To the editor,

The meta-analysis by Li et al. [1] published in the journal evaluated randomized controlled trials (RCTs) comparing minimally invasive versus open approaches to transforaminal lumbar interbody fusion (TLIF). No known meta-analysis on this topic has included only results from RCTs, and therefore, this would represent the highest level of evidence available. Yet upon review of this article, there are serious methodological issues that nullify the conclusions.

The authors state that 7 RCTs were included in this review. Yet only 2 of the 7 studies were actually RCTs—the studies of Serban et al. [2] and Wang et al. [3] The remaining 5 studies utilized prospective or retrospective nonrandomized comparisons [4,5,6,7,8]. Among the nonrandomized studies that were incorrectly included, Table 1 includes samples of text that clearly identifies each as a nonrandomized study. This is especially concerning since the authors reported that duplicate verification of study eligibility was performed. Further, in Figs. 2 and 3 (Li et al. [1]) in their meta-analysis, they state that 5 of 7 studies had low risk of bias as it relates to random sequence generation. Clearly, the authors have made egregious errors in the classification of the included studies.

Table 1 Listing of nonrandomized studies described as randomized controlled trials and locations of study design descriptions within each manuscript
Full size table

In Fig. 5 (Li et al. [1]) of their meta-analysis, another major error presents itself. It is noted in the forest plot that hospital stay was 2.2 days longer with minimally invasive TLIF in the Serban study [2]. Yet, hospital stay was actually 2.2 days shorter with minimally invasive TLIF in this study. This additional error fundamentally impacts the meta-analysis results since reporting of correct data would have led to a different conclusion—that minimally invasive TLIF is associated with a shorter hospital stay relative to open TLIF.

Given the major and obvious flaws in study selection and data analysis and consequently, the potential risk of other less obvious deficiencies, the authors and editors are encouraged to retract this article.