Our deep dive into the epidemiological models for the Wuhan coronavirus outbreak has brought to light a serious and important mismatch between the public messaging by the World Health Organization and US public health authorities, and the models they’re using internally to make projections about the course of the disease.
The incubation period and latent period of the virus appear from both clinical evidence (here and here) and epidemiological modeling to be much shorter than the two weeks stated publicly by the WHO and reproduced in reassuring reports widely circulated in the English-language press and by health authorities in the West.
Furthermore, despite what has been widely reported in the press, it seems very likely that patients with the Wuhan coronavirus are sometimes contagious before they exhibit symptoms, according to both a case study in The Lancet and unconfirmed reports in a European epidemiology journal.
These two factors — the fact that infected people can spread the disease without showing symptoms, and a shorter time between getting infected and becoming contagious — would make for a far faster spreading virus than what the more reassuring press reports suggest.
But before we explain why things may be worse than we initially thought, some quick definitions of technical terms used below:
- Ascertainment rate: The percentage of cases that are known to medical authorities.
- Latency period: The period of time that elapses between the virus infecting one host organism, and the time at which that host organism will typically transmit its infection to others. Note that this is different from the incubation period: the time between an infection and the first symptoms.
- Reproductive number (Ro): The average number of people who will become infected by transmission from one other person, over the life of an infection. The reproductive number of a virus in a largely uninfected population with little immunity, taking no special precautions, is referred to as a disease’s “basic reproductive number,” and may range from as low as 1.5 in an Ebola outbreak, to as high as 18 for a highly contagious virus like measles.
- Case fatality rate (CFR): The percentage of infected hosts who die of the infection. This may range from a small fractional percentage for the common cold, through nearly 100% for untreated HIV infection. It often varies considerably according to the quality of medical treatment received.
Author’s Note: I’m an expert in computational medicine, and I develop lifesaving drugs for a living. But I’m not a medical doctor, or a virologist or epidemiologist — just someone who can read the published science and understand it. We value accuracy above all, so if you have any corrections or clarifications, please get in touch with us via email or the comments section of this article. Doubly so if you’re someone whose work we’ve cited.
Reassuring claims by Western health authorities
In discussing the second Wuhan coronavirus patient confirmed in the USA, in Chicago, The Wall Street Journal (among other outlets) has reported that “People are most likely not contagious before symptoms develop,” citing “health authorities.”
“Authorities in Chicago expressed doubt that the woman who became the U.S. second confirmed case had transmitted the virus, because she didn’t have symptoms while flying,” the report continued.
Meanwhile, the New York Times, Public Health England, and even the United States’ CDC, while not explicitly claiming that asymptomatic people are not infectious, have discussed contagion in terms of symptomatic mechanisms and do not directly mention the possibility that patients may become contagious before they develop symptoms.
The Times, the Washington Post, the WSJ, USA Today, and Popular Science have reported the 14-day incubation period claim, all citing health authorities.
The widely reported claim of a 14 day incubation period and no asymptomatic transmission is equivalent to a claim that both the incubation and latent period are approximately 14 days. But based on a close reading of multiple widely cited epidemiological models, including the WHO’s own work, this claim does not appear to be correct, and the implications are concerning.
The meaning of incubation period and latent period values, and how they’re estimated
As discussed in our Wuhan coronavirus article, much hinges on the value of the latent period and the incubation period, and the relationship between them.
Longer incubation periods mean more undiscovered cases of the disease will be out there, and the effect of protective measures will take longer to show up in case numbers.
Longer latent periods lead to higher estimates of the basic reproductive number, Ro.
If the latent period is shorter than the incubation period, people can be contagious before they have symptoms, a frightening possibility.
The SARS and MERS coronaviruses both exhibit a latent period approximately equal to their incubation period at 2-7 days, while common cold coronaviruses are known to be transmissible before symptoms emerge.
Among these parameters, the incubation period is traditionally considered easier to estimate by epidemiologists, because a valuable data point can be obtained by knowing the date of symptom onset and likely time/place of transmission for even a single patient.
Directly estimating the latency period, in contrast, requires tracing the chain of transmission through multiple people so that the time of transmission is known for both one patient, and patient(s) to whom they transmitted the disease. Such measurements are also vulnerable to attribution errors at both timepoints, unlike direct estimates of the incubation period.
Latent periods can also be estimated indirectly, by exploiting the relationship between the latent period and reproductive number of a virus and the epidemic’s rate of growth. However, such estimates are highly dependent on the accuracy of the estimates of growth rate and reproductive number. While reproductive number can be estimated based on case tracking, the growth rate is harder to ascertain, as some epidemiological simulations have estimated that only 5% of coronavirus cases have been confirmed, while the percentage of cases confirmed may not be constant over time.
As we reported initially, epidemiological estimates of the latency period of the Wuhan strain have tended to be comparable to the latency period of the SARS virus, measured in days, not weeks.
However, health authorities have been claiming that the incubation period is approximately two weeks, and claiming that the latent period is similar. As we learn more, the proposition that the latent period is as long as two weeks has become difficult to square with the evidence.
Why the WHO’s 14-day latent and incubation period claims are now hard to believe
The Lancet report of asymptomatic transmission after much less than 14 days puts paid to the 14-day latent period claim all by itself.
So do clinical reports of the virus’s incubation period. Although they have ranged significantly, several recent reports suggest that the widely reported 14-day period may be a significant overestimate.
So does the epidemiology of the situation in Wuhan. The first patients became symptomatic on approximately December 8, less than seven weeks ago. If we assume a two week latent period, use the very highest estimates of the virus’s basic reproductive number (4.0) — even though the Read group generated this number by assuming a shorter latent period — and assume that no measures taken since then have been effective at all, then the epidemic could double in size each week, and have grown approximately one hundred fold from then to now.
With over 2,000 confirmed cases, this would imply that both the initial zoonotic outbreak was very large and the ascertainment rate very high. However, a survey of the Wuhan disease trajectory from the epidemiology journal Eurosurveillance concluded that recent evidence led away from the conclusion of a very large zoonotic outbreak. Moreover, both the WHO and the Read group have estimated that the ascertainment rate is very low.
Finally, consider the observed rate of growth. While the long-latent-period model would suggest that the epidemic might double in size each week, the number of confirmed cases has doubled not once but four times, from 121 to over 2000, in the last seven days, and tripled in the three days before that.
Taken at face value, these growth rates suggest that either the Wuhan strain has a shorter latent period along the lines of the SARS strain, or that there has been a sudden eight-fold increase in the percentage of existing cases that doctors are detecting. The latter interpretation seems difficult to believe.
A two week old report from the Wuhan Municipal Health Commission was cited in the Eurosurveillance report as reporting the transmission of the Wuhan coronavirus from one patient to another only five days after the first patient was infected and well ahead of any symptoms. Automatic translation of this report suggests that the explosive claim of direct confirmation of a 5-day latent period is not now present in the report, and that either the Eurosurveillance team mistakenly translated it, the claim has since been removed, or the Eurosurveillance team mistakenly cited the wrong report from Wuhan. However, a similar case (possibly the same one) has now been reported in The Lancet.
Accordingly, all the major epidemiological models that we evaluated in our comparative analysis of epidemiological models have assumed shorter latent and incubation periods than the 14 day number. The Read group assumed 4 days, the Majumdar group, 6-10 days, the Althaus group 7-14 days, and even the WHO itself, 6.8 to 8.4 days.
So even as health authorities told the press and the public about a 14 day number, the WHO was using the 6.8 to 8.4 day number in its own epidemiological modeling.
Why authorities may be sticking with the two-week latent period
It’s possible that health authorities may want to publicize a longer estimate of the incubation period in order to encourage the general population take more precautions, like a longer shelter-in-place period for possibly exposed persons. The 14 day number has been widely reported as a shelter-in-place period for travelers returning from Wuhan to other parts of China, and the CDC has used the same number as a period-of-concern recommendation for travelers in similar positions in the USA. However, this in no way justifies misleading the public about the timecourse of the disease, even unintentionally.
With multiple lines of evidence confirming that asymptomatic Wuhan coronavirus patients can be contagious and that the incubation and latency periods are significantly shorter than 14 days, the more optimistic public messaging by US health authorities is increasingly at odds with the published projections coming from researchers investigating this outbreak.