Response to the letter to the editor: Confounding by indication in comparing posterior longitudinal ligament resection and preservation

Article information

Asian Spine J. 2026;20(5):1025-1026
Publication date (electronic) : 2026 October 22
doi : https://doi.org/10.31616/asj.2026.0903.r2
1Division of Neurosurgery, Department of Surgery, Faculty of Medicine Vajira Hospital, Navamindradhiraj University, Bangkok, Thailand
2Division of Neurosurgery, Department of Surgery, Chulabhorn Hospital, Chulabhorn Royal Academy, Bangkok, Thailand
Corresponding author: Padungcharn Nivatpumin Division of Neurosurgery, Department of Surgery, Chulabhorn Hospital, Chulabhorn Royal Academy, Bangkok, Thailand Tel: +66-819172700, Fax: +66-76254430, E-mail: padungcharn_26@hotmail.com
Received 2026 October 2; Accepted 2026 October 12.

Dear Editor,

We thank the authors for their careful reading of our study [1] and for their thoughtful comments. We agree with most of their points and respond to each one below.

Groups were chosen by diagnosis and risk, not by chance

We agree. This was a retrospective study, and surgeons did not assign patients to posterior longitudinal ligament (PLL) resection or PLL preservation at random. The choice depended on the patient’s disease and risk. As stated in our paper, the PLL was preserved in patients with axial neck pain only, trauma with an intact PLL and disc, and ossification of the posterior longitudinal ligament with double-layer sign (because of the high risk of cerebrospinal fluid leakage). Most patients with cervical spondylosis were in the resection group, and most trauma patients were in the preservation group. This is “confounding by indication” [2]. It means the two groups were different from the start, so the differences we saw may come from the disease and its difficulty, not only from the PLL procedure.

Retractor time and complications

We agree that our wording should be more careful. Our paper already lists selection bias and the small sample as limitations, and we thank the authors for explaining what this means for the results. Retractor time was longer in the resection group, but the more complex disease in that group may also explain this. For complications, the difference between the groups was not statistically significant (odds ratio 2.074; 95% confidence interval, 0.628–6.846; p=0.231). So, our statement that PLL resection “could potentially result in postoperative complications” is only a possibility that needs more study. It is not proof that PLL resection itself causes complications. We will state this clearly: our results show an association between the groups, not a cause and effect.

Stratified analysis and the trauma subgroup

We thank the authors for this suggestion. We had only 18 complications in total, so a complex multivariable model could give unstable results. We agree that simple descriptive reporting by diagnosis is a better choice. The trauma subgroup (15 resection and 27 preservation patients) is the most similar pair of groups. Even so, the groups may still differ, because an intact PLL and disc were required to preserve the PLL. For reference, the complication types were: dysphagia (8 of 68 vs. 2 of 36), recurrent laryngeal nerve palsy (3 of 68 vs. 2 of 36), hoarseness (2 of 68 vs. 0 of 36), and neuropathy (1 of 68 vs. 0 of 36) (resection vs. preservation). All complications were temporary and resolved within 1 month.

The two different p-values (0.787 and 0.231)

Thank you for pointing this out. The two values answer slightly different questions.

• p=0.787 (Table 1 of our study [1]) compares the full list of complication types (none, neuropathy, dysphagia, hoarseness, recurrent laryngeal nerve palsy) between the two groups. As described in our Methods, this was done with the Mann-Whitney U test.

• p=0.231 (Table 2 of our study [1]) comes from a logistic regression that asks a simpler question: did the patient have any complication, yes or no? The odds ratio of 2.074 is the same as the simple (unadjusted) ratio from our raw numbers (14 of 68 vs. 4 of 36), so this model was not adjusted for other factors.

We agree that the Mann-Whitney U test is not the usual choice for a complication table, and that this was not explained well. To check, we re-tested “any complication” (14 of 68 vs. 4 of 36) with standard tests. The chi-square test gave p=0.224 and Fisher’s exact test gave p=0.283. All of these tests lead to the same conclusion: the difference in complications was not statistically significant. We will state the test used for each p-value in the corrected text.

Number of patients (102 vs. 104)

We thank the authors for finding this error. The correct total is 104 patients (68 PLL resection and 36 PLL preservation). Every part of Table 1 of our study [1] adds up to 104 (sex, diagnosis, and complications). The “102” in the abstract and the text is a typing error. The same applies to “62 males,” which should be 64 (64 males and 40 females). We will ask the journal to publish a correction.

Conclusion

We thank the authors again. We accept that our findings should be read as differences between groups that were defined by diagnosis, and we will say so more clearly in the interpretation of retractor time and complications. Larger studies that compare similar patients, ideally prospective, are needed to learn whether PLL resection truly affects retractor time or complications.

Notes

Conflict of Interest

No potential conflict of interest relevant to this article was reported.

Author Contributions

All authors conceived, drafted, critically revised, and approved the final version of this letter.

References

1. Niljianskul N, Nivatpumin P. Resection of the posterior longitudinal ligament in anterior cervical decompression surgery: a retrospective study of the clinical and radiographic outcomes in Thailand. Asian Spine J 2026;20:60–9. https://doi.org/10.31616/asj.2025.0134.
2. Kyriacou DN, Lewis RJ. Confounding by indication in clinical research. JAMA 2016;316:1818–9. https://doi.org/10.1001/jama.2016.16435.

Article information Continued