Ganna Rozhnova is an Assistant Professor at the Department of Epidemiology at the Utrecht Medical Center (UMC). She has worked on various aspects of the modeling of childhood infections, HIV, and influenza. Recently, she has focused on the impact of contact tracing strategies in managing the COVID-19 pandemic.

Can you tell us about your path from statistical physics to epidemiology? What led you to this field?

By the end of my postgraduate studies in theoretical physics at Lisbon University, I knew that I would like to do research where I could apply methods from physics and mathematics to solve ‘real-world’ problems. By ‘real-world’ problems I mean, for instance, the spread of infectious diseases or evolution of viruses such as influenza and HIV. I quickly realized that there were a lot of opportunities for researchers with my background to contribute to these areas. I felt it was easier to learn biology for a person with a technical background than to learn mathematics for a biologist. For my PhD studies, I was awarded a fellowship from the Portuguese Foundation for Science and Technology to investigate the spatio-temporal dynamics of childhood infections, – a long-standing problem where physicists could offer a different approach based on methods from statistical physics. Since then I have become involved in even more applied research and established collaborations with various stakeholders in public health. In my current appointment as Assistant Professor in the University Medical Center Utrecht (UMCU), I work on problems which could influence public health policy and decision making in the Netherlands and abroad. For example, in the past year, I contributed to developing vaccination strategies against cytomegalovirus, investigated the prospects of eliminating HIV in the population of Dutch men who have sex with men, and examined the impact of self-imposed prevention measures and short-term government-imposed social distancing on the COVID-19 epidemic.

How would you explain your research to a non-specialist?

I call myself an infectious disease modeller, a researcher who builds mathematical models to predict how various pathogens spread in the population and to devise strategies and interventions that affect transmission. Mathematical models can be of different types and complexity, but none of them is a crystal ball. A model always rests on certain assumptions, and it needs good data (e.g. surveillance data, demographic data, pathogen genetic data etc.) for adequate parameterisation. A model yields a quantitative prediction of the most likely evolution of the disease under those assumptions, and also a quantitative estimate of the uncertainty of that prediction. In simple terms, it means I could build a model to predict the number of COVID-19 infections in the Netherlands and calculate the uncertainty around this number.

You recently authored a paper in PLOS Medicine on the impact of prevention measures on mitigating COVID-19 epidemic. What is, in your opinion, the main insight this research has to offer? What can countries learn from this? 

Our study shows that it is important to maintain awareness of COVID-19 because it can help to increase adherence to self-imposed protective measures such as handwashing, mask-wearing and self-imposed social distancing. As a consequence, these self-imposed measures could minimise the total number of persons who get COVID-19 and the number of infected persons at the peak of the epidemic. If nearly all population in the Netherlands adopted self-imposed measures, we would not have to confront the possibility of secondary lockdowns as well as the possibility that we may find our medical systems overwhelmed during the peaks of epidemics. Our analyses also support that self-imposed measures should be adopted preferably as a suite rather than any specific measure only. Overall, it appears to be a relatively cheap solution that would not disrupt economical and societal fabric as much as a lockdown does.

About your article on the effectiveness of contact-tracing strategies for COVID-19 in Lancet: what are the main results and implications? 

Contact tracing is one of the key strategies to control the spread of SARS-CoV-2 after most lockdown measures are lifted. In this study, we evaluated the impact of timeliness and completeness in various steps of a contact tracing strategy for improving contact tracing effectiveness. Our main finding is that minimising testing delay, i.e. the time between a person developing symptoms and receiving a positive test result, has the largest impact on reducing transmission of SARS-CoV-2. This suggests that resources should be invested in testing infrastructure and optimising access to testing so that people can receive their test results as soon as possible (ideally within 24 hours of developing symptoms). If the testing delay is 3 days or longer, conventional contact tracing won’t stop transmission of the virus. Contact tracing effectiveness can be further enhanced with digital tracing based on mobile app technology by reducing delays in the contact tracing process and optimising contact tracing coverage. In particular, we found that digital tracing will be successful in stopping the spread of the virus even with a testing delay of 2 days if about 80% of contacts are traced.

Could you tell us about the particular challenges related to COVID-19 in comparison to other diseases you have been working on?

I started to work on COVID-19 before the first COVID-19 cases appeared in Europe anticipating that public health policymakers in different countries would be seeking recommendations on how to delay and/or flatten the peak of the epidemic. The main challenge is that the COVID-19 pandemic develops very quickly, and we have very little time to provide answers. For example, the first lockdown was introduced within two weeks, and the decisions had to be made very fast. To make a meaningful scientific contribution, it has to be timely. A study that addresses a pressing question e.g. the impact of opening schools, could become irrelevant within a few weeks. It is also hard to gather COVID-19 data quickly, as some of them are not publicly available. Equally hard is monitoring relevant literature, as the number of COVID-19 publications has grown very fast.

How have the pandemic and the lockdown in the Netherlands impacted your research?

The negative impact is that our human subject research was on halt for many months. For example, I am the PI of the project funded by the Aidsfonds to investigate the impact of a potential HIV cure on people’s lives and the Dutch epidemic. In this project, we need to interview people, but this was not possible because of the physical distancing measures in the UMCU. The positive impact is that we started to collaborate more on the COVID-19 topics. The Infectious Disease Modelling group from the Julius Center for Health Sciences and Primary Care in the UMCU has produced several high impact publications and we really enjoyed working together. We miss everyday interaction, but remote collaboration has worked for us. It was also nice as I could get specific funding for COVID-19 modelling.

And how has it impacted on the rest of your work and life?

I would point out the challenge of setting up an online course on the Mathematical Modeling of Infectious Diseases, where I contribute as a lecturer and project supervisor. We had to be creative in quickly making new content for the course which would be interesting and accessible for students. The assessment of the course by the students was positive, but we really missed the usual in-class interaction with our students.

In your opinion, what contributions can modelling bring to an uncertain world like the one we live in? What is to be learnt from the ordeal we have been (and still are) going through?

In the near future, mathematical models will continue to be instrumental in predicting, planning, and helping to make decisions regarding the COVID-19 pandemic. As time goes on and the information on COVID-19 accumulates, it will be possible to build more detailed COVID-19 models. However, there are still many factors that continue to be unknown. These range from the biology of the virus to social behaviour during the epidemic, so the uncertainty that affects model forecasts is also inevitable. It is important to understand that mathematical models do not predict the future, but rather describe various future scenarios of the epidemic that arise from the assumptions of each model. Despite this, modelling is the most transparent and rigorous way to take advantage of the available COVID-19 data to anticipate different scenarios and try to avoid them or to prepare for them.