Long COVID, Biomarkers, and Health Policy


Reports of so-called “long COVID” cases have been abundant in medical journals and the media. The emerging medical term for these conditions is “post-COVID syndrome.” Some reports suggest that around 14% of people who have contracted the virus experience one or more symptoms for at least three months following their acute infection. The most common symptoms are shortness of breath, “brain fog,” fatigue, and achiness. These lingering conditions are most common among adults ages 25-70, and, surprisingly, the highest reported rate is among those 25-34, where 18% of respondents with positive tests still complained of symptoms 12 or more weeks after their diagnosis. Meanwhile, women report suffering from ongoing issues more often than men, and Blacks report long COVID less often than whites.

The fact that adults in their prime suffer most from post-COVID syndrome—and not the elderly—raises several intriguing issues. First, like with any self-reported condition, cultural factors come into play: Women are more likely than men to seek medical care (and seem to be more in tune with their ailments in general). That, more than any biological mechanism linked to two X-chromosomes, likely explains the gender difference. Medical stoicism may also account for why Blacks report lingering symptoms less often than whites, despite more severe acute infections, on average. Mortality selection—the possibility that those with the worst potential long COVID die before they get a chance to experience the syndrome—likely does not play a large role in why those who would be seemingly the healthiest report the highest rates of long COVID. More than 90% of those reporting one or more symptoms were not sick enough to have been hospitalized.

Indeed, one of the more common experiences is “exercise intolerance”—becoming winded very easily from physical activity. This may mean that those who were in the best condition, and thus had a high subjective baseline of physical performance, were most likely to feel out of sorts for not performing at their usual level. Ditto for the second most common symptom: Brain fog. Those whose jobs were the most mentally demanding may have been acutely aware when they were a half-step slower in, say, recalling necessary information. The final cluster of symptoms, however, is less easily explained by prior, baseline conditions – those stemming from dysfunction of the autonomic nervous system, such as dizziness or a racing heart.

This is not to say that long COVID is all in people’s heads, but rather that it may, in fact, be underdiagnosed among populations that are too worried about other problems—like economic stressors, for instance, or other chronic conditions that come with old age. As the numbers of acute COVID sufferers declines in the United States and other countries, reports of people dealing with post-COVID syndrome may continue to rise for some time. Public policy needs to be able to deal with these folks in a fair and balanced way.

Important questions that policymakers need to consider include: Should individuals with long COVID be viewed as a vulnerable population like other socially recognized vulnerable populations? Should long COVID be protected as a pre-existing condition under the Affordable Care Act? Should other health policies be formulated specifically to meet the needs of these individuals, for example, should they be given priority access to care and resources such as disability insurance if their ongoing symptoms interfere with their ability to work? When resources are scarce, and when cultural or socioeconomic reporting biases are evident, how can we provide help to populations that need it while not creating social inequalities?

The biomarker revolution purports to offer a solution to this predicament. If we had a molecular indicator of post-COVID syndrome, we could cut through potential reporting biases and objectively assess whether someone who complained of long COVID actually was afflicted. Or would we?

In their broadest definition, biomarkers encompass any bodily measure from your weight on a scale to cotinine in one’s hair indicating nicotine exposure to the global methylation of one’s DNA as an indicator of biological “age.” Biomarkers such as cardiac troponin signaling a recent heart attack can be used to confirm diagnoses. Others, like a blood alcohol test or a urine-based test for metabolites of opioids are used to ferret out health behaviors that individuals have an incentive to misreport.

Other biomarkers are meant to be indicative of future risk, though this line blurs at times. For example, having short telomeres (the ends of chromosomes) is not a disease per se, but may indicate a high level of aging stress, which, in turn, may predict a host of other age-related conditions. And, though we now officially refer to hypercholesteremia and hypertension as “diseases,” they really are biomarkers associated with cardiovascular problems such as stroke, congestive heart failure, and myocardial infarction rather than signaling symptoms per se.

Still other markers are meant to provide insight into conditions that are difficult to pin down. For example, recent analysis has explored gene expression data that predict clinical depression and related phenomena such as suicidality or mania.1 Depression is diagnosed based on items ranging from a patient’s tendency to weep to sleep difficulties to somaticized anxiety in the form of gastrointestinal pain, for example. Reporting of these symptoms, of course, has its own cultural logic and biases. To get around this, researchers conducted within-subject analyses, following the same patients over years as their symptoms waxed and waned. Thus, they were not looking for biomarkers that predicted differences across individuals (with all of the biases that might entail) but rather differences over time within people, whose underlying cultural tendencies to report distress, anxiety or sadness is presumably constant. Moreover, they used supposedly objective measures—like suicide—to bolster claims of “objectivity.”

This is not a fool-proof approach to eliminating bias in the discovery of biomarkers. However, it could be the case that among different groups, different markers may be better or worse predictors. For instance, a particular group may tend to be so stoic that there is insufficient variation in the self-reported outcome or, alternatively, too much noise in their reporting, thus conflating other ailments. Indeed, scholars of depression found that biomarkers were better predictors for females when broken out by the specific mood disorder being predicted.1

Finally, we need to keep in mind that biomarkers that successfully predict an odds-difference in contracting a disease in the future is entirely different from predicting a major share of population variation in an outcome. The BRCA I and II mutations, for example, are very predictive of the likelihood that a woman will suffer from breast cancer during her lifetime. However, the vast majority of breast cancers cannot be predicted by these genetic variants. In other words, if we relied on BRCA tests alone to decide who should get mammograms, we would miss most cases.

These caveats aside, within-subject approaches to biomarker discovery are very promising for not only predicting and diagnosing conditions that are difficult to pin down but also for understanding their underlying biology. For example, it turns out that what we call “depression” may be several different diseases that have different biochemical signatures and, thus, would respond best to customized treatments.

Similar analysis has been conducted for long COVID, and distinct biological signatures have been found,2 though so far the analyses have not been systematic. In some sufferers, the viral proteins can still be found in the gastrointestinal tract. In others, biomarkers of inflammation are elevated. And, in a third tranche of patients, post-COVID syndrome is thought to be an autoimmune phenomenon, whereby the body’s defenses have turned on itself. There is molecular evidence for this hypothesis as well.2

To the extent that symptoms of long COVID come and go and that scientists are able to conduct within-subject searches for biomarkers that correlate with those changes, there is promise that we may arrive at a biological understanding of this syndrome and a way to improve objectivity in its diagnosis. Acute COVID has strained our health care system and catalyzed innovation at the same time (in the form of mRNA vaccines). Long COVID will be no different—the search for a magic indicator for this syndrome may help move biomarker research forward in a way that has lasting implications for our understanding, diagnosis, and treatment of many mental and physical diseases. How biomarkers should or should not be used for allocating medical and non-medical resources is a whole other question, of course.


[1] Le-Niculescu H, Roseberry K, Gill SS, et al. Precision medicine for mood disorders: objective assessment, risk prediction, pharmacogenomics, and repurposed drugs. Molecular Psychiatry. 2021. https://doi.org/10.1038/s41380-021-01061-w

[2] Nalbandian A, Sehgal K, Gupta A, et al. Post-acute COVID-19 syndrome. Nature Medicine. 2021. https://doi.org/10.1038/s41591-021-01283-z

Conley D. Long COVID, Biomarkers, and Health Policy. Milbank Quarterly Opinion. June 2, 2021. https://doi.org/10.1599/mqop.2021.0602

About the Author

Dalton Conley is the Henry Putnam University Professor in Sociology at Princeton University and a faculty affiliate at the Office of Population Research and the Center for Health and Wellbeing. He is also a research associate at the National Bureau of Economic Research (NBER), and in a pro bono capacity he serves as dean of health sciences for the University of the People, a tuition-free, accredited, online college committed to expanding access to higher education. He earned an MPA in public policy (1992) and a PhD in sociology (1996) from Columbia University, and a PhD in Biology from New York University in 2014. He has been the recipient of Guggenheim, Robert Wood Johnson Foundation and Russell Sage Foundation fellowships as well as a CAREER Award and the Alan T. Waterman Award from the National Science Foundation. He is an elected fellow of the American Academy of Arts and Sciences and an elected member of the National Academy of Sciences.

See Full Bio