Vaccines save Sydney's bacon, yet again. This time Omicron bacon.

Hints that the Omega wave in Sydney is less severe in disease impact than Delta 

Further decoupling of hospitalisation numbers from daily cases 

Graph 1.To allow for the usually 11 days of delay before newly infected cases turn into severe disease, the points for average daily cases are adjusted for the 11 day lag. Thus average daily cases plotted use a centred 7 day calculation window of reported new cases, with the averaging window lagging by 11 days prior to the day of hospital numbers plotted on the graph.

There is much talk about the less severe nature of the Omicron variant COVID wave.

A good lead on this topic is this tweet (linked pdf here):

The Australian Omicron wave is attracting it's fair share of attention, with much valuable chatter on Twitter.

Several previous posts at GMO Pundit have mentioned changes in disease severity during the rollout of vaccination in Australia.

Twitter discussions reveal that a good way to display these changes is by plotting the logarithm of disease metrics against time.

The graph displayed above does this for New South Wales epidemic numbers for the last half of 2021.

The graph plots log (average daily cases) in blue and log (reported hospitalisation numbers) in orange against the date at which hospital numbers were reported.

It can be seen that the orange curve representing cases progressively gets lower and lower than the blue curve as the weeks of the epidemic rage. This indicates that out of every 100 infections there is less and less time spent in hospital. The trend is either fewer people in hospitals, or shorter stays in the wards, or both. 

For those worried by logarithm's and lag corrections,  here is  a direct plot for NSW of the same kind of information (thanks dbRaevn on Twitter)

Graph 2. Direct number plots without lag adjustment.

Watch the gap.

Getting back to Graph 1 at the top, the gap between the two curves is also plotted in grey. This gap number log (cases)-log(hospital numbers). This dramatically displays the overall time trend from July to December of  decreased epidemic severity .

Most interestingly, the grey curve indicates that acceleration of gap widening is occurring in the latest days of the epidemic.

Is this further widening of the gap caused by a shift from mainly Delta variant (severe disease) to mainly Omicron (with less severe disease)? This will get much attention over the next two weeks.

Whatever the answer to this question, the grey graph curve powerfully emphasises the effective progressive mitigation of epidemic severity achieved from strong vaccine rollout.

Vaccines saved Sydney's bacon yet again.

Afternote.

1 Jan 2022, updated 5 Jan 22

The Pundit did some further work to visualise these trends .

First he realised the lag time should better be examined at 5 days for hospitalisation. Here (Graph 3) is the same graph approach using a 5 day lag and a 3-day centred window for daily average cases.

Graph 3. 3-day case average window, 5-day lag adjustment for hospitalisation, updated 6 Jan 2022

Similar patterns are seen to those shown in the first graph, but perhaps the trend to gap widening at the end of the date series started at the end of November. The overall gap pattern might be waxing and waning of vaccine immunity with a vaccine booster effect kicking in at the end of November.

 

 ICU numbers also show pronounced uncoupling from case numbers.

A similar examination was carried out on ICU occupancy in New South Wales (Graph 4) . This time ICU numbers were adjusted for an 8-day lag after symptom onset with the following pattern seen:

Graph 4. Updated 6 Jan 22. 3-day centered case average window, 8 day ICU lag adjustment

Here the gap between ICU curve in orange and daily case numbers in blue starts widening at the start of November.

To the Pundit this early onset means that Omicron variant proliferation could not explain all the uncoupling seen in November and December.

It keeps alive the proposition that widespread vaccine booster uptake is a main explanation for these welcome late 2021 relative improvement of ICU occupancy trends in the epidemic (with the emphasis on relative compared to case numbers).

Of course the Pundit realises (from several Twitter threads) that several other data nerds are discussing these trends using slightly different approaches, but he does commend the usefulness of differences between logs -- the grey gap curves on these graphs --- as a way of quickly finding out these patterns that is simple to do for people who can juggle spreadsheets.

It also seems prudent to emphasise that exponential growth in total infections  may well obliterate these welcome disease severity blunting trends , and that crisis overload of hospitals is a horrible scenario to consider. Discussions from the UK Financial Times suggest that normal ward system overload rather that ICU overload maybe the greater challenge during an Omicron wave:

6 Jan 22.

Here is Matt Hopcraft's modelling of NSW trajectories. It confirms the trends noted above, and illustrates critical levels of hospital ICU occupancy. The pattern occurring with the Omicron wave is not that which Matt expected for Delta, but perhaps Matt does not factor in an extension to "uncoupling" trends starting in November.