Married couples' risk of same disease: cross sectional study

Hippisley-Cox et al (1) observed significant spouse similarities for
disease in a large sample of 8386 couples recruited through general
practice. They suggest shared environmental factors may cause these
similarities, but reject assortative mating as an explanation.

In a sample of the Netherlands Twin Register (2), we could not
replicate their spouse similarities for asthma, depression, diabetes and
cardiovascular disease, possibly due to our smaller sample size of 2152
spouse pairs. However, when examining health behaviour in a larger sample
we found good associations between spouses for smoking, alcohol problems
and exercise behaviour, even after controlling for age and BMI of both
spouses. Interestingly, the duration of the relationship influenced these
associations between the spouses (see figure). Except for alcohol
problems, spouse similarities in health behaviour decreased as the
duration of the relationship increased. This would suggest that assortment
for these factors is based on similarity at the time dating began (3) and
highlights the importance of determining similarities in disease status at
the time of dating, as suggested by Hippisley-Cox et al.

Figure: Risk of smoking (current smoking and ever smoked), exercise (regular sport participation and cycling) and alcohol problems in women whose partner had the same lifestyle compared to those whose partner did not. The logistic regression analyses were adjusted for age and body mass index of both males and females.

Assortative mating may further be based upon social factors and
personality traits. Indeed, in our sample we found significant spouse
correlations for educational attainment, an indicator of social economic
status, which is also related to disease development. These spouse
correlations increased as the duration of the relationship increased
(r=.292, r=.356, r=587 for <_5/>5 and > 15 yrs duration
respectively), possibly due to convergence of phenotypes of the spouses or
to a higher divorce rate in dissimilar pairs (3). Significant spouse
correlations were also found for inhibition, a personality trait
associated with increased risk behaviour, but these correlations were
unaffected by duration of relationship (r=.386, r=.334, r=.373 for <_5/>5 and > 15 yrs duration respectively). These results suggest
different mechanisms underlie spouse similarities for health behaviour,
social factors and personality traits. However, the fact that spouse
similarities were found for this wide range of variables indicates that
assortative mating should not be hastily dismissed as a cause for spouse
similarities in disease.

It should be noted that any association between spouses does not
exclude genetic effects. Hippisley-Cox et al. (1) assume that because
spouses are unrelated, genes do not influence the association. However,
the similarity of spouses may be an example of an active genotype-
environment correlation which occurs when a particular genotype is
associated with the selection or creation of a particular environmental
circumstance (4).

Gonneke Willemsen. Lecturer, epidemiology.

Jacqueline M. Vink. Researcher, genetic epidemiology.

Dorret I. Boomsma. Professor, genetic epidemiology.

Dept. of Biological Psychology,
Vrije Universiteit Amsterdam,
Van der Boechorststraat 1,
1081 BT Amsterdam,
The Netherlands

1. Hippisley-Cox, J Coupland, C., Pringle, M., Crown, N., Hammersley,
V. Married couples’ risk of same disease: Cross-sectional study. BMJ 2002:
325; 1-5. (21 Sept)

2. Boomsma, D.I., Vink, J.M., Beijsterveldt, C.E.M.v., Geus de, E.J.C.,
Beem, A.L., Mulder, E.J.C.M., Riese, H., Willemsen, A.H.M., Bartels, M.,
Berg van den, M., Derks, E.M., Graaff, S.C., Kupper, H.M., Polderman,
J.C., Rietveld, M.J.H., Stubbe, J.H., Knol, L.I., Stroet, T. and Baal,
G.C.M. (in press). “Netherlands Twin Register: a focus on longitudinal
research.” Twin Research.

3. Price, R.A. and Vandenberg, S.G. (1980). “Spouse similarity in
American and Swedish couples.” Behavior Genetics 10(1): 59-71.

4. Rowe, D.C. (2002). Assesing genotype-environment interactions and
correlations in the postgenomic era. Behavioral Genetics in the
postgenomic era. R. Plomin, Defries, J.C., Craig, I.W., McGuffin, P.
Washington, American Psychological Association: 71-86.

The results presented by Hippisley-Cox et al [1] make me wonder
whether infectious agents might be more important in causing the various
diseases they mention than we have previously thought. After all, we have
only understood relatively recently about the importance of H pylori in
peptic ulcer disease. The role of C pneumoniae infection in heart disease
is still not completely understood, but could turn out to be important
[2]. Perhaps there are many more diseases out there with infectious
aetiologies waiting to be discovered. This would certainly explain high
concordance among married couples.

1. Hippisley-Cox J, Coupland C, Pringle M, Crown N, Hammersley V.
Married couples' risk of same disease: cross sectional study. BMJ 2002;
325: 636

2. Leinonen M, Saikku P. Evidence for infectious agents in
cardiovascular disease and atherosclerosis. Lancet Infect Dis 2002; 2: 11-
7

The study on married couples’ risk of same disease is of large
interest as it adds important and otherwise scarce evidence on the impact
of shared environmental exposures on the prevalence of common diseases.
However, there are some limitations that not addressed by the Authors.

One is the problem of the hidden time effects in the case of blood
pressure measurements, which were performed only once. As Altman (1999:79)
concludes, the blood pressure is a particularly variable measurement. It
varies considerably between individuals but also over time for each
individual. There is marked variation over 24 hours (circadian variation)
as well day-to-day variation. It is also difficult to measure blood
pressure. The method used in the study was not mentioned at all. There are
new techniques to perform measurements via an indwelling catheter in the
arm linked to a small tape recorder, which provides more accurate
information. Observer bias might also affect the results of blood pressure
measurements. Since the blood pressure has a strong circadian rhythm being
highest in the morning the advice is commonly to take repeat measurements
of blood pressure from the same subject at the same time of day (Altman
1999:148). The Authors did not provide this information and the
conclusions on the increased risk of hypertension shared by couples might
thus be biased.

The problem of the hidden time effects is also present due to the
impact of seasons on the utilisation of care due to asthma and chronic
obstructive disease. Both individual and population data might vary by
month of the year, e.g. levels of IgE are affected by the seasons.

We might also expect that there are variations in the practice among
the participating general practices such as various routines for follow-up
visits, i.e. the way in which the secondary demand on health care is
upcoming, which might contribute to various patterns of utilisation. The
problem of utilisation was partly mentioned by the Authors who pointed to
the fact that they did not use consultation data and could not adjust for
the different frequencies at which some groups of patients consult they
general practitioner. Such differences might also arise e.g. due to
distance to the general practices from the patients place of residence,
due to non-random environmental factors, i.e. geographical differences
affecting availability.

It would be of interest to examine if the same factors affect the
married couples in Sweden, whether there might be systematic health care
systems effects, i.e. the degree of coverage of travel costs by the health
insurance system could contribute to the various utilisation patterns in
the two countries which would affect the shared environmental risks of
couples.

Reference:
Douglas G Altman (1999). Practical statistics for medical research.
Chapman & Hall/CRC, USA (Originally published in London, 1991).