Epidemiology of Myasthenia Gravis in Sweden 2006–2016

Abstract Introduction Reported incidence and prevalence rates of Myasthenia Gravis (MG) vary widely and are assumed to have increased over the last few decades. We conducted a nationwide register‐based study on the current incidence and prevalence of MG and MG subgroups in Sweden. Methods Data were acquired from four Swedish Health Registers in order to identify patients with MG. Incidence and prevalence rates were calculated for the years 2006–2016, using population numbers provided by Statistics Sweden. Results In 2016, the incidence of MG in Sweden was 2.9 per 100,000 inhabitants (95% CI: 2.5–3.2/100,000) and the crude prevalence was 36.1 per 100,000 inhabitants (95% CI: 34.9–37.3). There was a significant increase in Myasthenia Gravis prevalence from 2006 to 2016. Prevalence rates of all MG subgroups but thymoma‐associated MG increased over the same period of time. Conclusions The incidence and prevalence of Myasthenia Gravis have increased over time in Sweden, and the rates are high in comparison with other countries.

There is a trend where older epidemiological studies on MG tend to report the lower incidence and prevalence rates (Carr et al., 2010;Phillips, 2003), and increasing incidence rates over time have been shown in particular regarding late-onset MG (Matsuda et al., 2005;Pakzad, Aziz, & Oger, 2011;Vincent, Clover, Buckley, Grimley Evans, & Rothwell, 2003). These increasing numbers may be partly explained by improved diagnostic techniques as well as emerging research-driven knowledge about the disease. Improved treatment regimens as well as an aging population are other likely contributing factors. Nevertheless, true regional differences do exist, especially with regard to certain MG subgroups, for example, MG with antibodies against muscle-specific tyrosine kinase (MuSK + MG), which is more common in Southern Europe than in Northern Europe (Niks, Kuks, & Verschuuren, 2007;Tsiamalos, Kordas, Kokla, Poulas, & Tzartos, 2009).
Accurate epidemiological information is important as it contributes to the understanding of disease mechanisms and provides a framework for planning the effective distribution of healthcare resources. Furthermore, knowledge of geographical variations is important when using results of MG studies from other countries in a clinical setting.
The epidemiology of MG in Sweden has traditionally been a sparsely studied subject, and reported prevalence rates vary markedly. Kalb et al reported MG prevalence to be 14.1 per 100,000 inhabitants in the county of Stockholm 1998 (Kalb, Matell, Pirskanen, & Lambe, 2002). Fang et al reported an MG prevalence of 24.8 per 100,000 inhabitants in Sweden 2010 (Fang et al., 2015), and we reported an MG prevalence rate of 19.9 per 100,000 inhabitants in Jönköping county 2014 (Westerberg, Landtblom, & Punga, 2018

| Study design and ethical approval
This is a Swedish, nationwide register-based cohort study approved by

| Study population
Patients with (a) classification codes for MG; ICD-9 code 358A or ICD-10 code G70.0, in the NPR and/or (b)

| MG subgroups
Myasthenia Gravis onset was defined as the date of first MG diagnosis in the registers or the first date of prescription of pyridostigmine or ambenonium-whichever occurred first.
The MG cohort was divided into four subgroups:

| Incidence and prevalence rates
Annual population statistics on age and gender was obtained for Sweden and the 25 counties from Statistics Sweden (https://www. scb.se/en/findi ng-stati stics/ stati stics-by-subje ct-area/popul ation/) for the years 2006-2016.

| Statistical analysis
The crude incidence and prevalence rates were calculated using denominators derived from Statistics Sweden from the years 2006 to 2016. Incidence and prevalence rates were calculated per 100,000 inhabitants and 95% confidence intervals (95% CI) were calculated using Poisson distribution. Student's t test (for parametric data) and Mann-Whitney U test (for non-parametric data) were used for analyzing continuous variables. A p-value < .05 was considered significant. The statistical analyses were performed in GraphPad Prism version 6.0h for Mac (Graphpad software, www.graph pad.com).

| Incidence
In total, there were 2,598 (1, As shown in Figure 1a, MG incidence rates in Sweden 2006-2016 varied between 2.1 and 2.9/100,000 per year (average 2.5/100,000), with a tendency toward higher incidence rates during the later years.
Subgroup incidences (Table 1) varied from year to year with no consistent pattern over time, although there was a slight increase in LOMG incidence rates from 2006 to 2016.

| Regional incidences
There was no clear or consistent difference in incidence rates between different Swedish regions over the years, although there was an increase in Sweden as a whole from 2006 to 2016. When comparing the three largest regions (Malmö, Gothenburg, and Stockholm; each with inhabitants of more than one million), there were in general slightly higher incidence rates in the southernmost region

| Gender incidences
Gender incidence rates for all MG and subgroups are shown in Table 1. Female incidences were consistently higher in EOMG over the years. Male incidences were slightly higher in LOMG.

| Prevalence
The prevalence of MG increased over time from 2006 to 2016 in Sweden as well as regionally within Sweden (Figure 3)

| Regional prevalences
The prevalence rates differed between regions; however, as some regions are very small, a difference of one single patient could cause striking differences in prevalence, and a comparison between all regions would therefore not be meaningful. There was a difference in average prevalence rates between the three largest regions (with populations over one million) with higher AllMG, EOMG, JOMG, and TOMG prevalence rates in the Malmö region than in the Gothenburg and Stockholm regions and higher LOMG prevalence rates in the Gothenburg than in the Stockholm region.
The prevalence rates in these regions 2006-2016 are shown in

| Gender prevalence
The female: male ratio for MG patients in Sweden was fairly stable over the years both for AllMG and in the subgroups, with the highest ratio of women in the EOMG subgroup (71%-73%). There was a slight tendency toward lower proportions of females with AllMG in When comparing the three largest regions, there was in general a higher proportion of women with MG in these regions than in all Sweden, especially in the southernmost region (Malmö; Figure 5).
Furthermore, there was a slightly higher proportion of women in both the LOMG and the EOMG subgroups in Malmö compared to Stockholm and Gothenburg.

| D ISCUSS I ON
This is the first nationwide study of MG incidence in Sweden. It is also one of few studies to document incidence and prevalence of different MG subgroups within the same population. Fang et al. (2015) have previously reported the prevalence of MG in Sweden in 2010, but nationwide prevalence rates over time have not been previously described.
Here, we report a Swedish incidence of MG of 2.9 per 100,000 in 2016, which is a rate similar to what has been reported in other recent studies (Aragones et al., 2014;Breiner et al., 2016;Gattellari, Goumas, & Worthington, 2012;Lai & Tseng, 2010). Furthermore, we report a current nationwide crude prevalence of 36.1 per 100,000, which is among the higher prevalence rates reported, however still in agreement with some other studies (Aragones et al., 2017;Breiner et al., 2016). This prevalence is also much higher than the reported MG prevalence of 14.1 in 100,000 in Stockholm, Sweden in 1998.
Accordingly, we found a significant increase in prevalence over the study period (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016), which is in line with a reported rise in regional prevalence rates of MG over time (Breiner et al., 2016;Carr et al., 2010;Holtsema et al., 2000;Lai & Tseng, 2010;Pallaver et al., 2011). although it can also lead to a higher risk of over-diagnosis. As the method to identify the patients was identical throughout the study and the total time-span was only 11 years, it is doubtful that increased disease awareness had a serious impact. There was also no significant increase in incidence, which would be an expected first sign of an improved diagnostic set-up.
Improved documentation and coverage by the national registers could lead to increasing prevalence figures over time; however, a parallel increase in incidence measures would similarly be expected here.
When the three largest health regions of Sweden were compared, we noted higher incidence and prevalence rates in Malmö, for all subgroups except LOMG. There are no previous descriptions of regional variations in MG within Sweden and the cause of the variations we saw in this study is unclear. Possibly, socioeconomic factors, The study has some limitations. As the register-based methodology has the advantage of a supposed almost total nationwide coverage, there are disadvantages to the accuracy of the MG diagnosis. Despite using a validated MG identification method in this study (Fang et al., 2015), there is a risk of including individuals with a false diagnosis, as we did not have access to individual patient charts and therefore could not confirm that the diagnostic criteria of the Myasthenia Gravis Foundation of America were met. This might to some extent lead to an overestimation of the incidence and prevalence rates, while the increase in the rates over the years is probably reliable. Furthermore, we chose a pragmatic subgrouping method, limited to age at onset and the presence of a thymoma or not. It would have been desirable to consider antibody status and distribution of symptoms to obtain a more precise subgrouping; however, these data are unavailable in currently obtainable health registers.
As MG is uncommon, comparisons of smaller regions become statistically unreliable, which is why we chose to compare only the largest health regions. Another option would have been to cluster the smaller areas into larger ones for comparison. However, we chose to exclude them from the comparative analysis as we could not logically justify the clustering method.
Despite the above limitations, we believe that this study represents important and reliable information on Swedish MG epidemiology and that the reported rising prevalence rates described also in other countries indicate a need to improve awareness of MG among healthcare providers.

| CON CLUS ION
Over the time period of 2006-2016, incidence and prevalence rates of Myasthenia Gravis have increased in Sweden, and the prevalence rate of 36.1 per 100,000 in 2016 is high in comparison with many other countries. Yet, there seem to be regional differences within the country.