Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
China Medicine

China, Scientists Dismiss Harvard Study Suggesting COVID-19 Was Spreading in Wuhan in August (reuters.com) 58

Beijing dismissed as "ridiculous" a Harvard Medical School study of hospital traffic and search engine data that suggested the new coronavirus may already have been spreading in China last August, and scientists said it offered no convincing evidence of when the outbreak began. From a report: The research, which has not been peer-reviewed by other scientists, used satellite imagery of hospital parking lots in Wuhan -- where the disease was first identified in late 2019 -- and data for symptom-related queries on search engines for things such as "cough" and "diarrhea." The study's authors said increased hospital traffic and symptom search data in Wuhan preceded the documented start of the SARS-CoV-2 pandemic in December 2019.

"While we cannot confirm if the increased volume was directly related to the new virus, our evidence supports other recent work showing that emergence happened before identification at the Huanan Seafood market (in Wuhan)," they said. Paul Digard, an expert in virology at the University of Edinburgh, said that using search engine data and satellite imagery of hospital traffic to detect disease outbreaks "is an interesting idea with some validity." But he said the data were only correlative and -- as the Harvard scientists noted -- cannot identify cause.

This discussion has been archived. No new comments can be posted.

China, Scientists Dismiss Harvard Study Suggesting COVID-19 Was Spreading in Wuhan in August

Comments Filter:
  • by omnichad ( 1198475 ) on Tuesday June 09, 2020 @03:55PM (#60165764) Homepage

    This is how we get junk science reporting. They should not even publish findings at such a preliminary stage. It's going to get published in a way that makes them look incompetent later.

    • The number of people I know, including myself, who had serious upper respiratory viruses last year is hard to ignore. I was nearly hospitalized, my SP02 was hovering just above 90%. I had similar symptoms, this would be April -> July last year. I know a dozen or so people who had similar symptoms in November.

      But there's absolutely no way to know they were remotely related to SARS-CoV-2. I am not sure how they are factoring out the possibility that there are other viruses out there that are also pretty ba

      • by iikkakeranen ( 6279982 ) on Tuesday June 09, 2020 @04:36PM (#60165892)

        Considering the speed at which this virus spreads when there's no social distancing in place (doubling every two days or so), it's fairly certain that the same virus was not going around in the US much earlier than thought. Or anywhere else for that matter - we had unrestricted international travel so it would've swept across the globe in a way impossible to ignore.

        • Yeah this defies all logic and sense.

          If I infect 5 people over a week they're all going to be in one place. You'll get little localized hot spots. If I infect 1 person who infects 1 person who infects 1 person who infects 1 person over 6 months just picturing this as a brownian motion simulation you're going to get way more spread geographically.

          It defies sense that one region of a country wouldn't see any spread for 6 months and then very very suddenly move to europe. And then very very suddenly move t

          • Well, it could have festered in an isolated community for several months -- China is still very rural. But that would not have affected hospital car parks.

            One thing is certain, those Harvard reviewers will do well applying for federal research grants.

          • I've been thinking about this. The only way that makes sense is that it is an older strain. It's possible that this has mutated since humans have had it. And now we have a version that doesn't affect kids less and spreads more rapidly.

        • by gweihir ( 88907 )

          Very much so. You cannot overlook something with exponential growth with a factor around 1.5/day for long. It is just not possible.

      • Well you could get an antibody test which will then show you for sure that you didn't have SARS-CoV-2, which you almost certainly didn't have.
        • That's not true. Many antibody tests have a 3%+ false positive rate. So if you live somewhere with say 1-2% prevalence of the virus using lab-grade antibody tests your Walgreens test is 3x more likely to be a false positive than a real positive.

          Most consumer antibody tests are kind of worthless right now. But are helpful for large scale testing. Abbot does have an antibody test with 99.8% specificity so only 0.2% false positives. But again, when you're testing everybody that's still a lot of false po

    • by 93 Escort Wagon ( 326346 ) on Tuesday June 09, 2020 @04:14PM (#60165832)

      The big science publishing houses have definitely done a lot of jenky, unethical stuff - but the current trend of just throwing all sorts of crap up on Arxiv, or "publishing by press release", doesn't make for good science.

      We really need a well thought out, researcher-driven, peer-reviewed system for publishing research results.

      • by DogDude ( 805747 ) on Tuesday June 09, 2020 @05:00PM (#60165966)
        Well, this is the system in action. Somebody posts crap, and scientists call them out on it. Now, the researchers who did this likely won't get jobs in their fields, and the lab that they belong to, and Harvard both lose some credibility points. Just because the media published some unverified junk doesn't mean that the system isn't working.
        • and I will refer you to the "MMR vaccine causes autism" idiocy, published in The Lancet.

          And not removed until 11 years later.

          I'm not sure the system is working, not while vested interests, big money, and policital agendas are involved.

    • Re: (Score:3, Insightful)

      by lkjlkjlkj ( 6410030 )
      Scientists publishing preliminary findings isn't the problem. Journalists re-publishing preliminary findings as "scientists find that definitely for sure [insert preliminary finding here]" are the problem. If anything, scientists *under*-publish their negative/low-power studies resulting in publication bias [wikipedia.org]
    • by vlad30 ( 44644 )
      The problem in this situation is we have a government that covered up the outbreak from the start I made a comment earlier about PPE being bought up as early as November 2019 https://slashdot.org/comments.... [slashdot.org] this information adds to that. A full investigation would need not only these satellite images and PPE sales and shipments info. It needs to includes statements from Chinese citizens however any with info are very likely to keep quiet or have been "reeducated" The official released info has already bee
    • by gweihir ( 88907 )

      Indeed. Especially because this is pretty irrelevant to the currently still pressing medical matters. Proper peer-review would definitely have been a good idea, but I guess some people just could not resist making a splash (and making asses out of themselves).

    • This is how we get junk science reporting.

      Give Msmash credit where due; these weren't scientists or even Chinese scientists. The exact words were:

      China, scientists

      It's clearly military procurement nomenclature; for example: grenade, fragmenting.

    • This is not even junk science. It's 'intelligence analysis' presuming to uncover that China has been hiding stuff from us.
      The writers understood that anti-China articles like that are career making moves, even if the article gets shot down. Like claiming the US protests have Russian and Venezuealan origins.

  • by kbahey ( 102895 ) on Tuesday June 09, 2020 @04:19PM (#60165848) Homepage

    You can see some satellite pictures in this article [bbc.com].

    Maybe there was something in Wuhan in summer/autumn of 2019, per this study.

    But if it was, I really doubt that it is SARS-CoV-2 as we know it since January 2020.

    Why? Because if it was the same infectiousness and virulence, it would have caused everything we saw in Italy, Spain, New York, Sao Paolo, Iran, and elsewhere in the span of a month or six weeks.

    Maybe it was a previous less virulent or less infectious strain? Maybe it was something else.

    • Was going to post the same thing - doubtful it was the same thing, or we would have seen the infection spread MUCH earlier and further.

    • The article also claims higher traffic at children's hospitals. Which also doesn't track with SARS-CoV2.

    • by crunchygranola ( 1954152 ) on Tuesday June 09, 2020 @08:26PM (#60166340)

      I have read the paper, checked the references, and took a close look at the trends they purport to find.

      They have 140 parking lot occupany measurements over a 28 month period for 6 different parking lots, or a bit less than measurement per hospital parking lot per month. They then combine these 6 occupancy graphs and apply LOESS local regression fitting to generate a single curve that they claim represents a hospital parking lot occupancy trend.

      A couple of things about using LOESS regression. First it requires fairly large, densely sampled data sets in order to produce good models [nist.gov]. Second LOESS makes it really easy to extract a pattern or trend and because it is a proper stats method it is often taken for granted that the pattern so derived is meaningful. [fromthebot...heheap.net] but that does not mean that it is.

      If you look at the fitted curve they generate overlayed on the original sparse parking lot graphs, the underlying data is extremely spikey, jumping all over the place, with no visually evident match to their trend curve. The use a 0.40 smoothing span, i.e. 11 months of data in each fit, so they are getting a smooth curve but not capturing much local behavior. They provide no information about the "goodness of fit". So this is curve fitting for beginners, just dump the date into a stats package and if it produces a pretty curve, publish!

      What is more the curve shows no response at all to the actual pandemic outbreak in December. When the stuff hits the fan, outbreaks occur, there is a lockdown, no inflection at all, it just continues its gentle meander, you could not detect the actual pandemic outbreak on it - which is to be expected since it really represents a year's worth of parking lot behavior

      It gets worse.

      They present data on flu cases, which precisely matches the search surges on "cough" (which is the *major* symptom of COVID-19) right up to the actual December pandemic outbreak, when it suddenly has a huge vertical spike. So the flu entirely explains all cough searches before the well known outbreak. That leaves their "diarrhea" search, which is associated with COVID-19, not flu, but is not a distinctive or prominent symptom. For that, there is a modest rise in August, but this elevated level then remains perfectly flat for four months, then spikes exactly at the same moment as cough, with the actual outbreak.

      The final annoying thing about the paper is the dishonesty of their write-up:

      While we cannot confirm if the increased volume was directly related to the new virus, our evidence supports other recent work showing that emergence happened before identification at the Huanan Seafood market. These findings also corroborate the hypothesis that the virus emerged naturally in southern China and was potentially already circulating at the time of the Wuhan cluster.

      This would have flown if the paper had been published in February, when we still sorting out the origins of the pandemic. But by late February we knew that the Huanan Seafood Market, the first case cluster identified in late December was not the origin of the outbreaks as infections not related to the market could be trace back a month earlier. So they are not actually contributing anything at all in this regard.

      But they also imply that recent work about the emergence is somehow consistent with this studies claims. That is pure BS. We have an extremely powerful dataset [nextstrain.org] consisting of the thousands of SARS-CoV-2 genomes from all over the world that coalesce on a single "first infection" in early November. We know that all the COVD-19 cases in the world stem from that first infection. There are no examples of a genome that split off before that event. Within a month of that first infection the virus was present in

      • I think the purported implication of the flu correlation would be that they weren't testing for COVID-19 so it would have been listed as something else, like the flu.

        And the flu is actually a disease, a pattern of symptoms, not to be confused with influenza infection. Flu is often caused by influenza, but it is also caused by... wait for it... coronaviruses! And influenza often only causes the Common Cold.

        The flu is already what you have if you're infected with COVID-19 and have a certain pattern of symptom

        • Thanks for the BS. Nothing in this post is true.

          Flu and the influenza virus have the same relationship as COVID-19 and SARS-CoV-2, which is a clinical symptom complex and a specific virus respectively.

          The common cold is a mild syndrome caused by a few dozen viruses, four of which are coronaviruses, none of which are SARS-CoV-2. Flu, the common cold, and COVID-19 are never the same, ever.

          https://science.slashdot.org/story/20/06/09/1856238/china-scientists-dismiss-harvard-study-suggesting-covid-19-was-spreadi

      • by kbahey ( 102895 )

        Thank you.

        Very informative analysis.

      • by scrout ( 814004 )
        LOESS and this methodology is used in so called climate modeling...using 30 years of data to describe millennial changes. Still interesting use of available data looking at parking lots. There was also a story about cell phone connections being terminated in the area?
        • Given the very dense set of data which represent actual trends LOESS modeling is well suited for this purpose. Of course you are lying about "30 years of data to describe millennial changes" and you know it (... or could you be so ignorant that you don't? Inquiring minds want to know.)

  • by Toddlerbob ( 705732 ) on Tuesday June 09, 2020 @04:42PM (#60165914)

    I could sort of understand writing a news story about non-peer reviewed work if urgency is needed to get evidence, any evidence out there. But this? What's the urgency? Why can't we wait for "peer review?" Rutger Bregman is right. Our news system has to stop reporting stuff just because it's "edgy." Wait for the peer review! Especially when one way or the other, it's not going to affect how we'll deal with the situation.

    • by gweihir ( 88907 )

      The "news" system is for-profit. The worse the news, the better for their profit. They thrive on drama, fear and panic. Hence they are actively looking how they can have more of that.

  • Pre-COVID19 (Score:4, Interesting)

    by crow ( 16139 ) on Tuesday June 09, 2020 @04:51PM (#60165950) Homepage Journal

    The virus we're facing now is believed to be a hybrid between a bat coronavirus and a pangolin coronavirus. That means some person or animal was infected with both at the same time, and a random glitch in viral reproduction caused a copy to jump between the two and create the new hybrid we face today. It's perfectly plausible that in August there was an outbreak of one of the two precursors to the pandemic, and sometime in November or December someone caught both that precursor and the other, resulting in the virus we know all to well.

    As others have said, it couldn't be COVID-19, or we would have had global infections noticed in October and November, not January.

    • by mveloso ( 325617 )

      Maybe this is how SARS-COV-2 came about - a coronavirus outbreak, and an infected person went to the market and it crossbred in a human?

    • WTF? Viruses don't breed. There is no such thing as a hybrid virus.
      • by crow ( 16139 )

        Actually, you are wrong.

        During the RNA replication, occasionally there is a glitch, and the copy jumps mid-stream to a different copy of the virus. As long as the segment in question is close enough, it can jump between two different viruses and merge portions of both of them.

        There are lots of stories on this if you Google it, such as this one at ArsTechnica:

        https://arstechnica.com/scienc... [arstechnica.com]

        • Interesting. Thanks for the correction! I searched for "hybrid virus" and got nothing on the first few pages of results but computer viruses and genetic engineering.

          Seems like if one cell is infected by two different virus species at the same time they can combine somehow?

          • by crow ( 16139 )

            Yup! And in your defense, it was a surprise to me that it works that way when I first saw the article.

  • by spitzak ( 4019 ) on Tuesday June 09, 2020 @07:50PM (#60166278) Homepage

    Doesn't genetic sequencing of the various variations of the virus disprove this? If it started earlier (at least if it was transmitted from more than one person) then wouldn't there be greater genetic variety? They can also show ancestry by how the variations compare to each other and I though showed pretty conclusively that there was only one variety in December in Wuhan.

    • No, there are limited samples from China, and the original source hasn't been found.

      When you find the source, you can do all that.

      China doesn't share data on the source. They only shared limited data on the strains that had spread to other regions.

      • by spitzak ( 4019 )

        I thought it could be pretty accurately determined at what time there was one variety of the virus only using samples from outside China.

Keep up the good work! But please don't ask me to help.

Working...