Sex ratio of multiple sclerosis in Canada: a longitudinal study

Summary

Background

Incidence of multiple sclerosis is thought to be increasing, but this notion has been difficult to substantiate. In a longitudinal population-based dataset of patients with multiple sclerosis obtained over more than three decades, we did not show a difference in time to diagnosis by sex. We reasoned that if a sex-specific change in incidence was occurring, the female to male sex ratio would serve as a surrogate of incidence change.

Methods

Since environmental risk factors seem to act early in life, we calculated sex ratios by birth year in 27 074 Canadian patients with multiple sclerosis identified as part of a longitudinal population-based dataset.

Findings

The female to male sex ratio by year of birth has been increasing for at least 50 years and now exceeds 3·2:1 in Canada. Year of birth was a significant predictor for sex ratio (p<0·0001, χ²=124·4; rank correlation r=0·84).

Interpretation

The substantial increase in the female to male sex ratio in Canada seems to result from a disproportional increase in incidence of multiple sclerosis in women. This rapid change must have environmental origins even if it is associated with a gene–environment interaction, and implies that a large proportion of multiple sclerosis cases may be preventable in situ. Although the reasons why incidence of the disease is increasing are unknown, there are major implications for health-care provision because lifetime costs of multiple sclerosis exceed £1 million per case in the UK.

Introduction

There is a consensus in many countries that the incidence and prevalence of multiple sclerosis has been increasing. However, this change is often attributed to inconsistencies in ascertainment and epidemiological methods. Restudy of populations often identifies additional cases that were missed during the first study. Differing views have been difficult to resolve because occurrence of several simultaneous changes that affect perception of incidence have had potential to confound the results.

Prevalence and incidence of multiple sclerosis have varied in reported studies and are related, at least in part, to the geography of the disease.1, 2 Although there seems to be a general trend over the years towards increased incidence and prevalence, results in different countries have not been easy to compare; some regions have reported increased incidence,3, 4, 5, 6, 7, 8, 9, 10, 11, 12 whereas others have shown no change or decreased incidence.13, 14 Ascertainment and study design undoubtedly confound multiregion comparisons as well as differential increases in survival rates for prevalence. Ways of overcoming or reducing ascertainment bias include uniformly sampling the same population over time or establishing changes with an internal variable in the same study group. Showing that changes have taken place in a subset of patients enables use of an internal contemporaneous matched control group, and a candidate for such a subset is the measurement of sex ratio.

Studies from several countries, including the USA,¹⁰ Australia,⁹ and Japan,¹⁵ have shown the female to male sex ratio of multiple sclerosis cases to increase over time when serial cross-sectional comparisons were made. We have shown that the rate of disease has been increasing in Canada. A heightened risk for later born female children was seen in very large pedigrees with sibships often spanning two decades or more in birth timing.¹⁶ Several lines of evidence, especially migration studies,17, 18, 19, 20, 21 have pointed toward determination of risk for the disease in early life. With the clear recognition that environmental factors in the disease act at a broad population level we have been studying the timing of this effect in Canada, albeit through approaches that are necessarily indirect. A month-of-birth effect with a May peak and a November nadir in risk,²² a maternal parent-of-origin effect in half-siblings,²³ and the finding of an excess concordance and age of onset correlation for dizygotic twins compared with siblings (Ebers G, Sadovnick A, unpublished) imply that gestational timing might be a factor.

With these considerations in mind, we sought to assess whether the female to male ratio of patients with multiple sclerosis in Canada has changed over the years in a population-based sample stratified by year of birth.