This is crucial in determining how to respond to the epidemic, but it is far from easy to estimate.
Many countries are struggling to accurately count all deaths from the coronavirus.
One of the most important questions about a new infectious disease like Covid-19 is how deadly it is. After months of collecting data, the researchers are finally getting closer to a reliable answer.
Researchers speak of infection fatality rate (IFR) to characterize the lethality of a new disease. This is the proportion of infected people who will die from it, including those who are not screened or have no symptoms.
“The IFR is one of the important numbers, next to the collective immunity threshold, and has obvious implications for the scale of an epidemic and the seriousness with which we must consider it,” says Robert Verity, epidemiologist at Imperial College London.
Accurately calculating the death rate is a challenge in any epidemic, as it relies on knowing the total number of people infected – not just confirmed cases – as well as the exact number of deaths attributable to the disease. But the death rate is particularly difficult to determine for Covid-19, the disease caused by the SARS-CoV-2 virus, said Timothy Russell, mathematician and epidemiologist at London School of Hygiene and Tropical Medicine. This is partly because many infected people have mild or no symptoms, and therefore goes undetected, but also because the time between infection and death can be very long, up to two months. Many countries are also struggling to accurately count all deaths from the virus. Death registers suggest that many deaths are not included in the official figures.
The data collected at the start of the pandemic overestimated the mortality of the virus, and later analyzes underestimated it. Today, many studies – based on different and complementary methodologies – estimate that in most countries the mortality of Covid-19 is between 0.5 and 1%. “The studies that seem reliable to me converge on this range,” says Timothy Russell. But some researchers argue that this convergence may just be a coincidence.
To fully understand the severity of the epidemic, scientists need to understand the lethality of the virus in different groups of people. The risk of dying from Covid-19 varies considerably depending on age, ethnicity, access to care, socioeconomic status and underlying health conditions. According to these researchers, it is necessary to look in more detail at these different groups.
The mortality rate is also specific to a given population and changes over time as doctors improve treatments for the disease, which further complicates its estimation.
It is important to determine the actual death rate so that governments and individuals can adopt the appropriate responses to the epidemic. “If the death rate estimate is too low, the community may not be well prepared and may not react strongly enough. Conversely, if the estimate is too high, it can provoke an overreaction which will be at best costly, and at worst will worsen the harmful effects of radical measures such as confinement, “explains Hilda Bastian, doctoral student in medicine at Bond University, Australia.
First diverging estimates
The first indications of virus mortality were taken from the total number of confirmed cases in China. At the end of February, the World Health Organization roughly estimated that the mortality rate was 3.8% among the confirmed cases (we speak of case fatality rate, or CFR). The death rate among confirmed cases reached 5.8% in Wuhan, the epicenter of the epidemic. But these estimates exaggerated the dangerousness of the disease, as they did not take into account the many people who had probably been infected but had not been tested, masking the true extent of the epidemic.
The researchers tried to fill this gap by estimating the mortality rate from models predicting the spread of the virus. The results of these first analyzes were around 0.9%, with a range ranging from 0.4 to 3.6%, according to Robert Verity. Its own modeling estimated the ta
Overall mortality rate in China is 0.7%, and up to 3.3% among people aged 60 and over.
Timothy Russell’s team also used data from the cruise ship Diamond Princess, quarantined in Japanese waters in early February, to estimate the death rate in China. Almost all of the 3,711 passengers and crew members were tested, which enabled the researchers to count the total number of infections (almost 700) – including asymptomatic cases (18% of cases) -, and death in a known population. The fatality rate among Diamond Princess passengers reached 1.2%. Comparing with the estimated death rate for confirmed cases in China, his team estimated an overall death rate of 0.6%.
“The purpose of these studies was to quickly obtain rough estimates of the Covid-19 mortality rate,” says Robert Verity. But the researchers also had to make complex estimates, which remain to be verified, on the number of confirmed cases and the actual number of people infected. “These preliminary estimates should be updated as soon as better data becomes available,” he said.
On the trace of antibodies
Large-scale surveys to test the proportion of individuals who have developed antibodies to SARS-CoV-2, known as seroprevalence surveys, should help refine mortality rate estimates. Around 120 seroprevalence surveys are underway worldwide. But the results of early serological surveys have only confused the picture, suggesting that the virus was far less deadly than previously thought.
In one of the first studies, 919 people were tested in the German city of Gangelt, where a major outbreak was observed. Of these individuals, approximately 15.5% had antibodies to the virus, five times higher than the number of proven Covid-19 cases in the city at the time. This figure was used to deduce a mortality rate of 0.28%. But this study was based on a relatively small number of people.
The mortality rate (IFR) is the proportion of people infected with the Covid-19 virus who will die from the disease. Estimates are for limited regions, and may vary depending on demographics, health system and study methodology.
Others among these first seroprevalence studies did not correctly take into account the lack of sensitivity and specificity of the serological tests used, nor the representativeness bias between the sampled populations and the actual populations, judge Robert Verity.
These problems could have artificially inflated the estimates of the total number of infected people and thus lower the estimate of the mortality rate of the virus, according to the researcher. Likewise, the fact that not all deaths from Covid-19 are counted – a problem that affects many countries because not all people who die are tested for the virus – can also bias the death rate, says Gideon Meyerowitz-Katz, doctoral candidate in epidemiology at the University of Wollongong, in Australia.
Larger seroprevalence studies have been published in recent weeks. They provide higher mortality rate estimates than early studies. A survey of more than 25,000 people in Brazil, published on medRxiv, provided a death rate of 1%. Another study, which tested more than 60,000 people across Spain, reports an antibody prevalence of 5%, but the results have yet to be officially analyzed. The team did not calculate the death rate, but based on the results, Robert Verity estimates that the death rate in Spain is around 1%.
Several researchers, including Timothy Russell and Robert Verity, find it significant that an increasing number of studies in different regions of the world converge on estimates of the mortality rate in the range of 0.5 to 1%. But other scientists are still cautious. “This trend may just be a fluke,” said Gideon Meyerowitz-Katz.
Marm Kilpatrick, researcher in infectious diseases at the University of California at Santa Cruz, also reports most of the data from serological studies have not been published in peer-reviewed scientific journals. It is difficult to know when and how they were collected and to correctly calculate a mortality rate that takes into account the time between infection and death, he said.
Marm Kilpatrick and others look forward to large surveys that will estimate mortality rates across all age groups and taking into account pre-existing conditions, which would provide a more accurate picture of mortality from Covid-19. One of the first studies to consider the effect of age was pre-published last week. Based on Geneva seroprevalence data, this study estimates the mortality rate at 0.6% for the total population, and 5.6% for people aged 65 and over.
These results have not been reviewed by a reading committee, but Marm Kilpatrick says the study addresses many of the problems encountered in previous seroprevalence surveys. “This study is what should be done with all the serological data,” he judges.