Coronavirus fatality rate estimated by Imperial scientists

Imperial researchers have released estimates on the proportion of people who may die from the disease.

** UPDATE (30-03-2020): This research (Report 4) has now been peer reviewed and published in Lancet Infectious Diseases. **

According to the latest estimates from the team, from the MRC Centre for Global Infectious Disease Analysis at Imperial, one percent of people with the disease will die from their infection.

Our estimates suggest the impact of the unfolding epidemic may be comparable to the major influenza pandemics of the twentieth century Professor Neil Ferguson Report co-author

This is the fourth report from the team, under the umbrella of J-IDEA, the Abdul Latif Jameel Institute for Disease and Emergency Analytics.

Since the emergence of the new coronavirus (COVID-19) in December 2019, the team have adopted a policy of immediately sharing research findings on the developing pandemic. The new report presents an assessment of the severity of the likely health impact of COVID-19.

A key measure of severity of any outbreak is the case fatality ratio (CFR) - the proportion of people with a disease who will eventually die as a result of infection. Quantifying this can be challenging.

Cases of infection are detected through surveillance, which typically occurs when ill people seek healthcare. Depending on the demands a healthcare system is under, and the capacity to undertake testing, confirmed case numbers reported during an outbreak only ever represent a fraction of the true levels of infection in the community.

In addition, it can take weeks for the final clinical outcome of someone infected with a respiratory virus such as COVID-19 to be known and reported.

For the new report, the researchers used statistical models that combined data on deaths and recoveries reported in China and in travellers outside mainland China, as well as infections in repatriated citizens.

Dr Ilaria Dorigatti, co-author of the report, said: “The estimates published in today’s report rely on limited data and the next few weeks will provide valuable information on the outcome of current infections, which will allow us to refine our estimates and fill our knowledge gaps on the severity of this new virus.”

Dr Lucy Okell, report co-author, added: "It's critical to work out the chance of dying from COVID-19 for those who become infected, and who is most at risk, but our estimates remain uncertain at the current time. Most cases reported so far remain in hospital or quarantine and we don’t yet know how many will eventually recover or die.  Deaths tend to be reported promptly but there is less information on how many people have recovered from the virus. We are likely to be missing cases who have milder symptoms but there is limited information on how common these are. However, our team has tried to account for these factors, and we are continuing to compile as much data as possible on the outbreak to reduce uncertainty about this critical issue."

Play video

Professor Neil Ferguson, co-author of the report, explained: “Understanding the likely impact of the unfolding pandemic caused by the COVID-19 virus on human health will be critical to informing the decisions made by countries in the coming weeks in how best to respond to this new public health threat. Our estimates – while subject to much uncertainty due to the limited data currently available – suggest that the impact of the unfolding epidemic may be comparable to the major influenza pandemics of the twentieth century. It is therefore vital that countries across the world continue to work together to accelerate the development and testing of effective treatments and vaccines, on the fastest possible timescale. Surveillance and data sharing also need to be further enhanced to allow the spectrum of disease severity caused by this virus to be better understood.”

Fatalities among severe cases

The team estimate that people dying now from the virus on average first developed symptoms three weeks ago. This means the numbers of deaths reported today should be compared with the numbers of cases reported over two weeks ago, to gain insight into the true proportion of people who might die from their infection.

Professor Azra Ghani, report co-author, said: “Assessing the severity of an emerging disease is always challenging. This is because we generally detect those that are most ill first and only later get a sense of how many people may have milder disease or carry the infection but not have any symptoms. However, by piecing together data from various sources – including from the most severe cases reported from China, from international travellers to Wuhan who return infected, and from testing undertaken on repatriated citizens – and by correcting for delays in reporting, we estimate a fatality ratio for all infections of approximately 1%. It is important to remember that this estimate remains very uncertain. However, by monitoring the outcome in some of these populations in the coming days and weeks we should very soon be in a position to get more precise estimates that can inform public health planning.”     

The team’s analysis presents estimates of the CFR for three bands of symptom severity. In Hubei province in China, only people with relatively severe symptoms are being prioritised for testing. Their central estimate of CFR for these cases is 18%, but with high uncertainty.

Only one in 19 people infected are being tested

Many countries have been looking for COVID-19 cases in travellers returning from mainland China and have been testing not just people with pneumonia (a key criterion for testing in China), but also with fever or other milder respiratory symptoms. Therefore, the severity of infections detected in travellers is lower than seen in reported cases in Hubei. The central estimates for CFR in cases detected in travellers outside mainland China lie in the range 1.2%-5.6% depending on the methods and data used. However, given the limited data so far available, there is substantial uncertainty around these values.

To make an overall assessment of the potential global health impact of this pandemic, the team needed to estimate the CFR across all people who will be infected, irrespective of symptom severity.

This required estimates of the number of people currently infected, including those with mild or no symptoms. From the results of testing of all passengers (irrespective of symptoms) on several repatriation flights to Japan and Germany, the team estimate that 1.3% of the population of Wuhan currently had the infection on January 31^st.

Comparing this figure with reported case numbers, the team estimate only about 1 in 19 people infected with COVID-19 in Wuhan are being tested for infection and therefore being reported as confirmed cases. In turn this allowed them to estimate that the CFR of COVID-19 in people with detectable infection (including some who may never develop symptoms) is approximately 1% (uncertainty range: 0.5-4%).

Professor Christl Donnelly, report co-author, said: "At the end of an epidemic that is completely observed, the case fatality ratio is simply the number of deaths divided by the number of cases. However, understanding the severity of a newly discovered pathogen is much more difficult, despite being an urgent priority.  To get an estimate during an ongoing epidemic, we have had to allow for mild cases being less likely to be detected in symptom-based surveillance as well as the fact that the clinical outcome (death or recovery) is not yet known for some confirmed cases that remain hospitalised. Assumptions are always required to underpin our estimation methods, so we have explored the impact of alternatives. Our best estimate is that the fatality ratio for all infections is roughly 1%, but there remains uncertainty and we will update our analyses as further data become available."

Lower fatality ratio than Ebola

All CFR estimates should be interpreted cautiously at the current time as the proportion of both deaths and cases being detected is unclear. Not all infected people may have symptoms and severe cases are more likely to be detected. All estimates rely on currently limited data on the time intervals from symptom onset to death or recovery. Additional studies to assess the extent of mild or asymptomatic infection will also allow estimates of CFR to be refined. The next two weeks will allow refinement of the CFR estimates presented here.

Professor Steven Riley , report co-author added: "As more data becomes available, we are able to estimate the severity of this novel coronavirus. Using a variety of data, and correcting for known biases, we estimate that approximately 1 percent of people infected with this virus will die as a result of that infection. Although this is lower than diseases such as Ebola or SARS, it is considerably higher than seasonal influenza."

Data painstakingly collected

In addition, the CFR estimates reported by individual countries in the coming months will vary depending on the sensitivity of those countries’ surveillance systems to detect cases of differing levels of severity, the clinical care offered to severely ill cases, and the method used to derive the estimate.

Dr Natsuko Imai , report co-author, explained: "Our team pieced together multiple sources of data to understand the potential health impact of the epidemic. These data have been painstakingly collected from hundreds of media reports and public health websites through a huge team effort. Everyone, from researchers, technical staff, and students have all pitched in to help with the response. What has also struck us is the openness and speed at which countries are publicly sharing information on these COVID-19 cases. This sharing of information between countries will be invaluable for the international response to this epidemic.”

Adapted from a press release from the MRC Centre for Global Infectious Disease Analysis

-

The full report is available here