Longer term follow-up found ~25% of infections were asymptomatic.
In the US, we're now doing significant amounts of testing, but we're still primarily testing only those with severe symptoms. Even limiting our sample to people presenting with severe matching symptoms of COVID-19, we're finding ~20% PCR-positive for SARS-CoV-2.
Beyond that, we also have good evidence that the Basic Reproduction Number is somewhere between 2-3. Given a seeding of cases in mid-to-late January, even without social distancing measures, it would be unlikely to have 10% of the population infected yet.
Its not a significant amount of testing if we're only testing those with severe symptoms
Here in SF at least there are scores of people I know who all were sick with a "weird flu" at some point from February to now who would all kill for a test to show that they have COVID immunity. These are people who could potentially go back into society and do more work / enjoy their lives instead of being shuttered inside with anxiety
It is currently _IMPOSSIBLE_ to get any kind of test to prove that
Like how does it work? Would love to go do that
it’s free and they do it for everybody
Do we know this for sure? I've read that we're seeing re-infections of people who previously had it, but it wasn't clear exactly what was happening.
the hardest thing about this is finding real information and then accepting it.
Is a pretty hard line to take when all the facts are evolving and we’re learning more every day
> These are people who could potentially go back into society
Nobody knows if they could go back or not but the statement that they might is somehow the most dangerous form of misinformation.
Drawing the line at ~5M inhabitants, the following countries/territories have conducted more tests per million population than the US: The UAE, Norway, Switzerland, Germany, Portugal, Italy, Ireland, Austria, Hong Kong, Australia, Spain, New Zealand, Israel, Czechia, Singapore, Canada, Belgium, South Korea, and Russia.
There's the old adage "nine women can't make a baby in a month". Sometimes there are real-world limits to scaling, and the U.S. does have an incredible amount of tests to create.
And it's not like Portugal, Italy, Spain, or the Czech Republic are particularly small or particularly wealthy.
https://ourworldindata.org/grapher/full-list-cumulative-tota...
(I'm using cumulative because the daily numbers are missing for Germany, France, and the Netherlands)
I ask because the data I have shows the US solidly in the middle for the big EU countries over the past month (Germany and France don't show up on the daily graphs)
- Italy started earliest, but the daily per-capita tests have been higher every day except Apr 4: https://ourworldindata.org/grapher/full-list-daily-covid-19-...
- Austria, Czechia, and Portugal have been higher than the US more often than not: https://ourworldindata.org/grapher/full-list-daily-covid-19-...
- Belgium is on par; the UK & the Netherlands are definitely worse: https://ourworldindata.org/grapher/full-list-daily-covid-19-...
https://www.worldometers.info/coronavirus/country/south-kore...
Since they have coronavirus relatively under control and have been doing extensive testing and contact tracing for months, it's plausible that they've caught most cases. It's wishful thinking to believe the infection fatality rate is an order of magnitude lower.
Here's the result of randomized testing in Iceland, which also has the virus under control and has done even more testing per capita (https://www.nejm.org/doi/full/10.1056/NEJMoa2006100?query=fe...)
Randomized testing was still finding 0.6% of the population (outside those otherwise quarantined already) actively infected. This means even in Iceland, less than half of infections were being caught.
Iceland's CFR right now is 0.74% using deaths/recovered (or if you use an ultimate 20% hospitalization fatality rate, around 0.87%). If they missed half of infections, you get an IFR down to under 0.5%, though I'll admit they are doing better by keeping their most vulnerable population from being infected. (note the low infection rate for people 70+ at https://www.covid.is/data).
So no, not an order of magnitude lower, but 3x lower (0.7%) is looking pretty reasonable. Imperial College's latest estimate is 0.66% for China (https://www.medrxiv.org/content/10.1101/2020.03.09.20033357v...).
I'd just like to point out that this is still a far cry from the wishful thinking a lot of people are putting forth, claiming an IFR of < 0.1%, that 30% of the population have already had the disease and that herd immunity is both imminent and feasible.
These interpretations seem less likely to be true today than they did two weeks ago, and even then they were making assumptions on the optimistic side.
An IFR of 0.5% and a hospitalization rate in the high single digits is still a big frickin problem for society, when the disease has a reproduction number greater than 2 (and in the absence of countermeasures, more likely in the region 2-5).
South Korea had 5 deaths per day on both 2/29 and 4/13 that’s a horrible sign.
Looking at the raw data, I don't think Korea ever had below 3% using deaths/recoveries. Iceland is at 0.75% right now. Two weeks after peak, Korea was at a crude CFR (deaths/total confirmed) at 1.1%
Also, deaths/recoveries is the least useful metric to compare countries as people with minimal symptoms recover first, again look at the South Korea or China graphs of the number of infected over time.
and meanwhile germany's data points toward a similar conclusion as the south korean data
singapore also has a 1.5% deaths/recovered... and it's been higher in the past if you've been following closely (as high as 2%... question is where have all the new cases been coming from?)
icelands data could easily be skewed if they avoided a nursing home getting infected, given their low number of cases, and even there 1% fatality seems plausible.
Germany and Singapore are also missing many infections. A recent serological study in Germany actually argued for 0.4%.
You are correct that IFR is skewed by who the population is and what interventions are done. But then again, so is the often cited flu benchmark (where we have targeted vaccinations of at-risk groups).
(And age is a huge thing to be aware of skewing the data. e.g. the pediatric IFR from covid is on the order of seasonal flu)
And taiwan is at 1.5% CFR as well?
And still 60 or so unresolved diamond princess cases with ~7 in critical condition?
Every country is missing large numbers of cases, so CFR doesn't mean much - randomized testing is what is needed.
Imperial College's paper (linked above which gives a 0.7% population IFR) uses Diamond Princess as an input. The relative risk ratio they give for someone age 70 (mean age on Diamond Princess) is something like 4.5x (IFR ~3%), so you'd natively guess about 80 deaths from the 2,666 passengers.
The CDC is currently estimating an R0 of 5.7, which has likely been repressed by shelter in place orders. But even if it was only 4 before any interventions, there's still a chance there there are closer to 20M cases in the US rather than the 660,000 now.
My point is just that there's still a lot we don't know, and small changes in our understanding could implicate massive changes in the reality on the ground.
This would dovetail with the fact that places like sub-Saharan Africa and India haven't been hit very hard yet; they also happen to be _very_ young compared to Western countries.
If that's the case, it'd explain why developed countries are bearing the brunt of the pandemic. A very interesting example of similarly developed countries with comparable (but not equal) populations and population density, but different vaccination regimes are Portugal and Spain. Looking at their [2]respective [3]charts, the differences are stark. A similar difference can be seen between Ecuador and Argentina, even though the Greater Buenos Aires area is very population dense.
If this pans out, following the trends and [4] this map would validate that hypothesis.
[1]https://nypost.com/2020/04/14/coronavirus-death-rates-lower-...
[2]https://www.worldometers.info/coronavirus/country/portugal/
As someone else pointed out, the BCG needs a booster. While growing up I was given the initial one and two boosters. Most countries stopped doing that, which could also explain the effect. There's also the matter of which strain was used to create the vaccine: apparently some strains were better at fighting tuberculosis than others, so the same might apply (if the BCG vaccine actually made a difference) to COVID-19.
Anyway, my objective wasn't to start a conspiracy theory. We already have enough of those going around.
Obvious to me, perhaps dumb, question - if countries that were slower to change vaccination regimes have some advantage, then wouldn't we see older people in the countries that did change also at an advantage? Yet all reporting seems to be that older people are harder hit.
Based on other available data the severity of the Covid symptoms are dependent on both the length of exposure and total exposure to the virus. A cruise ship with recirculated air likely means prolonged exposure to high concentrations of the virus that won't fit most other patients exposure.
May be about double that. Per wiki article, I noticed this a few days ago (https://en.wikipedia.org/wiki/2019%E2%80%9320_outbreak_of_no...)
"Initial estimates of the basic reproduction number (R0) for COVID-19 in January were between 1.4 and 2.5,[381] but a subsequent statistical analysis has concluded that it may be much higher.[382]"
From that link wiki provided (https://wwwnc.cdc.gov/eid/article/26/7/20-0282_article) (edit: and note this is published on the CDC website)
"...we calculated a median R0 value of 5.7 (95% CI 3.8–8.9)"
So this sucks harder then.
Source: https://www.contagionlive.com/news/asymptomatic-influenza-in...
> Study results revealed an overall pooled prevalence for asymptomatic carriers of 19.1% for any type of influenza, 21.0% for influenza A, and 22.7% for influenza A(H1N1). For subclinical carriers, the overall pooled prevalence of was found to be 43.4% for any type of influenza, 42.8% for influenza A, and 39.8% for influenza A(H1N1).
> Given a seeding of cases in mid-to-late January, even without social distancing measures, it would be unlikely to have 10% of the population infected yet
This sounds like circular logic - how we know it indeed started in January and not before? How do we know how many had it asymptomatically if we didn't test 99% of the population? We're making assumption based on knowledge gathered by select sample of 1% chosen by severity of symptoms - how can we make conclusion about how it works in the rest of the population and what's the dynamics there?
[1] https://www.sccgov.org/sites/covid19/Pages/dashboard.aspx
Indeed, without serological testing, we're necessarily making some assumptions. But there's other evidence that points this way, too.
Genetic analysis based on samples from around China and the world point to a single index patient in Hubei Province around late November. China's case tracing has not gotten to an index patient, but it appears to be pretty close, and also points to an initial case around the same time.
If there was a single case in late November, then we'd expect a few hundred cases by January. Some number of these will have traveled, but many would not.
While the US clearly had major blind spots in testing, we were explicitly looking for symptomatic cases from China. The first detected case was Jan 21. There may have been a few earlier, but genetic analysis of cases in Washington point to an index patient in the Seattle area around that time.
The BPR is estimated from the known cases, so it cannot tell you anything about a hypothetical large number of unknown cases.
Antibody tests in a recent German study of households gave 15% infection rate.
However, if we apply the estimated mortality rate (0.37%) based on that to 11500 deaths in hot-spot New York, it would amount to over 3 million cases, again roughly 15% of the population.
Now perhaps Germans are so much healthier and their healthcare is so much better, making their real mortality rate so much lower. It's all speculation at this point, either way.
Also, the particular antibody tests they used appears to have a much higher FP rate than they claimed. Another study found 4% FPs rather than <1% like the press release for the Gangelt testing claimed.
If COVID-19 can spread to 15% of that particular population within that timeframe, and the estimate of a basic reproduction number of 2-3 predicts that this is not possible, then the estimate must be wrong.
> Also, the particular antibody tests they used appears to have a much higher FP rate than they claimed. Another study found 4% FPs rather than <1% like the press release for the Gangelt testing claimed.
Perhaps, but that doesn't put much of a dent into the results.
A FP rate that high will make a very significant difference in the results. (How large a difference depends on whether the 15% is a raw number or if it's been adjusted based on the expected test specificity and sensitivity. One would hope nobody would just report the raw numbers. But since all we've gotten from this study is basically a press release, expecting scientific rigor seems ill advised.)
I don't understand what's so hard about measuring population infection rate.
Assuming the population infection rate is between 1-10%, we would only need to do around ~500 randomized tests to achieve a 95% confidence interval of +/- 1%.
For example, let's say we tested 500 random NYC residents for COVID and found that 10 were positive -- a 2% infection rate. The standard error (binomial approximation) for this sample is 0.6%. So, by doing a mere 500 randomized tests, we have a 95% confidence interval of 0.8%-3.2%.
Given that overall population infection rate is so important in planning for re-opening, why are we not doing randomized testing in hotspots like NY on a regular basis? Am I missing something?
1. Efficacy of a test can only be judged against specific priors, not any possible circumstance.
2. Likewise, the goal of a test is to assist in making a particular decision, not all possible decisions.
Specifically 9% false positive rate is not useful for testing general population where effect size is expected to be on the order of 1-10%. However it is very useful in testing groups where expected effect size is much larger for example "all symptomatic people who came to hospital" at 50% or "all contacts of a known case" at 20%. The numbers are made up to illustrate the math.
Importantly our goal is not to detect all infected people ending the epidemic in XYZ days flat, it is to reduce the viral spread factor below 0.5, halving the epidemic every XYZ days. Thinking in absolutes is counter-productive.
Bearing all this in mind, the test can be useful. For example let's take a pool of people who are symptomatic, and say we expect 50% were in fact infected. The false positives (9% of the healthy 50% == 4.5%) will be outnumbered by true positives (* 99% of the infected 50% == 49%). So now you're looking at 49% vs 4.5%, a respectable 11:1 accuracy under the given priors. Not bad, for some applications.
And here's one good application: test a group of people from the pool of the currently symptomatic, quarantine them for two weeks, then release them into the wild without self-isolation rules. They will all be healthy due to quarantine, and 10/11 will be immune. If we keep releasing 91% immune groups of people into the general population the virus will die off.
* the antibody test I was referring to has 99% true positive and 9% false positive rates.
In a hypothetical scenario of 1% of the population actually having COVID and 9% testing false positive, you could ask all 10 positive results to self-quarantine and that'd probably be a pretty effective way of shutting down the virus without asking the whole population to stay home.
Once the virus spreads more, it gets even more reasonable. (If it gets crushed more and there are very few cases, it does get a little extreme to call it useful though).
99,0000.09 = 8,910 false positives 99,0000.91 = 90,090 true negatives 1,0000.99 = 990 true positive 1,0000.01 = 10 false negative
So going through your plan will isolate 8,910 + 990 = 9900 people (9.9% of the population), catching 990 actual cases and letting loose 10 cases.
In other words we can isolate 10% of the population and reduce the number of carriers by 990:10 ratio. This seems more effective than the current 100% isolation. All we need is the tests now.
Plus people to then do what they are told...
[1] https://www.wsj.com/articles/health-authorities-roll-out-new...
https://news.usc.edu/168497/antibody-testing-covid-19-pandem...
Happy to change my mind -- but there simply hasn't been any effort to use population testing in this way -- which is one of the only useful forms of testing. Otherwise, why test people in the hospital? It doesn't change treatment. Finding asymptomatics is actually useful -- and random sampling seems critical for understanding whether we are simply fucked or actually fubar'd.
[0] varies by region, but see for example this graph of London: https://pbs.twimg.com/media/EVlO-28XQAA30xu?format=jpg&name=...
Here is an article by a Stanford epidemiologist calling for random testing a month ago: https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-a...
We need data to act objectively. As far as I can tell US state governments are doing all they can to prevent random data from being released.
I see plenty of motivation in some states to try to minimize the potential magnitude of the problem, or avoid "bad" numbers of infection rates.
I don't see a motivation for "milking" the crisis. The current actions by governments are restricting economic activity and revenue streams. Many states only make money on sales tax and that's probably not going super well right now.
I agree that there is hesitation to do widespread testing, but I don't know which government would intentionally cause a recession using this virus as an excuse.
I'm sure Trump was just itching to shut down international travel, recommend reduced productivity, and mail everyone checks. He just needed a good excuse.
Another point of reference is this: according to [1], a back-of-the-envelope estimate of the number of Covid-infected is 100 times the number of dead. It's about 8000 now in NYC, so about 800k infected. That's about 10% of NYC's population, exactly the number you mentioned.
[1] https://www.solipsys.co.uk/new/BackOfTheEnvelopeCOVID19.html...
That said, it's a whole lot more likely that you're right than the December-California-epidemic folks, just saying that that type of anecdotal evidence is fairly easy to come by in the winter.
The problem is that according to the data, none of us that had the "weird cold" in February ever went to the hospital and tested positive for COVID-19. So it was probably something else, but it is weird that there was a widespread "weird cold" in New York right before the COVID-19 cases, isn't it?
By some estimates we need over 80% of the population to get it in order to achieve herd immunity, so if only 10% of the population has it we're not very close at all to this being over.
Does it mean the period for which a patient is infectious before they become symptomatic?
Or do some people never become symptomatic?
And if it's the second, do they cease to be infectious like symptomatic individuals after a time? Or are they infectious long term (aka carriers)?
That makes sense and that's a useful word, thanks!
I've heard it takes about a week before you get symptoms and then they last for 2 days on average. So even if everyone developed symptoms only 22% of people with the disease would have symptoms on a given day...
We also know from contact tracing and mass testing instances that only something like half of cases stay asymptomatic forever. It is likely that the people in this study are either asymptomatic because they no longer have the disease or because they haven’t developed symptoms yet. Also, they might just not be recognizing their symptoms. Even without covid-19, way more than 7.5% of homeless people are typically coughing.
We’d be about 1/7 the way there. Meaning we’d see another 6x current deaths to get through this. That’s not even close to acceptable.
And the 10% aren’t realistic, anyway. At least nationally. Assuming nationwide rates of 10%, then scaling up by death numbers, would put infection rates far above 100% in New York.
I didn't say that. In fact, I said the exact opposite: to get to 10% nationwide while explaining the rather large per capita differences in deaths, you'd need either negative infection rates in some places (say, Alaska), or >100% infection rates in hotspots. Neither is possible.
It seems the more important point here is that the majority of the COVID-positive individuals were asymptomatic, putting another datapoint towards the conclusion that there are orders of magnitude more people that have this disease than have tested positive.
We need more studies to gather data on these asymptomatic cases if we want to reopen the economy soon. Imagine if 10+% of the population already had COVID and where immune, we'd be much closer to heard immunity than we currently think.