Abstract
Dear Editor,
We have carefully read the study by Zhu Z et al entitled “Risk factors for osteoporotic vertebral compression fracture and evaluation of clinical outcomes of minimally invasive vertebral augmentation”. The authors adopted an integrated research design of “prediction-intervention-validation” and made valuable explorations in the field of individualized management of osteoporotic vertebral compression fractures. However, we would like to discuss several key methodological limitations in the article, in order to discuss with the authors and provide reference for subsequent research.
On the “indication confusion” caused by non-random allocation. The most innovative part of this study is the stratification of prospective cohort patients based on the predicted risk score, and the tendency to assign PKP or PVP surgery based on this. However, this risk score-based treatment allocation strategy is essentially a non-random observational comparison, which introduces the typical “indication confounding”. As Kyriacou emphasized, when the choice of treatment is closely related to the severity of the patient’s baseline disease, the observed differences in outcomes may be partly or even entirely due to this baseline imbalance, rather than the effect of the treatment itself. 1 In this study, patients who were judged to be “high risk” and therefore more likely to receive PKP already had characteristics such as poorer bone density and higher age, which could affect postoperative pain recovery, functional improvement, and even the risk of re-fracture, independent of the choice of surgical procedure. Although the authors mentioned that the final decision was made after considering many factors, they did not use statistical methods such as propensity score matching to balance the baseline risks of the 2 groups of patients, which weakened the causal inference strength of the conclusion that “PKP is better than PVP”. To establish the evidence-based basis for the optimal surgical procedure under risk stratification, the ideal solution is to conduct randomized comparisons within the same risk level.
Potential bias due to uncontrolled surgical technique variables (eg, approach).The study compared the outcomes of PVP and PKP in detail, but did not report or analyze a technical variable that may have a significant impact on the results: the surgical approach (eg, unilateral vs bilateral, transpedicular vs extrapedicular). Different approach choices directly affect the amount of cement injected, the distribution pattern, and the biomechanical stability of the reconstructed vertebral body.2,3 A large number of technical documents show that bilateral approach can usually achieve more symmetrical and more sufficient vertebral body filling, which may bring better pain relief and lower risk of bone cement leakage; while unilateral approach has the advantages of short operation time and less radiation exposure. If there is a systematic difference in the proportion of unilateral/bilateral approaches selected in the PKP group and the PVP group, then to what extent can the observed inter-group differences in clinical and imaging results be attributed to the core surgical difference of “PKP vs PVP”, and to what extent is it mixed with the technical effects of “different approaches”, which is currently unknown. This makes the interpretation of the results uncertain. Future comparative effectiveness studies should fully record, report, and control for these important technical covariates at the study design or statistical analysis stage.
In summary, the study by Zhu Z et al provides an important preliminary framework and inspiration for the precise treatment of OVCF. We believe that future studies can build on this basis to construct a more solid, reliable, and directly clinically practical evidence chain for individualized treatment decision-making by introducing randomized comparisons to strengthen causal inference within the risk stratification framework, and by standardizing the reporting and analysis of key surgical technique variables.
