Effectiveness of Neural Mobilisation on Pain Intensity, Functional Status, and Physical Performance in Adults with Musculoskeletal Pain – A Systematic Review with Meta-Analysis

Objective To investigate up-to-date evidence of the effectiveness of neural mobilisation techniques compared with any type of comparator in improving pain, function, and physical performance in people with musculoskeletal pain. Data sources The following sources were consulted: PubMed, Web of Science, CENTRAL, CINAHL, Scopus, and PEDro databases; scientific repositories; and clinical trial registers. The last search was performed on 01/06/2023. Methods Two reviewers independently assessed the studies for inclusion. We included randomised, quasi-randomised, and crossover trials on musculoskeletal pain in which at least one group received neural mobilisation (alone or as part of multimodal interventions). Meta-analyses were performed where possible. The RoB 2 and the Grading of Recommendations Assessment, Development and Evaluation tools were used to assess risk of bias and to rate the certainty of evidence, respectively. Results Thirty-nine trials were identified. There was a significant effect favouring neural mobilisation for pain and function in people with low back pain, but not for flexibility. For neck pain, there was a significant effect favouring neural mobilisation as part of multimodal interventions for pain, but not for function and range of motion. Regarding other musculoskeletal conditions, it was not possible to conclude whether neural mobilisation is effective in improving pain and function. There was very low confidence for all effect estimates. Conclusions Neural mobilisation as part of multimodal interventions appears to have a positive effect on pain for patients with low back pain and neck pain and on function in people with low back pain. For the other musculoskeletal conditions, results are inconclusive.


⨁◯◯◯
Very low c ,i,j   Neural mobilization techniques appear to have no effect on flexibility but the evidence is very uncertain.
*The risk in the intervention group (and its 95% confidence interval) is bas ed on the as s umed ris k in the comparis on group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; SMD: s tandardis ed mean difference GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies clos e to that of the es timate of the effect.
Moderate certainty: we are moderately confident in the effect es timate: the true effect is likely to be clos e to the es timate of the effect, but there is a pos s ibility that it is s ubs tantially different.
Low certainty: our confidence in the effect es timate is limited: the true effect may be s ubs tantially different from the es timate of the effect.Very low certainty: we have very little confidence in the effect es timate: the true effect is likely to be s ubs tantially different from the es timate of effect.

Explanations
a. Eight of the ten RCTs were clas s ified at high ris k of bias ; the two qRCTs were clas s ified at critical ris k of bias .b.There is a high heterogeneity between s tudies (large variation effect, confidence intervals do not overlap, s tatis tical tes t for heterogeneity is p < 0.01, and the I2 is large).Even in the s ubgroup analys es performed, heterogeneity remains high and s ignificant among s tudies of the s ame s ubgroup.c.All s tudies were conducted in adults with low back pain, but the characteris tics of the condition were not homogeneous acros s s tudies .Some trials only included people with chronic pain, while others included people with acute or s ubacute low back pain.Regarding the intervention, the s tudies included NM as part of a multimodal intervention or as a s ingle intervention, as well as s liding or tens ioning techniques .Different co-interventions were included in each trial.With res pect to the outcome, they meas ured it with an appropriate meas uring ins trument.However, the comparis on interventions varied greatly between trials .Cons idering all thes e factors , it is not pos s ible to guarantee the generalizability, trans ferability, applicability, and external validity of the res ults .d. Sample s ize > 400 participants .e. Vis ual analys is us ing a funnel plot is inconclus ive regarding publication bias (Figure 1) and the res ults of the Egger and Begg tes ts did not reject the null hypothes is of s ymmetry, (t = -1.74,p-value = 0.11, intercept = 1.66; z = -1.92,p-value = 0.054, res pectively).The application of the Trim and Fill Method es timated two as ymmetric s tudies on the right s ide of the funnel plot (Figure 2), which adjus ted the global confidence interval for a nons ignificant effect (effect s ize = -0.61,95% confidence interval: -1.62; 0.39, z = -1.19,p-value = 0.23), s howing that the res ults found previous ly are not a robus t es timate of the real s ize and s ignificance of the effect.The p-curve analys is did not point to any s elective reporting on individual trials (Figure 3).f.Eight of the nine RCTs were clas s ified at high ris k of bias ; the two qRCTs were clas s ified at critical ris k of bias .g.There is a high heterogeneity between s tudies (large variation effect, confidence intervals do not overlap, s tatis tical tes t for heterogeneity is p < 0.01, and the I2 is large).However, in the s ubgroup analys is ives tigating multimodal intervention vs ers us s ingle intervention, no heterogeneity was identified for the s ingle intervention group.h.Vis ual analys is us ing a funnel plot is inconclus ive regarding publication bias (Figure 4) and the res ults of the Egger and Begg tes ts did not reject the null hypothes is of s ymmetry, (t = -0.89,p-value = 0.39, intercept = 0.48; z = -1.01,p-value = 0.31, res pectively).The application of the Trim and Fill Method es timated 1 as ymmetric s tudy on the right s ide of the funnel plot (Figure 5) which did not change the s ignificance of the res ults (effect s ize = -0.92,95% confidence interval: -1.69; -0.15, z = -2.34,p-value = 0.01), s howing that the res ults found previous ly s eem a robus t es timate of the real s ize and s ignificance of the effect.The p-curve analys is did not point to any s elective reporting on individual trials (Figure 6).i.Three of the four RCTs were clas s ified at high ris k of bias ; the qRCT was clas s ified at critical ris k of bias .j.There is a high heterogeneity between s tudies (large variation effect, confidence intervals do not overlap, s tatis tical tes t for heterogeneity is p < 0.01, and the I2 is large).

Figure 4 .
Figure 4. Funnel plot with Trim and Fill method (low back pain -functional status)