False Hope and Biased Data: Why COVID-19 transmission is worse than we think
On April 2, it was announced that 1% of the Australian population had been tested for COVID-19. This follows from recently announced trends in Australia which imply a reduction in the daily growth rate of new cases of COVID-19 from 25% down to 9%, leading to praise for the Australian people and their embrace of social distancing measures. Claims of ‘flattening the curve’ are abound. Unfortunately, this is likely a case of false hope for two key reasons:
- It is far too early for us to see the effects of the more extreme social distancing measures; and
- Our method for testing cases of COVID-19 is producing biased information which grows more inaccurate by the day.
Because of these two factors, we really don’t know how Australia is faring in the war against this global pandemic, and while our government is clearly aware of this, we risk the Australian population underestimating the devastating impact this virus is yet to have, and this can only contribute to making the problem worse.
This first point is fairly straightforward. We know that the incubation period of COVID-19 can be as long as two weeks, with some cases being even longer. Considering that social distancing only began to be properly implemented two weeks ago, with some of the more drastic measures coming into play on March 29 with the two-person limit, we cannot currently measure the full impacts of these policy changes. Fortunately for us, we should expect these ramped-up measures to be reducing the growth of infections. The problem is that the numbers we have are indicative of a bigger problem: biased data.
To understand how we have come to produce a measurement bias in our data of COVID-19 cases, we have to begin by looking at the criteria which must be met in order to be tested. While this has varied since the outbreak, the most recent guidelines as of April 2 are the least restrictive while demonstrating the same problem. To summarise, in order to be tested for COVID-19, you must meet the following criteria:
- You have returned from overseas in the past 14 days and have developed a respiratory illness with or without fever;
- You have been in close contact with a confirmed COVID-19 case in the past 14 days and have developed a respiratory illness with or without fever;
- You have severe community-acquired pneumonia and there is no clear cause; or
- You are a healthcare worker who works directly with patients and you have a respiratory illness and a fever.
There are two key observations we can make here. The first and most significant is that we are only testing those who display moderate-to-severe symptoms. This is a practical approach to testing for the virus given the limitations around testing kits to date, however, given that as many as 60% of infected individuals display no signs of infection, and further that most cases only display mild symptoms, we are clearly underestimating the number of cases Australia-wide. We can refer to this as a severity bias where most cases are only recorded if they are serious.
The second key observation is that the criteria are biased by international travel. Criteria one requires recent overseas travel and a display of moderate-to-severe symptoms, meaning it will not cover all cases of infection from overseas. However, criteria two can only be as good at identifying cases as criteria one given that it depends on contact with previously confirmed cases. If the infected can only be tested when they have come in contact with a confirmed case derived from international travel, and all cases of infection from international travel are not accounted for, there will be a subgroup of people who should be tested on the basis of criteria two who cannot be. While criteria three and four also provide means of testing for the virus, criteria three suffers from the aforementioned severity bias, and criteria four applies to the health sector — people who are likely aware of the risks of the virus and will more effectively self-isolate and quarantine than the average Australian.
What this means is that there is a group of infected individuals who have not been meeting the testing criteria, and this group of individuals will be growing exponentially with the virus, curbed only slightly through criteria three and four. Consequently, while we may be seeing a reduction in the growth of infections by the current testing criteria, the true infection growth rate is most likely higher than it has ever been in Australia. What we are currently seeing is a ‘flattening curve’ of infections meeting the criteria, while the curve of infections not meeting the criteria continuing to grow exponentially, meaning the curve of infections in Australia continues to grow exponentially, though perhaps at a slower rate.
Even if we had tested 20% of the Australian population, the number of COVID-19 tests we are doing would not give us a good indication of total cases since we are only testing one subgroup of possible infections. This is something which was notably referenced by NSW premier Gladys Berejiklian late last week, and it has been key in the decision of State and Federal Governments to implement heightened social distancing restrictions. It ultimately highlights the measurement bias of COVID-19 cases in Australia.
What does this mean for us?
My intentions of this article were not to instil fear, but rather to ensure that we are taking COVID-19 seriously. If it makes you feel better, this measurement bias actually means that COVID-19 is less deadly than it currently appears; given that we are underestimating the number of cases, our current estimations of the death-rate will be over-exaggerated. Fortunately, the heightened social distancing measures implemented nation-wide begin to nullify the problem with this measurement bias, though only if we all work to individually act upon these social distancing measures. By assuming that we could all have the virus, we can prevent Australia from becoming the next Italy or the United States by reducing our social interactions, and thus transmissions.
False hope leads to a false sense of security. At the end of the day, as with all things, how this pandemic plays out in Australia depends upon our ability to take personal responsibility for ourselves.
Note: As this article implies, all data on COVID-19 should be taken with a grain of salt given the problematic nature of data collection internationally. The author has used data to demonstrate the magnitudes of his arguments, and they should be taken as estimations of such.
