Contact details

Dr T Deirdre Hollingsworth

Junior Research Fellow
School of Public Health

Norfolk Place
St Mary's Campus

Tel: +44 (0)20 7594 3007
Email:

Dr Deirdre Hollingsworth

I am an Imperial College Junior Research Fellow in infectious disease epidemiology. The main focus of my research is developing mathematical models to understand the role of human movement in the transmission of malaria to inform the design of effective public health interventions. I also work on HIV and outbreaks of emerging infections.

Human movement and infectious disease transmission 

Local and global movements of people affect the speed and directions of the spatial spread of infectious diseases. The importance of movement patterns will depend on the characteristics of the disease. For example, heterogeneities in movement patterns may affect  the rate of spread for diseases with a slow epidemic growth rate, such as SARS, but may have minimal effect in a fast growing epidemic such as influenza (Hollingsworth et al Emerging Infectious Diseases 2007).  When planning local elimination programmes for malaria patterns of overnight travel, in particular seasonal movements to and from high transmission areas, will affect the degree of intervention required. I am gathering data and building models to address some of these issues. 

Malaria

In recent years there has been a resurgence of efforts to mitigate the mortality due to malaria with the ultimate aim of eradication. As part of the malaria research group in our Department, I work with Azra Ghani, Neil Ferguson, Jamie Griffin and others on developing models of endemic malaria transmission in areas of Africa to identify strategies which will be effective under particular local conditions. Immunity to malarial symptoms and disease develops with repeat infection, resulting in low rates of disease in the highest transmission areas and high rates of disease for intermediate levels of transmission. This interaction between infection, immunity and disease complicates the prediction of the long term impacts of intervention policies. Mathematical models can be used both to test hypotheses about these interactions and to inform the design of effective intervention strategies.

HIV

I also work on models of HIV transmission which consider variability in the life history of disease and the implications for evolution and control. These models can be used to study the evolution of virulence and the importance of the short period of high infectivity at the outset of an HIV infection (acute and primary infection) in transmission of the disease. This research is performed in collaboration with Christophe Fraser, George Shirreff and Bill Hanage and with colleagues in the Rakai Health Sciences Program and the HIV Monitoring Foundation in the Netherlands.

Evolution of HIV

Analysis of heterosexual couples from who transmitted to shows that they have similar viral load set-points (Hollingsworth, Laeyendecker et al PloS Pathogens 2010). This suggests that there are viral factors which play an important role in pathogenesis, despite high levels of within-host mutation and virus variability. Since viral load is a key determinant of both transmission and survival, it is important to identify viral factors which determine viral load, if they exist. Such factors could then be targeted for vaccination or treatment. This observation does not discount the importance of immune factors in determining viral load set-point, but suggests a balance between the two. 

Figure 1 Viral load set-point of index partner versus that of the secondary case in couples with strong support for transmission. Couples are stratified by male to female transmission (green triangles), male to female transmission (blue circles) and unknown direction of transmission (red diamonds, plotted as female against male viral load, since the index partner could not be identified). For further details see Hollingsworth, Laeyendecker et al PloS Pathogens 2010.

 In addition to the possible consequences for pathogenesis, our observation that viral load is 'heritable' from one person to the next is an important test of our hypothesis that HIV virulence can evolve at a population level (Fraser, Hollingsworth et al PNAS 2007). 

Viral load, which is a measure of the amount of virus in the blood of infected individuals, varies considerably over the course of an infection (Figure 2). Viral load set-point (or viral loads during the asymptomatic period)  are a key determinant of life expectancy and infectiousness (Figure 2). The opportunities for onward transmission are a balance between these two processes. Those with high viral loads are highly infectious but do not live long enough to infect many others. Those with low viral loads live longer and so have more opportunities to transmit but are not so infectious. Therefore, it is individuals with intermediate viral loads who have the potential to transmit most often (Fraser, Hollingsworth et al PNAS 2007; Figure 3, below).  This suggests that HIV may have already adapted to the human population to optimise transmission potential (or individual R₀). An important pre-requisite for this hypothesis to be correct is that viral load is ‘heritable’ from one person to the next, which we have now shown (Hollingsworth, Laeyendecker et al PLos Pathogens 2010).

Figure 2 Schematic of viral loads over the course of an infection. During the long asymptomatic period of infection, which lasts several years, viral loads are relatively steady and the average viral load in this period is defined as the ‘viral load set-point’. Those with high viral load set-points have short life expectancies and high infectiousness. Those with low viral load set-points have long life-expectancies and low infectiousness.

Figure 3 Coincidence of optimum transmission potential and observed viral load set-point distribution. The curve (right hand axis) represents the mean number of new infections across their lifetime for individuals with each viral load set-point. It is optimum for intermediate viral load set-points. The proportion of individuals in a cohort of dutch men cohort (black bars ) and a Zambian heterosexual transmission study (white bars) with each range of viral load is also shown (right axis). The most commonly observed viral loads are similar to those which optimise lifetime opportunities for transmission. For further details, see Fraser, Hollingsworth et al PNAS 2007. 

Recent studies in the Netherlands, Italy and in a cross-European study have shown HIV viral load set-points changing over 10-20 year timescales. Other studies, for example in Switzerland, North America and France, have shown no trend in virulence. Our analyses provide evidence that natural selection to optimise transmission fitness may be driving these changes. In addition, they raise the possibility that population level interventions, such as mass treatment, could change which viral loads lead to most onward transmissions and are the most 'fit'. This could then affect HIV virulence over long timescales. Extensive studies on larger populations are required to further test these hypotheses.

Emerging infections

As a member of the MRC Centre for Outbreak Analysis and Modelling I was part of the response to the 2009 H1N1 influenza pandemic. I have previously worked on reanalysis of the 2003 SARS epidemic with Roy Anderson, Christl Donnelly and collaborators in the Netherlands and Hong Kong.

Travel advisories

The WHO issued several travel advisories during the course of the 2003 SARS epidemic, advising avoidance of non-essential travel to affected areas. These were controversial because of their adverse economic impact and the uncertainty about their effectiveness. International airlines reported serious drops in passenger numbers, but the change was gradual (Fig 4). Our analysis suggested that within-country control measures for SARS were extremely effective in curtailing international spread. For influenza within-country control is much more difficult due to the rapid epidemic growth rate and pre- or a-symptomatic transmission. If within-country control measures fail following the emergence of a potentially pandemic variant of influenza, our analysis shows that only complete travel restrictions, rapidly implemented, would have a major impact on global spread.  This result has subsequently been supported by other modelling studies.

Figure 4 Daily SARS case number in 2003 (thin red bars) and monthly daily percentage change (over previous year) in passenger volume (thicker bars). Four air passenger datasets are shown: global passengers (black), Asia-Pacific passengers (red), Hong Kong airport passengers (green) and Beijing international Airport (yellow). For further details see Hollingsworth et al Nature Medicine. (2006).

Background

Before moving to Imperial I did a PhD in plant epidemiology, supervised by Adrian Stacey and Professor Chris Gilligan, in the Epidemiology and Modelling Group within the Department of Plant Sciences, University of Cambridge. Prior to my PhD I did an MSc in Mathematical Modelling and Numerical Analysis, supervised by Andrew Fowler, and a BA  in Mathematical Sciences, both at Lincoln College, Oxford University. Between these two degrees I did an MMus (Master of Music) in 'cello performance.

Teaching

I teach on the MSc in Modern Epidemiology and Imperial College’s professional short course, Introduction to Mathematical Models of the Epidemiology & Control of Infectious Diseases, which is aimed at public health professionals, policy-makers, researchers and health economists who want to learn about the basic principles and practical applications of mathematical modelling and modern quantitative methods.

Selected Publications