Mortality trends

Table 1 shows that the annual age-standardised mortality rate (per 100 000 people) fell from 528.4 in 2015–2019 to 497.0 in 2020–2021. This represents a 5.9% fall in the mortality rate. This figure is similar for males and females (for an explanation of why the percentage change for persons is slightly smaller than the figure for either males or females, see online supplemental appendix 2). Looking at causes of death, we find declines in mortality across the board. In absolute terms, the mortality drop is largest for respiratory diseases (down 9.4 deaths per 100 000), cancer (down 7.5 deaths per 100 000) and heart disease (down 8.4 deaths per 100 000). In proportional terms, the mortality drop is largest for respiratory diseases (down 20.5%) and the respiratory disease subcategory of influenza and pneumonia (down 41.8%).

Strikingly, while Australia experienced an average of around 600 influenza deaths each year in 2015–2019, influenza claimed fewer than 50 lives in 2020. In 2021, fewer than five Australian deaths were recorded from influenza (see online supplemental appendix 2 for details).

In figure 1, we analyse the seasonal nature of these effects, plotting weekly deaths across the year. During 2015–2019, the weekly death rate shows a marked seasonal effect, rising from around 9.5 deaths per 100 000 in summer to around 11.5 deaths per 100 000 in winter. By contrast, the years 2020–2021 saw less seasonality in Australian deaths, with the weekly mortality rate averaging around 9 deaths per 100 000 in summer and 10 deaths per 100 000 in winter.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Australia’s weekly death rate per 100 000 (age-standardised).

Figure 2 shows the six mobility trends indicators, with a horizontal line depicting the baseline. As these data show, the COVID-19 pandemic significantly reduced visits to retail and recreation, parks, transit stations and workplaces, and substantially increased the amount of time spent in residences. The decreased time in public places and workplaces coincided with the two lockdown periods: April–May 2020 (when all Australians were under lockdown) and August–September 2021 (when two-thirds of Australians were under lockdown). Visits to grocery stores and pharmacies show a slightly different pattern: spiking as Australians stocked up at the start of lockdowns, and steadily rising over the 2 years (perhaps reflecting increased demand for medicines).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Trends in mobility, lockdowns and mortality.

Alongside these mobility indicators, we show the share of the Australian population under lockdown. We also show how age-adjusted mortality compared with previous years. To do this, we take the age-adjusted death rate for each week in 2020 and 2021, and estimate the ratio of that mortality rate to the mortality rate for the corresponding week in the period 2015 –2019. Thus, a ratio of 1.1 indicates that the death rate was 10% higher than in prior years, while a ratio of 0.9 indicates that the death rate was 10% lower than the historical average. This chart shows that the death rate fell sharply as social distancing increased. By mid-2020, the death rate was around 10% below its historical average. The death rate then returned close to its historical average in late-2020, before dropping (that is, improving) in late-2021, a period in which Australia’s largest two states were under lockdown.

How does mobility track mortality? In figure 3, we present scatterplots of the mobility levels (relative to baseline) against the death rate (relative to prior years). Each dot denotes the figures for a single week. These data show that higher rates of activity in retail and recreation, grocery and pharmacy, transit stations, and workplaces were associated with higher mortality rates, while greater levels of activity in parks and more time spent in residences were associated with lower levels of mortality.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Mobility trends and mortality (each dot depicts a single week).

The simple bivariate regressions are significant at the 1% level for retail and recreation, grocery and pharmacy, transit stations, workplaces and residences. The association between time spent in parks and mortality is statistically insignificant. Finally, we show the relationship between mortality and the share of the population under official lockdown. Consistent with the mobility data, this relationship is negative and statistically significant at the 3% significance level.

Explaining the mortality decline

Australia was among a small number of countries that saw a decline in overall age-standardised mortality during 2020–2021. This decline in mortality was due to declines in both communicable and non-communicable causes of death. Using Google mobility data, we provide suggestive evidence that the drop in deaths tracked reductions in movement outside the home. In the short term, government lockdowns and social distancing appear to have significantly reduced overall mortality at least in the short term, which may help shape future public policy.

While the reduction in mortality from infectious diseases can explained by social distancing measures, the reduction in deaths from non-communicable conditions such as cancer and heart disease are more surprising. In the case of cardiovascular disease (which contributes to around one third of the reduction in deaths), possible explanations could be people spending greater time indoors over winter months (which are traditionally known to have high excess deaths),9 faster access to emergency services,10 lower pollution11 and a reduction in cardiovascular events following influenza.12

Explanations for the reductions in cancer deaths are more difficult to identify and might be due to potential limitations in the coding of deaths for people with multimorbidities.13 A recent analysis indicates that unlike other countries COVID-19 did not impact on provision of many healthcare services. For example, a recent analysis has indicated that COVID-19 pandemic did not disrupt cancer care in Australia.14 It is possible that reductions in influenza may also have played a role, given evidence that influenza mortality rates are higher among cancer patients than the general population.15

It will be important to continue track mortality in countries such as Australia to see if there are long-run mortality impacts of lockdowns, for example, from the disruption of breast cancer screening services16 or declining levels of physical exercise17 both of which may lead to an increase in mortality rates over the longer term.

Although our analysis of overall deaths includes coroner-reported deaths, our cause-specific mortality figures are based only on doctor-certified deaths, excluding coroner-reported deaths and so exclude deaths from suicides, accidents and assaults (see online supplemental appendix 2). A recent international analysis of suicides in the early stages of the pandemic (which included data from Australia) has shown that rates ‘remained largely unchanged or declined in the early months of the pandemic’.18 Future research should look separately at suicides and homicides when full data become available.

It should also be noted that excess deaths can be calculated using a variety of different methodologies which can produce some variation in results, although consistently 2020 is a year of below average mortality for many diseases.19 Further, the Australian Bureau of Statistics is not the only source of data on deaths. The World Mortality Dataset20 is a valuable resource used by numerous international organisations to compare excess deaths across countries that could be used in future analyses.

Another caveat is that we are considering only the impact on mortality and do not capture the impact on morbidity from poor mental health or from family, domestic and sexual violence.21 For example, a recent Australian study using a quasi-experimental design has shown that lockdowns were ‘associated with a modest negative change in overall population mental health’.22 Quality-adjusted life-years could be a useful broader metric, capturing both the negative and positive health impacts of home working and virtual interactions in a single measure. This could provide the basis of an evaluative framework of COVID-19 pandemic response policies.23 Additional impacts of lockdowns such as disruption to economic activity and schooling also need to be considered.

Beyond the pandemic, it is also worth considering whether a move to virtual working and other online activities may be a means to reduce mortality from a range of diseases such as respiratory infections during the winter months. One must also weigh against the potential benefits in mortality, the impact on other aspects of health such as general well-being from social interaction. While large-scale randomised studies24 have shown that home working can increase productivity, the potential for it as a health intervention in randomised studies has received less attention.25 Given the mortality benefits observed here at a population level, it would worth exploring if the health and social impacts of home working and virtual activities could be evaluated using large-scale randomised trials.