Your Lifetime Risk of Severe Long-Covid Is Likely <5%


What data from the U.K., Sweden, Germany, and Denmark show
Image by

If you are young and healthy, your risk of severe or fatal Covid-19 is minuscule. Even so, one concern remains: long-Covid or the post-Covid-19 syndrome, characterized by having at least one persistent symptom for at least 3 months after Covid-19, not explainable by alternative diagnosis.

But how worried should we be about the severity of long-Covid?

Previously in “Current Long-COVID Statistics Are Missing the Background Prevalence,” I described that even before Covid-19 emerged, 10–30% of the general population typically lived with symptoms similar to long-Covid, such as fatigue, breathlessness, headache, and sleep disturbance. So, minus the background prevalence, the true prevalence of long-Covid is likely:

  • one or two in ten after symptomatic Covid-19;
  • one or two in five after severe Covid-19;
  • close to zero in ten after asymptomatic Covid-19.

Some factors further modify such prevalence: more severe Covid-19, female sex, pre-existing diseases, and older age increase the risk of long-Covid, whereas vaccination decreases the risk.

But there’re two more crucial questions I promised to follow up: how severe are long-Covid symptoms, and how long does recovery take? In this article, I’ll discuss the former. I’ll be covering studies from four countries to draw a conclusion but feel free to skip to the end for it.

(i) Data from the U.K

A comprehensive study from the U.K conducted 10 longitudinal analyses from different cohorts, with a sample size of 48,901 people, of which 6,907 had Covid-19, mostly symptomatic and non-severe.

This study noted a 7.8–17% prevalence of long-Covid. But if we count diagnosed rather than self-reported long-Covid, the prevalence was just 0.27% of Covid-19 cases. This could be due to people not seeking healthcare despite having long-Covid symptoms, probably because symptoms were manageable and not of sufficient severity to seek healthcare.

In fact, if we count severe long-Covid that disrupt day-to-day activities, the prevalence was 1.2–4.8% of Covid-19 cases: 1.2% in the mean age of 20, which increased to 4.8% in the mean age of 63.

Long-Covid prevalence also follows an inverted U-shaped association with age: highest rate in 45–69 years old, but the same rate in ≥80 and 18–24 years old. Long-Covid risk also increased with female sex, pre-existing diseases, and poor general/mental health.

Although this study did not include a control group analysis, at least they use a stricter long-Covid diagnosis of symptoms lasting at least 3 months post-Covid, not due to alternative causes. That said, if we deduct some level of background prevalence, the prevalence of severe long-Covid limiting day-to-day activities could be less than 1.2–4.8% of Covid-19 cases.


But the U.K Office for National Statistics (ONS) gave a more worrisome estimate back in March/April 2021, which numerous long-Covid advocates and researchers have cited.

From 3-month onwards, 60.6% of long-Covid cases (self-reported) said that their symptoms limit their daily activities in some form. As much as 18.1% of long-Covid respondents said their symptoms are severe enough to limit their daily function a lot. Such estimates did not change much as of March 2022.

But this research (unpublished) relied on voluntary self-reporting, which is prone to sampling bias. So, people with more severe long-Covid may be more likely to report their symptoms than those with very mild or without long-Covid. Not to mention that such estimates lack a control group analysis.

(ii) Data from Sweden

Another well-done study from Sweden tracked 2,149 healthcare workers (around 33–56 years old) over time, of which 323 and 50 developed non-severe and severe Covid-19, respectively, but severe cases were unanalyzed.

Results revealed that at 2-month, 26% of Covid-19 survivors reported moderate-to-severe long-Covid symptoms compared to 9% in the non-Covid control group, giving an excess of 17%. At 8-month, such numbers dropped to 15% and 3%, respectively, an excess of 12%.

But if we count moderate to marked disruption in a disability scale, measuring impairments in daily activities, the numbers were 11% and 2% in the Covid-19 and non-Covid group, respectively, an excess of 9%.

Assuming marked disruption is rarer than moderate disruption — which was true at 2-month, but no data at 8-month — the prevalence of severe Covid-19 could be less than 9% in this study, presumably at 5%.

(iii) Data from Germany

A study from Germany studied 1,267 people from 341 households; 404 people were <14 years old, 140 people were 14–18 years old, and 723 people were >18 years old. About 50% of them had Covid-19, mostly non-severe.

At one-year follow-up, moderate-to-severe long-Covid were more common in infected than uninfected women (36.4% vs. 14.2%; excess of 22.2%), men (22.9% vs. 10.3%; excess of 12.6%), and 14–18-year-old adolescent girls (32.1% vs. 8.9%; excess of 23.2%).

But it’s not any more common in infected than uninfected children and adolescent boys, suggesting that boys up to 18 years and children below 14 years won’t get moderate-to-severe long-covid at one-year post-Covid.

Notably, this study only recruited households with at least one member with confirmed Covid-19. While this ensures proper control comparison since household members tend to be more similar than unrelated people, it could also introduce sampling bias. In fact, this study also found that the risk of moderate-to-severe long-Covid of a person increased by 12% for every additional moderate-to-severe long-Covid person in their household.

So, some confounding variable is likely involved. For instance, parents with more severe long-Covid symptoms may report that their child had the same or may negatively affect their child’s health. Or it could be vice-versa where long-Covid in the child add further stress to long-Covid in the parents. Or certain genetic factors predispose certain households to more severe long-Covid.

It may thus be reasonable to discount the prevalence of moderate-to-severe long-Covid in this study by 10% to generalize it to the general population; that is, an excess of 12% in women, 3% in males, and 13% in adolescent girls. Again, assuming severe is rarer than moderate long-Covid, the prevalence of excess severe long-Covid could be lesser, presumably <10%.

(iv) Data from Denmark

(iv) In one nationwide study in Denmark, 61,002 Covid-positive and 91,878 Covid-negative cases were investigated. Participants were at least 15 years old, with a median age of 50, and 61% were females.

At 6–12-month, 29.6% of Covid-19 survivors experienced long-Covid symptoms versus 13% in non-Covid controls. Of the Covid-19 survivors, 4% were hospitalized and 96% were not, and previously hospitalized survivors had a higher risk of long-Covid, consistent with many other studies showing increasing long-Covid rates and severity with increasing Covid-19 severity.

At 6–12-month, 12% of Covid-19 survivors took at least one sick leave compared to 7.7% in the non-Covid control group, giving an excess of 4.3%. For full-time sick leave, the numbers were 9.4% and 6.5%, respectively, an excess of 2.9%. And full-time sick leave is likely a better indicator of severe long-Covid than self-reported long-Covid symptoms.

At 6–12-month, Covid-19 survivors also had 2.5%, 1.2%, 1%, and 0.6% excess cases of diagnosed chronic fatigue syndrome, anxiety, depression, and post-traumatic stress disorder. Such disorders are also a fair proxy of severe long-Covid since they impose major impediments in daily functioning.

But this study had 260,637 non-responders. Compared to non-responders, responders (or participants) were more often females, older (50–70 years old), working in healthcare, and living outside the capital. So, some level of sampling bias could be present, given that female sex and older age are known risk factors of long-Covid in this study as well as others. As a result, people with long-Covid might be more likely to participate in this study than those without, overestimating the long-Covid prevalence to some extent.

One could argue that people with severe long-Covid may be less likely to respond to study participation since it could require a huge amount of effort. But 260,638 non-responders is a big number; 430,173 individuals were invited to participate, so that over a 50% decline rate. Most of them shouldn’t have long-Covid so severe that they can’t participate in the study.

The most prudent estimate

After researching and writing this article, I think severe long-Covid that markedly disrupts daily functioning occurs at the rate of 1–5 cases per 100 Covid-19 cases. Such rate increases by up to ~2–3-times with more severe Covid-19, female sex, pre-existing diseases, older age, and no vaccination.

Even so, long-Covid is still a potentially debilitating condition that can rob one’s quality of life indefinitely, a risk best avoided. So, avoid indoor crowds without masks and proper air ventilation, and get vaccinated, especially if you are at high risk of severe Covid-19 or severe long-Covid.

Given the massive scale at which Covid-19 inflicts people, sometimes more than once per person, long-Covid is a serious matter at the public health level. Risks, even if low, can add up to enormous heights.

At the individual level, however, the risk of severe long-Covid should be multiplied by the risk of getting Covid-19, which varies greatly depending on geography, public health policy, personal behavior, and vaccination status.

Assuming 5–10% of the general population gets Covid-19 at one year, the risk of severe long-Covid would be 0.05–0.5% (0.05–0.1 * 0.01–0.05 = 0.0005–0.005) per year per person. But assuming everyone gets at least one episode of Covid-19 at some point in their life, their lifetime risk of severe long-Covid would be 1–5% (1 * 0.01–0.05 = 0.01–0.05) per person.

Again, such risks would increase, probably by about 2–3-times, with more severe Covid-19, female sex, pre-existing diseases, older age, and no vaccination.

But in the end, these numbers are just estimates. Risks are complicated, which depend on one’s risk profile and situations. Scientific data are always shades of grey; answers are always not clear-cut. After all, no study is perfect, and the best we can do is infer something practical from dissecting several studies.

Comments / 16

Published by

MSc Biology student | 5x first-author academic papers | 100+ articles on coronavirus | Freelance medical writer


More from Shin

Comments / 0