Recent Sage reports have stressed that we’re still waiting on a major piece of information: the average stay in hospital for Omicron patients. It could be a game-changer. If the stay halves, the Covid-handling capacity of the NHS doubles and the risk of hospitals being overwhelmed falls quite a lot. As SPI-M says in a note published on 23 Dec:
‘A reduced length of stay would allow more capacity within hospitals to manage this, and, to a first approximation, this would scale linearly with the change, i.e. halving the length of stay would permit double the admissions.’
Might this all-important figure already be available? A Sage paper dated 22 December had some distribution graphs showing how many days Covid patients were spending in hospital. From this data, it’s possible to calculate the average length of hospital stay, by age.
The Spectator data team has done so (methodology here) and it shows a significant drop in hospital stay for all ages: now about five days for all over-50s. This is less than half the previous time the over-80s were in hospital for, with significant reductions for all age groups.
The above is calculated from a scan of the Sage graphs (the data behind its charts is, alas, not released). It’s likely to understate any Omicron effect given it looks at all UK hospitalisation – so for the first two weeks at least we’re looking at mostly Delta. But a significant drop in average stay can still be seen: whether an Omicron effect or something else, it’s significant.
When the South Africans worked out that length of hospital stay was so sharply cut by Omicron, they told the world – pretty quickly – saying on 4 Dec:
‘A significant early finding in this analysis is the much shorter average length of stay of 2.8 days for SARS-CoV-2 positive patients admitted to the COVID wards over the last two weeks compared to an average length of stay of 8.5 days for the past 18 months.’
Sage is being more coy. Indeed it says its age analysis ‘was available for participants to read’ at its last two Sage meetings but was ‘not considered or discussed’. Why not, if it’s such an important metric and such potentially significant news? Sir John Bell, Regius Professor of Medicine at the University of Oxford, recently spoke of an average hospital stay of three days, not five: if so, even better news. But it’s unclear where that figure came from.
Let’s recap. The case for lockdown was made by the Sage scenarios pointing to a vast increase in hospital Covid occupancy (peaking at anything from 40,000 to 120,000, vs 10,000 at present) unless further restrictions were introduced. This case was resisted by Cabinet members who suspected these scenarios could be based on what’s becoming known as ‘dodgy data’ – unduly pessimistic assumptions that may soon be disproven by real-world results. The three biggest factors are: case numbers, inherent severity of Omicron and length of stay in hospital.
In the last ten days, we have seen:
- Omicron falling fast in South Africa, with no lockdown
- South African hospital occupancy peaking at about half of its Delta peak, in part due to patients needing less time in hospital
- Cases still rising fast here, as are hospitalisations
- UKHSA study saying Omicron is between 50 per cent and 70 per cent less likely to hospitalise patients
- Hospital figures lifted by ‘incidental’ admissions (i.e. almost a third of Covid hospital admissions are patients being primarily treated for something else)
If hospital stay duration for the over-80s is dropping it might not have much effect – since mass vaccination, not many over-80s have ended up in hospital for Covid in the UK. Below, using NHS age bands, we see that just over half of all hospital admissions are now aged 64 and under.
There is no doubt that UK hospital figures are surging – but due to Covid-positive hospital patients being treated for something else, the real picture will take a while to settle. The chart we at The Spectator have so far been using to track the case/hospitalisation ratio is below: it uses Sage methodology and, three weeks on from Omicron, it has not (so far) recorded a drop.
A technical note on methodology can be found here.