By David Tuller, DrPH
*This is a crowdfunding month at UC Berkeley. If you’d like to support my work, the link is here.
I posted two days ago that a Lancet journal, eClinicalMedicine, had just published a letter of mine, as well as an unsatisfactory corrigendum to the study I had criticized. That study, “Effects of therapeutic interventions on long COVID: a meta-analysis of randomized controlled trials,” from Tan et al, reported “high-certainty evidence” in favor of exercise interventions, and suggested they “should be prioritized.”
My letter challenged that perspective. The corrigendum related to just one concern I had raised, and I found it unsatisfactory.
As it turned out, the record wasn’t complete. I had assumed the corrigendum was, in effect, the response to my letter. But at some point since then, the journal posted a direct response to my letter from the meta-analysis authors, in which they thanked me for my observations, among other remarks.
Here is one passage from their response that jumps out:
“We must acknowledge that certain expressions used in our writing, such as “high-certainty evidence” and “should be prioritized,” may have come across as overly assertive and could potentially lead to misinterpretation by readers. It would indeed have been more appropriate to phrase these as “evidence of moderate certainty” or “may be considered a preferred option.” We deeply appreciate your valuable feedback regarding these less precise formulations. Your comments contribute significantly to the rigor of academic discourse.”
I’m glad the authors agree that they should have toned down their language, although their substitute phrasings are still too positive, in my view. What I find especially perplexing is that they had an opportunity to fix these conclusions with the corrigendum—and they didn’t. As it stands, the article continues to assert that there is “high-certainty evidence” for exercise interventions, and that such interventions “should be prioritized.”
As I wrote in the initial version of my letter to the journal, that claim is “preposterous.” (I toned my letter down a bit for publication, upon request.)
On a second point: In my letter, I offered an example of a study whose participants did not have Long COVID—a point clear from the title itself, which indicated they were “coronavirus disease 2019 patients in the acute phase.” The authors have now removed that study from the meta-analysis and re-done everything to accommodate that change.
They also say they have conducted a thorough review of the remaining studies and found no other such aberrations. However, their review must have missed another study, said to be of “post-COVID-19 patients,” in which the criteria included “recovery from COVID-19 for at least 20 days.” Twenty days out does not meet anyone’s definition of Long COVID. The World Health Organization’s definition of what it calls post-COVID condition requires symptoms at least persisting or appearing at least two months after an initial infection. Anything sooner could easily be part of the acute infection.
The inclusion of this study in the analysis highlights one of the points I made in the letter. The trial populations are so heterogeneous that combining them all does not lead to clarity. That’s one reason the reported results are hard to take at face value.
Below is the authors’ response to my letter.
**********
Authors’ reply: Comments regarding “Effects of therapeutic interventions on long COVID: a meta-analysis of randomized controlled trials”
On behalf of all authors, I would like to sincerely thank David Tuller for the valuable feedback on our paper published in eClinicalMedicine.1 Your careful review of the literature inclusion criteria and evidence quality demonstrates a high degree of academic rigor, for which we express our deep respect.
You correctly pointed out that the study by Rodriguez-Blanco et al.2 included patients in the acute phase of COVID-19, which is inconsistent with our predefined criteria for the “Long COVID” population in the study protocol. We sincerely apologize for not identifying and excluding this study during the literature screening process. Upon receiving your comments, we have re-examined all included literature and removed the data from this study in the relevant analyses. We are pleased to note that after its exclusion, the main conclusions regarding the benefits of exercise training on key outcomes in patients with Long COVID, such as the 6-min walk test and fatigue, remained substantively unchanged. The pooled effect sizes and heterogeneity indices showed only minor variations, and the overall results remained robust. We fully recognize the stringent requirements for accuracy and transparency in academic research.
A Corrigendum has been issued to show the changes to the Article and the impact on the results and conclusions. Furthermore, we have taken this opportunity to conduct a secondary review of the remaining included literature (50 studies in total), confirming that all other references strictly adhere to the pre-defined PICO inclusion criteria, with no similar oversights detected. All revisions in this study have been cross-verified by two independent investigators.
We also appreciate your insights regarding the variations in the definition and inclusion criteria for Long COVID. We must acknowledge that when setting the inclusion criteria for “Long COVID patients,” we did not strictly adhere to the WHO’s precise definition.3 This is primarily due to the inherent breadth in the definition of “Long COVID” across current studies in this field, where different studies employ varying inclusion criteria. If systematic reviews were to include only studies that conform to a specific strict definition, at this stage, the evidence available for synthesis would be extremely limited. Such an exclusionary strategy might overlook a substantial body of early-stage studies containing potentially relevant information, thereby failing to provide timely, evidence-based references for clinical practice. Therefore, we adopted a relatively broad inclusion strategy aimed at comprehensively capturing interventional studies characterized by “Long COVID” or “post-COVID-19 condition” within the current global research landscape. We thank you for raising this point, which prompted us to provide a clearer explanation regarding these studies.
We highly acknowledge your concerns regarding the generally high risk of bias in many of the included studies and the low certainty of evidence according to the GRADE framework. In fact, during the design and execution of our study, we were fully aware of the methodological challenges commonly faced by early-stage research in this field. To address heterogeneity and potential bias, we employed random-effects models and conducted sensitivity analyses. The purpose of using the GRADE framework was precisely to objectively present the limitations of the current evidence and guide readers in interpreting the results cautiously.
Nevertheless, we must acknowledge that certain expressions used in our writing, such as “high-certainty evidence” and “should be prioritized,” may have come across as overly assertive and could potentially lead to misinterpretation by readers. It would indeed have been more appropriate to phrase these as “evidence of moderate certainty” or “may be considered a preferred option.” We deeply appreciate your valuable feedback regarding these less precise formulations. Your comments contribute significantly to the rigor of academic discourse.
We understand that as an emerging field, intervention research for Long COVID is still in its exploratory stages, where challenges such as difficulties in implementing blinding and high rates of loss to follow-up are often unavoidable. Furthermore, differences in exercise training program designs led to unavoidable heterogeneity. Nonetheless, we observed that across multiple studies evaluating exercise interventions, the direction of effect for improving specific health outcomes was consistent, and some results were statistically significant. Therefore, while transparently reporting their limitations, we chose to include the existing studies in the synthesis to provide as systematic and realistic a reference as possible under the current circumstances for future research.
We once again thank David Tuller for the constructive comments, which are crucial for enhancing the rigor and transparency of our study.
References
1.Tan, C. ∙ Meng, J. ∙ Dai, X. ∙ et al.
Effects of therapeutic interventions on long COVID: a meta-analysis of randomized controlled trials
eClinicalMedicine. 2025; 87, 103412
2. Rodríguez-Blanco, C. ∙ Bernal-Utrera, C. ∙ Anarte-Lazo, E. ∙ et al.
Breathing exercises versus strength exercises through telerehabilitation in coronavirus disease 2019 patients in the acute phase: a randomized controlled trial
Clin Rehabil. 2022; 36(4):486-497
Epub 2021 Nov 16
3. Soriano, J.B. ∙ Murthy, S. ∙ Marshall, J.C. ∙ et al., WHO Clinical Case Definition Working Group on Post-COVID-19 Condition
A clinical case definition of post-COVID-19 condition by a Delphi consensus
Lancet Infect Dis. 2022; 22(4):e102-e107
Epub 2021 Dec 21
