Intro: What follows is the third block of a series of daily posts that I am putting up on facebook, and have been asked to put up here for wider access. The first block covering days 1-14 is here, and the second covering days 15-30 is here. As you’ll twig from the title, I’m picking positives wherever possible. The COVID-19 situation isn’t positive, it’s absolutely dire. But as well as the frontline medical professionals there are a lot of very busy scientists working away on this in the background who I think we should pay tribute to. Best wishes to everyone at this difficult time.
Statement of interest: I am not, and never have been, medically qualified. I have an undergraduate degree and PhD in human physiology. This was more than 20 years ago, and since then I’ve worked in other disciplines. Consequently whilst I am comfortable reading the medical literature and distilling it into a few simple sentences, I am not going to comment in any detail on any of the science I’m referring to, and all of the posts will be short. Seeing as my blog posts are usually far too long and geeky, you’ll be glad of this… But I will always state my reference and provide a link. This last bit is important!
The phrase “following the science” seems to be wheeled out by our shambling idiot of a PM whose main qualifications for running the country are studying greek and being so dishonest he got sacked from his first job as a journalist. But seeing as this is supposed to be a moment of reprieve, here is a helpful analysis of why the phrase “following the science” doesn’t either reflect what is happening or mean much in the first place.
Today, bee stings. Bit of a tenous one this, it’s an observational study from China of 121 individuals from an apitherapy clinic, where they had been given progressively increasing doses of bee venom in order to become immune to it, having previously had severe reactions. Anyway, none of them got COVID-19, despite being in relatively high risk situations of exposure. It’s biologically plausible; generalised stimulation of the immune system (innate immunity) is definitely a thing. Animal studies are needed to see if immunity to bee stings confers wider immunity. Immunology is complicated and i never studied much of it. Here’s a picture from an article on apitherapy that reminded me why!
Readable (yes, really, I promise!) editorial in the Lancet on the social contract.
NICE (the National Institute for Health and Care Excellence) is the body that provides guidance to the NHS on good clinical practice, medicine and health promotion. They’re producing a steady stream of guidance notes on COVID-19. One out today is particularly good news; children whose immune systems are compromised aren’t at particular risk of severe COVID-19, and their prognosis isn’t significantly worse than the (usually good) prognosis of children in general.
Today, clinical trials. Science can easily suffer from reporting bias; the tendency to report positive results, but not studies with negative findings. Thankfully, this is much less likely to happen in medicine than in days gone by. Avoiding reporting bias is vital, because a trial demonstrating that a drug doesn’t work is just as important as a trial demonstrating success. So before trials can recruit participants, they must be publically registered on a clinical trials register. This makes it much easier to check up on any studies that mysteriously vanish instead of publishing results that might affect a pharmaceutical company share price, and the big journals will refuse to publish trials that weren’t pre-registered. The two biggest public registers are the World Health Organisation one (WHO ICRTP) and a US one called clinicaltrials.gov. There is so much traffic on the WHO one its search function has been temporarily disabled, but as of today (30th April) there are 1087 COVID-19 trials listed on the American one.
More on models. If you read the link on day 43 you’ll know that the simplest models involve 3 basic categories of people; susceptible, infected, recovered (a SIR model). But to better understand the spread of COVID-19 and how to stop it, it’s useful to divide people into a few more categories. So it’s useful to add data on how many people are infected but undiagnosed, because in the absence of any lockdown conditions, these people might behave differently to those who were infected but displayed symptoms. It’s also useful to add other categories, such as infected and needing hospital treatment. This paper describes the 8 category model the Italians are using to base their decisions on, and will doubtless be being pored over by modelling geeks the world over seeking to refine their own.
Lockdowns and flu. In the UK, we’ve been lucky that COVID-19 hit us at the very end of the winter flu season which puts huge pressures on the NHS every year. Our death toll would have been even higher if COVID-19 had hit us a couple of months earlier. But in other countries it was pretty much opposite. This paper from China demonstrates that the COVID-19 interventions massively decreased the transmission of seasonal influenza as well as COVID-19. On the downside, it also means that whatever our short term exit strategy looks like in the UK, it is likely that fairly stringent measures are likely to be required again in the run up to next winter in order to decrease the surge pressure on the NHS.
Models again today. Nice clear article on how the Imperial College model (that the UK government is basing its response on) got developed, and what the alternative types of model are. Key to us easing restrictions is models that will give us a reasonable confidence that the pandemic won’t get uncontrollably worse (Reproduction number Ro above 1). The data that get put into these models will ideally be based on the experience of other countries a few weeks ahead of us in the process, in combination with a massive ramping up of our ability to undertake testing and contact tracing. Since neither that data nor the testing/contact tracing are in place yet, it’s still not the time to be planning a party. Nobody likes lockdown, but it’s better than being the guinea pigs for experiments on when to let us all out again. The model itself and lots of related geekery are also available from Imperial’s website.
In order to safely ease restrictions, we need our reproduction number (Ro) to get below 1, and to stay there. The reproduction number is how many people you are likely to infect if you yourself have COVID-19. If Ro is less than 1, the virus will disappear. If it’s more than 1, the number of people infected increases (and tighter restrictions will be needed again). One of the key reasons why COVID-19 have reached pandemic scale is that some people experience no symptoms at all. Recognising the significance of this, China began reporting these infections separately from the beginning of April. This is important data to feed into future models of how the virus may spread, and further emphasises the need for widespread testing.
Moments of reprieve, day 41. Back to ethics and trust today, this time how it relates to contact tracing apps etc. A readable blog from the Ada Lovelace Institute. “While data can save lives at times of global public health crisis (and is already helping to do so), it can only do this effectively if its use, management and governance, even at times of crisis, is underpinned by clear rules (grounded in law, ethics and human rights) about how best to use data; and trust in institutions to use data well.“
They have also published a fairly long (56 pages) analysis of the potential use of technology during the transition out of lockdown (“Exit through the app store?”). It doesn’t make happy reading, but it does set out the key challenges that need to be overcome in order for technologies such as immunity certificates and contract tracing apps to play a useful role in exiting lockdown. The BMJ’s comments on contact tracing are also interesting, and reveal the level of distrust at the UK government’s strategy on this.
I’m mostly steering clear of pre-prints (papers that haven’t yet been through peer review), because I lack the relevant expertise and don’t want to say anything dangerous. But today’s paper is sufficiently neutral that it feels ok. It’s about barcoding. The researchers took 2000 complete gene sequences from those submitted to the COVID database, and looked at how similar they were to each other, a sort of family tree (we looked at this on day 14). They found that sequences fitted into 5 broad families, which could be described by a barcode of just 10 units. They then tested their barcoding system to look at another 4000 sequences, and found it correctly predicted which family each was in 96% of the time. Since it’s much quicker to barcode than determine the whole sequence, we can use this approach to detect whether new infections are from local transmission or imported cases.
Not a journal paper today, but trust, ethics and moral values. A nice simple blog from the Nuffield Council on bioethics. “It is not partisan politics to question or criticise Government policy. It would be reasonable to challenge any Government that asked for trust and failed to show exactly why it was trustworthy… The Government can trust us to make the effort, but it needs to more clearly demonstrate its trustworthiness in telling us why we should make it.”
Simple enough bit of science today. Yeah, this is actually me being slack because it’s now Thursday… Researchers tested over 1000 staff working for the NHS in Newcastle and then correlated the results with the occupations of those staff. Rates of infection in front line clinical staff were very similar to those in non-patient facing roles, which suggests that PPE is effective (or was a month ago when the data was being collected before the PPE started running out?). The numbers of staff getting COVID over the period of the study also follows the flattening curve, suggesting that they were picking it up in the community like the rest of us, and that lockdown restrictions are decreasing rates of infection in the community.
On why we should be really careful on the ways in which the media (of all political persuasions) are over-interpreting the results of modelling studies. Some really important points in this article (bonus points for anyone that clicks on the link and reads it). Firstly, the point of these models is to estimate the impact of interventions in reducing death and disease. They are NOT designed to give precise estimates of total numbers of deaths. Second point, they are only as good as the data that goes into them and the assumptions made. These will change over time. Thirdly, we shouldn’t try and interpret models beyond their range of applicability; Italy has the oldest population in Europe (median of 45.9, compared to 37 in China), other countries or states have particularly high levels of comorbidities. “the public reporting of estimates from these models, in scientific journals and especially in the media, must be appropriately circumspect and include key caveats to avoid the misinterpretation that these forecasts represent scientific truth”
No science today, just a short and straightforward piece from a paediatric intensive care consultant in the BMJ. George Bernard Shaw suggested that, “Both optimists and pessimists contribute to society. The optimist invents the aeroplane, the pessimist the parachute.” At the moment we need all of the aeroplanes and all of the parachutes we can muster. There is no doubt that things are bad. They may yet get worse. But we can do this. We already are.
Hopefully people have read the Sunday Times article on what a shit show the UK response to COVID-19 has been. Lots of reasons countries have had different capacities to deal with the pandemic (and on day 31, I highlighted Iceland). One of the most obvious factors is recent direct experience of other epidemics, which will always have a more lasting impact than theoretical preparedness exercises. This is certainly evident in Asia, but today’s journal paper looks at Canada. They were the most affected country outside Asia during the SARS outbreak in 2003. The key lessons learnt were the need for a public health agency (we already have these in the UK), and the need to empty out hospitals in the lead in to the outbreak (which we also did fairly well in the UK) and do a lot of consultations digitally. A key thing they’re doing markedly better than us is testing; Canada are running at 2-3 times the rate of testing per head of population than we are in the UK (data from 19th April, 14.04/1000 in Canada, 5.54/1000 UK, here). Although to be fair, high testing rates are also occurring in many countries without recent epidemic experience.
I’ve mostly been picking positive scientific advances for these daily posts. But it feels right to talk about mental health, and what psychologists have discovered when studying the impacts of quarantine. Most of the studies on the issue have been carried out during previous quarantines, so haven’t always been designed with carefully measured control groups for comparative analysis, and we need to be careful about drawing conclusions. However, common symptoms were exhaustion, detachment from others, anxiety, irritability, insomnia, poor concentration, indecisiveness and deteriorating work performance. Meanwhile, facebook continues to be its usual mix of people leading apparently perfect lives, over-achievers and doomsayers. Turn it off for a few days and see if you feel better. Lots of good resources on managing our mental health available here. Sending virtual hugs. xxx
Standing on the shoulders of giants. Today, read about June Almeida, who left school at 16, became an expert in imaging viruses and discovered the first human coronavirus in the 1960s. Her obituary and Wikipedia entry are both very readable.
The fact that COVID-19 can survive on smooth surfaces for several days was widely reported in the media a few weeks ago. This paper in the journal ‘Lancet microbe’ extends that piece of work, and also looks at the effectiveness of various disinfectants in killing COVID-19. The data on disinfectants is encouraging; all standard household disinfectants were effective. Full list if you want to read the small print before you buy: bleach (1:99), ethanol (70%), hand soap solution, iodine, chloroxylenol, chlorhexidine, benzalkonium chloride. This list covers the active ingredients of most disinfectants on the market.
Some countries have very low rates of COVID-19 infection and there’s a lot of speculation about why this is. Most of the speculation seems to be based on political leanings and fortuitous correlations, and has bog all to do with public health interventions. So today’s journal paper is about Iceland, their public health approach, and the results in terms of numbers of cases. They began targeted testing of people deemed high risk (international travellers) early, and then undertook contact tracing for all the positive cases, followed by quarantining them. Iceland actually started their population screening (to detect general spread) relatively late. The cool thing in the paper is that they have done a lot of gene sequencing which sheds light on which strains they have circulating and whether any more are arising (one person actually had been infected more than once, and had two different strains of the virus!). Iceland does have huge advantages; it’s not a major international transport hub, and they have a high testing capacity relative to population. But they do also seem to be doing a pretty effective job at containing the virus.