doi:10.1093/brain/awq134
Brain 2010: 133; 1889–1899
| 1889
BRAIN
A JOURNAL OF NEUROLOGY
Geographic variations of multiple sclerosis
in France
Agnes Fromont,1,2 Christine Binquet,3 Erik A. Sauleau,4 Isabelle Fournel,3 Audrey Bellisario,3
Johan Adnet,3 Alain Weill,5 Sandra Vukusic,6 Christian Confavreux,6 Marc Debouverie,7
Laurence Clerc,8 Claire Bonithon-Kopp3 and Thibault Moreau1,2
1 Department of Neurology, University Hospital of Dijon, 3 rue du Faubourg Raines, 21000 Dijon, France
2 Centre of Epidemiology of the Populations, Burgundy University, EA 4184, 1 Boulevard Jeanne d’Arc, BP 1542, 21079 Dijon, France
3 INSERM, CIE1, Dijon, F-21000, France; Clinical Investigation Centre—Clinical Epidemiology/Clinical Trials, Dijon University Hospital, Dijon,
F-21000, France; Burgundy University, Dijon, F-21000, France
4 Biostatistics and Methodology Unit, Public Health Department, Strasbourg University Hospitals, 1 Place de l’Hôpital BP 426, 67091 Strasbourg,
France
5 French Public Health Insurance, 75986 Paris cedex 20, France
6 Department of Neurology, University Hospital Pierre Wertheimer, 59 Boulevard Pinel, Hospices civils de Lyon, 69677 Bron, France
7 Nancy University, Metz University Paul Verlaine, Paris University Descartes, EA 4360 Apemac, 54000 Nancy, France
8 French Public Health Insurance, Direction of the Medical Department, 21000 Dijon, France
Correspondence to: Prof. Thibault Moreau,
Department of Neurology,
3 rue du Faubourg Raines,
21000 Dijon, France
E-mail: [email protected]
France is located in an area with a medium to high prevalence of multiple sclerosis, where its epidemiology is not well known.
We estimated the national and regional prevalence of multiple sclerosis in France on 31 October 2004 and the incidence
between 31 October 2003 and 31 October 2004 based on data from the main French health insurance system: the Caisse
Nationale d’Assurance Maladie des Travailleurs Salariés. The Caisse Nationale d’Assurance Maladie des Travailleurs Salariés
insures 87% of the French population. We analysed geographic variations in the prevalence and incidence of multiple sclerosis
in France using the Bayesian approach. On the 31 October 2004, 49 417 people were registered with multiple sclerosis out of
the 52 359 912 insured with the Caisse Nationale d’Assurance Maladie des Travailleurs Salariés. Among these, 4497 were new
multiple sclerosis cases declared between 31 October 2003 and 31 October 2004. After standardization for age, total multiple
sclerosis prevalence in France was 94.7 per 100 000 (94.3–95.1); 130.5 (129.8–131.2) in females and 54.8 (54.4–55.3) in males.
The national incidence of multiple sclerosis between 31 October 2003 and 31 October 2004 was 7.5 per 100 000 (7.3–7.6);
10.4 (10.2–10.6) in females and 4.2 (4.0–4.3) in males. The prevalence and incidence of multiple sclerosis were higher in NorthEastern France, but there was no obvious North–South gradient. This study is the first performed among a representative
population of France (87%) using the same method throughout. The Bayesian approach, which takes into account spatial
heterogeneity among geographical units and spatial autocorrelation, did not confirm the existence of a prevalence gradient
but only a higher prevalence of multiple sclerosis in North-Eastern France and a lower prevalence of multiple sclerosis in
the Paris area and on the Mediterranean coast.
Received November 21, 2009. Revised April 26, 2010. Accepted April 27, 2010. Advance Access publication June 15, 2010
ß The Author (2010). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.
For Permissions, please email: [email protected]
1890
| Brain 2010: 133; 1889–1899
A. Fromont et al.
Keywords: multiple sclerosis; prevalence; incidence; epidemiology
Abbreviations: ALD = affection de longue durée; CNAMTS = Caisse Nationale d’Assurance Maladie des Travailleurs Salariés
Introduction
Multiple sclerosis prevalence is heterogeneous worldwide. Multiple
sclerosis is known to be more prevalent in temperate than in
tropical regions. In Europe, there are dramatic differences between
its prevalence in Northern and Southern countries; prevalence
varies from 224 per 100 000 inhabitants in Orkney and the
Shetland Islands, to 40–90 per 100 000 in Italy (Rosati, 2001).
These variations may reflect a combination of environmental
conditions, cultural and dietary habits as well as genetic particularities. This increase in prevalence with increasing latitude has been
called ‘the latitudinal gradient’ (Kurtzke, 1975). In Europe, a
North–South prevalence gradient has also been described within
a number of countries including Ireland (McGuigan et al., 2004),
the UK (Williams and McKeran, 1986; Phadke and Downie, 1987;
Roberts et al., 1991; Forbes and Swingler, 1999) and Germany
(Lauer et al., 1984), with a higher prevalence in the Northern
regions of these countries. However, these gradients are a subject
of debate because they could be due to methodological artefacts.
Indeed they are often highlighted by studies comparing some specific regions of a country that are not representative of the country
as a whole (McGuigan et al., 2004), or even hospital cohorts,
sometimes with data collected at different periods of the year.
Moreover, most of these studies do not account for age and
gender distribution in the areas compared. Adjusting for these
factors and using adequate Bayesian approaches would improve
the clarity of the geographical variations in multiple sclerosis
prevalence (Besag et al., 1991; Banerjee et al., 2004; Lang and
Brezger, 2004).
France is situated in the middle of Western Europe and is considered a country of medium to high multiple sclerosis prevalence
(Debouverie et al., 2007b). At present, the prevalence of multiple
sclerosis in France is estimated to be around 50 cases for 100 000
inhabitants (Debouverie et al., 2007b). However, this estimation
relies on data collected at a regional level or on non-representative
samples. The method for estimating the prevalence of multiple
sclerosis needs to be more clearly defined. To our knowledge
only one study has been performed at the national level
(Vukusic et al., 2007). However, this study was based on a
sample of 7% of the French population, insured by the Caisse
Collective de la Mutualité Sociale Agricole, which insures
mostly farmers, agricultural employees and their families. This
study brought to light a North-East/South-West gradient.
However, as the population in this study was not representative,
the existence of this gradient in the population at large needs to
be confirmed. Thus, in order to obtain a better estimate of
multiple sclerosis prevalence in France as well as its geographic
variations, we initiated a study at the national level based on
data from the main French National Health Insurance system,
which insures 87% of the French population. This is the first
large-scale study using the same methodology at the national
level.
Materials and methods
Setting
This cross-sectional study was conducted in the French metropolitan
territory (including Corsica) located between latitudes 42 and
51 north. There were 60 028 292 inhabitants in France in 2004
(www.insee.fr). France covers 535 280 km2 divided into 96 French
administrative areas called ‘départements’ (Fig. 1). These départements
are grouped into 22 regions (Fig. 1). The population density of these
French administrative areas varies greatly (from 15 inhabitants/km2 in
Lozère to 20 569 inhabitants/km2 in the Paris administrative area),
as does age distribution.
Data sources
In France, multiple sclerosis is one of the 30 long-term illnesses
(affections de longue durée; ALDs) for which patients are covered
for 100% of their health care costs. Once the diagnosis of multiple
sclerosis has been established by a neurologist, a detailed request is
addressed to the health insurance system by the neurologist to obtain
ALD status for the patient. This status is authorized if the patient
needs disease-modifying treatment and/or if permanent disability
exists and the patient needs symptomatic management. All the data
corresponding to the ALD request and to the subsequent health care
are systematically collected by the French National Health Insurance
System covering 96% of the French population that includes three
main independent subsystems: the Caisse Collective de la Mutualité
Sociale Agricole (mentioned above), which covers farmers and their
families (7% of the French population); the Régime Social des
independents, which provides coverage to self-employed professionals
and their families (2% of the French population); and the Caisse
Nationale d’Assurance Maladie des Travailleurs Salaries (CNAMTS),
which insures private salaried employees, civil and non-civil servants
and their families. The CNAMTS covers 87% of the French population.
To reach our objectives, we used data collected by the CNAMTS
from patients with multiple sclerosis, residing in metropolitan France
(including Corsica) alive at 31 October 2004 and classified as having
ALD status for multiple sclerosis.
Data obtained from the CNAMTS for patients with multiple sclerosis
included in our study were: date of birth, gender, French administrative area of residence, date of request for ALD status related to
multiple sclerosis, date of death and new ALD status for multiple
sclerosis granted between 31 October 2004 and 31 October 2005,
and between 31 October 2005 and 31 October 2006.
The number of insured people in 2004 stratified in 5-year age
groups and according to gender for each French administrative area
was obtained from the CNAMTS. In accordance with French law, this
study was approved by the National Commission of Data Processing
and Civil Liberties.
Statistical analysis
Prevalence
The overall prevalence of multiple sclerosis in France according to ALD
status for the disease was defined as the number of patients with
Epidemiology of multiple sclerosis in France
Brain 2010: 133; 1889–1899
| 1891
Figure 1 Administrative divisions of France (regions and départements).
multiple sclerosis and ALD status identified in the CNAMTS database,
divided by the number of subjects insured by the CNAMTS.
In order to study the geographical variations of multiple sclerosis
prevalence, we calculated the standardized prevalence and the standardized prevalence ratio for each of the 96 départements individually
and grouped in 22 regions, and their 95% confidence intervals (CI).
The same calculation was made for the 22 regions of France. These
standardized prevalence ratios were estimated using a similar approach
to that used to estimate standardized incidence ratios (Breslow and
Day, 1980). The standardized prevalence ratio was defined as the
ratio between the observed number of multiple sclerosis subjects according to the CNAMTS database and the expected number. In each
French administrative area, the expected number was estimated by
multiplying the gender- and age-specific prevalence of the overall
CNAMTS population by the number of people in each age and
gender stratum of the area.
Incidence
The overall incidence of multiple sclerosis in France according to ALD
status for the disease between 31 October 2003 and 31 October 2004
was defined as the number of new ALD status granted for multiple sclerosis during this period divided by the number of subjects insured by the
CNAMTS. We calculated the standardized incidence and the standardized incidence ratio for each of the 96 French administrative areas and
their 95% CI using the same strategy as that applied for the calculation
of the standardized prevalence ratio. The same calculation was made
for the 22 regions of France. We calculated the incidence of new ALD
status for multiple sclerosis granted between 31 October 2004 and
31 October 2005, and between 31 October 2005 and 31 October 2006.
Spatial model
See Supplementary material for full details.
1892
| Brain 2010: 133; 1889–1899
We tested for a North-East/South-West gradient in multiple sclerosis
prevalence. We used the standardized prevalence ratio coefficient
of variation, obtained by dividing the standard deviation by the
mean of the 96 standardized prevalence ratios corresponding to
each administrative area. However, the large differences in population
size between areas may have induced variations in the accuracy of
standardized prevalence ratio from one département to another
making it difficult to distinguish random fluctuations from true
variations in standardized prevalence ratio. Thus, we used the
Potthoff–Whittinghill test (Potthoff and Whittinghill, 1966) in order
to determine the extent of these discrepancies. This test is used to
check geographical heterogeneity among all the observed number
of multiple sclerosis cases, while taking into account differences in
population size. Moreover, we had to take into account the fact
that contiguous areas may not be independent as they probably
share more risk factors than more distant administrative areas. To
detect the existence of this phenomenon, called spatial correlation,
we applied the Moran test (Cliff and Ord, 1981; Upton and
Fingleton, 1985).
Because results from the preceding tests were significant, we
applied a hierarchical Bayesian Poisson model (Olsen et al., 1996;
Maiti et al., 1998; Pascutto et al., 2000). This model allowed us to
estimate the relative prevalence (or relative risks) in département ‘i’
compared with the mean value for the population as a whole, while
accounting for spatial autocorrelation and adjusting for gender
and age. Posterior distributions of relative risks were obtained
by Monte Carlo Markov chain sampling techniques. More or less
informative parameters for hyperpriors were compared and the
model leading to the most conservative results was retained
(Colonna, 2006). We tested the interaction between the geographical
area and gender in order to highlight a potential difference in
geographical distribution of cases according to gender. Models were
compared using the deviance information criterion, a Bayesian version
of the Akaike criterion (Spiegelhalter et al., 2002). The same methodology was applied to study geographical variations in multiple sclerosis
incidence.
Disease mapping
Maps of standardized prevalence ratios, standardized incidence ratios
and smoothed relative risks at the level of the French administrative
area were established to visualize the geographical distribution of
multiple sclerosis prevalence. Standardized prevalence ratios and standardized incidence ratios of the 96 French administrative areas were
classified into three groups according to their value and significance:
(i) French administrative areas with a prevalence or incidence significantly lower than the mean national level; (ii) areas with a prevalence
or incidence no different from the mean national level; and (iii) areas
with a prevalence or incidence significantly higher than the mean
national level. Smoothed relative risk estimates were also classified
into three groups according to their posterior distribution with respect
to the number 1: (i) 95% of the posterior distribution below 1 indicating a lower prevalence or incidence than the mean national level; (ii)
95% of the posterior distribution above one corresponding to a higher
prevalence or incidence than the mean national level; and (iii) the
other cases.
We used R 9.2.1 (http://cran.univ-lyon1.fr/) for calculating Moran
and Potthoff–Whittinghill tests, Philcarto V5.05 (http://philcarto.free
.fr) for mapping, SAS for preparing the data for Bayesian analyses
and finally BayesX (http://www.stat.uni-muenchen.de/bayesX) for fitting Bayesian models.
A. Fromont et al.
Results
Description of the population
On the 31 October 2004, the CNAMTS insured 52 359 912 out of
the 60 028 292 inhabitants of France (87%). Among inhabitants
insured by the CNAMTS, 49 417 patients with multiple sclerosis
were registered as having ALD status for their disease. The sex
ratio (female to male) was 2.6. The distribution of ages at which
ALD status for multiple sclerosis was granted is shown in Table 1.
The mean age at the time of request for ALD status for multiple
sclerosis was 40.1 11.7 years. The years when ALD status was
first granted for multiple sclerosis are shown in Table 1. The first
ALD status allocated to a patient still alive in 2004 was requested
in 1987.
Prevalence
The national prevalence of multiple sclerosis in France was
94.7 per 100 000 patients covered by the CNAMTS (95% CI:
94.3–95.1), 130.5 per 100 000 in females (95% CI:
129.8–131.2) and 54.8 per 100 000 in males (95% CI: 54.4–55.3).
Standardized multiple sclerosis prevalence by region is shown in
Table 2.
When considering départements, the standardized prevalence
ratio of ALD multiple sclerosis among people covered by the
CNAMTS in France varied from a minimum of 0.73 (95% CI:
0.58–0.92) in Corse du Sud (a département in the Southern
area of France) to a maximum of 1.75 (95% CI: 1.51–2.01) in
Territoire de Belfort (in the Eastern area). The same trends were
found for both genders with a standardized prevalence ratio
varying from 0.73 (95% CI: 0.55–0.95) in Corse du Sud to 1.90
(95% CI: 1.61–2.22) in Territoire de Belfort among females; and
from 0.56 (95% CI: 0.39–0.77) in Ardèche to 1.53 (95% CI:
1.25–1.87) in Ardennes among males. The map of standardized
prevalence ratios showed a higher prevalence of ALD multiple
sclerosis in North-Eastern France for the overall population and a
lower one in the South-West (Fig. 2). The same trends were
observed when the data were stratified according to gender
(data not shown).
In our study, the P-value of Potthoff–Whittinghill test was equal
to 0.01, showing a statistically significant spatial heterogeneity.
The Moran test was of 0.3698 (P510 4) leading us to use a
model taking into account spatial autocorrelation. The results of
these two tests led us to use a Bayesian model. This model fitted
the data rather well as the relative difference between the
standardized prevalence and its Bayesian estimation was below
20% for almost 85% of départements. When using the
Bayesian approach, the North-Eastern part of France also
seemed to have a higher prevalence than the rest of France.
The smoothed relative risks varied from 1.22 (95% Bayesian
credible interval: 1.02–1.49) in Meurthe et Moselle, to 1.45
(95% Bayesian credible interval: 1.15–1.86) in Haute Marne.
Two main regions seemed to have a lower risk for the overall
population: Ile de France [e.g. 0.84 (95% Bayesian credible interval: 0.70–0.99) in Essonne] and a part of the Mediterranean coast
Epidemiology of multiple sclerosis in France
Brain 2010: 133; 1889–1899
| 1893
Table 1 Characteristics of patients with multiple sclerosis with chronic disease status (ALD) granted by the French
National Health Insurance System
Date of first multiple sclerosis ALD request
1987–95
1995–2001
2001–04
Age at first multiple sclerosis ALD request (years)
1–19
20–29
30–39
40–49
50–59
60
Invalidity
At least one act of physical therapy between 2003 and 2004
Total
N = 49 417
N (%)
Females
N = 35 833 (72.5%)
N (%)
Males
N = 13 584 (27.5%)
N (%)
18 698 (37.8)
18 774 (38.0)
11 946 (24.2)
13 517 (37.7)
13 556 (37.8)
8760 (24.5)
5180 (38.1)
5218 (38.4)
3186 (23.5)
840
8345
15 066
14 156
7799
3212
12 509
19 480
601
5977
10 912
10 238
5645
2460
8334
14 371
238
2368
4154
3918
2154
752
4175
5109
(1.7)
(16.9)
(30.5)
(28.7)
(15.8)
(6.5)
(25.3)
(39.4)
(1.7)
(16.7)
(30.5)
(28.6)
(15.8)
(6.9)
(23.3)
(40.1)
(1.8)
(17.4)
(30.6)
(28.8)
(15.9)
(5.5)
(30.7)
(37.6)
Table 2 Prevalence of multiple sclerosis ALD in each region of France in 2004
Overall
Administrative
region of
France
Affiliation
to the
CNAMTS
N (%)a
Lorraine
Champagne Ardenne
Picardie
Bourgogne
Franche Comte
Nord pas de Calais
Alsace
Centre
Auvergne
Basse Normandie
Haute Normandie
Bretagne
Limousin
Aquitaine
Midi Pyrénées
Rhône Alpes
Pays de Loire
Poitou Charente
Ile de France
PACA
Languedoc Roussillon
Corse
Total France
2 008 606
1 107 627
1 614 238
1 330 427
973 046
3 627 506
1 597 543
2 077 615
1 065 898
1 179 272
1 619 807
2 381 323
556 268
2 416 622
2 158 845
5 184 135
2 837 379
1 319 698
10 935 798
4 176 207
1 981 166
210 886
52 359 912
(86.3)
(83.2)
(86.1)
(82.4)
(86.1)
(90.2)
(89.3)
(83.9)
(80.5)
(81.8)
(89.9)
(79.4)
(78.1)
(79.6)
(80.8)
(88.4)
(85.1)
(78.5)
(97.3)
(89.8)
(81.1)
(77.9)
(87.2)
Number of
ALD for
multiple
sclerosis
2507
1348
1853
1568
1115
3949
1799
2158
1118
1140
1510
2233
542
2165
1900
4450
2374
1151
9139
3561
1655
183
49 418
Multiple
sclerosis
standardized
prevalence
123.7
122.9
116.8
116.2
115.3
113.3
112.0
103.1
101.5
98.0
94.7
94.7
93.2
87.9
87.4
87.3
86.7
85.5
84.3
84.1
83.1
81.1
94.7
Male
95% CI
121.2–126.2
119.5–126.2
114.1–119.5
113.3–119.1
111.8–118.7
111.5–115.1
109.4–114.7
100.8–105.3
98.4–104.5
95.1–100.9
92.3–97.1
92.7–96.7
89.2–97.2
86.0–89.8
85.3–89.4
86.0–88.6
84.9–88.5
83.0–88.1
83.4–85.2
82.7–85.5
81.1–85.2
75.1–87.1
94.3–95.1
Multiple
sclerosis
standardized
prevalence
70.8
73.4
65.4
62.9
63.2
71.7
65.8
57.4
47.1
54.7
57.2
52.3
43.1
50.6
47.8
53.5
45.2
46.8
50.4
49.5
49.5
50.6
54.8
Female
95% CI
68.1–73.5
69.6–77.1
62.4–68.3
59.8–66.1
59.5–66.9
69.6–73.7
62.8–68.7
54.9–59.7
44.1–50.1
51.5–57.8
54.4–60.0
50.1–54.5
39.1–47.2
48.5–52.7
45.6–50.0
52.0–55.0
43.3–47.0
44.1–49.5
49.4–51.4
47.9–51.8
47.2–51.8
43.8–57.5
54.4–55.3
Multiple
sclerosis
standardized
prevalence
171.5
166.5
163.5
162.6
162.2
151.3
154.6
143.9
149.7
136.2
128.7
131.4
136.2
120.1
122.2
118.2
123.9
118.9
115.7
114.6
112.0
108.5
130.5
95% CI
167.5–175.5
161.1–171.8
159.0–167.9
157.8–167.4
156.5–167.8
148.4–154.2
150.3–159.0
140.3–147.5
144.7–154.8
131.4–140.8
124.8–132.7
128.2–134.6
129.6–142.8
117.1–123.1
119.0–125.4
116.0–120.2
121.0–126.8
114.9–123.0
114.2–117.2
112.3–116.8
108.8–115.3
98.9–118.1
129.8–131.2
a Percentage of people affiliated to the CNAMTS in the general population.
[0.73 (95% Bayesian credible interval: 0.57–0.92) in Bouches du
Rhône—Fig. 3]. When considering females alone, an additional
département in the south of France (Herault) seemed to have a
lower prevalence (data not shown). For females, the North-Eastern
as well as the central part of France had the highest prevalence
of multiple sclerosis. For males, fewer départements in the
North-East of France (Haute Marne and Haute Saône) had
a higher prevalence, and a lower prevalence was also observed
1894
| Brain 2010: 133; 1889–1899
A. Fromont et al.
Figure 2 Multiple sclerosis standardized prevalence ratio for each département according to their significance (France, 2004).
Départements in red correspond to areas with a higher prevalence than the mean French level. Départements in green correspond to
areas with a lower prevalence than the overall French mean level.
in the Western départments of Loire Atlantique and Vendée (data
not shown). Interactions between gender and French administrative area did not improve the model. The deviance information
criterion of the model including the interaction with gender was
31 791.5, and without interaction with gender was 31 739.7,
indicating that geographic variations could not be considered as
different according to gender: the lower deviance information
criterion being considered the most valid. We tested the
North-East to South-West gradient of multiple sclerosis hypothesized from the results of Vukusic et al. (2007) by including trend
for latitude (X) and longitude (Y), either linearly (terms in X and Y)
or quadratically (terms in X, Y, X2, Y2 and XY). Neither improved
the fit to the data (deviance information criterion were 31 740
and 31 741, respectively).
Incidence
The incidence of ALD multiple sclerosis in France between
31 October 2003 and 31 October 2004 calculated among
French inhabitants covered by the CNAMTS was 7.5 per
100 000 (95% CI: 7.3–7.6); 10.4 (95% CI: 10.2–10.6) in females
and 4.2 (95% CI: 4.0–4.3) in males. The standardized incidence of
ALD multiple sclerosis in each region is shown in Table 3. The map
of standardized incidence ratios revealed a higher incidence of
multiple sclerosis in North-Eastern France, in two départements
in the north of France (Somme and Oise) and in two départements in central France (Eure et Loire and Loire et Cher) (Fig. 4).
In these high incidence areas, standardized incidence ratios varied
from a minimum of 1.26 (95% CI: 1.02–1.53) in Bas-Rhin to a
Epidemiology of multiple sclerosis in France
Brain 2010: 133; 1889–1899
| 1895
Figure 3 Smoothed relative risk of multiple sclerosis prevalence for each département compared to the mean French national level (both
genders, Bayesian model, France, 2004).
maximum of 2.37 (95% CI: 1.48–3.58) in Territoire de Belfort
(two départements in the North-Eastern area of France). Some
départements appeared to have a lower incidence of multiple
sclerosis. Among them, we found two départements on the
Atlantic coast: one in Western France (Loire Atlantique) and the
other in the South-West (Gironde); and three départements in
central eastern France (Ain, Rhone and Saône et Loire). In these
low incidence areas, standardized incidence ratios varied from a
minimum of 0.48 (95% CI: 0.27–0.78) in Ain to a maximum of
0.80 (95% CI: 0.69–0.93) in Paris. As the number of new ALD for
multiple sclerosis was rather low in 2004, we choose to not
present the results of the smoothed relative risk estimation for
incidence. We only used the Bayesian approach to test the interaction between gender and French administrative area. The deviance information criterion of the model including the interaction
with gender was 1624.4, compared to 1616.8 without interaction
with gender; the lower deviance information criterion being considered the most valid. The incidence of ALD multiple sclerosis in
France between 31 October 2004 and 31 October 2005, and
between 31 October 2005 and 31 October 2006 was the same
as during the previous 12-month period between 31 October
2003 and 31 October 2004 (data not shown).
Discussion
Only one reliable study has estimated the national prevalence of
multiple sclerosis in France (Vukusic et al., 2007). In 2003, multiple sclerosis prevalence was estimated to be 65.5 per 100 000
(95% CI: 62.5–67.5). However, this study was performed among
the population covered by the Caisse Collective de la Mutualité
Sociale Agricole, which insures only 7% of the French population,
1896
| Brain 2010: 133; 1889–1899
A. Fromont et al.
Table 3 Incidence of multiple sclerosis ALD in each region of France in 2004
Overall
Administrative
region of
France
Affiliation to
the CNAMTS
n (%a)
Picardie
Lorraine
Alsace
Centre
Champagne Ardenne
Franche Comte
Auvergne
Bretagne
Haute Normandie
Nord pas de Calais
PACA
Basse Normandie
Midi Pyrenees
Ile de France
Limousin
Poitou Charente
Bourgogne
Corse
Languedoc Roussillon
Aquitaine
Pays de Loire
Rhone Alpes
Total France
1 614 238
2 008 606
1 597 543
2 077 615
1 107 627
9 73 046
1 065 898
2 381 323
1 619 807
3 627 506
4 176 207
1 179 272
2 158 845
10 935 798
556 268
1 319 698
1 330 427
210 886
1 981 166
2 416 622
2 837 379
5 184 135
52 359 912
(86.1)
(86.3)
(89.3)
(83.9)
(83.2)
(86.1)
(80.5)
(79.4)
(89.9)
(90.2)
(89.8)
(81.8)
(80.8)
(97.3)
(78.1)
(78.5)
(82.4)
(77.9)
(81.1)
(79.6)
(85.1)
(88.4)
(87.2)
Number
of ALD
for multiple
sclerosis
181
227
161
188
100
85
89
196
132
289
308
83
158
816
39
91
90
15
134
153
174
323
4032
Multiple
sclerosis
standardized
incidence
11.0
10.8
9.7
9.0
8.9
8.6
8.1
8.1
8.0
7.7
7.3
7.0
7.0
7.0
6.9
6.8
6.8
6.8
6.6
6.2
6.1
6.1
7.5
Male
95% CI
10.2–11.8
10.1–11.6
8.9–10.5
8.3–9.6
8.0–9.8
7.7–9.6
7.3–9.0
7.5–8.6
7.3–8.7
7.2–8.1
6.9–7.7
6.3–7.8
6.4–7.5
6.7–7.2
5.8–8.0
6.1–7.5
6.0–7.5
5.0–8.5
6.0–7.2
5.7–6.7
5.6–6.6
5.8–6.4
7.3–7.6
Multiple
sclerosis
standardized
incidence
6.0
6.2
5.9
5.2
4.8
3.7
4.1
4.3
4.4
4.7
4.2
3.7
4.5
4.0
2.6
3.2
3.1
5.6
3.7
3.0
3.3
3.5
4.2
Female
95% CI
5.1–6.9
5.4–7.0
3.1–6.8
4.5–6.0
3.8–5.8
2.8–4.6
3.2–5.0
3.7–4.9
3.7–5.2
4.1–5.2
3.7–4.7
2.8–4.5
3.9–5.2
3.7–4.3
1.6–3.6
2.5–3.9
2.4–3.8
3.3–7.9
3.1–4.4
2.5–3.6
2.8–3.7
3.1–3.9
4.0–4.3
Multiple
sclerosis
standardized
incidence
15.6
15.0
13.2
12.4
12.6
13.2
11.7
11.4
11.2
10.5
10.1
10.0
9.2
9.7
10.6
10.0
10.0
7.8
9.1
8.9
8.7
8.5
10.4
95% CI
14.3–17.0
13.8–16.2
11.9–14.4
11.3–13.4
11.1–14.1
11.6–14.8
10.3–13.2
10.5–12.3
10.1–12.4
9.8–11.2
9.4–10.7
8.8–11.3
8.3–10.1
9.3–10.1
8.7–12.4
8.7–11.1
8.8–11.2
5.2–10.4
8.2–10.0
8.1–9.7
7.9–9.5
7.9–9.1
10.2–10.6
a Percentage of people affiliated to the CNAMTS in the general population.
essentially farmers and their families. Using the data of the main
French health insurance system (the CNAMTS), which covers 87%
of the French population, we decided to replicate the Vukusic
et al. (2007) study in order to estimate multiple sclerosis prevalence in France more accurately and describe its geographical variations. In our study, multiple sclerosis prevalence was estimated to
be 94.7 per 100 000 among subjects covered by the CNAMTS.
This prevalence was 130.5 per 100 000 females and 54.8 per
100 000 males. Multiple sclerosis incidence among this population
in 2004 was 7.5 per 100 000 (10.4 per 100 000 females and 4.20
per 100 000 males). Incidence remained stable in 2005 and 2006.
Surprisingly, our estimate of multiple sclerosis prevalence and
incidence suggested that the duration of ALD status for multiple
sclerosis was around 13 years. This value is certainly underestimated because before 1995, only patients with multiple sclerosis
and a disability were registered as ALD. Between 1995 and 2000,
several disease modifying treatments obtained authorization
from the French health care authorities. Before prescribing
disease-modifying treatment, neurologists require ALD status
from the health authorities for their patients in order to provide
patients with 100% coverage of their health care costs related to
multiple sclerosis. Since 1995, disease modifying treatment can be
prescribed as soon as definite relapsing multiple sclerosis, or as
soon as secondary progressive multiple sclerosis with superimposed relapses, is diagnosed [The IFNB Multiple Sclerosis Study
Group., 1993; Jacobs et al., 1996; PRISMS (Prevention of
Relapses and Disability by Interferon beta-1a Subcutaneously in
Multiple Sclerosis) Study Group, 1998; Consensus Conference
Organized by French Federation of Neurology, 2001], and since
2002 several disease modifying treatments can be prescribed after
a first demyelinating event (Jacobs et al., 2000). Also, from 1995
onwards, disease-modifying treatments were widely prescribed
and an obvious increase in the incidence of ALD declarations
was observed. This underestimation of prevalence before 1995
and the dramatic increase in incidence since the arrival of
disease-modifying treatments has led to this 13-year duration
calculated for multiple sclerosis-related ALD. In addition, since
2001, multiple sclerosis has been diagnosed earlier thanks to validated diagnostic criteria (McDonald et al., 2001) and widespread
access to magnetic resonance imaging. In 2004, however,
disease-modifying treatment had already been widely prescribed
by neurologists for several years. Neurologists therefore had
extensive disease modifying treatment experience and had standardized disease modifying treatment prescription procedures
according to validated recommendations. For progressive multiple
sclerosis, ALD status is obtained early because of permanent
disability. For all these reasons, we therefore believe that most
French patients with multiple sclerosis had ALD status in 2004.
To confirm the validity of our data, we compared our results
with those of the Lorraine Multiple Sclerosis Regional Network,
Epidemiology of multiple sclerosis in France
Brain 2010: 133; 1889–1899
| 1897
Figure 4 Multiple sclerosis standardized incidence ratio for each département according to their significance (France, 2004)
Départements in red correspond to areas with a higher incidence than the mean French level. Départements in green correspond
to areas with a lower incidence than the overall French mean level.
which includes all the private and hospital neurologists of the
Lorraine Region, using capture recapture methodology, in a
population-based cohort in an area of high prevalence. In
2004, the prevalence estimated by the Lorraine Network was
120 per 100 000 (95% CI: 119–121) (Debouverie et al., 2007a).
This result was very close to that calculated from the CNAMTS
data (124 per 100 000; 95% CI: 121–126). Similar values were
also obtained for females and males using both data sources.
The comparison between these values allowed us to validate
our data.
Thus, it seems reasonable to conclude that the number of new
multiple sclerosis-related ALD, after an increase between the
mid-1990s and the early 2000s, was relatively stable thereafter.
Our incidence, which was stable during 2004–06, confirms this
stability.
Vukusic et al. (2007) have shown differences in multiple sclerosis distribution in France. They observed a higher prevalence in
North-Eastern France in a particular population comprising essentially farmers and their families. In our study, we also confirmed
this higher prevalence in North-Eastern France in a large population representative of the French population. Despite the
differences between the populations studied (older age and
lower proportion of females in the population described by
Vukusic et al. in 2007) the same tendency of geographical
distribution was found.
One could argue that this higher prevalence could be related to
differences in gender and age distribution according to area.
However, this result was observed after adjusting for these two
variables. Another explanation could be that there were considerable differences in population sizes among the compared areas,
1898
| Brain 2010: 133; 1889–1899
which might have made it difficult to distinguish between random
and true variations in standardized prevalence ratios. Moreover,
neighbouring areas may not be independent as they probably
share more risk factors than more distant administrative areas
do. In order to account for this spatial autocorrelation as well as
population size heterogeneity, we used the Bayesian approach.
This type of model can provide smoothed relative risks that are
more accurate than standardized prevalence ratios. An alternative
method to account for spatial autocorrelation would have been to
use a generalized additive model, including a multidimensional
smoothing surface (Wood, 2006) with a structured variance–covariance matrix to account for heterogeneity. However, such
model is not applied routinely and requires specific methodological
development and validation without a clear benefit in terms of
results. Thus we preferred to use Bayesian validated approach.
With this latter approach, we showed that the higher prevalence
of multiple sclerosis was observed in North-Eastern France,
confirming that this could not be considered as a mere artefact.
Vukusic et al. (2007) pointed out that prevalence diminished
from the North-East to the South-West of France suggesting the
existence of a geographical gradient. We were not able to show
such a gradient for France formally, but rather areas of high and
lower prevalence. However, many other authors in non-European
countries have highlighted gradients of gradually decreasing
prevalence from North to South. For example, this tendency was
observed in the US (Kurtzke, 2008) as well as in the former Soviet
Union (Boiko et al., 1995). However, a higher prevalence was
observed in the South of Norway compared to the North of
the country (Rosati, 2001). These trends have not yet been
explained. These geographical variations may be related to genetic, environmental or socioeconomic factors such as cultural, dietary, income and interacting with each other. Several hypotheses
may be evoked to explain the higher multiple sclerosis prevalence
in the North-Eastern France. Indeed, it could be related to environmental factors, for example amount of sunshine. In a review,
Ebers et al. (2008) showed that ultraviolet radiation measured by
satellite was inversely proportional to multiple sclerosis prevalence.
Sunshine is the main determinant of vitamin D concentration
(Holick et al., 1987). In France, Chapuy et al. (1996) showed a
significant relationship between latitude and vitamin D concentration (r = 0.79) as well as between vitamin D concentration and
sunshine (r = 0.72). In Tasmania, multiple sclerosis risk was lower
for patients who had spent a lot of time in the sun during childhood (Van der Mei et al., 2003). This low risk of multiple sclerosis
correlated with actinic damage measured on the dorsal side of the
hand, which reflects past exposure to sunshine. Moreover vitamin
D seems to interact with genetic material through a vitamin D
response element located on the promoter region of HLA-DRB1
gene (Ramagopalan et al., 2009). This vitamin D response element
can influence gene expression and imparts vitamin D sensibility to
HLA-DRB1*15. In patients bearing HLA-DRB1*15, a lack of vitamin D during childhood could allow auto reactive T cells to escape
thymic deletion and thus increase immune disease risk. Other factors like racial mixing may also be implicated in geographical variations of multiple sclerosis. Low multiple sclerosis prevalences were
observed in areas with a high immigrant rate like in Paris (17%)
and on the Mediterranean coast (10%) (the national level of
A. Fromont et al.
immigrant’s rate being 8%) (www.insee.fr). Conversely, in North
Eastern regions like Burgundy and Champagne Ardennes where
multiple sclerosis prevalence is high, the immigration rate is low
(6%). However this hypothesis is probably not sufficient for explaining the geographical distribution of multiple sclerosis. Other
factors should be also considered like living in a rural or urban
area. While an excess of multiple sclerosis in rural populations
has been shown in studies from Sweden (Sallstrom, 1942),
Norway (Swank et al., 1952), Denmark (Hyllested, 1960) and
Lower Franconia (Bammer, 1960), other studies have suggested
an excess in urban areas (Beebe et al., 1967; Leibowitz and Alter,
1973). The availability of care facilities or socioeconomic factors
could also be involved, as there are also considerable
socioeconomic disparities between areas of high and low multiple
sclerosis prevalence.
In our study using Bayesian methodology, which takes into
account differences in population sizes in geographical units as
well as spatial autocorrelation, we found no gradient. Only a
higher prevalence in North-Eastern France and a lower prevalence
in Paris and its suburbs and on the Mediterranean coast were
observed. This distribution is not yet explained. Further studies
that explore environmental and socioeconomic factors are
needed to understand better the reasons for such geographical
variations. Among variables that could influence this distribution,
population movement, percentages of non-Caucasian subjects,
degree of urbanization and socioeconomic peculiarities seem
important.
Acknowledgements
We extend ours thanks to the ARSEP (Association pour
la Recherche sur la Sclérose En Plaques). We also thank
Mrs Valérie Jooste for her suggestions.
Supplementary material
Supplementary material is available at Brain online.
References
Multiple Sclerosis. Consensus Conference Organized by the French
Federation for Neurology. 7–8 June 2001. Rev Neurol 2001; 157:
902–1192.
Bammer H. Untersuchungen über multiplen sklerose in Unterfranken.
Stuttgart: Thieme Publisher; 1960.
Banerjee S, Carlin B, Gelfan A. Hierarchical modeling and analysis for
spacial data. Vol. 101. Florida: Chapman and Hall/CRC Press LLC
Boca Raton; 2004.
Beebe GW, Kurtzke JF, Kurland LT, Auth TL, Nagler B. Studies on the
natural history of multiple sclerosis. 3. Epidemiologic analysis of the
army experience in World War II. Neurology 1967; 17: 1–17.
Besag J, Yorke J, Mollié A. Bayesian image restoration with two
applications in spatial statistics. Ann Inst Stat Math 1991; 43: 1–59.
Boiko A, Deomina T, Favorova O, Gusev E, Sudomoina M, Turetskaya R.
Epidemiology of multiple sclerosis in Russia and other countries of the
former Soviet Union: investigations of environmental and genetic
factors. Acta Neurol Scand Suppl 1995; 161: 71–6.
Epidemiology of multiple sclerosis in France
Breslow N, Day N. Statistical methods in cancer research. The design and
analysis of cohort studies Vol. 2. Lyon: IARC Scientific Publication;
1980.
Chapuy M, Preziosi P, Maamer M, Arnaud S, Galan P, Hercberg S, et al.
Prevalence of vitamine D insufficiency in an adult normal population.
Osteoporos Int 1996; 7: 439–43.
Cliff A, Ord J. Spatial processes. In: Models and applications, AJ Scott
(eds). London, Pion, 1981.
Colonna M. [Influence of a priori parameters on bayesian relative risks
estimations. Spatial distribution of bladder cancer in the urban area of
Grenoble]. Rev Epidemiol Sante Publique 2006; 54: 529–42.
Debouverie M, Pittion-Vouyovitch S, Louis S, Roederer T, Guillemin F.
Increasing incidence of multiple sclerosis among women in Lorraine,
Eastern France. Mult Scler 2007a; 13: 962–7.
Debouverie M, Rumbach L, Clavelou P. [The organisation of health care
and epidemiology of multiple sclerosis in France]. Rev Neurol (Paris)
2007b; 163: 637–45.
Ebers GC. Environmental factors and multiple sclerosis. Lancet Neurol
2008; 7: 268–77.
Forbes RB, Swingler RJ. Estimating the prevalence of multiple sclerosis in
the United Kingdom by using capture-recapture methodology. Am J
Epidemiol 1999; 149: 1016–24.
Holick MF, Smith E, Pincus S. Skin as the site of vitamin D synthesis
and target tissue for 1,25-dihydroxyvitamin D3. Use of calcitriol (1,25dihydroxyvitamin D3) for treatment of psoriasis. Arch Dermatol 1987;
123: 1677–83a.
Hyllested K. Disseminated sclerosis in Denmark. Prevalence and geographic distribution. Copenhagen: Jorgensen Publ; 1960.
Jacobs LD, Beck RW, Simon JH, Kinkel RP, Brownscheidle CM,
Murray TJ, et al. Intramuscular interferon beta-1a therapy initiated
during a first demyelinating event in multiple sclerosis. CHAMPS
Study Group. N Engl J Med 2000; 343: 898–904.
Jacobs LD, Cookfair DL, Rudick RA, Herndon RM, Richert JR,
Salazar AM, et al. Intramuscular interferon beta-1a for disease
progression in relapsing multiple sclerosis. The Multiple Sclerosis
Collaborative Research Group (MSCRG). Ann Neurol 1996; 39:
285–94.
Kurtzke JF. A reassessment of the distribution of multiple sclerosis. Part
one. Acta Neurol Scand 1975; 51: 110–36.
Kurtzke JF. Some contributions of the Department of Veterans Affairs to
the epidemiology of multiple sclerosis. Mult Scler 2008; 14: 1007–12.
Lang S, Brezger A. Bayesian P-splines. J Comput Graph Stat 2004; 13:
183–212.
Lauer K, Firnhaber W, Reining R, Leuchtweis B. Epidemiological investigations into multiple sclerosis in southern Hesse. I. Methodological
problems and basic epidemiologic characteristics. Acta Neurol Scand
1984; 70: 257–65.
Leibowitz U. Alter M. Multiple sclerosis: clues to its cause. Amsterdam:
North Holland; 1973.
Maiti T. Hierarchical Bayes estimation of mortality rates of disease
mapping. J Statist Plan Inference 1998; 69: 339–48.
McDonald WI, Compston A, Edan G, Goodkin D, Hartung HP,
Lublin FD, et al. Recommended diagnostic criteria for multiple
Brain 2010: 133; 1889–1899
| 1899
sclerosis: guidelines from the International Panel on the diagnosis of
multiple sclerosis. Ann Neurol 2001; 50: 121–7.
McGuigan C, McCarthy A, Quigley C, Bannan L, Hawkins SA,
Hutchinson M. Latitudinal variation in the prevalence of multiple
sclerosis in Ireland, an effect of genetic diversity. J Neurol Neurosurg
Psychiatry 2004; 75: 572–6.
Olsen SF, Martuzzi M, Elliott P. Cluster analysis and disease mapping–
why, when, and how? A step by step guide. BMJ 1996; 313: 863–6.
Pascutto C, Wakefield JC, Best NG, Richardson S, Bernardinelli L,
Staines A, et al. Statistical issues in the analysis of disease mapping
data. Statist Med 2000; 19: 2493–519.
Phadke JG, Downie AW. Epidemiology of multiple sclerosis in the northeast (Grampian region) of Scotland–an update. J Epidemiol Community
Health 1987; 41: 5–13.
Potthoff R, Whittinghill M. Testing for homogeneity: the binomial and
multinomial distributions. Biometrika 1966; 53: 167–82.
PRISMS (Prevention of Relapses and Disability by Interferon beta-1a
Subcutaneously in Multiple Sclerosis) Study Group. Randomised
double-blind placebo-controlled study of interferon beta-1a in relapsing/remitting multiple sclerosis. Lancet 1998; 352: 1498–504.
Ramagopalan SV, Maugeri NJ, Handunnetthi L, Lincoln MR, Orton SM,
Dyment DA, 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.
Roberts MH, Martin JP, McLellan DL, McIntosh-Michaelis SA,
Spackman AJ. The prevalence of multiple sclerosis in the
Southampton and South West Hampshire Health Authority. J Neurol
Neurosurg Psychiatry 1991; 54: 55–9.
Rosati G. The prevalence of multiple sclerosis in the world: an update.
Neurol Sci 2001; 22: 117–39.
Sallstrom T. Das vorkommon und die Verbreitung der multiple sklerose in
Schweden. Acta Med Scand 1942; suppl. 137: 1–141.
Spiegelhalter D, Best N, Carlin B, Van der Linde A. Bayesian measures of
model complexity and fit (with discussion). J Royal Statist Soc 2002;
64: 583–639.
Swank RL, Lerstad O, Strom A, Backer J. Multiple sclerosis in rural
Norway its geographic and occupational incidence in relation to
nutrition. N Engl J Med 1952; 246: 722–8.
The IFNB Multiple Sclerosis Study Group. Interferon beta-1b is effective
in relapsing-remitting multiple sclerosis. I. Clinical results of a multicenter, randomized, double-blind, placebo-controlled trial. Neurology
1993; 43: 655–61.
Upton G, Fingleton B. Spatial data analysis by example. John Wiley,
Chichester 1985.
Van der Mei IA, Ponsonby AL, Dwyer T, et al. Past exposure to sun, skin
phenotype, and risk of multiple sclerosis: case control study. BMJ
2003; 327: 334–6.
Vukusic S, Van Bockstael V, Gosselin S, Confavreux C. Regional variations in the prevalence of multiple sclerosis in French farmers. J Neurol
Neurosurg Psychiatry 2007; 78: 707–9.
Williams ES, McKeran RO. Prevalence of multiple sclerosis in a south
London borough. BMJ (Clin Res Ed) 1986; 293: 237–9.
Wood S. Low-rank scale-invariant tensor product smooths for generalized additive mixed models. Biometrics 2006; 62: 1025–36.