❗️What're some of India's elite COVID19 calculators (@NITIAayog, @ICMRDELHI, majority of media) getting wrong? Here are some monumental blunders in elementary analysis of numbers related to COVID19 cases in India.
Rule 1: If facts are misrepresented, analysis will be wrong! /1
Rule 1: If facts are misrepresented, analysis will be wrong! /1
1. An ecological fallacy: 1st, we keep/kept deducing about the spread of infection in India by inferring global data (avg.).
Same at level of India, focused on average rather than local level data barring few states/districts. Need extensive analysis of data at local level. \2
Same at level of India, focused on average rather than local level data barring few states/districts. Need extensive analysis of data at local level. \2
Examples: We haven't done surveillance test or case study in India to estimate % of asymptomatic case, avg period of incubation, estimate for how much % of popn show symptoms in which week, avg recovery period from day of symptoms onset, etc. We apply arbitrary estimation .... \3
while ignoring "Global data is time & measure dependent, still emerging. Covid19 is still in progress, just started spread in high popn area of south Asia, Africa, South America, etc. while slow down in other parts are due to extensive measures like lockdown/isolation, etc." \4
2. Excluding facts and misrepresentation: Ignoring for now heavy fudging of data by state govts, we're systematically testing "only serious covid19 symptoms since early days, except few numbers from early days (which probably accounts for multiple tests of same individual?). \5
ICMR guidelines in early days was to test cases with symptoms which on May 18 was changed to serious symptoms (continuous high fever, breathing trouble, etc.) Previously, our testing number was too low. Our testing capacity increased multifold recently, still far from required \6 
(# of cases) No estimate for % of asymptomatic cases in India.
❗️No rough estimate for # of +ve symptomatic cases.
🧐What we have are:
official (mininum) numbers of "serious symptomatic cases" and "deaths of covid19 patients officially diagnosed by hospitals".
\7
❗️No rough estimate for # of +ve symptomatic cases.
🧐What we have are:
official (mininum) numbers of "serious symptomatic cases" and "deaths of covid19 patients officially diagnosed by hospitals".
\7
❗️Est of R0 is only lower bound on actual R0. Current R0 is min value for serious spreading, i.e., "covid19 w/ serious symptoms", not even for "covid19 w/ symptoms".
Doubling rate is for "officially recorded cases w/ serious symptoms".
\8
Doubling rate is for "officially recorded cases w/ serious symptoms".
\8
Each country has own testing criteria & ethical practices, comparing countries based on # of +ve cases, R0, doubling rate, etc. makes no sense at all (even IF assumed official data are real), leads to wrong conclusions like @NITIAayog/@ShamikaRavi👇. Compare states w/in India.\9
@MoHFW_INDIA believed 20% cases in COVID19 patients show serious symptoms (probably based on some old @WHO's reporting).
We are heavily undertesting and underreporting. If you take these into a/c then estimates for # cases in India will be monumentally large than current! ..\10
We are heavily undertesting and underreporting. If you take these into a/c then estimates for # cases in India will be monumentally large than current! ..\10
There are strong evidences that even COVID19 deaths are heavily underreported by blaming some other medical conditions in comorbidities patients and systematically since deceased ones are not diagnosed even if they showed symptoms before death. \11
Pretty plots and calculations for parameter using formula requires very limited skill. Analysis/inferences are lot more than that.
Crucial to see area-wise testing capacity and flaws in guidelines (duration for isolation, testing criteria, etc). \12
Crucial to see area-wise testing capacity and flaws in guidelines (duration for isolation, testing criteria, etc). \12
❗️Community transmission was on before March 23, policies and guidelines ignored this crucial information.
❗️Systematic (high) under-reporting of deaths counts. No post-death COVID19 test even if deceased had shown symptoms of COVID19. Ironically, burials will be as if COVID19 +ve.
Flattening the curve means flattening the curve!
@ShamikaRavi, plz study STATS 101, EPI101. We are doing adverse of "flattening". Kindly stick to plotting graphs. Let analysis for subject experts for sake of PUBLIC HEALTH. A humble, kind request!
@ShamikaRavi, plz study STATS 101, EPI101. We are doing adverse of "flattening". Kindly stick to plotting graphs. Let analysis for subject experts for sake of PUBLIC HEALTH. A humble, kind request!
