This article was published in an Elsevier journal. The attached copy
is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
Author's personal copy
Transactions of the Royal Society of Tropical Medicine and Hygiene (2008) 102, 207—208
available at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/trst
Modelling sexually transmitted infections: less is
usually more for informing public health policy
David G. Regan∗, David P. Wilson
National Centre in HIV Epidemiology and Clinical Research, The University of New South Wales,
316 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia
Available online 29 October 2007
disease transmission since Bernoulli’s smallpox modelling in 1760. Their use has become
widespread for exploring how epidemics can be prevented or contained. Here we discuss the
importance of modelling the dynamics of sexually transmitted infections, the technology-driven
dichotomy in methodology, and the need to ‘keep it simple’ to explore sensitivity, to link the
models to reality and to provide understandable mechanistic explanations for real-world policy-
makers. The aim of models, after all, is to influence or change public health policy by providing
rational forecasting based on sound scientific principles.
© 2007 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights
Mathematical models have been used to investigate the dynamics of infectious
Interventions aimed at preventing and/or containing infec-
tious disease outbreaks are generally expensive and their
outcomes are unknown prior to implementation. In addition
to providing theoretical insights into transmission dynamics,
mathematical models provide the only means of estimat-
ing the potential for intervention strategies to achieve
their intended aims. Models of sexually transmitted infec-
tions (STI) have been used since the early 1970s to study
their spread in populations. A widely used model of gonor-
rhoea transmission was one of the earliest to support the
still controversial hypothesis that so-called core groups are
responsible for the persistence of STIs (Yorke et al., 1978).
The emergence of the global HIV pandemic in the early 1980s
focused the attention of modellers as the need to estimate
its scope and to evaluate containment strategies became
paramount (May and Anderson, 1987). Whilst HIV models
∗Corresponding author. Tel.: +61 2 9385 0900;
fax: +61 2 9385 0920.
E-mail address: email@example.com (D.G. Regan).
have understandably dominated the STI modelling litera-
ture, other STI models are having significant influence in
public policy. For example, modelling has recently played an
integral role in the evaluation of mass vaccination strategies
against human papillomavirus, a STI associated with the vast
majority of cervical cancer cases (Barnabas et al., 2006).
Modelling is also being used in several countries to evaluate
the potential benefits of systematic screening for Chlamy-
dia trachomatis, a STI associated with serious sequelae in
women such as infertility and ectopic pregnancy.
In general, two distinct approaches have been adopted
for modelling STIs and the underlying processes of part-
nership formation necessary for transmission. One is the
deterministic approach where the population is com-
partmentalised and progresses through different states
according to defined rates. The other is the stochastic
approach where individuals and their contacts are explic-
itly tracked. Rapid advances in computational power have
driven a shift from population-based models towards com-
plex individual-based models. However, this shift has not
necessarily led to better models or a better modelling
0035-9203/$ — see front matter © 2007 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.
Author's personal copy
208 D.G. Regan, D.P. Wilson
paradigm. Deterministic models are simple to develop and
understand and they enable the use of a powerful set of
analytical tools. Simple compartmental models, based on
the Kermack—McKendrick SIR model, continue to be used
extensively and to provide transparent explanations of the
mechanisms by which diseases spread. They enable us to
predict the future course of an epidemic and to evaluate
strategies to contain their spread (Wilson et al., 2006).
Models of this type are not suitable for addressing circum-
stances for which there are only a ‘handful’ of individuals or
when individual-level intervention strategies (such as con-
tact tracing) or interactions (e.g. partner concurrency) are
the focus for investigation. However, the potential of the
individual-level approach is rarely achieved in a timely and
meaningful fashion. For large populations and well estab-
lished epidemics, compartmental models typically provide
as much, or more, insight than stochastic models.
With modern high performance computing, there is a
strong temptation to forfeit simple models for complex
stochastic ones that may be perceived as more ‘realistic’
or ‘lifelike’. There are many reasons why this temptation
should be resisted, whenever possible, for addressing most
practical real-world problems. First, it is extremely diffi-
cult to obtain reliable parameter estimates even for simple
models. In complex models there are manyfold more param-
eters to be estimated and many of them cannot be linked
easily to biologically or socially measurable outcomes. Reli-
able data for estimating parameters at the individual level
are often not available and data on network structures are
often limited and biased. Second, because complex mod-
els are computationally expensive, it is unwieldy and often
infeasible to explore the full parameter space when con-
ducting uncertainty and sensitivity analyses. These analyses
are essential to establish the robustness of any mathemati-
cal model of disease transmission (Blower and Dowlatabadi,
1994). With complex models, sensitivity analyses are often
reduced, by necessity, to limited scenario analysis and point
estimates of parameter values. Third, complex models can
be likened to ‘computer games’. They are comprised of
tens of thousands of lines of code that cannot be readily
understood by anyone except (perhaps) for their author,
and the mechanisms for their observed responses to par-
ticular sets of inputs can rarely be disentangled. Fourth,
because of the large investment in development and imple-
mentation of complex models, practitioners may be less
inclined or able to adapt their models or to start afresh when
addressing new research questions and may be tempted
to apply their existing models inappropriately to new
Consequently, we view the current trend towards com-
putationally intensive, computer game-like, individual-level
modelling to be largely misguided and often responsible for
diluting the insight attainable. Modelling can be used as a
powerful tool to enable informed decisions and predictions
of various strategies for policy-makers. But it is important
to keep in mind that simple relationships and constructs
that address the primary research question under investi-
gation will always be the most useful. Reproducing a game
of reality will usually not lead to an easily understandable
and actionable result to influence public health policy.
Funding: The National Centre in HIV Epidemiology and
Clinical Research is funded by the Australian Government
Department of Health and Ageing and is affiliated with
the Faculty of Medicine, University of New South Wales,
NSW, Australia. Dr David Regan is supported by a National
in Population Health. Dr David Wilson is supported by a
University of New South Wales Vice Chancellor’s Research
Fellowship and a grant from the Australian Research Council
Conflicts of interest: None declared.
Ethical approval: Not required.
Barnabas, R.V., Laukkanen, P., Koskela, P., Kontula, O., Lehtinen,
M., Garnett, G.P., 2006. Epidemiology of HPV 16 and cervical
cancer in Finland and the potential impact of vaccination: math-
ematical modelling analyses. PLoS Med. 3, 624—632.
Blower, S.M., Dowlatabadi, H., 1994. Sensitivity and uncertainty
analysis of complex models of disease transmission: an HIV
model, as an example. Int. Stat. Rev. 62, 229—243.
May, R.M., Anderson, R.M., 1987. Transmission dynamics of HIV
infection. Nature 326, 137—142.
Wilson, D.P., Kahn, J., Blower, S.M., 2006. Predicting the epi-
demiological impact of antiretroviral allocation strategies in
KwaZulu-Natal: the effect of the urban—rural divide. Proc. Natl.
Acad. Sci. USA 103, 14228—14233.
Yorke, J.A., Hethcote, H.W., Nold, A., 1978. Dynamics and control
of the transmission of gonorrhea. Sex. Transm. Dis. 5, 51—56.