Letter to the Editor concerning “Percutaneous versus traditional and paraspinal posterior open approaches for treatment of thoracolumbar fractures without neurologic deficit: a meta-analysis’’ by Sun, XY., Zhang, XN. & Hai, Y. Eur Spine J (2016). doi:10.1007/s00586-016-4818-4
Authors:
Yongzhao Zhao, Guoxin Fan, Yanjie Zhu, Shisheng He
With great interest, we read the study entitled ‘Percutaneous versus traditional and paraspinal posterior open approaches for treatment of thoracolumbar fractures without neurologic deficit: a meta-analysis.’ published in European Spine Journal in October 2016 [1]. In this article, the authors evaluated differences in outcome variables between percutaneous, traditional, and paraspinal posterior open approaches for traumatic thoracolumbar fractures without neurologic deficit. It is a very interesting study. Nevertheless, we have some queries and would like to communicate with the authors.
1.
The perfect literature search strategy was essential for a great meta-analysis. However, the authors just described the search strategy using the keywords, such as “thoracolumbar fractures”, “lumbar fractures”, “percutaneous”, “minimally invasive”, “open”, “traditional”, “posterior”, “conventional”, “pedicle screw”, “sextant”, and “clinical trial”, without the issue of the completeness of the search strategy report for databases. Therefore, we suggest that the authors provide us a complete search strategy to strengthen the credibility of the study.
2.
As described in statistical analysis, the random-effects model was used if p < 0.05 was presented in the analysis. In the meta-analysis, significant heterogeneity was observed in the paraspinal subgroup analysis of intraoperative fluoroscopy (p = 0.04), paraspinal subgroup analysis of postoperative visual analog score (VAS) with follow-up time less than 1 week (p = 0.0006) and postoperative VAS with follow-up time from 1 week to 3 months (p < 0.00001). However, the authors chose the fixed-effects model other than random-effects model which was inconsistent with the methods described in statistical analysis. And we recalculated the datum using random-effects model and obtained results which differed from the Sun et al. study (Fig. 1). Therefore, in view of the significant heterogeneity, we suggest that “fixed-effects model” should be replaced by “random-effects model”.
Fig. 1
Forest plot showing (a) paraspinal subgroup analysis of intraoperative fluoroscopy (b) paraspinal subgroup analysis of postoperative visual analog score (VAS) with follow-up time less than 1 week (c) postoperative VAS with follow-up time from 1 week to 3 months
×
3.
The assessment of publication bias was an important part in a meta-analysis. The authors declared that funnel plots were generated to determine publication bias. However, there were no relevant funnel plots in the results. As shown in Fig. 2, we generated the funnel plot of the analysis of postoperative VAS with follow-up time from 1 week to 3 months, and significant publication bias was observed. Obvious publication bias could heavily reduce the convincing of the results. Hence, we advise authors to add the funnel plots to complete this meta-analysis.
Fig. 2
Funnel plot of postoperative VAS with follow-up time from 1 week to 3 months
×
4.
In the meta-analysis, most included studies were carried out in China, therefore, the ethnic factor should be considered. We suggest that the authors should mention it in the conclusion.
Letter to the Editor concerning “Percutaneous versus traditional and paraspinal posterior open approaches for treatment of thoracolumbar fractures without neurologic deficit: a meta-analysis’’ by Sun, XY., Zhang, XN. & Hai, Y. Eur Spine J (2016). doi:10.1007/s00586-016-4818-4
