Lateral elbow tendinopathy (LET) or "tennis elbow" is a common disorder with a prevalence of at least 1.7% 1, and occuring most often between the third and sixth decades of life. Physical strain may play a part in the development of LET, as the dominant arm is significantly more often affected than the non-dominant arm. The condition is largely self-limiting, and symptoms seem to resolve between 6 and 24 months in most patients 2.

A number of interventions have been suggested for LET. Steroid injections, non-steroidal anti-inflammatory drugs or a regimen of physiotherapy with various modalities, seem to be the most commonly applied treatments 3. However, treatment effect sizes seem to be rather small, and recommendations have varied over the years. In several systematic reviews over the last decade 45, glucocorticoid steroid injections have been deemed effective, at least in the short-term. But in later well-designed trials evidence is found that intermediate and long-term effects of steroid injections groups yield consistently and significantly poorer outcomes than placebo injection groups, and physiotherapy or wait-and-see groups 67. Nevertheless, steroid injections have been considered as the most thoroughly investigated intervention, with 13 randomized controlled trials comparing steroid injections to either placebo/local anaesthetic or another type of intervention 5. Non-steroidal anti-inflammatory drugs (NSAIDs) have been found to achieve smaller short-term effect sizes than steroid injections 8, and topical application seems to be the best medication administration route 8 For oral administration of NSAIDs for LET, evidence is inconclusive from two heterogeneous trials only 9. The positive short-term results of anti-inflammatory therapies in LET appear to partly contradict the recent paradigm in tendinopathy research, where LET is thought to be mainly a degenerative disorder with minimal inflammation 1011.

Exercise therapy and stretching exercises have been used either alone or in conjunction with manipulation techniques or physical interventions. Although the sparse evidence makes it difficult to assess the separate effect of active exercises or stretching 12, four studies have found that either exercises alone 13, or in conjunction with a physiotherapy package, are more effective than placebo ultrasound therapy or wait-and-see controls. Also exercise therapy, particularly eccentric exercises, have been found effective in the intermediate term in tendinopathies of the Achilles, patellar or shoulder tendons 14151617. There is some evidence suggesting that joint manipulation or mobilisation techniques either of the wrist, elbow or cervical spine may contribute to short-term effects in LET 181920.

Among the physical interventions, ultrasound therapy has been considered to offer a small benefit over placebo from two small trials 12, but a well-designed and more recent trial did not find significant effects of ultrasound therapy in LET 21. Reviewers have arrived at different conclusions for the effect of acupuncture 2223. In reviews of physical interventions for LET, conclusions may vary between reviews because of differences in the treatment procedures. A good example of this is the negative conclusion of the LET review for extracorporeal shockwave therapy (ESWT) by Buchbinder et al. 24, where a later review with in-depth assessments of treatment intervention protocols 25, found that a subgroup of trials with proper treatment procedures and adequate timing of outcomes gave a positive result.

Low level laser therapy (LLLT) has been available for nearly three decades, and scattered positive results have been countered by numerous negative trial results. Several systematic reviews have found no significant effects from LLLT, in musculoskeletal disorders in general 26, and in LET in particular 122327. In this perspective it may seem futile to perform yet another systematic review in this area. But none of these reviews evaluated the results separately for the different LLLT treatment procedures, laser wavelengths or doses involved. Neither did they implement evidence of the newly discovered biomodulatory mechanisms which are involved when LLLT is applied. During the last 5–6 years the annual number of published LLLT reports in Medline has increased from 25 to around 200. We recently made a review of this literature, and concluded that LLLT has an anti-inflammatory effect in 21 out of 24 controlled laboratory trials, and a biostimulatory effect on collagen production in 31 out of 36 trials 28. Both of these effects were dose-dependent and could be induced by all wavelengths between 630 and 1064 nm with slight variations in therapeutic dose-ranges according to the wavelength used. The anti-inflammatory effect was seen in higher therapeutic dose-ranges than the biomodulatory effect on fibroblast cells and collagen fibre production. Diagnostic ultrasonography of tendinopathies has revealed that partial ruptures and tendon matrix degeneration are underdiagnosed if only physical examinations are made. Consequently, the stimulatory LLLT-effect on collagen fibre production should probably be beneficial for tendon repair. Another interesting feature was that LLLT with too high power densities or doses (above 100 mW/cm²), seemed to inhibit fibroblast activity 29 and collagen fibre production 30. Six years ago we showed in a systematic review of tendinopathy, that the effect of LLLT is dose-dependent 31. At the time, the accompanying editoral suggested that the advanced review design could become the new standard for reviewing empirical therapies with unknown optimal doses and procedural differences 32. Steroids induce a down-regulation of cortisol receptors, and we recently discovered that the cortisol antagonist mifepristone completely diminished the anti-inflammatory effect of LLLT 33. All these recent findings from the LLLT literature, prompted the World Association for Laser Therapy (WALT) to publish dosage recommendations and standards for the conductance of systematic reviews and meta-analyses last year 34. One of the issues that has lacked attention is the validity of LLLT-application procedures in tendinopathy. To our knowledge there are only three valid irradiation techniques for LLLT in tendinopathies: a) direct irradiation of the tendon, b) irradiation of trigger points and c) irradiation of acupuncture points.

In this perspective and as our previous tendinopathy review 31 is becoming outdated, there seems to be a need for a new in-depth review of the effects of LLLT in LET where possible confounders are analyzed and subgroup analyses are performed.

In this review, we found that most RCTs of LLLT for LET were of acceptable methodological quality. This finding is in line with previous reviews 122327, although there were some differences between reviewers in methodological scores for individual trials. RCTs of LLLT are of similar methodological quality and include similar sample sizes as RCTS included in recent reviews of corticosteroid injections 5 and topical or oral NSAIDs 8. Two of the previous reviews of LLLT for LET found only six RCTs 1223, whereas an earlier review found ten RCTs 27, and excluded one RCT for methodological shortcomings 43. We used broader searching criteria in our review and had no language restrictions. This resulted in 18 potentially eligible RCTs. We excluded one RCT for not meeting the inclusion criteria of specific endpoints 43 and another two RCTs for complete lack of blinding 44 and a lack of an LET control group 42. None of the previous LET reviews assessed the LLLT regimen for procedural errors, while our procedural assessments resulted in exclusion of another two RCTs with grave procedural errors, such as leaving the tendon insertion and acupoints unirradiated 40 and giving adequate LLLT to the placebo group 61. These exclusions resulted in 13 RCTs being eligible for our review which is twice the number of RCTs included in two of the previously published reviews1223.

Previous LET-reviews of LLLT 122327 and pharmacological interventions like NSAID 8 or corticosteroid injections 5 have not assessed possible bias from for-profit funding sources or publication bias. Our analysis revealed that bias from for-profit funding was largely absent in the available LLLT material and that trials were performed by independent research groups receiving funding from internal sources or non-profit organisations. This feature of the LLLT literature is definitely different from pharmacological pain treatments where up to 83% of trials may be industry-funded 62. A second feature of the LLLT-literature is that publication bias seems to go in a negative direction. This is distinctly different from the drug trials 6364 where positive results have been found to account for up to 85% of the published trials in single journals 63, although this bias seems to be lesser or absent in high impact journals 64. Our review suggests that LLLT trials reporting negative results are more likely to be published than trials with positive results. To our knowledge we are the first to demonstrate such bias, but such negative publishing bias is probably not unique to LLLT, and it may also be present for other electrophysical agents including TENS and acupuncture. We were surprised to see how large well-designed positive trials of LLLT 5150 were published in unlisted journals or journals with low-impact factor, and how small negative trials 46, often with grave methodological 42 or procedural flaws 40 were published in higher ranking journals. This may reflect a predominance of RCTs designed using drug-research methodology paradigms without due consideration given to adequacy of the technique used in delivering LLLT, leading to under dosing and negative outcome bias 65. In addition, it has been that documented drug sponsorship of research activities may influence guideline panels, journal editors and referees 6667 leading to negative views on non-drug treatments such as LLLT as reflected in editorials in pain journals 68 and national medical journals 69.

Despite these concerns, we believe that the positive overall results of this review need to interpreted with some caution. They arise from a subgroup of 7 out of the 13 included trials 48495051555657. These 7 trials had a narrowly defined LLLT regimen where lasers of 904 nm wavelength with low output (5–50 mW) were used to irradiate the tendon insertion at the lateral elbow using 2–6 points or an area of 5 cm² and doses of 0.25–1.2 Joules per point/area. The positive results for this subgroup of trials were consistent across outcomes of pain and function, and significance persisted for at least 3–8 weeks after the end of treatment, in spite of several factors which may have deflated effect sizes.

For the red 632 nm wavelength which has a poorer skin penetration ability 70, a single trial 60 with a higher dose (6 Joules) seemed to be equally effective as the lower doses of 904 nm used in the seven positive trials. These LLLT-doses are well within the therapeutic windows for reducing inflammation, increasing fibroblast activity and collagen fibre synthesis, and the dosage recommendations suggested by WALT 71.

The negative results for the 830 nm GaAlAs and 1064 nm NdYag lasers can be attributed to several factors such as too high doses, too high power density or the inclusion of patients with poor prognosis from long symptom duration and prior steroid injections. These wavelengths have previously been found effective in some tendon animal studies and in other locations such as shoulder tendinopathies 7273. At this time it is not possible to draw firm conclusions about the clinical suitability of wavelengths 820, 830 and 1064 nm in LET treatment, but the lack of evidence of effects indicates that they cannot be recommended as LET treatment before new research findings have established their possible effectiveness. The lack of effect for these lasers may also serve as a reminder that higher doses is not always best. We have been witnessing a tendency where newly developed lasers with these wavelengths are being marketed with ever-increasing power and power densities. This may be inappropriate because current knowledge about LLLT mechanisms and dose-response patterns at higher powers is inconsistent or lacking.

The positive results for combining LLLT of 904 nm wavelength with an exercise regimen, are encouraging. We would have thought that exercise therapy could have erased possible positive effects of LLLT, but the results showed an added value in terms of a more rapid recovery when LLLT was used in conjunction with an exercise regimen. This may indicate that exercise therapy can be more effective when inflammation is kept under control. Adding LLLT to regimens with eccentric and stretching exercises reduced recovery time by 4 and 8 weeks in two trials 4856. For this reason, LLLT should be considered as an adjunct, not an alternative, to exercise therapy and stretching.

Based on the above findings, LLLT should be considered as an alternative therapy to commonly used pharmacological agents in LET management. Cochrane-based reviews of NSAIDs 8 and corticosteroid injections 5 have found evidence of short-term effects within 4 and 6 weeks, respectively. The short-term reduction in pain intensity after corticosteroid injections may appear to have a more rapid onset and may also be larger in effect size than after LLLT. But on the other hand, the available LLLT-material is confounded by factors capable of deflating effect sizes. In this perspective, there is a need for more high quality trials with head-to-head comparison of short-term effects between LLLT and corticosteroid injections. In the longer term, NSAIDs seems to be ineffective and corticosteroid injections seem to be harmful both at 26 and at 52 weeks 6. For LLLT there are some significant long-term effects found at 8, 12 and 24 weeks after the end of treatment.

The available material suggests that LLLT is safe and effective, and that LLLT acts in a dose-dependent manner by biological mechanisms which modulate both tendon inflammation and tendon repair processes. With the recent discovery that long-term prognosis is significantly worse for corticosteroid injections than placebo in LET, LLLT irradiation with 904 nm wavelength aimed at the tendon insertion at the lateral elbow is emerging as a safe and effective alternative to corticosteroid injections and NSAIDs. LLLT also seems to work well when added to exercise and stretching regimens. There is a need for future trials to compare adjunctive pain treatments such as LLLT with commonly used pharmacological agents.

The authors declare that they have no competing interests.

JMB had the original idea, which was developed through lengthy discussions with contributions from RABL-M, JJ, CC, AEL, AS and MIJ. The literature search, including handsearching, was performed by all members of the author team. The first draft was written by JMB, RABL-M and JJ, and revised by AS and MIJ. Methodological assessments of trials were performed by JMB, AEL, CC, AS. The statistical analysis was performed by JMB, RABL-M, JJ and MIJ. The final linguistic revision was performed by MIJ and all members of the author team read and commented on the manuscript before submission.