How common is a long COVID?Why research gives different answers

Clinical epidemiologist Ziyad Al-Aly has access to a treasure trove that many researchers can dream of. Millions of electronic medical records from the US Department of Veterans Affairs (VA) provide medical care to national military veterans.

With this data in hand, Al-Aly and his colleagues at VA St. Louis Healthcare System in Missouri discuss the long-term effects of COVID-19. Cardiovascular disease¹ To Diabetes²..They also tackled the challenge of studying long COVID, a condition in which people experience symptoms months after the acute SARS-CoV-2 infection appears to have resolved, and recently published findings.³ It surprised some researchers.The team has only previous vaccinations It reduces the risk of developing long COVID after infection by about 15%. this is Significantly less than other estimates^FourThis suggested that the vaccine halves the risk.

It is the result of a type of whiplash that people doing long-term COVID studies are accustomed to, as data from various studies report inconsistent results. Depending on how the syndrome is defined, the type of data used to study it, and how those data are analyzed, both the general public and policy makers are working on different answers to the basic questions. .. How often are long COVIDs? And how do vaccinations, reinfections, or the latest SARS-CoV-2 mutants affect the risk of developing this condition?

Answers to these questions can be used to develop COVID-19 policies, but steady infusions of sea sewing studies can also cause confusion. Al-Aly says. Al-Aly adds that less uncertainty doesn’t give you much confidence. “People don’t react much when they say” between 15% and 50%. ” “

Slippery definition

Part of the problem is the definition of long COVID associated with more than 200 symptoms, the severity of which can range from inconvenient to debilitating. The syndrome can last for months or years and has a dire tendency to relapse months after apparent recovery.

So far, there is no consensus on how to define and diagnose long COVIDs.World Health Organization Attempt Consensus, published in 2021It has not proven to be popular with patient advocates and researchers, and research continues to use various criteria to define the condition. Its prevalence estimates can range from 5 to 50%.

Studies of such complex conditions should be large enough to reflect the extent of symptoms and possible effects of characteristics such as age and severity of acute SARS-CoV-2 infection. This is where analytics like Al-Aly offer many benefits. Data from large medical networks can provide huge sample sizes. Al-Aly’s study of long-term COVID (post-vaccination infection) after a “breakthrough” infection included more than 13 million records. Although 90% of those people were men, Al-Aly points out that there were still 1.3 million women left in the analysis, and many other studies can be gathered.

Big advantage

TheoVos, an epidemiologist at the institute, said that with the types of data available in many of these and some health records, researchers can perform complex statistical analyzes to obtain demographics of people infected with the coronavirus. It can be carefully matched against an uninfected control group. Health indicators and ratings at the University of Washington in Seattle. I have been studying COVID for a long time using various data sources.

However, there are drawbacks. “People mistake the scale of their research for its quality and validity,” says Walid Gerad, a physician studying health policy at the University of Pittsburgh, Pennsylvania.

In particular, Gellad is worried that electronic medical record-dependent research will be confused by behavioral differences. For example, he says, those who receive acute COVID-19 are more likely to report long-term COVID symptoms than those who do not seek treatment for acute COVID-19.

In addition, medical records and health insurance claims may not reflect demographically diverse populations, said Maimuna Majumder, a computational epidemiologist at Harvard University School of Medicine in Boston, Massachusetts. This is especially likely in the United States, where health insurance coverage varies widely, she says. “The number of data points considered is so large that we often mistakenly assume that these data are representative,” she says. “But this is not always the case.”

Majumder also wonders if researchers could underestimate the number of people with long COVIDs by studying billing data, as many may not seek medical care for their condition. I think.

Coding lesson

Another issue is how symptoms are recorded in billing and electronic medical records. Doctors often record some symptom and condition codes, but rarely list the codes for all the symptoms a patient is experiencing, and the choice of code for a particular condition varies from doctor to doctor. It may be. This can make a difference in whether COVIDs are reported and how long they are reported. “Electronic medical records definitely contain useful information,” says Gellad, who says the VA study is particularly well-designed. “But they may not be the best to answer the question of how common something is.”

Other methods also have pitfalls. Some studies rely on self-reporting, such as the COVIDSymptomStudy app developed by King’s College London and the data science company ZOE, also in London. According to app data, vaccination reduced the risk of COVID lengthening more than 28 days after an acute infection by about half.^Four.. However, because people with symptoms are more likely to participate, studies in which people voluntarily self-report their symptoms can be biased, says Gellad. Also, surveys that rely on smartphone apps may not be able to fully collect data from underprivileged communities.

One of the most useful data sources is the Office for National Statistics (ONS), said Nisreen Alwan, a public health researcher at the University of Southampton, UK. In May, ONS reported: Subspecies of SARS-CoV-2 that people are infected with can affect the risk of developing long-term COVID.. Among the double-vaccinated participants, those who appear to have COVID-19 caused by the Omicron BA.1 mutant are more likely than those whose infection was probably caused by the Delta mutant. , 4-8 weeks after infection, was about 50% less likely to develop long COVID symptoms. .. This finding is consistent with the results of the June 18 paper.^Five Based on ZOE data.

Find a common thread

Alwan, who has a long COVID and advocated the collection of data on symptoms, enrolled a group of people with the utmost care to represent the UK population and then asked them questions in the ONS study design. Admire the infection status and symptoms.

Other aspects of the study plan, such as whether controls are used, can have a significant impact on results, Alwan said. However, explaining different methods and definitions does not have to stall research. “It’s not new,” she says. “It’s what we had before COVID because of other conditions.”

For Al-Aly, the discrepancy between the findings is not surprising or criticized. Epidemiologists often interweave evidence from multiple data sources and analytical methods, he says. For example, researchers can look for trends even when it is difficult to accurately quantify the impact of vaccination on long-term COVID risk. “You look for a common thread,” says Al-Aly. “The common point here is that vaccines are better than no vaccines.”