Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
shananalla Profile picture
@shananalla
Sep 13, 2020 • 19 tweets • 13 min read • Read on X
Scrolly
Tamil Nadu provides detailed information on its fatalities in its daily bulletin
I scraped that data (Jul-1 to Sep-10).
This thread contains a preliminary analysis, and a comparison with Karnataka/Odisha (at the end) @epigiri
Age-Profile
mean age: 63.1 yrs

Very clear clustering in the 60-80 yrs range
The Mean age has increased significantly from July-Sep
Gender-Profile

Male : 71.9%
Female: 27.6%

Fraction of Males in daily deaths has increased slowly over time
Co-Morbidiites:
63.7% of deaths had co-morbidities
36.2% of deaths had NO co-morbidities

Common ones,

Diabetes : 41%
Hypertension : 29%
Kidney disease: 6%
Admission-Death interval
The time between a patient's admission to hospital and death is a measure of quality of hospital-care as well as early-detectionn

46% of TN's deaths occur within 2 days of admission
mean Admission-Death interval ~ 5.4 days
The distribution has a long tail (some patients spend several weeks in hosp.)

it has risen significantly from July-Sep
Death-Reporting inteval
A quantitative measure of size of TN's reporting "backlog". Its the time between death and actual reporting in state bulletin

Mean ~ 2 days

It has fallen from July-Sep
Chennai vs Rest of TN (RoTN):

Chennai's mean age of fatalities is consistently higher than RoTN

CHN : 65.1 yrs
RoTN: 61.4 yrs
Fraction of Males is slightly lower in Chennai

CHN : 70%
RoTN: 72%
Co-morbidities

A much higher fraction of Chennai's deaths had co-morbidities

CHN : 72%
RoTN: 60%
Admission-Death interval
Consistently higher in Chennai, corresponding to availability of better hospital care in the capital

Mean,
CHN : 6.7 days
RoTN: 4.8 days
Death-Reporting interval

Chennai reports deaths more prompty vs RoTN

CHN : 1.7 day
RoTN: 2.2 days
Comparison with Karnataka and odisha:

- TN has the highest mean-age of deaths, OD lowest
- OD has the highest fraction of Male deaths
- TN's deaths with no comorb. are higher
- TN's A-D inteval is higher than KA
- TN's reporting lag is lower

One important caveat here is that TN and KAR are in different stages of their epidemics, and most of KAR's metrics (mean age,A-D interval,R-D interval) have improved from Aug to Sep.

Common:
- Mean age of deaths is rising with time in all 3 states
One thing thats hard to miss is the very high fraction of diabetics among deaths in all 3 states. Why does diabetes increase risk of death so much? @anupampom @amitsurg @giridar100 @drcheruvarun
Bulletins -> stopcorona.tn.gov.in/daily-bulletin/
Source code -> github.com/grill05/covid1…
Parsed TN deaths dataset (CSV format) -> github.com/grill05/covid1…
@anupampom @muradbanaji @rijomjohn @amitsurg @giridar100 @menonbiophysics @oommen @rukmini @covid19indiaorg @spacem0nk3y @countthedead @ShamikaRavi @BhramarBioStat @thattai @vishuguttal @rsidd120 @jamewils @UllasKolthur @drpugazh @kprabhdeep @gkangind @jebujohn @vincentrk
@vrsrini @nikhilrampal1 @rprasad12 @urvashi1 @sripal_chandra @patralekha2011 @rahuldave
Tamil Nadu reported 518 deaths on July 22, including 444 deaths from its "backlog" (mostly from Chennai city)

No details were provided for these 444 deaths in Chennai, so they were excluded from the above analysis (clip from July 22 TN bulletin)

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with shananalla

shananalla Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @shananalla

shananalla Profile picture
Jan 12, 2022
This is an elegant analysis, but it can't be applied directly to India, as domestic SGTF data is neither available, nor reliable (lots of BA.2)

I use a variation using TPR to show
GT(𝚶) << GT(Δ) in Tamil Nadu,partly explaining recent explosive growth🧵
Case counts cannot be used, since testing has increased in the state recently. This would tend to inflate omicron's est. advantage

Fortunately, state bulletins report district-wise TPR, so that can be used instead

ImageImage
Essentially, TPR in early part of wave can be expressed as
Aexp(rᵈ t)+Bexp(rᵒ t)

where A,B are constants, and rᵈ,rᵒ is epidemic growth rate of Δ/𝚶 (See embedded thread)

Optimization is then used to estimate A,B,rᵈ,rᵒ
for all TN districts
Image
Read 7 tweets
shananalla Profile picture
Aug 9, 2021
Understanding reinfections is critical as a large share of our population has already been infected

A summary of the (limited) data, some extrapolation, and what it means for the 3rd wave 🧵

For conciseness, NI(X) will denote "protection due to infection against outcome X"
The are 4 heads under which I'll examine the data

a) (protection against) asymptomatic infection
b) "" symptomatic infection
c) severe disease/hospitalization(/death)
d) Transmission
a) Asymp. infections:

In pre-Δ era, NI(asymp) was 81-89% in various UK/danish studies (doi:10.1056/NEJMoa2034545 etc)

A large SA community study which involved regular PCR swabbing found NI(asymp)= ~84% vs β

β is atleast as bad(immune-escape) as Δ
doi.org/10.1101/2021.0…
Read 13 tweets
shananalla Profile picture
Jul 4, 2021
ICMR-NIV neutralization study vs B.1/Delta
, among Covishield vaccinees, who were

A) uninfected, 1-dose (n=31)
B) uninfected, 2-dose (n=31)
C) (previously) infected, 1-dose (n=15)
D) infected, 2-dose (n=19)
E) Vaccine breakthrough (n=20)
Previously infected had much higher titres (even 1-dose) vs previously uninfected (even after 2-doses)

18/31 (58%) uninfected showed no Nab after 1-dose

5/31 (16%) uninfected showed no Nab after 2-dose @singhak_endo

All previously infected seroconverted after 1-dose
Higher IgG vs RBD-S1 in previously infected;1-dose

These findings add to the growing evidence that infected enjoy strong protection after 1-dose

NTAGI needs to take a call on whether 2nd dose can be skipped for this subgroup
This will allow faster coverage of high-risk groups
Read 4 tweets
shananalla Profile picture
Jul 3, 2021
The long-awaited Covaxin Phase-3 data came out this morning ( doi.org/10.1101/2021.0… )

Headline results are good efficacy against symp. infection, along with v. good efficacy against severe disease

Some other things I found interesting, and some links with other Indian data:
First, a note on the CIs

The study was designed to evaluate efficacy against symp. infection. The design/sample size was based on that

This means that any subgroup analysis (by age/variant/comorbidities etc) will "automatically" have wider CIs

This is v. much expected
SAFETY:

SAE rates were low (~0.3%)

AE were similar in vaccine/placebo groups
Read 15 tweets
shananalla Profile picture
Jul 2, 2021
This thread describes a SIR model for Delhi that uses google mobility data and genomic surveillance to refine its output

Using known parameters, it reproduces the epi-curve, variant frequencies etc fairly well

A general introduction to SIR is below
It has been calculated elsewhere that
Alpha has a ~50-60% Tx advantage over B.1
Delta has a ~60-70% advantage over alpha

So, R0b~2.5, R0a~4,R0d~7
A common generation time (5 days) is used throughout

For dates of introduction, I use data from CSIR paper (doi.org/10.1101/2021.0…)
From that,
Alpha was ~10% of all infections on Feb 1
and Delta was ~10% in 3rd week of march

This gives time of introduction

Using known R0 etc, solving the differential equations numerically gives this output for infections due to B1/∝/𝛥

Read 6 tweets
shananalla Profile picture
Jul 2, 2021
A brief description of SIR epidemic models for later reference (with extensions to variants, reinfections, mobility etc):

One of the simplest way to mathematically model the evolution of an epidemic is to use the SIR model

In this model, there are 3 classes of individuals:
a) "Susceptible": Fraction of pop. thats never been infected
b) "Infected": Fraction of pop. currenly infected
c) "Recovered": Fraction of pop. that has recovered as is now immune

Further,
G is the generation time (how long someone is infectious)
R0,M etc as below
We can write differential equations describing the rate at which people move from one compartment to the next

For eg: people recover at a rate 1/G
get infected at a rate R0/G, etc as described below
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(