Early sensitive indicators of COVID-19 harm mitigation can detect vaccination impact in the New South Wales outbreak

 

(The health metric data for this post will be updated as new health reports are published and the commentary will be revised if trends change.)

Revised 7/09/2021 

Key points

Comparing all infections before and after 16 June 2021, there has been a clear shift in the frequency profile of infection from older people to younger age groups, indicating rising vaccine induced immunity in the elderly is having impact in the community

Statistical indicator ratios for disease severity (e.g. odds staying away from hospital/being forced to enter hospital) appear to be improving each week, fitting with (marginally) better protection against severe harm of the most vulnerable groups over time being successfully achieved by a targeted immunisation program. Further scrutiny may confirm these trends, although this may take some weeks of further monitoring.

Springtime in Australia. The first green shots of vaccination impact are here

The Pundit has been searching for indicators that the New South Wales Covid-19 outbreak causing so much anguish in Sydney is at least turning the corner by showing some moderation in health damage.

Gladys Berejiklian, the Premier of New South Wales,  has been repeatedly arguing that infection case numbers themselves are not the most important indicator of the status of the epidemic.

Premier Berejiklian definitely has a point. Many infections do not to cause obvious harm and it's the minor fraction of patients with severe symptoms who are the main cause for worry. Metrics about the most severely threatened patients are clearly more important than total case numbers.

That's what this post is about-- measures of severe infection outcomes such as mortality rates  or numbers of patients needing to be admitted to an intensive care unit in an hospital.

These relevant numbers are available but only mortality figures are issued at press conferences each day, and the form in which information about intensive care unit numbers are presented in recent press releases is not all that useful for answering the questions being considered in this post. 

On the last weeks of August 2021 detailed comparisons of age profile of infection before and after the major 2021 Greater Sydney COVID-19 Delta variant outbreak became available as infection age-profile graphs in NSW government weekly surveillance reports.

The most recent of these is below (ending 21 August 2021, week 33 of the epidemiology year).

It shows the proportions of COVID-19 infections in different age-groups. The black line is a reference value based on known age distribution of age from the census. If the observed infection proportions (red and blue bars) match this profile, then all matching age groups are equally susceptible to infection.

For the outbreak starting 16th June 2021, which is almost exclusively caused by a Delta variant strain of the virus, it is clear that the age-groups older than 40 years are less frequently infected than expected. These include the most heavily vaccinated age groups who were given high priority in the vaccination program.

In contrast, in this same post-16 June outbreak period, people in the age range 10 to 39 were more likely to be infected than expected. These simple expectations, as mentioned, are represented by the black line and are based solely on relative numbers of persons in the age-groups and assume all age-groups have equal infection susceptibility.

The pattern of infection was markedly different in the time period before the July 16 outbreak (lower part of graph).

In the earlier period, overall, larger proportions of older people were infected than were infected during later period, and conversely there were decreased proportions of younger people getting infected. 

Its this contrast of patterns in the two time periods is strong evidence that vaccination is having a preventative effect on infection frequency in the community. 

More people in younger age categories are getting infected, and fewer older people are getting infected in the latest wave, which is in some ways good because the young are less likely to suffer severe consequences of disease. This has already been widely noted .

In a time of universal gloom, this discovery by NSW public health workers cheers the Pundit up. 

It has provided him with the motivation to look for other indications that favourable health outcomes are continuing  to improve over time. This is a work in progress at the moment. Any progress will be reported in an updates list below.

But first, a little explanation about the choice of statistical measure needed to track this anticipated progress.

Capturing benefits of vaccination from health statistics.

There are two types of epidemic control health interventions taking place in Sydney:

1. Limiting social interactions to prevent transmission and 
2. A rapidly accelerating vaccination campaign to improve immunity to bad COVID-19 disease.

Immunisation is currently giving around 66% first dose protection for eligible participants in New South Wales. Protection is even higher in the most vulnerable older age-groups. This should be enough to ensure measurable health improvement trends over the coming days and weeks. The issue being tackled here is how to spot them.

The frequent emphasis on epidemic case numbers is understandable, but case numbers themselves are not the most sensitive and early measure of vaccination impact.

Vaccination protection gives most effect to the most severe medical conditions, namely death, admission to intensive care units or hospitalisation. Protection by vaccines against infection, especially with infection by the Delta SARS CoV2 variant causing the Sydney outbreak, is less dramatic than protection against severe outcomes of disease such as hospitalisation.

Heather Scobie and colleagues at CDC also provide good evidence on stronger vaccine protection against severe disease compared to infection protection.

This is summarised in vaccine effectiveness parameters used to model the COVID-19 outbreaks

A recent report by Antonelli and colleagues September 2021 Lancet Infectious Diseases  shows what to expect in terms of COVID-19 disease characteristics in either vaccinated or unvaccinated individuals,

Here is graphic from that report:

The odds ratio used here is a number indicating how different the group of infected previously vaccinated people in the study were from those getting infected who had not been vaccinated. An odds ratio =1.0 means no difference between vaccinated and unvaccinated people in patter of disease. The study found that there were distinct differences.

The low odds ratios for hospitalisation mean that vaccinated people were much less likely to go to hospital if they get infected as compared to unvaccinated people.

Conversely, coronavirus infections in vaccinated people were more likely to be symptom free than infections of unvaccinated people (odds ratio above 1).

Thus vaccination should improve the ratio of hospitalised numbers / non hospitalised numbers over time during a period in which vaccination coverage in improving.

A work in progress: Tracking the odds of good health in New South Wales.

For the purposes of this current blog posting, evidence like that in the UK just mentioned is taken as a starting point exploring statistics such as likelihood of hospitalisation as indicators of actual effective outcomes in the ongoing NSW Delta variant outbreak.

Useful metrics being considered are deaths per week, new ICU admissions each week, new hospitalisations, and new total cases (infections) and their relative values (eg ODDS of hospitalisation each week/ given you are diagnosed as infected that week). In thinking about this the Pundit relised that the Odds ratio, chances of staying away from hospital given a COVID-19 positive test/Chances of going to hospital  when testing positive is a precise and sensitive indicator of vaccine effectiveness in the Sydney community. It's precise because large numbers are counted, case detection in Sydney is efficient, and hopefully NSW health keep accurate and meaningful hospital statistics, and although they do have delays in releasing reports to the public, those reports are comprehensive.

The graph below shows the progression of a favourable health outcome odds ratio over time for the New South Wales outbreak , and is followed by a table of weekly event counts on which those odds were calculated, taken from weekly government reports.

The trend in odds ratio for staying away hospital and for staying away from the ICU if testing COVID-19 positive in Sydney appears to be improving.

The Pundit reckons its worth betting on vaccination at those odds.

So Punters, get vaccinated please. Don't bet on getting infected.

Later post on this topic (with updated stats for hospitalisation)

Steady predictable trend of improvement in the odds of staying out of hospital in Sydney

Further undated note 19/09/2021

.

Mortality rates are an obvious indicator of vaccination effectiveness, and are a lagging indicator. They are however published quickly, and the time series can be tracked almost in real time. To check on the conclusions raised above about hospitalisation improvements the Pundit has looked at time series for fatality rates per COVID-19 diagnosis. Here they are:

Three ratios were calculated. One based on cumulative fatality times 100/cumulative outbreak diagnoses. The second was a similar cumulative value but adjusted for lags using the average cumulative mortality values for the prior two weeks. The third was lag adjusted but based on weekly event counts, not the cumulative counts. 

The main pattern displayed is that after week 31, mortality ratios are showing an improvement, which may indicate improved protection against death from expanded vaccination coverage. Note that before week 31, values are based on low numbers of deaths per week and the information about these sad occasions is noisy in a statistical sense.

(As with all discussion here, these comments are tentative, and may need correction as more epidemic evidence emerges.)

Note Added 19-Sep-2021

These observations are echoed by clinical experience in Victoria

This article is from the September 19 issue of The Age Digital Edition. To subscribe, visit "https://www.theage.com.au".

EXCLUSIVE

Aisha Dow

Health editor

Previously healthy patients in their 40s and younger are filling Victorian hospital wards as the state encounters a new pandemic of the young and unvaccinated.

New data provided to The Sunday Age has revealed the drastically changing face of coronavirus in Victoria, with 61 per cent of those hospitalised in the latest outbreak aged 40 or younger, compared to about 20 per cent in 2020.

As parents fall ill at higher rates and the Delta strain spreads through extended families, asymptomatic children with COVID-19 are being taken in at a Melbourne hospital because no one in their family is well enough to care for them.

Doctors are also concerned about the number of pregnant women who are catching the virus. They say many of the people falling seriously ill with the disease in the state’s third wave have never been in hospital before and do not have underlying health conditions, a contrast to 2020 when patients tended to be older and with more complicated medical histories.

Data from the Royal Melbourne Hospital, which treated about 150 COVID patients in wards from July to mid-September , put the median age of those needing hospitalisation at around 46. Last year it was over 60 years and many patients were infected in nursing homes outbreaks.

‘‘ Most cases are otherwise well young people and at that median age, between 40 to 50 years old, who are getting very sick,’’ said Dr Kasha Singh, an infectious diseases physician at the Royal Melbourne Hospital. She said the typical patient was now a young person who had not been vaccinated, who had been sick for a few days and was finding it increasingly difficult to breathe. ‘‘ Other family members being infected is really common.’’

The Victorian Health Department confirmed that a number of unaccompanied children who had COVID-19 had been cared for at Northern Hospital while their parents were hospitalised.

Doctors who look after the sickest COVID patients in Melbourne are also looking after a higher proportion of younger people. Almost 50 per cent of those admitted with coronavirus to Victorian ICUs between August 4 to September 17 were aged 40 or younger. That included two children under 10 and 28 people in their 20s and 30s who required ventilation.

In 2020, just 25 per cent of the ICU admissions were for those aged 40 or below. The trend is inextricably tied to vaccination. Australia’s rollout saw older people largely prioritised and protected first ....

Further information

Michela Antonelli, PhD Rose S Penfold, BMBCh Jordi Merino, PhD Carole H Sudre, PhD Erika Molteni, PhD Sarah Berry, PhD et al. 2021. Risk factors and disease profile of post-vaccination SARS-CoV-2 infection in UK users of the COVID Symptom Study app: a prospective, community-based, nested, case-control study Lancet Infectious Diseases Open Access Published:September 01, 2021 DOI:https://doi.org/10.1016/S1473-3099(21)00460-6

https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(21)00460-6/fulltext

Bruxvoort, Katia and Sy, Lina S. and Qian, Lei and Ackerson, Bradley K. and Luo, Yi and Lee, Gina S. and Tian, Yun and Florea, Ana and Takhar, Harpreet S. and Tubert, Julia E. and Talarico, Carla A. and Tseng, Hung Fu, Real-World Effectiveness of the mRNA-1273 Vaccine Against COVID-19: Interim Results from a Prospective Observational Cohort Study. Available at SSRN: https://ssrn.com/abstract=3916094

Heather Scobie and others, MMWR 70:37, page 1284.

Monitoring Incidence of COVID-19 Cases, Hospitalizations, and Deaths, by VaccinationStatus — 13 U.S. Jurisdictions, April 4–July 17, 2021

Averaged weekly, age-standardized rates (events per 100,000 persons) were higher among persons not fully vaccinated than among fully vaccinated persons for reported cases (112.3 versus 10.1), hospitalizations (9.1 versus 0.7), and deaths (1.6 versus 0.1) during April 4–June 19, as well as during June 20–July 17 (89.1 versus 19.4; 7.0 versus 0.7; 1.1 versus 0.1, respectively). Higher hospitalization and death rates were observed in older age groups, regardless of vaccination status, resulting in a larger impact of age-standardization on overall incidence for these outcomes.