Govt. should increase testing centers (allow pvt too) but this absurd focus by @scroll_in newspeddlers on "tests per million" is outlandish: shows ignorance about sample sizes (they don't vary much with population size for large N); the focus has to be on sample quality not size. 
1. social graph for diseases like this. The disease spreads from certain nodes. Finding one +ve case at the airport before he/she spreads the infection all over (root node for large cluster) is better than finding 100 cases deep inside the country. So not all tests are equal.
2. Sampling doesn't runs on linear scales. Assume some sort of stable state equilibrium with random tests: for the govt to get the approx. rate of infection, the required sample size doesn't increase increase linearly with the population (so "Test per million" doesn't say much)
3. Tests come with risks.10% of a locality gets tested - shows up at the nearest testing center. People might end up spreading/getting infection in that crowd. And it wastes tests which could've been spared for those who had a higher probability of testing +ve; showing symptoms.
4. Number of international passengers & their families - the international traveller % is very low for India, main focus should be that the bulk of the testing happens within 2-3 social degrees of this group for now. And institutions/orgs they attended or work in.
5. testing with the kit is the last stage of a bayesian net (temperature screening, other symptoms) with multiple steps (doctors zooming in properly on those showing symptoms) If anything, the # which matters - India had a 1.32% +ve rate; means that many kits are getting wasted.
6. High rate of accurate zoom in - French docs seem to be doing that part well: 10% +ve, proportion of kits getting wasted isn't as much. Number of tests run per million needs to rise rapidly after large clusters are detected - so the usage of those kits needs to be optimized.
Lethargic start aside (airport/flight restrictions), govt now seems to be doing whatever it can (but allow the pvt centers to test) - more important to find infected clusters to test early (not numbers)
DON'T consume doc-time/test-kits pointlessly; others might really need it.
DON'T consume doc-time/test-kits pointlessly; others might really need it.
this is the article: scroll.in/pulse/956110/i…
TL;DR
People/Admin should just listen to epidemiologists, docs & biostatistics folk - not newspeddlers. Decades of neglect of MBBS-seats & public-hospitals are going to bite badly; but clearly even richer countries are struggling.
TL;DR
People/Admin should just listen to epidemiologists, docs & biostatistics folk - not newspeddlers. Decades of neglect of MBBS-seats & public-hospitals are going to bite badly; but clearly even richer countries are struggling.
இந்த பாடங்கள் தெரியாதவர்களுக்கு செவிசாய்க்க வேண்டாம். கவலை மற்றும் முன்னெச்சரிக்கை தேவை, பீதி அல்ல. சோதனை மையங்களில் கூட்டம் நோய்களை பரப்பக்கூடும்.
Focus should be for public health folk to find clusters in towns where this might've spread: # of tests need not be high for now
Focus should be for public health folk to find clusters in towns where this might've spread: # of tests need not be high for now
Not exactly the thing as a disese which is spreading rapidly, but a good example of how a required sample size for a survey doesn't keep increasing linearly with population size in the case of surveys, for instance.
many non-linear factors at work for an epidemic; but still.
many non-linear factors at work for an epidemic; but still.
best person for this is @anupampom - think he's already mentioned the private-testing part
But this is not such a simple thing... "tests per million" doesn't mean all that much; in fact right now the focus should be to prepare the system for the peak of the epidemic
But this is not such a simple thing... "tests per million" doesn't mean all that much; in fact right now the focus should be to prepare the system for the peak of the epidemic
Also look at @sabal_abla timeline, doc at KGMC in Lucknow. Has some info on this. All other doctors and bio-science folk on our TL, please add your inputs. That's what the public needs. Not some bewakoof scroll journo's recommendations leading to a panicky crowd -> more spread.
* doc from KGMC
unroll @threadreaderapp
For a lot more details:
And this is the real mistake ⬇️ - not a low "tests per mil" metric. Test anyone flagged positive by the kind of flowchart China used (or some modified version of it). For all you know, they might've come into contact with an infected person unknowingly.
Adding excellent info by @spinesurgeon - follow his entire thread. Touches on many of the points which Anumap made. Great info spinesurgeon.
Important: same thing mentioned in the very first tweet; expanding test centers by including pvt ones. No need to stick to an 1897 set of rules.
More folk congregating at the same centers will only spread the disease.
More folk congregating at the same centers will only spread the disease.
Clarification: that sampling (generally to arrive at a %) is not meant to be analogous to testing (ideally want to catch every case).
Though district-level sampling might (hopefully not) be required to zoom in on clusters to do comprehensive testing.
Though district-level sampling might (hopefully not) be required to zoom in on clusters to do comprehensive testing.
