further studies are needed with a more accurate collection of occupational data to reinforce the evidence of the possible role of agricultural working exposure. The excess of MS risk in female agricultural workers could indicate a role of pesticides, also found in a study showing that the prevalence of multiple sclerosis was significantly higher in districts with greater pesticide use as compared to those with lower pesticide use ( Parron et al., 2011 ). This finding is somewhat contradicted by the protective effect of solar UV radiation, shown in a number of studies, as outdoor or agricultural workers are exposed to more sunlight. However, the costal and high latitude climate in Denmark with frequent overcast skies attenuates the sunlight exposure compared with many other areas. A recent Australian study showed an increased risk for developing the first demyelinating event in women exposed to livestock or farming ( Valery et al., 2013 ), and in a Danish study working within the agricultural segment, especially in dairies was associated with an increased MS risk ( Horwitz et al., 2013 ). With its low aetiologic fraction that approximates zero, it has, however, no share in the high and increasing incidence of MS in women.

Access to medical care and social support in Denmark is not influenced by the person׳s socioeconomic status or education, although there could be differences in health knowledge and behaviour and differences in psychosocial stress in different socio-economic strata. Our study could not identify these factors with the accessible variables we were investigating. Denmark is a country with high social equality, and this may eventually blur the effects of lifestyle factors.

The present study has some strength. The study population was large and well characterised. Using population based registries as sources of data ensured that information about socio-economic factors was collected independently of the person׳s disease status, responsiveness, and memory, thereby limiting the possibility of bias.

The increasing incidence of MS in women represents the major part of the total incidence increase in Denmark. Our study did not reveal any physical, chemical or socioeconomic factors which could explain the gender discrepancy in the incidence increase over the last 40 years.

The aetio-pathogenesis of MS is not yet fully known, and the complexity of the possible environmental factors that may contribute to the disease risk makes identification of these factors difficult. Socio-economic indicators are sometimes strongly linked to lifestyle and can vary with country and geographical area, which can further complicate epidemiological studies.

Identification of risk factors will hopefully lead to possibilities of a better prevention of this chronic disease.

Conflict of interest

Melinda Magyari has served on scientific advisory board for Biogen Idec; TEVA, has received honoraria for lecturing from Biogen Idec, Merck Serono, Sanofi-Aventis, Teva; has received support for congress participation from Biogen Idec, Merck Serono, Novartis, Genzyme.

Nils Koch-Henriksen has received honoraria for lecturing and participation in advisory councils, travel expenses for attending congresses and meetings, and financial support for monitoring the Danish MS Treatment Register from Bayer-Schering, Merck-Serono, BiogenIdec, TEVA, Sanofi-Avensis and Novartis. The Danish Multiple Sclerosis Registry is funded by the Danish Multiple Sclerosis Society.

Claudia Christina Pfleger has served on scientific advisory board for Novartis, Amirall, Biogen Idec; has received support for congress participation from Biogen Idec, Teva and Novartis.

Per Soelberg Sorensen has served on scientific advisory boards Biogen Idec, Merck Serono, Novartis, Genmab, TEVA, Elan, GSK; has been on steering committees or independent data monitoring boards in clinical trials sponsored by Merck Serono, Genmab, TEVA, GSK, Bayer Schering, and he has received funding of travel for these activities; has received speaker honoraria from Biogen Idec, Merck Serono, TEVA, Bayer Schering, Sanofi-aventis, Genzyme, and Novartis.

The study was founded by the Danish Multiple Sclerosis Society.


Economic support was obtained from the Danish Multiple Sclerosis Society (F-22737-01).


  • Amaducci et al., 1982 L Amaducci, C Arfaioli, D Inzitari, M. Marchi. Multiple sclerosis among shoe and leather workers: an epidemiological survey in Florence. Acta Neurol Scand. 1982;65:94-103
  • Ascherio and Munger, 2007 A Ascherio, KL. Munger. Environmental risk factors for multiple sclerosis. Part II: noninfectious factors. Ann Neurol. 2007;61:504-513 Crossref
  • Bach, 2002 JF. Bach. The effect of infections on susceptibility to autoimmune and allergic diseases. N Engl J Med. 2002;347:911-920 Crossref
  • Bager et al., 2004 P Bager, NM Nielsen, K Bihrmann, et al. Childhood infections and risk of multiple sclerosis. Brain. 2004;127:2491-2497 Crossref
  • Bager et al., 2006 P Bager, NM Nielsen, K Bihrmann, et al. Sibship characteristics and risk of multiple sclerosis: a nationwide cohort study in Denmark. Am J Epidemiol. 2006;163:1112-1117 Crossref
  • Beebe et al., 1967 GW Beebe, JF Kurtzke, LT Kurland, TL Auth, B. Nagler. Studies on the natural history of multiple sclerosis. 3. Epidemiologic analysis of the army experience in World War II. Neurology. 1967;17:1-17
  • Bentzen et al., 2010 J Bentzen, EM Flachs, E Stenager, H Bronnum-Hansen, N. Koch-Henriksen. Prevalence of multiple sclerosis in Denmark 1950-2005. Mult Scler. 2010;16:520-525 Crossref
  • Bland and Altman, 1995 JM Bland, DG. Altman. Multiple significance tests: the Bonferroni method. BMJ. 1995;310:170 Crossref
  • Conradi et al., 2011 S Conradi, U Malzahn, F Schroter, et al. Environmental factors in early childhood are associated with multiple sclerosis: a case-control study. BMC Neurol. 2011;11:123 Crossref
  • Fleming and Cook, 2006 JO Fleming, TD. Cook. Multiple sclerosis and the hygiene hypothesis. Neurology. 2006;67:2085-2086 Crossref
  • Gale and Martyn, 1995 CR Gale, CN. Martyn. Migrant studies in multiple sclerosis. Prog Neurobiol. 1995;47:425-448
  • Ghadirian et al., 2001 P Ghadirian, B Dadgostar, R Azani, P. Maisonneuve. A case-control study of the association between socio-demographic, lifestyle and medical history factors and multiple sclerosis. Can J Public Health. 2001;92:281-285
  • Horwitz et al., 2013 H Horwitz, B Ahlgren, E. Naerum. Effect of occupation on risk of developing MS: an insurance cohort study. BMJ Open. 2013;3:e002894 Crossref
  • Hughes et al., 2013 AM Hughes, RM Lucas, AJ McMichael, et al. Early-life hygiene-related factors affect risk of central nervous system demyelination and asthma differentially. Clin Exp Immunol. 2013;172:466-474 Crossref
  • Jensen and Rasmussen, 2011 VM Jensen, AW. Rasmussen. Danish Education Registers. Scand J Public Health. 2011;39:91-94
  • Koch et al., 2013 MW Koch, LM Metz, SM Agrawal, VW. Yong. Environmental factors and their regulation of immunity in multiple sclerosis. J Neurol Sci. 2013;324:10-16 Crossref
  • Koch-Henriksen, 1989a N. Koch-Henriksen. An epidemiological study of multiple sclerosis. Familial aggregation social determinants, and exogenic factors. Acta Neurol Scand Suppl. 1989;124:1-123
  • Koch-Henriksen, 1989b N. Koch-Henriksen. An epidemiological study of multiple sclerosis. Familial aggregation social determinants, and exogenic factors. Acta Neurol Scand Suppl. 1989;124:1-123
  • Koch-Henriksen and Sorensen, 2010 N Koch-Henriksen, PS. Sorensen. The changing demographic pattern of multiple sclerosis epidemiology. Lancet Neurol. 2010;9:520-532 Crossref
  • Kotzamani et al., 2012 D Kotzamani, T Panou, V Mastorodemos, et al. Rising incidence of multiple sclerosis in females associated with urbanization. Neurology. 2012;78:1728-1735 Crossref
  • Landtblom et al., 1996 AM Landtblom, U Flodin, B Soderfeldt, C Wolfson, O. Axelson. Organic solvents and multiple sclerosis: a synthesis of the current evidence. Epidemiology. 1996;7:429-433 Crossref
  • Lauer and Firnhaber, 1985 K Lauer, W. Firnhaber. Epidemiological investigations into multiple sclerosis in Southern Hesse. III. The possible influence of occupation on the risk of disease. Acta Neurol Scand. 1985;72:397-402
  • Leibowitz et al., 1966 U Leibowitz, A Antonovsky, JM Medalie, HA Smith, L Halpern, M. Alter. Epidemiological study of multiple sclerosis in Israel. II. Multiple sclerosis and level of sanitation. J Neurol Neurosurg Psychiatry. 1966;29:60-68 Crossref
  • Levin et al., 2005 LI Levin, KL Munger, MV Rubertone, et al. Temporal relationship between elevation of epstein-barr virus antibody titers and initial onset of neurological symptoms in multiple sclerosis. JAMA. 2005;293:2496-2500 Crossref
  • Magyari et al., 2013 M Magyari, N Koch-Henriksen, CC Pfleger, PS. Sorensen. Reproduction and the risk of multiple sclerosis. Mult Scler. 2013;12:1604-1609 Crossref
  • Magyari et al., 2014 M Magyari, N Koch-Henriksen, CC Pfleger, PS. Sorensen. Gender and autoimmune comorbidity in multiple sclerosis. Mult Scler. 2014;20(9):1244-1251
  • McDonald et al., 2001 WI McDonald, A Compston, G Edan, et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol. 2001;50:121-127 Crossref
  • Miller et al., 1960 H Miller, A Ridley, K. Schapira. Multiple sclerosis. A note on social incidence. Br Med J. 1960;2:343-345 Crossref
  • Munger et al., 2009 KL Munger, T Chitnis, A. Ascherio. Body size and risk of MS in two cohorts of US women. Neurology. 2009;73:1543-1550 Crossref
  • Munger et al., 2013 KL Munger, J Bentzen, B Laursen, et al. Childhood body mass index and multiple sclerosis risk: a long-term cohort study. Mult Scler. 2013;10:1323-1329 Crossref
  • Orton et al., 2010 SM Orton, SV Ramagopalan, D Brocklebank, et al. Effect of immigration on multiple sclerosis sex ratio in Canada: the Canadian collaborative study. J Neurol Neurosurg Psychiatry. 2010;81:31-36 Crossref
  • Parron et al., 2011 T Parron, M Requena, AF Hernandez, R. Alarcon. Association between environmental exposure to pesticides and neurodegenerative diseases. Toxicol Appl Pharmacol. 2011;256:379-385 Crossref
  • Pedersen, 2011 CB. Pedersen. The Danish civil registration system. Scand J Public Health. 2011;39:22-25
  • Ponsonby et al., 2012 AL Ponsonby, RM Lucas, IA van der Mei, et al. Offspring number, pregnancy, and risk of a first clinical demyelinating event: the AusImmune study. Neurology. 2012;78:867-874 Crossref
  • Ponsonby et al., 2005 AL Ponsonby, I van dM, T Dwyer, L Blizzard, B Taylor, A. Kemp. Birth order, infection in early life, and multiple sclerosis. Lancet Neurol. 2005;4:793-794 Crossref
  • Ramagopalan et al., 2009 SV Ramagopalan, W Valdar, M Criscuoli, et al. Age of puberty and the risk of multiple sclerosis: a population based study. Eur J Neurol. 2009;16:342-347
  • Riise et al., 2011 T Riise, J Kirkeleit, JH Aarseth, et al. Risk of MS is not associated with exposure to crude oil, but increases with low level of education. Mult Scler. 2011;17:780-787 Crossref
  • Russell, 1971 WR. Russell. Multiple sclerosis: occupation and social group at onset. Lancet. 1971;2:832-834 Crossref
  • Sadovnick et al., 2005 AD Sadovnick, IM Yee, GC. Ebers. Multiple sclerosis and birth order: a longitudinal cohort study. Lancet Neurol. 2005;4:611-617 Crossref
  • Stenager et al., 2003 E Stenager, H Bronnum-Hansen, N. Koch-Henriksen. Risk of multiple sclerosis in nurse anaesthetists. Mult Scler. 2003;9:427-428 Crossref
  • Valery et al., 2013 PC Valery, RM Lucas, DB Williams, et al. Occupational exposure and risk of central nervous system demyelination. Am J Epidemiol. 2013; (Apr 12)
  • Visscher et al., 1981 BR Visscher, VA Clark, R Detels, RM Malmgren, NL Valdiviezo, JP. Dudley. Two populations with multiple sclerosis. Clinical and demographic characteristics. J Neurol. 1981;225:237-249 Crossref


a Danish Multiple Sclerosis Center, Department of Neurology, and the Danish Multiple Sclerosis Registry, Rigshospitalet, and University of Copenhagen, Copenhagen, Denmark

b Clinical Institute, Department of Clinical Epidemiology, University of Aarhus, Denmark

c Department of Neurology, Aalborg University Hospital, Aalborg, Denmark

d Danish Multiple Sclerosis Center, Department of Neurology, Rigshospitalet and University of Copenhagen, Copenhagen, Denmark

e Danish Multiple Sclerosis Registry, Rigshospitalet, and University of Copenhagen, Denmark

lowast Corresponding author. Tel.: +45 35459839.