Australian COVID-19 response may have saved 16 000 lives

THE COVID-19 global pandemic has already led to more than 33 million infections and more than one million deaths worldwide as of 29 September 2020. Australia had an extremely successful response to the first wave with only 196 deaths by 31 July.

The total deaths in Australia substantially increased to 875 as of 29 September as a result of the second wave of infections in Melbourne. However, even considering this second wave, the rate of 34 deaths per million population remains substantially lower than countries in Europe and the Americas.

Part of the successful response has been early use of social distancing and restrictions on public gatherings. However, these measures can have high social and economic costs as we have seen with the prolonged lockdown in Melbourne. Moreover, there is a risk that the success of the Australian response and the resulting low mortality rate may have led to a degree of complacency about the virus, particularly in jurisdictions outside of Victoria.

As we do not know how many people would have died if Australia’s response had been different, it is not possible to determine the actual number of deaths avoided by the implementation of social distancing and other measures such as early border closures. Here, we use data on all-cause mortality from England and Wales during the peak period of COVID-19 mortality from March to May 2020 to estimate the number of excess deaths that could have resulted in Australia if we had an outbreak of similar size and nature as that in England and Wales. Both regions are island nations and have a geographical advantage in containing outbreaks over non-island countries. However, in contrast to Australia, as of 29 September there had been over 42 000 deaths and a death rate of 618 per million persons in the UK as a result of COVID-19.

Estimating relative risk of all-cause mortality in England and Wales during the outbreak peak

We estimated the increased relative risk of all-cause mortality in England and Wales from the COVID-19 outbreak by dividing the total deaths from all causes for week 11 to 21 of 2020 (week commencing March 9 to week commencing May 18) by the mean number of deaths for the same weeks averaged over the years 2014–2018. We used this 5-year period rather than 2015–2019 owing to the unavailability of comparable Australian weekly death data from the Australian Bureau of Statistics (ABS) for 2019.

We estimated relative risks stratified by age group and sex due to well documented differences in COVID-19 case fatality by age and sex. We based our age groupings on those provided in publicly available deaths data provided by the UK Office of National Statistics. These estimates are shown in Box 1.

Box 1: Estimating the relative risk for death in England and Wales during weeks 11 to 21 2020, and Italy during March 2020

1. Data source: Office of National Statistics website. Deaths registered weekly in England and Wales, provisional (viewed July 2020)

2. Data source: Italian Istituto Nazionale di Statistica. Italian estimates calculated by dividing the number of deaths across 6868 communes in Italy for the month of March in 2020 by the mean number of deaths in March over the 5-year period from 2015 to 2019.

We used all-cause mortality rather than COVID-specific mortality for two reasons.

First, there has been known undercounting of COVID-19 cases and COVID-related deaths in many countries including the UK.

Second, in a large outbreak there is the possibility of increased deaths that are not directly caused by COVID-19 but are due to the impact of the outbreak on the health system and people’s use of health services. This is supported by data from the Office of National Statistics showing that in weeks 13 to 20 of 2020, there was an excess of non-COVID-19 deaths of up to 32% compared with the last 5 years.

Applying estimated relative risk for all-cause mortality in England and Wales to the Australian population

We estimated the baseline risk of all-cause mortality in the Australian population by estimating the weekly mean number of deaths from any cause in age ´ sex strata for weeks 11 to 21, averaged over the 5-year period 2014–2018 (the most recent years that weekly Australian deaths data were available from the ABS). We then estimated the total deaths that could have been experienced if Australia had experienced a similar outbreak to England and Wales by multiplying the baseline expected number of deaths by the age-specific relative risks for men and women (Box 2).

Box 2 Applying the relative risk estimate to the Australian population

Data source: Number of deaths for weeks 11 to 21, 2014-2018 by age and sex were provided on request from the Australian Bureau of Statistics.

This resulted in an estimated additional 16 313 deaths in Australia: 9295 deaths in men and 7018 deaths in women. In contrast, by 24 May (the end of week 21), there had been 103 deaths in Australia attributed to COVID-19.

Even taking into consideration new estimates of total excess deaths of 1340 by the end of May in Australia from ABS provisional mortality statistics (a figure which includes deaths due to other non-COVID-19 causes as well as undiagnosed COVID-19 deaths), there is still an enormous difference between our estimate and actual reported deaths.

Is it reasonable to apply the relative risk from England and Wales to Australia?

In epidemiology it is often considered that relative measures of effect like the relative risk are more likely transferable between populations than absolute measures that vary considerably based on baseline risk of disease. However, if there is any evidence of effect modification of the relative risk by characteristics that differ between populations, then applying the relative risk to a different population may not be appropriate.

In terms of COVID-19, there is good evidence that the impact of COVID-19 on mortality differs by both age and sex. We have accounted for these effects by using age-specific relative risks for men and women for our calculations. The relative risk of dying from COVID-19 also differs across socio-economic status, ethnicity, comorbidities and health system resourcing. It is unclear whether these differences would remain once baseline mortality risk is accounted for, as we have done in our calculations. Nevertheless, we will briefly consider differences in the above factors between Australia and the UK.

Comparison of income inequality using the Gini coefficient (where 1 represents complete inequality and 0 complete equality) shows fairly equivalent income inequality in the UK, with a score of 0.36 compared with 0.33 in Australia. In terms of ethnicity, Australia has a more ethnically diverse population than the UK. In terms of comorbidities, data from the Global Burden of Disease study show that the prevalence of the majority of comorbidities known to be associated with increased COVID-19 mortality are lower in the Australian population and this better health status is supported by the higher life expectancy in Australia compared to England and Wales. In terms of health system resourcing, Australia has a better resourced health system with a greater number of hospital beds, doctors and nurses per 1000 inhabitants (Box 3).

Box 3: Comparison of comorbidities and life expectancy between Australia, the UK and Italy

* Data from Institute for Health Metrics and Evaluation (IHME). GBD Compare Data Visualization. Seattle, WA: IHME, University of Washington, 2018 (viewed July 2020).
** Data from Organisation for Economic Co-operation and Development (OECD). Life expectancy at birth. 2019 (viewed July 2020).
***Data from Organisation for Economic Co-operation and Development (OECD). OECD Data, and here (viewed July 2020).

This difference in comorbidities and health system resourcing could result in overestimation of the number of deaths in Australia by applying the England and Wales relative risk estimates. However, although the relative risk of dying appears to be higher for people with comorbidities compared to those without, the relative risk for populations with higher baseline mortality risk might not differ from those with lower baseline mortality risk (ie, effects of comorbidities are already accounted for by baseline mortality risk). Differences in comorbidities and health resourcing between the nations may therefore already be accounted for by our use of baseline mortality risk in our calculations. Evidence supporting this contention is found from an analysis of the Italian data on mortality risk during the pandemic for people aged 65 years and older (Box 1, right-hand column). These estimates are remarkably similar to those from England and Wales, despite the lower prevalence of comorbidities in Italy compared with the UK (Box 3) and Italy’s higher life expectancy (Box 3), which is also higher than that of Australia. Italy also has more equivalent health system resourcing in terms of hospital beds and the number of doctors per 1000 inhabitants (Box 3).

Conclusion

In contrast to other modelling studies, we have used empirical evidence to more directly estimate the number of excess deaths that may have occurred if Australia experienced a COVID-19 outbreak of a similar extent to that seen in England and Wales in March and April.

Our estimate of more than 16 000 deaths underlies the importance of Australia’s response to the pandemic, which has used a combination of extensive testing and contact tracing, mandatory quarantine of people returning from overseas and an early partial shutdown.

One notable component of the overall Australian response has been that led by First Nations people, resulting in Indigenous people being significantly under-represented in the country’s COVID-19 cases despite a higher risk status.

The subsequent breakdown in hotel quarantine leading to community transmission and a substantial number of deaths in Melbourne highlight that there is no room for complacency in Australia’s response, and that continued efforts are needed to avoid the large excess mortality observed in many other nations.

Dr Fiona Stanaway is a clinical epidemiologist with expertise in ethnic inequalities and ageing. She researches the impact of socio-economic status and social relationships on the health of older migrants in Australia.

Les Irwig is an epidemiologist with an interest in assessing risk, weighing benefits and harms of health interventions, and communicating these to health professionals and the public.

Armando Teixeira-Pinto is a Professor of Biostatistics at the University of Sydney and a Researcher at the Centre for Kidney Research, Westmead Children’s Hospital.

Associate Professor Katy Bell is a clinical epidemiologist with expertise in health technology assessment. She researches how medical tests for screening, diagnosis and monitoring may be used to benefit health without causing harm.

The statements or opinions expressed in this article reflect the views of the authors and do not represent the official policy of the AMA, the MJA or InSight+ unless so stated.