ReviewThe changing demographic pattern of multiple sclerosis epidemiology
Introduction
Multiple sclerosis (MS) is a demyelinating immune-mediated disease of the CNS, and is the leading cause of disability in young and middle-aged people in the developed world. It is also one of the best studied neurological diseases in terms of epidemiology. The most striking epidemiological characteristic is the apparent uneven distribution of the disease across the world. The traditional view, based on many early studies and reviews, is that MS is particularly prevalent where white people of Nordic origin live, in temperate zones, and in high-income countries. Conversely, MS is uncommon where non-whites live, in low-income countries, and in tropical zones. Thus, three factors are in effect: population genetics; the interplay between genes and geographically determined physical environment; and socioeconomic structure, including availability of medical facilities.1 Clearly, these factors that are not easy to study separately.
The idea of a single and dominant environmental cause of MS has largely been abandoned, although there are prominent proponents of an infectious viral cause.2 Although strong associations between HLA alleles and MS have been known for decades,3 until recently the standpoint had been that, although many genes seem to affect the risk of MS, genome screenings had failed to detect a single major locus.4 However, new research indicates a strong effect of certain MHC type II genes, which is masked by other genes via epistasis.5, 6 Shared environmental exposure is unlikely to be the cause of increased risk of MS in close biological relatives of patients with MS, because the risk of MS among non-biological relatives (adoptees) closely matches that of the background population.7 Concordance in identical twins is about 25%,8, 9, 10 far higher than that in dizygotic twins. The lack of full concordance in identical twins could at first be taken as proof of environmental effects; however, it has been suggested that random factors can affect gene expression, based on observations in single cells.11, 12 Moreover, twins share their environment in utero and in childhood, and attempts to separate discordant monozygotic twins in terms of differential exposure have failed.13, 14
If a population gene pool is stable, apart from adaptations caused by migration and selection, any short-term marked changes in incidence of MS will add weight to the notion of aetiological environmental factors and will stimulate hypotheses formulation. This was the case with the observations of MS occurrence in the Faroe Islands in the North Atlantic. The islands, presumed to be MS naive until World War II, experienced a near epidemic of MS from 1943 onwards.15, 16 The investigators suggested that MS was a rare, late result of a single, ubiquitous, systemic infection by a specific agent, carried by British troops, who occupied the islands during the war, and to which the Faroese population hitherto had been naive. Others have questioned this view,17 which depends on the assumption that there were no cases of MS in the Faroe Islands before World War II.
Prevalence (number of patients with MS alive at a specific date per 100 000 population) can be indicative of several other factors besides the true frequency of MS, including diagnostic accuracy and ascertainment probability, both of which are connected with the level of medical services in the country. The latter also accounts for another important component of prevalence: survival time. These factors have improved in most parts of the world throughout the past five decades.
The incidence rate (number of new cases per 100 000 population per year) is a better measure of the risk of MS, because it is independent of survival time and is not affected by insufficient ascertainment probability before the incidence period in question. Furthermore, it indicates changes in population risk sooner and more directly than does prevalence. The observed numbers included in incidence studies are usually smaller than in prevalence studies, resulting in statistically uncertain estimates, particularly when small time-space units have been surveyed.
Because incidence of MS peaks at about 30 years of age, and prevalence peaks at about 50 years of age, the crude incidence rates also depend on the age distribution of the population. In recent decades, adjustment of rates by standardisation to a hypothetical standard population has been more common in incidence and prevalence studies. Zivadinov and colleagues18 analysed published incidence and prevalence studies, in which data sufficiently detailed to calculate standardised rates were given. They found that, after standardisation, the latitudinal effect on adjusted prevalence weakened and the effect on incidence disappeared, which seriously challenged the presence of a latitudinal effect in the northern hemisphere within the range studied (40–60°N). Serious concerns about the validity of the diagnosis in prevalence studies have thus been raised.19
In this Review, we focus on the changes in prevalence, incidence, and sex ratio over the past few decades. We discuss whether the rising incidence rates represent a true increase in the incidence of MS, why the growth in incidence in some surveys is apparently confined to women, and we present evidence that challenges the widely accepted notion of a latitudinal gradient in the northern hemisphere.
Section snippets
Search strategy and selection criteria
From PubMed, we retrieved 1877 articles written in English and published since 1965, by use of the terms “multiple sclerosis” and “incidence” or “prevalence” in the title or abstract, and supplemented this search with articles from the authors' personal archives. The abstracts were read, and 226 original articles in which the words “incidence” or “prevalence” referred to the frequency of the disease were selected, avoiding double registration if more than one paper referred to the same survey.
Geographical distribution
Systematic epidemiological surveys were rare until the mid 20th century. However, in 1922, Davenport described a higher frequency of MS in men drafted to the US Army from northern states, and noted the high risk in people of Scandinavian ancestry, and thus suggested that there was a latitudinal effect as well as a possible racial effect.20 Conclusive contributions and reviews of the distribution of MS in the latter half of the 20th century have been published by Kurtzke.21, 22, 23 He divided
Time effect
Prevalence invariably increases with repeated surveys in the same area, for several reasons. The first survey might have missed old cases in particular, diagnosed years or decades earlier, but as new cases accumulate with time, the effect of missing old cases will become diluted and less important, and new data will enable approximation of the true prevalence. Moreover, the first surveys might have stimulated attention to the disease in the region studied. Prevalence and incidence estimates
Sex ratio
At the beginning of the 20th century, MS was thought to be a predominantly male disease.206 However, this might not have indicated the true situation at that time. Men were more important for the family as breadwinners and hence probably more likely to receive a diagnosis if affected by MS. Since then, and long after considerations for such an effect have become obsolete, there has been a steady increase in the incidence of MS in women. In a Canadian study207 that comprised all registered MS
True increase in incidence or better ascertainment and more rapid diagnosis?
Three factors could have increased the probability of an early diagnosis over the past few decades. First, MRI is now used as standard whenever MS or clinically isolated syndrome is suspected, often with a low clinical threshold. Second, availability of disease-modifying treatments makes an early diagnosis relevant, as opposed to the reluctance in earlier times to make and communicate a diagnosis in younger people with few symptoms and no current disability. Third, the more sensitive McDonald
Conclusions
The most striking changes in demographic epidemiology of MS, ascertained over the past decades, are (1) an increase in prevalence of MS, predominantly due to longer survival; (2) a probable true increase in incidence of MS in many places, particularly in women, leading to higher male to female sex ratios of MS; and (3) a dismissal of the notion of a latitudinal gradient in Europe and North America. The increase in the female:male sex ratio strongly indicates the existence of an environmental
References (214)
- et al.
Histocompatability determinants in multiple sclerosis with special reference to the clinical course
Lancet
(1973) - et al.
Genetics of multiple sclerosis
Lancet Neurol
(2004) Environmental factors and multiple sclerosis
Lancet Neurol
(2008)- et al.
A prevalence study of multiple sclerosis in L'Aquila, central Italy
Clin Neurol Neurosurg
(1988) - et al.
Trends in prevalence and incidence of multiple sclerosis in Bajo Aragon, Spain
J Neurol Sci
(2003) - et al.
Bayesian mapping of multiple sclerosis prevalence in the province of Pavia, northern Italy
J Neurol Sci
(2006) Medical facilities and the prevalence of multiple sclerosis
Acta Neurol Scand
(1965)Epidemiologic evidence for multiple sclerosis as an infection
Clin Microbiol Rev
(1993)- et al.
Epistasis among HLA-DRB1, HLA-DQA1, and HLA-DQB1 loci determines multiple sclerosis susceptibility
Proc Natl Acad Sci USA
(2009) - et al.
Epistasis: multiple sclerosis and the major histocompatibility complex
Neurology
(2009)