Eligible studies
Our search strategy yielded 1130 studies from EMBASE (n=580), PubMed (n=331), and Cochrane Central Register of Controlled Trials (n=219). Only sixteen RCTs were included.15,21,25,33-45 Three of 16 studies included follow-up of less than 1 year,15,38,39 five were duplicate reports of the same set of patients.40-44 8 RCTs with 1391 patients were included in the meta-analysis. There were 997, 248, and 558 patients allocated to OMD, TMD, and TED, respectively (Figure 1). Characteristics of 8 studies were shown (Table 1). 21,25,33-37,45
Risk of Bias of Individual Studies
Figure 6-7 shows the graph of risk of bias. In the included studies, 7 studies indicated “randomly allocating”. Only 3 trials described methods of randomization. All trials doesn’t mentioned it use the blind method, because no studies could perform blinding. 6 studies found no significant other bias.
Pooled Weighted Outcomes and Direct Meta-Analysis
Compared with OMD, TMD is superior in ODI (SMD=-3.43 95% CrI -4.64 to -2.21, Figure 2), and there is no statistical difference in other aspects (Figure 3-5). Compared with OMD, the TED was associated with higher pain relief (SMD=-1.10 95% CrI -1.85 to -0.34, Figure 3), functional outcome (SMD=-5.17 95% CrI -8.04 to -2.31, Figure 2), and shorter hospital stays (SMD=-0.70 95% CrI -1.05 to -0.35). There is no statistical difference in other outcomes (Figure 4-5).
Network Meta-Analyses and Rank Probabilities
We established a network for comparisons of OMD, TMD, and TED. According to our results, OMD, TMD, and TED showed no statistical difference in terms of VAS, ODI, complications, reoperation, operation time, and duration of hospital stay (Table 2A-F).
Using a ranking system that indicates the best treatment probability, we developed a histogram figure (Supplemental Figure S2) to show the probability to be the best, the second-best, the third best among 3 treatments. It was found that agents with greater value in the histogram had greater probabilities for better outcomes. The cumulative probabilities of TED (95%), TMD (4%), and OMD (1%) being associated with the highest level of pain relief came from the network (Figure S2A). TED (77%), TMD (22%), and OMD(1%) were associated with the highest cumulative probabilities of disability prevention (Figure S2B). In other words, the most likely clinical efficacy was found with TED, followed by TMD, and finally OMD. TMD (54%), TED (23%), and OMD (23%) are associated with the lowest cumulative probability of reoperation ( Figure S2C). According to the cumulative probabilities, TED (58%), TMD (36%), and OMD (6%) are associated with the lowest complication rates (Figure S2D). That is to say, TED was the most likely to have the highest safety, followed by TMD, and then OMD. TED (56.3%), TMD (39.3%), and OMD (4.4%) are associated with the least cumulative probability of operation time. TED (87.7%), TMD (10.3%), and OMD (2%) are associated with the shortest cumulative probability of duration of hospital stay. By using the 12 criteria recommended by the Cochrane Back Review Group, we assessed bias risk in all original studies.29 Figure 7-8 summarizes the ratings from all included studies.
An analysis of the funnel plot (Supplemental Figure S3) revealed no apparent publication bias. As our study had a limited sample size, no sensitivity or scenario analysis was conducted.
Quality Assessment
All pooled outcomes were evaluated using GRADE quality assessment. Results are shown in Table 3.