Grateful for the hard work of dedicated public health scientists at @PHE_uk like @kallmemeg and unsung others who work overtime to produce excellent reports on the variant of concern B1.617.2 🇮🇳

🧵analysis of vax effectiveness, and why interpretation of reduced VE limited.

PHE did a 'test negative case control study' w/ logistic regression as I outlined yday.

From test and vax databases they retrospectively created a control cohort of 99k who tested negative and compared to 6.4k test positive for B117 & 1k for B1.617.2


PHE found a statistically significant dip in symptomatic protection after 1 vax dose.

51% ➡️ 33% against B1.617.2 variant (yellow).
For both Pfizer and Ox/AZ (21 days after 1st dose).

No statistically significant change after 2 vax doses (14 days after 2nd dose).

This is expected from first principles: symptomatic protection is highly correlated to antibody (Ab) neutralisation level.

Mutations that reduce neutralisation will have greater efficacy impact in the linear response part of the curve, where Ab lower - like after 1 dose.

The mechanism for variants to 'evade' antibody immunity via mutations in spike protein is described here 👇

(T-cells role appears more in priming, regulation and severe disease protection.)

However there are a couple of limitations to the PHE study. We are undoubtedly very lucky they could publish something this important so rapidly and comprehensively! However, no scientific study is perfect and we have to read this study like others with nuance.

In defining B117 and B1.617.2 PHE used PCR S-gene target positive (B117) and negative ( B1.617.2) as proxies when full sequencing was not available.

PHE justify this as from the sequenced pool 87.5% of S-gene positive were B.1.617.2 and 99.7% S-gene negative were as B117.

1/3 of B.1.617.2 cases were classified on this PCR S-gene basis.

However, variants B1.351 🇿🇦 and P.1 🇧🇷 also have this trait, and both have well studied reduction in Ab neutralisation/Ab immune evasion particularly B1.351.

Given relatively low event numbers involved after 2 doses in the study, misclassification of 1 or 2 cases will have large effects on statistical power and confidence intervals - let alone change the VE point estimate⬆️.

Excellent CI charts from @steber232!

A sensitivity analysis restricting to fully sequenced cases should ideally have been provided in the supplementary. (NB 10-15% misclassification if 1/3 S-G positive for the 1 dose Ox/AZ group will not have changed much).

The bigger discrepancy between this study and previous ones potentially comes from handling/exclusion of previous infected who have a acquired a level of natural immunity.

ONS survey found 65% effectiveness for 1 dose >21 days after ONS excluded prev Ab or PCR positives.

PHE only excluded cases w/ prior PCR positive result (not Ab). 7% of England's population have tested PCR positive over pandemic. But Serology surveys and models (w/ waning Ab) place true figure ~4x higher.

Most phase 3 vax trials excluded or separated Ab+ for analysis.

A substantial fraction of PHE control group will have actually been previously infected and acquired natural immunity. The median ages of variant cases was 30-40 with 1/3 aged 16-29 = groups w/ highest exposure. Despite 'exclusion' 20%+ likely in control previously infected.

This will have significant confounding effect on vax effectiveness (VE) calculation vs control.

Multiple well conducted studies eg SIREN and CHARM (US marines) found prior infxn 80-85% protective against re-infection.

This prior immunity in the control group could reduce the observed baseline infxn rate by 16-17%+.

And correspondingly cause underestimation of Vax effectiveness against symptomatic disease in B1.617.2 for 1 dose with VE ~44%. (case control adjustment wd vary)

Now vaxxed groups wd also benefit from prior infxn immunity - boosting VE higher than otherwise the case. But differential infection rates overall (effectiveness) between baseline 0 infxn/vax and prior infx is greater so will bias toward underestimation of VE.

Taking this into account I'm not certain that confidence intervals for 1 dose vax effectiveness B117 vs B1.617.2 won't overlap ➡️ the difference might not be statistically significant.

Though my prior belief remains if Ab 'escape' 1 dose will be more affected than 2 doses.

PHE study also had 2 dose Ox/AZ symptomatic effectiveness lower than previous analysis: 89% [78,94] 14 days post 2 dose.

But as @kallmemeg noted the weighted distribution of cases in new report had AZ recipients much sooner after doses. 74% AZ in past 28 days v 46% Pfizer.

Ox/AZ vax mechanism is comparatively slow release. Non-replicating adenovirus vector can continue to persist and stimulate immune system:…

Especially T-cells:…


Thus the effectiveness of Ox/AZ may not have fully peaked at the time of assessment . When JnJ- also adenovector based- analysed vax effectiveness against B1.315 in South African trial they used a >28 day cut off for analysis.


- 1 dose VE vs B1.617.2 is likely reduced but this is less certain than presented.

- VE will be higher than headline reported results.

- 2 dose minimally affected - may take 10ks matched cases to say if difference at all.

• • •

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

Keep Current with Devan Sinha

Devan Sinha 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!


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 @DevanSinha

22 May
This is good news.

The longer it takes to statistically tell difference between vax effectiveness against variants the smaller the actual dip (if any) in protection will be.

In randomised control trials we can be confident in vax effect after only 100+ infections because selection bias and confounding variables between the vaxxed and unvaxxed comparison populations are (e)limited by the randomisation process.

This is not the case when we do retrospective observational studies - like vax effectiveness against variants in the field.

The statistical analysis is more challenging. There are biases in who is vaxxed or infected, living, working, mixing patterns, medical health/immunity.

Read 7 tweets
16 May
JCVI extension of vax dosing interval to 12 weeks is a great public health decision.

3.5X higher peak antibody levels w/ Pfizer if delayed boost vs 3 week.

Really important for protection against variants and duration esp in vulnerable🧵

We kept seeing 1 dose vax after prior infxn boosting Ab tires 3X higher than 2 doses at the standard 3 week interval. Some of us wondered if this was from optimised prime-boost interval at the time...

It now looks like a probable explanation.


3 weeks on short side of intervals, where immune response hasn't completely matured after prime stimulus.

Delay may selectively expand population of long lived plasma cells (B cell derivatives) too.

Delayed boost shown to have better Ab response across prev vax types👇

Read 15 tweets
29 Apr
Fantasy commentary letter in Lancet (not a peer reviewed study)!

It's an unreasonable proposition for most countries like the UK to achieve elimination + constraints of reality negates claimed economic and liberty benefits if attempted...

Despite pre-flight screening and quarantine Australia and New Zealand have failure rate of 4.8 per 100k travellers. SARS-COV-2 cases slip through!

Those cases slipping through results in necessary lockdown (regional or state wise) to eliminate community transmission.

Up to January 31, 2021:
Australia 7 failures, with 1 causing 800 deaths
New Zealand 9 failures, 1 causing an outbreak of 3 deaths

more since

Read 18 tweets
28 Apr
PHE study on household transmission after vax:

40-50% reduction if break through infxn

From contact tracing of 4.1k households where index vaxxed >21days and contact positive 2-14 days of index

4 Jan 4 - 28 Feb from 365k households w/ single index case and 1m+ contacts

1/ Image
No statistical difference between Ox/AZ and Pfizer efficacy for onward transmission

No statistical difference in effect by age

Well controlled:
adjusted for age of index case and contact, sex, region, calendar week, index of multiple deprivation quintile, household type


Pillar 2 symptomatic case only so may miss asymptomatic cases (=overestimate efficacy).
Assumed contact symptom >2days of index was transmission but cd be co-primary (=underestimate),
and all 2-14 days are household transmission not independent (=underestimate)

Read 4 tweets
30 Mar
Updated meta-analysis B117 mortality w/ new PHE sequenced Pillar1/2 study Oct-Dec

Major limit= underpowered
Only 36/2.8k deaths B117 cohort v 40 non-VOC
HR 0.9 [0.57,1.41]

(W/ updated LSHTM and Exeter papers, but now excl hospitalised HR studies)

🔺c40% mortality risk

Study still found B117 increased hospitalisation risk
HR 1.34 [1.07,1.66]

Unless in hospital mortality is lower in B117 (no clear difference) then over all fatality rate will necessarily be higher

A few other considerations 👇


From pathophysiology understandable why B117 might have higher morbidity/mortality.

Higher binding affinity to ACE receptor means more tissues or systems w/ lower expression may be affected

More replication (⬆️viral load) means more tissue damage and systemic inflammation

Read 4 tweets
15 Mar
COVID on the brain 🧠

Stroke like brain injury can be a life altering complication of SARS-COV-2 infection

Let's see how often is happens, what it looks like, and why.

A retrospective study was done in my own hospital trust :)

3,403 PCR+ COVID inpatients between March and May 2020

167 had neurological symptoms warranting neuroimaging

38 revealed vascular (ischemic or haemorrhagic) abnormality = 1.12% of total C19 inpatients

Mean age 59.7

Onset of neuro symptoms mean 10.1 days from admission

Even w/out COVID, many patients admitted for other reasons have strokes. However, in this COVID patient cohort when age adjusted the rate is ~5x ⬆️

87 vs 16 per 1000 hospitalised inpatients annualised rate

Read 10 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

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

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!