Type of study and design

This quantitative SR employed an observational, retrospective, and secondary design. The review adhered to the guidelines outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) [19]. Additionally, the SR was registered at the International Prospective Register of Systematic Reviews (PROSPERO) by the National Institute for Health Research with the registration number CRD42024530324 (registration date of March 28^th 2024) [20].

Selection criteria

The study adhered to the PICOS approach (patient, intervention, comparison, outcome, and type of studies), focusing on QT patients who received LLLT alone or in addition to other physical therapy treatments that were compared with other physical therapy interventions, medical treatments, or placebos. The primary outcome measure was pain intensity assessment using scales such as the visual analogue scale (VAS) or the numeric pain rating scale (NPRS). Secondary outcomes encompassed alterations in grip strength and disability, evaluated through scales or questionnaires such as the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire. The studies analysed included randomised and nonrandomised clinical trials in English or Spanish.

The exclusion criteria comprised case studies, literature reviews, SRs on LLLT unrelated to this study, research involving individuals with QT coexisting with other musculoskeletal disorders or neurological conditions, and studies characterised by incomplete or inaccessible data.

To ascertain eligibility, two researchers (TCH and MJG)) independently evaluated potential studies by scrutinising titles, abstracts, and full texts. A third researcher (HDB) resolved any selection disparities.

Literature review

Pubmed, Scopus, Web of Science (WoS), EBSCOhost, Embase, Cochrane, and the Physiotherapy Evidence Database (PEDro) databases. Additionally, the grey literature was explored through the Google Scholar search engine [21].

Keywords derived from the Medical Subject Headings (MeSH) dictionary were used, including “Lasers,” “Laser Therapy, ” “Phototherapy, ” “Low-Level Light Therapy, ” “Laser Class IIIb, ” “Musculoskeletal Pain, ” “De Quervain Disease, ” “Tendi- nopathy,”and “Tenosynovitis.” The terms were combined using Boolean search syntax to construct the search algorithm: ((“Lasers”OR “Laser Therapy”OR “Phototherapy”OR “Low-Level Light Therapy” OR “Laser Class IIIb”) AND (“Musculoskeletal Pain” OR “De Quervain Disease” OR “Tendinop- athy” OR “Tenosynovitis”)).

Data extraction

Three researchers used the Rayyan web tool for the initial screening of article titles and abstracts across multiple databases during the data extraction phase [22], with searches for each database retrieved in nbib, ris, or ciw formats. The three researchers independently evaluated the selected studies and transferred the data to a standardised Microsoft Excel form. Titles and abstracts of potential RCTs or non- RCTs were carefully examined for inclusion. Data extracted from the included reviews encompassed author details, country of origin, publication year, participants, number of groups, selection criteria, treatments, relevant outcomes, assessment instances, conclusions, and funding sources.

Risk of bias and methodological quality

The methodological quality of RCTs was initially assessed using the PEDro scale [23]. The Cochrane Collaboration’s Risk of Bias 2 (RoB2) tool was applied to analyse bias across six domains within the included studies [24]: (a) bias arising from the randomisation process; (b) bias due to deviations from the intended intervention; (c) bias due to missing outcome data; (d) bias in outcome measurement; (e) bias in the selection of the reported result; and (f) overall bias. Studies with a PEDro score of five or less or two or more high-risk bias assessments were categorised as low-quality [25]. The assessment of bias agreement among three researchers (AM, GH, and HDB) was measured using the kappa statistic [26].

Statistical analysis

Statistical analysis was conducted using Review Manager software (RevMan 5.4). Heterogeneity among the studies was evaluated using the chi-squared (x²) test and the I² statistic. The study weight was calculated as the inverse of its variance, with the combined effect determined as the weighted average of individual effects. Heterogeneity was considered significant when the I² coefficient was greater than 50%. [27, 28]. Depending on the level of heterogeneity detected in the meta-analyses, either the Mantel-Haenszel fixed-effect method or the DerSimonian and Laird random-effects method was employed to compute the pooled effect, expressed as weighted mean difference (WMD) or standardised mean difference (SMD), along with 95% confidence intervals (Cl), for the reported outcomes of interest. Publication bias was planned to be assessed using funnel plots and Egger’s test [29, 30]

An electronic search for randomised controlled trials (RCT s) examining LLLT intervention in QT was conducted using the Quality of evidence

The evaluation of evidence quality used the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) approach, considering parameters such as study limitations, inconsistency, indirectness, imprecision, and publication bias [27]: (a) study limitations arising from blinding, allocation deficiencies, or overestimation of treatment effects; (b) inconsistency determined by heterogeneity (I² > 50%) in main outcomes; (c) indirectness stemming from significant deviations in treated individuals or when compared to less common interventions; (d) imprecision, entailing uncertainty characterised by broad confidence intervals extending across the line of no effect in the meta-analysis, along with the requirement of an optimal sample size for relevance (n > 400); (e) when there were fewer than three relevant studies, publication bias could occur.

Evidence levels were assigned, from high to very low, with an initial high-quality rating attributed to each level owing to the exclusive inclusion of RCTs. Factors affecting GRADE domains can lead to a decrease in evidence quality. The significance of the evidence was juxtaposed to ascertain alignment with minimally clinically important differences (MCID). Researchers used the GRADEpro GDT tool to compile a summary table of evidence (www.gradepro.org) to ensure a thorough synthesis.

Literature search

The search encompassed seven databases, yielding 1,319 articles as of the last update on March 10, 2025: PubMed (n = 190), Scopus (n = 482), WoS (n = 251), EBSCOhost (n = 126), Embase (n = 340), Cochrane Central (n = 131), and PEDro database (n = 1). Alternative methods, primarily manual searches on Google Scholar, identified 359 articles. After removing duplicates, 738 articles were selected for further analysis. Screening of titles and abstracts led to the preliminary selection of 15 articles, out of which eight were excluded: three studies on HILT in QT, four on LLLT in other hand con ditions, and one SR focusing on physical modalities for hand tenosynovitis. The alternative databases produced 21 documents, including 16 duplicates, with one study on LLLT in other hand conditions, one protocol study, and one webpage about the effects of LLLT. Appendix 1 presents an overview of the studies excluded from the databases, registries, and alternative methods, while Appendix 2 details the search strategy. Ultimately, eight RCTs were included in this review [16-18, SI- 35]. Figure 1 illustrates the search strategy using the PRISMA flowchart.

Figure 1

Prisma flow chart

Quality assessment and risk of bias

The methodological quality assessment using the PEDro scale yielded an average score of 5.3 (± 1.3), with four studies rated as low quality (score < 6) (Table 1). The criteria with the lowest scores were therapist blinding (criterion 6), assessor blinding (criterion 7), and intention-to-treat analysis. In contrast, the highest scores were for adequate follow-up (criterion 8) and group comparisons at the end of treatment (criterion 11). Appendix 2 presents an evaluation of the PEDro scale.

Figure 2 depicts the RoB evaluation with the RoB2 tool, administered by three researchers with a significant concordance (k = 0.76) [26]. The criteria with the highest RoB were deviations from intended interventions (50%) and bias in outcome measurement (37.5%). No bias was observed in “missing outcomes” or “bias due to selective reporting.” The overall bias was assessed as 25%.

Figure 2

Risk of bias summary: assessment of bias in each study

Characteristics of the included studies

Table 1 summarises the characteristics of the included studies, specifying the author, country, number of participants, study groups, interventions, treatment sessions, outcomes assessed, and findings. The studies were conducted in India [16, 17, 32, 33, 35], Turkey [31,34], and Egypt [18] between 2002 and 2023. The total number of participants was 320, including 27 men and 246 women with an average age of 35.3 years (± 6.3). Two studies did not specify participant gender [31,35]. A total of 129 subjects received LLLT, either alone [16, 17, 35] or combined with other interventions such as splint immobilisation [31,34], therapeutic US [32], K-Tape [33], and therapeutic exercise [18]. The controls included 191 participants treated with wrist splints [31,34] and US [17, 33, 35]. Some controls were treated with corticosteroid injections, decompression surgery [35], and Langerhans Cell Histiocytosis (LCH) injections [32]. Only one study included a placebo control for LLLT [31].

Outcomes

Pain was assessed using tools such as the VAS [16-18, 32, 33-35] and the Ritchie Index for tenderness [16, 17]. Three studies evaluated grip strength using a sphygmomanometer or manual dynamometry [16, 17, 34]. Two studies employed USG to gauge the thickness of the synovial sheath within the initial compartment, measuring both anteroposterior and mediolateral diameters [16, 17]. Only one study evaluated disability using the DASH questionnaire [35]. Other outcomes assessed included patient-reported improvement using the Verbal Scale Global Improvement (VSGI) [31,34], quality of life using the Patient-Rated Wrist Evaluation (PRWE) [33], and blood cortisol levels [18].

LLLT dosage

Table 2 outlines the LLLT parameters used in the studies. All lasers were infrared As-Ga-Al diodes with wavelengths of 830 nm, except for one study that used 904 nm [33]. The mean treatment power ranged between 30 and 100 mW, with spot applications over the radial styloid process. Only one study reported a combination of punctual and scanning applications [17]. The spot size varied between 1.57, 4.9, and 12.6 cm². Treatment time ranged from 100 to 600 s, with a mean of 284 s (± 200.6). Energy dosage ranged from 3 to 24 J, with a mean of 16.3 J (± 16.5).

Meta-analysis

Pain intensity: The meta-analysis used six studies to scrutinise the analgesic effects of LLLT in contrast to alternative treatments (Figure 3A) [17, 18, 32-35]. The Dersimonian- Laird random effects model assessed the WMD derived from the MD, with differences in favour of LLLT observed (WMD = -0.98; 95% Cl: -1.91, -0.04; p = 0.04; EG (n) = 107, CG (n) = 124) with substantial heterogeneity [27, 28]. Sensitivity analysis by excluding studies with the lowest methodological quality showed a greater analgesic effect (WMD = -1.50 cm; 95% Cl: -2.11, -0.88; p = 0.01; EG (n) = 93, CG (n) = 93), but heterogeneity remained (Figure 3D).

Figure 3

Forest plots for pain intensity (visual analogue scale) (3A), strength (3B), and synovial sheath thickness (ultrasonography) at the end of treatment (3C). Sensitivity analysis excludes studies with lower methodological quality to manage heterogeneity (3D)

Figure 4 shows the evaluation of publication bias for pain intensity [29, 30]. The distribution for each mean difference (MD) measure extends from the middle of the funnel to the top, indicating good or fair precision given the homogeneous sample sizes, accompanied by a symmetrical MD distribution, although with values crossing the line of no effect. The Egg- er’s regression test ruled out publication bias (p = 0.161), suggesting the absence of a small study effect [30]. The evidence on the analgesic effects of LLLT was assessed as important, albeit with low certainty (Table 4).

Figure 4

Evaluation of publication bias for the pain intensity outcome using a funnel plot (left) and Egger's regression method (right)

Handgrip strength: The meta-analysis included four studies (Figure 3B) [16, 17, 32, 34]. The DerSimonian and Laird random effect analyses were applied to ascertain the SMD. No statistically significant differences favouring either group were discerned (SMD = -0.07; 95% Cl: -0.92, 0.78; p = 0.87; EG (n) = 76, CG (n) = 77), and substantial heterogeneity was present [27, 28]. The quality of the evidence was not assessed since there was no difference between the groups to warrant a recommendation grade.

Synovial sheath thickness: Two studies were included in the meta-analysis (Figure 3C) [16, 17]. Due to the absence of heterogeneity, the Mantel-Haenszel fixed-effect method was used. No differences were observed between the groups for the reduction in anteroposterior diameter (WMD = -0.07 mm; 95% Cl: -0.05, 0.12; p = 0.47; EG (n) = 28, CG (n) = 28) and mediolateral diameter (WMD = 0.00 mm; 95% Cl: -0.08, 0.08; p = 0.94; EG (n) = 28, CG (n) = 28) measured by USG. The quality of the evidence was not assessed since there was no difference between the groups to warrant a recommendation grade [29].

Disability: A meta-analysis was not conducted for disability because it was only reported as a relevant outcome by one study [35].

LLLT and pain reduction

Lllt effectively reduces VAS-determined pain in QT patients when applied alone [16, 17, 35] and in combination with other physical therapy treatments such as splint immobilisation [31,34], US [32], K-Tape [33], and therapeutic exercise [18]. The results were significantly in favour of LLLT within and between groups. Nevertheless, these values fall below the reported minimal clinically important difference (MCID) of 1.37 cm for VAS [36]. Furthermore, it has been suggested that a change of 2.8 cm (± 2.1) for patients with baseline VAS pain scores of at least 6.7 can represent a clinically significant change in pain severity (MIC) from the patient’s perspective [37]. The meta-analysis did not reflect this MIC, but intragroup comparisons before and after LLLT treatment did.

Although LLLT is thought to have analgesic effects, particularly in nociceptive and neuropathic pain, the specific un derlying mechanisms are controversial. The widely accepted hypothesis suggests that the oxidised cytochrome c oxidase (CCO) enzyme, the terminal part of the electron transport chain, absorbs radiation because of its photosensitivity to light [11, 14]. CCO acts as a photoacceptor for red and infrared wavelengths through copper and iron chromophores, leading to increased cellular metabolism and ATP production. Additionally, LLLT promotes nitric oxide release through photodissociation from CCO, increasing the rate of ATP production and promoting vasodilation, thereby modulating reactive oxygen species (ROS) [38, 39]. ROS are implicated in inflammatory, neuropathic, and persistent pain, promoting neuronal excitability in nociceptive pathways and triggering mitochondrial dysfunction and neuroinflammation [38]. Additionally, ROS contribute to the activation of transient receptor potential channels (TRP), rendering transient receptor potential vanilloid 1 (TRPV1) more sensitive and activating transient receptor potential A1 (TRPA1) [40].

Lllt induces analgesia by stimulating endogenous opioid peptide synthesis, such as ß-endorphin, while simultaneously reducing the activity of bradykinin and C fibres [41]. Additionally, LLLT initiates morphological alterations in neurons, leading to a reduced mitochondrial membrane potential and a deceleration or inhibition of nerve conduction, thereby attenuating nociceptive transmission [10, 13, 41]. Neurons of the dorsal root ganglion (DRG) specifically exhibit these effects [42]. Furthermore, LLLT mitigates the release of pro-inflammatory neuropeptides, such as substance P and calcitonin gene-related peptide (CGRP), which play pivotal roles in neurogenic inflammation [39, 41].

Evidence suggests that LLLT may increase serotonin (5-HT) production in the central nervous system, favouring the activation of descending pain modulatory pathways [39, 41]. Through the 5-HT1a, 5-HT1b, 5-HT2c, 5-HT3, and 5-HT4 receptors, the 5-HT pathway from the rostral ventromedial medulla changes how pain signals are sent In the dorsal horn [43]. The diminution of signalling molecules Implicated In the Inflammatory cascade, including nuclear factor kappa B (NF-kB), which can hinder prostaglandin E2 (PGE2), tumour necrosis factor-alpha (TNF-a), cyclooxygenase-2 (COX-2), and interleukin 1 (IL-1), may also explain the analgesic effects of LLLT [39].

Heterogeneity observed for pain intensity

The I² index determines heterogeneity, which was evident in the pain intensity meta-analysis. This heterogeneity indicated variability among the studies, which affects the validity of the conclusions drawn from combining the RCT results [27, 28]. These variations may be due to differences in sample size, randomisation, or the measurement of outcomes of interest [27, 28].

Despite attempts to perform a sensitivity analysis by excluding studies with lower methodological quality [35], significant heterogeneity persisted. Additionally, excluding studies can increase publication bias, leading to a small study effect that overestimates the results. Consequently, the authors opted to retain the initial analysis results.

LLLT and synovial sheath thickness

Lllt has been shown to effectively reduce synovial sheath thickness by the end of treatment, though it does not surpass control treatments such as US [17, 35]. This application is grounded in the anti-inflammatory effects of LLLT, supported by various physiological mechanisms. Indeed, wavelengths between 632 and 904 nm induce significant anti-inflammatory effects comparable to nonsteroidal anti-inflammatory drugs (NSAIDs) [44]. These effects include the inhibition and/ or attenuation of inflammatory mediators and pain markers, such as TNF-a, IL-ip, IL-6, COX-2, and PGE2 [44, 45].

In certain scenarios where corticosteroid injections are used to treat QT, the anti-inflammatory efficacy of LLLT may be compromised [35, 46]. Corticosteroid injections are successful in 73.4% of cases with two applications but have a higher failure rate in women with a high body mass index (BMI > 30 kg/m²) [46]. Therefore, LLLT could be an effective alternative to steroid injections in overweight women with QT.

There is evidence that both red and infrared LLLT have a dose-dependent anti-inflammatory effect, with a median average power output of 25 mW [44]. According to the ArndtSchultz law [47], biostimulation is observed at doses ranging from 0.05 to 10 J/cm², with an optimal range between 0.5 and 4 J/cm², which has been associated with reduced pain and inflammation [44, 48]. A dose of 0.6 J/cm², with an irradiation time of at least 16 s, yields the lowest energy densities for anti-inflammatory effects, while a dose of 3 J/cm² for 300 s yields better effects [44].

LLLT and grip strength

Although LLLT therapy focuses on pain reduction, this review demonstrates an increase in grip strength at the end of treatment, although it was not shown to be superior to control treatments. Pain influences the ability to generate muscle strength due to reflex inhibition and structural changes associated with the clinical condition [49, 50]. Moreover, a reduction in strength of up to 20-30% in a painful limb appears to be typical in chronic pain patients [51].

The assessment of grip strength is essential for understanding hand function. Therefore, the authors recommend maintaining it as a relevant outcome variable for future studies.

LLLT and disability

Despite the close relationship between pain and disability, only one study using the DASH questionnaire considered this variable as a relevant outcome [35]. Pain, particularly in chronic cases, commonly restricts functional capacity, leading to heightened pain perception due to physical inactivity and reduced mobility [52]. However, the relationship between the duration, intensity, extent, and meaning of pain does not always show a linear correlation with disability [52, 53]. The authors recognise the complexity of this construct and highly recommend the inclusion of functional outcomes, such as patient-reported outcome measures (PROMs) [54], in (RCTs) and clinical practice.