By Prof. Ben Cowling, School of Public Health, The University of Hong Kong
COVID-19, caused by the virus SARS-CoV-2, has been spreading in humans for at least 6 months. We have just passed the milestone of 10 million confirmed cases worldwide and the milestone of 500,000 deaths. However, these numbers underestimate the number of infected people and fatalities, respectively.
COVID-19 infections are often mild, and sometimes occur without any symptoms. In order for an infected person to become a "laboratory-confirmed" case, a number of things must happen. In many parts of the world, an infected person would need to develop symptoms that are serious enough that they seek medical attention, and the doctor is able to provide a laboratory test for COVID-19. In some places the test may only be available to doctors in certain hospitals or clinics.
Among persons infected with the virus that causes COVID-19, not everyone would test positive in the clinic or hospital. There have been numerous media reports of patients strongly suspected to have COVID-19 but repeatedly testing negative. Not all laboratory testing approaches are the same. Some types of laboratory tests are “better” than others at detecting the virus. Some patients may have relatively lower amounts of the virus, which is more difficult to detect.
Apart from identification of cases in clinics and hospitals, some parts of the world have been more proactive in finding cases. For example China tested millions of people in Wuhan in late May, regardless of whether or not each person had any symptoms, in an ambitious plan to identify and then eliminate all remaining infections in the city. In some parts of the world arriving travelers are placed into quarantine for 14 days, and tested regardless of whether or not they have symptoms. These approaches have allowed the detection of COVID-19 in asymptomatic persons.
Some of these asymptomatic cases will go on to develop symptoms a few days later, and this is termed “pre-symptomatic” detection of the virus. Others may recover from the infection without ever developing symptoms.
Identification and counting of cases is based on laboratory testing to detect the COVID-19 virus in a person. However, this only works while the infection is present. Once a person has recovered, the virus will disappear and a test for the virus will come back negative.
There is another approach to identify infected persons, and that relies on detection of signs of infection in the blood, rather than on detection of the virus. Around 2-3 weeks after infection, most infected people will develop “antibodies” as part of the immune response to that infection. These antibodies should provide some degree of protection against re-infection.
Wuhan has reported around 50,000 confirmed cases and around 4,000 deaths from COVID-19 to date. However, the true number of infections is likely to be higher than 50,000, since not every infected person would become a confirmed case.
One way to estimate how many people have been infected would be to extrapolate from the laboratory-confirmed cases, adjusting for the proportion that might be mild or asymptomatic. One study by HKU estimated that there were around 200,000 cases in Wuhan if also extrapolating to mild cases, and it is likely that there were a large number of undetected asymptomatic infections as well.
One recent study by HKU reported the prevalence of antibodies in 452 Hong Kong residents who returned from Hubei province in early March, 80% of whom were returning from Wuhan. That study identified antibodies in around 4% of the returning persons. Among persons returning from Wuhan, 16/364 (4.4%) were seropositive. Among persons returning from elsewhere in Hubei province, 1/88 (1.1%) were seropositive. Among all 452 persons tested, 17 were positive, which was 3.8%.
The authors extrapolated the 4.4% seroprevalence to the entire population of Wuhan, which would correspond to around 500,000 infections. Of course, the returning travelers might be different in various ways from the residents of Wuhan, but an estimate of 500,000 infections is quite plausible. However, the authors also extrapolated the 3.8% to Hubei province as a whole, corresponding to 2.2 million infections. It might perhaps have been more reasonable to extrapolate the 1.1% seroprevalence outside of Wuhan to the 48 million people who live in Hubei province but not Wuhan city, and that would correspond to another 500,000 infections for 1 million in total (rather than another 1.7 million infections for 2.2 million in total). However with a small number of persons sampled from the rest of Hubei province outside Wuhan, the sample size may be insufficient for a robust estimate of the number of infections in the province as a whole.
Another serologic study in Wuhan estimated seroprevalence of 3.8% in 714 healthcare workers in Wuhan, 3.8% in 346 staff of the hotels accommodating healthcare workers, and 3.2% in 219 family members of healthcare workers, based on samples collected in early April. These are special populations and seroprevalence in these groups may not correspond to seroprevalence in the population as a whole, although these estimates are very consistent with the other information available.
One final point regards mortality statistics. A number of studies have estimated that the risk of mortality among infected persons is in the range 0.5% to 1% overall, but with considerable variation by age with elderly at highest risk of severe disease. Given that 4,000 deaths have been reported, an estimate of 500,000 infections in Wuhan would correspond to 0.8% of infections being fatal. Unrecognised deaths, i.e. deaths due to COVID-19 which were not recorded as COVID-19 deaths, most likely have occurred in every part of the world that COVID-19 has spread. However, it is unlikely that a large number of unrecognised COVID-19 deaths have occurred in Wuhan.