-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Coronavirus
- Thread starter Wenche
- Start date
Don't underestimate the effects of drug addiction problems. "Crack-babies", "Meth-heads", "Heroin-fiends", etc.
I will go out on a limb and say that drug addicts tend to be worse parents, and drug addicts are more inclined to eschew science in favor of dramatic conspiracy theories.
BC had one new case and one death today. Still 21 in hospital and 5 in intensive care. Most things open now. Dentists slowly coming back on line. Some medical screening procedures (mamograms etc) still real slow. No gatherings over 50 people. But hey, lots better than a couple months ago. Still 193 active cases in the province. Population of BC is a little over 5 million not including temporary foreign workers and foreign students...of which there are a lot of each this time of year.
This piece is a really good catch, and very solid thinking - those who haven't seen or who have been busy trying to calculate IFR might take a closer look at this in Scientific American: So How Deadly Is COVID-19?
Dr Dalton makes a series of cogent statements towards the end of the piece (although he is far too kind to Ioannidas in mentioning that his study was "ferociously criticized"):
"So, the mortality rate, instead of being a fixed number that distills the true essence of the virus’s danger, is actually a protean, organic, fluid metric. The rate of fatalities among COVID-19 cases “is not a biological constant,” according to a team of Oxford researchers. “Instead, it reflects the severity of the disease in a particular context, at a particular time, in a particular population.” Even with perfect data, the mortality rate is a living number, changing all the time, that is in part a reflection of ourselves. With these limitations in mind, we should be wary of using any one estimate of mortality in shaping our response to the pandemic."
But from my perspective as an emergency physician, precisely how deadly the virus is doesn’t matter right now, because the virus is deadly enough. I’ve stood on the front lines of the pandemic, and I know that this virus is no house cat. Every day for weeks, my colleagues and I have faced wave after wave of COVID patients in their 30s, 50s or 80s, many of them extraordinarily ill. Some of these people have died. Its virulence is astonishing, at least among hospitalized patients. Experienced physicians know that this is nothing like the flu."
Norbert
TSLA will win
I can't really follow this train of thought. How do "experienced physicians" know it is "nothing like the flu", if not by making making a quantitative analysis similar to calculating IFR? By looking only at the CFR in their own personal office?
What did "Oxford" produce on this question? I remember one report that made a mess of quoting other studies without critical evaluation, and apparent errors in looking at the chinese numbers (previous post on this thread). This article is a similar mess, not even making a distinction between IFR and CFR, and giving most space to the debunked Stanford study. I don't understand why this article is in the Scientific American? It seems hardly newspaper quality.
Their projections trend up slightly from here only to peak in August and fall presumably due to low herd immunity required when R_t is so close to 1 to bring it back under one. This site has had the projections I trust the most right from when they launched, but I don’t think they’ve nailed the processes for the reopening phase yet. It’s so up in the air might be best to treat the modeling as several separate, possible scenarios.
In the US R_t seems to be stubbornly close to 1 and trending up slowly due to shut down fatigue or more interactions from a reopening economy - I certainly wouldn’t dare try to nail down the cause. But masks, hand washing, and general awareness of avoiding disease (and spreading of disease) seem to be sufficient to take the real sting out. The protests will be good information, if the stats stay relatively unmoved the rest of the reopening over the summer is unlikely to move the needle either.
If the summer looks like covid-projections forecasts then I think the thing to watch, in the US, is the return to school in September. That is a big shock to the disease transmission system, a whole highly connected sub population just springs into existence. Assuming the US doesn’t do the dumbest thing I’ve heard discussed which is k-12 remote learning. The childcare role of school is critical for the rest of the economy imo. Worst case turns into a kind of damned if you do, damned if you don’t. But schools have moved a few notches up my essential list compared to a few months ago when I was weighting their potential heightened transmission rate.
Last edited:
Norbert
TSLA will win
I'd be probably be a bit more cautious about saying numbers are going down. It depends a lot on the last three days, for which in some categories the data is still incomplete. Also there was sort of a peak on May 30, and coming down from that peak doesn't (yet) mean it is going down in general. But looks more optimistic than a few days ago, and improvement may be immanent.
AlanSubie4Life
Efficiency Obsessed Member
Yeah, I like this model too, but in addition to a lot of unknowns about reopening, it doesn’t seem to react fast to some things, as according to their website they do this: "We utilize daily deaths data reported by each region to forecast future reported deaths. After some additional validation techniques (to minimize a phenomenon called overfitting), we use the learned parameters to simulate the future and make projections."
So, it has some issues, which I'm sure they try to address, but I'm not sure how...
As an example, look at Alaska Rt, juxtaposed with actual data. It clearly is not correct and the model doesn't work at all for small samples like this - because there have been zero deaths for a long time, I think.
Another problem (other than small sample size issues which suffer from stochastic issues, etc.) is the changing nature of treatment & behavior over time. If deaths come in lower than expected, due to improvements in treatment, or better protection of the elderly, I suspect the model will underestimate the actual number of cases (because IFR goes down but not sure the model can figure this out easily...). You could argue that this isn't that big a deal since fewer deaths per infection is definitely a success. However, the possibility remains that it could provide low estimates for community transmission & case growth because it thinks there are fewer infections...and then if it becomes a high enough disease burden that it starts breaking into the elderly populations...could be bad...
They have updated their IFR estimate recently, I believe, so they are aware of the issue of course, but this is definitely an issue that is hard to model, since it depends a lot on who gets infected, and that is very difficult to predict!
I'm not that sure about how much the protests will move it relative to other reopening. I can't figure out how many people are protesting nationwide, but perhaps it is in the low single millions (I have no idea - it's clearly a very large movement but not sure the numbers are well known right now)? A few million is a relatively small number compared to the number of people who will eventually need to go back to work. And they're outside, mostly young (so maybe asymptomatic and perhaps less contagious and possibly even less vulnerable). So I think it's likely to be a less serious driver of outbreaks than the general reopening. I'm sure there will be cases associated with them...but how many I'm not sure.
We'll see. I think we're already seeing the numbers move in some states, and that is very likely not due to the protests - it is too early.
I'm still hoping we'll just effectively eliminate the virus by the time school rolls around. I feel like if we just try to fake it till we make it, it's going to end in a serious problem when school arrives. And yes, it is important for kids to go back to school in order for businesses to work correctly.
That's true. It's definitely still a short term downward trend in hospitalization.
Well said. I’m not expecting the protests to suddenly fill the hospitals with covid but in another week or two we will have some data to sink into and given how sudden the change from isolation to large scale interaction was for the admittedly small population of protesters the lack or presence of a break in trend will be insightful.
RobStark
Well-Known Member
Covid-19 Projections dot com went as high as 199k American deaths by August 4, down to 160k, and now at 162k by August 4.
IHME was stuck at 127k for a while (after they switched to a hybrid model and doubled their forecast), then 135k for a while, and now 140k by August 4.
MTL_HABS1909
Active Member
It’s funny that the people protesting/rioting in the streets are the same hypocrites who were going on and on about social distancing and shitting all over Elon for wanting to reopen Fremont.
And the source for this "data"?
linux-works
Active Member
I wish I was in better physical shape and not at risk for these health issues. I would be out there protesting, too.
the timing could not be worse for public gatherings like this, but I am proud of those who do take the risk for a greater good that needs to happen.
the timing could not be worse for public gatherings like this, but I am proud of those who do take the risk for a greater good that needs to happen.
You clearly didn't read this piece very closely. Sorry you didn't like it but he actually makes a very careful distinction between case fatality rate and IFR. And since he is a physician he's talking about the clinical presentation of illness which is distinct from influenza something you would not know because you're not a clinician. You do not see kidney failure, pathological clotting, strokes, and the like in influenza (regrettably he did not go into details on these clinical issues). Not sure what turned you off but I thought you jumped the gun on this one.
Last edited:
linux-works
Active Member
mxlr: just stop.
just STOP. before you make yourself look even more like an ass.
just STOP. before you make yourself look even more like an ass.
Doggydogworld
Active Member
It has to be more than herd immunity. They show TX new infections peaking on July 3rd with cumulative infections on that date of only 1.6%. They show TX Rt at 1.04, down from 1.06 in mid-May. I don't see how a sub-2% immune population reduced R by more than 6%. Their model has picked up (or was force-fed) some other improvement in R.
Or not. Look at their charts for Sweden. They have Rt dropping below 1.0 for 3.5 weeks, then rising to 1.0 or greater for 6 weeks (1.02 today):
View attachment 548587
Yet they show infections dropping steadily from late March through September. These models have lags, of course, but how can 6 weeks of Rt >= 1.0 not produce growth?
Attachments
Norbert
TSLA will win
I'm surprised that your only explanation is that I didn't read it carefully enough.
He uses a single term "mortality rate" and defines it as "a ratio of the number of deaths from the virus divided by the number of infections". Usually I see the term "mortality" used and explained as deaths per population, but I'll put that aside. His definition suggests that he would be talking about the IFR, deaths per "true" infections. But most if not all of his numbers are CFRs, deaths per diagnosed cases. He talks a lot about the difference between known and unknown cases, but uses that as one explanation for the differences in CFRs found. When he talks in one paragraph about the "true mortality rate", which could have been the start of talking about the IFR, then it is more as the idealized CFR, so to speak. There he mentions factors such as age, which also apply to the IFR, not only the CFR.
However, he doesn't establish the IFR as a separate entity which (usually) requires statistical evaluations (based on serological studies with antibody testing), as opposed to actually diagnosed cases. He doesn't mention the serological findings in New York or Spain. He only mentions Ioannidis and the Stanford study, without much explanation of what they did, and that its methods were criticized. But Stanford wasn't criticized because there would be a problem with making serological studies in general. He even says it would be unfortunate that some researchers are even trying, as if the for example the age variance would be a secret regarding IFR studies. New York and New Jersey are only mentioned in the context of "all-cause mortality" and "excess deaths", as if New York would be relevant only in that context.
In other words, it is an article about CFR values being "fuzzy" and inconclusive, while talking its way around the existence of serious IFR studies.
This is what happens when idiots play the game of telephone. Go back to the original source. Michigan is using more accurate numbers now. Twitter poster lies and says people testing negative are being counted in Covid death toll. Reality is untested people with classic Covid symptoms like glassy lung are counted.
Similar threads
