Methods
Study populations and sampling procedure
The
serum repository results from the third
population-based health examination survey conducted
in Switzerland within the framework of the
multicenter WHO-MONICA (MONItoring trends and
determinants in CArdiovascular disease) project in
19923 [28]. The survey was based
on a random sample of the adult resident population
of two Swiss regions, composed of the cantons of
Vaud and Fribourg (mainly French-speaking, age range
2574) on one hand and of the canton of Ticino
(Italian-speaking, age range: 35 to 64) on the other
hand. The two-stage sampling procedure consisted
first in drawing a random sample of communes
(administrative divisions enclosing a village or a
city and its surrounding territory) stratified by
size (number of inhabitants) and, secondly in
drawing a random sample of residents in the
population files of the selected communes. The
sampling fraction of communes were 51 out of 651 in
Vaud-Fribourg and 25 out of 147 in Ticino. The
population samples drafted in these two regions were
of 3300 and 2000, of whom 1742 and 1510 persons
respectively accepted to participate to the survey,
corresponding to overall participation rates of 53%
and 76% in the two areas. The participants were
invited to attend a health examination in their
commune of residence and to complete a
self-administered questionnaire about their
socio-demographic characteristics, dietary intakes,
and lifestyle habits. In addition to anthropometric
and blood pressure
measurements, a venous blood sample was drawn from
each subject for determination of blood lipid
concentrations. The blood samples were allowed to
clot at room temperature, centrifuged within two
hours of harvest, and the serum transferred and
stored at -20°C.
Socio-economic indicators
Educational level was
categorized in four classes according to the highest
degree the subjects had achieved and the age at
which they completed their education, i.e. i) low :
mandatory school only (age ≤ 16); ii) medium:
apprenticeship (age > 16); iii) high: secondary or
professional school (age > 18) and iv) very high:
university (age > 22).
Serologic testing
Samples
were assayed by ELISA (HerpeSelect, Focus
Technologies, Inc., Cypress, CA) for HSV-1 (Herpes
Simplex Virus type 1) and HSV-2 IgG,
according to the manufacturer instructions [18,29].
All samples with an Index Value between 0.81.2 (OD
standardized using a reference serum set by the
manufacturer to provide a cut off value arbitrarily
given the Index value of 1) were tested again using
the serum supernatant after centrifugation (15000
rpm during 5 min on a Eppendorf 5804 microfuge) and,
if still undetermined, were resolved by immunoblot (HerpeSelect
1 & 2 Immunoblot, Focus Technologies, Inc).
Statistical Analysis
The data from the two
study regions were pooled together for the
overlapping age range (3564) to increase
statistical power. Complete data from the
Vaud-Fribourg region extending to the age range
2574 were used to analyze age patterns of HSV
prevalence. In order to account for the regional
differences in sampling fractions and participation
rates in the MONICA survey, the HSV seroprevalence
estimates were weighted by sex and age so as to be
representative of the populations of the two study
regions at the time of the survey (end of 1992). All
the statistical analyses (i.e. estimation of
prevalence rates and odd ratios, calculation of
confidence intervals, logistic regression) were
performed using the procedures adapted to weight
survey data ("svy commands") in the Stata 6.0 (Stata
Corp, College Station, TX) software.
The matching of the
epidemiological data from the MONICA survey with the
results of the serological testing for HSV-1
and HSV-2 antibodies allowed to analyze the
prevalence of herpes across the different
socio-demographic strata of the Swiss population and
to identify factors influencing this prevalence.
Variations of the seroprevalence rates between
groups according to various characteristics (sex,
age, marital status, education level, place of
residence, etc) were first examined in
cross-tabulations and bivariate logistic regression
models. All variables were then entered into a
multivariate logistic regression model in order to
test whether they were predictors of the HSV
status independently of each other. The contribution
of a given explanatory variable was tested by the
Wald test.
A
classification tree analysis [30]
was performed using the S-PLUS (StatSci, a Division
of Math Soft, Seattle, USA) software in order to
define population risk categories for HSV-2 (Herpes
Simplex Virus type 2) on the basis of the
different socio-demographic factors. This technique
consists in successively splitting the whole set of
subjects into increasingly homogeneous subsets,
separating as far as possible those with a positive
HSV-2 status from those with a negative
HSV-2 status. Each split is conditioned by an
optimal cut-off in the levels of a single factor and
the process is continued as long as it is feasible
or statistically significant. Tree-based models
capture interactions between factors more easily
than logistic regression models. This procedure
generated subsets of subjects with increasing
seroprevalence rates, i.e. reflecting different risk
levels of acquiring HSV-2. Neither HSV-1
serostatus nor sex were used as discriminating
factors because we wanted to analyse sex-specific
differences in HSV-2 seroprevalence rates
according to HSV-1 serostatus across these
risk categories.
Ethical considerations
Serum samples had been
drawn from the volunteers under a consent involving
cardiovascular research but not explicitly for
assays such as Herpes simplex antibodies.
However, taking into account the potential
relationship between Herpes simplex infection
and atherosclerosis, the anonymisation of the
database and the fact that the serum samples had
been drawn 9 years before performing the study, the
performance of HSV-1 and HSV-2 assays
on this serum repository was authorised by the
Institutional Review Board of the School of
Medicine, CHUV, Lausanne.
Results
Characteristics of the study population
Of the 3235 serums
available in the serum repository, 3120 could be
matched in the MONICA database with one of the 3252
participants of the health examination survey.
Fifteen samples had an identification number not
corresponding to any of the participants probably
because of transcription errors. On the other hand,
132 participants had no serum sample in the
repository because of problems in serum collection
(n = 107, missing values for blood lipids as well)
or serum storage (n = 25).
Prevalence of antibodies against
HSV-1 and -2 infection
Overall, for the pooled
data concerning adults in the age range 3564, the
infection rate was 80.0 % for HSV-1 (Herpes
Simplex Virus type 1) and 19.3 % for HSV-2
(Tables
1
and
2). Infection rate
was higher in women than in men for HSV-1
(82.7% versus 77.2 %, p = 0.004) and for HSV-2
(20.9 % and 17.7 %, p = 0.073), but the difference
was statistically significant for HSV-1 only
(Tables
1 and
2). Seroprevalence
rates were higher in the Ticino region than in the
Vaud-Fribourg region for HSV-1 (87.7 % versus
77.0%, p < 0.001), as was HSV-2 prevalence
(20.5% versus 18.9 %, p = 0.309), but statistical
significance was reached for HSV-1 for both
genders and for HSV-2 for females only (see
also Table
5, multivariate analysis).
Herpes Simplex Virus
seroprevalence and variables other than age
For sociodemographic
variables other than age, the association with
HSV infection was analyzed using the pooled data
of both regions within the age range 3564.
Sex-specific seroprevalence rates and unadjusted
odds-ratios are shown in Tables
1 and
2. Adjusted odds
ratios, 95% confidence intervals and p values
resulting from multivariate logistic regression
models of HSV-2 seroprevalence are shown on
Table
5.
HSV seroprevalence and
marital status
Marital status was
associated with HSV infection in both sexes. In
males, a lower HSV-2 seroprevalence was
observed in married subjects compared to all other
categories, significantly so in widowed subjects as
compared to married subjects (Table
1). In females,
HSV-2 seroprevalence was significantly higher in
all categories (single, separated/divorced, or
widowed) as compared to married subjects (Table
2).
This observation was supported by the multivariate
analysis (Table
5, p = 0.013 and
0.0004 for males and females respectively by the
Wald test for the overall contribution of the
marital status variable).
HSV seroprevalence and socio-economic status
Several socio-economic
variables were collected in the MONICA survey: level
of education, economic sector, employment grade and
job qualification level. While similar trends in HSV
seroprevalence rates were observed according to all
these variables, the level of education showed the
strongest association and was therefore selected for
the present analysis.
The
seroprevalence of HSV-1 (Herpes Simplex
Virus type 1) decreased with increasing level of
education in both genders (Tables
1 and
2). In contrast,
the seroprevalence of HSV-2 increased
significantly with increasing level of education
(Tables
1 and
2). In males, the
odds of being infected by HSV-2 were higher
in subjects with high or very high education levels
(see definition in material and methods) compared
with subjects with low or medium education levels
(Table
1), while in
females, subjects with a low education level stood
out with a low HSV-2 seroprevalence as
compared with all higher education level subjects
(Table
2). Again, this
observation withstood the test of the multivariate
analysis for males, and nearly so for females (Table
5, p = 0.045 and
0.064 for males and females respectively by the Wald
test for the overall contribution of the education
level variable).
Herpes Simplex Virus
seroprevalence and size of place of residence
As the two-stage
sampling procedure of the MONICA survey consisted
first in drawing a random sample of communes (see
definition under method section) stratified
according to their size, we could analyze the
association between commune size and HSV
seroprevalence. As seen in Tables
1 and
2, HSV-2
seroprevalence was lower among residents from small
communes (<1500 inhabitants), when compared to
residents from larger towns. This trend for a lower
odd ratio of being infected by HSV-2 was
significant among females. Again, this conclusion
was supported by the multivariate analysis (Table
5, p = 0.53 and
0.02 for males and females respectively by the Wald
test for the overall contribution of the place of
residence variable).
Interaction between HSV-1
and HSV-2 infection
HSV-1(Herpes
Simplex Virus type 1)-seropositive individuals
had a lower HSV-2 seroprevalence rate
compared to HSV-1-uninfected subjects, and
this difference was statistically significant in
women and of borderline significance in men (Tables
1 and
2). The inverse
association between HSV-1 and HSV-2
serological status persisted after adjustment for
age and the other sociodemographic variables in
females (Table
5, p = 0.30 and
0.006 for males and females respectively by the Wald
test for the overall contribution of the HSV-1
serostatus variable). The potential role of sex and
HSV-1 status in protection against HSV-2
transmission was more closely examined in population
risk groups for HSV-2 infection defined by
classification tree analysis (see methods). Young
adults 25 to 34 years old and living alone formed
the lowest risk group (HSV-2 seroprevalence:
3.9%) and those living with a partner formed the
second lowest (13.3%).
The highest
risk group comprised the adults aged 35 years and
more who were separated, divorced or widowed
(31.6%) while those single or married had an
intermediate risk depending of their educational
level (less than high: 17.0%; high or very high:
23.6%). Figure
1 shows the HSV-2 seroprevalence rate in
these five groups according to sex and HSV-1
serostatus. The protective effect of HSV-1
infection against HSV-2 was mostly
contributed by women in the high risk group for
HSV-2 infection with an odds ratio of 0.37 (95%
confidence interval 0.190.75; p = 0.005).
Herpes Simplex Virus seroprevalence in
pregnant women
Women in the MONICA
study were asked whether they had been pregnant
within the twelve months before participating to the
study. Among those 55 answering that they had been,
72.7% were seropositive for HSV-1 and 9.1%
for HSV-2.
Herpes Simplex Virus prevalence and age
Since a larger age range
of participants was sampled in the Vaud-Fribourg
region, the relationship between age and Herpes Simplex Virus
seroprevalence was analysed in detail in this
region.
