Speak Data, page 7
Such mistakes have been really harmful, especially in the conversation around vaccines. You had all of these maps floating around in which Africa was green or some light-shaded color, and the dark red was reserved for the rich countries of the world. For someone who looks at these maps, it’s understandable that they might come away with the idea that it’s not a problem for poor countries to not have access to vaccines. This situation has been common, and really tragic.
Data is power, and the service you provided during the pandemic was a public help. One could say that this is what the government should have been doing, at least in the United States. At the same time, no one asked you to do this. It’s just something that you just decided to do. In a perfect world, do you think Our World in Data should even exist?
No, it shouldn’t be us who do this work. During the pandemic, it was always my expectation that, at some point, we wouldn’t be responsible for this work anymore. When we began tracking COVID-19 data, we started a bit reluctantly, thinking that surely the relevant health organizations were going to take over. And then we found ourselves in this position where suddenly everyone was relying on us. We fully expected that some institution with the mandate to do this work would come and pick it up from us and then we could end it. But that didn’t happen. So, no, I really don’t think it should be us.
You were talking about how missing data and uncertainty is not rendered in datasets. We’ve been struggling with this a lot. There are some countries that still test and have free PCRs, or there’s different mechanisms to report positive cases, or different restrictions, or COVID-19 is treated differently. Yet, all that being true, we still plot data on a map as if that context didn’t exist, and as if missing data is not part of the picture. We think that missing data is as important as a data point. And rendering the uncertainty of a chart is as important as rendering what’s certain. It’s hard for the public, though, to trust data if we admit to how much we don’t know. It’s an ever-present tension, right?
For sure. It’s the key tension to navigate. And it’s difficult because the audiences are so different. Sometimes you have people who believe everything that is printed in a chart and take any chart as a perspective from God. That it’s just the pure truth. Then you have people who don’t believe any data, ever. The challenge is to find that middle ground where, even in the best cases, the data is imperfect and wrong in some sense but still more helpful than anything else we’ve got to understand the world.
What’s your personal definition of data?
Wow. I’ve never thought about the definition of data. I think it’s something like—structured information about something that matters?
That’s a good one.
I’ll probably regret this after we hang up.
It should be said that the data visualizations you produce at Our World in Data are marvels of simplicity and intuitiveness. You’re also always very clear about where you’re getting your source material and how you’re analyzing it. I wish other media outlets would take a similar approach.
A big guiding principle of ours is to keep it simple. If a line chart does it, then it’s a line chart. If a bar chart does it, then it’s a bar chart.
Social media has helped, I think, improve the design. You can quickly share your work online, for instance, and see what people understand or don’t understand, what kinds of questions they come back with, what kinds of wrong conclusions they’re taking from a chart. It’s fast to iterate. That was a big change.
Most importantly, and I try to get this across to everyone on the team: We’re standing on the shoulders of others. We’re a platform for others to present the data that’s out there. We shouldn’t understand our work as beginning with us and ending with us. We’re always trying to see our work as a place to build on and from. So we make it easy for people to pull our data into their tools, or to explore the data on our site and visualize it in the ways they want to.
NOTES
1 “Max Roser,” University of Oxford, https://www.ox.ac.uk/research/research-in-conversation/our-place-world/max-roser.
3
Vital Signs: Data and Health
Vital Signs
As data obsessives, we would be the first to admit that we have an unusually intimate relationship with quantification. For us, data is both a professional pursuit and a personal passion—the frame, filter, and language through which we understand the world. It’s a creative material with which we express ourselves. And when life is confusing, or scary, we tend to turn to data for answers. It never let us down.
That is, until the COVID-19 pandemic. Until I, Giorgia, got long COVID.
In total, long COVID robbed me of more than three years of happy, healthy life. While today my health is much improved, my diagnosis continues to affect me in profound ways. And it has fundamentally changed my relationship with data. When I think back to a few years ago, my mind fills with memories of fear and agony. Sensations of dizziness and nausea were my almost daily reality. Pain constantly pulsed through my body, and my limbs felt simultaneously as heavy as concrete and weak as jelly. It was as if a machine were squeezing my skull, and extreme exhaustion often overtook me.
In no way were these symptoms just a lingering cough, or a few weeks of fatigue after an acute infection. They were so serious that at times I was completely bed-bound, an unwilling prisoner in my own unruly body.
Long COVID is a chronic illness that, as of this writing, has affected an estimated seventeen million people in the United States.1 Despite its deceptive name, this mysterious condition is not just one disease, but multiple afflictions that attack various parts of the body, including the respiratory, circulatory, and nervous systems. This makes the illness particularly difficult to identify and treat. For reasons not yet understood, long COVID affects women and Latinx people disproportionately, as well as those with underlying health conditions.
I, Giorgia, first got COVID-19 in March 2020, just as New York was going into lock-down. My case was mild, and I was not hospitalized. Like many who got sick in those early days, I experienced what felt like a bad flu. But a few weeks after I seemed to recover, strange symptoms emerged: extreme fatigue, frequent low-grade fevers, general temperature dysregulation, chills, heart palpitations, brain fog, burning sensations all over my body, and more. And my symptoms persisted.
In December 2021, I got COVID-19 again. Excruciating nerve pain began to radiate up and down my side. I visited more doctors and took more tests without getting any answers. I was awash in data, but meaning was elusive. I tried more than a dozen medications, injections, and physical therapies, but this new pain never went away. My doctors were confused when I wanted them to be alarmed. After more inconclusive results, they told me that I was probably just stressed and should take a break from work. Or I should try to push through and exercise. Or maybe I should start antianxiety meds.
Then, after a third infection, my symptoms became entirely debilitating. Unrelenting chest tightness and tachycardia, dizziness while being upright, frequent nausea and headaches, systemic reactions to most foods, tinnitus, severe insomnia, a persistent feeling of being poisoned, blurry and double vision, and exhaustion relegated me to bed with the lights off for days at a time.
I’m not a medical expert, but I had to become one to try to figure out what was happening to me. And as I’ve done so many times before, I turned to data: the tool I’ve always had to help me cope with life when I am afraid, confused, and looking for answers. I started logging all of my symptoms. I tracked everything in an enormous spreadsheet: my symptoms’ intensity, whether they came on suddenly or gradually, when new symptoms appeared, the medications and supplements I was taking, the treatments I was trying, what I did that day, if I felt stressed, what I ate and drank, and scores of biometrics from my newly purchased smartwatch. Days on my data canvas became thick with color: red for bad, green for good. Most months, there was far more red than green.
No matter how much data I collected or how many correlations I tried to draw, answers eluded me. At the same time, the act of tracking my data brought its own perverse pleasure. I assiduously kept the spreadsheet updated, noting with as much precision as possible every twinge of pain, every haze of brain fog. My spreadsheet was the only thing I could control in a life I no longer recognized. I thought that if I collected enough data, I would eventually figure out what was wrong with me. In a sea of uncertainty, it was the life raft I clung to in hopes of finally returning to my predictable, knowable life.
But still, I was sinking. And dry land was nowhere in sight.
In a moment of extreme desperation, I started to question whether the data I was collecting was still serving me. Was something that had given me a semblance of control in these years of uncertainty actually becoming my enemy? In that moment of extreme desperation, I decided to shift focus. I started a new data collection, this time populated with qualitative and quantitative information focused not on the bad, but on the good. Instead of tracking my pain and symptoms, I started to track my progress: when I was able to climb stairs, take a walk around the block, or sleep through the night. As I got better, my logs became longer and more specific. I noted when I was able to go out to dinner with my boyfriend without feeling adverse effects the morning after, or hang out with my friends and feel at home in my body. Finally, I also retired my smartwatch—a device that, truthfully, was giving me more alarming than encouraging news. I stopped logging symptoms, and therefore no longer gave them constant attention.
At first, this new approach felt scary. Monitoring my body every day had given me a semblance of control. Now I felt naked and unprotected by the statistical biometrics that had for so long described and defined me as a living, breathing human being. Yet through it all, I’ve realized that changing how I look at things also changes how things look. I’m no longer paying constant attention to my symptoms. It doesn’t mean that the symptoms have gone away; far from it. I just refuse to give them the same priority in the mental picture I paint of myself.
In 2024, Giorgia’s story of long COVID was published in The New York Times as a print and digital guest op-ed titled “1,374 Days: My Life with Long COVID”
The spreadsheet Giorgia used to track her symptoms
A visual representation of Giorgia’s long COVID symptom tracking
To be clear: We’re not suggesting this approach can “cure” long COVID, or even worse, that these illnesses are just “in people’s heads.” Positive thinking cannot cure chronic illness. Today Giorgia is doing much better thanks to a combination of medical treatments and alternative therapies undertaken in close consultation with her doctors. But this new dataset, the one she decided to give attention to every day, reshaped the way she saw her personal health journey. It’s also taught us both something fundamental about life and data. The world is made of data—not just the data we produce with our phones or our credit cards, but the data we decide to direct our attention to at any given moment. It’s important to acknowledge that we do have a choice regarding which data to collect, which data to store, which data to present, and how.
As Giorgia’s journey takes her further into the complex landscape of health care, it’s clear to both of us that data can be both a tool of clarity and a source of confusion. How we choose to measure our health—and which metrics we prioritize—can significantly influence not only the treatment we receive, but also our perception of well-being. The decisions we make about data collection and interpretation hold profound consequences for patients, practitioners, and policymakers alike.
NOTES
1 Alice Burns, “As Recommendations for Isolation End, How Common Is Long COVID?,” KFF, https://www.kff.org/coronavirus-covid-19/issue-brief/as-recommendations-for-isolation-end-how-common-is-long-covid/.
David Putrino
David Putrino is a decorated academic with a PhD in neuroscience who marshals interdisciplinary research to tackle some of the big medical mysteries of our time, from depression to Parkinson’s disease. His mission is simple: “Use technology to save lives.” That usually means using data too, and lots of it. More recently, David and his lab have also turned to investigating mysterious but debilitating chronic diseases, including long COVID. In this conversation, David explains how he came to this work; how he thinks about patient data versus patient experience; and getting comfortable with the uncomfortable.
Your background is unique. You’re not a typical physician or researcher. How did you get into this work?
My title at Mount Sinai is director of rehabilitation innovation. My clinical training is as a physiotherapist. Initially, I was interested in helping people recover from neurological injuries. I worked a lot with stroke and spinal cord injuries. I also—interestingly, weirdly—got pulled into sports performance. I was training the adult nervous system to change, which is actually very hard. Whether you’re dealing with a stroke victim or an elite performer, training that nervous-system change requires similar strategies and similar uses of technology. I did my PhD in experimental neuroscience, trying to understand how the brain controls movement.
I’m interested in disruptive innovation. When I started at Mount Sinai, my pitch to my new employers was that I was going to partner with industry, partner with people who have ideas that I think deserve the main stage right now. I’m going to run rapid, pragmatic clinical trials for them, and then we’re going to adopt the technologies that we prove work. To take an idea from bench to bedside in the medical world takes, in the United States, an average of seventeen years. And I think that number is just ridiculous. It should be much smaller. So I’ve created a division within our department that does nothing other than try to get new treatments to patients as quickly as possible.
We’ve grown from one small center to five centers across three different hospitals in the Mount Sinai Health System. We’ve brought in more than $30 million in funding. And we continue to grow. If we see a good idea, we get that good idea into the hands of patients as quickly as we can. We don’t want patients to be told that there’s nothing more that we can do. Even if we have a patient who is not responding to an array of therapies, we’re never out of ideas. We’ve always got two or three more things that we can try.
On the clinical side, we’ve always got an answer for them: This is what we’ll try next, or this is the next thing we’re going to test for. On the research side, we collect data in a large registry. We rarely run highly controlled RCTs [random clinical trials] because for the most part, even though we’re told that RCTs are the gold standard for doing things, they rarely provide actionable information. What they usually do is say, hey, when you control for every element that makes something useful, yes, you might get a little bit of signal here and things might improve. But what I want to know is: For the three hundred stroke patients who come through my clinic in a year, if I give this to everyone, what percentage are going to respond? I believe a lot of mono-therapeutic approaches are designed by people who want to sell you one drug to solve your one problem. Unfortunately, biology is more complicated than that.
How did you get interested in long COVID?
Our interest in long COVID arose out of necessity. Our lab was well poised in March 2020 when things hit the fan here in New York. We had a lot of technology that we knew could help patients, so we immediately put it to use. For instance, we had an app called Precision Recovery. We were using it specifically for stroke survivors: After they were discharged from the hospital, we would check their blood pressure, make sure they were doing okay. So when COVID hit, we quickly made the decision to reprogram our app to start asking questions about respiratory symptoms and risk of respiratory failure. We started deploying it to thousands of people in New York who were symptomatic with COVID and concerned that they might end up dying of their symptoms. We onboarded our first patient on March 15, 2020. Soon we had about seven thousand people on the platform.
Those numbers just kept growing. By about late April, we started to notice that 10 to 15 percent of the acute people we were monitoring weren’t getting better. They kept reporting symptoms, but the symptoms started to change. Now they were talking about fatigue. Cognitive impairment. Many other symptoms. All that stuff now seems commonplace, but these people were reporting it before we had a name for it. It wasn’t an identifiable disease, it was just a collection of symptoms—but a collection of symptoms that were highly consistent across hundreds of individuals. And they weren’t going away. So as we started seeing them, we started listening. As we listened to symptom presentation, we built this structure. Fatigue: What questionnaire can we use to measure fatigue? What can we use to measure post-exertional malaise, or PEM? So that was initially how we started measuring long COVID and its severity. We listened.
I get suspicious whenever I hear someone say that they’ve “cured” long COVID because we know that long COVID is not just one thing. It’s many things. We know that in all complex chronic illness is this big mess of things going wrong with the body all at once. So we’ve started this very multidisciplinary approach. We collect a lot of data because we need to be systematic in coming up with the highest-priority treatments. And then as treatments either succeed or fail, we go down the line of what needs to change.
There are always measurable aspects of human health, but also more subjective and personal perceptions. How do you balance the data from tests you perform on a patient with the more anecdotal data that a patient shares with you?
