A common feature of a growing epidemic is that the % of cases detected by surveillance systems typically declines (just as the % detected typically increases as epidemics are brought under control)... 1/
In week ending 24 Sep, ONS estimated around 8,400 new infections per day in England (ons.gov.uk/peoplepopulati…). And on 24 Sep, the 7-day average for daily reported cases in UK was around 6,800. 2/
By week ending 16 Oct, ONS estimate was around 35,000 new infections per day in England (ons.gov.uk/peoplepopulati…), with 7-day average for UK cases by 16 Oct at around 18,500. 3/
Level of under-reporting isn't anything like mid-March (where <5% of symptomatic cases were being reported in UK), but above suggests that gap between estimated infections and reported cases is growing, so need to be increasingly cautious about interpreting raw case patterns. 4/
), but there is also a 'bottom-up' method, which my @cmmid_lshtm colleagues have been using to track R... 2/
The basic idea is that R depends on four components: duration of infectiousness; opportunities for transmission (i.e. contacts); transmission probability during each opportunity; and population susceptibility... 3/
First, let's be clear about difference between a 'scenario' and 'forecast'. Scenarios explore specific 'what if' questions, e.g. 'What if we don't introduce any control measures?' - Below are some examples from the March Imperial UK modelling report (imperial.ac.uk/mrc-global-inf…). 2/
In contrast, epidemic forecasts provide an answer to the question 'What do we think is most likely to happen?' More on scenarios vs forecasts here: washingtonpost.com/outlook/2020/0… 3/
The COVID-19 pandemic has shown power of open data and analytics in research, but these activities often aren't recognised in traditional academic metrics. New perspective piece with @rozeggo & @sbfnk: journals.plos.org/plosbiology/ar…. I'd also like to highlight some examples... 1/
A short thread about a dead salmon and implausible claims based on epidemic curves... 1/
A few years ago, some researchers famously put an Atlantic salmon in an fMRI machine and showed it some photographs. When they analysed the raw data, it looked like there was evidence of brain activity... wired.com/2009/09/fmrisa… 2/
Now of course there wasn’t really any activity. It was a dead salmon. But it showed that analysing the data with simplistic methods could flag up an effect that wasn’t really there. Which leads us to COVID-19... 3/
'Herd immunity' has been reached during previous epidemics of influenza, measles and seasonal coronaviruses. But it's subsequently been lost (and then regained). What are some of the reasons for this? 1/
Here we're using technical definition of 'herd immunity', i.e. sufficient immunity within a population to push R below 1 in absence of other control measures. But reaching this point doesn't mean R will stay below 1 forever. Here four things to be aware of... 2/
A: Population turnover. Over time, new births mean an increase in % of population susceptible. This will eventually lead to R>1 and new (but smaller) outbreaks - the more transmissible the infection, the sooner this recurrence will happen. More:
How would a 'protect the vulnerable and let everyone else go back to normal' approach to COVID play out? I see three main scenarios, each with important consequences to consider... 1/
Scenario A: Let's suppose it's possible to identify who's at high risk of acute/chronic COVID-19. Then somehow find way to isolate these people away from rest of society for period it would take to build immunity in low risk groups and get R below 1 & infections low... 2/
This would mean isolating at least 20% of UK population (if use over 65 as age cutoff) and this period of isolation could be several months (or longer if rest of population continues to be cautious, reducing the overall rate of infection and hence accumulation of immunity). 3/