Download Sibship structure and risk of infectious

International Journal of Epidemiology, 2014, 1607–1614
doi: 10.1093/ije/dyu118
Advance Access Publication Date: 11 June 2014
Original article
Miscellaneous
Sibship structure and risk of infectious
mononucleosis: a population-based
cohort study
Klaus Rostgaard,1* Trine Rasmussen Nielsen,1,2 Jan Wohlfahrt,1
Henrik Ullum,3 Ole Pedersen,4 Christian Erikstrup,5 Lars Peter Nielsen6
and Henrik Hjalgrim1
1
Department of Epidemiology Research and 2Department of Regulatory and Medical Affairs,
Statens Serum Institut, Copenhagen, Denmark, 3Department of Clinical Immunology, Copenhagen
University Hospital, Copenhagen, Denmark, 4Department of Clinical Immunology, Naestved Hospital,
Naestved, Denmark, 5Department of Clinical Immunology, Aarhus University Hospital, Aarhus,
Denmark and 6Department of Microbiologic Diagnostics and Virology, Statens Serum Institut,
Copenhagen, Denmark
*Corresponding author. Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, DK-2300
Copenhagen S, Denmark. E-mail: [email protected]
Accepted 19 May 2014
Abstract
Background: Present understanding of increased risk of Epstein-Barr virus (EBV)-related
infectious mononucleosis among children of low birth order or small sibships is mainly
based on old and indirect evidence. Societal changes and methodological limitations of
previous studies call for new data.
Methods: We used data from the Danish Civil Registration System and the Danish
National Hospital Discharge Register to study incidence rates of inpatient hospitalizations
for infectious mononucleosis before the age of 20 years in a cohort of 2 543 225 Danes
born between 1971 and 2008, taking individual sibship structure into account.
Results: A total of 12 872 cases of infectious mononucleosis were observed during 35.3
million person-years of follow-up. Statistical modelling showed that increasing sibship
size was associated with a reduced risk of infectious mononucleosis and that younger
siblings conferred more protection from infectious mononucleosis than older siblings. In
addition to this general association with younger and older siblings, children aged less
than 4 years transiently increased their siblings’ infectious mononucleosis risk. Our results were confirmed in an independent sample of blood donors followed up retrospectively for self-reported infectious mononucleosis.
Conclusions: Younger siblings, and to a lesser degree older siblings, seem to be important in the transmission of EBV within families. Apparently the dogma of low birth order
in a sibship as being at the highest risk of infectious mononucleosis is no longer valid.
Key words: Epstein-Barr virus, infection pattern, delayed infection, birth order, Denmark
C The Author 2014; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association
V
1607
1608
International Journal of Epidemiology, 2014, Vol. 43, No. 5
Key Messages
• Within sibships each additional sibling, especially younger siblings, is associated with lowered risk of infectious
mononucleosis.
• The smaller the age difference between sibs, the larger the mutual reduction in risk of infectious mononucleosis.
• An additional sibling may be associated with up to 25% reduction in risk of infectious mononucleosis, despite the nu-
merous alternative sources of EBV transmission
• Similar results have been shown for risk of multiple sclerosis and EBV-related Hodgkin lymphoma.
Introduction
Infection with Epstein-Barr virus (EBV) is ubiquitous, and
by adulthood 90% or higher of persons are infected, as
illustrated by the prevalence among controls in numerous
case-control studies.1–4 Primary infection with EBV in
childhood is usually asymptomatic or accompanied only
by mild symptoms. On the other hand, primary EBV infection in adolescence or adulthood is often accompanied by
infectious mononucleosis (IM), the estimated proportion in
restricted populations ranging from 25% to 77%.5,6 All
this suggests that host factors, e.g. genetic constitution, in
interaction with age-dependent environmental and behavioural factors, influence risk of IM. These host factors and
age-dependent factors are very incompletely understood,
e.g. the mechanism(s) of EBV transmission among preadolescent children are not well known.7
Studies have shown that sibship structure and interaction may influence the age at which individuals become
exposed to infectious agents.8,9 For EBV, current understanding proposes that age at primary infection, and by implication the risk of IM, is higher for children of lower
birth order and smaller sibships. This understanding
has mainly been articulated in papers on cancer and autoimmune diseases, see e.g. Westergaard et al.,10,11 Vianna
et al.,12 Hernan et al.13 and references in these.
Remarkably, this interpretation seems to rest primarily on
inferences made from studies of other infectious agents,
such as hepatitis A, hepatitis B and polio virus,14–16
whereas direct evidence is rather limited.17 Moreover, in
many Western societies marked secular changes have taken
place during recent decades, which may have changed the
role of family structure in infectious disease pressure in
childhood. For instance, the proportion of children attending pre-school has vastly increased,18 pre-school exposure
replacing older sibs as an important source of infections.19,20 Accordingly, older findings for infections other
than EBV may no longer be applicable to the epidemiology
of clinical EBV infections in general and its presumed association with sibship structure in particular. Based on many
epidemiological observations, we have hypothesized the
existence of a set of common risk factors for IM, EBVpositive Hodgkin lymphoma and multiple sclerosis
that include constitutionally determined ‘inadvertent’
immunological responses to EBV.21 If that hypothesis is
true, putative associations between sibship structure and
IM risk will also be of relevance and an inspiration in
aetiological studies of Hodgkin lymphoma and multiple
sclerosis.
We undertook the present study to investigate the significance of sibship structure on the age-specific risk of IM
at age 0–19 years in a large population-based cohort in
Denmark.
Methods
The nationwide Danish Civil Registration System (CRS)
was implemented on 1 April 1968. All Danish citizens
have since been assigned unique identification numbers
(the CRS number), by which the CRS continuously monitors individual vital status, emigration status, identity of
parents and residence.22
Using the CRS, we established a cohort of all Danish
singletons born between 1 January 1971 and 31 December
2008. For each cohort member we identified sibs, defined
as children born to the same mother as the individual. For
all persons, information about dates of birth was obtained
to allow the construction of variables such as the age difference between a proband and an exposing sib.
The Danish National Hospital Discharge Register was
established in 1977 and has since recorded 99.9% of all
discharges from Danish non-psychiatric hospitals.23,24 For
each hospitalization, the register contains information on
dates of admission and discharge, and discharge diagnoses
classified according to the International Classification of
Diseases, 8th revision (ICD-8) from 1977 to 1993, and the
10th revision (ICD-10) thereafter. In the Register we identified all inpatient hospitalizations of IM, classified as code
075* in ICD-8 and code B27* in ICD-10.
The study was approved by the Danish Data Protection
Board, J.nr 2008-54-0472. According to Danish law,
International Journal of Epidemiology, 2014, Vol. 43, No. 5
informed consent is unnecessary for conducting a purely
register-based study.
Statistical analysis
The cohort of singletons was followed for hospitalization
for IM from 1 January 1977 or from date of birth, whichever occurred last, until the date of diagnosis of IM, death,
emigration, age 20 or 31 December 2008, whichever came
first. Exclusively in incidence calculations (Figures 1 and
2), the cohort was extended to include all persons in the
CRS born since 1957 and alive in 1977.
Log-linear Poisson regression was used to model incidence rates and rate ratios (RR) and thereby assess the effects of sib-ship structure. The variables investigated
included the number of sibs with a specific age and the
number of sibs with a specific signed age interval to the
proband. At the crudest, the latter amounted to counting
the number of older and younger sibs of the proband. All
these variables were treated as time-dependent variables,
e.g. changing when new members were born into the sibship. All analyses were adjusted for birth cohort (calendar
year of birth), sex-specific age (1-year groups) and sexspecific period (calendar year). Of importance, the identification problems normally relevant to the simultaneous
inclusion of these three intimately related time scales for
effect estimation are irrelevant for confounder control.
Based on present understanding of the epidemiology of
EBV and consistent with the incidence data, risk factor
125
Females
Males
1609
analyses were carried out for children aged less than 7
years and for older children and adolescents separately as
well as combined.
As a preliminary step, we analysed data restricted to
children being followed up while they had at most one sibling, to find out which sibling variables and categorizations
of these to use. The starting point of this exercise was a detailed table of rate ratios by the age of the child crossed
with the age of the child’s sibling. This set-up is similar to
that of age-period-cohort analyses, and similar techniques
were used to obtain a parsimonious and interpretable data
description. To limit the number of parameter estimates to
present while, on the other hand, having no a priori hypothesis, we used the corrected Akaike criterion (AICC)
and the Bayesian information criterion (BIC) with the
number of outcome events as n,25 both yielding a post hoc
justification of our choices. The final set of predictors was
number of sibs aged 0, 1, 2 and 3 years and the number of
sibs with a specific age interval to the proband.
All analyses were performed using the SAS statistical
software package (versions 9.2 and 9.3; SAS Institute,
Cary, NC, USA); 95% confidence intervals (CIs) were
based on Wald tests. Tests for effect modification were
evaluated by inclusion of an interaction term, and all tests
were likelihood-ratio based.
175
1977−1986 females
1987−1996 females
1997−2008 females
1977−1986 males
1987−1996 males
1997−2008 males
150
125
100
100
75
75
50
50
25
25
0
0
0
10
20
30
AGE
Figure 1. Age- and sex-specific incidence of hospitalizations for infectious mononucleosis per 100 000 person-years, based on follow-up of
all Danes born in 1957 and later.
0
10
20
30
age
Figure 2. Age- and sex-specific incidence of hospitalizations for infectious mononucleosis per 100 000 person-years in three time periods,
based on follow-up of all Danes born in 1957 and later.
1610
International Journal of Epidemiology, 2014, Vol. 43, No. 5
Table 1. Demographic characteristics of the study cohort by gender, year of diagnosis and number of younger and older sibs
Characteristics
Age 0-6 years
Number of cases
Gender
Male
Female
Year of diagnosis
1977-86
1987-96
1997-2008
Number of younger siblings
0
1
2
3þ
Number of older siblings
0
1
2
3þ
Age 7-19 years
Person-years
Number of cases
Person-years
1 696 (62%)
1 033 (38%)
7 506 373 (51%)
7 153 702 (49%)
4 607 (45%)
5 536 (55%)
10 553 733 (51%)
10 065 941 (49%)
773 (28%)
981 (36%)
975 (36%)
4 512 871 (31%)
4 477 039 (31%)
5 670 166 (39%)
873 (9%)
4 500 (44%)
4 770 (47%)
2 951 394 (14%)
7 776 257 (38%)
9 892 023 (48%)
1 886 (69%)
765 (28%)
74 (3%)
4 (0.1%)
10 985 531 (75%)
3 305 069 (23%)
337 506 (2%)
31 969 (0.2%)
5 312 (52%)
3 503 (35%)
1095 (11%)
233 (2%)
9 895 256 (48%)
7 642 617 (37%)
2 391 608 (12%)
690 192 (3%)
1 305 (48%)
988 (36%)
322 (12%)
114 (4%)
6 595 641 (45%)
5 372 110 (37%)
1 963 720 (13%)
728 604 (5%)
4 361 (43%)
3 995 (39%)
1 376 (14%)
411 (4%)
9 340 658 (45%)
7 549 295 (37%)
2 724 443 (13%)
1 005 277 (5%)
Results
The distribution at person-years of risk and the observed
number of outcomes in the studied cohort of singletons are
shown in Table 1. Overall, 12 872 persons were hospitalized for IM during 35.3 million person-years of follow-up
contributed by 2 543 225 Danes. Age- and gender-specific
incidence rates of IM are shown in Figure 1. IM displayed
a bimodal age-specific incidence rate distribution in boys
with peaks around the ages of 4 and 16 years, whereas for
girls, the age-specific incidence rate distribution of mononucleosis was unimodal and peaked at 15 years. The
4-year peak for boys is statistically ‘real’, e.g. a trend test
restricted to age 4–10 years yields a declining incidence
rate of 10% per year (RR ¼ 0.90, 95% confidence interval ¼ 0.88–0.92; P ¼ 3 x 1017). The age- and sex-specific
incidence rates of IM did not change materially over the
study period, see Figure 2.
Table 2 shows the rate ratio of IM according to the
number of sibs aged 0, 1, 2 and 3 years and the number of
sibs with a specific age interval to the proband for two age
categories of the proband. Having a very young sib conveyed a substantial increase in relative risk of IM, with risk
estimates around 1.5. Otherwise, each extra sib conveyed a
multiplicative protective effect which varied by sibling age
interval. Hence, the protective effect was largest if the sib
was around the same age as the proband, and negligible if
the sib was at least 9 years older than the proband. A reduction of the detailed categorization of age interval to
each sib to just counting the number of younger and older
sibs was not acceptable statistically (0–6 years old
P ¼ 0.02, 7–19 years old P ¼ 0.0001, 0–19 years old
P ¼ 0.00002); however, for completeness these estimates
are also presented together with a birth order effect estimate in Table 2. Data were compatible with a common set
of sibling parameters for the two categories of attained
age, i.e. 0–6 and 7–19 years (P ¼ 0.06), which therefore is
the model used and referred to later. We found no interaction between calendar period (1977–86, 1987–96,
1997–2008) and sibship structure (P ¼ 0.26, data not
shown), nor between sex and sibship structure (P ¼ 0.77,
data not shown).
Discussion
The present analyses showed that family structure has retained importance to the epidemiology of IM and therefore
EBV infection/transmission regardless of the changing roles
of sibs and other children in modern affluent societies. An
increasing proportion of children attend day care, which
constitutes an important source of infection for other infectious diseases.19,20 This should imply a trend towards earlier infection with EBV and thus a decreased occurrence of
IM in adolescents. Despite having a 31-year-long study
period, the incidence pattern of hospitalized adolescent
cases did not markedly change over time, as seen in other
hospitalized or monospot-tested populations including
adults.26–28
Under the assumption that the observed age-specific
rate of hospitalized IM mirrors (by proportionality) the
pattern for IM in general, see Visser et al. for a recent
International Journal of Epidemiology, 2014, Vol. 43, No. 5
1611
Table 2. Rate ratio (RR) of infectious mononucleosis by sibship characteristics. Example: if a boy has a one-year old sib who is
three years younger than him and two sibs who are both six years older than him his RR of infectious mononucleosis will be
1.57x0.82x0.96x0.96 ¼ 1.19 compared to a boy with the same characteristics (age, period, sex, birth cohort), but no sibs at all
Sibship
Per sib aged 0-3 years
Age 0
Age 1
Age 2
Age 3
Per sib by age interval
At least 9 years younger
6-9 years younger
3-5 years younger
0-2 years younger
0-2 years older
3-5 years older
6-8 years older
At least 9 years older
Summary effects of sibs
Number of younger sibsa
Number of older sibsa
Birth orderb
Age 0-6 years
RR (95% CI)
Age 7-19 years
RR (95% CI)
Age 0-19 years
RR (95% CI)
1.04 (0.82–1.32)
1.45 (1.17–1.81)
1.21 (0.98–1.49)
1.05 (0.87–1.26)
1.14 (0.95–1.37)
1.44 (1.24–1.67)
1.55 (1.36–1.77)
1.22 (1.07–1.39)
1.20 (1.07–1.34)
1.57 (1.42–1.73)
1.44 (1.31–1.59)
1.16 (1.05–1.28)
NA
1.19 (0.69–2.03)
1.01 (0.81–1.27)
0.80 (0.66–0.95)
0.81 (0.74–0.90)
0.94 (0.87–1.02)
0.90 (0.81–1.00)
1.02 (0.95–1.10)
0.81 (0.75–0.87)
0.85 (0.80–0.90)
0.81 (0.77–0.85)
0.76 (0.72–0.80)
0.89 (0.85–0.94)
0.94 (0.90–0.98)
0.97 (0.92–1.02)
0.96 (0.93–1.00)
0.81 (0.76–0.87)
0.85 (0.80–0.90)
0.82 (0.79–0.86)
0.75 (0.72–0.79)
0.87 (0.84–0.91)
0.94 (0.90–0.97)
0.96 (0.91–1.00)
0.98 (0.94–1.01)
0.88 (0.74–1.04)
0.93 (0.89–0.98)
0.93 (0.89–0.97)
0.80 (0.78–0.83)
0.95 (0.93–0.97)
1.00 (0.98–1.03)
0.80 (0.78–0.83)
0.95 (0.93–0.96)
0.99 (0.97–1.01)
All estimates are mutually adjusted and adjusted for age, gender, calendar period and birth cohort.
NA, not applicable.
a
Obtained by replacing the ‘age interval to proband’ variables in the model.
b
Obtained by replacing all the sibship variables in the model.
example,28 our observations suggested that young siblings
aged less than 4 years constitute an important source of
primary EBV infections within families. Accordingly, the
statistical analyses showed that exposure to a younger
sibling in this age range was associated with a transiently
increased risk of IM. Most likely, the exposure to
younger siblings aged 0–3 years also carries an increased
risk of asymptomatic EBV infection. This would explain
why, in the long run, exposure to a younger sibling was
associated with a decreased risk of IM. Specifically, by
being infected by the younger sibling, the older sibling
(proband) would no longer be at risk of IM in adolescence when it (IM) is most common. The long-term protective effect varied by the age difference between siblings
in such a way that the farther apart from each other in
age, the less likely they were to confer protection on each
other. Presumably the closer in age, the more likely sibs
are to be in contact with each other in a way conducive
to EBV transmission.
Obviously contagion can go in both directions, i.e. from
an older to a younger sibling, as also suggested by the analyses showing a reduced risk of IM by older siblings displaying a qualitatively similar sibling interval-dependent
variation as the younger sibling effect. Interestingly, the
estimates of the protective effect of sibs in different age
intervals relative to the proband suggested that it is more
common that a younger sib infects an older sib (giving rise
to asymptomatic disease) than vice versa. This interpretation is consistent with that of two recent studies reporting
that sibling contact reduced the risk of IM29 and increased
the risk of EBV seropositivity,30 respectively. Moreover, it
is also supported by the distribution of older and younger
siblings in young persons tested for primary EBV infection
at Statens Serum Institut (see Supplementary data available
at IJE online).
If the hypothesis of a set of common risk factors for IM,
EBV-positive Hodgkin lymphoma and multiple sclerosis21
is true, then the results of the present study are also of relevance for the study of Hodgkin lymphoma and multiple
sclerosis. The common use in epidemiological studies of
birth order as a proxy for childhood infections seems to
be inferior to counting the number of older as well as
the number of younger sibs, mutually adjusted.31 In the
present study, birth order, i.e. number of older sibs, carried
a much weaker effect than the number of younger sibs, to
which it was linked in unpredictable ways. Moreover, we
found no overall trend in relation between birth order and
IM when not adjusting for other sibship characteristics.
We also note that large recent studies29,32 of sibship structure and risk of multiple sclerosis that have made the
1612
distinction between older and younger siblings have found
associations very similar in size and direction to ours, not
necessarily appreciating the finding. Similarly, a stronger
protective role for younger siblings than for older siblings
has been found in EBV-related Hodgkin lymphoma in
younger adults, whereas neither of these factors was associated with risk of EBV-unrelated Hodgkin lymphoma.33
Little is known about risk factors for IM and hence potential confounders of the studied relations. It has been
hypothesized that the individual’s response to EBV is influenced by a number of factors, including age at infection,
infectious load, efficiency of the immune system in controlling the virus infection, and the genetic composition of the
individual.34 Socio-economic affluence in childhood has
previously been implicated as a risk factor for IM in adolescence.17 To some extent this association has been
believed to reflect the presumed association with small sibship size and isolated upbringing and thus delayed exposure to EBV. Presumably sibship structure is intimately
linked with age at infection and infectious load in such a
way as to make these mediators between sibship structure
and IM rather than confounders to be adjusted for.
Hypothetically, genes linked to sibship structure (e.g. regulating fecundity) could be in linkage disequilibrium with
genes involved in, for example, handling EBV infection.
However, very little is known about the genetics of IM at
the time of writing.7,21 The efficiency of the immune system in controlling the virus infection will be determined
partly by genetics, partly by the timing, type and size of
various stimuli.35 Again we consider the latter factors
mostly mediators of the association between sibship structure and IM.
The Danish population is ethnically very homogeneous;
Denmark has one of the most equal income distributions in
the world;36 and hospital treatment is free in Denmark.24
All this mitigates hospitalized IM cases being disproportionately recruited from well-connected affluent social
classes with low thresholds for seeking medical attention.
All in all, we see little potential for confounding of our results by any of these risk factors.
Our study has a number of strengths and limitations to
consider. Among the strengths are the population-based
setting and the size of the studied cohort. Also, secular
changes in patterns and severity of infection could be taken
into account by adjusting for possible calendar period
and birth cohort effects. The age distribution of these
hospitalized IM cases conformed to expectations regarding
the full population of all incident IM cases. We can think
of only two realistic limitations of the present study. First,
many parents would agree that the threshold for reacting
to disease symptoms in their offspring is increased with
increasing birth order.37,38 If so, this would translate into
International Journal of Epidemiology, 2014, Vol. 43, No. 5
later-born siblings being at lower risk of being hospitalized
in case of getting IM than earlier-born. What we have
found in general is the opposite. This means that our finding of younger siblings as being more protective than older
siblings cannot be an artefact of this hypothesized birth
order-dependent hospitalization threshold effect. Post hoc
modelling yielded the same conclusion about the absence
of a birth order effect in the hypothesized direction (data
not shown). By the same token, although we did not validate the diagnoses of IM, diagnostic misclassification is unlikely to influence our findings materially. The fraction of
misclassified IM cases in the Danish National Patient
Register is likely to be substantial, but no specific study on
this topic exists. Second, if we apply our estimates of the
effect of the 0–3-year-old siblings as direct sources of
hospitalized IM, to IM in general, the result may be exaggerated. According to Aaby and coworkers (see e.g.
Nielsen39), a case of infectious disease is more likely to be
severe when the source of the infection is inside the household than outside the household, due to prolonged and
close contact with the source of contagion, leading to exposure to a larger dose of the infectious agent. So a severe,
i.e. hospitalized, case of IM would be more likely to originate from within-family contagion than a mild case of IM.
However, a comparison with the association between
occurrence and age at self-reported IM and sibship structure in a cohort of 18 906 Danish blood donors40 showed
remarkably similar magnitude and direction of associations between the two cohorts (see Supplementary data,
available at IJE online). Thus, our results are unlikely to
be an artefact of using exclusively hospitalized IM as
outcome.
Conclusion
In conclusion, family structure has retained importance to
the epidemiology of IM regardless of the changing roles of
sibs and other children in modern affluent societies. We believe that the presented model captures the typical modes
of infection in an environment that is affluent yet with exposure to many other children very early in childhood, e.g.
in kindergartens. We are therefore confident that the most
important finding of this study, that increasing sibship size,
especially by younger sibs, reduces the risk of IM in adolescence, is true of hospitalized IM as well as IM in general,
in modern affluent societies. Seemingly, the dogma of the
firstborn in a sibship as being at the highest risk of IM is
no longer valid.
Supplementary Data
Supplementary data are available at IJE online.
International Journal of Epidemiology, 2014, Vol. 43, No. 5
Funding
This work was supported by the Danish Cancer Society [grant number DP08155] and the Danish Council for Independent Research
[grant number 09-072164].
Conflict of interest: None.
References
1. Hanlon P, Avenell A, Aucott L, Vickers MA. Systemic review
and meta-analysis of the sero-epidemiological association between Epstein-Barr virus and systemic lupus erythematosus.
Arthritis Res Ther 2014;16:R3.
2. Almohmeed YH, Avenell A, Aucott L, Vickers MA. Systematic
review and meta-analysis of the sero-epidemiological association
between Epstein-Barr virus and multiple sclerosis. PLoS One
2013;8:e61110.
3. Holl K, Surcel HM, Koskela P et al. Maternal Epstein-Barr virus
and cytomegalovirus infections and risk of testicular cancer in
the offspring: a nested case-control study. APMIS 2008;116:
816–22.
4. Sjöström S, Hjalmars U, Juto P, et al. Human immunoglobulin G
levels of viruses and associated glioma risk. Cancer Causes
Control 2011;22:1259–66.
5. Crawford DH, Macsween KF, Higgins CD et al. A cohort study
among university students: identification of risk factors for
Epstein-Barr virus seroconversion and infectious mononucleosis.
Clin Infect Dis 2006;43:276–82.
6. Balfour HH Jr, Odumade OA, Schmeling DO et al. Behavioral,
virologic, and immunologic factors associated with acquisition
and severity of primary Epstein-Barr virus infection in university
students. J Infect Dis 2013;207:80–88.
7. Balfour HH Jr. Genetics and infectious mononucleosis. Clin
Infect Dis 2014;58:1690–91.
8. Blaser MJ, Nomura A, Lee J, Stemmerman GN, Perez-Perez GI.
Early-life family structure and microbially induced cancer risk.
PLoS Med 2007;4:e7.
9. Lagiou P, Trichopoulos D. Parental family structure,
Helicobacter pylori, and gastric adenocarcinoma. PLoS Med
2007;4:e25.
10. Westergaard T, Andersen PK, Pedersen JB et al. Birth characteristics, sibling patterns, and acute leukemia risk in childhood: a
population-based cohort study. J Natl Cancer Inst 1997;
89:939–47.
11. Westergaard T, Melbye M, Pedersen JB, Frisch M, Olsen JH,
Andersen PK. Birth order, sibship size and risk of Hodgkin’s disease in children and young adults: A population-based study of
31 million person-years. Int J Cancer 1997;72:977–81.
12. Vianna NJ, Polan AK. Immunity in Hodgkin’s disease: importance of age at exposure. Ann Intern Med 1978;89:550–56.
13. Hernan MA, Zhang SM, Lipworth L, Olek MJ, Ascherio A.
Multiple sclerosis and age at infection with common viruses.
Epidemiology 2001;12:301–06.
14. Green MS, Zaaide Y. Sibship size as a risk factor for hepatitis A
infection. Am J Epidemiol 1989;129:800–05.
15. Hsieh CC, Tzonou A, Zavitsanos X, Kaklamani E, Lan SJ,
Trichopoulos D. Age at first establishment of chronic hepatitis
B virus infection and hepatocellular carcinoma risk. A birth
order study. Am J Epidemiol 1992;136:1115–21.
1613
16. Weinstein L. Influence of age and sex on susceptibility and clinical manifestations in poliomyelitis. N Engl J Med 1957;
257:47–52.
17. Niederman JC, Evans AS. Epstein-Barr virus. In: Evans AS,
Kaslow R.A (eds). Viral Infections of Humans. Epidemiology and Control. 4th edn. New York and London: Plenum,
1997.
18. UNESCO. Beyond 20/20 WDS. Report Folders. 2011. http://
stats.uis.unesco.org/unesco/ReportFolders/ReportFolders.aspx
(7 February 2012, date last accessed).
19. Kamper-Jørgensen M, Wohlfahrt J, Simonsen J, Grønbæk M,
Benn CS. Population-based study of the impact of childcare attendance on hospitalizations for acute respiratory infections.
Pediatrics 2006;118:1439–46.
20. Koch A, Homøe P, Pipper C, Hjuler T, Melbye M. Chronic suppurative otitis media in a birth cohort of children in Greenland:
population-based study of incidence and risk factors. Pediatr
Infect Dis J 2011;30:25-29.
21. Rostgaard K, Wohlfahrt J, Hjalgrim H. A genetic basis
for infectious mononucleosis: Evidence from a family
study of hospitalized cases in Denmark. Clin Infect Dis
2014;58:1684-89.
22. Pedersen CB. The Danish Civil Registration System. Scand J
Public Health 2011;39:22–25.
23. Andersen TF, Madsen M, Jørgensen J, Mellemkjær L, Olsen JH.
The Danish National Hospital Register. A valuable source of
data for modern health sciences. Dan Med Bull 1999;46:
263–68.
24. Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient
Register. Scand J Public Health 2011;39:30–33.
25. Hurvich CM, Tsai CL. Model selection for extended quasilikelihood models in small samples. Biometrics 1995;51:
1077–84.
26. Tattevin P, Le Tulzo Y, Minjolle S et al. Increasing
incidence of severe Epstein-Barr virus-related infectious
mononucleosis: surveillance study. Clin Microbiol 2006;
44:1873–74.
27. Morris MC, Edmunds WJ. The changing epidemiology of infectious mononucleosis? J Infect 2002;45:107–09.
28. Visser E, Milne D, Collacott I, McLernon D, Counsell C, Vickers
M. The epidemiology of infectious mononucleosis in Northern
Scotland: A decreasing incidence and winter peak. BMC Infect
Dis 2014;14:151.
29. Ponsonby AL, van der Mei I, Dwyer T et al. Exposure to infant
siblings during early life and risk of multiple sclerosis. JAMA
2005;293:463–69.
30. Higgins CD, Swerdlow AJ, Macsween KF et al. A study of risk
factors for acquisition of Epstein-Barr virus and its subtypes.
J Infect Dis 2007;195:474–82.
31. Grulich AE, Vajdic CM, Falster MO et al. Birth order and risk of
non-Hodgkin lymphoma — true association or bias? Am J
Epidemiol 2010;172:621–30.
32. Bager P, Nielsen NM, Bihrmann K, Frisch M et al. Sibship
characteristics and risk of multiple sclerosis: A nationwide cohort study in Denmark. Am J Epidemiol 2006;
163:1112–17.
33. Hjalgrim H, Smedby KE, Rostgaard K et al. Infectious mononucleosis, childhood social environment, and risk of Hodgkin
lymphoma. Cancer Res 2007;67:2382–88.
1614
34. Khan G, Miyashita EM, Yang B, Babcock GJ, Thorley-Lawson
DA. Is EBV persistence in vivo a model for B cell homeostasis?
Immunity 1996;5:173–79.
35. Cozen W, Hamilton AS, Zhao P et al. A protective role for early
oral exposures in the etiology of young adult Hodgkin lymphoma. Blood 2009;114:4014–20.
36. Wikipedia. List of countries by income equality. http://en.wikipedia.org/wiki/List\_of\_countries\_by\_income\_equality
(13
February 2012, date last accessed).
37. Søndergaard G, Biering-Sørensen S, Michelsen SI, Schnor O,
Andersen AM. Non-participation in preventive child health
International Journal of Epidemiology, 2014, Vol. 43, No. 5
examinations at the general practitioner in Denmark: a
register-based study. Scand J Prim Health Care 2008;
26:5–11.
38. Tessler R. Birth order, family size, and children’s use of physician
services. Health Serv Res 1980;15:55–62.
39. Nielsen NM, Hedegaard K, Aaby P. Intensity of
exposure and severity of whooping cough. J Infect
2001;43:177–81.
40. Pedersen OB, Erikstrup C, Kotzé SR et al. The Danish Blood
Donor Study: a large, prospective cohort and biobank of medical
research. Vox Sang 2012;102:271.