R3 Seminar Recap: Effectiveness of Paxlovid in Protecting Against Long COVID: EHR Insights
RECOVER researchers shared study results which investigated whether taking Paxlovid during a COVID-19 infection could help prevent Long COVID.
During the April 8th RECOVER Research Review (R3) Seminar, RECOVER researchers Hannah Mandel, MS, Fei Wang, PhD, Sandy Preiss, MS, and Abhishek Bhatia, MS presented findings from two different studies of electronic health records (EHRs). Rainu Kaushal, MD (Cornell University), Dr. Wang’s co-author, joined the presenters for the panel discussion and audience Q&A. Both studies examined whether taking the antiviral drug Paxlovid during a COVID-19 infection could help prevent Long COVID. Although Paxlovid is known to reduce the risk of severe COVID-19, there has been limited research into how Paxlovid might impact a person’s risk of developing Long COVID. The studies presented are two of the largest to date that have explored the role of Paxlovid in Long COVID prevention.
Watch the R3 recording below or on YouTube
Ms. Mandel (NYU Langone Health) provided an overview of the EHRs that RECOVER uses for Long COVID research. RECOVER researchers can access EHRs through three research networks. These networks allow researchers to access tens of millions of health records which represent people from different backgrounds and all walks of life across the U.S.
The presented studies analyzed data from the research networks National COVID Cohort Collaborative (NC3) and PCORnet (The National Patient-Centered Clinical Research Network). Both studies used an approach called trial target emulation, in which researchers use statistical methods to analyze EHR data in a way that is similar to conducting a clinical trial. This method has several benefits:
It reduces biases and confounding (which occurs when multiple factors could influence the relationship being studied, therefore producing study results that could be inaccurate).
It is a way to explore the effects of potential treatments that is faster and less expensive than a clinical trial.
It allows researchers to understand how potential treatments are being prescribed in the real world.
PCORnet data reveals Paxlovid reduces Long COVID risk for people with high-risk health conditions
Dr. Wang (Cornell University) presented findings from the study that used PCORnet data to compare two groups of patients: those who received Paxlovid within 5 days of their COVID-19 infection and those who did not receive the drug. Researchers examined how often patients died, were hospitalized, or developed Long COVID.
The study included EHR data from nearly 500,000 patients. Researchers examined a period between March 1, 2022, and February 1, 2023 (a time when Paxlovid was increasingly used to treat COVID-19). All EHRs used in the study included information that indicated the patient had COVID-19. Researchers used two methods to identify people with COVID-19:
They collected formal COVID-19 diagnoses, which are noted in EHRs using specific codes called ICD-10 codes.
They also used a computer program called a machine learning algorithm that analyzed symptoms and other information to identify people who are likely to have the condition. Formal diagnosis of Long COVID can be limited, especially early in the pandemic when diagnosis codes were not yet developed and healthcare providers were less familiar with symptoms of the condition. Therefore, this additional method of identifying patients is important to ensure that research includes more people who have been affected by Long COVID.
Researchers found that taking Paxlovid during a COVID-19 infection is effective in reducing risk of Long COVID, hospitalization, and death. The study team also separated patients into groups according to whether they had other health conditions that put them at a higher risk for developing Long COVID. In the group of patients found to have a higher risk, Paxlovid reduced the rate of Long COVID by 2.99 events for every 100 people. However, among the patients who had a lower risk, researchers found no statistically significant reduction in risk for developing Long COVID.
N3C data shows taking Paxlovid yields a small benefit in preventing Long COVID
Co-authors Sandy Preiss (RTI International) and Abhishek Bhatia (University of North Carolina at Chapel Hill) presented findings from the study that analyzed EHR data from N3C. Mr. Preiss explained how real-world evidence produced from data sources like EHRs can help inform clinical trials. Because real-world evidence shows how medicines are actually prescribed, it can help researchers make decisions on clinical trial designs and study hypotheses.
Like the PCORnet study, the N3C study identified people who were likely to have Long COVID, using a machine learning algorithm. Because Long COVID is a complex condition that affects everyone differently, researchers also used three common symptom clusters (groups of symptoms)—cognitive symptoms, respiratory symptoms, and fatigue symptoms—to identify people who were likely to have Long COVID.
Mr. Bhatia described how the study team accounted for differences between a traditional clinical trial and the target trial emulation that analyzed EHR data. For example, in a clinical trial, patients are randomized, or assigned to a specific treatment. Researchers do their best to remove all other factors that could impact the effect of the treatment so they can better understand how the treatment works. This controlled environment is different from the real-world healthcare that is reflected in EHRs. In this target trial emulation, researchers used multiple statistical methods to eliminate confounding factors, or factors that could impact researchers’ analysis and understanding of whether Paxlovid helps to prevent Long COVID.
Researchers in this study found that taking Paxlovid reduced the frequency of Long COVID by 6 percent for the population studied. When looking at the specific symptom clusters, researchers saw that Paxlovid protected against symptoms related to cognition and fatigue, but not respiratory symptoms.
Study results agree on Paxlovid’s role in Long COVID prevention
Mr. Bhatia noted that the results from the two studies are complementary and reach a consensus that Paxlovid can help reduce the risk of Long COVID. Both studies are currently in pre-print, which means they are publicly available online while the results continue to be peer-reviewed by other researchers to confirm accuracy and scientific rigor. Mr. Bhatia also recommended that future research should account for the wide range of symptoms associated with Long COVID and explore tailored treatments for different symptoms.
Studies that analyze EHR data do have limitations, Dr. Wang explained. For example, structured EHR data may be incomplete, and researchers do not have a way to determine whether people actually took Paxlovid after being prescribed the medication. However, EHR studies allow researchers to examine large groups of people from all walks of life, meaning that results will be relevant for a wide range of people.
To find recordings and transcripts of more R3 seminars, visit the RECOVER YouTube channel and the R3 webpage.