How much has Australia really flattened the curve of coronavirus? Until we keep better records, we don't know

Since March 29, we have congratulated ourselves for flattening the curve of the COVID-19 epidemic in Australia.

But if we ask more detailed questions, such as “which curves have flattened?”, “by how much?”, and “how was it done?”, the answers are not immediately obvious.

Data for addressing these questions are poor, and by identifying their limitations we can improve them and provide better tools for future decision making.

Read more: How to flatten the curve of coronavirus, a mathematician explains

Which curves have flattened?

The typical curve we are talking about is one that shows the number of new COVID-19 cases per day across the country. Flattening it means fewer new cases each day, so the number of current cases comes down, reducing demand on services.

We can debate exactly how flat we would like this curve, but everyone broadly agrees flattening is the way to go.

Australia’s curve is built from daily reports from each state and territory. New cases are logged to the date of the report, not the date of the test result or when the swab was taken, nor when symptoms appeared, and it certainly isn’t the date of infection.

It can be up to two weeks from infection to a case appearing in our curve, and this time lag is very important when considering if policies and community actions have desired effects.

Local versus imported cases

New Zealand data show the importance of separating out infections which happened overseas from “local” transmission. Its total curve shows a handful of cases per day in mid-March then rises steeply to a peak of over 70 cases per day in late March. By Easter Sunday, transmission had fallen 86% from the peak.

In the graph above, the medium blue line represents people who arrived recently in the country and most likely were infected overseas. The light blue line shows people who had not been out of the country, so they were infected in New Zealand. You can see the overseas infections peak six days before the high point for local transmission.

Border controls were tightened on March 19, which led to the drop in overseas transmissions. The later drop in the local transmission curve is due to two main things. Tighter quarantine of arrivals from overseas reduced the spread to the broader population, and the community lockdown following the declaration of a state of emergency on March 25 further reduced local transmission.

Different states treat data in different ways

It’s harder to see the big picture of what’s happening in Australia, because different states and territories report their figures in different ways. One way to approximate a national curve is to combine figures from New South Wales, Victoria, Tasmania and the ACT, which is about 70% of the country’s cases.

The Australian curves look similar to New Zealand’s but the drop in local cases is not as steep, falling to about half the peak by April 8.

How was this done? As in New Zealand, border control measures reduced the number of infected people coming into the country.

The “early signs of flattening the curve” at the end of March were largely due to the drop in cases from overseas, which had already fallen 25%. The peak and fall in local transmissions was due to quarantine of overseas arrivals, isolation of confirmed cases, social distancing and hygiene measures. These occurred later because of the two-week lag from action to effect.

Have Australia’s curves fallen further since the end of this chart, as New Zealand’s have done?

Local transmission data are vital

The Australian epidemic moved into a different phase at the start of April when local transmission overtook daily cases from overseas. Good data on local transmission are now more important than ever.

Their use could chart progress against targets and potentially compare infection rates across jurisdictions. This potential seems unlikely to be realised.

The ABC, the Guardian and covid19data.com.au have painstakingly collated state and territory information from a variety of formats. However, they cannot rectify inconsistencies in reporting across jurisdictions or the categories used to describe cases, which can prevent comparisons.

Inconsistencies cloud the picture

These inconsistencies and changes mean we cannot say with confidence how much Australia’s important local transmission curve has continued to move in the right direction since April 10.

For example, over the Easter weekend New South Wales reclassified between 71 and 76 previously reported cases from local transmission to a new combined grouping of overseas/interstate transmission. More changes to New South Wales categorisations were made over the following three days.

A best estimate of local transmission across the four jurisdictions is about 30 cases per day. Tasmania alone has reported over ten cases per day on average in recent days following the outbreak in the northwest of the state.

Although Australia’s border controls, rigorous testing and contact tracing, social distancing and hygiene have mitigated what would otherwise have become a national catastrophe, we can still improve our ability to track what’s going on.

Better records can help steer our future response

One way to resolve these inconsistencies would be to adopt a national unit record system like New Zealand’s. This would mean each record represents one case and includes information about date of report and possible source of infection. The record can be altered to reflect new information without affecting the total number or apparent timing of cases.

The two-week lag from infection to reporting presents challenges. Fencing off part of Tasmania where a lot of new infections have been reported is encircling a spot that was hot 10–14 days ago. More broadly, we are guiding our “scalable proportional response” using a chart that is necessarily out of date.

Our response is more like steering a container ship that takes two weeks to respond to the controls than tacking in a dinghy to meet the tide and the wind of the moment. So far, we have struggled to record where we have been and cannot pinpoint our present position or direction.