Multiple Sclerosis in US Veterans of the
Vietnam Era and Later Military Service:
Race, Sex, and Geography
Mitchell T. Wallin, MD, MPH,1William F. Page, PhD,2and John F. Kurtzke, MD1
We identified 5,345 cases of multiple sclerosis (MS) among US veterans who first entered military service between 1960
and 1994, and who were “service-connected” for MS by the Department of Veterans Affairs (VA). Two controls per case
were matched on age, date of service entry, and branch of service. Available for service and VA files were demographic
and military data for 4,951 cases and 9,378 controls. Versus white men, relative risk of MS was significantly higher for
all women, at 2.99 for whites, 2.86 for blacks, and 3.51 for those of other races. This was a significant increase from our
prior series of veterans of World War II and the Korean Conflict, where white women had a relative risk of 1.79. Risk
for black men was higher now (0.67 vs 0.44), while other men remained low (0.30 vs 0.22). Residence at service entry
in the northern tier of states had a relative risk of 2.02 versus the southern tier, which was significantly less than the 2.64
for the earlier series. Residence by individual state at birth and service entry for white men further supported this
decreasing geographic differential. Such marked changes in geography, sex, and race in such a short interval strongly
imply a primary environmental factor in the cause or precipitation of this disease.
Ann Neurol 2004;55:65–71
Multiple sclerosis (MS) is an inflammatory demyelinat-
ing disease of the central nervous system. It affects ap-
proximately 300,000 persons in the United States and
up to 2 million worldwide.1Despite intensive research
over the last century, no specific cause has been iden-
tified. Epidemiological studies have been important in
this disease in focusing research efforts by identifying
disease patterns, assessing risk factors, and evaluating
The US veteran population provides an exceptional
resource for such studies. In eight previous articles, we
have explored the epidemiology of MS using an unusu-
ally large cohort of MS cases and preillness matched
controls comprising US veterans of World War II
(WWII) and the Korean Conflict (KC).2–9Risk factors
for developing MS from these studies included female
sex, white race, northern latitude, urban residence,
high socioeconomic status, and Scandinavian ancestry
of the populations, but not of the patients themselves.
In this report, we examine data on a newly assem-
bled cohort of US veterans with first entry into mili-
tary service between 1960 and 1994. We compare
them with our earlier WWII-KC cohort, which had
been assembled in a similar manner. The uniqueness of
this new cohort lies in its size, racial diversity, national
representation, and relative recency, with, as before,
preillness matched controls from the military to repre-
sent the specific population at risk.
Subjects and Methods
The new Vietnam era (VNE) and later service case–control
series was assembled in a manner similar to the WWII-KC
series.2Starting with a list of all veterans receiving “service-
connection” for MS from the Department of Veterans Af-
fairs (VA), we identified 5,345 new cases who first entered
military service between 1960 and 1994 and who were alive
in 1995. The VA has established the following periods of
service for “Wartime veterans:” WWII: December 7, 1941
to December 31, 1946; KC: June 27, 1950 to January 31,
1955; VNE: August 5, 1964 to May 7, 1975; and the Gulf
From the1Neuroepidemiology Section, Neurology Service, Depart-
ment of Veterans Affairs Medical Center, and Department of Neu-
rology, Georgetown University Medical School; and
Follow-up Agency, National Academy of Sciences, Washington, DC.
Received Jun 5, 2003, and in revised form Aug 15. Accepted for
publication Aug 19, 2003.
Address correspondence to Dr Wallin, Neurology Service, VAMC,
Supplementary material to this article (Appendix Tables) can be
viewed online at www.interscience.wiley.com/jpages/0364-5134/
The opinions and assertions contained herein are those of the au-
thors and are not to be construed as reflecting the views or positions
of the Department of Veterans Affairs, Georgetown University, the
National Academy of Sciences, the Institute of Medicine, or the
National Research Council.
This article is a US Government work and, as such, is in the public do-
main in the United States of America.
Published 2003 by Wiley-Liss, Inc., through Wiley Subscription Services
War: August 2, 1990 to September 30, 2003.10A decision
for service-connection for MS requires definitive evidence
of clinical signs upon examination attributable to MS dur-
ing or within 7 years after military service. Copies of all
relevant medical information are in the veteran claims
folder for all service connection requests. The decision was
made without regard to rank, race, sex, or financial status.
Roughly half of the cases (and, of course, controls) served
in the VNE. Sixty-five random case records were reviewed
for diagnostic accuracy (M.T.W. and J.F.K.) and 97% (bi-
nomial 95% confidence interval, 89.3–99.6%) met the
Schumacher11and Poser12criteria for definite MS. All data
retrieval and initial analyses were performed by the Medical
Follow-up Agency-National Academy of Sciences under our
A total of 10,683 controls were selected from the auto-
mated personnel files of the Defense Manpower Data Center
and a 1% sample of military personnel held by the Medical
Follow-up Agency-National Academy of Sciences depending
on date of service entry. The controls were matched on a 2
to 1 basis for age, month and year of entry into military
service, and branch of service. The VNE and later service
series contains substantial numbers of women and minorities
because of the changing demographics of the military (see
Appendix Table 3; Appendix Tables 1–3 can be viewed on-
line at www.interscience.wiley.com/jpages/0364-5134/supp-
Data from computerized military records were available
for most subjects, and hard-copy military records also were
abstracted and computerized to obtain additional informa-
tion. The numbers of MS cases and controls analyzed were
4,951 and 9,378, respectively. Race and sex were taken di-
rectly from the computerized record, as was home of record,
which is the same as residence at time of entry into service
for almost all subjects.
The contiguous United States was divided into the same
three geographic tiers used previously to allow for stratifica-
tion by residence location. The northern tier was defined as
states north of 41 degrees to 42 degrees north latitude; the
southern tier as states south of 37 degrees north latitude as
well as California from Fresno south. The middle tier in-
cluded all other states and northern California.
For comparisons between individual case–control (CC) ra-
tios with the WWII-KC series, adjustment was made by di-
viding the specific CC ratio by the total CC ratio for the
comparison of which this was a component. For small sub-
sets, this approximates the standard odds ratio. This study
has been approved by the VA Medical Center Washington,
DC Institutional Review Board. All statistical analyses were
performed using the Statistical Analysis System computer
There were 394 MS cases and 1,305 controls with
missing data and/or who were unable to be linked be-
tween Department of Defense and VA databases.
These cases were eliminated from the study cohort,
leaving a total of 4,951 MS cases and 9,378 controls
(Table 1). As expected, the predominant racial group
for MS cases was white, but there were a substantial
number of blacks and “other” race persons. This last
classification includes Asian and Native American vet-
erans as the major groups.
Table 1 shows the adjusted CC ratios for MS by
sex and race according to period of service. Black
males in the VNE and later service cohort had a sig-
nificantly higher risk for developing MS compared
with the WWII-KC cohort. The other male race
groups had fairly similar risk ratios between the two
All female groups in the VNE and later service co-
hort had a higher risk for developing MS than the
WWII-KC cohort. Differences for white females and
all females as a group were statistically significant.
Comparison of the nonwhite groups between cohorts is
Table 1. Multiple Sclerosis in US Veterans by Sex and Race According to Period of Service
Sex and Race
Vietnam and laterWWII and KCa
No. of cases,
No. of cases,
aData of Table 1 from Kurtzke and colleagues (1979)2
bAdjusted CC ratios.
WWII ? World War II; KC ? Korean Conflict; CI ? confidence interval; CC ? case–control.
Annals of NeurologyVol 55No 1January 2004
limited with the low numbers in the WWII-KC series.
With the differing proportions between cohorts for all
subsets of race and sex, we also calculated these data as
the relative risk of MS versus that for white males for
each service cohort (Table 2).
Adjusted CC ratios for MS by sex, race, and tier of
residence at entry into service are shown in Table 3.
There is a decreasing risk for MS when proceeding
from north to south in each sex-race group, and of
fairly equal degree for each subset. The low number of
nonwhite MS veterans in the WWII-KC cohort pre-
cluded calculating such ratios for other-race males and
nonwhite females. To compare north versus south di-
rectly, we summarized these data as north to south ra-
tios in Table 4. There is a 2 to 1 difference for the
total VNE and later service cohort. This gradient was
more pronounced in the WWII-KC cohort, at 2.6 to
1. Thus, although higher MS risk in the northern
states than in the south persists, it is now to a signifi-
cantly lesser degree.
Figure 1 shows the distribution by state and tier of
residence at service entry for the WWII-KC cohort and
the corresponding distribution for the new series. To
avoid confounding by sex or race, data for both are
limited to white males. Because of low numbers
(case ? control ? 10 or less) in some comparisons,
results for Maryland and Washington, DC and for Ne-
vada and Utah were combined for all assessments. The
latitude gradient observed in the WWII-KC series is
still apparent but less pronounced. Figure 2 is a scat-
terplot of these adjusted CC ratios, with the x-axis pre-
senting the data from the VNE and later cohort, and
the y-axis presenting the data from WWII-KC. The
Spearman’s coefficient of correlation (r) of 0.73 is very
highly significant, indicating that, whereas difference
are clearly lessening, the overall distribution of MS in
this country has remained similar over a 40 to 50-year
Table 5 presents the adjusted CC ratios for residence
at birth and service entry, stratified by geographic tier
and data source. The Pacific column had to be ex-
cluded for the recent cohort, because California was
not divisible by birth location into its two tiers, middle
and south. The north to south ratio was 2.27 birth and
Table 2. Relative Risk of Multiple Sclerosis versus White
Males by Race and Sex from Adjusted Case–Control Ratios
Sex and Race Vietnam and laterWWII and KC
Data from Table 1.
WWII ? World War II; KC ? Korean Conflict.
Table 3. Multiple Sclerosis in US Veterans by Tier of Residence within the Contiguous US at Entry into Service by Sex and Race
Sex and Race
North TierMiddle TierSouth Tier
No. of cases,
No. of cases,
No. of cases,
Vietnam and later
World War II and Korean Conflicta
aData from Table 2 in Kurtzke and colleagues (1979)2
bIncludes black female, and other race men and women.
CC ? case–control.
Wallin et al: MS in US Veterans
2.07 at entry for the new cohort, in which the ratio
had been 1.97 at entry with the Pacific column in-
cluded. The complete tier data for the WWII-KC co-
hort showed north to south ratios of 2.50 at birth
and 2.46 at entry into active duty.4The correlation
between the adjusted CC ratios by state for the new
cohort at birth and entry into active duty was very
strong (Spearman’s r of 0.84) (Fig. 3). This also
serves to support the inference that geographic move-
ment between birth and service entry in either cohort
was not primarily responsible for the north to south
gradients in risk, or for their significant decrease in
the recent cohort.
This relatively modern era US veteran cohort has pro-
duced what we believe to be important information on
the changing distribution of MS by race, sex, and ge-
ography. The north to south gradient persists but is
much less pronounced than for our earlier WWII-KC
cohort. There has been a notably increased risk for de-
veloping MS in white females compared with the pre-
vious cohort. In the VNE and later cohort, the relative
risks for black and other race females were also signif-
icantly higher than for white males and appeared sim-
ilar to that for white females.
The rarity of MS among certain racial groups such
as the native Siberians, North American Indians, and
Fig 1. Adjusted case–control ratios (?100) for white male
veterans service-connected for multiple sclerosis (MS), according
to state and tier of residence within the coterminous United
States at entry into active duty (EAD): (top) Vietnam-era and
later service cohort; (bottom) World War II–Korean Conflict
cohort. Data in Appendix Table 1.
Fig 2. Scatterplot of adjusted case–control ratios (?100) for
white male veterans service-connected for multiple sclerosis
(MS) of Vietnam-era and later service by state of entry into
active duty (EAD) versus white males of World War II–Ko-
rean Conflict (WWII-KC) by EAD; California is divided into
its two parts as in Figure 1. Spearman’s r ? 0.73. Data in
Appendix Table 1.
Table 4. Multiple Sclerosis in US Veterans
Sex and Race
WWII and KC
(ratio north: south)
Ratios of residence within northern tier of the contiguous US versus
southern tier from adjusted case-control ratios by sex and race.
Data from Table 3.
aIncludes black women, and other race men and women.
WWII ? World War II; KC ? Korean Conflict.
Annals of Neurology Vol 55 No 1 January 2004
Japanese is held as evidence for a strong influence of
genetics on the disease. Yet, many of these groups have
not recently been studied intensively. Note our own
findings for all nonwhite women veterans. The specific
races of our “other race” females are not yet available,
but the expectation is that they will be mostly Asian.
Two recent reports found MS to be present among
ethnic groups felt earlier to be resistant to developing
MS. Mirsattari and colleagues14presented a population
based study of aboriginals in Manitoba, Canada, where
an MS prevalence rate of 40 per 100,000 was found.14
Seven aboriginals with MS were identified in the study,
most with an aggressive disease course and predomi-
nance of neuromyelitis optica. Grønlie and colleagues
found an increasing incidence of MS over the past 30
years in northernmost Norway.15The counties that
were surveyed were those where most of the Sami peo-
ple (Lapps) lived, six of whom had diagnoses of MS in
the 1993 prevalence survey for a crude rate of 73 per
MS has been described as distributed throughout
the world within three zones of high, medium, and
low frequency.16High-frequency areas, with preva-
lence rates of 30 and above per 100,000 population
currently include most of Europe into the former So-
viet Union, Israel, Canada, and the entire United
States, plus New Zealand and southeastern Australia.
This also seems to include the easternmost part of
Russia. Evidence has indicated the existence of geo-
graphic gradients within North America,16,17Austra-
lia,18and Europe.19,20Although time trend analyses
have shown many of these gradients to persist,15,21,22
diffusion of these gradients over one or two genera-
tions also has been reported,17a pace too rapid for
genetic influences to be effective. There is now little
difference between north versus south in Europe, and,
as noted, the difference in the United States is de-
creasing even though half our new cohort is from the
Vietnam era (1964–1975) not long after the WWII-
Korean Conflict periods.
Some have argued against the existence of recent
latitude gradients for MS. Robertson and colleagues23
and Forbes and colleagues24have stated that latitude
gradients in the United Kingdom may be an artifact
of methodological differences. Savettieri and collea-
geus25have concluded with 20 years of prevalence
data that Sicily now is a high-risk region for MS, but
it was not so reported in earlier data. This high rate,
Table 5. Multiple Sclerosis in US White Male Veterans
Birth Entry into Service
Case ControlAdjusted CCCase ControlAdjusted CC
Vietnam and latera
WWII & Korean Conflictb
North: south ? 2.27North: south ? 2.07
North: south ? 2.50North: south ? 2.46
Ratios of residence of birth and at entry into service, by tier of residence in the contiguous US.
aData of Appendix Table 2, excludes Pacific column: entry into active duty north:south ratio with Pacific column is 1.97.
bData of Kurtzke and colleagues (1985)4; entry into active duty north: south ratio without Pacific column is 2.49 (Appendix Table 1).
Fig 3. Scatter plot of adjusted case-control ratios (?100) for
white male veterans service-connected for MS of Vietnam-era
and later service by state of entry into active duty (EAD) vs.
state of birth. Spearman r ? 0.84. Data in Appendix Table 2.
Wallin et al: MS in US Veterans
they argue, is in contrast with the gradient hypothe-
sis; they believe the high frequencies observed on the
island are related to the genetic makeup of the pop-
ulace. Geography, however, is a marker of the envi-
ronmental influences that we would argue are playing
a major role in initiating this disease. It is our im-
pression that, at least in the Occident, there is a clear
diffusion of this disease over time, with a decrease or
disappearance of previously identified geographic vari-
MS, then, is predominantly endemic among white
populations of Europe and its former colonies such as
Canada, the United States, Australia, and New Zea-
land. Whether or not this is related in whole or in part
to the geographic distribution of the races and varia-
tions in racial susceptibility, or to diagnostic practices,
survey methods, and reporting procedures has been
questioned repeatedly. The admixture of race within
the United States may be argued to be a confounder
that could weaken our conclusions. There has been ev-
idence for increasing genetic admixture of nonwhites
and whites over the past few decades.27,28Harrison and
Bennett reported there were 150,000 interracial mar-
ried couples in the United States in 1960.29This num-
ber increased to 1 million in the 1990. Therefore, there
may have been some effect of racial admixture between
white, black, and other races in our cohort, but it was
likely minimal because of the time frame (1960–1994)
of collection of our MS cases. Most subjects in our
cohort were born before 1970. Although one could ar-
gue as to what the racial terms really mean, these cat-
egories were applied equally across cases and controls.
We believe that our conclusions are independent,
and potential confounders have largely been dealt with
by our case–control matching and stratification of re-
sults. For the north to south gradient, the consistent
gradient effect across all race and sex subgroups makes
bias an unlikely explanation for the diffusion of the
gradient over time. Moreover, there are sufficient com-
parable data to conclude that there has been a striking
geographic variation in the risk of MS in North Amer-
ica and Europe, even though some of the details may
This report has shown that MS risk has changed in
the US veteran population, and thus presumably in all
Americans, over a single generation. The risk for devel-
oping MS has significantly increased for white women,
and now all women regardless of race, have signifi-
cantly higher risk of MS than do white males in this
country. This change is too rapid for genetic or hor-
monal influences and again supports the need to look
further at the interaction of both the macroclimate (en-
vironment) as well as the microclimate (genetics) to
solve the enigma which is MS.
This work was supported in part by a Veterans Affairs Merit Review
Grant (V688-1532, J. K.; V688P-1997, J. K.) and the Neuroepide-
miology Research Program, Veterans Affairs Medical Center–Wash-
This study was presented at the annual meetings of the American
Neurological Association, Boston, MA, 10/15–18/00; the 2001
Congress of Epidemiology, Toronto, Canada, 6/13–16/01; and the
American Epidemiology Society, New York, 3/22–23/02.
1. Anderson DW, Ellenberg JH, Leventhal CM, et al. Revised es-
timate of prevalence of multiple sclerosis in the United States.
Ann Neurol 1992;31:333–336.
2. Kurtzke JF, Beebe GW, Norman JE Jr, et al. Epidemiology of
multiple sclerosis in US veterans. 1. Race, sex, and geographic
distribution. Neurology 1979;29:1228–1235.
3. Norman JE Jr, Kurtzke JF, Beebe GW. Epidemiology of mul-
tiple sclerosis in US veterans. 2. Latitude, climate and the risk
of multiple sclerosis. J Chronic Dis 1983;36:551–559.
4. Kurtzke JF, Beebe GW, Norman JE Jr. Epidemiology of mul-
tiple sclerosis in US veterans. III. Migration and the risk of MS.
5. Kurtzke JF, Page WF, Murphy FM, et al. Epidemiology of
multiple sclerosis in US veterans. IV. Age at onset. Neuroepi-
6. Page WF, Kurtzke JF, Murphy FM, et al. Epidemiology of
multiple sclerosis in US veterans. V. Ancestry and the risk of
multiple sclerosis. Ann Neurol 1993;33:632–639.
7. Page WF, Mack TM, Kurtzke JF, et al. Epidemiology of mul-
tiple sclerosis in US veterans. VI. Population ancestry and sur-
name ethnicity as risk factors for multiple sclerosis. Neuroepi-
8. Kurtzke JF, Page WF. Epidemiology of multiple sclerosis in US
veterans. VII. Risk factors for MS. Neurology 1997;48:
9. Wallin MT, Page WF, Kurtzke JF. Epidemiology of multiple
sclerosis in US veterans. VIII. Long-term survival after onset of
multiple sclerosis. Brain 2000;123:1677–1687.
10. Klein RE. Period of service in Veteran Population 2001 ad-
justed to Census 2000. VA Special Topic Report, September 30,
2002. Available at: http://www.va.gov/vetdata/demographics/; ac-
11. Schumacher GA, Beebe GW, Kibler RF, et al. Problems of ex-
perimental trials of therapy in multiple sclerosis: report by the
panel on the evaluation of experimental trial of therapy in mul-
tiple sclerosis. Ann NY Acad Sci 1965;122:552–568.
12. Poser CM, Paty DW, Scheinberg L, et al. New diagnostic cri-
teria for multiple sclerosis: guidelines for research protocols.
Ann Neurol 1983;13:227–231.
13. SAS Institute. SAS procedures guide. Release 6.03 edition.
Cary, NC: SAS Institute, 1988.
14. Mirsattari SM, Johnston JB, McKenna R, et al. Aboriginals
with multiple sclerosis: HLA types and predominance of neu-
romyelitis optica. Neurology 2001;56:317–323.
15. Grønlie SA, Myrvøll E, Hansen G, et al. Multiple sclerosis in
North Norway, and first appearance in an indigenous popula-
tion. J Neurol 2000;247:129–133.
16. Kurtzke JF, Wallin MT. Epidemiology. In: Multiple sclerosis:
diagnosis, medical management, and rehabilitation. Burks JS,
Johnson KP, eds. New York: Demos Press, 2000:49–71.
17. Pryse-Phillips WEM. The incidence and prevalence of multiple
sclerosis in Newfoundland and Labrador, 1960–1984. Ann
Annals of Neurology Vol 55No 1January 2004
18. Hammond SR, McLeod JG, Millingen KS, et al. The epidemi-
ology of multiple sclerosis in three Australian cities: Perth,
Newcastle and Hobart. Brain 1988;111:1–25.
19. Lauer K. Multiple sclerosis in the old world. The new old map.
In: Multiple sclerosis in Europe. An epidemiological update.
Firnhaber W, Lauer K, eds. Darmstadt: LTV Press, 1994:
20. Sumelahti ML, Tienari PG, Wikstrom J, et al. Regional and
temporal variation in the incidence of multiple sclerosis in Fin-
land 1979–1993. Neuroepidemiology 2000;19:67–75.
21. Kurtzke JF. Multiple sclerosis in time and space-geographic
clues to cause. J Neurovirol 2000;6:S134–S140.
22. Forbes RB, Swingler RJ. Estimating the prevalence of multiple
sclerosis in the United Kingdom by using capture-recapture
methodology. Am J Epidemiol 1999;149:1016–1024.
23. Robertson N, Compston A. Surveying multiple sclerosis in the
United Kingdom. J Neurol Neurosurg Psychiatry 1995;58:2–6.
24. Forbes RB, Wilson SV, Swinger RJ. The prevalence of multiple
sclerosis in Tayside, Scotland: do latitude gradients really exist?
J Neurol 1999;246:1033–1040.
25. Savettieri G, Ragonese P, Aridon P, et al. Epidemiology of mul-
tiple sclerosis in Sicily. Neurol Sci 2001;22:175–177.
26. Kurtzke JF. Epidemiology of multiple sclerosis. Does this really
point toward an etiology? Lectio Doctoralis. Neurol Sci 2000;
27. Waters MC. Immigration, intermarriage, and the challenges of
measuring racial/ethnic identities. Am J Public Health 2000;90:
28. Reitnauer PG, Go RC, Acton RT, et al. Evidence for genetic
admixture as a determinant in the occurrence of insulin-
dependent diabetes mellitus in U.S. blacks. Diabetes 1982;31:
29. Harrison R, Bennett C. Racial and ethnic diversity. In: Farley
R, ed. State of the Union: America in the 1990s. Vol 2.
Social Trends. New York: Russell Sage Foundation Press,
30. Noonan CW, Kathman SJ, White MC. Prevalence estimates
for MS in the United States and evidence of an increasing trend
for women. Neurology 2002;58:136–138.
Wallin et al: MS in US Veterans