Advertisement
Research Article

The Epidemiology of Multiple Sclerosis in Scotland: Inferences from Hospital Admissions

  • Adam E. Handel,

    Affiliations: Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom, Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom

    X
  • Lynne Jarvis,

    Affiliation: National Health Service Scotland, Edinburgh, United Kingdom

    X
  • Ryan McLaughlin,

    Affiliations: Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom, Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom

    X
  • Anastasia Fries,

    Affiliations: Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom, Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom

    X
  • George C. Ebers,

    Affiliations: Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom, Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom

    X
  • Sreeram V. Ramagopalan mail

    sreeramr@well.ox.ac.uk

    Affiliations: Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom, Department of Clinical Neurology, University of Oxford, Oxford, United Kingdom

    X
  • Published: January 27, 2011
  • DOI: 10.1371/journal.pone.0014606

Abstract

Background

Multiple sclerosis (MS) is a neurological disorder with a highly characteristic disease distribution. Prevalence and incidence in general increase with increasing distance from the equator. Similarly the female to male sex ratio increases with increasing latitude. Multiple possible risk factors have been hypothesised for this epidemiological trend, including human leukocyte antigen allele frequencies, ultraviolet exposure and subsequent vitamin D levels, smoking and Epstein-Barr virus. In this study we undertook a study of medical records across Scotland on an NHS health board level of resolution to examine the epidemiology of MS in this region.

Methods and Results

We calculated the number and rate of patient-linked hospital admissions throughout Scotland between 1997 and 2009 from the Scottish Morbidity Records. We used weighted-regression to examine correlations between these measures of MS, and latitude and smoking prevalence. We found a highly significant relationship between MS patient-linked admissions and latitude (r weighted by standard error (rsw) = 0.75, p = 0.002). There was no significant relationship between smoking prevalence and MS patient-linked admissions.

Discussion

There is a definite latitudinal effect on MS risk across Scotland, arising primarily from an excess of female MS patients at more Northerly latitudes. Whether this is a true gradient or whether a threshold effect may apply at particular latitude will be revealed only by further research. A number of genetic and environmental factors may underlie this effect.

Introduction

Multiple sclerosis (MS) is a complex neurological disorder characterised by demyelination and axonal loss.[1] MS shows a marked variation by latitude throughout the world, with increasing prevalence and incidence with increasing distance from the equator.[2], [3] This latitudinal effect has been observed in a large number of different studies in the Northern hemisphere and also holds true in the Southern hemisphere. Much of this latitude gradient appears to stem from an excess of female MS patients at higher latitudes.[4], [5]

Multiple genetic and environmental risk factors have been hypothesised to underlie the distribution of MS, including human leukocyte antigen (HLA) allele frequencies, ultraviolet radiation and vitamin D levels, smoking, and infection with Epstein-Barr virus.[6], [7], [8], [9] It is likely that many of these interact together on a population-wide level to determine MS prevalence and incidence.[8], [10], [11]

The Northern parts of Scotland have previously been shown to have a particularly high prevalence of MS.[12] This appears to be responsible for much of the latitudinal variation in MS across the United Kingdom.[13] However recent data on the prevalence of MS in Scotland is not available. Given suggestions that the latitude gradient in MS is decreasing,[5] we undertook an analysis of MS hospital admissions stratified by NHS health board in Scotland in order to examine epidemiological trends in MS in fine detail.

Methods

Data on hospital admissions for MS between 1997 and 2009 were analysed using a file of linked hospital admission statistics, built from the Scottish Morbidity Records (SMR01) system. SMR01 records all inpatient and day case discharges from non-obstetric and non-psychiatric specialties in NHS Hospitals in Scotland. Probability matching methods have been used to link together individual hospital records for each patient, thereby creating “linked” patient histories. Up to six diagnoses (1 principal, 5 secondary) are recorded on SMR01 returns. All six diagnoses have been used to select Multiple Sclerosis. The following code was used from the International Statistical Classification of Diseases and Other Health Problems, tenth revision (ICD10): Multiple Sclerosis - G35. These diagnoses are made by the lead physician responsible for each patient's care during the admission.

We show data for the areas of the 14 Scottish Strategic Health Boards. MS indirect admission rates were calculated using observed average annual numbers of MS patients from each area with one or more hospital admissions as the numerators and the expected number of average annual MS admissions from census data as the denominators, before multiplying this by the overall MS admissions rate for Scotland. Using data linkage, we identified each person only once for MS, regardless of how many admissions each person had, and recorded their residence at first known admission for MS. To adjust for differences in the age structure of different areas, age-standardisation was undertaken using the indirect method and the age-specific rates in the Scottish population. All rates are expressed per 100 000 population with 95% lower and upper confidence limits. Rates were calculated separately for males and females as well as both sexes combined. We also calculated rates for other autoimmune diseases (rheumatoid arthritis, ulcerative colitis, Crohn's disease and type 1 diabetes mellitus) and a neurological disease (motor neuron disease).

Smoking data was obtained from the Atlas of Tobacco Smoking in Scotland, produced by the NHS Health Scotland, ISD Scotland and ASH Scotland.[14] Population smoking prevalence estimates were used from 2003/2004 and were based upon all age groups combined. Latitudes for each NHS health board were defined as the latitude of the administrative centre for that health board. As anonymised data were used we followed the ethical principles of existing UK data protection legislation and guidance, including two National Statistics (NS) Protocols on Data Access and Confidentiality, and Data Matching and so specific ethical approval was not required for this study.

We performed Pearson and linear weighted regression to compare the admissions for MS, latitude and smoking using MATLAB R2009a. Least-squared weighted regression was conducted based both on standard error estimates and a weighting value derived from NHS health board population numbers. χ2 tests were used to compare observed and expected hospital admissions throughout Scottish health boards.

Results

MS admissions by NHS health board

The overall number of patients admitted with MS (as an absolute value over the 13 year period and a yearly average) and indirect rate per 100,000 population are shown in table 1. Overall 11,094 individual patients were admitted with MS in Scotland between 1997 and 2009. This was equivalent to a yearly average of 853 and an indirect rate of 16.87 per 100,000 population. However, it was clear that the admission rates varied considerably between NHS health boards, with the highest rate in NHS Orkney Islands (37.38 [95% CI 25.40–49.37]) and the lowest in NHS Forth Valley (12.54 [95% CI 5.60–19.47]).

thumbnail

Table 1. Overall admissions data for MS in Scotland between 1997 and 2009.

doi:10.1371/journal.pone.0014606.t001

The data for male and female patients admitted with MS are shown in tables 2 and 3. The indirect rate for MS in females across Scotland was significantly higher than in males (males 10.69 vs. females 22.64, χ2 = 4.28, d.f. = 1, p = 0.04).

thumbnail

Table 2. Male only admissions data for MS in Scotland between 1997 and 2009.

doi:10.1371/journal.pone.0014606.t002
thumbnail

Table 3. Female only admissions data for MS in Scotland between 1997 and 2009.

doi:10.1371/journal.pone.0014606.t003

Correlation of MS admissions with latitude

Overall MS hospital admissions showed a strong correlation with latitude (figure 1a; r = 0.76, p = 0.001; r weighted by SE (rsw) = 0.75, p = 0.002; r weighted by population (rpw) = 0.71, p = 0.004). This held true when analysed separately by gender, although the male data was not significant when weighted by population (figure 1b; males – r = 0.77, p = 0.001; rsw = 0.74, p = 0.002; rpw = 0.52, p = 0.06; females – r = 0.77, p = 0.001; rsw = 0.76, p = 0.002; rpw = 0.76, p = 0.001).

thumbnail

Figure 1. Scatter plots of MS admission rates against latitude.

(A) Overall data. (B) Male (blue) and female (red) specific data.

doi:10.1371/journal.pone.0014606.g001

Correlation of MS admissions with smoking prevalence

There was a nominally significant inverse correlation with smoking but this was lost in the weighted regression (figure 2a; overall – r = −0.55, p = 0.04, rsw = −0.49, p = 0.08; rpw = −0.25, p = 0.39; figure 2b; males – r = −0.52, p = 0.06, rsw = −0.47, p = 0.09; rpw = −0.21, p = 0.47; females – r = −0.56, p = 0.04; rsw = −0.49, p = 0.08; rpw = −0.25, p = 0.38).

thumbnail

Figure 2. Scatter plots of MS admission rates against smoking prevalence.

(A) Overall data. (B) Male (blue) and female (red) specific data.

doi:10.1371/journal.pone.0014606.g002

Sex ratio

There was no significant correlation between the sex ratio and latitude in MS patient-linked admissions except when weighted by population (figure 3; r = 0.32, p = 0.27; rsw = 0.41, p = 0.15, rpw = 0.75, p = 0.002). There was no relationship between the sex ratio and smoking prevalence (r = 0.27, p = 0.36; rsw = 0.36, p = 0.21; rpw = 0.14, p = 0.64).

thumbnail

Figure 3. Scatter plots of female-to-male sex-ratio of MS admission rates against latitude.

doi:10.1371/journal.pone.0014606.g003

Other autoimmune and neurological diseases

There was no significant overall correlation between indirect admissions rate and latitude for several autoimmune diseases (data not shown; rheumatoid arthritis – rsw = 0.29, p = 0.39; Crohn's disease – rsw = 0.06, p = 0.81; type 1 diabetes mellitus – rsw = 0.26, p = 0.37) or a neurological disease (motor neuron disease – rsw = 0.26, p = 0.37). Indirect admissions rates for ulcerative colitis did show a nominally significant correlation with latitude (rsw = 0.57, p = 0.03) but this was lost after correction for multiple hypothesis testing.

Discussion

We present here the most up to date available map of the distribution of MS in Scotland. Our findings have shown that the latitude gradient of MS prevalence and incidence observed still exists in Scotland.[2] As suggested by previous work,[2] much of this latitude gradient is as a result of an increase in female MS risk at increasing latitudes. It must be noted, however, that, although our data does support a latitudinal gradient, the data itself could also argue for a “threshold” effect around 57°N. Until more is understood about the aetiological origin for this effect of latitude on MS prevalence, conclusions regarding this must remain tentative.

Smoking does not appear to underlie this latitudinal effect, especially given the apparent inverse correlation of smoking prevalence with latitude. Although there is some suggestion that the sex-ratio in MS admissions may be inversely related to smoking prevalence, this is quite unlikely to be a real effect given that where MS risk is highest (NHS Orkney Islands), smoking prevalence is lowest and so, regardless of the ratio between male and female smoking behaviour, it is difficult to envision how the absence of a susceptibility factor could result in a sex-specific increase of disease risk.[15] There are clear limitations in using background population measures of smoking prevalence as a surrogate for exposure in patients with MS. In particular, effects on the level of individual patients will not be detected in this sort of population-based data. Future studies should re-examine this question in relation to MS-specific smoking data.

Limitations of record linkage studies using routinely collected administrative data are well known, and include the facts that the data are limited to hospitalised patients and that information about some variables of potential interest, such as social circumstances and ethnicity, are generally unavailable. It is also worth noting that the sample size of regions primarily driving the latitudinal effect (e.g. the Orkney and Shetland Islands) and relatively small and so this effect will need to be confirmed in future work. The fact that latitudinal effects were not seen for other autoimmune and neurological diseases argues strongly against the gradient we observed being solely due to underlying trends in hospital services or background population admissions rates.

The environmental risk factor that most likely explains the latitudinal effect observed in this study is the effect of ultraviolet irradiation on vitamin D biosynthesis. Vitamin D deficiency has already been associated with risk of MS and controls the expression of HLA-DRB1*1501, the key genetic risk factor in MS.[10], [16] However, it is likely that MS prevalence will be explained only by considering a complex interplay of genetic and environmental risk factors, some of which may not yet have been identified [8].

The latitudinal variation in MS prevalence may have important implications for neurological service provision. For example, although the number of MS-relevant clinics per week correlates well with the total number of admissions as one would expect with the tendency of admissions to follow hospital resources (r = 0.69, p<0.01), these show a trend towards inverse correlation with MS indirect admissions rates, probably a better estimate of background disease prevalence (r = −0.52, p = 0.06) [17]. MS disease distribution should be considered when evaluating neurology services in Scotland in the future.

In conclusion, the distribution of MS across Scotland shows nearly three-fold variation, at least as measured by hospitalisation rates. The higher rates in the North than South of Scotland persist; they are consistent with earlier findings from studies in different parts of the UK; and they are consistent with the more general finding across the world that MS prevalence increases with increasing distance from the equator in both hemispheres. The high prevalence of MS in the North of Scotland should be recognised in funding of care for MS and for studies of disease prevention.

Acknowledgments

We would like to acknowledge all members of the Ebers group for helpful ideas and discussion.

Author Contributions

Conceived and designed the experiments: AEH GCE SVR. Analyzed the data: AEH LJ RM AF SVR. Wrote the paper: AEH LJ RM AF GCE SVR.

References

  1. 1. Noseworthy JH, Lucchinetti C, Rodriguez M, Weinshenker BG (2000) Multiple sclerosis. N Engl J Med 343: 938–952.
  2. 2. Alonso A, Hernán MA (2008) Temporal trends in the incidence of multiple sclerosis: a systematic review. Neurology 71: 129–135.
  3. 3. Staples J, Ponsonby A, Lim L (2010) Low maternal exposure to ultraviolet radiation in pregnancy, month of birth, and risk of multiple sclerosis in offspring: longitudinal analysis. BMJ (Online first).
  4. 4. Handel AE, Giovannoni G, Ebers GC, Ramagopalan SV (2010) Environmental factors and their timing in adult-onset multiple sclerosis. Nat Rev Neurol 6: 156–166.
  5. 5. Koch-Henriksen N, Sorensen PS (2010) The changing demographic pattern of multiple sclerosis epidemiology. Lancet Neurol 9: 520–532.
  6. 6. Ascherio A, Munger KL (2007) Environmental risk factors for multiple sclerosis. Part I: the role of infection. Ann Neurol 61: 288–299.
  7. 7. Ascherio A, Munger KL (2007) Environmental risk factors for multiple sclerosis. Part II: Noninfectious factors. Ann Neurol 61: 504–513.
  8. 8. Handel AE, Handunnetthi L, Giovannoni G, Ebers GC, Ramagopalan SV (2010) Genetic and environmental factors and the distribution of multiple sclerosis in Europe. Eur J Neurol.
  9. 9. Beretich B, Beretich T (2009) Explaining multiple sclerosis prevalence by ultraviolet exposure: a geospatial analysis. Mult Scler 15: 891–898.
  10. 10. Ramagopalan SV, Maugeri NJ, Handunnetthi L, Lincoln MR, Orton SM, et al. (2009) Expression of the multiple sclerosis-associated MHC class II Allele HLA-DRB1*1501 is regulated by vitamin D. PLoS Genet 5: e1000369.
  11. 11. Simon KC, van der Mei IA, Munger KL, Ponsonby A, Dickinson J, et al. (2010) Combined effects of smoking, anti-EBNA antibodies, and HLA-DRB1*1501 on multiple sclerosis risk. Neurology 74: 1365–1371.
  12. 12. Dean G, Goodall J, Downie A (1981) The prevalence of multiple sclerosis in the Outer Hebrides compared with north-east Scotland and the Orkney and Shetland Islands. J Epidemiol Community Health 35: 110–113.
  13. 13. Robertson N, Compston A (1995) Surveying multiple sclerosis in the United Kingdom. J Neurol Neurosurg Psychiatry 58: 2–6.
  14. 14. Whyte B, Gordon D, Haw S, Fischbacher C, Scott N, et al., editors. (2007) An atlas of tobacco smoking in Scotland: a report presenting estimated smoking prevalence and smoking-attributable deaths within Scotland: Health Scotland. 64 p.
  15. 15. Hernan MA, Olek MJ, Ascherio A (2001) Cigarette smoking and incidence of multiple sclerosis. Am J Epidemiol 154: 69–74.
  16. 16. Munger KL, Levin LI, Hollis BW, Howard NS, Ascherio A (2006) Serum 25-hydroxyvitamin D levels and risk of multiple sclerosis. JAMA 296: 2832–2838.
  17. 17. NHS Quality Improvement Scotland (2007) Review of services available to those with neurological conditions. Scott-Moncrieff 1–88.