Why did it take so long for us to pay attention to pulse ox’s racial bias?
Volume 7, Issue 3 | January 18, 2023
In This Issue
Don’t forget to report negative test results, too
We still don’t know enough about COVID in China
Market value for in vitro Dx comes back to Earth
Machine learning in Dx - progress and potential
Being a vaccinator is hard. Don’t give up.
New and Noteworthy
Um . . . it’s not just the positive results that count, you know.
In the eight weeks since NIH launched MakeMyTestCount.org to provide a way for you, me, and the masses to anonymously report their COVID test results, more than 24,000 people have gone online to use the site. Over three quarters of reported results have been positive tests, and most of the reporters are women, so the data is definitely skewed.
China’s COVID Crisis Continues
One of the challenges in assessing the impact of COVID in China is the lack of credible information coming out of the country. This week we got two signs of just how challenging the situation is.
First, Chinese officials made an extraordinary update to the COVID death count: The official estimate went from 37 to almost 60,000. Officials later walked this number back, saying that 5,500 of those deaths were from COVID - a designation that China strictly limits to deaths from respiratory failure - and 54,000 were people who had died with COVID from cancer and cardiovascular diseases. (Commentary: COVID seriously screws with the lining of blood vessels, so we’re raising an eyebrow at those cardiovascular diseases.)
Second, we continue to learn more about test results from flights out of China. As previously reported, almost 25% of passengers on two flights from eastern China to Italy had COVID. Now we learn that three different variants were present on those flights: BA.5.2.48, BF.7.14, and BQ.1.1 Commentary: More data is needed, but the biggest concern here is that viral spread through the mostly non-vaccinated, infection-naive, older Chinese population could create a new variant more aggressive than the Omicron that we have come to hate.
Three years of pandemic market value - we’re back to where we started from
The aggregate market value of public global in vitro diagnostic companies increased dramatically in the first year of the pandemic. Smaller public companies especially benefited - those with diagnostic revenues under $1 billion were worth twice as much in early 2021 as they had been in late 2019. By the end of 2022 that appreciation - and more - had evaporated, with many newer companies losing 60-80% of their peak market value. These smaller companies are more pure-play diagnostics - and thus more representative of the diagnostic industry specifically - than the larger ones, most of which are diversified and perform closer to the S&P 500 index overall.
Back in 2016 we created an index of public diagnostic companies' market value to track the financial fortunes of the global industry. The latest update is shown above. (A brief note on methodology: A company is included if it has clinically relevant IVD revenues (i.e., excluding imaging). Our objective was to be as inclusive as possible, so for those companies with diverse market participation, we reduce their market value by their diagnostic revenues as a percent of their total revenue. Included companies, total revenue, diagnostic participation, and foreign exchange rates are re-adjusted each calendar year.)
Slow, steady progress applying machine learning in diagnostics
Artificial intelligence (AI) and machine learning (ML) are hot topics at the moment. The press has been energized by the power of ChatGPT to create text (research reports, term papers, etc.) that can fool experts and thus, in theory, pass the Turing test.
Progress in using these technologies in the diagnostics field has mostly been confined to medical-image interpretation, and a recent paper demonstrates a high point of this skill. It features a system named Sybil, which can predict lung-cancer risk up to six years in the future with ~90% accuracy after looking at a single low-dose chest CT. ML is reliable and indefatigable at performing routine tasks flawlessly in situations where human attention can flag, especially when the incidence of salient conditions is very rare. Image analysis falls neatly into this category. The real potential of ML is to integrate massive amounts of data from disparate sources to come up with predictions (e.g., diagnoses) that rely on complex interactions not readily apparent from any one data set. Eric Topol reviews current progress and potential in this week’s Ground Truths blog.
Food for Thought
Racial bias in pulse oximetry
Over these last three pandemic years, one small medical device became ever more commonplace - a pulse oximeter (often called a pulse ox), which is used to measure the saturation (percentage) of oxygen in the blood. Traditionally, pulse oxes are used on the fingertip, but now lots of wrist-based digital watches and fitness trackers have incorporated them into wristbands.
The pulse ox is a powerful tool to measure whether a person is getting enough oxygen - but it’s often inaccurate when used on people of color. The device works by sending light through the skin, and in individuals with darker skin, it can overestimate the amount of oxygen that’s present in the blood. That mistake can be deadly - if medical personnel don’t realize that someone’s not getting enough oxygen, critical treatment with supplemental oxygen will be delayed.
According to STAT News, the FDA has asked Hypoxia Labs to evaluate current device vendors and help to create a solution for this long-standing problem. Hypoxia was one of the first labs to acknowledge that the issue existed- in this paper 18 years ago. Progress is finally beginning to appear, as multiple device manufacturers come up with new solutions.
Commentary: It’s about time. Racial (and gender) disparities are not new, and especially when they can impact life-and-death decisions, it should not take a crisis (pandemic or other) to create change. This example is particularly frustrating, since COVID has disproportionately affected people of color. We’re glad to see manufacturers swiftly step up to address the discrepancy, which we hope will soon be a thing of the past.
Convincing people to get vaccinated can be hard. Hang in there.
Commentary: One of our readers’ comments on our vaccine article from a couple of weeks ago stuck with us. It came from an MD who is suffering from “vaccinator fatigue”: the exhaustion that comes from continuing to offer the COVID vaccine, only to be put off time and again.
First off, we totally hear you. How do vaccinators find the energy to do their work every day in a world where 28% of people in the US think they know someone who died from the COVID vaccines? As of December 2022, 9 people have been confirmed to have died as a result of COVID vaccination - from blood clots associated with the J&J vaccine. An investigation into strokes in people over age 65 as a result of the bivalent Pfizer vaccine is ongoing, but right now it looks like the vaccine doesn’t present any increased risk of stroke in that group. Have adverse events arisen from the vaccines? Yes. Might vaccines cause overactive immune system activity and create other disorders? Yes. In our mind, however, the risks of COVID itself, plus the 10%+ chance of long COVID, justifies the small vaccine risk.
Sometimes, in the face of this, there’s nothing else you can do but take a break and walk away for a while. And at this point, there is frankly nothing we can do for those who have made up their mind against vaccines.
But for those still on the fence, there are ways we can communicate that will help. Repetition helps. Being candid about adverse effects is necessary - without it, we lose the trust of all we treat. And we need to hang in there and continue to focus on the hope that vaccines provide. Even if they can’t completely prevent infection, they are still really good at keeping people from getting sick enough to go to the hospital or to die - especially older folks. That was the original goal for these vaccines, after all.
Quick Hits
In 2022, COVID remained the leading cause of death for law-enforcement officers for the third year in a row, just ahead of firearms-related deaths. Fortunately, the number of COVID fatalities in this group is down 83% from 2021.
The recent Ebola outbreak in Uganda is over. When the outbreak began in September 2022, no vaccine was available for the strain of virus involved. While candidate vaccines have since been developed, they have not yet been used - the outbreak was contained due to the efforts of Ugandan public health teams including contact tracing and testing.