The first SARS-CoV-2 infections in humans occurred during fall 2019, but not enough evidence exists to confirm a global spread before early 2020

A video posted on Facebook on 8 May 2020 claims in its title that the “coronavirus quickly spread around the world as early as October 2019”. The video is mostly based on a peer-reviewed study published on 5 May by François Balloux and colleagues in Infection, Genetics, and Evolution. In the study, the researchers sequenced an unprecedented 7,666 samples of the SARS-CoV-2 virus from patients with confirmed diagnoses of COVID-19 to determine how the virus has evolved as it has spread through human populations^[1].

Reviewers contacted by Health Feedback, including Balloux, agree with the video’s claim that propagation among humans started sometime between October and December 2019. The title, however, suggests that the virus was spreading internationally as early as October 2019. This particular claim was found unlikely and unsupported by one of our reviewers.

According to the video, “a new study found that the coronavirus began spreading between humans late last year”. It reports that the findings from the research paper “indicate that the virus was transmitted [between humans] in late 2019, but before it was identified in a documented case”. The video then concludes: “the first documented case may not be related to the first jump of the virus from an animal to a human.” By analyzing the occurrence of mutations in the genomes of different SARS-CoV-2 samples, the study authors were able to establish a phylogenetic tree for the virus and reconstruct its evolutionary history (see also this thread on Twitter from the study’s senior author). The team determined that the most recent common ancestor (MRCA) of all analyzed SARS-CoV-2 samples must have emerged sometime between October and December 2019. This means that SARS-CoV-2 was infecting humans sometime during that window of time.

The video correctly reports the time window of the beginning of the outbreak established in the study and other members of the scientific community also reached similar conclusions^[2,3]. Andrew Rambaut, professor of molecular evolution at the University of Edinburgh, also published a phylogenetic study on the widely used virology.org platform, which similarly estimated that the MRCA emerged around November 2019. Another similar result was obtained by Sebastian Duchene who studies virus and bacteria evolution at the University of Melbourne.

The video also claims that the “researchers working on the study believe that people in Europe and the U.S. were being infected weeks or even months before first cases were documented”. It is indeed safe to say that SARS-CoV-2 was present in a given population before the first cases of COVID-19 were identified there. Swift detection of COVID-19 cases was hampered in the early stages of the pandemic by the similarity of symptoms to other common infections such as the flu or common cold, to the high proportion of asymptomatic carriers, and to the relatively long incubation time of up to 14 days. Recent data indicates that the virus was present in the U.S. earlier than initially thought. For example, Mount Sinai reported that the virus was likely circulating in the New York City area in late January 2020 even though the first reported case only dates to late February 2020. And on the West Coast, autopsies revealed cases of COVID-19-positive patients who had died in early February 2020, a month before the first officially recorded COVID-19 case.

However, the video states in its title that the virus was already propagating worldwide by fall 2019, specifically “as early as October 2019”, but this claim is not clearly supported by current evidence. Balloux, the study’s senior author, says that “There were clearly cases in Europe and probably in the U.S. from December 2019 onwards, and possibly November. The earliest reported case in China dates back to mid-November, and outside China to 14-22 December 2019 (France)^[4].” Other scientists state that the video’s title places the date of international spread too early. Firstly, Samuel Alizon, an evolutionary ecologist studying infectious diseases at the National Center for Scientific Research (CNRS) in France, notes that “This title is misleading because the authors report a 95% credibility interval between 6 October 2019 and 11 December 2019, which means the middle of the interval (i.e. November) should be in the title, not one of the boundaries.” As Alizon further explains: “Even by assuming such a rapid spread and an early date, it would mean that 30 days after the origin, there were only about 1,000 infections, most likely near the place of origin of the epidemic. This makes it unlikely to find it rapidly all around the world [by October 2019].”

The earliest indication of the presence of SARS-CoV-2 in France could be December 2019. A scientific paper described the case of a patient hospitalized on December 27 2019 whose biological samples tested positive for an RT-PCR based COVID-19 detection assay[4]. However, some researchers urge caution over this finding. Gaetan Burgio, a geneticist at the Australian National University, deemed the positive signals from the RT-PCR rather weak. Other scientists acknowledged the potential importance of this result but pointed out that only a sequencing of the viral genome apparently present in the samples would help to determine whether SARS-CoV-2 was already present in France in December 2019. Reuters reported similar investigations in Italy where researchers have undertaken a retrospective analysis of flu-like cases from November 2019 to determine whether they were in fact COVID-19. However, many scientists find it unlikely that the coronavirus spread as early as fall 2019. “I think it extremely unlikely that the virus was present in Europe before January,” said Paul Hunter from the University of East Anglia in an interview with Reuters.

Similarly, it has not been confirmed that SARS-CoV-2 was present in the U.S. as early as October 2019. In fact, Trevor Bedford, from the Fred Hutchinson Cancer Research Center, presented retrospective testing of thousands of samples from patients who had suffered flu-like illness in January and February 2020 and noted that none of them turned out positive for SARS-CoV-2 in January. Additionally, the testing of 2,888 nasopharyngeal samples and 148 bronchoalveolar lavage samples collected between 1 January 2020 and 26 February 2020 in the San Francisco Bay Area revealed that SARS-CoV-2 was only detected in the samples harvested during the second half of February^[5].

Furthermore, Bedford added that the genomic comparisons of SARS-CoV-2 genomes shows that “there were multiple introductions driving the U.S. epidemic and the earliest was in January”. The press reported on 15 May that an individual in Washington State with a positive serological test for COVID-19 carried out in May 2020 presented symptoms consistent with the disease as early as December 2019. However, Bedford explained on Twitter that, given the known epidemiological data, the most likely explanation is rather that this person had suffered from a flu-like illness in December 2019, followed by an asymptomatic SARS-CoV-2 infection later in 2020 when the virus was already circulating on the West Coast.

Finally, it is necessary to identify the formation of a lasting transmission chain from person to person to conclude that the virus is spreading globally. As Alizon explains, “you have to distinguish isolated cases and the epidemic wave. It is possible to have isolated cases before the beginning of the epidemic wave because random effects are very strong early in an epidemic and transmission chains can vanish by chance. The epidemic wave then corresponds to the transmission chain(s) that did not go extinct. Once the epidemic wave has started, chance has less effect and extinction in the absence of control measures is very unlikely.” In other words, the identification of isolated cases of SARS-CoV-2 in different countries is not sufficient to claim that the virus is actually spreading. Therefore, local chains of transmission must be identified before such a claim is made.

In summary, the video accurately reports the finding from Balloux’s team that SARS-CoV-2 was present among humans before the first officially reported case in China in December 2019, probably sometime between October and December 2019. This finding has also been confirmed by other studies. The video’s title, however, overstates the confidence in the possibility that the virus had spread globally by October 2019. The study itself only lists this as the earliest possible date of initial propagation between humans, not as a date of worldwide propagation. Diagnostic and phylogenetic data from late 2019 and early 2020 are scant. “In general, the date of origin very strongly depends on the virus genomes available, and currently the [earliest] genome dates from the end of December [even though] there are documented cases around the end of November. With more genomes from December, we could get a much more precise picture,” says Alizon. Some data indicate that SARS-CoV-2 spread to other parts of the world, such as the U.S., in early 2020. Other scientists have reported isolated cases in Europe, such as in France, in late 2019 albeit without a very high level of confidence and without identifying lasting local chains of transmission between people. It is thus unlikely, and unsupported so far, that the virus was globally spreading in October 2019.

 

Claim 1. “Coronavirus quickly spread around the world as early as October 2019” (Title)

This title is misleading because the authors report a 95% credibility interval between 6 October 2019 and 11 December 2019[…]

In general, the date of origin very strongly depends on the virus genomes [that were] available [for analysis], and currently the [earliest] genome dates from the end of December 2019 [even though] there [were] documented cases around the end of November. With more genomes from December, we could get a much more precise picture.

Even if the virus emerged in early October, the epidemic doubling time has been estimated by Rambaut (2020) and Volz (2020) to be initially in the order of 7 days. In Europe, it can go down to three days. Even by assuming such a rapid spread and an early date, it would mean that 30 days after the origin, there were only about 1,000 infections, most likely near the place of origin of the epidemic. This makes it unlikely to find it rapidly all around the world.

Claim 2. “People in Europe and the U.S. were being infected weeks or even months before first cases were documented” (2’54” in video)

You have to define what a documented case is. But more importantly, you have to distinguish isolated cases and the epidemic wave. It is possible to have isolated cases before the beginning of the epidemic wave because random effects are very strong early in an epidemic and transmission chains can vanish by chance. The epidemic wave then corresponds to the transmission chain(s) that did not go extinct. Once the epidemic wave has started, chance has less effect and extinction in the absence of control measures is very unlikely.

Claim 3. “The first documented case may not be related to the first jump of the virus from an animal to a human” (3’07” in video)

Again you need to define what a documented case is. Usually you will indeed notice an outbreak once it has begun to spread, so tracing back its origin requires long and hard work.

 

Claim: “People in Europe and the U.S. were being infected weeks or even months before first cases were documented” (2’54” in video)

There were clearly cases in Europe and probably in the U.S. from December 2019 onwards, and possibly even November 2019. The earliest confirmed case in China dates back to mid-November, and outside China to 14-22 December 2019 (France)[4].

Claim: “The first documented case may not be related to the first jump of the virus from an animal to a human” (3’07” in video)

It would be unusual that we can identify the real “patient zero” for an epidemic. It is possible that the first cluster identified in Wuhan was a few transmissions away from the patient zero.

[In addition], I wrote a thread covering essentially the same material but in a more casual tone.

 

REFERENCES

• 1 – van Dorp et al. (2020) Emergence of genomic diversity and recurrent mutations in SARS-CoV-2. Infections, Genetics and Evolution.
• 2 – Li et al. (2020) Transmission dynamics and evolutionary history of 2019‐nCoV. Journal of Medical Virology.
• 3- Giovanetti et al. (2020) The first two cases of 2019‐nCoV in Italy: Where they come from? Journal of Medical Virology.
• 4 – Deslandes et al. (2020) SARS-CoV-2 was already spreading in France in late December 2019. International Journal of Antimicrobial Agents.
• 5 – Hogan et al. (2020) Sample Pooling as a Strategy to Detect Community Transmission of SARS-CoV-2. Journal of the American Medical Association.