Multiple sclerosis (MS) is an immune-mediated demyelinating disease involving the central nervous system. MS is the most common neurologic disease affecting young adults, especially women of childbearing age, which makes reproductive factors an issue of major interest1 from a clinical, as well as from an etiologic, point of view.
During pregnancy, a complex alteration of the immune system takes place to avoid maternal rejection of the fetus.2,3 This mechanism, by which tolerance to the fetus is established, also seems to have a favorable effect on the course of certain autoimmune diseases,4 including MS.5–7 This in turn could suggest that the biologic changes associated with pregnancy might influence the course of MS and perhaps also the development of MS.
An imbalance in the immunologic mechanisms responsible for the tolerance of the fetus has been suggested as a possible cause of miscarriage, preterm delivery, and preeclampsia.2,8–10 Pregnancy complications and infertility could therefore indicate malfunctions of the female immune system not only during pregnancy, but also during nonpregnant periods, which again could influence the woman's risk of autoimmune diseases.
Previous studies on the association of childbearing patterns and MS risk have produced conflicting results, variously finding no association between parity and MS risk,11–13 a slight protective effect of pregnancy,7,14 or a triggering effect (precipitating the onset of MS).15 Episodes of preeclampsia,14,16 hyperemesis,16 and spontaneous abortion16,17 have been reported among women who subsequently developed MS, but small numbers of participants and the general lack of appropriate control groups in these studies prevent causal interpretations.
By using information from several nationwide registers, we established a population-based cohort of all Danish men and women born between 1935 and 1989 and alive in 1968 or later. We examined associations between reproductive history and subsequent risk of MS. Unlike prior investigations of reproductive factors and MS, we also studied associations of liveborn children with MS risk among men. This allowed us to evaluate the significance of factors associated with parenting but not related to the biologic effect of pregnancy, such as lifestyle and socioeconomic status.
By means of data from the Danish Civil Registration System (a continuously updated demographic database covering the entire Danish population),18 we identified study cohorts comprising all Danish women and men (4.4 million) born between 1935 and 1989 and alive in 1968 or later. We followed these persons with respect to both reproductive factors and diagnosis of MS. Information about reproductive history and pregnancy complications, as well as MS outcomes in the cohorts, was drawn from the Civil Registration System and other nationwide registers.
Information about liveborn children was obtained from the Civil Registration System, which was established on 1 April 1968. As described in detail elsewhere,18 the Civil Registration System contains continuously updated information about vital status, residence, and close-to-complete registration of all children of women and men born in Denmark since 1935 and alive on 1 April 1968 or later. Thus, for each member of the study cohorts, we had information about number of children, which we defined operationally as the number of pregnancies that resulted in at least one liveborn child (ie, multiple births counted as only one pregnancy).
Information about pregnancy complications and pregnancy losses was obtained from the Danish Hospital Discharge Register, a nationwide register containing data about all nonpsychiatric hospital admissions in Denmark since January 1977 (outpatients included since 1995). Diagnoses are coded according to the WHO international classification of diseases (ICD), using its eighth revision from 1977 to 1993 and its 10th revision since January 1994.19 We obtained information about spontaneous abortions (ICD8: 643; ICD10: O03), missed abortions (ICD8: 634.61, 634.62, 634.63, 634.69,645.1; ICD10: O021), and ectopic pregnancies (ICD8: 631; ICD10: O00), as well as information about pregnancies complicated by hyperemesis (ICD8: 638, 762.49; ICD10: O21), gestational hypertension (ICD8: 637.00, 637.02; ICD10: O12, O13, O16), or preeclampsia (ICD8: 637.03–637.99, 639.99, 762.19–762.39, 762.99; ICD10: O14, O15). Information about induced abortions was obtained from the National Register of Induced Abortions, which has been in operation since the law on induced abortion was implemented in Denmark on 1 October 1973.20 Since 1995, the register has been a part of the Danish Hospital Discharge Register (ICD8: 640, 641, 642; ICD10: O04, O05, O06). Cases of stillbirths (which until 2004 was defined as fetal death after the 27th week of gestation and since 2004 as fetal death after the 21st week of gestation21) were identified using the Danish Medical Birth Register, established on 1 January 1973.22 Women with infertility problems were identified using information about all women who were either referred to Danish public hospitals or private fertility clinics in the period 1968–1998 or registered with a diagnosis of infertility (ICD8: 628; ICD10: N97) in the Danish Hospital Discharge Register from 1977 to 1998.23 As most of the above registers were established in the mid 1970s, virtually complete information on pregnancy complications, pregnancy losses, and infertility was available only for women born in 1955 or later. Thus, using the nationwide registers, we had close-to-complete information about reproductive history for all cohort members through 2004, except for information about infertility, which was available only until 1998.
MS cases in the study cohort were identified in the Danish Multiple Sclerosis Register, which was established in 1956 as a continuation of a nationwide MS prevalence study carried out a few years earlier. The register has since collected information about MS patients from all Danish hospital departments of neurology and neuropathology, practicing neurologists, MS rehabilitation centers, death certificates, and (since 1977) the Danish Hospital Discharge Register.24 All cases have been reclassified by neurologists at the Danish Multiple Sclerosis Register, and we included in the present study all cases fulfilling the diagnostic criteria of Allison or Poser24 including possible MS (when other diseases are ruled out), but not cases of clinically isolated syndrome or poorly documented cases. Appearance of first symptoms of MS and date of diagnosis of MS are only recorded by calendar year. Consequently, in all analyses, we defined the date of first symptoms as 1 July in the recorded year of first symptoms, and the date of diagnosis as 1 July in the year of diagnosis. If the MS patient died or emigrated in the same year as diagnosis, the date of diagnosis was defined as the day before death or emigration.
Throughout, we analyzed men and women separately. Each cohort member contributed person-years at risk for MS from his or her 15th birthday or 1 April 1968, whichever came later, and until MS diagnosis or the date of death, emigration, disappearance, or 1 January 2005, whichever came first. We considered only pregnancies occurring before the diagnosis of MS as potential risk factors. Persons diagnosed with MS before 1 April 1968, or before the age of 15 years were excluded; thus, the most recent birth cohort participating in the study comprised men and women born in 1989, ie, those who turned 15 years old in 2004. Person-years at risk and MS outcomes were stratified according to time-dependent values of age; birth cohort and calendar period (all in 1-year groups); marital status (unmarried, married, separated/divorced, or widowed); and each of the following live birth-related variables: number of children (0, 1, 2, 3, 4+), age at first live birth (<20, 20–24, 25–29, 30+ years), and time since birth of the most recent live-born child (<2, 2–4, 5–9, 10–14, 15–19, 20+ years).
For the period 1982–2004, we also adjusted for socioeconomic factors obtained from the Integrated Database for Labor Market Research, a database that contains annually updated information about socioeconomic factors for the entire Danish population from 1980 to the present.25 Specifically, person-years at risk and MS outcomes were stratified according to annual values of the highest household educational level among adults in a household (basic school, high school, vocational education, short higher education, medium higher education, and long higher education), and according to annual values of the relative household income (<50%, 50%–74%, 75%–124%, 125%–149%, ≥150% of the average household income for persons of the same sex and birth year). To account for possible changes in income associated with incipient MS, we used information about income for the calendar year 2 years before the year of observation.
For women born in 1955 or later, we further stratified person-years and MS outcomes according to information about spontaneous abortions (0, 1+), missed abortions (0, 1+), stillbirths (0, 1+), induced abortions (0, 1+), ectopic pregnancies (0, 1+), hyperemesis (0, 1+), gestational hypertension (0, 1+) and preeclampsia (0, 1+). The effect of infertility was investigated using a time-dependent variable indicating whether a woman had been evaluated for infertility problems (0, 1).
Poisson Regression Analysis
The statistical analysis of the resulting table of stratum-specific MS incidence rates was carried out as a log-linear Poisson regression analysis, yielding incidence rate ratios (RRs) of MS with likelihood-based 95% confidence intervals (CIs). Adjustments for age, birth cohort, and calendar period were performed for all RR calculations using cubic splines restricted to be linear in the tails.26 Adjustment for other variables was performed by using the categorizations described earlier in the text. In trend tests, we treated categorical quantitative variables as continuous variables by replacing each category with the person-years-weighted median of the original detailed variable in that category. Trend tests were carried out only after acceptance (P ≥ 0.05) of the corresponding model reduction from a categorical to a linear description of the association between MS and the variable in question.
In the cohort of 2.14 million Danish women born between 1935 and 1989, 6332 cases of MS were identified in the Danish Register of Multiple Sclerosis between 1968 and 2004. Women who had given birth to at least one child were at 24% reduced risk of developing MS compared with nulliparous women (RR = 0.76 [95% CI = 0.71–0.82]) (Table 1). Among parous women, the more children the women had given birth to, the lower the risk of MS (change per additional child, RR = 0.87 [0.84–0.91]; test for trend P < 0.001). Thus, compared with 1-child mothers, women with 2, 3, and 4 or more children were at 11% (RR = 0.89 [0.83–0.96]), 21% (0.79 [0.72–0.87]), and 38% (0.62 [0.53–0.72]) reduced risk of MS, respectively, adjusted for age, birth cohort, period, marital status, and age at birth of first child (Table 1).
We observed an inverse association between the age at birth of the first child and risk of MS. Thus, compared with the reference group of women who had their first child at the age of 20–24 years, women who gave birth to their first child at the age of 25–29 years or at the age of ≥30 years were at 9% (RR = 0.91 [0.85–0.98]) and 23% (0.77 [0.68–0.86]) reduced risk of MS, respectively, adjusted for age, birth cohort, period, marital status, and number of children (Table 1). The change per additional year increase in age was: RR = 0.98 [0.97–0.99]; test for trend P < 0.001.
The risk of MS increased with time since birth of the most recent child. Compared with the reference group of women who gave birth to their most recent child 10–14 years ago, the risk of MS was 21% (RR = 0.79 [0.70–0.90]) lower among women who had their most recent child <2 years ago, and 11% (1.11 [0.98–1.26]) higher among women who had their most recent child more than 20 years ago (Table 1).
In a subcohort of 1.39 million Danish women born between 1955 and 1989, information about adverse pregnancy outcomes was available (Table 2). None of the pregnancy factors examined, including infertility status, was associated with an unusual risk of MS.
In the cohort of 2.24 million men born between 1935 and 1989, 3426 cases of MS were identified in the MS register between 1968 and 2004. Associations between children (yes/no), number of children, age at birth of first child and time since birth of most recent child, and the subsequent risk of MS were similar to those seen for women. Men who had at least one child were at reduced risk of MS (RR = 0.89 [0.80–0.98]) compared with men without children and, among men with at least one child, the more children a man had fathered the lower was his risk of MS (change per additional child, RR = 0.89 [0.84–0.94]; test for trend P < 0.001). The inverse associations between age at birth of the first child and risk of MS seen in women were also seen for men, just as the risk of MS increased with time since the birth of the most recent child (test for trend P < 0.001) (Table 1).
To evaluate the robustness of our findings in women, we repeated all analyses of the possible impact of live births using 3 alternative definitions of MS. In the main analysis, presented in Table 1, we included possible cases of MS in our MS definition, and the onset of MS was defined by the year of diagnosis recorded in the Danish Multiple Sclerosis Register. In the first robustness analysis, we excluded patients with possible MS from our definition of MS cases. In the second robustness analysis, we included possible MS cases but the onset of MS was defined by the year of first MS symptoms recorded in the Danish Multiple Sclerosis Register. In the third robustness analysis, we defined MS outcomes based on the date of first hospital admittance with MS recorded in the Danish Hospital Discharge Register (eTable, http://links.lww.com/EDE/A480). Associations between children (yes/no), number of children, age at first child's birth, time since most recent child's birth, and RR of MS were similar regardless of the MS definition used.
To address possible confounding influences by differences in socioeconomic status among participants in our study, we conducted a supplementary analysis restricted to the period 1982–2004, in which we adjusted for 2 socioeconomic variables (highest household educational level and relative household income). Compared with results in Table 1, results of this supplementary analysis were virtually unchanged (data not shown).
Finally, to address the possibility of reverse causality as an explanation for the above findings, we repeated the analysis shown in Table 1 for children (yes/no) by backdating the date of MS diagnosis by 5 years, assuming that reproductive behaviors in women destined to develop MS would not be affected to any great extent more than 5 years before MS diagnosis. In this analysis, there was no indication of an overall reduction in MS risk among persons with children (for women, RR = 0.95 [0.88–1.03]; for men, 1.08 [0.98–1.20]).
Women who had given birth to at least one child were at reduced risk of MS compared with nulliparous women. Furthermore, the higher the parity and the older the woman was when she gave birth to her first child, the lower was her risk of MS. Additionally, the risk of MS increased with time since the birth of the most recent child. However, almost identical results were observed for men, suggesting that our findings among women do not reflect a biologic effect of pregnancy. Adverse pregnancy outcomes and pregnancy complications, as well as prior infertility evaluation, were not associated with risk of MS.
Our observations accord well with findings in some previous studies that showed an inverse association between pregnancy and MS risk7,14 and no increased risk of MS following stillbirth, abortions, or preeclampsia.15,16 However, other studies observed that neither parity nor age at first childbirth was associated with MS,24,25 and others have found that the risk of MS was increased in the year following pregnancy.11,12,15 However, none of these previous studies considered associations between reproductive factors and MS risk among men.
Women with higher education tend to have lower parity and to have their first child at an older age compared with women with shorter education.27 Accordingly, an association of socioeconomic status with risk of MS could contribute to our findings. However, in a supplementary analysis, we adjusted for 2 socioeconomic variables, household educational level, relative household income, and estimates were essentially unchanged.
Previous studies have shown that women with MS are no more likely to have spontaneous abortions, congenital malformations, stillbirths, prematurity or preeclampsia than healthy women.28–30 However, compared with the general population both women and men with MS seem to have a higher frequency of childlessness,7,31 which could reflect lowered fertility or voluntary childlessness. Whether MS has an impact on fertility is still debated,6,32 but the present observations indicate that women who undergo evaluation for infertility are not at any elevated risk of developing MS. The desire to have a child might, however, be influenced by the severity and progression of the disease. A recent study in Germany showed that mothers with MS had slightly shorter disease duration and less impairment according to the Expanded Disability Status Scale score compared with childless women of the same age with MS.33 In general, men and women reporting long-standing illness and poor health are less likely to marry34 and are more likely to either not have children or have only few children.17,35 Consequently, having many children and having children at a relatively advanced age presumably are markers of good health.36
Recent studies have shown that a substantial proportion of MS patients may have had symptoms suggestive of MS several years before evaluation for the first demyelinating event.37 Additionally, MS patients have been reported to have contacts with their general practitioners for a variety of symptoms as long as 10 years before diagnosis.38 Accordingly, our findings in Table 1 of a higher risk of MS among childless women and men could reflect an association between reduced reproductive activity and declining health during a premorbid stage of MS. We evaluated the possibility of such reverse causality by backdating the time of MS onset to the reported date of MS diagnosis minus 5 years. Findings in this analysis did not corroborate the overall reduction in MS risk associated with parenthood seen in Table 1. Rather, the negative findings in the supplementary analysis combined with the parenthood-associated reduction in MS risk seen in Table 1 suggest that, for yet unclear reasons, women and men who were destined to be diagnosed with MS within the next 5 years had fewer children than other persons. Perhaps, subtle symptoms or premorbid immunologic changes in the years before MS diagnosis somehow affected their wish or ability to become parents.
We obtained information on MS from the Danish Multiple Sclerosis Register, where all reports undergo individual evaluation by experienced neurologists. This register has been estimated to be more than 90% complete and to have a diagnostic validity of 94%.24 In addition to verified cases of MS, possible cases of MS are also recorded in the Danish Register of Multiple Sclerosis and all MS cases are recorded with the year of first symptoms as well as with the year of diagnosis. In the main analysis, we considered all the cases recorded in the Danish Register of Multiple Sclerosis, and we used year of diagnosis to define MS outcomes. In supplementary analyses, we excluded possible MS cases, used the year of first MS symptoms instead of year of MS diagnosis, including possible cases, and, finally, restricted MS outcomes to those that were recorded in the Danish Hospital Discharge Register; risk estimates were virtually unchanged.
Information about live births and induced abortions was collected from national registers that are likely to be accurate and virtually complete. The validity and completeness of other reproductive variables may be lower.39 However, in a previous study, more than 70% of women reporting spontaneous abortions could be traced under codes for spontaneous abortion in the files of the Danish Hospital Discharge Register.40
We followed a cohort comprising all Danish residents born between 1935 and 1989 and identified their recorded reproductive histories and subsequent MS diagnosis in national registries. Thus, by design our cohort study is free of selection biases that may complicate interpretation of findings in case-control studies. Moreover, similar findings in the 2 sexes indicate no biologic effect of pregnancy on the risk of MS. Thus, the present parallel analyses in women and men serve as a useful addition to prior studies, raising questions about possible biologic interpretations of the link between parenthood and MS risk previously suggested on the basis of studies restricted to women.
In conclusion, our findings argue against a biologic role of pregnancy in the etiology of MS. The observed increase in risk of MS among childless women and men could be explained by reverse causality—ie, reduced reproductive activity in persons with yet undiagnosed MS.
We thank the two anonymous reviewers whose suggestions improved the manuscript considerably.
1. Houtchens MK. Pregnancy and multiple sclerosis. Semin Neurol
2. Saito S, Shiozaki A, Nakashima A, et al. The role of the immune system in preeclampsia. Mol Aspects Med
3. Shigeru S, Akitoshi N, Subaru MH, et al. The balance between cytotoxic NK cells and regulatory NK cells in human pregnancy. J Reprod Immunol
4. Ostensen M, Villiger PM. The remission of rheumatoid arthritis during pregnancy. Semin Immunopathol
5. Confavreux C, Hutchinson M, Hours MM, et al. Rate of pregnancy-related relapse in multiple sclerosis. Pregnancy in Multiple Sclerosis Group. N Engl J Med
6. Birk K, Rudick R. Pregnancy and multiple sclerosis. Arch Neurol
7. Runmarker B, Andersen O. Pregnancy is associated with a lower risk of onset and a better prognosis in multiple sclerosis. Brain
. 1995;118(pt 1):253–261.
8. Makhseed M, Raghupathy R, El-Shazly S, et al. Pro-inflammatory maternal cytokine profile in preterm delivery. Am J Reprod Immunol
9. Piccinni MP. T cells in normal pregnancy and recurrent pregnancy loss. Reprod Biomed Online
10. Raghupathy R, Kalinka J. Cytokine imbalance in pregnancy complications and its modulation. Front Biosci
11. Hernan MA, Hohol MJ, Olek MJ, et al. Oral contraceptives and the incidence of multiple sclerosis. Neurology
12. Operskalski EA, Visscher BR, Malmgren RM, et al. A case-control study of multiple sclerosis. Neurology
13. Thorogood M, Hannaford PC. The influence of oral contraceptives on the risk of multiple sclerosis. Br J Obstet Gynaecol
14. Villard-Mackintosh L, Vessey MP. Oral contraceptives and reproductive factors in multiple sclerosis incidence. Contraception
15. Leibowitz U, Antonovsky A, Kats R, et al. Does pregnancy increase the risk of multiple sclerosis? J Neurol Neurosurg Psychiatry
16. Tillman AJ. The effect of pregnancy on multiple sclerosis and its management. In: Woltman HW, ed. Multiple Sclerosis and the Demyelinating Diseases: Proceedings of the Association for Research in Nervous and Mental Diseases, December 10 and 11, 1948, New York
. Baltimore: The Williams and Wilkins Company; 1950:548–582.
17. Poser S, Raun NE, Wikstrom J, et al. Pregnancy, oral contraceptives and multiple sclerosis. Acta Neurol Scand
18. Pedersen CB, Gotzsche H, Moller JO, et al. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull
19. Andersen TF, Madsen M, Jorgensen J, et al. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull
20. Knudsen LB. Induced abortions in Denmark. Acta Obstet Gynecol Scand Suppl
21. Fødselsregisteret 2004. The Danish National Board of Health. Available at: http://www.sst.dk/publ/tidsskrifter/nyetal/pdf/2005/04_05.pdf
22. Knudsen LB, Olsen J. The Danish Medical Birth Registry. Dan Med Bull
23. Jensen A, Sharif H, Olsen JH, et al. Risk of breast cancer and gynecologic cancers in a large population of nearly 50,000 infertile Danish women. Am J Epidemiol
24. Koch-Henriksen N, Rasmussen S, Stenager E, et al. The Danish Multiple Sclerosis Registry. History, data collection and validity. Dan Med Bull
25. Petersen JK. The Danish demographic database—longitudinal data for advanced demographic methods. I: research report 15. Odense, Denmark: Demografisk Forskningscenter; 2000.
26. Harrell FE Jr. Regression Modeling Strategies: With Application to Linear Models, Logistic Regression, and Survival Analysis
. New York: Springer; 2001.
27. Dano H, Hansen KD, Jensen P, et al. Fertility pattern does not explain social gradient in breast cancer in Denmark. Int J Cancer
28. Abramsky O. Pregnancy and multiple sclerosis. Ann Neurol
29. Ferrero S, Pretta S, Ragni N. Multiple sclerosis: management issues during pregnancy. Eur J Obstet Gynecol Reprod Biol
30. Mueller BA, Zhang J, Critchlow CW. Birth outcomes and need for hospitalization after delivery among women with multiple sclerosis. Am J Obstet Gynecol
31. Koch-Henriksen N. An epidemiological study of multiple sclerosis. Familial aggregation social determinants, and exogenic factors. Acta Neurol Scand Suppl
32. Cavalla P, Rovei V, Masera S, et al. Fertility in patients with multiple sclerosis: current knowledge and future perspectives. Neurol Sci
33. Twork S, Wirtz M, Schipper S, et al. Chronical illness and maternity: life conditions, quality of life and coping in women with multiple sclerosis. Qual Life Res
34. Cheung YB, Sloggett A. Health and adverse selection into marriage: evidence from a study of the 1958 British birth cohort. Public Health
35. Mountifield R, Bampton P, Prosser R, et al. Fear and fertility in inflammatory bowel disease: a mismatch of perception and reality affects family planning decisions. Inflamm Bowel Dis
36. Grundy E, Kravdal O. Reproductive history and mortality in late middle age among Norwegian men and women. Am J Epidemiol
37. Gout O, Lebrun-Frenay C, Labauge P, et al. Prior suggestive symptoms in one-third of patients consulting for a “first” demyelinating event. J Neurol Neurosurg Psychiatry
38. Hamilton WT, Hall GH, Round AP. Frequency of attendance in general practice and symptoms before development of chronic fatigue syndrome: a case-control study. Br J Gen Pract
39. Nickelsen TN. Data validity and coverage in the Danish National Health Registry. a literature review [in Danish]. Ugeskr Laeger
40. Buss L, Tolstrup J, Munk C, et al. Spontaneous abortion: a prospective cohort study of younger women from the general population in Denmark. Validation, occurrence and risk determinants. Acta Obstet Gynecol Scand