"Despite these strengths, the study has certain limitations that should be considered. About 57% of the original participants completed the second survey. Selection bias cannot be excluded, both from non-response at Phase I and from attrition between Phase I and Phase II. In addition, participants were recruited from a cohort of plasma donors after SARS-CoV-2 infection, which may limit generalizability to the broader population with SARS-CoV-2 infection. The stepwise variable-selection approach has known limitations and may result in overly optimistic model estimates16. However, in our study, several sensitivity analyses using different modelling approaches yielded broadly similar findings. Some adjusted estimates were based on sparse exposure counts and had wide confidence intervals, limiting their precision and quantitative interpretability. The study also used a self-administered online survey. Because participants had to interpret the questions on their own without the chance to ask for clarification, this approach may also have led to misinterpretation of some items. At the same time, self-reported data were an important component of early PCC research and remain valuable for capturing symptoms and functional impact that may not be fully reflected in routinely collected clinical data41,42. However, the absence of additional clinical or laboratory validation limited the ability to independently verify some reported symptoms. Higher PCFS grades were uncommon in our cohort and were therefore combined for the regression analyses. This may have reduced the ability to detect differences between participants with greater functional impairment. No formal symptom-level exclusion algorithm for alternative diagnoses was applied. Symptom information for the acute phase was collected retrospectively, and recall may have been influenced by participants’ current symptom status at the time of the respective survey."