You are here

Article Text

Article menu

Download PDFPDF

Multiple sclerosis
Research paper
The month of birth effect in multiple sclerosis: systematic review, meta-analysis and effect of latitude
  1. Ruth Dobson1,
  2. Gavin Giovannoni1,
  3. Sreeram Ramagopalan1,2,3
  1. 1Queen Mary University of London, Blizard Institute, Barts and the London School of Medicine and Dentistry, London UK
  2. 2Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK
  3. 3Department of Clinical Neurology, John Radcliffe Hospital, Oxford, UK
  1. Correspondence to Dr Ruth Dobson, Queen Mary University of London, Blizard Institute, Newark Street, London E1 2AT; ruth.dobson{at}qmul.ac.uk

Abstract

Background Month of birth has previously been described as a risk factor for multiple sclerosis (MS). This has been hypothesised to be related to maternal vitamin D levels during pregnancy, although conclusive evidence to support this is lacking. To date, no large studies of latitudinal variation in the month of birth effect have been performed to advance this hypothesis.

Methods Previously published data on month of birth from 151 978 MS patients were compared to expected birth rates. A linear regression model was used to assess the relationship between latitude and observed:expected birth ratio of MS patients for each month.

Results Analysis of all reported data demonstrated a significant excess of MS risk in those born in April (observed:expected 1.05, p=0.05) and reduction in risk in those born in October (0.95, p=0.04) and November (0.92 p=0.01). A conservative analysis of 78 488 patients revealed an excess MS risk in those born in April (1.07, p=0.002) and May (1.11, p=0.0006), and a reduced risk in those born in October (ratio 0.94, p=0.004) and November (0.88, p=0.0002). A significant relationship between latitude and observed:expected ratio was demonstrated in December, and borderline significant relationships in May and August.

Conclusions Month of birth has a significant effect on subsequent MS risk. This is likely to be due to ultraviolet light exposure and maternal vitamin D levels, as demonstrated by the relationship between risk and latitude.

https://doi.org/10.1136/jnnp-2012-303934

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    It has long been thought that the development of multiple sclerosis (MS) is a result of a complex interaction between genes and environment.1 ,2 One of the environmental factors implicated in MS development is vitamin D deficiency.2 Vitamin D is formed by the skin when it is exposed to sunlight, and ultraviolet (UV) light exposure has also been associated with MS risk3 ,4 It is not yet understood how and when vitamin D acts to modulate MS risk, although there is increasing evidence that this occurs through gene–environment interactions.5–7

    The ‘month of birth’ effect, where those born in the winter have a reduced MS risk and those born in the spring have an increased risk, has been interpreted as indicating a pre-natal role for vitamin D in modulating MS risk.1 The month of birth effect was first described in 1987,8 and the first large-scale study by Willer et al,9 which hypothesised that this effect was linked to maternal vitamin D levels, was performed in 2005. It is thought that maternal vitamin D levels during pregnancy affect the immune status of the developing foetus, and hence modulates subsequent MS risk.1 Interest in month of birth as a potential influence on MS risk has since increased exponentially.

    A number of studies examining the variation in MS risk associated with an individual's month of birth have been performed. These have enrolled varying numbers of participants, although more recently national MS registers have facilitated large-scale population-based studies.9–12 Studies have been performed at a range of latitudes, meaning that there is a large inter-study variation in the change in UV light exposure between seasons. At latitudes greater than approximately 52° from the equator, insufficient UV light of the correct wavelength (UVB; 290–315 nm) reaches the skin between October and March to enable vitamin D synthesis during the winter months.13 It would therefore be expected that studies examining those populations living at latitudes greater than 52° would observe a significant month of birth effect, whereas those at lower latitudes would not. However, no large-scale study has studied this potential variation to date.

    We therefore set out to review and integrate the existing data on month of birth and subsequent MS risk by performing a systematic review and meta-analysis. By interrogating the available data for both risk of MS and any interaction between population latitude and the effect of month of birth on MS risk, we hope to more accurately describe the magnitude of this phenomenon, in addition to demonstrating the effect of seasonal UV light variation on the month of birth effect.

    Methods

    Inclusion criteria

    Inclusion criteria were pre-specified. Papers had to be published after 2000, include both MS and healthy control (HC) groups, and provide either month or season of birth data for each population. The relative risk of MS for each month of birth had to be described compared to an HC population, rather than relative to a reference month. In order to be included in the latitudinal analysis, papers had to provide information about the geographical location of the population studied.

    Search strategy

    PubMed and Web of Science were searched using the terms ‘MS’ AND ‘month of birth’, ‘MS’ AND ‘month’ and ‘MS’ AND ‘season’. Papers were then evaluated using the inclusion criteria described above. Additionally, the references of evaluated articles were screened for additional publications meeting the inclusion criteria. The numbers of observed and expected births for MS patients and HCs in each month were recorded for each dataset. The OR was then calculated using the observed and expected MS birth rate for each month.

    Where seasonal data only were given, the months used to define each season were examined. Prior to performing the search, the decision had been taken to use the UK Met Office definitions for each season14: winter: December, January, February; spring: March, April, May; summer: June, July, August; autumn: September, October, November. Papers that provided seasonal data not adhering to these categories15 ,16 were excluded at this point. Data given according to month of birth were combined for seasonal analysis. Only one study gave data purely by season of birth.17

    Selection for latitudinal analysis

    The geographical location of the populations used in each of the included papers was extracted from the original paper. Google maps (http://www.maps.google.com) were then used to determine the latitude. In those papers where databases from a large geographical area (such as an entire country) were used to determine population characteristics, the mid-point latitude of the geographical area was used in the analysis. In those papers where two or more geographically distinct regions were studied (ie, different countries/continents)9 ,11 and separate population figures were given for the distinct regions, these were analysed as separate datasets. The Italian cohort studied by Menni et al11 originated from three separate areas of Italy, and in this case the latitude of the central region of the three was used in the analysis. The single paper18 where the datasets from three countries were combined into a single analysis was excluded from the latitudinal analysis. In one paper,11 data were given covering the entire geographical area of the USA. These data were excluded from the latitudinal analysis due to the large area covered—the latitude of the USA ranges from 18.5°N (Hawaii) to 71.2°N (Alaska).

    Statistical analysis

    Month and season of birth

    Review Manager 5.1 was used for the initial analysis. The generic inverse variance model was used with the observed and expected MS births in each month and season. In the initial analysis the paper by Sadovnick et al19 was excluded, as the authors state that they use an identical dataset to that used by Willer et al.9 Additionally the paper by Ramagopalan et al18 was excluded, as the population used in this paper was mostly encompassed by those studied by Willer et al9 and Salzer et al.20

    A population-conservative analysis excluded additional papers where there was a reasonable suspicion that duplicate populations were being examined. Papers excluded in this analysis were those by Willer et al9 (UK data), Ramagopalan et al, Sadovnick et al and Bayes et al.17–19 In each case, in order to ensure maximum case ascertainment, the paper citing the highest number of cases was retained and all others excluded.

    A geographically-conservative analysis examined the effect of month of birth in those populations with a clear and consistent difference in UV radiation between months. This analysis selected those studies where the latitude associated with the population was greater than 52°. Included papers were those by Menni et al11 (Danish data only), Willer et al,9 Saastamoinen et al,10 Disanto et al12 and Salzer et al.20 Ramagopalan et al18 was excluded from this analysis, as it was not possible to estimate the latitude of the population, as samples from three countries were used in the study. Additionally, Sadovnick et al19 was excluded from this analysis as the authors state that the dataset used was previously used by Willer et al.9 The figure of 52° was chosen because at latitudes of about 52° and above, there is no UV light of appropriate wavelength for the cutaneous synthesis of vitamin D during October–March.13 People living at these latitudes would therefore be expected to have significant variation in vitamin D levels over the course of a normal year.

    Finally, an overall-conservative analysis was performed where all of the studies that were excluded in the population- and geographically-conservative analyses were excluded. Included papers in this section of the analysis were Menni et al11 (Danish data only), Willer et al9 (Canadian data only), Saastamoinen et al,10 Disanto et al12 and Salzer et al.20

    Effect of latitude

    A linear regression model was used for this analysis (PASW V.18 (SPSS)). Observed:expected (O:E) MS births/month were regressed on the population latitude, which had been determined as described above. The dependent variable was O:E MS births/month, and the independent variable latitude and the contribution of latitude to the equation O:E ratio≈(latitude×X)+constant were assessed for each month in turn using a linear regression model. An additional independent variable for sample size was then added to the model in order to assess whether this affected the results obtained. The papers by Sadovnick et al19 and Ramagopalan et al18 were excluded from this analysis, as the authors specified that they had used a dataset overlapping with that used by Willer et al.9

    Results

    Included studies

    The initial search generated 38 results. Three papers were excluded as they did not include a control group, two papers calculated risk compared to a reference month rather than a reference population, and two papers gave seasonal analysis using a different definition of seasons to the pre-specified definition. Details of the screening and inclusion process are given in figure 1. The 10 remaining papers were considered for inclusion in the analysis.9–12 ,17–22 Details of the included papers are given in table 1. A total of 172 918 MS births were identified, of which 151 978 were included in the analysis. There was little significant heterogeneity in any of the analyses (I2 range 0–91%, with 6 months having an I2 of 0% for at least one of the analyses performed).

    View this table:
    Table 1

    Included studies

    Figure 1
    Figure 1

    Selection of studies included.

    Month and season of birth

    Nine studies gave information on month of birth and subsequent MS risk.9–12 ,18–22 The studies by Ramagopalan et al18 and Sadovnick et al19 were excluded from this analysis for the reasons described in the methods. The O:E ratio of MS by month of birth is given in table 2. When all studies were included there were significantly fewer observed MS births than expected in October (O:E=0.95, p=0.04) and November (O:E=0.92 p=0.01). There were more MS births than expected in April (O:E=1.05, p=0.05) (table 2 and figure 2A). There was a significant variation in MS births when looking at seasonal data (figure 3A).

    View this table:
    Table 2

    Observed:expected MS cases by month of birth.

    Figure 2
    Figure 2

    (A) Variation in observed:expected multiple sclerosis (MS) births over the year calculated using all studies. The points represent the absolute values and the bars the 95% CI. (B) Variation in observed:expected MS births over the year calculated using the overall-conservative selection strategy. The points represent the absolute values and the bars the 95% CI.

    Figure 3
    Figure 3

    (A) Variation in observed:expected multiple sclerosis (MS) births between seasons calculated using all available studies. The points represent the absolute values and the bars the 95% CI. (B) Variation in observed:expected MS births between seasons calculated using the overall-conservative selection strategy. The points represent the absolute values and the bars the 95% CI.

    When the population-conservative analysis was performed, much of the significance was lost. The only month in which there was a significant deviation of MS births from expected was November, where there were significantly fewer MS births than expected (O:E=0.93, p=0.04) (table 2).

    When the geographically-conservative analysis was performed, the effect of UV variation over the course of the year was highlighted. There were significantly more observed MS births in April than expected (O:E=1.08, p=0.001), May (O:E 1.11, p=0.007) and June (O:E=1.06, p=0.05), and significantly fewer in October (O:E=0.94, p=0.006) and November (O:E=0.89, p=0.004) (table 2). Conversely, when only those studies performed at <52°N were selected (Menni et al (Italian data),11 Givon et al21 and Salemi et al22), the month of birth effect was lost, bar a borderline significant increase in observed MS births in June (O:E=1.21, 95% CI 1.01 to 1.44, p=0.04).

    Finally, the overall-conservative analysis was performed (cases=78 488). In this analysis there were significantly more observed MS births in April than expected (O:E 1.08, p=0.004) and May (O:E 1.09, p=0.002), and significantly fewer in October (O:E 0.95, p=0.03) and November (O:E 0.90, p=0.03) (table 2 and figure 2B). There was an increase in observed MS births in spring and a decrease in autumn (figure 3B).

    Effect of latitude

    When linear regression was carried out using latitude as the predictor variable, a significant relationship was seen between latitude and the ratio of O:E MS births for December (p for latitude to predict O:E ratio=0.039). A borderline significant prediction p value was seen for May (p=0.093) and August (p=0.076) (figure 4).

    Figure 4
    Figure 4

    Scatterplot demonstrating the linear relationship between observed:expected multiple sclerosis births for May, a month where an overall significant month of birth effect can be seen.

    However, when study size was added as an additional predictor variable into the model, the effect of latitude was lost (data not shown). Given that the three southernmost studies were also the three smallest studies, it is therefore difficult to know whether the latitudinal interaction is a function of study size, or is indeed a genuine effect. It must also be taken into account when interpreting this data that latitude may not act as a linear variable in terms of its effect on month of birth.

    Discussion

    Through combining existing datasets for month of birth and subsequent MS risk, this study provides the most robust evidence to date that the month of birth effect is a genuine one. While this has previously been shown in a number of studies, many of these studies have used similar or overlapping datasets, such as that developed by the Canadian Collaborative Project on the Genetic Susceptibility to Multiple Sclerosis, used in several studies.9 ,18 ,19 While these data support the month of birth effect being a result of UVB (and hence vitamin D) variation, it could result from any factor that varies in a similar seasonal and latitudinal manner. It must be noted that there is a large body of evidence supporting the importance of vitamin D in MS,23 and so maternal vitamin D levels would appear to be the most likely explanation for this effect.

    In this study we have performed a conservative analysis which excludes studies where there is evidence that the patient data used may be either wholly or partially overlapping. While this reduced the number of cases overall, this ‘population-conservative’ analysis ensured that effects exclusive to individual datasets, a potential source of error, were not subject to magnification in the overall analysis. In the population-conservative analysis, the month of birth effect was lost in all months bar November. This is likely to be due to the fact that the excluded studies were all performed at a latitude >52°N, and so the UV variation over the course of a year was significantly reduced in the remaining studies.

    When the studies that were performed at a latitude of <52°N were excluded, the month of birth effect once again became apparent. There was a highly significant increase in MS births in both April and May, and a reduction in October and November. Only one study12 has previously shown a significant reduction in all of these months. This finding was complemented by the demonstration that the month of birth effect is almost entirely lost when selecting those studies performed at <52°N. However, it must be noted that the geographically conservative analysis has the potential to overestimate some population-specific risks, due to the high probability of duplicate data in the analysis. Any deviation present in these datasets will therefore be exaggerated in this analysis.

    No studies from the southern hemisphere were included in this analysis. This was not a deliberate selection criterion, but instead reflects the imbalance in the origin of such studies. While studies into the month of birth effect in the southern hemisphere do exist—including Staples et al (Australia)24 and Fragoso et al (Brazil)25—the data were not presented in a manner that could be used in this analysis. There remains a need for further studies in the southern hemisphere in order to confirm if the reversal in the month of birth effect noted by Staples et al24 exists in other countries.

    The overall conservative analysis had the least number of participants, 78 488. However, it is likely to be the most appropriate analysis, as both potential duplicate datasets and those studies performed in areas with low variation in UV exposure during the year were excluded. The highly significant increase in MS risk in those born in April and May remains clear, as does the reduction in risk in those born in October and November. By pooling data and performing a meta-analysis, the month of birth effect can be extended from that previously described. This effect is highlighted when looking at the differences in risk of MS stratified by season of birth.

    The significant interaction seen between latitude and O:E MS births highlights the latitude-specific nature of the month of birth effect. While it has previously been suggested that the month of birth effect is dependent on latitude,25 with a loss of effect at lower latitudes, this is the first study to demonstrate a significant interaction between the two. The fact that study size confounds the relationship is likely to result from the small populations used in the southernmost studies, rather than systematic selection bias.

    However, it must be borne in mind that this study is a meta-analysis of existing data, and therefore has weaknesses in keeping with this methodology. Publication and selection bias are potential problems, although studies showing both no effect and significant effects were included, and there was no evidence of bias in funnel plots. Additionally, although steps were taken to attempt to exclude duplicate data, it may be that some remains, influencing the results. The lack of southern hemisphere studies is a significant limitation, as a demonstration of the reversal of the month of birth effect would strengthen confidence in this finding considerably. Combining the data from individual months into seasons (as defined by the UK Met Office) may have introduced bias into the results, either strengthening or weakening the association between season of birth and MS. However, combining the data in this way allowed the addition of a further study. These results must therefore be taken in the context of the other results presented here.

    In conclusion, this study, which uses the largest number of patients to date, confirms and extends the month of birth effect seen in MS. Through the demonstration of an interaction between month of birth effect magnitude and latitude, it supports ambient UV radiation, and hence maternal vitamin D levels, as pre-natal environmental modulators of MS risk. This finding, which supports concepts hypothesised some years previously,1 surely adds weight to the argument for early intervention studies to prevent MS through vitamin D supplementation.1

    References

    1. ↵
      1. Ebers G
      . Environmental factors and multiple sclerosis. Lancet Neurol 2008;7:268–77.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Ramagopalan SV,
      2. Dobson R,
      3. Meier UC,
      4. et al
      . Multiple sclerosis: risk factors, prodromes, and potential causal pathways. Lancet Neurol 2010;9:727–39.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Orton SM,
      2. Wald L,
      3. Confavreux C,
      4. et al
      . Association of UV radiation with multiple sclerosis prevalence and sex ratio in France. Neurology 2011;76:425–31.
      OpenUrlCrossRef
    4. ↵
      1. Ramagopalan SV,
      2. Handel AE,
      3. Giovannoni G,
      4. et al
      . Relationship of UV exposure to prevalence of multiple sclerosis in England. Neurology 2011;76:1410–4.
      OpenUrlCrossRef
    5. ↵
      1. Ramagopalan SV,
      2. Hanwell HE,
      3. Giovannoni G,
      4. et al
      . Vitamin D-dependent rickets, HLA-DRB1, and the risk of multiple sclerosis. Arch Neurol 2010;67:1034–5.
      OpenUrlCrossRefPubMed
      1. Agliardi C,
      2. Guerini FR,
      3. Saresella M,
      4. et al
      . Vitamin D receptor (VDR) gene SNPs influence VDR expression and modulate protection from multiple sclerosis in HLA-DRB1*15-positive individuals. Brain Behav Immun 2011;25:1460–7.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Ramagopalan S,
      2. Maugeri N,
      3. Handunnetthi L,
      4. et al
      . Expression of the multiple sclerosis-associated MHC class II Allele HLA-DRB1*1501 is regulated by vitamin D. PLoS Genet 2009;5:e1000369.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Shimura MKT,
      2. Miura T
      . Season of birth in some neurological disorders: multiple sclerosis, ALS, senile dementia. Prog Biometeorol 1987;6:163–8.
      OpenUrl
    9. ↵
      1. Willer C,
      2. Dyment D,
      3. Sadovnick A,
      4. et al
      . Timing of birth and risk of multiple sclerosis: population based study. BMJ 2005;330:120.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Saastamoinen KP,
      2. Auvinen MK,
      3. Tienari PJ
      . Month of birth is associated with multiple sclerosis but not with HLA-DR15 in Finland. Mult Scler 2012;18:563–8.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Menni C,
      2. Lowell WE,
      3. Bentzen J,
      4. et al
      . Short and long term variation in ultraviolet radiation and multiple sclerosis. Int J Environ Res Public health. 2012;9:685–97.
      OpenUrlCrossRefPubMed
    12. ↵
      1. Disanto G,
      2. Chaplin G,
      3. Morahan JM,
      4. et al
      . Month of birth, vitamin D and risk of immune mediated disease: a case control study. BMC Med 2012;10:69.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Webb AR,
      2. Kline L,
      3. Holick MF
      . Influence of season and latitude on the cutaneous synthesis of vitamin D3: exposure to winter sunlight in Boston and Edmonton will not promote vitamin D3 synthesis in human skin. J Clin Endocrinol Metab 1988;67:373–8.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
    15. ↵
      1. Fernandes de Abreu DA,
      2. Babron MC,
      3. Rebeix I,
      4. et al
      . Season of birth and not vitamin D receptor promoter polymorphisms is a risk factor for multiple sclerosis. Mult Scler 2009;15:1146–52.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Sotgiu S,
      2. Pugliatti M,
      3. Sotgiu M,
      4. et al
      . Seasonal fluctuation of multiple sclerosis births in Sardinia. J Neurol 2006;253:38–44.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Bayes HK,
      2. Weir CJ,
      3. O'Leary C
      . Timing of birth and risk of multiple sclerosis in the Scottish population. Eur Neurol 2010;63:36–40.
      OpenUrlCrossRefPubMed
    18. ↵
      1. Ramagopalan SV,
      2. Link J,
      3. Byrnes JK,
      4. et al
      . HLA-DRB1 and month of birth in multiple sclerosis. Neurology 2009;73:2107–11.
      OpenUrlCrossRef
    19. ↵
      1. Sadovnick AD,
      2. Duquette P,
      3. Herrera B,
      4. et al
      . A timing-of-birth effect on multiple sclerosis clinical phenotype. Neurology 2007;69:60–2.
      OpenUrlCrossRef
    20. ↵
      1. Salzer J,
      2. Svenningsson A,
      3. Sundstrom P
      . Season of birth and multiple sclerosis in Sweden. Acta Neurol Scand 2010;122:70–3.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Givon U,
      2. Zeilig G,
      3. Dolev M,
      4. et al
      . The month of birth and the incidence of multiple sclerosis in the Israeli population. Neuroepidemiology 2012;38:64–8.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Salemi G,
      2. Ragonese P,
      3. Aridon P,
      4. et al
      . Is season of birth associated with multiple sclerosis? Acta Neurol Scand 2000;101:381–3.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Smolders J
      . Vitamin d and multiple sclerosis: correlation, causality, and controversy. Autoimmune Dis 2011;2011:629538.
      OpenUrlPubMed
    24. ↵
      1. Staples J,
      2. Ponsonby AL,
      3. Lim L
      . Low maternal exposure to ultraviolet radiation in pregnancy, month of birth, and risk of multiple sclerosis in offspring: longitudinal analysis. BMJ 2010;340:c1640.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Fragoso YD,
      2. Shearer KD,
      3. Adoni T,
      4. et al
      . Month of birth does not seem to interfere with the development of multiple sclerosis later in life in Brazilian patients. Neuroepidemiology 2012;39:70–1.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Contributors RD, GG and SVR conceived the idea of this paper. RD performed the literature search and analysis, and drafted the initial manuscript. SVR assisted with data analysis. SVR and GG provided input into the final draft of the manuscript. All authors provided intellectual contribution to the article.

    • Funding This work was supported by MS Society of Great Britain and Northern Ireland, grant number 940/10. RD is funded by an Association of British Neurologists/MS Society of Great Britain Clinical Research Fellowship. SVR is supported by the Medical Research Council (MRC). GG receives grant support from the MRC, National MS Society, MS Society of Great Britain and Northern Ireland, AIMS2CURE and the Roan Charitable Trust.

    • Competing interests GG has received research grant support from Bayer-Schering Healthcare, Biogen-Idec, GW Pharma, Merck Serono, Merz, Novartis, Teva and Sanofi-Aventis. GG has received personal compensation for participating on Advisory Boards in relation to clinical trial design, trial steering committees and data and safety monitoring committees from: Bayer-Schering Healthcare, Biogen-Idec, Eisai, Elan, Fiveprime, Genzyme, Genentech, GSK, Ironwood, Merck-Serono, Novartis, Pfizer, Roche, Sanofi-Aventis, Synthon BV, Teva, UCB Pharma and Vertex Pharmaceuticals.

    • Provenance and peer review Not commissioned; externally peer reviewed.