Objective To evaluate the effectiveness of exercise compared with other conservative interventions in the management of lateral elbow tendinopathy (LET) on pain and function.
Design Systematic review and meta-analysis.
Methods We used the Cochrane risk-of-bias tool 2 for randomised controlled trials (RCTs) to assess risk of bias and the Grading of Recommendations Assessment, Development and Evaluation methodology to grade the certainty of evidence. Self-perceived improvement, pain intensity, pain-free grip strength (PFGS) and elbow disability were used as primary outcome measures.
Eligibility criteria RCTs assessing the effectiveness of exercise alone or as an additive intervention compared with passive interventions, wait-and-see or injections in patients with LET.
Results 30 RCTs (2123 participants, 5 comparator interventions) were identified. Exercise outperformed (low certainty) corticosteroid injections in all outcomes at all time points except short-term pain reduction. Clinically significant differences were found in PFGS at short-term (mean difference (MD): 12.15, (95% CI) 1.69 to 22.6), mid-term (MD: 22.45, 95% CI 3.63 to 41.3) and long-term follow-up (MD: 18, 95% CI 11.17 to 24.84). Statistically significant differences (very low certainty) for exercise compared with wait-and-see were found only in self-perceived improvement at short-term, pain reduction and elbow disability at short-term and long-term follow-up. Substantial heterogeneity in descriptions of equipment, load, duration and frequency of exercise programmes were evident.
Conclusions Low and very low certainty evidence suggests exercise is effective compared with passive interventions with or without invasive treatment in LET, but the effect is small.
PROSPERO registration number CRD42018082703.
- tennis elbow
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Lateral elbow tendinopathy (LET), (‘tennis elbow’), is the most common musculoskeletal disorder of the elbow causing significant disability and productivity loss. The prevalence of LET ranges from 1% to 3% in the general population and up to 29% in certain occupations.1 The diagnosis of LET is based on the presence of pain over the lateral humeral epicondyle that may radiate distally into the forearm.2 Conservative management is suggested as first-line treatment and may include: exercise, orthoses, manual therapy, physical therapy modalities, acupuncture and multimodal physiotherapy treatment.2 3 Load management and different modes of exercise (ie, eccentric, isometric, concentric) appear to be the most popular treatment approaches with or without additive interventions, such as ‘passive’ physical therapies or injections.4 5 Three previous systematic reviews advocated the effectiveness of eccentric exercise in the management of LET at short-term follow-up; however, their superiority compared with other interventions remains inconclusive.6–8
This review intended to evaluate the effectiveness of exercise compared with other conservative interventions in the management of LET. We also tried to synthesise the evidence regarding exercise type, mode and dosage aiming to inform clinical practice.
We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines9 in search strategy and reporting.
We included only randomised controlled trials (RCTs) comparing the effectiveness of any type of exercise used alone or as an additive intervention compared with another type of conservative intervention in adult patients diagnosed with LET. Outcome measures included: (1) Patient-Rated Tennis Elbow Evaluation (PRTEE),10 Tennis Elbow Function Scale,11 Nirschl/Pettrone pain score,12 the Disability of the Arm, Shoulder and Hand questionnaire (DASH),13 Pain Free Function Questionnaire (PFFQ)14; (2) Global Rating of Change (GROC); (3) pain on activity/loading in Visual Analogue Scale or Numeric Pain Rating Scale and (4) pain-free grip strength (PFGS). Follow-up was categorised into very short-term (≤2 months), short term (>2 months≤3 months), mid term (>3 to<12 months) and long term (≥12 months).15
Information sources and search
MEDLINE, PubMed, CINAHL, EMBASE, PEDro, ScienceDirect, Cochrane Library and Grey literature databases were systematically searched from inception to November 2019 (online supplemental material 1).
Study selection and data extraction
Search results were imported into EndNote V.X7 and two independent reviewers (SK and IV) evaluated them against the eligibility criteria in a two stage process.16 Data extraction was conducted by the same reviewers independently using a standardised data extraction form. Discrepancies were resolved by discussion with a third reviewer (VK). Based on the Consensus on Exercise Reporting Template checklist, the most relevant details from exercise interventions were extracted.17
Risk of bias in individual studies
Two independent reviewers (SK and IV) assessed risk of bias using the criteria advised by the revised Cochrane risk-of-bias tool 2 for randomised trials.18 Criteria were rated as ‘low risk’, ‘high risk’ or ‘some concerns’ of bias and then revised the ‘overall’ risk of bias for each study. A third reviewer (VK) was consulted when necessary.
Data analysis, synthesis and summary of findings
Outcome data were transformed to 0–100 point scales (ie, pain, function) or ratios of symptomatic:non-symptomatic elbow for PFGS (x100%) for analyses. To compute success rates (GROC), patients who rated themselves as ‘completely recovered’ and ‘much improved’ were counted as successes.
As measures of treatment effect, we calculated risk ratios with 95% CIs for dichotomous data and mean difference (MD) or standardised mean difference where applicable with 95% CIs for continuous data. A random-effects model was included to pool studies’ outcomes due to clinical and methodological heterogeneity. Data were entered and treatment effect estimates were calculated using the Cochrane Review Manager software V.5.3 (The Cochrane Collaboration). To assess publication bias, we planned to generate funnel plots for meta-analyses including at least 10 trials.16
We decided post hoc protocol registration to subgroup and pool studies comparing exercises with or without physiotherapy or home exercise programme to another type of intervention (ie, modalities, corticosteroid injections) or a control group. Sensitivity analyses (post hoc registration) were performed examining unexpectedly large treatment effect sizes and ‘leave-one-out’ exclusion, as well as studies presenting significant heterogeneity at baseline for participant characteristics. Two independent reviewers (SK and VK) evaluated the certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology15 (box 1).
Criteria for downgrading the quality of outcomes using the Grading of Recommendations Assessment, Development and Evaluation approach
Evidence started at high certainty and was downgraded for the following reasons15 :
Risk of bias (>75% of studies do not have low risk of bias).
Inconsistency (large heterogeneity based on the similarity of point estimates, statistical heterogeneity and I2 >50%).
Imprecision (included population inadequately reflects selection criteria of the review, estimate of the effect is sufficiently imprecise if the upper or lower confidence limit crosses the effect size of 0.5 in either direction).
Indirectness (studies with indirect comparison).
Publication bias (asymmetry in funnel plots where a sufficient number of studies was available).
When one trial was available, the evidence was graded as low certainty, and if it also had high risk of bias the evidence was graded as very low certainty.19 79
For elbow function, we defined the minimal clinically important difference (MCID) of the PRTEE as a mean of 11-point change or 37% of the baseline score,10 while for pain intensity it was defined as a 30% improvement from the pooled weighted mean of the baseline.19 The MCID of PFGS has been reported to be 7 kg in a group of patients with LET11 which represents a 18% change of the pooled mean normative value of grip strength (38.4 Kg) for men and women aged 40–50 years across geographically relevant studies.20–25 Therefore, we set the clinical significance of PFGS at 20% change in the ratio (%) between the affected and unaffected sides.
Study selection and characteristics
Study characteristics, outcome measures and main results are presented in online supplemental material 3. The median number of participants randomised per trial was 45 (25th and 75th percentiles 31–101). The total number of participants was 2123 (mean age 46.6 years, 46% female) and the sample size ranged from 17 to 198. Diagnostic criteria for LET were presence of tenderness on palpation over the lateral humeral epicondyle and resisted middle finger extension in 28 studies (93%).26 28–31 33–55 Ten studies (33%) included pain provoked by gripping or decreased PFGS.27–29 32 33 35 42 45 52 55 Eleven studies (37%) used stretching of forearm extensor muscles for inclusion.28 29 32 35 40–42 45 48 52 53
Exercise as an intervention was used alone, in combination with physiotherapy, and as a home exercise programme (HEP). Nine of the included trials compared exercise with invasive treatment,26 28 31 34 36 43 49 50 55 six trials with a wait-and-see policy group (WSP),26 36 38 43 50 53 three trials with forearm support band (FSB),31 44 47 four trials with physiotherapy with or without modalities,32 35 52 54 two trials with ultrasound,40 41 four trials made a direct or indirect comparison of eccentric and concentric exercises,33 39 45 51 one trial with cryotherapy,27 one trial with the addition of Pilates-based exercises,29 one trial with the addition of scapular strengthening exercises,42 one trial with delayed exercise initiation and use of nonsteroidal anti-inflammatory drugs (NSAIDs),37 one study with propioceptive neuromuscular facilitation (PNF) stretching,48 one trial with neural tension techniques of the radial nerve30 and one with manipulation techniques of the wrist.46
Risk of bias within studies
The overall risk of bias was rated as ‘some concerns’ in 30% of trials (n=9/30), and as ‘high’: in 70% (figure 2).
Most of the studies (63%) were rated with ‘some concerns’ in domain 1 due to lack of information of concealment of the allocation prior to group assignment. We assessed the effect of adhering to intervention as high risk in 66% of studies mainly due to lack of blinding of participants and therapists and lack of appropriate analyses to estimate the effect of adherence. However, in most studies, it was not possible to blind participants and personnel in an exercise or invasive intervention. Missing outcome data and inappropriate analyses methods meant 43% of the studies had ‘high risk’ of bias. Bias in measurement of the outcome was judged as ‘some concerns’ in 83% of the patient-rated outcomes in included studies and as ‘low risk’ in all studies using observer-reported outcomes. Only 2/30 studies had prespecified, registered outcomes and analysis plans and were judged as ‘low risk’ in selective reporting.
Main results of meta-analyses and GRADE of evidence
All outcomes were rated as very low or low-certainty evidence. Forest plots for all comparisons and outcomes are presented in online supplemental material 4.
Exercise with or without physiotherapy or HEP compared with corticosteroid injection(s)
Eight RCTs26 28 31 34 36 43 47 50 made a direct or indirect comparison of exercise and corticosteroid injections as an intervention for LET. The maximum duration of follow-up ranged from 7 weeks to 1 year and the mean age of the participants (n=933) ranged from 41.9 to 51.6 years. Two studies31 36 were excluded from quantitative synthesis due to indirectness and heterogeneity of intervention and comparators, and two studies49 50 due to attrition bias (29% and 70%, respectively).
Three studies26 28 43 evaluated self-perceived recovery using a GROC scale showing a significant difference in favour of exercise (physiotherapy ±HEP) compared with corticosteroid injections in self-perceived recovery (table 1).
From the four studies5 7 34 37 which evaluated pain intensity we excluded one34 from quantitative synthesis due to significant heterogeneity of the comparators (measurement of pain at rest compared with pain during activities during the day or last week). A statistically, but not clinically, significant difference was found in favour of exercise compared with corticosteroid injections in mean change in pain rating (table 1).
Three studies26 28 33 evaluated the PFGS at short-term, mid-term and long-term follow-up and were included in quantitative synthesis. There was a statistically and clinically significant difference in favour of exercises (physiotherapy ± HEP) compared with corticosteroid injections in PFGS ratio (table 1).
In elbow disability the quantitative synthesis of four studies5 26 33 43 using the PRTEE, PFFQ, or modified PFFQ suggested only statistically significant differences at short-term, mid-term and long-term follow-up (table 1).
Exercise with or without physiotherapy or HEP compared with WSP
Six RCTs26 36 38 43 49 53 evaluated the use of exercises with or without physiotherapy or HEP compared with a WSP for patients with LET. The maximum duration of follow-up ranged from 3 weeks to 1 year and the mean age of the participants (n=729) ranged from 41.9 to 49.1 years. We excluded one study with high risk of bias49 from quantitative synthesis due to attrition bias (29%).
Four studies26 36 43 53 evaluated self-perceived recovery by using 6-point GROC scale showing a statistically significant difference at short-term and no difference at mid-term and long-term occasions (table 2).
From five RCTs evaluating pain intensity, we excluded one study38 from quantitative synthesis due to significant heterogeneity of the comparators (measurement of pain at maximum voluntarily contraction and maximum elongation). There were statistically, but not clinically, better results in favour of exercises (physiotherapy ± HEP) in mean change in pain rating at short-term and long-term follow-up, but not at the mid-term follow-up (table 2).
Four RCTs28 36 43 53 evaluated PFGS, but substantial heterogeneity of the measurement of grip strength (mean value in kilograms) excluded one study53 from quantitative synthesis. There were no significant differences in PFGS at all follow-up time points (table 2).
Five RCTs28 36 38 43 53 evaluated elbow disability by using the PRTEE, DASH, PFFQ or modified PFFQ. There was only a statistically significant difference in favour of exercise (physiotherapy ±HEP) compared with WSP at the long- term follow-up (table 2).
Eccentric compared with concentric exercise alone or in combination with or without physiotherapy or HEP
Four RCTs33 39 45 51 made a direct or indirect comparison of eccentric and concentric exercises for LET. The duration of follow-up ranged from 6 weeks to 1 year and the mean age of the participants (n=269) ranged from 42.8 to 51.0 years.
Four RCTs33 39 45 51 evaluated pain intensity of which one39 was excluded from quantitative synthesis due to significant heterogeneity of the outcome measure (measurement of pain at maximum voluntarily contraction and elongation). There were no significant differences between eccentric alone compared with concentric (with or without eccentric exercises) at short-term follow-up for change in pain, pain-free grip strength or elbow disability (table 3).
Exercise with or without physiotherapy or HEP compared with Cyriax physiotherapy
Three RCTs32 35 52 evaluated the effectiveness of exercise (physiotherapy ± HEP) compared with Cyriax physiotherapy as an intervention for LET. The duration of follow-up ranged from 2 weeks to 2 months and the mean age of the participants (n=140) ranged from 32.0 to 38.2 years. Cyriax physiotherapy included deep transverse friction of the common extensor tendons and a Mill’s manipulation, while exercise included an eccentric strengthening component.32 35 52 Two studies made an indirect comparison of exercise to Cyriax physiotherapy by adding low level laser therapy32 or phonophoresis,35 while one study52 directly compared Cyriax physiotherapy to a supervised exercise programme. Due to substantial heterogeneity of data reporting among studies a quantitative synthesis was not performed. Contradictory results were present at 1-month follow-up for all outcome measures. One study32 showed no significant differences between interventions in pain, function and PFGS, one study52 presented favourable results for exercise in pain and function, while another study35 found favourable results (1 and 2 months) for Cyriax physiotherapy in pain, function and PFGS.
Exercise with or without physiotherapy or HEP or FSB compared with FSB alone
Three RCTs31 44 47 evaluated exercise and FSB as an intervention for LET. Quantitative synthesis was not possible due to lack of data availability from one study47 and indirectness of intervention in another31 (online supplemental material 5, box 1).
There were no significant differences between exercise (physiotherapy ±HEP) and FSB in self perceived improvement, PFGS, at all follow-up occasions. There were statistically better results for FSB alone in mean change in pain rating at very short-term and no difference between interventions at mid-term and long-term follow-up .
FSB alone performed statistically better in elbow disability compared with either exercise (physiotherapy ± HEP) with or without FSB at very short-term follow-up; however, no difference between the comparators was found at mid-term and long-term follow-up.47
Exercise with or without physiotherapy or HEP compared with other types of active or passive interventions
Nine RCTs evaluated the effectiveness of exercises compared with other types of active or passive interventions. Three studies with high risk of bias were excluded from further analyses mainly due to inability to retrieve data from the publications.27 48 54 There were non-significant differences between exercises and cryotherapy,27 Pilates-based exercises,29 a delayed programme of isometric exercises after 1 month of oral NSAIDs,37 PNF stretching,48 wrist manipulation,46 physiotherapy programme54 and neural tension techniques30 in all comparators and follow-up occassions (online supplemental material 5). There were statistically significant differences in favour of exercises (physiotherapy ±HEP) compared with prolotherapy in mean change in pain rating and PFGS at short-term follow-up55 and in favour of the addition of scapular strengthening compared with standard care in all comparisons at very short-term follow-up42 (online supplemental material 5).
Sensitivity analyses that altered the effect estimate are presented in online supplemental material 6.
Description and details of exercise interventions
All included studies used an active exercise component; 50% (15/30) used additive physiotherapy interventions, 90% (27/30) employed an HEP and in 53% (16/30) of the studies a form of patient education was provided. No study provided complete information of the exercise intervention(s) within the publication or additional material. All studies provided the name of the exercise intervention and repetitions and sets for some exercises. However, studies partially described the materials (free weights, elastic bands, etc), the procedure, and the load used. Only 27% (8/30) of the studies provided rest time between sets, 70% (21/30) described time under tension, 90% (27/30) specified the duration of the intervention, 60% (18/30) reported who provided or supervised the intervention and 43% (13/30) gave information about fidelity or adherence of which 7/13 provided no further information, while 6/13 studies reported an exercise compliance range rate of 51%–99% (online supplemental material 7).
Main findings and comparison with other reviews
Thirty RCTs from 12 countries, including 2123 patients suffering from LET with a mean age of 46.6 years, and an overall ‘high risk’ of bias in 70% of the trials were analysed aiming to evaluate the effectiveness of exercise alone or as an additive intervention or HEP in patients with LET.
There is very low certainty and low certainty evidence of statistically significant benefits for an exercise component compared with corticosteroid interventions in all outcomes at all follow-up occasions. We found clinically significant differences in mean change in pain rating at mid-term follow-up and PFGS at all follow-up occasions. However, low and very low certainty evidence suggests interventions with an active component only outperform WSP in self-perceived improvement at short-term and mean change in pain rating and elbow disability at short- and long-term follow-up. Overall, interventions with an active treatment component do better than passive interventions with or without invasive treatment, but the effects are small. We rated all results as very low and low certainty and should be interpreted with caution due to statistical heterogeneity, high risk of bias, inconsistency, indirectness of interventions, and substantial differences at baseline pain and disability scores among studies.
The design and the objectives of our systematic review and meta-analysis did not allow a direct overall comparison with previous systematic reviews in LET evaluating interventions with an active component.6–8 56–58 As with primary research studies, systematic reviews vary greatly in quality and clarity of reporting. Recent systematic reviews on the effect of exercise interventions on LET6 8 did not adhere to guidelines addressing suboptimal reporting and improving systematic reviews’ quality.9 For LET most evidence has used quality assessment tools instead of a comprehensive risk of bias assessment. This creates the threat of different critical assessment findings affecting inferences about the reviews’ credibility.59 Only one narrative review8 evaluated the effectiveness of different types of exercises in LET and concluded that resistance exercises can reduce pain, increase strength and function in LET. The inclusion of non-RCTs and the lack of results’ translation to clinically relevant metrics may partially explain the substantial disagreement with the conclusions of our review.
Traditional passive treatments, such as transverse friction, and therapeutic ultrasound have shown short-term effects in some cases, but have not been maintained in the long-term follow-up in lower limb tendinopathy.60–63 Similarly, moderate evidence of no effect has been found for other commonly prescribed passive interventions, such as laser therapy, transcutaneous electrical nerve stimulation, extracorporeal shockwave therapy, pulsed electromagnetic and ultrasound in shoulder tendinopathy64 and inconclusive evidence for the management of LET.65
Comparably, we found inconclusive evidence between exercise and Cyriax physiotherapy for pain, function, and PFGS at short-term follow-up.32 35 52 It seems that passive interventions such as Cyriax physiotherapy, braces or wrist manipulation may offer some pain relief at short term, however, no evidence exists for their superiority against active treatments at mid-term and long-term follow-up. Most passive treatments are usually assessed in a combined treatment with other interventions, therefore, there is little evidence whether these passive approaches are effective as monotherapy.7 19 64 65
Exercise is a better option compared with corticosteroid injection(s)
We found a clinically significant benefit for the use exercise with or without physiotherapy and HEP compared with corticosteroid injections (very low and low certainty evidence) in patient-reported improvement and PFGS at all follow-up time points although elbow disability was statistically but not clinically improved. Increased clinical heterogeneity may explain this discrepancy: one study26 included patients with substantially greater disability scores. Note short-term effects for GROC, pain and PFGS had moderate statistical heterogeneity (I2: 66%, 62% and 56%, respectively), and mid-term effects for PFGS had considerable heterogeneity (I2=85%) so should be interpreted with caution.
Contemporary evidence from systematic reviews indicates that corticosteroid injections may provide short-term temporary pain relief in shoulder and lower limb tendinopathies66 67 and a significant (number needed to treat [NNT]=47), but transient pain reduction over placebo injections in LET.36 Patients who received corticosteroids presented worsening of pain symptoms at the long term compared with a WSP or physiotherapy68 or exacerbation of symptoms (number needed to harm [NNH]=1351) compared with placebo injections.28 36 56 69
Although corticosteroids are consistently shown as less beneficial for patients with LET, 9% of specialist surgeons and therapists report they would use corticosteroids as a first-line treatment,70 27% as a second-line option and 21% for most patients despite the objectively poor results—this research has clearly not yet translated to clinical practice.5 56 71
Exercise compared with WSP
In terms of clinical significance, exercise with or without physiotherapy or HEP were no better than WSP based on very low and low certainty evidence in all outcomes at all follow-up time points. Statistically significant differences with very low certainty evidence were presented in favour of exercise compared WSP only at short term in GROC, and short-term and long-term follow-up in pain and elbow disability. Quantitative synthesis showed that more than 76% of WSP and 81% of exercise patients self-rated themselves as recovered after mid-term follow-up (exercise NNT=20).
Moderate statistical heterogeneity in the short-term effects in mean change in pain rating, PFGS, and elbow disability (I2: 49%, 63% and 67%, respectively) along with significant clinical heterogeneity in PFGS and elbow disability baseline scores among studies indicates caution in the interpretation of these findings.
The ambiguous superiority of exercise with or without physiotherapy or HEP compared with a WSP in LET can be attributed to the interplay of several clinical domains among studies such as: severity and duration of symptoms, previous treatments, natural history, per-protocol or non-per-protocol cointerventions and exercise compliance. WSP with adequate advice and appropriate analgesic medication if required is reported to be effective for up to 85% of cases.26 36 43 We believe some caution is required before extrapolating this to suggest that most patients will benefit from WSP for two reasons. First, the effect of patient education61 72and additional use of non-protocol interventions, including medication, is not negligible in this group.26 36 However, three RCTs26 36 43 reported that about 15% of patients, regardless of allocation to WSP or an exercise group, failed to report successful outcomes after 1 year. Second, median symptom duration in studies examined was as high as 109 weeks—there are clearly patients for whom WSP has failed. Identifying those who are at risk of long-term disability early in their presentation would appear to be a priority.
There is little evidence comparing exercise with or without FSB compared with FSB alone
Quantitative synthesis in exercise with or without physiotherapy or HEP compared with FSB was not possible due to indirectness of interventions. Very low certainty evidence suggests that FSB does not produce clinically significant improvement compared with exercise with physiotherapy and HEP in all outcomes at all follow-up time points. Moreover, the addition of corticosteroids to FSB or to an HEP did not result in significant differences between groups at 1-year follow-up.31
The use of FSB or counterforce brace assumed that the applied compression over the common extensor muscle mass disperses stresses generated by muscle contraction, and that contact of the brace with the skin and underlying tissue may enhance neuromuscular performance.73 74 Despite the contradictory evidence for the effectiveness of FSB in pain and function compared with controls,75 these findings are not supported by our review at least for mid and long-term follow-up.
There is no benefit in using one exercise type compared with another, but the evidence is poor
Very low and low certainty of evidence suggests no difference between concentric combined with eccentric and eccentric exercise alone in mean change in pain rating, PFGS, and elbow disability in LET. Quantitative synthesis was conducted only for short-term follow-up presenting imprecision and inconsistency of the direction of treatment effects.33 39 45 51 Substantial missing outcome data and selective reporting in two studies with high risk of bias39 51 affected the validity of the results. Unbalanced cointerventions among comparators introduced significant imbalances in two studies33 51 where a wide range of additional treatments (stretching, ultrasound, transverse friction massage, heat, ice, FSB, self-ice massage, medication) were used according to the patient’s preferences.
Although eccentric loading regimes have been very popular among practitioners for the treatment of tendinopathies, evidence of their superiority compared with other loading programmes is inconclusive.76 77 A recent narrative review suggested that there is no convincing clinical evidence to demonstrate that isolated eccentric loading exercise improves clinical outcomes more than other loading approaches in tendinopathy.78 The findings of our review concur with this and findings in previous systematic reviews.7 8
Based on very low certainty evidence the addition of Pilates-based exercises is not recommended.29 The additional use of scapular strengthening exercises may provide statistically better results compared with eccentric exercise alone; however, the generalisability of these results should be interpreted with caution due to the limited evidence.42
Limitations and future research
We acknowledge as a limitation that we did not downgrade the evidence based on the pooled sample size and it is likely that chronicity is a confounder of outcomes and intervention success; however, we were unable to do a subgroup analysis on this domain. When one trial was available, the evidence was graded as low certainty and was subsequently downgraded in the case of high risk of bias. Although this was considered as the best available method, validity is lacking.19 79 Despite including 30 studies with over 2000 participants, there were no studies with a low risk of bias. Also, we made comparisons that were both direct and indirect which reflects effectiveness (clinical practice), but not efficacy. A random-effects model was used as recommended when direct comparisons are not available80; however, more precise estimates would be possible if direct comparisons were evident. Quantitative syntheses included a very low number of studies in most comparisons. For the same reason, we were unable to generate funnel plots to assess for publication bias.16 Finally, the lack of solid evidence for the MCIDs used for clinical significance should be considered in the clinical interpretation of our results.
We suggest future research should clearly describe all aspects of loading interventions prescribed and performed by participants. It is plausible that there is a differential response to loading interventions across the lifespan which have not yet been investigated. Further confounders likely include patient education, and the lack of standardised clinical diagnostic criteria. Although our primary outcome measures were in line with the core domains suggested by the International Scientific Tendinopathy Symposium Consensus,81 their increased number and variety were barriers for a direct interpretation of the findings. Future research will benefit from a core outcome set for comparison between studies and standardisation. Finally, the lack of documented MCID for grip strength needs to be addressed.
The available data reinforce the message that exercise should be considered the primary treatment for LET2; however, the magnitude of the effect is small compared with other conservative interventions. The paucity of research detail in terms of important exercise components such as load, dosage, time under tension, rest periods, acceptable pain level, as well as the considerable heterogeneity of equipment, duration and frequency were a barrier to provide clinical recommendations for the optimal exercise selection in LET. Only a few studies26 28 53 reported their exercise programme in enough detail to allow full replication without arbitrary assumptions on baseline load and progressions. Although the certainty of evidence is still low and very low (with a small effect), pooled results indicates exercise is effective for managing LET as compared with passive interventions with or without invasive treatment. Clinicians should thus recommend active (exercise) interventions in patients with LET including upper limb strengthening and stretching exercises as first-line treatment.
What is already known
Load management and different modes of exercise are the most popular treatment approach with or without additive interventions in patients with lateral elbow tendinopathy (LET).
The quality of evidence related to the effectiveness of exercise compared with other interventions have not been investigated.
What are the new findings
Exercise performs better than corticosteroid injections in all outcomes at all time points.
Exercise shows better outcomes than a wait-and-see policy only in self-perceived improvement at short-term, pain reduction and elbow disability at short-term and long-term follow-up.
No difference between concentric–eccentric and eccentric exercise alone was found.
Clinicians should recommend active (exercise) interventions in patients with LET including upper limb strengthening and stretching exercises as first-line treatment.
Contributors SK contributed in planning, conduct, data analysis, synthesis and reporting of the work. VK contributed in planning, conduct, data analysis, synthesis and reporting of the work. RW contributed in data synthesis and revising of the work. IV contributed in data extraction. SW contributed in revising the work. GG contributed in revising the work.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.