Letter to the editor on methodological considerations in the study of post-stroke facial palsy recovery patterns

To the Editor,

We read the recent article published by Professor Lu's team (Clinical Rehabilitation, 2025) ¹ with great interest. This study, based on the Bergen NORSTROKE Registry, ² provides valuable insights into the dynamic changes of early facial paralysis after stroke.

We hope to raise a possible methodological question. The study defined the recovery time as starting from the first in-hospital assessment, a practice that may introduce left truncation bias. ³ Patients who recover rapidly before admission—especially those with mild or transient symptoms—may be overlooked in the recovery analysis, which may lead to an underestimation of the early recovery rate. For instance, the study observed that 25% of the patients recovered on the second day, but this proportion might not reflect the true situation. Future research may consider taking the onset of symptoms as the starting point or adopt statistical methods such as inverse probability weighting, which may help reduce this bias.

Furthermore, we noted that after adjusting for the severity of movement disorders, the association between stroke subtypes and facial paralysis reversed—this is a typical case of Simpson's Paradox. ⁴ This finding suggests that the severity of stroke may have a more crucial impact on facial paralysis than the type of stroke, and also emphasizes the importance of correcting severity in prognostic models.

We appreciate the important work accomplished by the author and believe that the discussion of these key points will further enhance the influence of his research findings.