High-Intensity Laser Therapy for Musculoskeletal Disorders: A Systematic Review and Meta-Analysis of Randomized Clinical Trials

High-intensity laser therapy (HILT) is one of the therapeutic approaches used in the treatment of musculoskeletal disorders (MSD). The main objective of this study was to examine the effectiveness of HILT for reducing pain and improving functionality in people with MSD. Ten databases were systematically searched for randomized trials published up to 28 February 2022. Randomized clinical trials (RCTs) assessing the effectiveness of HILT on MSD were included. The main outcome measures were pain and functionality. In total, 48 RCTs were included in the qualitative synthesis and 44 RCTs in the quantitative analysis. HILT showed a decrease on the pain VAS (mean difference (MD) = −1.3 cm; confidence interval (CI) 95%: −1.6 to −1.0) and an improvement in functionality (standardized mean difference (SMD) = −1.0; CI95%: −1.4 to −0.7), with low and moderate quality of evidence, respectively. A greater effect was observed when compared with control than with other conservative treatments, both on pain (χ2 = 20.6; p < 0.001) and functionality (χ2 = 5.1; p = 0.02). Differences in the effectiveness of HILT were found depending on the location (χ2 = 40.1 p < 0.001), with further improved functionality in MSD of the knee and shoulder. HILT is an effective treatment for improving pain, functionality, range of motion, and quality of life in people with MSD, although these findings must be treated with caution due to the high risk of bias in the studies. Further clinical trials should be well designed to lower the risk of bias.


Introduction
The prevalence of musculoskeletal disorders approaches 36% in older adults in Europe [1]. This figure increases to almost 45% in the working-age Spanish population [2], making this condition an economical and public health issue [3]. Laser therapy is a standard treatment in clinical practice for the treatment of musculoskeletal disorders [4]. Although low-power laser (Class III) therapy has been employed for more than two decades [5], high-power laser (Class IV) therapy has been implemented in the clinical setting in the last years due to its greater depth of penetration and the possibility of delivering higher doses with lower exposure times [6]. Over the last few years, clinical trials have aimed at assessing the effectiveness of this new therapy for patients with musculoskeletal disorders.
One systematic review [7] and one systematic review with meta-analysis [8] for the evaluation of high-intensity laser therapy (HILT) in musculoskeletal disorders have been published, which included trials conducted up to 17 January 2018. Of the studies found, only 12 randomized trials met the criteria for inclusion, with 6 comparing HILT versus sham stimulation and 6 versus other treatments. The large number of clinical trials published after the end of the search period of this meta-analysis, the possibility to analyze the effectiveness of HILT on other outcome variables, and the lack of an analysis of effectiveness depending on the dosage call for an update of the level of evidence of HILT for patients with musculoskeletal disorders.

Data Extraction
Two researchers (JAG and RAF) performed the data extraction by using a chart specifically designed for this purpose that they agreed upon. A third researcher (JAC) compared both charts and presented the final data collection. The main outcome variables were pain and functionality. Pain was measured as the subjective perception of the patient reported on a visual analogue scale (VAS). Functionality was assessed from values recorded in specific scales for each region. When a trial employed several scales for measuring functionality, data were extracted only from one based on an order already stated in our protocol, which had been previously registered in PROSPERO. Additionally, adverse effects reported in the studies were recorded. The secondary variables were the ROM (degrees), quality of life based on the SF-36 scale, and muscle strength measured via dynamometry. Authors of the selected studies were contacted to obtain or clarify missing or unclear data if needed. Data available only in graphs were extracted using Graph Grabber 2.0.2 software for graph digitalization (https://www.quintessa.org/software/ (accessed on 19 September 2022)).

Assessment of Risk of Bias
The risk of bias was assessed based on recommendations by the Cochrane organization [10] using Review Manager (RevMan) (Computer program. Version 5.3. Copenhagen: The Nordic Cochrane Centre, the Cochrane Collaboration, 2014). Two independent reviewers (RAF and JAG) evaluated the risk of bias and a third investigator (DSM) resolved cases of disagreement. Seven items were addressed for evaluating the risk of bias, and the relevant risk was expressed in three levels (unclear, low, and high). Previously, the researchers had agreed on the following: for the item "blinding of participants", the risk would be qualified as unclear when participants were not blinded in all groups in studies with more than two arms; and for the item "selective reporting", studies without a registered protocol would be qualified as unclear or high risk depending on the final report. Finally, funnel plots for the main variable (pain on a VAS) were analyzed to evaluate publication bias.

Data Synthesis and Analysis
The inverse variance method was used for analyzing all variables (pain, functionality, ROM, strength, and quality of life). Statistical heterogeneity was evaluated using the Chisquared test (with statistical significance set at p < 0.10), and heterogeneity was measured calculating I2, with 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively [10]. Random effect and fixed effect analysis models were used when the heterogeneity I 2 was greater or lower than 50%, respectively. The mean difference (MD) was obtained for the pain VAS, strength, and quality of life outcomes, which were expressed on the same scale in the included studies. Average values of the VAS pain were calculated whenever this outcome was reported for different situations (at rest, during movement, etcetera). In terms of quality of life assessed via the SF-36 scale, the results were evaluated for each of its 8 items since the calculation of an overall score for this scale has not been validated [11]. The standardized mean difference (SMD) was estimated for the assessment of functionality and ROM, since their measurements vary depending on the different employed scales or different units (degrees and mm), respectively. For functionality scales where a higher score means less disability (i.e., Foot and Ankle Outcome Score and Constant Murley Score scales), this value was multiplied by −1 in order to align the direction of the effect. For the ROM variable, an analysis of the sum of all the movements (degrees) of the assessed joints was performed. Only the passive ROM was included from those studies reporting both the passive and active ROM. Confidence intervals were set at 95% (CI95%) for all variables. The analyzed results were those with the longest follow-up period for each of the included studies.
In the case of studies with more than two arms, splitting of the shared group was applied according to the Cochrane Group Guidelines [10] to avoid double count. In addition to the global analysis, an analysis by subgroups was conducted for all variables to account for the comparator (control/other therapy). For those studies comparing HILT versus control, an analysis of the main variables (pain and functionality) was conducted by subgroups to account for the follow-up period, dosage, and anatomical location of musculoskeletal pain. In the case of studies comparing HILT versus other treatments, an analysis by subgroups was performed to account for the other therapy. The RevMan 5.4.1 software was used for the quantitative analysis. The quality of evidence was classified for each outcome as high, moderate, low, or very low following the Grades of Recommendation Assessment, Development, and Evaluation (GRADE) method [12].
In the included RCTs, the treatment alternatives with which HILT was compared were: low-level laser therapy (LLLT) [
adverse effect in a participant consisting of an allergy to HILT [60].

Risk of Bias in the Included Studies
The two researchers that assessed the risk of bias (JAG and RAF) agreed upon 82% of the items, and disagreements were resolved by a third researcher (DSM). Figure 2 shows the summary of risk of bias for the 48 included studies. The majority of studies showed high performance bias, mainly resulting from the inability to blind the researcher and participants whenever HILT was compared with other therapy or no intervention. Low risk of bias was only found in 4 RCTs [16,36,42,56] (8.3% of included studies) for the item "blinding of personnel" and in 14 RCTs [14,16,21,23,24,25,35,36,42,53,56,58,59] (29.2%) for the item "blinding of participants". On the contrary, 31 RCTs [13,14,16,20,21,22,23,24,26,27,28,30,32,33,35,39,41,43,44,45,46,47,48,50,51,52,53,54,55,58,60] (64.6% of included studies) were rated as having a low risk of bias regarding the blinding of outcome assessment (detection bias). The items with the lowest risk of bias were those regarding the randomization generation and incomplete results, where high risk was observed in only three [14,51,59] and two [40,48] RCTs, respectively. Finally, the reporting bias was classified as unclear for the majority of included trials for not having previously registered their relevant protocol. The study by Naruseviciute et al. [40] was also classified as high risk for this item since their former protocol incorporated secondary variables for measuring functionality and quality of life that the final report did not include. The trial by Pekyavas et al. [48] was also rated with high risk for not reporting the VAS pain outcomes despite describing its estimation in their methods. The works by Angelova et al. [59] and Boyraz et al. [15], which were excluded from the pooled quantitative analysis due to insufficient data, were classified as having a high risk of bias due to being rated with low risk in only one and none of the assessed items, respectively. Conversely, all items were rated as low risk in the study by Cantero-Tellez et al. [16], and so were all the items but one in the trials by Nouri et al. [42] and Yesil et al. [53]. Twenty-three RCTs [14,16,21,22,23,24,27,28,30,33,35,36,39,41,42,43,45,50,53,55,56,58,60] (46.2%) presented moderate risk of bias with 4-5 items rated as low risk (see Supplementary Appendix S3 for the risk of bias for each included study). The risk of publication bias was considered The items with the lowest risk of bias were those regarding the randomization generation and incomplete results, where high risk was observed in only three [14,51,59] and two [40,48] RCTs, respectively. Finally, the reporting bias was classified as unclear for the majority of included trials for not having previously registered their relevant protocol. The study by Naruseviciute et al. [40] was also classified as high risk for this item since their former protocol incorporated secondary variables for measuring functionality and quality of life that the final report did not include. The trial by Pekyavas et al. [48] was also rated with high risk for not reporting the VAS pain outcomes despite describing its estimation in their methods. The works by Angelova et al. [59] and Boyraz et al. [15], which were excluded from the pooled quantitative analysis due to insufficient data, were classified as having a high risk of bias due to being rated with low risk in only one and none of the assessed items, respectively. Conversely, all items were rated as low risk in the study by Cantero-Tellez et al. [16], and so were all the items but one in the trials by Nouri et al. [42] and Yesil et al. [53]. Twenty-three RCTs [14,16,[21][22][23][24]27,28,30,33,35,36,39,[41][42][43]45,50,53,55,56,58,60] (46.2%) presented moderate risk of bias with 4-5 items rated as low risk (see Supplementary Appendix S3 for the risk of bias for each included study). The risk of publication bias was considered low since the distribution of the main variable (VAS pain) in a funnel plot did not show asymmetries (Figure 3). low since the distribution of the main variable (VAS pain) in a funnel plot did not show asymmetries (Figure 3).

Effect on Functionality
When comparing HILT versus control/sham control groups and other conservative interventions, the overall effect on functionality was superior with HILT (SMD = −1.0; CI95%: −1.4 to −0.7) with a high level of heterogeneity (I2 = 92%; p < 0.001) ( Figure 5). The quality of evidence for this outcome according to GRADE was moderate in terms of factors related to rating down (serious risk of bias and serious inconsistency or heterogeneity) and the factor related to rating up (magnitude of effect). In the analysis by subgroups, the difference between HILT and control/sham control (SMD = −1.5; CI95%: −2.0 to −0.9) was significantly greater (χ2 = 5.1; p = 0.02) than that observed when comparing HILT to other conservative treatments (SMD = −0.7; CI95%: −1.1 to −0.2). However, the level of heterogeneity was high (I2 = 92% and 93%, respectively; p < 0.001) in both subgroups ( Figure 5).
No significant differences were found between HILT and control/sham control groups depending on the follow-up period (post-immediate, 3-9 weeks, and 10-24 weeks), but the effect tended to be greater in the long term (Table 1 and Supplementary Appendix S3). When four dosages were compared (≤10 J/cm 2 , 10-50 J/cm 2 , 50-100 J/cm 2 , 100-300 J/cm 2 ), the greatest effect on functionality was achieved with 10-50 J/cm 2 , whereas 50-100 J/cm 2 did not produce a significant effect. However, differences between these four subgroups did not reach statistical significance (Table 1 and Supplementary Appendix S3). Significant differences (p < 0.001) were observed in the comparison by subgroups in terms of the anatomical location of musculoskeletal pain, where the greatest effect on functionality was recorded at the knee and shoulder, and with no changes in functionality at the foot and temporomandibular pain (Table 1 and Supplementary Appendix S3).

Effect on Secondary Variables: Range of Motion (ROM), Strength, and Quality of Life
Overall, HILT was effective in improving the ROM (SMD = 1.1; CI95%: 0.6 to 1.7). The quality of evidence for the ROM variable according to GRADE was moderate in terms of factors related to rating down (serious risk of bias and serious inconsistency or heterogeneity) and the factor related to rating up (magnitude of effect). A significant difference in the ROM outcome was observed in the subgroup analysis depending on the comparator (control/other treatments) (χ2 = 6,9; p < 0.01). When comparing HILT versus control, an increased in ROM outcome was observed in favor of HILT (SMD = 1.7; CI95%: 1.1 to 2.4), but no differences were found when comparing with other treatments (SMD = 0.2; CI95%: −0.7 to 1.1) ( Figure 6A). No significant differences were recorded in the overall effect on muscle strength (MD = 2.0 kg; CI95%: −0.3 to 4.4) or in the analysis by subgroups depending on the comparator (control/other treatments) ( Figure 6B).
The quality of evidence for the strength variable according to GRADE was moderate in terms of the factor related to rating down (serious risk of bias). HILT was more effective for improving quality of life than control, especially in the four items of the SF-36 questionnaire related to physical health (physical functioning, role physical, bodily pain, and general health) and in two items related to mental health (role of emotional and social functioning), with no differences in the two remaining items related to mental health (vitality and mental health) (Figure 7). The quality of evidence for the quality-of-life variable according to GRADE was moderate in terms of the factor related to rating down (serious risk of bias).   Of note, the previously registered protocol of this meta-analysis included a pooled quantitative analysis of adverse effects that in the end could not be conducted because only one adverse event was reported, consisting of allergy to HILT [60], and no adverse events were noted in control groups or others receiving conservative treatments.

Discussion
This meta-analysis showed that HILT is an effective treatment for improving pain and functionality in musculoskeletal disorders with low and moderate recommendation levels according to GRADE, respectively. The improvement in these variables was greater when comparing HILT versus control or sham than versus other conservative treatments. Some studies determined that a change of 1.4-2.0 cm on a musculoskeletal pain VAS can be considered clinically significant [61,62]. The difference of 1.9 cm on the pain VAS observed between HILT and control treatments can be considered clinically relevant, unlike the 0.7 cm difference observed between HILT and other conservative therapies. The standardized mean difference (SMD) is used to measure the magnitude of the effect, and an SMD of ≥0.8 is considered to represent a large effect [63]. The magnitude of the effect on functionality observed between HILT and control treatments can be considered large (SMD = 1.5), unlike the effect observed between HILT and other conservative therapies (SMD = 0.7). However, the improvement in functionality was large using HILT when comparing to shock wave therapy. This could be since the dose used in shock wave therapy (0.05 mJ/mm 2 ) used in the included study was below the recommended dose [64].
The effects on pain and functionality of HILT versus control (MD = 1.9 cm on the pain VAS and SMD = 1.5 on functionality) observed in this meta-analysis were greater than those in a former meta-analysis by Song et al. [8] for musculoskeletal pain, which reported a pain reduction of MD= 1.0 cm on the VAS and SMD = 1.0 for the effect of HILT on functionality. Additionally, Song et al. [8] did not find differences in functionality when contrasting HILT against other conservative treatments. Of note, the number of included RCTs (n = 44) in the present meta-analysis was substantially greater than the RCTs (n = 12) in the meta-analysis by Song et al. [8].
In the analysis by subgroups for the follow-up period, an opposite tendency in the effect of the treatment was observed between the main variables (pain and functionality) when comparing HILT versus control. The effect of HILT on functionality showed a tendency to improve over time, whereas the improvement in pain tended to decrease, although statistical difference was not reached. These results are in agreement with the former meta-analysis by Song et al. [8] and another meta-analysis that assessed the effect of HILT on knee osteoarthritis [65].
In the analysis by subgroups to account for the dosage, lower doses (≤10 J/cm 2 ) tended to achieve greater analgesic effects, although statistical difference was not reached. A study by Ezzati et al. [28] in patients with carpal tunnel syndrome also observed this effect, where after delivering two doses of HILT (8 J/cm 2 versus 20 J/cm 2 ), the analgesic effect of HILT was greater with 8 J/cm 2 . Along these lines, a preclinical study determined that a dose of 8 J/cm 2 yielded the greatest antinociception, possibly due to the activation of the endogenous opioid system [66]. However, medium doses (>10 J/cm 2 to 50 J/cm 2 ) tended to achieve greater effect on functionality. Additionally, the present meta-analysis did not observe significant effects on functionality employing high doses of 50 and 100 J/cm 2 . Nevertheless, more comparative trials delivering diverse dosages of HILT are necessary to determine the potential influence of this parameter on the therapy effectiveness.
In the analysis by subgroups depending on the location of pain, effect on pain by location was close to reaching statistical significance, with a clinically significant improvement in pain (1.6 cm to 3 cm on the VAS) for all anatomical locations except for the foot [53]. Two former meta-analyses assessing the effectiveness of HILT for treating two specific musculoskeletal disorders, knee osteoarthritis [65] and spinal disorders [67], which included six and nine RCTs, respectively, reported a lesser effect on pain reduction than the present meta-analysis. In terms of the effect on functionality, this meta-analysis found significant differences depending on the location, with the greatest effect observed at the knee and shoulder and without effect in temporomandibular and foot locations. Song et al. [8] reported the greatest effect at the neck and lower back, although no trial about the knee was included. The functionality results of this meta-analysis were very similar to those obtained by Alayat et al. in a previous meta-analysis about spinal disorders [67] and inferior to those observed in a former meta-analysis about knee osteoarthritis [65].
HILT could be considered a safe technique due to the absence of adverse effects reported by the authors and the similar number of abandonments in the experimental and control groups. However, 62.5% of the included trials did not specify these data. The previously published systematic reviews and meta-analyses [8,65,67] on HILT did not assess complications or adverse effects. The reported adverse effects of LLLT do not differ from those stated by patients exposed to placebo devices in trials [5]. However, the higher power that HILT employs makes it necessary to conduct further research where adverse effects or complications are systematically assessed. The analysis of secondary variables showed the effectiveness of HILT for improving both the ROM and quality of life with a "moderate" recommendation level according to GRADE, in contrast to muscle strength, which did not improve with the intervention and had a "moderate" recommendation level according to GRADE. To our knowledge, this is the first meta-analysis that assessed the effect of HILT on these variables.

Study Limitations
An important limitation of this review is the large heterogeneity in the results of the assessed variables, which lowered the certainty of evidence. Strength and some domains of quality of life as assessed via the SF-36 were the only variables with moderate or low heterogeneity. The previous meta-analysis by Song et al. [8] also reported this high level of heterogeneity. Additionally, the analysis by subgroups did not reduce heterogeneity. Although determining the factors accounting for this high heterogeneity or inconsistency of the results was not possible, it could stem from the large variability in demographic characteristics of the samples, with ages ranging from 28 to 64 years, as well as from their clinical characteristics, such as the diverse pathologies that included etiology, duration, and severity of symptoms. Additionally, protocols for applying HILT were also heterogeneous. Another important limitation was the high risk of bias of the included trials, mainly regarding the blinding of participants and the researcher delivering the interventions. Hence, these results must be regarded with caution since the effect of HILT could be overestimated due to the large placebo effect that high-technology medical devices can produce [68].

Conclusions
This updated meta-analysis supports the effectiveness of HILT in improving pain, functionality, ROM, and quality of life in patients with musculoskeletal pain, without side effects. However, the certainty of this evidence was classified as "low" and "moderate". The effectiveness of HILT varied depending on the location of the musculoskeletal disorder, and the greatest effect on pain was observed at the shoulder and knee. Future clinical research and reviews should be designed with a lower risk of bias in order to improve the certainty of this evidence.

Data Availability Statement:
The data presented in this study are available on reasonable request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.