A trio of researchers with Ikigai Research, Australian National University and the University of Melbourne respectively, has found evidence that suggests the true COVID-19 infection rate for 15 selected countries is on average 6.2% higher than official tallies have listed.
In their paper published in the journal Royal Society Open Science, Steven Phipps, Quentin Grafton and Tom Kompas describe analyzing infection data from 15 similar countries and using it to estimate true infection rates.
The current resurgence of the global pandemic has very starkly highlighted the fact that it is far from over.
Millions of people the world over have been infected and over 1 million have died, yet medical scientists are still not in agreement on true infection rates.
Instead, most countries and even local jurisdictions publish “known” numbers of infections and use such figures to determine infection rates.
In this new effort, the researchers suggest that such an approach leaves out many people that become infected but never show up on official tallies.
To gain a better perspective on true rates of infection, the researchers chose to focus on 15 countries that have similar approaches to testing, diagnosing and treatment of the disease – all of them developed western countries except for Korea.
To make better estimates of true infection rates for each of the countries in their study, the researchers used a technique called “backcasting” by which they studied official numbers of infections along with death rates (both known to be from COVID-19 and from unknown causes above what was normal for given areas).
They also used what are believed to be survival rates for the disease. Data was then ported to a computer model that provided graphs showing true estimated infection rates for all 15 countries and for all of them averaged together.
The graphs showed high variation between detection rates, from 5.7 for Italy, to 39.1 for South Korea.
The researchers also note that in all cases, the true estimated number of people infected with the virus was higher than the numbers given for each country.
As an example, they found that for Australia, the true estimated infection rate (for people that recovered) was approximately five times higher than official estimates at the end of August.
That would translate to 0.48% of the population having been infected, which would mean 130,000 people – far more than the government there has reported.
31 October), SARS-CoV-2 has spread to every corner of the world. More than 40 million people have been diagnosed with SARS-CoV-2 infection and more than a million people have died of COVID-19, the disease caused by SARS-CoV-2.
Not all cases, in particular if asymptomatic, have been diagnosed and the true number of infections and deaths is probably much higher. Relatively few large scale seroprevalence studies have been completed but the available seroprevalence data show that only a few places, like Mumbai and Manaus, have reached a high prevalence in the population, close to the level required for some kind of herd immunity (see Table 1).
[Herd immunity is defined as the proportion of a population that must be immune to an infectious disease, either by natural infection or vaccination, to provide indirect protection (herd protection) to those who are not immune to the disease (D’Souza 2020, Adam 2020).
Table 1 shows that countries hit hardest by the COVID-19 pandemic have higher seroprevalence rates but, without an effective vaccine, no country can count on any kind of herd immunity in the near future.
|Table 1. Seroprevalence data 2020|
|Italy*||Nationwide||May 25-July 15||2.5%||Sabbadini 2020|
|Italy||Lodi (red zone)||23%||Percivalle 2020|
|Denmark||Faroe Islands||0.6%||Petersen 2020|
|China||Wuhan||March 9-April 10||3.2-3.8%||Xu X 2020|
|US||New York City|
San Francisco Bay area
|March 23-April 1|
|US||New York State||14%||Rosenberg 2020|
|US US||NYC, Health care personnelNationwide in patients receiving dialysis||July 2020||13.7% 8.3%||Moscola 2020 Anand 2020|
Not peer-reviewed. Results have recently been questioned.
* Note that Italy’s national survey results are preliminary and probably an underestimation. The country only managed to collect 40% of the planned samples, with many people refusing to be tested. Insiders never believed these figures and favored a seropositivity rate of 5-10% like in Spain or in France. Now we have a new estimate of COVID-19 prevalence in Italy by Francesca Bassi and colleagues: 9%, corresponding to almost 6 million Italians (Bassi 2020).
The articles cited in Table 1 report some interesting findings:
- Wuhan – Seropositivity for IgM and IgG antibodies was low (3.2%-3.8%) even in a highly affected city like Wuhan (Xu X 2020).
- New York City – In New York, the prevalence of SARS-CoV-2 among health care personnel was 13.7% (5523 of 40,329 individuals tested) (Moscola 2020) which was similar to that among adults randomly tested in New York State (14.0%) (Rosenberg 2020).
- UK – Black, Asian and minority ethnic (BAME) individuals were between two and three times as likely to have had SARS-CoV-2 infection compared to white people. An interesting trend: young people aged 18-24 had the highest rates (8%), while older adults aged 65 to 74 were least likely to have been infected (3%).
- Mumbai – In a cross-sectional survey the prevalence of past SARS-CoV-2 infection in three areas in Mumbai was around 57% in the slum areas of Chembur, Matunga and Dahisar, and 16% in neighboring non-slums (Kolthur-Seetharam 2020). In some places of the world herd immunity may be within reach.
- Geneva – Young children (5–9 years) and older people (≥ 65 years) had significantly lower seroprevalence rates than other age groups (Stringhini 2020).
- Faroe Islands – At the beginning of the pandemic, small islands tended to have low seropositivity rates.
It is worth noting that we still have few nationwide population-based seroprevalance studies, that the sensitivity and specificity of serological tests being used can vary from place to place, and that some people might have been infected without showing detectable levels of antibodies. Based on all available serological studies, WHO has estimated that around 10% of the world population, or 760 million people, might have been infected as of October 2020. https://www.euronews.com/2020/10/05/around-10-of-the-world-s-population-may-have-had-covid-19-according-to-who
The mean incubation period of SARS-CoV-2 infection is around 5 days (Li 2020, Lauer 2020, Nie X 2020). The serial interval – defined as the duration of time between a primary case-patient having symptom onset and a secondary case-patient having symptom onset – has been estimated to be between 5 and 7.5 days (Cereda 2020).
SARS-CoV-2 is highly contagious, with an estimated basic reproduction number R0 of around 2.5-3.0 (Chan 2020, Tang B 2020, Zhao 2020). [R0 indicates the average number of infections one case can generate over the course of the infectious period in a naïve, uninfected population. Read the guide by David Adam (Adam 2020) for more precious information on R0.]
As with the earlier SARS and MERS outbreaks (Shen Z 2004, Cho SY 2016), the spread of SARS-CoV-2 is characterized by the occurrence of so-called “superspreaders events” where one source of infections seems responsible for a large number of secondary infections. (Wang L 2020) This phenomenon is well described by a recent study of SARS-CoV-2 transmission in Hong-Kong (Adam DC 2020).
The authors analyzed all clusters of infection in 1038 cases that occurred between January and April 2020 and concluded that 19% of cases were responsible for causing 80% of the additional community cases, with large clusters originating from bars, weddings, and religious ceremonies.
Interestingly, decreased delays in confirmation of symptomatic cases did not influence the rate of transmission (suggesting higher rate of transmission at or before symptom onset), whereas rapid contact tracing and quarantine of contacts was very effective in terminating the transmission chain.
Other authors (Endo 2020) have estimated a k of 0.1 outside China, meaning that only 10% of infected individuals transmit the virus (k or dispersion factor describes, in mathematical models, how much the disease tends to cluster).
A relatively low dispersion factor with few infected people causing most transmissions could explain some puzzling aspects of the beginning of the COVID-19 pandemic. For example, why the early introductions in Europe of SARS-Cov-2 in December 2019 (France) and in January 2020 (France, Germany) did not result in earlier outbreaks. Or why the large outbreak in Northern Italy in February 2020 did not lead to a similar rapid spread of the virus in the rest of the country.
Understanding the reasons underlining superspreader events can be key to the success of preventive measures, so the big question is, “Why do some COVID-19 patients infect many others, whereas most don’t spread the virus at all?” (Kupferschmidt 2020).
It is possible that some individuals simply shed more virus that others, or that there is much more shedding at a specific moment of higher contagiousness in the natural history of the infection, possibly when viral load is at its peak.
Environmental conditions also clearly play a role, with crowded, closed places where people talk loudly, shout, sing or exercise being at higher risk, possibly because of the higher production and diffusion of small particles like aerosols.
A “superspreader individual“ in a “superspreading setting“ may result in a very large number of infections, as seen in the Shincheonji church cluster in South Korea where, in March 2020, one single person was estimated to have generated more than 6000 cases.
A better understanding of superspreader events may help in defining the most effective measures to reduce SARS-CoV-2 transmission by reducing the likelihood of superspreading events. We will explore below the most common “hotspots” of SARS-CoV-2 infection, where the likelihood of single or multiple infections is higher.
More information: Steven J. Phipps et al. Robust estimates of the true (population) infection rate for COVID-19: a backcasting approach, Royal Society Open Science (2020). DOI: 10.1098/rsos.200909