On a daily basis, the grim number of American deaths attributed to COVID-19 scroll across our screens and make headlines in the news. But how do we know those deaths were caused by COVID-19 and not by pre-existing conditions? What do those numbers really mean, and where do they come from? Do the data tell the full story? And, are they trust-worthy?
In his article, “How Many Have Died?” published in Issues in Science and Technology – from the National Academies of Sciences, Engineering, and Medicine and Arizona State University – CMC Philosophy Professor Andrew Schroeder explores how we measure deaths caused by COVID-19 in the United States.
Although assessing COVID-19 mortality might seem to be a purely empirical undertaking, Schroeder writes, “To know the COVID death toll, we also need to address a complicated set of conceptual issues: what does it mean to attribute a death to COVID, or to say a death was caused by COVID?”
We asked Schroeder – whose research and teaching cover a range of topics in ethics, political philosophy, bioethics, the philosophy of disability, and the philosophy of science – what compelled him to dig into the complexities of how we measure the pandemic’s death toll and why it’s essential, from both a philosophical and scientific perspective. “So many important aspects of science require reflecting not just on facts, but on values, and what we care about,” he explained.
What led you to examine the COVID-19 death toll?
My goal was to bring some clarity to public discussions that, I think, were a bit confused and also confusing.
When I saw people arguing about COVID death statistics, it was pretty straightforward to take my prior experience and use it to hopefully clarify how such statistics are typically put together, and to flag some respects in which the standard statistics can be misleading.
I’ve worked with epidemiologists for a number of years, serving as an advisor for the Global Burden of Disease Study and co-editing a book [Measuring the Global Burden of Disease: Philosophical Dimensions] on philosophical issues that arise in epidemiological work.
As part of that work, questions about causation frequently came up. Globally, for example, there are a significant number of HIV+TB deaths: an HIV-patient contracts tuberculosis and then dies. In such cases, it’s often true that neither condition would have been fatal on its own. Without TB, the patient could have managed her HIV; and without HIV, the patient could have fought off TB. The combination, though, may be fatal. In such cases, should epidemiologists attribute that death to HIV? TB? Both? Neither? Or should they try to somehow divide it up, maybe counting it as half an HIV death, and half a TB death? All of these solutions lead to problems, and can tend to give health policy makers a distorted picture of what’s going on.
That’s one example, but there are hundreds more.
The challenges involved in assessing the COVID death toll are very similar to the challenges involved in assessing many other epidemiological statistics, and those statistics are frequently misinterpreted in the same way. It’s just that, for obvious reasons, people are paying a great deal of attention to COVID statistics right now.
What kinds of interdisciplinary dialogues do you think are necessary and most effective to measuring the pandemic’s toll?
The important thing to realize is that this is not purely a scientific endeavor. The pandemic has taken a toll in so many ways – economic, social, educational, and political, for example. If we focus, though, just on the pandemic’s death toll, measuring it of course requires extensive scientific work.
But we also can’t measure the death toll without thinking philosophically, without reflecting on what matters or what we value. Take the patient with pre-existing COPD who contracts COVID and dies. There is no experiment we can conduct or data we can collect to determine whether we should say COPD or COVID was the cause of her death. Of course, there is an important sense in which they both caused her death. But if we have record-keeping systems (like death certificates, and national statistics) that require picking a single or primary cause of death, there is no scientific way to decide which to pick. As I explain in the article, which we pick should depend on how we’re going to use the resulting statistics, which of course is in turn dependent on what we care about.”
How does this fit into your overall work? What is the underlying thread?
Science isn’t the purely objective, sterile, value-free endeavor that we typically make it out to be. So many important aspects of science require reflecting not just on facts, but on values, or what matters. To take just one example, think about defining terms. One of the first steps in doing research on severe weather events, child abuse, employment, or anxiety disorders is to figure out what counts as a severe weather event, child abuse, employment, or anxiety disorder. Scientists need definitions of these terms that will make their resulting analyses useful. (Setting aside technicalities, a study which counted my daughter as “employed” because she earns a couple dollars a week doing chores around the house is not going to yield useful information for government policy-makers.) But usefulness, of course, is in the eyes of the beholder: you and I might differ, for example, about what information we find useful in choosing between medical treatment options, or in thinking about whether a COVID lockdown is justified. And those differences will, in many cases, ultimately boil down to differences in what we value.
Defining terms is an especially clear example, but there are many other aspects of science that, similarly, require scientists to reflect not just on what is true, but also what is useful or important or valuable. And these, of course, are extremely tricky and often contested matters. This is the point where some people start to lose faith in science. Upon realizing that the results scientists get – the number of deaths they attribute to COVID, for example – can depend not just on the facts, but on what they care about or value, some people conclude that science has been tainted. Why should we trust scientists when they tell us how many people have died of COVID, or that a COVID vaccine is safe, or that GMO foods are safe to eat? This, I think, is the wrong reaction, and a dangerous one. Science that involves value judgments can still be trustworthy. Indeed, science can be trustworthy precisely because it involves value judgments. The key is for scientists to manage those value judgments in a responsible way. That’s what my current research aims to do: to figure out how scientists can make the value judgments that their work requires in a responsible way.
— Anne Bergman