Clinical epidemiologist Ziyad Al-Aly has access to a wealth that many researchers can only dream of: millions of sets of electronic medical records from the US Department of Veterans Affairs (VA), which provides health care to veterans of the world’s military.With this data in hand, Al-Aly, based in VA St. Louis Healthcare System in Missouri, and colleagues have studied the long-term effects of COVID-19, from heart disease1 to diabetes2. They also face the challenge of learning the long-term COVID – a situation in which people experience symptoms for months after the SARS-CoV-2 infection appears to have been resolved – and the recently published findings3 surprised some researchers. The team found that early vaccination only reduced the risk of getting long-term COVID-19 after infection by about 15%, which is significantly lower than other doses4, which suggested that vaccines reduce the risk by half.
It is a type of whiplash effect that people who follow a long COVID study are accustomed to seeing, as data from various studies report the effects of disagreement. Differences in the way the syndrome is described, the types of data used to research it and the way that data is analyzed have left both the public and policymakers facing different answers to the basic questions. How often is a long COVID-19? And how does a recent vaccination or re-infection of SARS-CoV-2 affect the risk of developing the condition?The answers to those questions can be used to develop COVID-19 policies, but the constant dripping of sawmill studies can also be confusing. said Al-Aly. Having such uncertainty does not lead to great trust, Al-Aly adds: “The public does not respond well when they say ‘between 15% and 50%’. ‘
Part of the problem is the long definition of COVID, linked to more than 200 symptoms, the severity of which can vary from discomfort to deterioration. The syndrome can last for months or years, and has a depressive tendency to recur, sometimes for months after the apparent recovery.To date, there is no agreement on how to define and diagnose a long-term COVID. The World Health Organization’s concerted effort, published in 2021, did not seem to appeal to patient advocates or researchers, and studies continue to use a range of procedures to explain the condition. Its frequency estimates can range from 5-50%.The study of this complex condition needs to be large enough to show the extent of symptoms and the potential impact of symptoms such as the age and severity of the severe SARS-CoV-2 infection. That’s where analytics like Al-Aly’s offer comes in with a host of benefits: data from major health care networks can provide samples of larger sizes. Al-Aly’s study of COVID long after the ‘success’ infection – following immunization – included records from more than 13 million people. Although 90% of those people were men, that still left 1.3 million women in the analysis, Al-Aly notes, in addition to many other studies that could be included.
These large numbers, along with other types of data available from other health records, allow researchers to perform complex statistical analyzes to carefully compare the number of coronavirus infections with the virus-free group, says Theo Vos, a dermatologist at the Institute for Health Metrics and Evaluation at the University of Washington in Seattle, who has worked with various data sources to learn more about COVID.But there are also obstacles. “People make the mistake of researching the quality and validity of research,” says WalidGellad, a medical policy researcher at the University of Pittsburgh, Pennsylvania.In particular, Gellad is concerned that studies relying on electronic health records will be tainted by behavioral differences. For example, compared to a person who does not seek medical attention because of the dangerous COVID-19, someone who wants it may be more likely to report long-term COVID symptoms, he says.
In addition, medical records and health insurance applications may not reflect the diverse population, says computer epidemiologist MaimunaMajumder at Harvard Medical School in Boston, Massachusetts. This is especially true in the United States, he says, when health insurance is very different. “The number of data points considered is usually so large that we mistakenly assume that this data must represent,” he said. “But this is not really the case.”
Majumder also wonders if studying claims data could lead researchers to reduce the number of people with long-term COVID, as many people may not seek medical help for their condition.Another problem is how symptoms are recorded in claims and electronic medical records. Doctors often record codes for several symptoms and conditions, but they rarely write code for all the symptoms a patient has, says Vos, and the selection of codes for a particular condition may vary from one doctor to the next. This can lead to differences in how COVID is reported and for how long. “Electronic health records have useful information on them, without a doubt,” said Gellad, who says the VA study was very well designed. “But to answer the question of how common an object is, it may not be the best one.”
Other methods also have their own pitfalls. Other studies rely on self-reporting, such as the COVID Symptom Study app developed by King’s College London and data science company ZOE, also based in London. Data from the app have shown that immunizations reduce the risk of people getting COVID 28 days or more after a viral infection by about 4 days. But studies where people voluntarily report their symptoms could be biased, because people with symptoms are more likely to participate, Gellad said. And studies that rely on smartphone apps may not fully capture data from poor communities.
One of the most important sources of data has been the UK Office for National Statistics (ONS), said NisreenAlwan, a public health researcher at the University of Southampton, in the UK. In May, the ONS reported that SARS-CoV-2 variants in infected individuals could contribute to their risk of developing long-term COVID. Of the two vaccinated participants, those thought to have COVID-19 caused by Omicron BA.1 variants were about 50% less likely to develop long-term symptoms of COVID four to eight weeks after infection than participants who may have had their illnesses caused by Delta variants. . . These findings are consistent with the results of 18 June paper5 based on ZOE data.
Alwan, a long-time COVID and data advocate for the situation, recommends the ONS study design, which includes registering a group of people carefully to represent the UK population, and following them up to ask about infection status and symptoms. control group is used, which can significantly affect outcomes, Alwan said. But counting different methods and definitions should not stop research. “That’s not new,” he said. “It’s something we had before COVID, in some cases.” For Al-Aly, the discrepancy between the results of the study is not surprising, nor does it condemn. Epidemiologists often combine evidence from multiple data sources with analytical methods, he says. Even if it is difficult to accurately estimate the effect of vaccination on long-term COVID risk, for example, researchers can look at trends. “You’re looking for a common thread,” Al-Aly said. “The common thing here is that vaccines are better than no vaccines.”
Source Journal Reference:Heidi Ledford, How common is long COVID? Why studies give different answers, Nature News (2022), https://www.nature.com/articles/d41586-022-01702-2
