Part 1. Detecting impact of vaccination on COVID-19 mortality ratios during the New South Wales outbreak

 

Updated with similar findings 20 Sept 2021.

Updated with similar and stronger findings 24 Sept 2021.

In recent posts the Pundit has been exploring ways of detecting improvement to health outcomes during the New South Wales Covid-19 outbreak.

This effort gave an early signal of vaccination progress from hospitalisation ratio numbers. 

Scrutiny of infection age profile changes over time gave a signal that over 50-year-olds may be benefiting from increasing vaccination rollout.

This post turns to grim statistics about death.  It's purpose is to support appreciation of the value from rapid vaccine rollout

Apart from loss of life being a horrible situation for affected families and friends, mortality needs careful statistical interpretation to reveal meaningful guidance.

The broad picture is that overall during 2020 in Australia the ratio of death to diagnosed infections was 2.75 percent. In 2021 the ratio is, so-far, lower, and is approximately below 1 percent.

The Pundit is interested in finding out about the evolution over time of the percentage of infected people who die from COVID-19, but detailed information to do this with best precision is not readily available. But relevant infection diagnosis numbers and death numbers are given each day in press conference reports.

We know that there is a lag between the time people get sick and the time of death, typically about 11 days.

We also know there is a lag between vaccination and the time when effective immunity is induced of about 14 days after the last vaccine dose.

A complication arises because two vaccine doses are needed for full immunity and there is partial protection from one dose of vaccine.

Putting these thoughts together, the Pundit came up with the following graphs of the situation in New South Wales.

First, a graph of  weekly numbers for mortality from locally contracted Covid-19 in New South Wales this year (updated with mortality for week 38 ending 26 09 2021 calculated as 6 days deaths x7/6):

Sadly mortality is bad but their are hints that weekly deaths may reach a plateaux. The numbers themselves give no strong indication of improvement in outcomes, and  in the early weeks 27 through 30 there are low numbers which are likely to be imprecise metrics by themselves. By the end of the period, COVID-19 mortality numbers are large each week and so weekly numbers themselves should be a meaningful metric during those weeks. .

From this mortality information, and recorded numbers of diagnosed COVID-19 infections mortality ratios were calculated at the end of each week and are presented in the next graphic:

Week 38 ends 26  Sept 2021

This graphic uses for calculations the cumulative number for mortality  up to the end of the week recorded on the horizontal scale. The crude case fatality ratio (CFR) represents cumulative numbers of cases and deaths for the same week. As there is a lag between getting ill and passing away, a lag adjusted case fatality ratio was also calculated. This used a denominator that is the average of the cumulative cases for the 2 weeks prior to the week in which the time interval for deaths finishes.

From both of these trendlines it seems likely that there has been a small improvement in mortality since a peak near week 31. Cumulative mortality by the end of this interval is likely to be about 0.7 percent. Since this may reflect some immunity established by vaccinations completed several weeks earlier,  and at that time vaccination protection was only partial (possibly about 30%) this may indicate an infection mortality ratio in the New South Wales outbreak for unvaccinated people of near 1 percent. 

Very definitely this indicates that infection the Delta virus outbreak in New South Wales is a very serious mortality risk that greatly exceeds that from from recent epidemics of influenza.

It also exceeds some low values for COVID-19 mortality hovering near  0.3 percent or less that some take to assert that COVID-19 is a relatively harmless disease.

Weekly estimates of mortality

An alternative approach to understanding mortality trends is to look at weekly totals rather than cumulative totals for health events.

Here is a graph using such a weekly value approach:

To adjust for the lag between infection and deaths, the denominator used for the risk ratio calculation was the average of  weekly new cases in the 2 preceding weeks to that indicated on the horizontal scale.

The numbers in this graph suggests that there as has been an improvement over time in the mortality ratio, with a lower chance of death when infected around week 35 of this time series as compared to 5 weeks earlier.

Its also informative to ask if ICU admissions reflect a similar trend to mortality.  Here is a graph that looks at this (added 24/09/2021): 

A lag of one week is used to adjust the ICU/diagnosed case weekly ratio shown here. Currently ICU admission numbers are only available to week 36. There is a trend to lower rates of ICU entry per infected case identified one week prior.

The progress of vaccination in New South Wales was not been dramatic until about the start of September, but most of the vulnerable elderly been vaccinated by mid-August so it is plausible to believe these graphs is detecting the early effects of vaccination on mortality and severe forms of disease. Other posts by the Pundit provide convincing information that virus transmission is already being reduced by the vaccine program.

The Pundit is eagerly hoping for continuing improvement in mortality and hospitalisation ratios, but the moderate pace of vaccination rollout suggests we will have to wait several weeks to see substantially more improvement in the mortality ratio.

In the meantime the Pundit is very encouraged by several indications that the pandemic is tangibly moderated already in NSW by vaccination. He will will remain optimistic, and keep you posted with weekly updates to these graphs.

To be continued.

See

Part 2. Vaccination is now saving 4 lives among the 1000 Sars-CoV2 infected people a day in New South Wales, and that's without factoring in infections it prevents.

Part 3. in a series. Death rates in the 2021 MelbourneCOVID-19 Delta 2 outbreak start out lower than rates at start of the Sydney2021 COVID-19 outbreak