Am delighted to say we've just published two papers in Nature arising from the ISARIC4C study. We used our pre-approved "sleeping" protocol to investigate the unexplained outbreak of hepatitis in kids last year. A thread... This is the best example I've seen of academic researchers working hand-in-glove with national public health agencies to tackle a new threat. Because we had the ISARIC4C (isaric4c.net) study infrastructure in place before the outbreak...

I've been thinking about this and the responses. There are many complex problems in the NHS but I can give you some simple, important facts that I know to be true, from my own observations. 1/n

https://twitter.com/JeremyFarrar/status/1450721827899289601

When a healthcare system fails, increasing numbers of people suffer and die needlessly. That's all. If you aren't a patient or staff, you don't see it. But this is happening, now, all over the UK. 2/n

Our discovery of 4 human genetic variants underlying life-threatening illness in Covid-19 is past peer review and has just been published: nature.com/articles/s4158…

Here's what we found: (1/n) There are 5 positions on the genome that are strongly associated with critical illness (i.e. getting so sick that you need intensive care) in our study:

We have found some robust new discoveries about Covid-19 in GenOMICC, which have potential therapeutic relevance. Readers outside of the genomics community will want to wait for peer review before acting on these findings. Preprint here for specialists:
medrxiv.org/content/10.110… By looking across the entire genome in critically ill patients we open up the possibility of discovering something completely new. We have found and replicated new, robust and potentially therapeutically-relevant genetic associations.

Which host genes play a role in viral replication/host response to SARS-CoV-2? We're systematically reviewing and meta-analysing the literature (12,000 papers so far!) to find out. Current top hit is PPIA, a gene which encodes cyclophilin A. All results: baillielab.net/covid Modern biological research often generates lists of genes implicated in the process being studied - the trouble is, it can be very difficult to assess the quality and relevance of these results.

Important new COVID data. Two things have recently altered my perception of the mechanisms of critical illness in COVID-19. Together, dexamethasone and this autopsy series (@davidadorward, Chris Lucas, @clarkdrussell @COVID_ICECAP) change everything 1/n
medrxiv.org/content/10.110… This shows that tolerance - the presence of a pathogen in the absence of tissue damage - is common in fatal COVID in humans. But it is tissue-specific. Some tissues have virus but no inflammation/damage; some, such as lung, have inflammation damage but little or no virus 2/n

I tend to get over-excited about science but I can't remember having my breath taken away like it was when @PeterHorby and @MartinLandray told the steering committee this result: dexamethasone substantially reduces mortality in severe COVID-19. 1/n Paper to follow very soon, obviously, but in my view it is better to release this information publicly now. It is up to clinicians to decide how to use it, and of course independent scrutiny is essential to convince the international community to change practice 2/n

The GenOMICC.org study could change our understanding of COVID-19 and prioritise treatments

Our aim is to recruit every intubated COVID patient

Extraordinary effort by thousands of people across the UK. Here are the latest figures 1/n

genomicc.org/uk/recruitment/ Some very small ICUs are managing to recruit lots of patients in extremely difficult circumstances. These are recruitment figures for the last 2 weeks:

GenOMICC is an open global collaboration to understand host genetic factors leading to death in critical illness. In the UK we are now recruiting in 85 ICUs, covering half of the ICU capacity nationwide.

This could make a difference in COVID-19.

1/n

genomicc.org/setup This is urgent public health work. Multiple therapies are being considered to target the host response (e.g. IL1 IL6 ACE2).

Genetic signals can give us direct evidence in humans to prioritise these therapies for inclusion in clinical trials.

Many COVID19 drug trials rolling, which is an amazing achievement. But low event rates, and variation in individual risk at recruitment. So we should brace ourselves for confusing results.
While that's happening, please remember that we *already know* how to reduce mortality 1/n In countries not affected much yet, we need to prepare to deliver the basics
- preventing nosocomial transmission
- antibiotic stewardship
- lung protective ventilation, early proning
- ECLS where it will help 2/n

Steroids are still in wide use for nCoV infection (as with SARS and MERS). But WHO recommend against, and the balance of evidence suggests harm. @clarkdrussell @jemillarni and I summarise the evidence in this paper, published a few minutes ago: thelancet.com/lancet/article… @clarkdrussell @jemillarni There's evidence of prolonged viral shedding, and increased mortality from studies (mostly observational) in other respiratory viruses, including SARS/MERS). We conclude that steroids should not be used for nCoV lung injury outside of a clinical trial

Here's our paper out today! We found 121 host genes that the flu virus needs to replicate itself. About half of them are new discoveries, and might make good therapeutic targets. nature.com/articles/s4146… This is important because if we design drugs to target these host genes/proteins, it will be harder for the virus to evolve resistance to our treatments. dx.doi.org/10.1126/scienc…

I'll be talking about how genomics might offer clues to design new treatments for sepsis at #WCIC19. Ultimately this needs global effort, led by local clinicians: download our free, open-source research protocol here: genomicc.org This work is generously funded by @stopsepsisnow , @wellcometrust and @icsupdates

Flu "snatches" the start (cap) of host RNAs in order to get its own transcripts processed and translated - a process that was thought to be random. This really careful analysis by @thefluzee shows that cap-snatching is *very* highly selective: biorxiv.org/content/10.110… What host transcripts would you avoid if you were a flu virus? We found that flu very strongly avoids snatching the host RNAs encoding many ribosomal subunits.

New, memory-efficient GWAS coexpression analysis now online.
1. Upload a bed file containing the top hits from a GWAS
2. Fine map some causal variants, and discover the cell-type specific expression signature of your disease or phenotype
baillielab.net/coexpression/r… Our manuscript reporting the method and applying it to 7 common traits is here:
ncbi.nlm.nih.gov/pmc/articles/P…

By doing simple things - recording clinical outcomes, establishing training programmes - NICST has improved ICU care across Sri Lanka But they have had very little direct funding and mostly the charity survives on donations of time and money from its workers