Q&A from February 10, 2026 R3 Seminar

Dr. Hornig: I think the more education we can provide to clinicians about how symptom patterns can vary over time is crucial. In my case, a specific organ domain was prominent at one time, but it may not hold true for other times. Understanding that symptoms may fluctuate among different individuals over time is vital. I use the metaphor of checking windshield wipers—they’re working or they’re not working. One must consider the larger picture of symptoms varying over time. It can be challenging when clinicians feel pressed for time, having only a couple minutes to get an elevator pitch from their patients about their issues at that moment. This system doesn’t accommodate integrative team approaches, which leads to significant frustration. Clinicians need time to properly assess and understand the multifaceted nature of Long COVID symptoms. Moreover, we can work toward dispelling persistent myths within the clinical community. The ongoing biomarker studies in RECOVER, as well as the pathobiology studies, are crucial for understanding if and how viral persistence is an issue, as well as addressing vascular problems and cardiovascular sequelae, cognitive decline, and other problems. The more we learn about these issues, the better equipped we will be to inform clinicians and improve patient care.

Dr. Henrich: Understanding Long COVID is critical for clinicians. The information from studies like this is valuable for revealing what symptoms to expect when we see patients in the clinic. It educates clinicians about the heterogeneous nature of the disease. This knowledge is vital for offering comprehensive care based on understanding individuals’ trajectories. If clinicians are prepared for prolonged, varying symptoms, they can provide more substantial support, including directing patients to clinical trials that advance our understanding of Long COVID while working toward patient well-being.

Dr. Thaweethai: This is a hard question to answer. It’s a little beyond the scope of the analysis that we conducted, which looked at participants for up to 15 months after their infection, so I can’t comment on specific rates of continuing to experience persistent symptoms. I think we can speak to, from a high-level point of view from the data, that many participants in RECOVER are still reporting numerous symptoms years out from their initial infection. Saying that I don’t know the number doesn’t mean that it’s not high or that it’s not something to be very concerned about, and I think it’s certainly going to be the subject of future analyses.

Dr. Hornig: In my experience, recrudescence of symptoms always came with some other new infection. Recently, I had my first reinfection with SARS-CoV-2; I had a common cold–type coronavirus, 229E, back in 2022 that put me in the hospital. I had bilateral pneumonia and reactivation, kind of like shingles, but without the rash (known as zoster sine herpete), only with itching. I’ve had this form of herpes zoster 4 times, and each time I’ve had those additional issues, it’s been a recrudescence of various symptoms. But the organ system involved has varied. The question becomes, is SARS-CoV-2 becoming reactivated in my gastrointestinal (GI) tract and leading to GI symptoms, or is it the overall impact on my immune system and maybe just a random effect of which organ is impacted? I’ve had some vascular and cardiovascular issues and so forth. This is why this research, and continued funding for this research, is so critical.

Dr. Henrich: I agree that we need to understand this better. We need this research to happen. I can always speak anecdotally. I’m a bit biased because in our clinical studies, we often have people who are on the sicker end or more persistent end, where we see that people who persist are unfortunately the ones who tend to keep persisting. Once you get to a couple of years, those are usually the folks who we see go much longer. Anecdotally, we have people with 5-plus years impacted that we are now enrolling in some of these studies. But I think having further research and further understanding of why some people fully recover and why some people take a long time but do recover is important. We certainly see relapses, especially when we administer questionnaires for different studies. It seems that there is some type of insult to the immune system or inflammation, whether it’s another illness, another infection, or potentially reinfection, that can trigger them. That’s anecdotally something we observe, and I think what’s frustrating is that we would love to have therapeutics that work for a large number of people that we can roll out to stop or manage these. We just need more research to do this and have the biggest impact on everyone.

Dr. Thaweethai: First, regarding the relapse or late-onset symptoms in general, it’s important to echo Dr. Henrich’s thoughts. It’s important for us to understand what’s going on and how it might be triggered by other infections or pathophysiologic mechanisms. It can be difficult for us to assess this in a rigorous statistical and epidemiological framework because there are so many possibilities, and if you test a million things, you’re going to find some false positives. We need to be thoughtful and rigorous when analyzing this, so we’re not misled by spurious findings. 

It’s complicated when analyses are longitudinal because you’re trying to characterize someone’s entire natural history. For instance, what would have happened if this person received this treatment? Would their symptoms have improved or gotten worse? It’s tricky, and there are many things to account for. This is the subject of future work, which can help us better understand the triggers. 

Regarding strains, we did not formally adjust for any strain because our cohort was restricted to people infected during the Omicron era or later. We didn’t include people with earlier strains in the analysis. But I think once we get into analyses that incorporate the post-acute individuals enrolled in RECOVER more than 30 days after their first infection, they could have been infected as early as 2020, and this is something we’ll certainly want to account for, to better understand whether specific findings are applicable to people with particular strains. It’s an excellent question, and it’s critical to address it.

Dr. Hornig: One of RECOVER’s long-term focuses is investigating symptoms that might tell us about the development of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). The more we do this, the more we will learn. We should be bringing all sorts of comparator groups into these studies (e.g., people who have infectious mononucleosis caused by Epstein-Barr virus). We should do follow-ups on those individuals, pre-pandemic or in the current era. I think looking at all the infectious stimuli that can cause similar symptom patterns is crucial. SARS-CoV-2 has its own unique pattern, but so do other viruses and their clinical features. The more we can identify the common threads, the better we will understand the differences and treat disorders like ME-CFS, which is largely initiated by an infectious and viral-like syndrome.

Dr. Henrich: Again, I think it’s mixed. First, there’s evidence that having the vaccine before infection can reduce the chances of developing Long COVID, which logically makes sense from an immunological standpoint. If you can reduce the severity of infection, it’s likely beneficial against developing Long COVID. However, from a clinical perspective, responses to vaccination after Long COVID are highly variable. Some people actually feel better, some feel worse, and many feel no difference. This variability highlights the complexity of these conditions. Although we might have some anecdotal insights, collecting more systematic data is essential for drawing more conclusive understandings.

Dr. Thaweethai: There’s a lot of evidence supporting what Dr. Henrich is saying—that being vaccinated before first infection can be protective against Long COVID. We published this in the RECOVER-Pediatrics study, demonstrating that adolescents vaccinated before infection were less likely to develop Long COVID later. However, it doesn’t answer what happens to people already experiencing Long COVID who get vaccinated. Our cohort includes people with varying vaccination status and symptom histories. Because of this complexity, we didn’t rigorously evaluate this question in our paper. The topic was significant enough that we aim to compare individuals with Long COVID who received the vaccine against those who didn’t. We must carefully account for various factors affecting those comparisons because someone’s underlying health can influence their vaccination decision. This is a complicated question, and RECOVER data are well positioned to answer it through the detailed symptom histories and immunological conditions we collect.

Dr. Hornig: I think of it in percentages; I think of myself as getting back to where I was before this reinfection a month ago, and I was about 85%–90% of my pre-pandemic self. However, I came into my first COVID infection with autoimmune diseases, so my baseline may have been a little different from other people’s, but I wasn’t sleeping 16 hours a day as I was when I had Long COVID. I wouldn’t even call that unrefreshing sleep; I just needed a whole lot more. Therefore, it’s vital to regain functionality. Getting back to one’s former level of function, even if it was at a lower level, is essential for me when I go to the doctor. I’m looking for assistance with what interferes most with my functionality. It’s common for older patients to feel their concerns are not heard, and that’s crucial for clinicians to understand.

Dr. Bevc: RECOVER is conducting a large multiomics study. Information about accessing data is covered in our previous R3 Seminar, with additional guidance on how to access and identify opportunities for research.

Dr. Thaweethai: Infection dates ranged from October 2021 to June 2023.

Dr. Thaweethai: This is explored further in Table 2 of the published manuscript. If you look, there is not a clear trend regarding age across the different profiles, suggesting that people of all ages can experience persistent Long COVID.

Dr. Thaweethai: This study included only participants who were prospectively followed from the first time they had COVID-19. Since RECOVER began enrolling participants at the end of October 2021, the only participants who could have been included in this particular study were those whose first infection was at that time or later. The Omicron variant became the predominant strain around that time. Therefore, this study included only participants who were infected with the Omicron variant, so it was not necessary to adjust for variant as a covariate in these analyses. This particular study population does not include participants who were infected with earlier variants. In this particular study, we did not analyze whether the incidence of Long COVID changed with different variants, but this will be incorporated into future longitudinal analyses of RECOVER data.

Dr. Thaweethai: Symptoms related to the nervous system were also included in the analysis. A full description of the symptoms at 3 and 15 months can be found in Supplementary Figure 1 of the manuscript.

Dr. Thaweethai: The threshold for classifying an individual as having probable Long COVID was an LCRI score of 11 or greater. In this particular study population, we found that 10.3% had an LCRI score of 11 or greater at 3 months after their first infection. Supplementary Table 4 of the manuscript has the percentage of participants in each trajectory profile who had an LCRI score of 11 or greater at 3, 6, 9, 12, and 15 months after first infection.

Dr. Thaweethai: In our analyses, we found that individuals with a higher LCRI score were not much more likely to miss subsequent visits. We explored this in Supplementary Table 2c of the manuscript. For example, if we look at participants who attended the 6-month visit and stratified them by LCRI score of 0, 1–10, 11–19, and 20+, we found that the proportion who missed the 9-month visit was only slightly higher for the 20+ group (11% for 0, 13% for 1–10, 13% for 11–19, and 17% for 20+). The direction of association was reversed when looking at participants who attended the 12-month visit and the percentage who missed the 15-month visit (12% for 0, 14% for 1–10, 12% for 11–19, and 7% for 20+). This indicates that there was only a weak association between having more symptoms with a Long COVID diagnosis and missing later visits. Of course, it is still possible that individuals who suddenly had worse symptoms were less likely to have visits and that this was not explained by LCRI at the prior visit, but this was unlikely given the patterns we had seen in the observed data.

Dr. Thaweethai: Pre-existing and new-onset symptoms were included when calculating the LCRI. Additional details can be found in Thaweethai et al. (2023) and Geng et al. (2025).

Dr. Donohue: We did not specifically examine which treatments people received in this work, but treatments and other factors are important questions for future studies. 

Dr. Thaweethai: Finite mixture modeling was used to group participants into distinct longitudinal profiles. Although profiles A and B were similar in some respects, we used the Bayesian information criterion to determine the optimal number of profiles. We found that each of the 8 profiles was sufficiently different from the others to be defined as unique trajectories. A full description of the statistical methods used in the analysis can be found in the manuscript and in the Supplementary Methods section of the Supplement.

Dr. Thaweethai: Participants in trajectory profiles A and B have been followed well beyond 15 months in the RECOVER study to understand the severity of their symptoms, as well as other clinical characteristics that change longitudinally. Follow-up of these participants, as well as those with other trajectory profiles, continues in RECOVER and will be the subject of future analyses.

Dr. Thaweethai: Formal statistical testing of this particular hypothesis was not evaluated in this study.

Dr. Donohue: We haven’t examined this yet, but that is a great question for future work. 

Dr. Thaweethai: Demographic and clinical characteristics of the study cohort, stratified by the 8 trajectory profiles, can be found in Table 2 of the manuscript. The frequency of symptoms reported, also stratified by the 8 trajectory profiles, can be found in Supplementary Figure 1 of the manuscript. Given the very large number of symptoms and demographic/clinical characteristics measured in this cohort, formal statistical testing was not performed to evaluate associations with each factor and each of the 8 trajectory profiles due to the risk of repeated testing.

Dr. Donohue: Anxiety and depression were measured in RECOVER, and Supplementary Figure 1 in the manuscript shows the respective frequencies of these and other symptoms at 3 and 15 months. For example, in profile A (persistent group), 35.9% of participants reported anxiety at 3 months, with 34.8% of people in profile A reporting it at 15 months. Neuropsychological symptoms are also of interest for future work. 

Dr. Thaweethai: Additional analyses are underway, looking at studies with longer follow-up beyond 15 months after infection.

Dr. Thaweethai: Participants in Profile A (persistent, high symptom burden) compared with those in Profile H (consistent, minimal to no symptom burden) were more often female (77% vs. 64%). The full distribution of gender and age, along with other demographic and clinical characteristics of the study cohort, stratified by the 8 trajectory profiles, can be found in Table 2 of the manuscript.

Dr. Thaweethai: The median age was not very different across the 8 trajectory profiles identified. The full distribution of age at first infection, stratified by the 8 trajectory profiles, can be found in Table 2 of the manuscript.