Outbreak of an infectious disease? Mathematics helps in making quick, informed decisions
A job thanks to COVID—something not many people can claim. But PhD candidate Vera Arntzen can. Over the past four years, she has mapped two crucial characteristics of SARS-CoV-2 infections. Thanks to her research, experts can now make well-informed decisions on matters like quarantine duration, which will be incredibly useful for future outbreaks.
Experts in infectious diseases—or ‘disease detectives’, as Arntzen calls them—rely on two key features when dealing with a new outbreak: the incubation period and the latent period. Taking action quickly is critical, but in the early stages of an outbreak, knowledge is scarce. ‘You have very little information, and the data you do have is full of biases,’ says Arntzen, who will defend her PhD thesis on 16 October.
The incubation period is the time between infection and the appearance of the first symptoms.
The latent period is the time between infection and when a person becomes contagious.
Infected at the wedding or the soccer field? Hard to say
Where do these data biases come from? With some diseases, pinpointing the exact moment of infection is relatively straightforward. For example, if half the guests at a wedding get food poisoning, it’s pretty clear where the infection happened. But SARS-CoV-2 is different. ‘It can spread through all sorts of contact—from having tea with a friend to attending your neighbour’s birthday party.’ Instead of a clear point of infection, there’s a window in which it could have occurred.
When researchers estimate the incubation period, they usually assume that the risk of infection is constant throughout this window, but that’s not realistic. Staying overnight with someone presents a higher risk than chatting in the park while walking your dog. Moreover, as Arntzen explains, the risk of infection is not constant at the start of an outbreak: ‘The risk increases rapidly over time, so the chance of infection is much higher towards the end of the window than at the beginning.’
Staying overnight with someone presents a higher risk than chatting in the park while walking your dog.
Estimating the latent period is even more difficult
When it comes to the latent period, the uncertainty lies not only in the infection moment but also in when a person becomes contagious. People are unaware when they become infectious, so there is very little data available. This makes estimating the latent period particularly challenging.
For SARS-CoV-2, in particular, estimating this period is crucial. There are many asymptomatic cases—people who show no symptoms but still go through a latent period during which they are infected but not yet contagious. Ideally, measures should be based on the latent period, Arntzen explains.
Measures came too late for the Delta variant
Once the biases are accounted for, what’s next? Arntzen refined the estimates for SARS-CoV-2 using a unique dataset from Vietnam. Vietnam was slow to start vaccinating, and due to their “zero-tolerance policy”, other virus variants were not circulating for some time. The estimates for incubation and latent periods from this dataset were specific to the Delta variant, free from the effects of immunity through vaccination or exposure to other variants.
Vietnam’s experience with infectious diseases
Arntzen’s research with the Vietnamese dataset had another unique aspect: Arntzen visited Vietnam twice. ‘What struck me the most in Vietnam is that severe infectious diseases are a daily reality there, while here they feel more distant. But we too need to find a way to live with infectious diseases. COVID won’t be the last pandemic. The best thing we can do is be prepared.’
Thanks to the Vietnamese dataset, Arntzen concludes in her thesis that the latent period of the Delta variant is shorter than previously thought, meaning there’s more time between the onset of contagiousness and the appearance of symptoms. That explains why many new cases with the Delta variant went undetected. Control measures, such as quarantine following contact tracing, came too late when people had been unknowingly infectious for a long time.
New software for future outbreaks
Arntzen ventured into uncharted territory with her research. Existing statistical models couldn’t handle the biases in the data, so she developed new ones. This software will help improve and speed up estimates of incubation and latent periods during future outbreaks. But each outbreak is different. ‘Every infectious disease is unique, and the data on early cases vary. It’s always a combination of using the best method and collecting the best data.’