In my previous post I examined terms and concepts associated with the metrics used in coronavirus dashboards. The pandemic reminds us that dashboards and metrics enable data-informed decisions and data-driven actions – if we understand what those metrics represent, and on what data they were based. In this post, I'll dive deeper into the terms and concepts surrounding coronavirus testing – a primary source of COVID-19 data.
Combating the coronavirus pandemic requires performing a widespread number of daily viral tests. This encompasses testing across three dimensions of distributed data collection:
- Geographic — Testing across locations, such as postal codes, cities and regions. This is too often done reactively to highlight hotspots where the uncontrolled spread of coronavirus threatens public health in a community. Geographically distributed testing can help identify problematic locations before they become a hotspot by tracking positivity rate changes.
- Demographic — Testing across attributes, such as age, race and preexisting health conditions. After more than six months of global analysis, we know coronavirus is more deadly to certain people (e.g., older than 65 years), but we shouldn’t just assume other demographics are safer.
- Symptomatic — Testing across the symptomatic spectrum. I’ll go into more detail on this in the next section, but for now the most important point is: We have to test more than just the sickest people. Reactively testing hospitalized patients with severe symptoms discovers an outbreak after it’s happened. And that is likely after those patients have already set off chains of transmission infecting more people.
Widespread testing faces many challenges. These include testing supply chain problems, backlogs at laboratories (which delays test results) and testing capacity that doesn't scale with larger outbreaks.
Research and development underway – helpful approaches
There are some efforts underway that may help with the issues raised by widespread testing.
- Pool testing — Combining viral-test samples from several people and conducting one laboratory test on the combined pool. This allows laboratories to test more samples with fewer testing materials. If a pool test result is negative, then all the samples can be presumed negative with a single test. This could be most useful in evaluating related groups, such as employees returning to a workplace or students returning to a campus. However, if a pool test result is positive, then each of the samples in the pool needs to be tested individually to determine which ones are positive. Therefore, pool testing works best in locations or situations where we expect a low number of positive test results.
- Surveillance testing — Collecting or estimating viral-testing information at a population level, rather than an individual level. Various techniques exist (e.g., random testing), but one that’s being deployed in some areas is testing sewage and waste water for the prevalence of coronavirus. This doesn’t tell you who has it, but might provide an early warning system. Again, this works best for locations such as workplaces and campuses.
- Rapid (lower quality) testing — Several companies and universities are working on developing alternative viral tests for coronavirus that will be fast, cheap, easy and in some cases will not require a laboratory. Although the quality of these tests could be much lower (i.e., far more false negatives), it could allow individuals to test themselves at home. And perhaps most important of all – it could enable recurring testing, which would overcome some of the challenges I'll describe in the next section.
What I call the symptomatic spectrum is essential to understanding coronavirus community spread – and why it can be so confusing. Here's the spectrum in descending order of severity:
- Symptomatic (severe) — As we slowly learn more about COVID-19, studies (many still pending peer review) suggest clusters of symptoms for measuring the severity of the disease. Required hospitalization is my oversimplification for the severe end of the spectrum.
- Paucisymptomatic (mild) — Some people develop symptoms too mild to seek medical attention. The medical term for this is paucisymptomatic, meaning presenting few symptoms. Such individuals can also be denied testing unless (or until) they get sicker.
- Pre-symptomatic — Since it can take up to 2 weeks after exposure to develop symptoms (hence 14-day quarantines), some people viral-test positive before exhibiting symptoms. Additionally, some infected people viral-test negative due to being tested too soon after exposure.
- Asymptomatic — This means never exhibiting symptoms. Most people in this category have not been infected. This represents the majority of any given population. However, some people who viral-test positive still never develop symptoms. And not all of these cases can be dismissed as false positives.
There’s been a lot of debate over the extent of the asymptomatic spread of coronavirus. Some medical experts believe that many of the so-called asymptomatic spreaders are actually either paucisymptomatic or pre-symptomatic. Furthermore, time (lag) again plays a crucial role.
Some people simply viral-test positive before symptoms appear and get classified as asymptomatic. And there is no follow-up to see if they were really pre-symptomatic or paucisymptomatic. This leads to inaccuracies in the estimates of how many infected people are symptomatic and, most important, still contagious.
Additionally, contact tracing is a time-sensitive job with a limited window of opportunity to catch people with an active infection. The success of the strategy proven effective in combating coronavirus relies on being able to quickly identify where people fall on the symptomatic spectrum. That strategy involves widespread testing and contact tracing while quarantining the exposed and isolating the infected.
Masking the problem
Understanding the symptomatic spectrum might also help with clearing up the confusion over wearing masks. In the early days of COVID-19, many medical experts were advocating against wearing a mask to combat the novel coronavirus primarily for two reasons:
- PPE shortages — If N95 and surgical masks were hoarded (like toiler paper was), then it could cause personal protective equipment (PPE) shortages for the frontline healthcare workers who are in direct contact with the sickest people.
- Symptomatic transmission — COVID-19, the disease caused by the novel coronavirus, is also known as SARS-CoV-2. SARS-CoV-1, which first appeared in 2003, has since caused occasional outbreaks around the world. While SARS-CoV-1 is a far deadlier disease, it's less easily transmissible – requiring sustained personal contact with a symptomatic person. SARS-CoV-1 was never declared a pandemic – even though it's deadly, it's very difficult to spread. SARS-CoV-2 was declared a pandemic after it proved to be less deadly but far more transmissible.
The first reason above is why cloth masks became and remain the primary recommendation. The second reason experts originally recommended not wearing masks was negated after we learned that the disease is transmissible by seemingly healthy people. Masks are highly recommended to help break chains of transmission originating with paucisymptomatic, pre-symptomatic and asymptomatic people.
Passing the test
Coronavirus is proving a challenging test for our planet. However, it’s a test we can pass. We’ve heard a lot about what we need to do. Most crucial is whether leaders at the forefront of the coronavirus response can remain committed to making data-informed decisions and taking data-driven actions while combating the pandemic.View the COVID-19 dashboard from SAS