Design

This SR followed PRiSMA guidelines and was registered in the National institute for Health Research’s international Prospective Register of Systematic Reviews (PRoSPERo) under id CRd42023387399 [22, 23].

The research question was structured using the PiCo approach (population, intervention, comparison, and outcome). The study focused on patients with SAiS who underwent HiLT, compared to those receiving other physical therapy interventions, with or without sham HiLT. The primary outcome assessed pain intensity changes using tools like the visual analogue scale (VAS), the numerical scale (NPRS), or other validated scales. Changes in RoM (measured by goniometry) and disability (measured with established questionnaires like the disability of Arm, Shoulder, and Hand Questionnaire, the Shoulder Pain and disability index [dASH], or the Constant-Murley Shoulder outcome Score, [CMS]) were considered secondary outcomes.

Search strategy

Medline (via PubMed), Web of Science, Scopus, CiNAHL, Science direct, Cochrane Library, the Evidence-Based Physiotherapy database (PEdro), and Google Scholar databases were searched for HiLT in SAiS RCTs (updated to January 8, 2025). The search was conducted employing a curated set of keywords extracted from the MeSH (Medical Subject Headings) dictionary. These keywords encompassed “Lasers”, “Laser Therapy”, “Phototherapy”, “High-intensity Laser Therapy”, “Class iV laser”,“Musculoskeletal Pai”, “Shoulder Pain”, “Shoulder impingement Syndrome”, “Joint diseases”, and “Rotator Cuff Tear Arthropathy”. These terms were combined using boolean connectors “oR” and “ANd” to create the search algorithm: (((((“Lasers”) OR (“Laser Therapy”)) OR (“Phototherapy”)) OR (“High-Intensity Laser Therapy”)) OR (“Class IV laser”)) AND (((((“Musculoskeletal Pain”) OR (“Shoulder Pain”)) OR (“Shoulder Impingement Syndrome”)) OR (“Joint Diseases”)) OR (“Rotator Cuff Tear Arthropathy”)). Furthermore, filters for “Clinical Trial” and “Randomised Controlled Trial” were used to ensure the identification of clinical trials in the search results.

Selection criteria

Three independent researchers (AC-B, FP-A, and ES-o) collectively assessed article titles and abstracts from the databases using the Rayyan web tool (https://www.rayyan.ai/), making inclusion or exclusion decisions based on predefined criteria. The review used the following inclusion criteria: (A) human clinical trials with a SAiS diagnosis, (B) studies in English, Portuguese or Spanish, (C) HiLT alone or in combination with other physical therapy modalities, (d) comparisons with other physical therapy treatments with or without placebo, and (E) changes in pain intensity measured by different scales or instruments as the main outcome. Exclusions comprised literature reviews, systemic reviews of HiLT and SAiS linked to other musculoskeletal or neurological disorders, and studies with incomplete or inaccessible texts. No time limitation has been set for the search, considering the recent emergence of HiLT and the potential limitation in the quantity of available studies.

Risk of bias (RoB)

The RCTs’ methodological quality was initially assessed using the PEdro scale (iCC 0.53–0.91) for an initial overview [24]. Additionally, researchers used the Cochrane Collaboration risk of bias 2 tool (RoB 2.0) to evaluate bias based on investigator judgment [25]. Studies with PEdro scores above five were categorised as having good (6–8 points) or excellent methodological quality (9–10 points). Using the RoB tool, studies with two or more high RoBs were deemed low quality [25]. Concordance in RoB evaluation among researchers was assessed using the kappa statistic [26].

Statistical analysis

For statistical analysis, the Review Manager Software (RevMan 5.4) from the Cochrane Collaboration was utilised [27]. Heterogeneity across studies was evaluated using the chi-square ( ²) test and the I² statistic, categorised as follows: unimportant (0–40%), moderate (30–60%), substantial (50–90%), or considerable (75–100%). depending on the degree of heterogeneity observed, the researchers employed either the Mantel-Haenszel fixed-effects method or the derSimonian and Laird random-effects method to calculate the pooled effect using mean differences (Mds) for the outcomes of interest, along with their corresponding 95% confidence intervals (Ci) [28].

Evidence recommendation

The evaluation of the quality of evidence employed the GRAdE approach (Grading of Recommendations, Assessment, development, and Evaluation), considering criteria such as RoB, inconsistency, indirectness of evidence, imprecision, and publication bias [29]:

RoB. Highs are present if there are deficiencies in blinding or hidden allocation criteria, potentially leading to an overestimation of treatment effects.

Inconsistency. depends on heterogeneity equal to or exceeding 50% for an interesting outcome.

Indirect evidence. identified when the characteristics of treated individuals deviated from those of the broader population with the health condition or when comparing HiLT treatment to less common interventions in the available evidence or clinical practice.

Imprecision. depends on whether the Cis for the pooled effect in meta-analyses intersect the line of no effect, which suggests a lack of clarity in favouring one of the two treatment groups. Furthermore, the optimal sample size will be evaluated, requiring a representative participant number exceeding 200 to consider the effect as clinically relevant.

Publication bias. depends on the number of studies related to the relevant outcome, with a cutoff of at least three studies to mitigate bias effectively.

Evidence levels, ranging from high to very low, were assigned, with an initial high-quality rating for each level due to the exclusive inclusion of RCTs in this review [29]. Factors affecting one or two GRAdE domains may potentially lead to a downgrade of the evidence quality by one or two levels. The assessment of evidence importance will involve a comparison to ascertain whether statistically significant weighted Mds align with the literature’s definition of minimal clinically important differences (MCid).

To ensure a comprehensive synthesis of evidence regarding HiLT and its effects in SAiS, researchers used the GRAdEpro GdT tool for guideline development to facilitate a summary table of evidence for outcomes that demonstrated statistical significance in the meta-analysis (https://www.gradepro.org).

Search results

The initial database search generated a preliminary 3,174 articles (PubMed = 42; Scopus = 483; Web of Science = 108; CiNAHL = 201; Science direct = 2,087; Cochrane Central = 201; the PEdro database = 7; and Google Scholar = 15). Following the removal of duplicate articles, 29 documents remained for further analysis. However, ten articles were excluded as they pertained to studies on HiLT for frozen shoulder (n = 5), LLLT in SAiS (n = 2), two case reports (n = 2), and HiLT in post-stroke patients with hemiplegic shoulder pain (n = 1). This led to a final selection of 19 articles for analysis [17–21, 30–43]. Figure 1 illustrates the PRiSMA flowchart outlining the article search and selection process.

Figure 1

PRiSMA flowchart diagram

Appendix 1 provides a summary of the search strategy and results for each database.

Methodological quality and risk of bias

Methodological quality assessed by the PEdro scale indicated an overall good-to-excellent quality, averaging 7 points (Table 1) [24]. Major methodological gaps were noted in the concealment allocation and blinding of participants and assessors. However, criteria such as random participant assignment, pre-treatment group comparisons, participant follow-up, and intergroup treatment comparisons were generally fulfilled.

Figure 2 shows the RoB 2.0 assessment conducted by three investigators (HdB, ACB, and NSV) on the included studies. The RoB assessments resulted in a consensus among evaluators, with a kappa coefficient of 0.90. The primary high RoB was linked to measurements in the outcome data (57.7%) and bias due to deviations from the intended intervention (26.0%). Conversely, bias in the measure of outcome data (0%), bias in missing outcome data (89.4%), and bias in the randomisation process (15.7%) were deemed to have a low RoB. Considering the presence of these results, they yielded an overall risk of (31.6%).

Figure 2

Risk of bias summary. The review authors’ judgments about each risk of bias item for each included study

Characteristics of the RCTs

Table 1 provides a concise overview of essential details from the RCTs, encompassing study groups, selection criteria, interventions, assessments, and relevant outcomes. These trials were conducted across various countries, including italy [17], iran [18, 41], Turkey [19–21, 32, 39, 40, 43], india [30, 31], Egypt [33, 36, 37], Taiwan [35], Spain [34], and Romania [38], spanning from 2009 to 2023.

A total of 871 participants diagnosed with SAiS were included, with an average age of 46.9 years (Sd ± 8.8), comprised of 379 women and 322 men, and two studies lacked gender information. Among these, 449 received HiLT, while 422 controls underwent conventional physical therapy. in the experimental group (EG), 141 patients exclusively received HiLT [17, 19, 32, 35, 40], and 308 participants received HiLT alongside US [18, 39], therapeutic exercises [18, 20, 21, 30, 31, 33, 34, 36–38, 43], TENS [18, 20, 21, 39], thermotherapy [21, 39], Kinesio Tape [20,42], and interferential current [39]. Controls (GCs) were treated with US [17, 18, 30, 31, 38–41], therapeutic exercises [18, 20, 21, 30, 31, 33, 34, 36–38, 43], interferential current [39], radiofrequency [32], thermotherapy [21, 41], Kinesio Tape [20, 41], and LLLT [41]. Furthermore, two studies employed placebo HiLT [34, 41, 42].

HiLT treatments primarily targeted the deltoid, with the scanning technique being the predominant approach in most studies [17–21, 30–32, 34, 35, 38, 39]. Six studies utilised a combination of scanning and spot techniques [33, 36, 37, 41–43], while the spot technique alone was used in one RCT [40]. in 12 studies, 1064 nm Nd:YAG lasers were used [17, 19–21, 32–34, 36, 37, 39, 42, 43], while dual-wavelength devices (808–930 nm) were employed in 3 studies [35, 38, 41], and four studies used three different wavelengths within the infrared spectrum [18, 30, 31, 40]. The maximum power output ranged from 1.5 to 3000 W, with 8 and 12 W being the most common. The average energy delivered typically fell within the range of 2500–3000 J. The number of treatment sessions varied, ranging from 10 to 15 sessions over a span of 3 to 4 weeks. Additional HiLT parameters, such as pulse rate, phase duration, energy density, and treatment duration, are summarised in Table 1.

Outcomes

In these studies, pain intensity was assessed via VAS [17–21, 30–35, 37, 37–43], while the SPAdi was employed to evaluate shoulder pain [20, 21, 30, 32, 40, 41] and disability. Shoulder RoM was quantified using goniometry [20, 31, 33, 35–37, 39, 42, 43]. Also, pain pressure threshold [39,40], functional assessment with CMS [17, 34, 35, 38, 42, 43], quality of life with SF-36 [42] and NHP [32], tendon thickness with ultrasonography [36, 37], and muscle strength with dynamometry [39] or isokinetic assessment [42] were important outcomes.

Table 2 shows the results and comparisons between the studies and control groups. Both groups experienced a significant reduction in pain (p < 0.05) during the evaluation sessions [17–21, 30–43]. However, HiLT demonstrates a more pronounced and lasting pain reduction after treatment and during follow-up sessions. For RoM and disability, both groups exhibit statistically significant differences before and after treatment (p < 0.05). HiLT notably enhances RoM, but findings regarding disability are mixed, with conflicting results in certain studies [20, 21, 39].

Meta-analysis

Pain intensity

Ten studies were included in both the comprehensive meta-analysis and subgroup analyses to assess the effectiveness of HiLT compared to other treatment modalities regarding pain intensity. Pain intensity was measured using the VAS and SPAdi subscale. The pooled effect was determined utilising the random-effects model of derSimonian and Laird [28]. At the end of the treatment, both the VAS (Md = –1.36 cm; 95% Ci = –1.96, –0.75; p < 0.01) (Figure 3A) and the SPAdi subscale (Md = –4.9%; 95% Ci = –8.4, –1.4; p < 0.01, Figure 3B) showed a statistically significant reduction in pain intensity at rest in favour of HiLT. However, subgroup analyses revealed no significant differences when comparing HiLT versus placebo (Md = –0.24 cm; 95% Ci = –0.63, –0.16; p = 0.24) (Figure 3E), HiLT versus conventional physical therapy at one month (Md = –0.7 cm; 95% Ci = –2.5, 1.1; p = 0.05) (Figure 3C) and three-month follow-up (Md = –1.43 cm; 95% Ci = –3.13, –0.8; p = 0.28) (Figure 3d). The I² coefficient derived from the analyses indicated a substantial level of heterogeneity across the studies, except for pain intensity measured using the SPAdi subscale and the comparison between HiLT and placebo, where heterogeneity was not statistically significant. The quality of evidence was deemed significant for pain intensity assessed with VAS but had low certainty, and it was not considered significant with a low level for the SPAdi subscale (Table 3).

Figure 3

Forest plots for pain intensity at rest (VAS) at the end of treatment (3A), pain intensity at rest (SPAdi, 3B), pain intensity at follow-up (1 month, 3C), pain intensity at follow-up (3 months, 3d) and HiLT versus placebo (3E)

Table 3

Summary of findings and quality of evidence (GRAdE) for interesting outcomes

Certainty assessment							No. of patients		Effect		Certainty (level of evidence)	Importancee
no. of studies	study design	RoB	inconsistency	indirectness	imprecision	publication bias	HILT	conventional PT	relative	absolute
									(95% CI)	(95% CI)
Pain intensity at rest (VAS; 0 to 10 cm)
10	RCT	seriousa (-1)	seriousb I2 = 93% (-1)	not seriousc (0)	not seriousd (0)	no (0)	294	293		MD 1.36 cm lower	⊕⊕○○ low	important
										(1.96 lower to 0.75 lower)
Pain intensity at rest (SPADI; 0 to 100%)
3	RCT	seriousa (-1)	seriousb I2 = 62% (-1)	not seriousc (0)	not seriousd (0)	no (0)	100	131		MD 4.9 % lower	⊕⊕○○ low	not important
										(8.4 lower to 1.4 lower)
Shoulder abduction (goniometry; 0 to 180°)
8	RCT	seriousa (-1)	seriousb I2 = 99% (-1)	not seriousc (0)	not seriousd (0)	no (0)	169	200		MD 15.3° higher	⊕⊕○○ low	important
										(4.4 higher to 26.1 higher)
Shoulder flexion (goniometry; 0 to 180°)
7	RCT	seriousa (-1)	seriousb I2 = 97% (-1)	not seriousc (0)	not seriousd (0)	no (0)	139	170		MD 12.8° higher	⊕⊕○○ low	important
										(2.5 higher to 23.1 higher)
Functionality (CMS: 0 to 100 points)
4	RCT	seriousa (-1)	not seriousb I2 = 0% (0)	not seriousc (0)	not seriousd (0)	no (0)	110	113		MD 4 points higher	⊕⊕○○ moderate	not important
										(1.7 higher to 6.2 higher)

[i] Ci – confidence interval, CMS – Constant-Murley Scale, Md – mean difference, PT – physical therapy, SPAdi – Shoulder Pain and disability index, RCT – randomised controlled trial, VAS – visual analogue scale

[ii] a The high RoB was linked to the measurement of outcome data (57.7%) and bias due to deviations from the intended intervention (26%).

[iii] b The heterogeneity determines the inconsistency, depending on the I² statistic ( 50%).

[iv] c Considering a direct comparison of interventions and outcomes relevant to the study, with applicability to the clinical context, it was found that the indirect evidence held little significance. d imprecision was assessed by examining the width of the confidence intervals (Cis) for the pooled Md, the crossing of the no-effect line in the meta-analysis, and the sample size (n < 200).

[v] e importance was gauged based on whether the pooled effect, expressed as Md, met a value recognised in the literature as a clinically important change (MCid).

Shoulder ROM

A meta-analysis grouped data on shoulder RoM for abduction [20, 30, 31, 33, 36, 39, 42], flexion [20, 30, 31, 36, 37, 39, 42], external rotation [20, 30, 33, 39, 41], and internal rotation [30, 33, 39, 42]. The results revealed statistically significant changes in RoM for abduction (Md = 15.3°; 95% Ci = 4.4, 26.1; p < 0.01, Figure 4A) and shoulder flexion (Md = 12.8°; 95% Ci = 2.5, 23.1; p < 0.01, Figure 4B) after the treatment. However, no significant differences were observed between the groups in terms of external rotation movements (Md = 3.8°; 95% Ci = –5.1, 12.6; p = 0.4, Figure 4C) and internal rotation (Md = 2.9°; 95% Ci = –2.5,0.37; p = 0.29, Figure 4d) at the end of the treatment. All analyses exhibited a considerable I2 index [29]. The evaluation of the evidence suggests important changes for shoulder abduction and flexion, although with a low level of certainty (Table 3). For the other CRoMs, researchers assessed the evidence as not important due to the non-statistically significant changes [29].

Figure 4

Forest plots for shoulder abduction (4A), shoulder flexion (4B), shoulder external (4C), internal rotation (4d), disability with SPAdi and dASH (4E and 4F), and functionality with CMS (4G) at the end of treatment

HiLT and shoulder pain

This SR shows the effectiveness of HiLT in reducing pain among SAiS patients. HiLT exhibits a significant reduction of –1.4 cm (95% Ci: –2.0, –0.8) when compared to other treatments, including isolated exercise [30, 31, 33, 36, 37, 42], exercise combined with phonophoresis [37], ultrasound [17, 18, 30], thermotherapy [21], TENS [21], and HiLT placebo [34]. Notably, this pain reduction surpasses the CMid threshold (CMid = –1.4 cm) established for individuals with rotator cuff injuries [44]. Nonetheless, while clinically meaningful reductions in pain intensity were noted even at 1- and 3-month follow-ups, there were no statistically significant differences compared to conventional physical therapy interventions. This implies that HiLT is effective in the short and medium term, but other treatments may produce similar effects, albeit with a delay. These findings regarding HiLT’s effects on pain intensity with VAS align with prior research on LLLT in shoulder tendinopathies. These studies have found that laser therapy reduces pain by –1.3 cm (95% Ci: –1.7 to –2.4) and –2.4 cm (95% Ci: –1.3 to –2.8) compared to other physical or exercise therapy methods and sham LLLT, respectively [11]. This suggests similar analgesic effects and supports the idea that these lasers share a common physical mechanism of action. However, more research is needed to directly compare the effectiveness of HiLT and LLLT, as no study in the review conducted such a comparison. These findings endorse laser therapy’s efficacy in alleviating shoulder pain, resulting in an important GRAdE rating despite some uncertainty due to bias and variability. Currently, HiLT is regarded as a clinically equivalent alternative to LLLT, making the choice between these approaches contingent on resource availability, cost considerations, or individual patient preferences.

The VAS outcomes align with those from the SPAdi sub-scale, where HiLT exhibits a greater pain reduction compared to the control treatment, with a difference of –5.0% (95% Ci: –8.4, –1.39). it’s worth noting that this difference falls below the CMid threshold of 14 to 20% that is established for the SPAdi [45]. it should be considered that only three studies utilised the SPAdi subscale [20, 21, 30], rendering the VAS results more reliable. Notably, no significant discrepancies in pain intensity at treatment completion emerged in the HiLT versus placebo comparison. Notably, this analysis hinged on just two studies [34, 42], highlighting the need for more extensive research for definitive outcomes. in patients with SAiS, an important clinical metric is the Patient Acceptable Symptom Status (PASS) Assessment, which employs VAS scores to identify the highest symptom level deemed acceptable by patients. An average PASS of 3 cm has been established as acceptable (95% Ci: 2.3, 3.7). While not directly assessed in the referenced RCTs, the pain intensity at the end of treatment aligns with the acceptable PASS for several RCTs [17, 18, 30, 31, 37]. in SAiS rehabilitation, exercise integration is crucial due to the central role of muscle weakness [46, 47]. Several RCTs in this study integrated exercises within HiLT protocols. Given that there is a strong link between SAiS and rotator cuff and postural muscle weakness, physical therapy is very important for three very important reasons: managing symptoms (using HiLT or other methods), strengthening muscles, and postural re-education [46].

HiLT and RoM

The results show that HiLT is better than traditional physical therapy at increasing the RoM of the shoulder in flexion and abduction by an average of 15.3° (95% Ci: 4.4–26.1) and 12.9° (95% Ci: 2.5–23.1), respectively. These improvements are greater than the minimal detectable change (MdC), which is 8° for flexion and 4° for abduction [48], and align with the MCid reported, ranging from 11° to 16° [49]. increased shoulder flexion and abduction influence scapular plane arm lifting, which is limited in SAiS patients [1, 5]. This movement improves movement patterns, strengthens the rotator cuff, and prevents injuries [50].

The thermal and analgesic benefits of HiLT increase RoM. Heating the tissues relaxes muscles and interrupts the cycle of painful muscle spasms [51, 52]. At the same time, pain reduction diminishes abnormal afferent information that affects movement performance [53]. Since it was applied directly to the deltoid, which controls abduction and flexion, the HiLT likely had a greater effect on RoM. importantly, focused HiLT administration to specific rotator muscles may improve shoulder rotation. The limited number of RCTs on the rotation RoM may make it difficult to draw conclusions.

Although HiLT can enhance movement, it is recommended to combine it with more targeted interventions, such as exercises or manual therapy, for optimal results [54].

HiLT and disability

According to SPAdi and CMS, HiLT is effective in reducing disability by –6.6% (95% Ci: –12.5, –0.6) and enhancing functionality by 4 points (95% Ci: 1.7, 6.2), respectively. in patients with rotator cuff injuries, the SPAdi values for disability are lower than the MCid (8 to 13 points, or 10 to 16%), while the CMS values for functionality are concordant with the MCid (2 to 16 points) [55, 56]. disability has gained increasing importance in clinical trials. This outcome is crucial, as treatments should not only evaluate the influence of symptoms on functionality, a relevant aspect for patients [57]. Although HiLT is primarily intended to alleviate pain, its secondary effect on disability is significant, as pain frequently induces a dread of moving that restricts functionality. Furthermore, this dread of movement frequently contributes to the condition’s chronicity [58–60].