Autologous blood and platelet‐rich plasma injection therapy for lateral elbow pain

Teemu V Karjalainen; Michael Silagy; Edward O'Bryan; Renea V Johnston; Sheila Cyril; Rachelle Buchbinder

doi:10.1002/14651858.CD010951.pub2

Autologous blood and platelet‐rich plasma injection therapy for lateral elbow pain

Authors' declarations of interest

Version published: 30 September 2021 Version history

https://doi.org/10.1002/14651858.CD010951.pub2

Collapse all Expand all

Abstract

available in

Background

Autologous whole blood or platelet‐rich plasma (PRP) injections are commonly used to treat lateral elbow pain (also known as tennis elbow or lateral epicondylitis or epicondylalgia). Based on animal models and observational studies, these injections may modulate tendon injury healing, but randomised controlled trials have reported inconsistent results regarding benefit for people with lateral elbow pain.

Objectives

To review current evidence on the benefit and safety of autologous whole blood or platelet‐rich plasma (PRP) injection for treatment of people with lateral elbow pain.

Search methods

We searched CENTRAL, MEDLINE, and Embase for published trials, and Clinicaltrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal for ongoing trials, on 18 September 2020.

Selection criteria

We included all randomised controlled trials (RCTs) and quasi‐RCTs comparing autologous whole blood or PRP injection therapy to another therapy (placebo or active treatment, including non‐pharmacological therapies, and comparison between PRP and autologous blood) for lateral elbow pain. The primary comparison was PRP versus placebo. Major outcomes were pain relief (≥ 30% or ≥ 50%), mean pain, mean function, treatment success, quality of life, withdrawal due to adverse events, and adverse events; the primary time point was three months.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

Main results

We included 32 studies with 2337 participants; 56% of participants were female, mean age varied between 36 and 53 years, and mean duration of symptoms ranged from 1 to 22 months. Seven trials had three intervention arms. Ten trials compared autologous blood or PRP injection to placebo injection (primary comparison). Fifteen trials compared autologous blood or PRP injection to glucocorticoid injection. Four studies compared autologous blood to PRP. Two trials compared autologous blood or PRP injection plus tennis elbow strap and exercise versus tennis elbow strap and exercise alone. Two trials compared PRP injection to surgery, and one trial compared PRP injection and dry needling to dry needling alone. Other comparisons include autologous blood versus extracorporeal shock wave therapy; PRP versus arthroscopic surgery; PRP versus laser; and autologous blood versus polidocanol.

Most studies were at risk of selection, performance, and detection biases, mainly due to inadequate allocation concealment and lack of participant blinding.

We found moderate‐certainty evidence (downgraded for bias) to show that autologous blood or PRP injection probably does not provide clinically significant improvement in pain or function compared with placebo injection at three months. Further, low‐certainty evidence (downgraded for bias and imprecision) suggests that PRP may not increase risk for adverse events. We are uncertain whether autologous blood or PRP injection improves treatment success (downgraded for bias, imprecision, and indirectness) or withdrawals due to adverse events (downgraded for bias and twice for imprecision). No studies measured health‐related quality of life, and no studies reported pain relief (> 30% or 50%) at three months.

At three months, mean pain was 3.7 points (0 to 10; 0 is best) with placebo and 0.16 points better (95% confidence interval (CI) 0.60 better to 0.29 worse; 8 studies, 523 participants) with autologous blood or PRP injection, for absolute improvement of 1.6% better (6% better to 3% worse). At three months, mean function was 27.5 points (0 to 100; 0 is best) with placebo and 1.86 points better (95% CI 4.9 better to 1.25 worse; 8 studies, 502 participants) with autologous blood or PRP injection, for absolute benefit of 1.9% (5% better to 1% worse), and treatment success was 121 out of 185 (65%) with placebo versus 125 out of 187 (67%) with autologous blood or PRP injection (risk ratio (RR) 1.00; 95% CI 0.83 to 1.19; 4 studies, 372 participants), for absolute improvement of 0% (11.1% lower to 12.4% higher).

Regarding harm, we found very low‐certainty evidence to suggest that we are uncertain whether withdrawal rates due to adverse events differed. Low‐certainty evidence suggests that autologous blood or PRP injection may not increase adverse events compared with placebo injection. Withdrawal due to adverse events occurred in 3 out of 39 (8%) participants treated with placebo versus 1 out of 41 (2%) treated with autologous blood or PRP injection (RR 0.32, 95% CI 0.03 to 2.92; 1 study), for an absolute difference of 5.2% fewer (7.5% fewer to 14.8% more). Adverse event rates were 35 out of 208 (17%) with placebo versus 41 out of 217 (19%) with autologous blood or PRP injection (RR 1.14, 95% CI 0.76 to 1.72; 5 studies; 425 participants), for an absolute difference of 2.4% more (4% fewer to 12% more).

At six and twelve months, no clinically important benefit for mean pain or function was observed with autologous blood or PRP injection compared with placebo injection.

Authors' conclusions

Data in this review do not support the use of autologous blood or PRP injection for treatment of lateral elbow pain. These injections probably provide little or no clinically important benefit for pain or function (moderate‐certainty evidence), and it is uncertain (very low‐certainty evidence) whether they improve treatment success and pain relief > 50%, or increase withdrawal due to adverse events. Although risk for harm may not be increased compared with placebo injection (low‐certainty evidence), injection therapies cause pain and carry a small risk of infection. With no evidence of benefit, the costs and risks are not justified.

PICOs

Population

Intervention

Comparison

Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

available in

Autologous blood or PRP injection for lateral elbow pain

Background

Lateral elbow pain, also known as tennis elbow or lateral epicondylitis, is a degenerative (age‐related structural change of tissue) tendon disease at the site where forearm extensor muscles attach to the outer part of the elbow. It is a common cause of elbow pain and disability, typically in middle‐aged people.

Autologous blood and platelet‐rich plasma (PRP) injections have been suggested to improve tendon healing. Autologous blood is derived from the person's own venous blood sample (blood taken from a vein), and PRP is a concentrate of plasma and platelets isolated from autologous blood.

This study aimed to review evidence regarding the benefits and harms of autologous blood or PRP injection for treatment of lateral elbow pain.

Study characteristics

We searched MEDLINE, CENTRAL, Embase, Clinicaltrials.gov, and WHO trials registries, unrestricted by date or language, on 18 September 2020.

We included 32 trials with 2337 participants. Mean age varied between 36 years and 53 years, and mean duration of symptoms ranged from 1 month to 22 months. Of 21 studies that reported gender, 56% of participants were female. Among the included studies, three studies were funded by manufacturers of the PRP centrifugation system; two studies were provided PRP kits; and one study received funding from PRP kit manufacturers.

Key findings

Comparison with placebo at three months revealed the following.

Pain (lower scores mean less pain) (8 studies, 523 participants; moderate‐certainty evidence).

Pain improved by 2% (3% worse to 6% better), or by 0.16 points on a zero to 10 scale.

• People who had placebo rated their pain as 3.7 points.

• People who had autologous blood or PRP injection rated their pain as 3.9 points.

Function (0 to 100; lower scores mean better function or less disability) (8 studies, 502 participants; moderate‐certainty evidence).

Function improved by 2% (5% better to 1% worse), or by 2 points on a zero to 100 scale.

• People who had placebo rated their function as 27 points.

• People who had autologous blood or PRP injection rated their function as 29 points.

Treatment success (4 studies, 372 participants; very low‐certainty evidence).

0% more people rated their treatment a success (11% fewer to 12% more), or zero more people out of 100.

• 65 out of 100 people considered treatment as successful after placebo injection.

• 67 out of 100 people considered treatment as successful after autologous blood or PRP injection.

Health‐related quality of life (higher scores mean better quality of life).

None of the studies measured this outcome.

Pain relief (> 30% or > 50%).

None of the studies reported this outcome at three months.

Withdrawals due to adverse events (1 study, 80 participants; very low‐certainty evidence).

5% fewer people withdrew from the study because of an adverse event (7.5% fewer to 14.8% more), or 5 fewer people out of 100.

• 7 out of 100 people withdrew from the study due to an adverse event after placebo injection.

• 2 out of 100 people withdrew from the study due to an adverse event after autologous blood or PRP injection.

Adverse events (typically transient injection site pain) (5 studies, 425 participants; low‐certainty evidence).

2% more people had adverse events (4% fewer to 11% more), or 2 more people out of 100.

• 17 out of 100 people reported adverse events after placebo injection.

• 19 out of 100 people reported adverse events after autologous blood or PRP injection.

Certainty of the evidence

For people with lateral elbow pain, moderate‐certainty evidence (downgraded for bias, i.e. methodological shortcomings in the included studies) shows that autologous blood or PRP injection probably provides little or no clinically important benefit for pain or function compared with placebo injection, and low‐certainty evidence (downgraded due to risk of bias, i.e. methodological shortcomings; and imprecision, i.e. too few data to estimate the precise difference) suggests that autologous blood or PRP injection may not cause higher risk for adverse events. We are uncertain whether autologous blood or PRP injection is associated with a higher proportion of people reporting treatment success, or if this treatment increases withdrawals due to adverse events.

Authors' conclusions

Implications for practice

Data in this review do not support the use of autologous blood or PRP injection for treatment of lateral elbow pain. These treatments probably provide little or no benefit for pain or function, and it is uncertain whether they improve treatment success or increase withdrawal due to adverse events. Although risk for harm may not be increased compared with placebo injection, there is always a small risk of infection and pain related to injection therapies, and as long as no evidence shows benefit, the costs or potential harms, even if minimal, are not justified.

Most of the participants in the included studies assessed their pain as low (< 3 on a 0 to 10 scale) after placebo injection. This is in line with the known benign natural course of the condition. However, patients with lateral elbow pain could have pain and disability that persist for a long time.

Implications for research

Future trials should consider comparing PRP injection only to placebo injection and should follow rigorous research standards to minimise the risk of bias. As long as no solid evidence is available on the efficacy of PRP compared to placebo, comparison to other treatment modalities provides little value.

The data in this review do not provide any viable hypotheses about whether some subgroups of people or some variety in treatment regimens (e.g. multiple injections) or in PRP preparations would yield more favourable outcomes.

Regarding pain and function, the included studies followed up more than 500 participants, findings were robust to selection and detection biases, and it is unlikely that new trials would show clinically important benefit in these outcomes for up to 6 months. However, at later time points, new studies may affect the estimates, although a biological rationale is missing for the late onset of possible effects.

Given that results at 12 months show imprecision, future trials could follow up with participants up to 1 year to improve the certainty of estimates for longer follow‐up. Trialists should consider using core outcome sets proposed for tendinopathy trials to facilitate aggregation of data in future meta‐analyses (Vicenzino 2020).

Summary of findings

Open in table viewer

Summary of findings 1. Autologous blood or PRP versus placebo at 3 months' follow‐up

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№. of participants (studies)	Quality of the evidence (GRADE)	Comments
Autologous blood or PRP versus placebo at 3 months' follow‐up
Patient or population: lateral elbow pain Setting: outpatient Intervention: autologous blood or PRP injection Comparison: placebo
Outcomes	Risk with placebo	Risk with autologous blood or PRP injection	Relative effect (95% CI)	№. of participants (studies)	Quality of the evidence (GRADE)	Comments
Pain (VAS, PRTEE) translated to 0 to 10, where 0 is no pain Follow‐up: 3 months	Mean pain in the placebo group was 3.7 points^a	Mean pain was 0.16 points better (0.60 better to 0.29 worse)	‐	523 participants (8 studies)	⊕⊕⊕⊝ Moderate^b	PRP probably provides little to no benefit for pain. Absolute benefit 1.6% better (6% better to 3% worse); relative benefit 2.3% better (9% better to 4% worse).^c Not clinically significant
Function (PRTEE, DASH, MMCPIE, Roles‐Maudsley), translated to 0 to 100, where 0 is best function, or no disability Follow‐up: 3 months	Mean function in placebo was 27.5 points^d	Mean function was 1.86 points better (4.97 better to 1.25 worse)	‐	502 participants (8 studies)	⊕⊕⊕⊝ Moderate^d	PRP probably provides little to no benefit for function. Absolute benefit 1.9% better (5% better to 1% worse); relative benefit 4% (11% better to 3% worse).^e Not clinically significant
Treatment success (> 25% improvement in pain or function) Follow‐up: 3 months	650/1000	670/1000 (582 to 765)	RR 1.0 (0.83 to 1.19)	372 participants (4 studies)	⊕⊝⊝⊝ Very low^b,e,f	We are uncertain whether PRP provides better treatment success. Absolute benefit 0% higher (11.1% lower to 12.4% higher); relative benefit 0% higher (17% lower to 19% higher)
Health‐related quality of life Not measured	See comment	See comment	‐	(0 studies)	See comment	Not measured in any of the included studies
Pain relief ≥ 30% or ≥ 50% Not measured at 3 months	See comment	See comment	‐	(0 studies)	See comment	Not reported in any of the included studies at 3 months
Withdrawal due to adverse events	77/1000	24/1000 (2 to 225)	RR 0.32 (0.03 to 2.92)	80 participants (1 study)	⊕⊝⊝⊝ Very low^b,g	We are uncertain whether PRP results in more people withdrawing due to adverse events. Absolute change 5.2% less (7.5% less to 14.8% more); relative change 68% less (97% less to 192% more)
Adverse events (pain and swelling at injection site and limitation of elbow movement following injection) Follow‐up: 12 months	168/1000	192/1000 (128 to 290)	RR 1.14 (0.76 to 1.72)	425 participants (5 studies)	⊕⊕⊝⊝ Low^b,f	PRP may not increase the number of people reporting adverse events. Absolute change 2.4% more (4% less to 12% more); relative change 14% more (24% less to 72% more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; DASH: Disabilities of the Arm, Shoulder and Hand; MMCPIE: Modified Mayo Clinic Performance Index for Elbow; OR: odds ratio; PRP: platelet‐rich plasma; PRTEE: Patient‐Rated Tennis Elbow Evaluation; RR: risk ratio; VAS: visual analogue scale.
GRADE Working Group grades of evidence. High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aMedian pain value from placebo groups in the included studies (excluding Mishra 2014, which reported percentage improvement). ^bDowngraded one level for risk of bias in the included studies. ^cRelative changes calculated relative to baseline in control group (i.e. mean difference divided by mean at baseline in the placebo group) (from Montalvan 2015 ‐ value for pain was 7 points on a 0 to 10 scale; for function from Krogh 2013 ‐ value was 47 points on a 0 to 100 scale). Absolute change calculated as mean difference divided by scale of the instrument, expressed as percentage. ^dMedian function from placebo groups at 3 months' follow‐up. ^eDowngraded one level for indirectness, as none of the studies measured global participant‐reported success directly but measured pain or function improvement cutoff values. ^fDowngraded one level for imprecision due to 95% CIs including both effect and no effect. ^gDowngraded evidence by two levels because of a small number of events leading to very wide confidence intervals, which overlap relative risk estimates of 0.75 and 1.25.

Background

Description of the condition

Lateral elbow pain is described by many analogous terms in the literature, including tennis elbow, lateral epicondylitis (or epicondylosis), rowing elbow, tennis elbow, lateral epicondylitis, tendonitis of common extensor origin, extensor tendinopathy, and peritendinitis of the elbow. For the purposes of this review, and in keeping with previous Cochrane systematic reviews for this condition, we will use the term 'lateral elbow pain'.

Lateral elbow pain is a common condition that causes pain in the lateral elbow and forearm. It affects 1% to 3% of the general population and up to 15% of workers in at‐risk industries, and is a common sports injury (Hume 2006; Ranney 1995; Walker‐Bone 2004). Men and women appear to be affected equally. The annual incidence in general practice is 4 to 7 per 1000 person‐years, with an incidence of 11 per 1000 person‐years in the 40 to 60‐year age group ‐ the age group most affected (Bot 2005).

Lateral elbow pain is thought to be an overuse injury at the common extensor origin at the lateral epicondyle. Histological studies have identified the presence of angiofibroblastic hyperplasia (fibroblast proliferation, vascular hyperplasia, and disorganised collagen) (Nirschl 1979). Although no histological studies of acute lesions are available, the presence of typical inflammatory symptoms such as night pain and early morning stiffness suggests there may be an early inflammatory component. In spite of the title 'tennis elbow', tennis is a direct cause in only 5% of cases. Other risk factors include repetitive wrist turning and hand gripping. People in strenuous occupations that involve repetitive use are at increased risk.

People with lateral elbow pain typically present with insidious onset of worsening pain and tenderness over the lateral epicondyle. Repetitive movement, lifting, and gripping often aggravate the pain. Examination findings include localised tenderness over the common extensor origin at the lateral epicondyle and elicitation of pain on resisted dorsiflexion of the wrist, middle finger, or both.

Acute lateral elbow pain usually lasts 6 to 12 weeks and often results in work absence (Mallen 2009). For most, it is a self‐limiting condition, but some episodes may persist for up to two years. One study found that 80% of patients with elbow pain already lasting longer than four weeks recovered after one year without any specific treatment (Bisset 2006). Prognostic factors at least moderately associated with poorer outcomes at one year include previous occurrence, high physical strain at work, a manual job, high baseline levels of pain and/or distress, and less social support. Depression and ineffective coping skills have also been found to strongly predict disability (Alizadehkhaiyat 2007). An ultrasound study determined that the presence of a lateral collateral ligament tear or of large (≥ 6 mm) intrasubstance tears was associated with poorer outcomes, but no relationship between tendon thickness or neovascularity and outcomes was seen (Clarke 2010).

Although lateral elbow pain is generally a self‐limiting condition, it results in significant disability, increased healthcare utilisation, lost productivity, and increased costs (Silverstein 2006). Therefore, treatment that shortens the duration of symptoms and disability has the potential to be of significant value in terms of reduced morbidity and costs to both the individual and the community.

Although many treatments are available for lateral elbow pain, the optimal evidence‐based treatment remains unclear. Currently used treatments include topical and oral non‐steroidal anti‐inflammatory drugs (Pattanittum 2013), orthotic devices (Borkholder 2004; Struijs 2002), physiotherapy modalities such as deep friction massage, exercise, and laser and ultrasound therapy (Bisset 2005; Bjordal 2008; Herd 2008; Kohia 2008; Smidt 2003), glucocorticoid injection (Assendelft 1996; Coombes 2010; Smidt 2002b), extracorporeal shock wave therapy (Buchbinder 2005), acupuncture (Green 2002), and surgery (Buchbinder 2011; Lo 2007). Less than 10% of patients with lateral epicondylitis undergo surgery (Nirschl 1979).

Description of the intervention

Autologous whole blood injection involves collection of the patient's blood, which is then injected directly back into the area of tendinopathy. Platelet‐rich plasma (PRP) injection, sometimes referred to as autologous conditioned plasma (ACP), or platelet concentrate, is a treatment by which platelet‐rich centrifuged blood is injected into the affected tendon (Kampa 2010). Autologous conditioned serum (ACS) is another type of autologous blood preparation that can be used. ACS differs from PRP in that it has a higher concentration of anti‐inflammatory cytokines, particularly naturally occurring interleukin‐1 receptor antagonists (IL‐1Ras), rather than platelets (Evans 2016).

No standardised nomenclature or method of preparation has been adopted for autologous blood products. Different classification systems have been proposed for comparison between different PRP preparations. One of the most widely reported is the PAW (Platelets, Activation, White cells) classification system, which is based on (1) absolute numbers of platelets, (2) the manner in which platelet activation occurs, and (3) the presence or absence of white cells in the injectable product (DeLong 2012). More recent classification systems incorporate additional measures, including concentration of red blood cells, the preparation method, and use of imaging‐guided injection (Lana 2017).

Little consensus has been reached on the optimal preparation process for autologous blood products. Centrifugation time and speed can vary, as can the volume of blood extracted and injected back into the body, as well as platelet and white blood cell content (Bennell 2017; Mautner 2015). PRP can be injected into the tendon without further treatment immediately after spinning, or it can be frozen and stored for later use (Kampa 2010). Frozen storage of PRP provides convenience when serial injections are used, but the act of freezing and thawing may have physiological effects on the blood product that alter its efficacy (Bennell 2017). Other modifications of the intervention include the addition of activating factors such as calcium to further enhance the release of cytokines and growth factors (Wehling 2007), or dry needling to cause fresh injury to the tendon.

The procedure is simple to perform, and theoretically at least, adverse effects, such as temporary pain or stiffness following the injection, should be minor (Kampa 2010).

How the intervention might work

Autologous whole blood or PRP injection has been proposed as treatment for chronic non‐healing tendon injuries including lateral epicondylitis. The rationale of action is based upon the hypothesis that platelets would release high concentrations of platelet‐derived growth factors and cytokines to stimulate angiogenesis and healing (Edwards 2003; Engebretsen 2010; Samson 2008; Suresh 2006).

Although platelets have traditionally been thought to be involved exclusively with haemostasis at sites of vascular injury, they are now known to play a role in tissue regeneration and healing through release of an abundant array of cytokines and growth factors such as transforming growth factor‐beta, vascular endothelial growth factor, platelet‐derived growth factor, and epithelial growth factor (Eppley 2004). These growth factors are known to be important in tissue regeneration and healing (Lee 2013). One study showed that injection of autologous blood into rabbit patellar tendons resulted in significantly stronger tendons than with non‐injection, although no histological differences were identified after 12 weeks (Taylor 2002).

Why it is important to do this review

Autologous whole blood and PRP have been used for over 20 years in a variety of surgical situations to reduce blood loss (Carless 2011); recently these modalities have been used to promote wound and bone healing (Griffin 2012; Martinez‐Zapata 2012; Martinez‐Zapata 2013; Samson 2008), as well as to treat chronic tendinopathy (Bell 2013; De Vos 2010). However, few rigorous controlled trials have been reported.

Based on a review of the procedure in 2009, the UK National Institute for Health and Clinical Excellence (NICE) stated that current evidence on the safety and efficacy of autologous blood injection for tendinopathy is inadequate in quantity and quality (NICE 2013). This statement was reiterated in a systematic reviews of the evidence (De Vos 2010; Kampa 2010), and in a 2010 International Olympics Committee consensus paper on use of PRP in sports medicine (Engebretsen 2010).

Several randomised studies have compared autologous blood or PRP injection with various treatments, with conflicting results. These products are used increasingly despite the lack of sound evidence supporting their efficacy and safety. This review is timely in seeking to determine whether further research is needed, and in assessing the value of these therapies for this condition.

Objectives

To review current evidence on the benefit and safety of autologous whole blood or platelet‐rich plasma (PRP) injection for treatment of people with lateral elbow pain.

Methods

Criteria for considering studies for this review

Types of studies

We included studies described as randomised controlled trials (RCTs) and trials describing quasi‐randomised methods of participant allocation. We included studies reported as full text, those published as abstract only, and unpublished data. We used no language or date restrictions.

Types of participants

We included adult participants (> 16 years) with lateral elbow pain as defined by trial authors. These criteria may include clinical features such as pain that is maximal over the lateral epicondyle and pain that is reproduced by tests including palpation of the lateral epicondyle or the common extensor origin of the elbow, gripping, resisted wrist, or second or third finger extension (dorsiflexion), as well as imaging results such as ultrasound or magnetic resonance imaging (MRI) showing the presence of focal hypo‐echoic areas or frank tears or alterations in the normal fibrillary pattern in the common extensor origin. However, studies that did describe particular features of lateral elbow pain were still eligible for inclusion.

In addition, we included participants with tendonitis at other sites, provided lateral elbow pain results were presented separately, or at least 90% of participants in the trial had lateral elbow pain.

We excluded participants with lateral elbow pain due to acute traumatic injury.

Types of interventions

Interventions: autologous whole blood, platelet‐rich plasma (PRP), or other autologous blood products including autologous conditioned serum.
Comparators included:
- placebo;
- no treatment;
- exercise and other physical therapy interventions including braces and orthotics;
- other injections (including glucocorticoid injection, hyaluronic acid injection, or cell‐based therapies such as stem cell therapy);
- surgical interventions;
- drug therapy (including analgesics and non‐steroidal anti‐inflammatory drugs); and
- supplements and complementary therapies.

Co‐interventions were eligible for inclusion provided they were applied equally in all treatment groups.

Trials that assess the additional benefit of platelet‐rich plasma or other autologous blood products in a surgical procedure compared to surgery alone will be excluded.

Types of outcome measures

There is considerable variation in the outcome measures reported in clinical trials of interventions for pain. However, there is general agreement that outcome measures of greatest importance to patients should be considered, and people with lateral elbow pain typically suffer from pain as suggested by the name of the condition.

The Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT) has published consensus recommendations for determining clinically important changes in outcome measures in clinical trials of interventions for chronic pain (Dworkin 2008). Reductions in pain intensity ≥ 30% and ≥ 50% reflect moderate and substantial clinically important differences, respectively, and it is recommended that the proportion of patients that respond with these degrees of pain relief should be reported.

NICE has recommended that trials of tendinopathy include functional and quality of life outcomes with minimum follow‐up of one year (NICE 2013).

Major outcomes

Participant‐reported pain relief: proportion reporting pain relief of 30% or greater, or 50% or greater
Mean pain or mean change in pain score on a visual analogue scale or a numerical rating scale, or subscore of a total function score, or other measure
Function/disability as measured by disease‐specific disability measures such as the Patient‐Rated Tennis Elbow Evaluation (PRTEE) questionnaire (Rompe 2007), or the upper limb‐specific Disabilities of the Arm, Shoulder and Hand (DASH) outcome questionnaire (Gummesson 2003), or other measure
Participant's perception of overall effect or success, as measured by a global rating of treatment satisfaction such as the Patient Global Impression of Change (PGIC) scale, or of overall treatment success, as defined in the trials (e.g. includes proportion without elbow pain; proportion with 25% pain or disability reduction)
Health‐related quality of life as measured by either generic measures (such as components of Short Form‐36 (SF‐36)) or disease‐specific tools
Proportion of withdrawals due to adverse events
Proportion with any adverse event

Minor outcomes

Other pain measures including proportion achieving pain score below 30/100 mm on a visual analogue scale (VAS); PGIC in pain much or very much improved
Grip strength (preferably pain‐free maximum grip strength)
Proportion with serious adverse events (defined as adverse events that are fatal, are life‐threatening, or require hospitalisation)

Timing of outcome assessment

For the purpose of this review, if multiple time points were reported, we grouped outcomes up to three weeks, greater than three weeks and up to six weeks, over six weeks to three months, over three months to six months, over six months to a year, and more than a year. If trials included outcomes at more than one time point within these time periods, we extracted the latest time point. Adverse event data were extracted at the end of the trials. Our primary time point was over six weeks to three months.

Search methods for identification of studies

Electronic searches

We searched the following electronic databases, unrestricted by date or language, on 18 September 2020.

Cochrane Central Register of Controlled Trials (CENTRAL) (via Ovid) (Appendix 1).
MEDLINE (Ovid 1946 to present) (Appendix 2).
Embase (Ovid 1947 to present) (Appendix 3).
Clinical trials registers such as ClinicalTrials.gov (http://clinicaltrials.gov/) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/), for ongoing trials (Appendix 4 Appendix 5).

Searching other resources

We screened reference lists of retrieved review articles and trials to identify potentially relevant studies.

Data collection and analysis

Selection of studies

Two review authors (TK, SC) independently reviewed the search results to identify trials that appeared to fulfil our inclusion criteria. All articles selected by at least one of the review authors were retrieved for closer examination. Review authors were not blinded to the journal nor the authors. Disagreement about inclusion or exclusion of individual studies was resolved by consensus, or if consensus was not reached, by a third review author (RJ).

Data extraction and management

Two review authors (TK, SC) extracted the following data from the included trials and resolved any differences by consensus.

Trial characteristics including size and location of the trial, and source of funding.
Characteristics of the study population including age and comorbidities.
Characteristics of autologous whole blood or PRP injection therapy such as dose and frequency of injections, schedule of treatment, total number of treatment sessions.
Characteristics of autologous blood product preparation and injection protocols, including a description of the centrifugation protocol (speed and time) and the number of centrifugations, use and type of activating agents, use of frozen or fresh PRP, leukocyte rich or poor, and injection characteristics (such as volume injected, frequency and total number of injections, injection approach, use of local anaesthetic and imaging such as ultrasound).
Characteristics of control interventions.
Risk of bias domains as outlined in Assessment of risk of bias in included studies.
Outcome measures: measurement scale and direction of the scale, mean and standard deviation, number of participants per treatment group for continuous outcomes (such as mean pain, function, quality of life), number of events and number of participants per treatment group for dichotomous outcomes (such as proportion with 30% or greater pain relief, treatment success, withdrawal due to adverse events, adverse events), as outlined in Types of outcome measures.

We noted in the Characteristics of included studies tables whether outcome data were not reported in a form suitable for meta‐analysis, and when missing data were calculated or estimated from a graph or were imputed.

Our a priori decision rules to extract data in the event of multiple outcome reporting in trials are as follows.

When trialists report both final values and change from baseline values for the same outcome, we extracted final values.
When trialists report both unadjusted and adjusted‐for‐baseline values for the same outcome, we extracted adjusted values.
When trialists reported data analysed based on the intention‐to‐treat (ITT) sample and another sample (e.g. per‐protocol, as‐treated), we extracted ITT‐analysed data.
For cross‐over RCTs, we extracted data from the first period only.

When trials did not include a measure of overall pain but included one or more other measures of pain, for the purpose of pooling data, we combined overall pain with other types of pain in the following hierarchy: unspecified pain; pain with activity; daytime pain. For disability, the hierarchy was Patient‐Rated Tennis Elbow Evaluation (PRTEE) questionnaire (Rompe 2007), followed by upper limb‐specific Disabilities of the Arm, Shoulder and Hand (DASH) outcome questionnaire (Gummesson 2003), then other measures. When studies used scales in the opposite direction to PRTEE (0 = worst function), we changed the direction of scores to ensure consistency in interpretation of results.

When multiple time points were reported within our time frames (up to six weeks; over six weeks to three months; over three months to six months; over six months to a year; over one year), we extracted the latest time point (e.g., if data were reported at four weeks, five weeks, three months, and six months, we extracted outcomes at five weeks, three months, and six months).

Assessment of risk of bias in included studies

Two review authors (TK, SC) assessed the risk of bias of each included trial and resolved any disagreements by consensus, and if consensus was not reached, by consultation with a third review author (RJ).

We assessed the following methodological domains, as recommended by Cochrane (Higgins 2017c).

Sequence generation (to determine if the method of generating the randomisation sequence was adequate, such as random‐number tables, computer‐generated random numbers, minimisation, coin tossing, shuffling of cards, and drawing of lots).
Allocation sequence concealment (to determine if adequate methods were used to conceal allocation, such as central randomisation and use of sequentially numbered, sealed, opaque envelopes).
Blinding of participants and personnel.
Blinding of outcome assessors for subjective self‐reported outcomes such as pain and function.
Blinding of outcome assessors for objective outcomes.
Incomplete outcome data.
Selective outcome reporting.
Other potential threats to validity, such as inappropriate analysis in cross‐over trials, baseline imbalance, inappropriate administration of an intervention (or co‐intervention), contamination, inappropriate interim analysis.

Each of these criteria was explicitly judged as having low risk of bias, high risk of bias, or unclear risk of bias (either lack of information or uncertainty over the potential for bias). We considered blinding of objective outcomes separately from blinding of subjective participant‐reported outcomes (e.g. pain, function). We presented figures generated by the risk of bias tool to provide summary assessments of the risk of bias.

Measures of treatment effect

When possible, analyses were based on ITT data (outcomes provided for every randomised participant) from individual trials. For each trial, we presented outcome data as point estimates with mean and standard deviation for continuous outcomes, and as risk ratios (RRs) with corresponding 95% confidence intervals (CIs) for dichotomous outcomes.

For continuous data, results were presented as mean differences (MDs). However, when different scales were used to measure the same outcome or concept, standardised mean differences (SMDs) were used. SMD was re‐expressed as a mean difference on a typical scale (e.g. 0 to 10 for mean pain) by multiplying by a typical among‐person standard deviation (e.g. standard deviation of the control group at baseline from the most representative trial) (Schünemann 2017b). We entered data presented as a scale with a consistent direction of effect across studies.

In the Effects of interventions results section and the 'Comments' column of the 'Summary of findings' table, we provided absolute and relative per cent differences and the number needed to treat for an additional beneficial outcome (NNTB), or the number needed to treat for an additional harmful outcome (NNTH) (NNTB or NNTH was provided only when the outcome showed a clinically significant difference). For dichotomous outcomes, NNTB or NNTH was calculated from the control group event rate, and relative risk using the Visual Rx NNT calculator (Cates 2008). NNTB or NNTH for continuous measures was calculated using the Wells calculator (available at the Cochrane Musculoskeletal Group (CMSG) Editorial Office) (http://musculoskeletal.cochrane.org/).

For dichotomous outcomes, the absolute per cent change was calculated from the difference in risks between intervention and control groups using GRADEpro (GRADEpro GDT 2015), and was expressed as a percentage. The relative per cent change for dichotomous data was calculated as risk ratio ‐ 1 and was expressed as a percentage. For continuous outcomes, the absolute difference was calculated as the mean difference between intervention and control groups in original measurement units, and was also expressed as a percentage (percentage of the measurement scale); the relative difference was calculated as the absolute change (MD) divided by the baseline mean of the control group from a representative trial. We assumed a minimal clinically important difference (MCID) of 1.5 points on a 10‐point continuous pain scale, and 10 points on a 100‐point scale, for function or disability for input into the calculator (Gummesson 2003).

Unit of analysis issues

When multiple trial arms were reported in a single trial, we included only the relevant arms but reported that there were multiple trial arms in the Characteristics of included studies table. If two comparisons from a three‐arm trial (e.g. PRP regimen 1 versus PRP regimen 2) were combined in the same meta‐analysis, we combined the two treatment groups if both regimens were relevant, and we compared the combined treatment group to the placebo group in the usual way.

If we identified trials that injected both forearms but trialists reported outcomes per participant without accounting for the bilateral correlation, we planned to report results from one arm when possible. If we were unable to obtain the data for a single arm, or to adjust the outcome data, we planned to include data as reported by trialists and to comment on the validity of such analyses, and to assess the effects of including such data by performing sensitivity analyses. For a cross‐over design, we planned to include data only from the first treatment episode.

If two comparisons (e.g. autologous whole blood versus placebo and PRP versus placebo) from one trial were combined in the same meta‐analysis, we halved the placebo group to avoid double‐counting.

Dealing with missing data

When data were missing or incomplete, we sought further information from the study authors.

In cases where individuals were missing from the reported results, we assumed the missing values to have a poor outcome. For dichotomous outcomes that measured adverse events (e.g. number of withdrawals due to adverse events), we calculated the withdrawal rate by using the number of patients who received treatment as the denominator. For dichotomous outcomes that measured benefits (e.g. proportion of subjects with 30% or greater reduction in pain), we calculated the proportion using the number of randomised subjects as the denominator. For continuous outcomes (e.g. pain), we calculated MD or SMD based on the number of patients analysed at the time point. If the number of patients analysed was not presented for each time point, we used the number of randomised patients in each group at baseline.

When possible, we computed missing standard deviations from other statistics such as standard errors, confidence intervals, or P values, according to the methods recommended in the Cochrane Handbook for Systematic Reviews of Interventions. If standard deviations could not be calculated, they were imputed (e.g. from other studies in the meta‐analysis) (Higgins 2017a; Higgins 2017b).

Assessment of heterogeneity

We assessed clinical and methodological diversity in terms of participants, interventions, outcomes, and study characteristics for the included studies to determine whether a meta‐analysis would be appropriate. We did this by observing these data from the data extraction tables. We assessed statistical heterogeneity by visually inspecting the forest plot to assess for obvious differences in results between studies, and by using I² and Chi² statistical tests. As recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017), interpretation of an I² value of 0% to 40% might 'not be important'; 30% to 60% may represent 'moderate'
heterogeneity; 50% to 90% may represent 'substantial' heterogeneity; and 75% to 100% represents 'considerable' heterogeneity. As noted in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017), we
considered that the importance of I² depends on (1) magnitude and direction of effects and (2) strength of evidence for heterogeneity. The Chi² test with a P value ≤ 0.10 was interpreted as indicating evidence of statistical heterogeneity. If we identified substantial heterogeneity, we reported this and investigated possible causes by following the recommendations provided in Section 9.6 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017).

Assessment of reporting biases

To determine whether reporting bias was present, we determined whether the protocol of the trial was published before patients were recruited for the study. For studies published after 1 July 2005, we screened the WHO ICTRP search portal, as described in Electronic searches. We checked trial protocols against published reports to evaluate whether selective reporting of outcomes was present (outcome reporting bias).

We planned to create and examine a funnel plot to explore possible small‐study biases and to examine the different possible reasons for funnel plot asymmetry, as outlined in Section 10.4 of the Cochrane Handbook for Systematic Reviews of Interventions and to relate this to review results (Sterne 2017). We compared the fixed‐effect estimate against the random‐effects model in the primary analyses to assess the possible presence of small‐sample bias in the published literature (i.e. in which the intervention effect is more beneficial in smaller studies). In the presence of small sample bias, the random‐effects estimate of the intervention is more beneficial than the fixed‐effect estimate. We planned to undertake formal statistical tests to investigate funnel plot asymmetry when more than 10 studies were included in a single meta‐analysis.

Data synthesis

For studies with similar participant and intervention characteristics and a common comparator, we pooled outcomes in a meta‐analysis using the random‐effects model as a default, and we performed a sensitivity analysis with the fixed‐effect model.

Because all blood products contain similar active biological factors although in different concentrations, it is likely that the mode of action is similar. We therefore elected to combine data in a single comparison, irrespective of whether the trial evaluated autologous blood or PRP. However, we did perform subgroup analyses to compare results for different blood products (as below).

Our main comparison was autologous or PRP versus placebo. Other comparisons included the following.

Autologous blood or PRP injection versus glucocorticoid injection.
PRP and dry needling versus dry needling alone.
PRP versus autologous blood.
Autologous blood or PRP versus extracorporeal shock wave therapy (ESWT).
PRP versus surgery.
Autologous blood plus tennis elbow strap versus exercise and tennis elbow strap.
PRP versus laser.
Autologous blood versus polidicanol injection.

Subgroup analysis and investigation of heterogeneity

We planned to carry out the following subgroup analyses to assess whether pain and function differ between the following groups at the primary time point of three months.

Participants who receive whole blood compared to those who receive PRP or autologous conditioned serum.
Participants with a lateral collateral ligament tear or a large (≥ 6 mm) intrasubstance tear compared to participants without these tears.
Use of freshly prepared versus frozen autologous blood product.
Use of leukocyte‐rich versus leukocyte‐depleted autologous blood product.

We used the formal test for subgroup interactions in Review Manager (RevMan 2014), and we applied caution in interpreting subgroup analyses, as advised in Section 9.6 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017). We compared the magnitude of effects between subgroups by assessing the overlap of CIs of the summary estimate. Non‐overlap of CIs indicated statistical significance.

Sensitivity analysis

We conducted a sensitivity analysis to assess the robustness of treatment effects of pain and function with regard to selection and detection biases, by excluding trials with potential for selection (inadequate or unclear random sequence generation or allocation concealment) and detection (unclear or inadequate participant blinding) bias from the meta‐analysis at the primary time point (three months for placebo; six weeks and six months for glucocorticoid comparisons).

Summary of findings and assessment of the certainty of the evidence

We presented the main results of the review in a 'Summary of findings' table, which provides key information concerning quality of evidence, magnitude of effect of interventions examined, and the sum of available data on outcomes (proportion reporting pain relief ≥ 30% or ≥ 50%; mean (or mean change in) pain; function; treatment success; health‐related quality of life; withdrawals due to adverse events; proportion of participants with adverse events). The comparison in the 'Summary of findings' table shows autologous blood or PRP injection versus placebo at three months.

Two people (TK, SC) independently used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to studies that contributed data to meta‐analyses for prespecified outcomes, and reported the quality of evidence as high, moderate, low, or very low. We used methods and recommendations described in Sections 8.5 and 8.7, and Chapters 11 and 12, of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017; Schünemann 2017a; Schünemann 2017b). We used GRADEpro software to prepare the 'Summary of findings' table (GRADEpro GDT 2015).

We justified all decisions to downgrade the quality of studies by using footnotes and made comments to aid the reader's understanding of the review when necessary. We provided absolute per cent difference and relative per cent change from baseline and, for outcomes with statistically significant differences between intervention groups, the number needed to treat for an additional beneficial or harmful outcome (NNTB or NNTH) in the 'Comments' column of the 'Summary of findings' table, as described in the Measures of treatment effect section above.

Interpreting results and reaching conclusions

We followed the guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Chapter 12) for interpreting results (Schünemann 2017b), and we were aware of distinguishing lack of evidence of effect from lack of effect. We based our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. We avoided making recommendations for practice; our implications for research suggest priorities for future research and outline remaining uncertainties in the area.

Results

Description of studies

Results of the search

The search, which was conducted up to 18 September 2020, yielded 350 records across databases (MEDLINE = 68; Embase = 111; CENTRAL = 117; Clinicaltrials.gov = 16; WHO ICTRP = 38). One additional record was identified by screening the reference lists of previously published systematic reviews. After duplicates were removed, 210 unique records remained. Of these, we retrieved 75 for full‐text screening on the basis of title and abstract. We deemed 32 trials eligible for inclusion (Arik 2014; Behera 2015; Branson 2016; Creaney 2011; Dojode 2012; Gautam 2015; Gedik 2016; Gosens 2011; Gupta 2019; Jindal 2013; Kazemi 2010; Krogh 2013; Lebiedziński 2015; Lim 2017; Linnanmäki 2020; Martin 2019; Martínez‐Montiel 2015; Merolla 2017; Mishra 2014; Montalvan 2015; Omar 2012; Ozturan 2010; Palacio 2016; Raeissadat 2014; Schoffl 2017; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Watts 2020; Wolf 2011; Yadav 2015; Yerlikaya 2018). Three trials are awaiting classification (see Characteristics of studies awaiting classification table). We identified 24 ongoing trials in clinical trials registries (see Characteristics of ongoing studies table). We excluded 16 studies, 15 of which were not randomised controlled trials and 1 that used the wrong intervention. A flow diagram of the study selection process is presented in Figure 1.

Figure 1

Study flow diagram.

Included studies

We have provided a full description of all included trials in the Characteristics of included studies table. We contacted the authors of five trials to request information about missing data for unreported or partially reported outcomes and received replies from three of them (Creaney 2011; Martin 2019; Martínez‐Montiel 2015).

Study design and setting

Thirty studies were randomised controlled trials (RCTs) and two were quasi‐randomised trials (Jindal 2013; Tetschke 2015). Twenty‐five studies had two intervention arms, and seven had three intervention arms (Branson 2016; Krogh 2013; Linnanmäki 2020; Ozturan 2010; Palacio 2016; Wolf 2011; Yerlikaya 2018).

The included trials were conducted in 19 different countries: Turkey (Arik 2014; Gedik 2016; Ozturan 2010; Yerlikaya 2018), India (Behera 2015; Dojode 2012; Gautam 2015; Gupta 2019; Jindal 2013; Yadav 2015), Australia (Branson 2016), UK (Creaney 2011; Stenhouse 2013; Watts 2020), The Netherlands (Gosens 2011), Iran (Kazemi 2010; Raeissadat 2014), Denmark (Krogh 2013), Finland (Linnanmäki 2020), Poland (Lebiedziński 2015), South Korea (Lim 2017), Spain (Martin 2019), Mexico (Martínez‐Montiel 2015), Italy (Merolla 2017), USA (Mishra 2014; Wolf 2011), France (Montalvan 2015), Egypt (Omar 2012), Brazil (Palacio 2016), Germany (Schoffl 2017; Tetschke 2015), and Greece (Thanasas 2011). The total duration of trials varied between four months and five years. Three studies were funded by manufacturers of the PRP centrifugation system (Gosens 2011; Mishra 2014; Montalvan 2015), two studies were provided with PRP kits (Krogh 2013; Schoffl 2017), one study received funding from PRP kit manufacturers (Watts 2020), and four studies were funded by research grants (Linnanmäki 2020; Martin 2019; Raeissadat 2014; Wolf 2011). The remaining 22 studies did not report a funding source.

Participant characteristics

The 32 trials had randomised 2337 participants to receive autologous blood, PRP, or the control intervention, with numbers ranging between 25 and 230 per trial. The mean age of participants ranged from 36 years to 53 years, and the mean duration of symptoms before study enrolment for the 13 studies that reported it ranged from 1 month to 22 months. Among the 22 studies that reported gender, 56% of participants were female.

Inclusion criteria varied between trials. Seven studies specified a clinical diagnosis of lateral epicondylitis (Arik 2014; Lebiedziński 2015; Linnanmäki 2020; Merolla 2017; Watts 2020; Wolf 2011; Yadav 2015), and 11 studies specified pain on resisted wrist extension as a specific inclusion criterion (Branson 2016; Gosens 2011; Kazemi 2010; Krogh 2013; Martin 2019; Mishra 2014; Omar 2012; Raeissadat 2014; Schoffl 2017; Tetschke 2015; Thanasas 2011). Three studies specified a positive Cozen's test, Maudsley test, and Mill’s manoeuvre (Dojode 2012; Palacio 2016; Yerlikaya 2018), and one study specified a positive Thomsen test (Ozturan 2010). Four studies confirmed the diagnosis of lateral epicondylitis based on ultrasound or MRI (Branson 2016; Krogh 2013; Lim 2017; Stenhouse 2013), and one study excluded other causes of elbow pain using X‐rays (Jindal 2013). Five studies specified an inclusion criterion of recalcitrant lateral epicondylitis, defined as failed conservative treatment (oral medication and physical therapy) for three to six months (Behera 2015; Creaney 2011; Gautam 2015; Gedik 2016; Martínez‐Montiel 2015).

Interventions

A detailed description of the interventions delivered in each trial is summarised in the Characteristics of included studies table. Of the 32 included trials, seven had three intervention arms (Branson 2016; Krogh 2013; Linnanmäki 2020; Ozturan 2010; Palacio 2016; Wolf 2011; Yerlikaya 2018).

Nine trials compared PRP injection to placebo injection (Behera 2015; Krogh 2013; Linnanmäki 2020; Martin 2019; Mishra 2014; Montalvan 2015; Palacio 2016; Schoffl 2017; Yerlikaya 2018), and two trials (out of which one ‐ Linnanmäki 2020 ‐ had three arms comparing autologous blood to PRP to saline) compared autologous blood injection to placebo injection (Linnanmäki 2020; Wolf 2011). Fifteen trials compared PRP to glucocorticoid injection (Arik 2014; Branson 2016; Dojode 2012; Gautam 2015; Gosens 2011; Gupta 2019; Kazemi 2010; Krogh 2013; Lebiedziński 2015; Martínez‐Montiel 2015; Omar 2012; Ozturan 2010; Palacio 2016; Wolf 2011; Yadav 2015); six trials compared autologous blood to glucocorticoid injection (Arik 2014; Branson 2016; Dojode 2012; Kazemi 2010; Ozturan 2010; Wolf 2011); and nine trials compared PRP to glucocorticoid injection (Gautam 2015; Gosens 2011; Gupta 2019; Krogh 2013; Lebiedziński 2015; Martínez‐Montiel 2015; Omar 2012; Palacio 2016; Yadav 2015). Four trials compared PRP to autologous blood injection (Creaney 2011; Linnanmäki 2020; Raeissadat 2014; Thanasas 2011), and one trial compared PRP injection and dry needling to dry needling alone (Stenhouse 2013). Two trials compared autologous blood or PRP injection plus tennis elbow strap and exercise versus tennis elbow strap and exercise alone (Gedik 2016; Lim 2017). One trial compared PRP injection to extracorporeal shock wave therapy (ESWT) (Ozturan 2010); two trials compared PRP injection to surgery (Merolla 2017; Watts 2020). One trial compared PRP injection to laser application (Tetschke 2015), and one trial compared autologous blood injection to polidocanol injection (Branson 2016).

Sixteen studies used a peppering technique (multiple passes to the tendon) to incite fresh tendon injury during injection (Behera 2015; Branson 2016; Gautam 2015; Gosens 2011; Gupta 2019; Krogh 2013; Martin 2019; Mishra 2014; Montalvan 2015; Raeissadat 2014; Schoffl 2017; Stenhouse 2013; Thanasas 2011; Watts 2020; Wolf 2011; Yerlikaya 2018). Omar 2012 did not describe the injection, and remaining trialists described injection without mentioning multiple passes of the needle.

Most participants received one injection. In five studies (Branson 2016; Martin 2019; Montalvan 2015; Stenhouse 2013; Tetschke 2015), participants were given two injections, and in one study (Ozturan 2010), those who did not improve with one injection were given a second injection.

Outcomes

An ORBIT matrix that shows outcomes measured and level of reporting for each outcome in each trial (rated as fully reported, partially reported, measured but not reported, unclear if measured, or not measured) is presented in Table 1.

Open in table viewer

Table 1. Outcome reporting bias In trials (ORBIT) matrix

Study ID	Participant‐reported pain relief ≥ 30%	Pain	Function or disability	Treatment success	Health‐related quality of life	Withdrawal due to adverse events	Adverse events
Arik 2014	?	Full	Full	Full	?	?	Full
Behera 2015	?	Full	Full	?	?	Full	Full
Branson 2016	Not measured	Not measured	Full	Full	Not measured	?	Full
Creaney 2011	?	?	Full	Full	?	Full	?
Dojode 2012	?	Full	?	Full	?	Full	Full
Gautam 2015	?	Partial	Partial	?	?	Full	?
Gedik 2016	?	?	Full	Full	?	Full	?
Gosens 2011	Not measured	Full	Full	Full	Not measured	Full	Full
Gupta 2019	?	Full	Full	Full	?	?	Full
Jindal 2013	?	Full	?	Full	?	Full	?
Kazemi 2010	Not measured	Full	Full	Not measured	Not measured	Full	Full
Krogh 2013	Not measured	Full	Full	Not measured	Not measured	Full	Full
Lebiedziński 2015	?	?	Full	Full	?	Full	Full
Lim 2017	?	Partial	Partial	Full	?	Full	Full
Linnanmäki 2020	Not measured	Full	Full	Not measured	Not measured	Not measured	Full
Martin 2019	Not measured	Full	Full	Full	Not measured	Full	Full
Martínez‐Montiel 2015	?	Full	Full	?	?	Full	?
Merolla 2017	?	Partial	Partial	Measured	?	Full	?
Mishra 2014	Full	Partial	Partial	Full	?	Full	Full
Montalvan 2015	Not measured	Full	Full	Not measured	Not measured	Full	Full
Omar 2012	?	Full	Full	?	?	Full	?
Ozturan 2010	Full	Full	Full	?	?	Full	Full
Palacio 2016	?	?	Full	Full	?	?	?
Raeissadat 2014	?	Full	Full	Full	?	Full	?
Schoffl 2017	?	?	Full	?	?	?	?
Stenhouse 2013	?	Full	Full	Full	?	Full	Full
Tetschke 2015	?	Full	Full	Full	?	Full	?
Thanasas 2011	?	Full	Full	?	?	Full	Full
Wolf 2011	?	Full	Full	?	?	Full	?
Watts 2020	not measured	Full	Full	Not measured	Not measured	Not measured	Full
Yadav 2015	?	Partial	Partial	?	?	Full	?
Yerlikaya 2018	?	Full	Measured	?	?	Full	Measured

'Full': sufficient data for inclusion in a meta‐analysis were reported (e.g. mean, standard deviation, sample size per group for continuous outcomes).
'Partial': insufficient data for inclusion in a meta‐analysis were reported (e.g. means only, with no measures of variance).
'Measured': outcome was measured but no outcome data were reported.
'Not measured': outcome was not measured by trialists.
'?': unclear whether the outcome was measured or not (as a trial protocol was unavailable).

Major outcomes

Participant‐reported pain relief 30% or greater, or 50% or greater

None of the trials reported pain relief > 30% (pre‐planned cutoff), but two trials reported pain relief > 50% (Mishra 2014; Ozturan 2010).

Pain

Twenty‐four trials measured overall pain (mean or mean change) using a 0 to 10‐point visual analogue scale (VAS), with 10 indicating worst pain, and two trials measured pain with the PRTEE pain subscale (Krogh 2013; Watts 2020). Six trials did not report measures of variance or did not clearly report them (Gautam 2015; Gupta 2019; Lim 2017; Merolla 2017; Mishra 2014; Yadav 2015). One trial did not measure pain (Branson 2016). It is unclear whether the five remaining trials that did not report pain measured pain or not, as these trials were not registered and no study protocols were found (Creaney 2011; Gedik 2016; Lebiedziński 2015; Palacio 2016; Schoffl 2017).

Function

Twenty‐nine trials measured function, six of which did not clearly report measures of variance (Gautam 2015; Gupta 2019; Lim 2017; Merolla 2017; Mishra 2014; Yadav 2015). One trial measured function but did not report the results (Yerlikaya 2018). Two trials did not measure function (Dojode 2012; Jindal 2013). Most trials used either the PRTEE questionnaire ‐ Arik 2014; Branson 2016; Creaney 2011; Gedik 2016; Krogh 2013; Merolla 2017; Mishra 2014; Palacio 2016; Watts 2020 ‐ or the DASH questionnaire ‐ Gautam 2015; Gosens 2011; Gupta 2019; Kazemi 2010; Lebiedziński 2015; Linnanmäki 2020; Martin 2019; Omar 2012; Schoffl 2017; Tetschke 2015; Wolf 2011; Yadav 2015. Three trials measured function using the Modified Mayo Clinic Performance Index for Elbow (MMCPIE) (Behera 2015; Lim 2017; Raeissadat 2014); one used the quick DASH (Martínez‐Montiel 2015); one used the Roles‐Maudsley score (Montalvan 2015); and one used upper extremity functional score to measure elbow function (Ozturan 2010). In one trial, elbow function was measured by the Nirschl staging system (Stenhouse 2013), and another trial used the Liverpool elbow score to measure elbow function (Thanasas 2011).

Treatment success

Eighteen trials reported some kind of assessment of treatment success, most of which measured proportion with 25% pain or disability reduction; one trial measured treatment success on the Global Rating of Change (GROC) (Branson 2016), and another trial included patient satisfaction with treatment results along with pain reduction (Gedik 2016). Three trials did not measure treatment success (Kazemi 2010; Krogh 2013; Montalvan 2015); it is unclear whether 11 trials measured treatment success or not, as there was no study protocol (Behera 2015; Gautam 2015; Martínez‐Montiel 2015; Omar 2012; Ozturan 2010; Schoffl 2017; Thanasas 2011; Watts 2020; Wolf 2011; Yadav 2015; Yerlikaya 2018).

Health‐related quality of life

None of the included studies measured or reported this outcome.

Withdrawal due to adverse events

Only two trials reported withdrawal due to adverse events (Martin 2019; Stenhouse 2013).

Adverse events

Eighteen trials reported adverse events, one trial measured but did not report adverse events (Yerlikaya 2018), and in 13 trials it is unclear whether or not adverse events were measured (Creaney 2011; Gautam 2015; Gedik 2016; Jindal 2013; Martínez‐Montiel 2015; Merolla 2017; Omar 2012; Palacio 2016; Raeissadat 2014; Schoffl 2017; Tetschke 2015; Wolf 2011; Yadav 2015).

Minor outcomes

None of the studies reported other pain measures or serious adverse events.

Nine trials reported mean grip strength (Arik 2014; Gautam 2015; Gedik 2016; Gupta 2019; Kazemi 2010; Linnanmäki 2020; Ozturan 2010; Merolla 2017; Yadav 2015).

Excluded studies

We excluded 16 full‐text articles; 15 were not RCTs, and one had only one participant who received different treatments in both arms. Full details can be found in the Characteristics of excluded studies table.

Risk of bias in included studies

A summary of the risk of bias of included studies can be seen in Figure 2, and details are provided in the Characteristics of included studies table. All included trials were susceptible to bias. Overall, 21 (66%) trials were susceptible to selection bias, 20 (62%) were at risk of performance bias, 20 (62%) were at risk of detection bias, seven (22%) were at risk of attrition bias, 25 (78%) were at risk of selective reporting bias, and five (16%) were at risk of other potential bias (Figure 3).

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Only 11 (34%) trials used appropriate methods to both generate and conceal their allocation sequence, and so we rated these at low risk of selection bias (Branson 2016; Gosens 2011; Gupta 2019; Krogh 2013; Lebiedziński 2015; Linnanmäki 2020; Martin 2019; Martínez‐Montiel 2015; Schoffl 2017; Watts 2020; Wolf 2011).

Eight (25%) trials did not clearly report their method of sequence generation (Arik 2014; Behera 2015; Creaney 2011; Gautam 2015; Omar 2012; Ozturan 2010; Palacio 2016; Yadav 2015), and 17 (53%) trials did not adequately report their method of allocation concealment (Arik 2014; Behera 2015; Creaney 2011; Dojode 2012; Gautam 2015; Gedik 2016; Lim 2017; Merolla 2017; Mishra 2014; Montalvan 2015; Omar 2012; Ozturan 2010; Raeissadat 2014; Stenhouse 2013; Thanasas 2011; Yadav 2015; Yerlikaya 2018). Therefore, the risk of selection bias in these trials was unclear. We judged three trials as having high risk of bias (Jindal 2013; Kazemi 2010; Tetschke 2015), as two were quasi‐randomised (Jindal 2013; Tetschke 2015), and one used a coin‐toss method of randomisation for only the first participant and sequential allocation for the rest of the sample (Kazemi 2010).

Blinding

We judged 11 (34%) trials to be at low risk of performance and detection bias because both participants and study personnel were successfully blinded (Creaney 2011; Gosens 2011; Krogh 2013; Linnanmäki 2020; Martin 2019; Martínez‐Montiel 2015; Mishra 2014; Montalvan 2015; Schoffl 2017; Wolf 2011; Yerlikaya 2018). Of these, seven were placebo‐controlled trials (Krogh 2013; Linnanmäki 2020; Martin 2019; Mishra 2014; Montalvan 2015; Wolf 2011; Yerlikaya 2018), one trial compared autologous blood to PRP (Creaney 2011), two trials compared PRP to glucocorticoid injection (Gosens 2011; Martínez‐Montiel 2015), and one trial compared PRP to dry needling (Schoffl 2017).

We judged 14 (43%) trials to be at risk of high performance and detection bias (Arik 2014; Gautam 2015; Gupta 2019; Jindal 2013; Kazemi 2010; Lebiedziński 2015; Lim 2017; Merolla 2017; Ozturan 2010; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Watts 2020; Yadav 2015). Four trials did not blind participants and study personnel, leading to high risk of bias in the assessment of both subjective and objective outcomes (Arik 2014; Gupta 2019; Ozturan 2010; Watts 2020). Seven trials had high risk of performance and detection bias for subjective outcomes only and measured no objective outcomes (Dojode 2012; Jindal 2013; Kazemi 2010; Lebiedziński 2015; Stenhouse 2013; Tetschke 2015; Thanasas 2011). One trial had high risk of performance bias and detection bias for subjective outcomes and low risk of detection bias for objective outcomes, as assessors were blinded (Lim 2017). Three trials had high risk of performance bias and detection bias for subjective outcomes and unclear risk of detection bias for objective outcomes, as it is unclear whether or not assessors were blinded (Gautam 2015; Merolla 2017; Yadav 2015).

We judged one trial to be at unclear risk of both performance and detection bias for subjective and objective outcomes (Omar 2012). Two trials had unclear risk of performance and detection bias for subjective outcomes and low risk of bias for objective outcomes, as no assessor‐reported outcomes were measured in this study (Behera 2015; Palacio 2016). In Omar 2012, study authors did not report whether participants and study personnel were blinded to treatment allocation, so we judged risk of performance and detection bias as unclear. We judged one trial to be at unclear risk of performance and detection bias for objective outcomes and at high risk of detection bias for subjective outcomes, as participants were unable to be blinded due to the nature of the intervention (injections compared to bandage and exercise) (Gedik 2016). Branson 2016 had low risk of performance bias and unclear risk of detection bias for both subjective and objective outcomes, as study personnel and participants were blinded for the first injection; however investigators do not report whether they were blinded for the second injection.

Incomplete outcome data

We judged 25 (78%) trials to be at low risk of attrition bias (Arik 2014; Behera 2015; Branson 2016; Creaney 2011; Dojode 2012; Gautam 2015; Gedik 2016; Gosens 2011; Gupta 2019; Jindal 2013; Kazemi 2010; Krogh 2013; Lebiedziński 2015; Lim 2017; Martínez‐Montiel 2015; Merolla 2017; Montalvan 2015; Omar 2012; Ozturan 2010; Palacio 2016; Raeissadat 2014; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Yerlikaya 2018). In 12 trials, there were no withdrawals (Arik 2014; Dojode 2012; Gautam 2015; Gupta 2019; Jindal 2013; Kazemi 2010; Krogh 2013; Martínez‐Montiel 2015; Merolla 2017; Omar 2012; Palacio 2016; Yerlikaya 2018). One trial reported only one withdrawal from the control group (Behera 2015); another trial reported one withdrawal from the autologous blood group (Thanasas 2011). In one trial, although there were more withdrawals in the control group, an ITT was performed and data from all withdrawals were used in the final analysis (Branson 2016). In seven trials, withdrawal numbers and reasons were similar across groups (Creaney 2011; Gedik 2016; Gosens 2011; Lebiedziński 2015; Montalvan 2015; Ozturan 2010; Stenhouse 2013). In one trial, almost similar numbers withdrew from both treatment arms and left the study to receive other treatments, hence data from those participants were not sought (Lim 2017). In one trial, two participants left the control group to undergo surgery and were excluded from the final analysis (Tetschke 2015).

We judged three (9%) trials to be at high risk (Mishra 2014; Schoffl 2017; Watts 2020). In Mishra 2014, withdrawal rates in the control group (19%) were twice as high as those in the intervention group (9.8%), reasons for withdrawal were not given, and study authors did not provide withdrawal numbers for each group for final follow‐up. Study authors for Schoffl 2017 reported that they excluded from the study those not achieving satisfactory results, and withdrawal rates were high (28%) for both groups. We judged four (12%) trials to be at unclear risk of attrition bias (Linnanmäki 2020; Martin 2019; Wolf 2011; Yadav 2015). In one trial, although reasons for withdrawal were similar across groups, attrition rates were unbalanced across groups, at 22.5% in the PRP group, 5% in the autologous blood group, and 18% in the placebo group (Linnanmäki 2020). Another trial had high attrition rates (> 30%) that were balanced between groups, but study authors did not provide reasons for withdrawal (Martin 2019). For two trials, authors provided overall withdrawal rates but did not provide group‐wise withdrawal numbers (Wolf 2011; Yadav 2015).

Selective reporting

Risk of selective reporting bias was low in five (16%) trials (Gosens 2011; Kazemi 2010; Linnanmäki 2020; Martin 2019; Montalvan 2015), high in five (16%) trials (Lim 2017; Martínez‐Montiel 2015; Mishra 2014; Palacio 2016; Yerlikaya 2018), and unclear in 22 (68%) trials (Arik 2014; Behera 2015; Branson 2016; Creaney 2011; Dojode 2012; Gautam 2015; Gedik 2016; Gupta 2019; Jindal 2013; Krogh 2013; Lebiedziński 2015; Merolla 2017; Omar 2012; Ozturan 2010; Raeissadat 2014; Schoffl 2017; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Watts 2020; Wolf 2011; Yadav 2015).

We judged Lim 2017 to be at high risk of selective reporting bias, as there was no protocol or trial registration, some outcomes were measured but were not reported, and measures of variance were not reported for any outcome data. No protocol or trial registration is available for Martínez‐Montiel 2015, and study authors did not give a clear description of measurement tools or the intervention used. In Mishra 2014, study authors did not provide measures of variance for self‐reported data, and due to lack of US FDA clearance on the PRP centrifuge, although the trial was registered, no details were provided at clincialtrials.gov. Yerlikaya 2018 did not report any numerical results for subjective and objective outcomes; this trial was not registered, and no study protocol is available.

We judged 19 trials (60%) at unclear risk of selective reporting bias due to lack of study protocol and trial registration (Arik 2014; Behera 2015; Creaney 2011; Dojode 2012; Gautam 2015; Gedik 2016; Gupta 2019; Jindal 2013; Lebiedziński 2015; Merolla 2017; Omar 2012; Ozturan 2010; Raeissadat 2014; Schoffl 2017; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Wolf 2011; Yadav 2015). Branson 2016 stated a secondary outcome (Stratford Pain‐Free Function Questionnaire) at trial registration but did not measure or report it in published results of the trial. Krogh 2013 failed to report secondary outcomes at all time points.

Other potential sources of bias

We judged 27 trials (84%) at low risk of other identified potential sources of bias (Arik 2014; Behera 2015; Branson 2016; Creaney 2011; Dojode 2012; Gautam 2015; Gupta 2019; Jindal 2013; Kazemi 2010; Krogh 2013; Lebiedziński 2015; Lim 2017; Linnanmäki 2020; Martínez‐Montiel 2015; Merolla 2017; Montalvan 2015; Omar 2012; Palacio 2016; Raeissadat 2014; Schoffl 2017; Stenhouse 2013; Tetschke 2015; Thanasas 2011; Watts 2020; Wolf 2011; Yadav 2015; Yerlikaya 2018). We judged two trials at high risk of other sources of bias (Gedik 2016; Mishra 2014). Gedik 2016 administered the intervention to 62% of control group participants during the study (4 weeks), and we judged this trial at high risk of bias due to contamination of results at three months and six months. In Mishra 2014, the study sponsor added a post‐hoc six‐month follow‐up for a subset of participants (n = 119; 52% of the planned sample), and results from this subset may be biased.

We judged three trials at unclear risk of other potential bias (Gosens 2011; Martin 2019; Ozturan 2010). In Gosens 2011, there was risk of contamination of results due to several re‐interventions, which were unplanned and unbalanced across the two groups. In Martin 2019, the number of participants with medial elbow pain was higher in the control group (19%) than in the intervention group (11%), leading to potential contamination in interpretation of results. In Ozturan 2010, administration of re‐interventions across intervention (70%) and control groups (10%) was not balanced, leading to possible contamination in interpretation of results.

Effects of interventions

See: Summary of findings 1 Autologous blood or PRP versus placebo at 3 months' follow‐up

See summary of findings Table 1 for the main comparison autologous blood or PRP injection versus placebo.

Autologous blood or PRP injection versus placebo

Two trials compared autologous blood injection to placebo (saline) injection (Linnanmäki 2020; Wolf 2011), and nine trials compared PRP injection to placebo (saline or local anaesthetic) injection (Behera 2015; Krogh 2013; Linnanmäki 2020; Martin 2019; Mishra 2014; Montalvan 2015; Palacio 2016; Schoffl 2017; Yerlikaya 2018). We judged the ten placebo‐controlled trials to be clinically similar with respect to inclusion criteria and baseline participant characteristics of mean pain, function, and treatment success, facilitating pooling of data in a meta‐analysis. Statistical heterogeneity was unimportant for these outcomes until six weeks, and thereafter Behera 2015 caused substantial heterogeneity in pain and function. The certainty of evidence was moderate for pain and function, low for adverse events, and very low for treatment success, participant‐reported pain relief of 30% or greater or 50% or greater, and withdrawal due to adverse events. The major outcomes are reported in summary of findings Table 1.

Benefits

Participant‐reported pain relief (≥ 30% or ≥ 50%)

No studies measured participant‐reported pain relief of 30% or greater, and no studies measured this outcome at 3 months. Mishra 2014 measured participant‐reported pain relief (≥ 50%) but reported the outcome selectively at 6 months for a subgroup of 119 participants who were followed up longer than the originally planned 3 months. At 6 months, very low‐certainty evidence (downgraded twice for bias and for small numbers of events) indicates that the proportion of participants with pain relief of 50% or greater may be higher with PRP injection compared with placebo; 46 out of 56 (82%) who received PRP injection reported pain relief of 50% or greater compared with 38 out of 63 (60%) who received placebo injection (risk ratio (RR) 1.36, 95% confidence interval (CI) 1.08 to 1.72) at 6 months. Results show absolute improvement of 22% (5% better to 43% better) and relative improvement of 36% (8% better to 72% better; Analysis 1.1).

Mean pain

Based on data from eight trials, we found no clinically important improvement in pain (minimal clinically important difference (MCID) 1.5 points on a 0 to 10 scale; higher is worse pain) at 3 months for autologous blood or PRP injection versus placebo (moderate‐certainty evidence; downgraded once for bias). Statistical heterogeneity was unimportant up to 3 months (I² = 7% to 33%) and was substantial (I² = 76% to 78%) at later follow‐up points, largely driven by one study (Behera 2015).

At 3 weeks, mean pain (0 to 10; higher is worse) was 2.8 points with placebo and 2 points worse (higher scores) (95% CI 0.65 better to 4.65 worse; 1 study, 19 participants) with autologous blood or PRP injection. At 6 weeks, mean pain was 4.8 points with placebo and 0.26 points worse (95% CI 0.14 better to 0.65 worse; 7 studies, 570 participants) with autologous blood or PRP injection. At 3 months (primary time point), mean pain was 3.7 points with placebo and 0.16 points better (95% CI 0.60 better to 0.29 worse; 8 studies, 523 participants; I² = 13%) with autologous blood or PRP injection. This corresponds with absolute improvement of 1.6% (6% better to 3% worse) and relative improvement of 2.3% (9% better to 4% worse). At 6 months, mean pain was 1.64 points with placebo and 0.45 points better (95% CI 1.5 better to 0.59 worse; 7 studies, 387 participants) with autologous blood or PRP injection. At 12 months, mean pain was 2.3 points with placebo and 0.69 points better (95% CI 1.78 better to 0.39 worse; 5 studies, 241 participants) with autologous blood or PRP injection (Analysis 1.2).

Function

Based on data from seven trials, we found no clinically important improvement in function (MCID 10 points on a 100‐point scale; higher is worse function) for autologous blood or PRP injection versus placebo (moderate‐certainty evidence; downgraded once for bias). Statistical heterogeneity was unimportant (I² = 0 to 9%) up to 3 months and substantial (78% to 82%) at later time points, driven by one study (Behera 2015). Removing Behera 2015 decreased I² to 0 to 3%.

At 3 weeks, function (0 to 100 scale; lower is better) was 24 points with placebo and 12.0 points worse (95% CI 5.33 better to 29.33 worse; 1 study, 19 participants) with autologous blood or PRP injection. At 6 weeks, function was 36.2 points with placebo and 1.3 points worse (95% CI 1.64 better to 4.25 worse; 7 studies, 473 participants) with autologous blood or PRP injection. At 3 months, function was 27.5 points with placebo and 1.86 points better (95% CI 4.97 better to 1.25 worse; 8 studies, 502 participants; I² = 0%) with autologous blood or PRP injection. This corresponds with absolute benefit of 1.9% (5% better to 1% worse) and relative benefit of 4% (11% better to 3% worse). At 6 months, function was 19.2 points with placebo and 1.15 points better (95% CI 8.6 better to 6.3 worse; 7 studies, 379 participants) with autologous blood or PRP injection. At 12 months, function was 20.4 points with placebo and 5.81 points better (95% CI 16.7 better to 5.05 worse; 4 studies, 203 participants) with autologous blood or PRP injection (Analysis 1.3).

Treatment success

Based upon very low‐certainty evidence (downgraded for bias, indirectness, and imprecision), we are uncertain whether autologous blood or PRP injection improves treatment success compared with placebo injection. Data from four trials show that 121 out of 185 (65%) rated their treatment as successful with placebo versus 125 out of 187 (67%) with autologous blood or PRP injection (RR 1.00, 95% CI 0.83 to 1.19; I² = 38%) for absolute improvement of 0% higher (11.1% lower to 12.4% higher) and relative change 0% higher (17% lower to 19% higher) (95% confidence intervals include both clinically important and unimportant change in treatment success with use of autologous blood or PRP injection (Analysis 1.4) (Martin 2019; Mishra 2014; Montalvan 2015; Palacio 2016).

Health‐related quality of life

No studies measured this outcome for this comparison.

Minor outcomes

None of the studies reported other pain measures, grip strength, or serious adverse events.

Harms

Withdrawal due to adverse events

Very low‐certainty evidence (downgraded once for bias and twice for very serious imprecision) suggests that we are uncertain whether autologous blood or PRP injection increased the risk of withdrawal due to adverse events.

Martin 2019 reported withdrawal due to adverse events. Six studies reported reasons for withdrawal, and reasons did not include adverse events (judged as zero events) (Behera 2015; Krogh 2013; Mishra 2014; Montalvan 2015; Wolf 2011; Yerlikaya 2018). Thus, the data from these six trials are not included in the pooled estimate (Analysis 1.5).

Data from Martin 2019 show withdrawal due to adverse events in 3 out of 39 (8%) with placebo versus 1 out of 41 (2%) with autologous blood or PRP injection (RR 0.32, 95% CI 0.03 to 2.92; 1 study), an absolute change of 5.2% fewer events (7.5% fewer to 14.8% more), and a relative change of 68% fewer events (97% fewer to 192% more) (95% confidence intervals show that autologous blood or PRP injection can cause both greater and lesser harm compared with placebo).

Adverse events

Low‐certainty evidence (downgraded for bias and imprecision) suggests that autologous blood or PRP injection may not increase risk for adverse events compared with placebo. Data from five studies show adverse event rates of 35 out of 208 (17%) with placebo versus 41 out of 217 (19%) with autologous blood or PRP injection (RR 1.14, 95% CI 0.76 to 1.72; 5 studies, 425 participants; I² = 0%), an absolute change of 2.4% more events (4% fewer to 12% more), and a relative change of 14% more events (24% fewer to 72% more) (Analysis 1.6) (Behera 2015; Krogh 2013; Martin 2019; Mishra 2014; Montalvan 2015).

Autologous blood or PRP injection versus glucocorticoid injection

Benefits

Participant‐reported pain relief (≥ 30% or ≥ 50%)

Ozturan 2010 reported proportion of participants with 50% or greater improvement in pain. We graded the evidence as low certainty (downgraded for bias and small numbers of events).

Pain relief favoured glucocorticoid injection at 6 weeks but not at 1 year. Pain relief rates were 18 out of 20 (90%) with glucocorticoid injection versus 3 out of 18 (17%) with autologous blood or PRP injection (RR 0.19, 95% CI 0.07 to 0.53) at 6 weeks. At 1 year, pain relief was 10 out of 20 (50%) with glucocorticoid injection versus 15 out of 18 (83%) with autologous blood or PRP injection (RR 1.67, 95% CI 1.03 to 2.71; number needed to treat for additional benefit (NNTB) 3, 95% CI 1.6 to 20) (Analysis 2.1)

Mean pain

We identified 13 trials reporting this outcome and noted considerable statistical heterogeneity up to 3 months (Arik 2014; Dojode 2012; Gautam 2015; Gosens 2011; Gupta 2019; Jindal 2013; Kazemi 2010; Krogh 2013; Martínez‐Montiel 2015; Omar 2012; Ozturan 2010; Wolf 2011; Yadav 2015). At 3 weeks, heterogeneity (I² = 91%) seemed to be driven largely by one study (Arik 2014), but at 6 weeks (I² = 90%) and at 3 months (I² = 71%), heterogeneity could not be explained by study or participant characteristics. After 6 months, statistical heterogeneity was moderate (I² = 61% at 6 months and 62% at 1 year), and with removal of Gupta 2019, heterogeneity dropped to 0% at 1 year. Standard deviation (SD) values reported by Gupta 2019 were unusually small, yielding large weight to this study at 1‐year analysis (Analysis 2.2). Study authors did not respond to queries; thus we report estimates with and without this study.

We downgraded this outcome to low‐certainty evidence for bias and inconsistency (effect sizes varied from no effect to clinically meaningful effect). Up to 3 months, PRP may not provide clinically important pain reduction when compared with glucocorticoid injection. At 3 weeks, mean pain (on a 0 to 10 scale; higher is worse) was 3.43 points with glucocorticoid injection and 2.06 points worse (95% CI 0.67 worse to 3.45 worse; 5 studies, 280 participants) with autologous blood or PRP injection. At 6 weeks, mean pain was 2.5 points with glucocorticoid injection and 0.99 points worse (95% CI 0.21 worse to 1.77 worse; 13 studies, 707 participants) with autologous blood or PRP injection.

At 6 months, 12 months, and greater than 1 year follow‐up, mean pain was better with autologous blood or PRP injection. At 3 months, mean pain was 3.1 points with glucocorticoid injection and 1.15 points better (95% CI 1.71 better to 0.59 better; 11 studies, 627 participants) with autologous blood or PRP injection. At 6 months, mean pain was 3.89 points with glucocorticoid injection and 1.55 points better (95% CI 2.21 better to 0.9 better; 8 studies, 427 participants) with autologous blood or PRP injection. At 12 months, mean pain was 4.6 points with glucocorticoid injection and 1.59 points better (95% CI 2.22 better to 0.97 better; 4 studies, 258 participants) with autologous blood or PRP injection. At greater than 1 year, mean pain was 4.24 points with glucocorticoid injection and 2.11 points better (95% CI 3.19 better to 1.03 better; 1 study, 100 participants) with autologous blood or PRP injection (Analysis 2.2).

Excluding Gupta 2019 (which reported unusually small SD values) did not change the results considerably. At 6 weeks, the mean difference was 0.99 points (95% CI 0.21 to 1.77) including this study and 0.80 points (95% CI 0.05 to 1.56) without the study. At 3 months, the mean difference was ‐1.15 points (95% CI ‐1.71 to ‐0.59) including this study and ‐1.04 points (95% CI ‐1.66 to‐0.42) without the study. At 12 months, the mean difference was ‐1.59 points (95% CI ‐2.22 to ‐0.97) with Gupta 2019 and ‐1.95 points (95% CI ‐2.54 to ‐1.35) without it.

Function

Fourteen studies measured function using various measures (Patient‐Rated Tennis Elbow Evaluation (PRTEE) questionnaire, Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire, Quick Dash) (Arik 2014; Branson 2016; Gautam 2015; Gosens 2011; Gupta 2019; Kazemi 2010; Krogh 2013; Lebiedziński 2015; Martínez‐Montiel 2015; Omar 2012; Ozturan 2010; Palacio 2016; Wolf 2011; Yadav 2015). We observed considerable statistical heterogeneity up to 6 months (I² = 93% at 3 weeks; I² = 86% at 6 weeks; I² = 78% at 3 months; I² = 87% at 6 months). At 3 weeks, heterogeneity seemed to be driven by one study (Arik 2014), but at later time points, heterogeneity could not be explained by study or participant characteristics. At 1 year, statistical heterogeneity was unimportant (I² = 22%), and at greater than 1 year follow‐up, there was only one study (Gosens 2011). Similar to mean pain, Gupta 2019 reported unusually small SD values; thus we report these values with and without these results.

We found low‐certainty evidence (downgraded for bias and imprecision) to suggest that PRP may not improve function compared with glucocorticoid injection up to 1 year. At 2 years, PRP may improve function compared to glucocorticoid injection. We considered downgrading for inconsistency, but further downgrading to very low seemed inappropriate, as in all studies, the direction of effect was the same.

At 3 weeks, function (0 to 100; lower is better) was 43.5 points with glucocorticoid injection and 16.5 points worse (95% CI 4.23 better to 37.15 worse; 3 studies, 170 participants) with autologous blood or PRP injection. At 6 weeks, function was 31.2 points with glucocorticoid injection and 6.1 points worse (95% CI 1.79 worse to 10.44 worse; 13 studies, 724 participants) with autologous blood or PRP injection. At 3 months, function was 33.4 points with glucocorticoid injection and 10.2 better (95% CI 6.21 better to 14.1 better; 12 studies, 635 participants) with autologous blood or PRP injection. At 6 months, function was 33.22 points with glucocorticoid injection and 5.07 points better (95% CI 12.66 better to 2.52 worse; 7 studies, 374 participants) with autologous blood or PRP injection. At 1 year, function was 32.2 points with glucocorticoid injection and 8.94 points better (95% CI 5.78 better to 12.1 better; 4 studies, 317 participants) with autologous blood or PRP injection. At greater than1 year, function was 36.5 points with glucocorticoid injection and 18.9 points better (95% CI 28.27 better to 9.53 better; 1 study, 100 participants) with autologous blood or PRP injection (Analysis 2.3).

Excluding Gupta 2019 did not change the results considerably. At 6 weeks, the mean difference was 6.11 points (95% CI 1.79 to 10.44) with, and 5.39 points (95% CI 0.00 to 10.78) without, the study. At 3 months, the mean difference was ‐10.19 (95% CI ‐14.16 to ‐6.21) with, and ‐9.80 (95% CI ‐15.03 to ‐4.57) without, Gupta 2019. At 1 year, the mean difference was ‐8.94 (95% CI ‐12.09 to ‐5.78) with, and ‐9.73 (95% CI ‐15.89 to ‐3.58) without, Gupta 2019.

Treatment success

Six trials reported some measure of treatment success (Arik 2014; Branson 2016; Dojode 2012; Gupta 2019; Jindal 2013; Lebiedziński 2015). We downgraded the evidence to very low for bias, imprecision, and indirectness. Only one study used patient‐reported global improvement (Branson 2016). Data from five trials suggest that PRP may not improve treatment success compared with glucocorticoid injection.

Treatment success rates were as follows: 76 out of 155 (49%) with glucocorticoid injection versus 27 out of 162 (17%) with autologous blood or PRP injection (RR 0.26, 95% CI 0.07 to 0.95; 5 studies, 317 participants) up to 6 weeks. This corresponds with number needed to treat for additional harm (NNTH) of 2.8 (95% CI 1.6 to 11). At 3 months, 43 out of 94 (46%) with glucocorticoid injection versus 69 out of 94 (73%) with autologous blood or PRP injection (RR 1.56, 95% CI 1.08 to 2.26; 3 studies, 188 participants) were reported. This corresponds with NNTB of 3.8 (95% CI 2.2 to 16.6). At 6 months, 46 out of 90 (51%) with glucocorticoid injection versus 44 out of 97 (45%) with autologous blood or PRP injection (RR 1.02, 95% CI 0.23 to 4.44; 3 studies) were reported. At 12 months, 57 out of 145 (39%) with glucocorticoid injection versus 90 out of 155 (58%) with autologous blood or PRP injection (RR 1.00, 95% CI 0.31 to 3.16; 2 studies, 199 participants) were reported. At greater than 1 year, 51 out of 95 (54%) with glucocorticoid injection versus 58 out of 104 (56%) with autologous blood or PRP injection (RR 1.0, 95% CI 0.31 to 3.16; 2 studies) were reported (Analysis 2.4).

Health‐related quality of life

Health‐related quality of life was not measured by any studies.

Minor outcomes

Six trials reported mean grip strength, but measured units were reported only by Arik 2014; thus we used standardised mean difference (SMD) to summarise the data (Arik 2014; Gautam 2015;Gupta 2019; Ozturan 2010; Kazemi 2010; Yadav 2015).

At 2 weeks, the SMD was ‐0.52 (95% CI ‐0.87 to ‐0.16; 3 studies, 170 participants), which back‐transforms to a mean reduction of 8.1 kg (95% CI 2.5 kg worse to 13.57 kg worse) with autologous blood or PRP injection. At 6 weeks, the SMD was ‐0.26 (95% CI ‐0.68 to 0.16; 6 studies, 348 participants), which back‐transforms to a mean reduction of 4,1 kg (95% CI 10.6 kg worse to 2.5 kg better) with autologous blood or PRP injection. At 3 months, the SMD was 0.56 (95% CI 0.19 to 0.93; 348 participants, 6 studies), which back‐transforms to a mean increase of 8.7 kg (95% CI 3 kg better to 14.5 kg better) with autologous blood or PRP injection. At 6 months, the SMD was 0.35 (95% CI ‐0.13 to 0.83; 2 studies, 68 participants), which back‐transforms to a mean increase of 5.5 kg (95% CI 2.03 kg worse to 12.95 kg better) with autologous blood or PRP injection. At 1 year, the SMD was 0.66 (95% CI 0.29 to 1.03; 2 studies, 118 participants), which back‐transforms to a mean increase of 10.3 kg (95% CI 4.5 kg better to 16.1 kg better) with autologous blood or PRP injection (Analysis 2.6).

Harms

Withdrawal due to adverse events

None of the studies measured this outcome.

Adverse events

Eight studies reported adverse events, but three had zero events; thus estimates were calculated based on five studies (Arik 2014; Dojode 2012; Krogh 2013; Lebiedziński 2015; Ozturan 2010). Very low‐certainty evidence (downgraded once for bias and twice for very serious imprecision) suggests that we are uncertain whether autologous blood or PRP injection increases adverse events compared with glucocorticoid injection. Statistical heterogeneity was unimportant (I² = 0%).

Adverse event rates were 24 out of 195 (12%) with glucocorticoid injection versus 46 out of 201 (23%) with autologous blood or PRP injection (RR 1.64, 95% CI 0.65 to 4.12; 317 participants, 5 studies) (Analysis 2.5).

PRP and dry needling versus dry needling alone

Benefits

Pain relief

The only study in this comparison did not measure this outcome (Stenhouse 2013).

Mean pain

Low‐certainty evidence (downgraded for bias and imprecision) suggests that PRP and dry needling may not improve pain compared with dry needling alone. The 95% confidence intervals include both clinically meaningful harm and benefit (1.5 points).

At 3 months, mean pain (0 to 10; higher is worse) was 6.02 points with dry needling alone and 0.14 points better (95% CI 2.13 better to 1.85 worse; 1 study, 28 participants) with PRP and dry needling. At 6 months, mean pain was 4.5 points with dry needling alone and 0.35 points better (95% CI 2.88 better to 2.18 worse; 1 study, 28 participants) with PRP and dry needling (Analysis 3.1).

Function

Low‐certainty evidence (downgraded for bias and imprecision) suggests that PRP may not improve function compared with dry needling alone. The 95% confidence intervals include both clinically meaningful harm and benefit (10 points).

At 3 months, function (0 to 100 scale, lower is better) was 28.7 points with dry needling alone and 2.8 points worse (95% CI 16.88 better to 22.48 worse; 1 study, 28 participants) with PRP and dry needling. At 6 months, function was 45.4 points with dry needling alone and 5.7 points worse (95% CI 14.36 better to 25.76 worse; 1 study, 28 participants) with PRP and dry needling (Analysis 3.2).

Treatment success

Stenhouse 2013 did not report this outcome.

Health‐related quality of life

Stenhouse 2013 did not report this outcome.

Minor outcomes

None of the studies measured minor outcomes for this comparison.

Harms

Withdrawal due to adverse events

Withdrawal rates were as follows: 1 out of 13 (8%) with dry needling alone versus 2 out of 15 (13%) with PRP and dry needling (RR 1.73, 95% CI 0.18 to 16.99; 1 study) (Analysis 3.3).

Adverse events

Very low‐certainty evidence (downgraded once for bias and twice for very serious imprecision) from Stenhouse 2013 suggests that we are uncertain whether PRP and dry needling increases adverse events compared with dry needling alone.

Adverse event rates were as follows: 1 out of 13 (8%) with dry needling alone versus 2 out of 15 (13%) with PRP and dry needling (RR 1.73, 95% CI 0.18 to 16.99; 1 study) (Analysis 3.4).

PRP injection versus autologous blood injection

Benefits

Pain relief

None of the studies measured this outcome.

Mean pain

Three studies measured pain for this comparison (Linnanmäki 2020; Raeissadat 2014; Thanasas 2011). Moderate‐certainty evidence (downgraded for bias) shows that PRP injection probably does not improve pain compared with autologous blood injection. The confidence intervals exclude clinically important benefit at all time points.

At 6 weeks, mean pain (0 to 10; higher is worse) was 1.9 points with autologous blood injection and 0.24 points better (95% CI 1.21 better to 0.73 worse; 3 studies, 169 participants) with PRP injection. At 3 months, mean pain was 2.1 points with autologous blood injection and 0.4 points better (95% CI 1.1 better to 0.3 better; 3 studies, 169 participants) with PRP injection. At 6 months, mean pain was 2.1 points with autologous blood injection and 0.28 points better (95% CI 1.04 better to 0.48 worse; 3 studies, 169 participants) with PRP injection. At 12 months, mean pain was 2.3 points with autologous blood injection and 0.05 points better (95% CI 1.12 better to 1.22 worse; 2 studies, 141 participants) with PRP injection (Analysis 4.1).

Function

Four studies measured function for this comparison (Creaney 2011; Linnanmäki 2020; Raeissadat 2014; Thanasas 2011). Moderate‐certainty evidence (downgraded for bias) shows that PRP injection probably does not provide clinically important benefit for function compared with autologous blood injection.

At 6 weeks, function (0 to 100; lower is better) was 31.2 points with autologous blood injection and 3.44 points better (95% CI 6.6 better to 0.28 better; 4 studies, 276 participants) with PRP injection. At 3 months, function was 21.4 points with autologous blood and 3.25 points better (95% CI 6.33 better to 0.17 better; 4 studies, 292 participants) with PRP. At 6 months, function was 18 points with autologous blood and 2.83 points better (95% CI 6.02 better to 0.37 better; 4 studies, 297 participants) with PRP injection. At 12 months, function was 17 points with autologous blood injection and 0.71 points better (95% CI 8.53 better to 7.11 worse; 2 studies, 140 participants) with PRP injection (Analysis 4.2).

Treatment success

Two studies measured treatment success for this comparison (Creaney 2011; Raeissadat 2014). Low‐certainty evidence (downgraded for bias and imprecision) suggests that PRP injection may not improve rate of treatment success compared with autologous blood injection.

Treatment success rates were as follows: 61 out of 90 (68%) with autologous blood injection versus 69 out of 101 (68%) with PRP injection (RR 1.03, 95% CI 0.77 to 1.37) (Analysis 4.3).

Health‐related quality of life

None of the studies measured this outcome.

Minor outcomes

None of the studies measured any minor outcomes for this comparison.

Harms

Withdrawal due to adverse events

None of the studies reported this outcome.

Adverse events

Very low‐certainty evidence (downgraded for bias and very serious imprecision) suggests that we are uncertain whether rate for adverse events differs between PRP and autologous blood injections.

Adverse events rates were as follows: 9 out of 77 (12%) with autologous blood injection versus 4 out of 62 (6%) with PRP injection (RR 2.25, 95% CI 0.9 to 5.62; 2 studies) (Analysis 4.4).

Autologous blood injection versus extracorporeal shock wave therapy (ESWT)

Benefits

Pain relief

The only study in this comparison reported proportion with 50% or greater pain improvement (Ozturan 2010). Low‐certainty evidence (downgraded for bias and imprecision) suggests that autologous blood injection may not provide better chance of pain relief compared with ESWT.

Pain relief rates were as follows: 8 out of 20 (40%) with ESWT versus 3 out of 20 (15%) with autologous blood injection (RR 0.38, 95% CI 0.12 to 1.21) at 6 weeks; 18 out of 20 (90%) with ESWT versus 16 out of 20 (80%) with autologous blood injection (RR 0.89, 95% CI 0.68 to 1.16) at 1 year (Analysis 5.1).

Mean pain

Low‐quality evidence (downgraded for bias and imprecision; only 1 study) suggests that autologous blood injection may not provide important benefit for pain compared with ESWT. The 95% confidence intervals exclude clinically important benefit.

At 6 weeks, mean pain (0 to 10 scale; higher indicates worse) was 4.42 points with ESWT and 0.63 points worse (95% CI 0.28 better to 1.54 worse; 1 study, 37 participants) with autologous blood injection. At 3 months, mean pain was 2.26 points with ESWT and 0.29 points worse (95% CI 0.75 better to 1.33 worse; 1 study, 37 participants) with autologous blood injection. At 6 months, mean pain was 2.21 points with ESWT and 0.23 points worse (95% CI 0.78 better to 1.24 worse; 1 study, 37 participants) with autologous blood injection. At 1 year, mean pain was 2.10 points with ESWT and 0.23 points worse (95% CI 0.61 better to 1.07 worse; 1 study, 37 participants) with autologous blood injection (Analysis 5.2).

Function

Low‐certainty evidence (downgraded for bias and imprecision; only 1 study) suggests that autologous blood injection may not provide important benefit for function compared with ESWT. The confidence intervals do not overlap with clinically important benefit at any time point.

At 6 weeks, function (0 to 100 scale; lower is better) was 30.0 points with ESWT and 3.8 points worse (95% CI 1.56 better to 9.16 worse; 1 study, 37 participants) with autologous blood injection. At 3 months, function was 18.1 points with ESWT and 1.4 points worse (95% CI 5.82 better to 8.62 worse; 1 study, 37 participants) with autologous blood injection. At 6 months, function was 19.2 points with ESWT and 1.5 points worse (95% CI 4.17 better to 7.17 worse; 1 study, 37 participants) with autologous blood injection. At 1 year, function was 19.5 points with ESWT and 0.9 points better (95% CI 5.98 better to 4.18 worse; 1 study, 37 participants) with autologous blood injection (Analysis 5.3).

Treatment success

Ozturan 2010 did not measure this outcome.

Health‐related quality of life

Ozturan 2010 did not report this outcome

Minor outcomes

At 6 weeks, grip strength was 33.2 kg with ESWT and 0.4 kg less (95% CI 4.9 more to 5.7 less; 37 participants) with autologous blood injection. At 3 months, grip strength was 36.9 kg with ESWT and 1.1 kg less (95% CI 2.84 more to 5.04 less; 37 participants) with autologous blood injection. At 6 months, grip strength was 37.2 kg with ESWT and 0.3 kg worse (95% CI 3.37 more to 3.97 less; 37 participants) with autologous blood injection. At 1 year, grip strength was 39.6 kg with ESWT and 2.3 kg better (95% CI 5.73 more to 1.13 less; 37 participants) with autologous blood injection (Analysis 5.4).

Ozturan 2010 did not report other minor outcomes.

Harms

Withdrawal due to adverse events

The only study in this comparison did not measure this outcome (Ozturan 2010).

Adverse events

Very low‐certainty evidence (downgraded once for bias and twice for very serious imprecision) suggests that we are uncertain whether autologous blood injection increases risk for adverse events compared with ESWT.

Adverse event rates were as follows: 12 out of 20 (60%) with ESWT versus 3 out of 20 (15%) with autologous blood injection (RR 0.25, 95% CI 0.08 to 0.75; 1 study) (Analysis 5.5).

PRP injection versus surgery

Benefits

Pain relief

Studies in this comparison did not measure this outcome.

Mean pain

Moderate‐certainty evidence from two studies shows that PRP injection probably does not improve pain compared with surgery (Merolla 2017; Watts 2020). At 24 months, we found low‐certainty evidence to suggest that PRP injection may increase pain compared with surgery. We downgraded the evidence for risk of bias to moderate certainty at 3 months and 6 months. At 6 weeks and 24 months, evidence was of low certainty (once for bias and once for imprecision). Statistical heterogeneity was 57% at 3 months, 51% at 6 months, and 88% at 12 months.

At 6 weeks, mean pain (0 to 10; higher is worse) was 5.4 with surgery and 0.8 points better with PRP injection (95% CI 0.38 better to 1.98 worse). At 3 months, mean pain was 3.8 points with surgery and 0.14 points better (95% CI ‐1.40 better to 1.12 worse; 2 studies, 153 participants) with PRP injection. At 6 months, mean pain was 2.6 points with surgery and 0.14 points better (95% CI 0.91 better to 1.20 worse; 159 participants, 2 studies) with PRP injection. At 12 months, mean pain was 1.8 points with surgery and 0.39 points better (95% CI 1.86 better to 2.64 worse; 153 participants; 2 studies) with PRP injection. At 24 months, mean pain was 2.1 points with surgery and 5.0 points worse (95% CI 4.02 worse to 5.98 worse; 1 study, 101 participants) with PRP injection (Analysis 6.1).

Function

Low‐certainty evidence from two studies suggests that PRP injection may not improve function compared with surgery (Merolla 2017; Watts 2020). At 24 months, PRP may result in deteriorated function. We downgraded the evidence to low for bias and imprecision.

At 6 weeks, function (0 to 100 scale; lower is better) was 49 with PRP and 7.00 points worse (95% CI 5.94 better to 19.94 worse; 56 participants) with surgery. At 3 months, mean function was 29.9 points with surgery and ‐0.59 points better (95% CI 19.63 better to 18.45 worse; 2 studies, 153 participants) with PRP injection. At 6 months, function was 22.7 points with surgery and 1.36 worse (95% CI 15.92 better to 18.63 worse;159 participants, 2 studies) with PRP injection. At 12 months, function was 17.5 points with surgery and 1.53 points worse (95% CI 13.27 better to 16.33 worse; 153 participants) with PRP injection. At 24 months, function was 21.2 points with surgery and 48.0 points worse (95% CI 40.2 worse to 55.8 worse; 1 study, 101 participants) with PRP injection (Analysis 6.2).

Treatment success

Neither study reported this outcome.

Health‐related quality of life

Neither study reported this outcome.

Minor outcomes

PRP probably does not improve grip strength compared with surgery (moderate‐quality evidence; downgraded for bias). After 6 months, PRP injection may decrease grip strength compared with arthroscopic surgery. The mean difference favoured surgery at every time point, and 95% confidence intervals did not overlap null effect at any time point. Grip strength is a measure of capacity rather than function/disability, and the clinically important difference is unclear in this condition.

At 3 months, grip strength was 48.4 kg with arthroscopic surgery and 1.0 kg worse (95% CI 0.99 better to 2.99 worse; 1 study, 101 participants) with PRP injection. At 6 months, grip strength was 50.2 kg with arthroscopic surgery and 26.8 kg worse (95% CI 29.03 worse to 24.57 worse; 101 participants) with PRP injection. At 12 months, grip strength was 47.3 kg with arthroscopic surgery and 23.7 kg worse (95% CI 25.59 worse to 21.81 worse; 101 participants) with PRP. At 24 months, grip strength was 48.4 kg with arthroscopic surgery and 25.6 kg worse (95% CI 27.31 worse to 23.89 worse; 101 participants) with PRP (Analysis 6.3).

Harms

Withdrawal due to adverse events

Studies in this comparison did not measure this outcome (Merolla 2017).

Adverse events

Watts 2020 reported one adverse event in the surgery group (wound debridement) and zero events in the PRP group; thus we could not estimate the relative risk (very low‐certainty evidence). Merolla 2017 did not report this outcome.

Autologous blood or PRP injection with tennis elbow strap and exercise versus tennis elbow strap and exercise alone

Two studies with 171 participants studied whether autologous blood or PRP injection improves clinical outcomes when added to tennis elbow strap and exercise (Gedik 2016; Lim 2017).

Benefits

Pain relief

Both studies did not measure this outcome.

Mean pain

Only Lim 2017 measured this outcome.

Low‐certainty evidence (downgraded for bias and imprecision) suggests that PRP injection plus tennis elbow strap and exercise may not provide clinically important improvement in pain compared with tennis elbow strap and exercise alone. At 4 weeks, mean pain (0 to 10 scale; higher is worse) had improved by 2.92 points with tennis elbow strap and exercise and by 1.14 additional points (95% CI 1.86 more to 0.42 more; 1 study, 120 participants) with PRP injection plus tennis elbow strap and exercise (Analysis 7.1).

Function

Although both studies provided data for this outcome, measurements were obtained at different time points and hence were not pooled. Low‐certainty evidence (downgraded for bias and imprecision) from Lim 2017 suggests that PRP injection plus tennis elbow strap and exercise may not provide clinically important improvement in function compared with tennis elbow strap and exercise alone. At 4 weeks, function (0 to 100 scale; lower is better) had improved by 8.42 points with tennis elbow strap and exercise and by 7.81 additional points (95% CI 12.71 more to 2.91 more; 1 study, 105 participants) with PRP injection plus tennis elbow strap and exercise.

Low‐certainty evidence from Gedik 2016 suggests that autologous blood injection may not improve function when added to tennis elbow strap and exercise. At 3 months, function (0 to 100; lower is better) was 8.6 points with tennis elbow strap and exercise and 1.6 points worse (95% CI 2.19 better to 5.39 worse; 1 study, 45 participants) with autologous blood injection plus tennis elbow strap and exercise. At 6 months, function was 3.9 points with tennis elbow strap and exercise and 2.46 points worse (95% CI 0.41 better to 5.33 worse; 1 study, 45 participants) with autologous blood injection plus tennis elbow strap and exercise. We downgraded the evidence for bias and imprecision ‐ only one study with few participants (Analysis 7.2).

Treatment success

Only Gedik 2016 provided data for this outcome. Very low‐certainty evidence suggests that we are uncertain whether autologous blood injection affects treatment success when added to tennis elbow strap and exercise. We downgraded the evidence for bias, imprecision, and indirectness; the confidence intervals include no effect, and instead of assessing subjective global success, Gedik 2016 researchers used own non‐validated measure. Treatment success rates were as follows: 13 out of 13 (100%) with tennis elbow strap and exercise versus 29 out of 32 (91%) with autologous blood injection plus tennis elbow strap and exercise (RR 0.93, 95% CI 0.79 to 1.08) at 3 months; 13 out of 13 (100%) with tennis elbow strap and exercise versus 31 out of 32 (97%) with autologous blood injection plus tennis elbow strap and exercise (RR 0.99, 95% CI 0.87 to 1.12) at 6 months (Analysis 7.4).

Health‐related quality of life

Both studies did not measure this outcome.

Minor outcomes

Low‐certainty evidence (downgraded for bias and imprecision) suggests that autologous blood injection may not improve grip strength when given in conjunction with tennis elbow strap and exercise. Grip strength is a measure of capacity rather than function/disability, and the clinically important difference is unclear in this condition.

At 3 months, grip strength was 29.1 kg with tennis elbow strap and exercise and 2.2 kg better (95% CI 7.1 better to 2.7 worse; 1 study, 45 participants) with autologous blood injection plus tennis elbow strap and exercise. At 6 months, grip strength was 30.9 kg with tennis elbow strap and exercise and 3.0 kg better (95% CI 8.85 better to 2.85 worse; 1 study, 45 participants) with autologous blood injection plus bandage and exercise (Analysis 7.3).

Harms

Withdrawal due to adverse events

Both studies did not measure this outcome.

Adverse events

Both studies did not measure this outcome.

PRP injection versus laser application

Benefits

Pain relief

None of the studies measured this outcome.

Mean pain

Compared with low‐lever laser application, PRP injection may not provide important pain improvement (low‐certainty evidence; downgraded for bias and imprecision; 95% CI overlaps with clinically important benefit).

At 3 months, mean pain (0 to 10; higher is worse) was 4.7 points with laser application and 1.0 point better (95% CI 2.13 better to 0.13 worse; 1 study, 56 participants) with PRP injection. At 6 months. mean pain was 3.6 points with laser application and 0.9 points better (95% CI 1.9 better to 0.1 worse; 1 study, 56 participants) with PRP injection. At 12 month, mean pain was 2.7 points with laser applications and 0.9 points better (95% CI 2.03 better to 0.23 worse; 1 study, 56 participants) with PRP injection (Analysis 8.1).

Function

Compared with low‐lever laser application, PRP injection may not improve function (low‐certainty evidence; downgraded for bias and imprecision; 95% CI overlaps with clinically important benefit).

At 3 months, function (0 to 100; lower is better) was 38.9 points with laser application and 9.1 points better (95% CI 20.03 better to 1.83 worse; 1 study, 56 participants) with PRP injection. At 6 months, function was 29.0 points with laser applications and 2.5 points better (95% CI 13.22 better to 8.22 worse; 1 study, 56 participants) with PRP injection. At 12 months, function was 26.7 points with laser application and 8.5 points better (95% CI 19.32 better to 2.32 worse; 1 study, 56 participants) with PRP injection (Analysis 8.2).

Treatment success

The only study in this comparison did not report this outcome (Tetschke 2015).

Minor outcomes

The only study in this comparison did not report any minor outcomes (Tetschke 2015).

Harms

Withdrawal due to adverse events

The only study in this comparison did not measure this outcome (Tetschke 2015).

Adverse events

Tetschke 2015 reported no adverse events in either group (estimates could not be calculated).

Autologous blood injection versus polidocanol injection

Benefits

Pain relief

The only study in this comparison did not measure this outcome (Branson 2016).

Mean pain

The only study in this comparison did not measure this outcome (Branson 2016).

Function

Autologous blood injection may not improve function compared with polidocanol injection (low‐certainty evidence; downgraded for bias and imprecision; 95% CI includes both clinically important benefit and harm at all time points).

At 6 weeks, function (0 to 100; lower is better) was 9.2 points with polidocanol injection and 4.4 points worse (95% CI 10.76 better to 19.56 worse; 1 study, 30 participants) with autologous blood injection. At 3 months, function was 19.9 points with polidocanol injection and 2.1 points better (95% CI 16.78 better to 12.58 worse; 1 study, 30 participants) with autologous blood injection. At 6 months, function was 28.9 points with polidocanol injection and 0.5 points worse (95% CI 15.21 better to 16.21 worse; 1 study, 30 participants) with autologous blood injection (Analysis 9.1).

Treatment success

Autologous blood injection may not improve treatment success rates compared with polidocanol injection (very low‐certainty evidence; downgraded for bias and twice for very serious imprecision; 95% CI overlaps large effect in both directions).

Treatment success (completely recovered or much improved) rates were as follows: 2 out of 16 (13%) with polidocanol injection versus 3 out of 14 (21%) with autologous blood injection (RR 1.71, 95% CI 0.33 to 8.83; 1 study) at 6 weeks; 6 out of 16 (38%) with polidocanol injection versus 5 out of 14 (36%) with autologous blood injection (RR 0.95, 95% CI 0.37 to 2.45; 1 study) at 3 months; and 13 out of 16 (81%) with polidocanol injection versus 9 out of 14 (64%) with autologous blood injection (RR 0.79, 95% CI 0.5 to 1.25; 1 study) at 6 months (Analysis 8.3).

Minor outcomes

The only study in this comparison did not measure this outcome (Branson 2016).

Harms

Withdrawal due to adverse events

The only study in this comparison did not measure this outcome (Branson 2016).

Adverse events

The only study in this comparison did not measure this outcome (Branson 2016).

Sensitivity analyses

We performed two sensitivity analyses to assess the effect of excluding studies with high or unclear risk for selection and detection bias for pain and function at the primary time point of 3 months. Sensitivity analyses were performed only for the primary comparison (autologous blood or PRP injection versus placebo).

Removing studies with inadequate or unclear allocation concealment ‐ Behera 2015; Mishra 2014; Montalvan 2015; Yerlikaya 2018 ‐ did not have a clinically important effect on pain estimates (0 to 10 scale; higher is worse) at 3 months. Clinically important benefit with autologous blood or PRP injection was unlikely both with (mean difference (MD) ‐0.16, 95% CI ‐0.60 to 0.29) and without (MD 0.40, 95% CI ‐0.27 to 1.08) studies with inadequate or unclear allocation concealment (Analysis 10.1).

Removing studies with inadequate or unclear allocation concealment ‐ Behera 2015; Mishra 2014; Montalvan 2015 ‐ did not have a clinically important effect on function estimates (0 to 100; higher is worse) at 3 months. Clinically important benefit with autologous blood or PRP injection was unlikely both with (MD ‐1.86, 95% CI ‐4.97 to 1.25) and without (MD ‐0.01, 95% CI ‐4.80 to 4.78) inadequate or unclear allocation concealment (Analysis 10.2).

Removing one study with inadequate or unclear participant blinding ‐ Behera 2015 ‐ did not have an important effect on pain estimates (0 to 10; higher is worse). Clinically important benefit was unlikely both with (MD ‐0.16, 95% CI ‐0.60 to 0.29) and without (MD 0.00, 95% CI‐0.47 to 0.47) the only study with inadequate or unclear participant blinding (Analysis 10.3).

Removing studies with inadequate or unclear participant blinding ‐ Behera 2015; Palacio 2016 ‐ did not have an important effect on function estimates (0 to 100; higher is worse). Clinically important benefit was unlikely both with (MD ‐1.86, 95% CI ‐4.97 to 1.25) and without (MD ‐0.23, 95% CI ‐3.74 to 3.29) studies with inadequate or unclear participant blinding (Analysis 10.4).

When we compared fixed‐effect estimates to random‐effects estimates (autologous blood or PRP versus placebo), we did not find evidence of small‐sample bias (Analysis 1.1; Analysis 1.2; Analysis 1.3; Analysis 1.4; Analysis 1.5; Analysis 1.6; only random‐effects estimates are shown).

Subgroup analysis

PRP versus autologous blood

Data from seven placebo‐controlled trials were available for analysis at the primary time point (3 months) comparing PRP versus placebo; two trials (with 98 participants receiving autologous blood) compared autologous blood to placebo (Linnanmäki 2020; Wolf 2011). We could include only mean pain and function in this analysis (Analysis 12.1; Analysis 12.2).

The type of injected product did not seem to modify the treatment effect (subgroup heterogeneity: I² = 0% in pain, I² = 22.6% in function). Mean difference in pain with PRP versus placebo was ‐0.19 (95% CI ‐0.63 to 0.25) and with autologous blood versus placebo was ‐0.12 (95% CI ‐1.40 to 1.15). Mean difference in function with PRP versus placebo was ‐2.30 (95% CI ‐5.24 to 0.64) and with autologous blood versus placebo was 0.50 (95% CI ‐6.56 to 7.55).

Leukocyte‐rich versus leukocyte‐poor PRP injection versus placebo

Data from up to seven trials were available for this subgroup analysis. We found no important differences between leukocyte‐rich and leukocyte‐poor PRP versus placebo in pain, function, treatment success, or adverse events.

At 3 months, mean difference in pain was ‐0.21 (95% CI ‐0.71 to 0.30; 292 participants, 3 studies) for leukocyte‐rich and ‐0.07 (95% CI ‐0.80 to 0.66; 193 participants, 4 studies) for leukocyte‐poor PRP versus placebo (Analysis 11.1); mean difference in function was ‐2.34 (95% CI ‐6.91 to 2.23; 272 participants, 3 studies) for leukocyte‐rich and ‐0.09 (95% CI ‐8.36 to 8.18; 132 participants, 3 studies) for leukocyte‐poor PRP versus placebo (Analysis 11.2); risk ratio for treatment success was 1.03 (95% CI 0.67 to 1.59; 275 participants, 2 studies) for leukocyte‐rich and 0.75 (95% CI 0.53 to 1.06; 107 participants, 2 studies) for leukocyte‐poor PRP versus placebo (Analysis 11.3); and risk ratio for adverse events was 1.14 (95% CI 0.71 to 1.84; 270 participants, 2 studies) for leukocyte‐rich and 1.15 (95% CI 0.53 to 2.51; 155 participants, 3 studies) for leukocyte‐poor PRP versus placebo (Analysis 11.4).

Other planned subgroup analyses could not be performed because included trials did not use frozen products.

Discussion

Summary of main results

Autologous blood or PRP injection versus placebo

Moderate‐certainty evidence indicates that autologous blood or platelet‐rich plasma (PRP) injection probably provides little or no improvement in pain or function for people with lateral elbow pain up to 3 months compared with placebo. Mean differences were clearly under minimal clinically important differences (MCIDs) across outcomes at all time points, and 95% confidence intervals suggest that clinically important benefit is unlikely.

The uncertainty of evidence was related to flaws in design or reporting of included studies, subjecting them to high or unclear risk of bias in various domains. We did not identify any studies comparing autologous blood or PRP injection to placebo with low risk of bias in all domains. However, as studies with risk of bias did not show benefit, we consider it unlikely that unbiased studies will show meaningful benefits.

At 12 months, uncertainty was greater regarding pain and function results, as 95% confidence intervals overlapped with clinically important benefit, and statistical heterogeneity was substantial (largely driven by Behera 2015, with unclear allocation concealment and participant blinding). Removing Behera 2015 decreased heterogeneity, and the estimate aligned with findings from the earlier time points showing no benefit with confidence intervals not including MCID values.

Sensitivity analyses suggest that estimates are robust to possible selection or detection bias; removing studies with inadequate or unclear allocation concealment or participant blinding supported findings from the primary analysis.

Very low‐quality evidence suggests that autologous blood or PRP injection may not improve treatment success, and low‐quality evidence suggests that autologous blood or PRP injection may increase the risk for adverse events. None of the studies in the primary comparison assessed health‐related quality of life. Although reported harms were mostly transient (injection site pain), they were not balanced by any clinically meaningful benefit.

One study found a difference in pain relief in a subset of participants who were followed up to 6 months based on a post‐hoc decision to continue follow‐up for a subset of participants (Mishra 2014). At 3 months, study authors did not report this outcome, and they did not respond to our queries. This may be a spurious finding related to alterations in study design and selective reporting; benefit was not corroborated when mean pain (or function) values were compared at any time point in that particular study, or in the meta‐analysis.

Other comparisons

When compared, low‐certainty evidence suggests that autologous blood or PRP injection is inferior to glucocorticoid injection for pain and function in the first 6 weeks of follow‐up. This transient effect vanishes by 3 months' follow‐up. Low‐certainty evidence suggests that autologous blood or PRP injection may result in greater improvement in pain and function after 6 months' follow‐up. However, confidence intervals overlapped with MCID for pain and function, suggesting that we are uncertain whether the difference is clinically important. Furthermore, as glucocorticoid injection exerts short‐term effects measured only in weeks (up to 6 to 8), the biological rationale for long‐term benefit or harm of glucocorticoid injection is unclear.

Based on moderate‐certainty evidence from four trials (Creaney 2011;Linnanmäki 2020; Raeissadat 2014; Thanasas 2011), PRP probably does not improve pain, function, or treatment success compared with autologous blood. Uncertainty is related to risk of bias in the included studies. This casts doubt over the biological rationale of concentrating platelet‐derived growth factors into the area of pathology to stimulate angiogenesis and healing. It also suggests that the extra cost incurred by the centrifugation process probably is not justified.

Subgroup analyses performed at the primary time point of this review (3 months) suggest that the type of PRP (leukocyte‐rich versus leukocyte‐poor) does not modify treatment effects with regard to mean pain or function, or rates of treatment success.

Other comparisons mainly supported our conclusion from the primary analysis. We did not find evidence of clinically important benefit when autologous blood or PRP injection was compared to extracorporeal shock wave therapy (ESWT), laser applications, or surgery, or when it was given in conjunction with dry needling or tennis elbow strap and exercise. Most of these comparisons included only one trial; thus the results were imprecise. Also, the included studies did not blind participants. Lack of blinding may bias outcomes, but as the comparators were also active treatments with potential placebo effects, it is difficult to say whether bias would alter findings to a considerable extent.

Overall completeness and applicability of evidence

The results of this review likely can be applied to a typical lateral elbow pain population. Studies included participants from 18 countries, and mean age was between 36 years and 53 years ‐ a typical age group for tendinopathy. All participants had lateral elbow pain (mean duration 1 month to 22 months) and were clinically diagnosed to have pain and soreness over the lateral humeral epicondyle. One study also included participants with medial elbow pain but they not report outcomes separately (Martin 2019). As only 11 (15%) participants had medial elbow pain at baseline, we included this study in the analyses.

As long as we have no evidence to indicate that autologous blood or PRP injection improves symptoms in this population (placebo comparison), comparisons to other active treatments offer little additional information. One study assessed the efficacy of PRP given in conjunction with dry needling, which may be analogous to comparison with placebo injection. It is unclear to what extent the observed improvements are related to needle injury and subsequent response in the tendinopathy area, and how much can be attributed to the natural course of the condition. All these aspects compromise the external validity of all other comparisons except placebo.

The results from this review are likely applicable to both PRP and autologous blood injections. We found moderate‐certainty evidence to show that treatment effects do not differ between PRP and autologous blood. Regarding the most pertinent comparison ‐ autologous blood or PRP injection versus placebo ‐ there was only one small study (19 participants receiving autologous blood or saline) injecting autologous blood; thus estimates were imprecise with autologous blood (Wolf 2011).

We found no evidence to support the hypothesis that leukocyte‐rich PRP would be beneficial as opposed to leukocyte‐poor PRP for this population (Fitzpatrick 2017). We did not perform other subgroup analyses, as we did not identify any previous evidence of possible interactions, and no studies used fresh versus frozen products.

For the primary comparison, study authors administered one injection of autologous blood or PRP except in one study (Martin 2019), which administered two injections to participants. Evidence that the number of injections may have any impact on outcomes is limited (Glanzmann 2015), and estimates from Martin 2019 were in line with those of other studies.

None of the included studies for the primary comparison measured health‐related quality of life; thus we could not calculate any estimate for this outcome. As the typical symptom is pain, and functional disturbances strongly correlate with pain (Rompe 2007), it is unlikely that generic health‐related quality of life measures would capture effects beyond pain and function measures. Estimates for pain relief and global perceived treatment success were imprecise; future trials may improve the certainty regarding these outcomes.

Quality of the evidence

Regarding the primary comparison autologous blood or PRP injection versus placebo, the certainty of evidence was moderate for mean pain and function, low for adverse events, and very low for pain relief, treatment success, and withdrawal due to adverse events.

We downgraded the evidence for pain and function once because none of the included trials had low risk of bias in all domains. At 12 months' follow‐up, the confidence intervals for mean pain and function overlapped with clinically important benefit; hence we downgraded the certainty of evidence (another step) to low.

Regarding pain relief, we found no data at 3 months (very low‐certainty evidence), but at 6 months, one study reported this outcome. We downgraded this outcome at 6 months due to bias and low numbers of events.

For adverse events, the certainty of evidence was low, as we downgraded the evidence two levels due to bias and imprecision because the confidence intervals overlapped with no effect. Regarding treatment success, we further downgraded the evidence to very low due to indirectness, in addition to bias and imprecision. None of the studies measured perceived global success but rather reported treatment success based on various definitions, usually using a cutoff value for pain or function score. None of the studies for the primary comparison reported health‐related quality of life; thus we could not calculate estimates (very low‐certainty evidence).

For PRP versus glucocorticoid injection, we downgraded the evidence for pain due to risk of bias and inconsistency; the statistical heterogeneity (I² = 72% to 77%) could not be explained by study characteristics or treatment protocols, and estimates from the included studies ranged from no effect to clinically important effect. For function, we downgraded evidence for bias and imprecision; although we considered downgrading for inconsistency, this was not done, as the direction of effect was consistent across studies.

For other comparisons, the certainty of evidence was low to very low due to bias (primarily detection bias due to lack of blinding) and imprecision. Only for PRP versus autologous blood (and PRP versus surgery) for mean pain, the confidence intervals for function were precise enough to exclude clinically important effects; thus the evidence was graded as moderate certainty.

Potential biases in the review process

We searched all relevant databases with no language restrictions, as well as registries for ongoing trials, and to the best of our knowledge, we identified all relevant trials. Two review authors independently assessed trials for potential inclusion, performed data extraction, and conducted risk of bias assessments, with a third review author adjudicating in case of discrepancy.

One review author (TK) is the study investigator for a trial included in this review (Linnanmäki 2020); another review author (RB) is the study investigator for an ongoing trial in this review (ACTRN12613000616774). To avoid any bias, these trials were independently assessed by two other review authors to discern whether they fulfilled inclusion criteria for this review. Neither review author was involved in data extraction nor risk of bias assessments for his or her own trial. We identified ten ongoing trials comparing PRP with placebo (ACTRN12613000616774; ChiCTR1900024425; EUCTR2013‐000478‐32‐ES; EUCTR2013‐004875‐12‐CZ; ISRCTN12951626; NCT01476605; NCT03984955; NCT03987256; NTR4569; NTR5005); once published, data from these trials may increase the precision of estimates. One study awaiting classification is a placebo‐controlled trial (Chiavaras 2014); similar to the ongoing trials mentioned above, results of this trial may affect estimates.

Agreements and disagreements with other studies or reviews

Several systematic reviews have assessed the effects of autologous blood or PRP injection for lateral elbow pain, with conflicting conclusions ranging from positive in Ahmad 2013; Arirachakaran 2016; Dong 2016; Mi 2017; and Murray 2015 to strong evidence against in de Vos 2014.

Some positive early reviews included non‐randomised trials or observational studies (Ahmad 2013; Rabago 2009); these review authors did not perform meta‐analyses. Some of the positive reviews base their conclusions on the comparison against glucocorticoid injection and not against placebo (Arirachakaran 2016; Mi 2017; Murray 2015). Regarding this specific comparison, their conclusions are in line with ours. However, the efficacy of treatment should first be assessed against placebo; only if the treatment is shown to work, comparison against other (effective) treatments is meaningful.

Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Analysis 1.1

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 1: Pain relief ≥ 30% or ≥ 50%

Navigate to figure in ReviewOpen in new tab

Analysis 1.2

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 2: Mean pain (VAS 0 to 10, PRTEE)

Navigate to figure in ReviewOpen in new tab

Analysis 1.3

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 3: Function (DASH, MMCPIE, Roles‐Maudsley)

Navigate to figure in ReviewOpen in new tab

Analysis 1.4

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 4: Treatment success (> 25% improvement in pain or function)

Navigate to figure in ReviewOpen in new tab

Analysis 1.5

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 5: Withdrawal due to AEs

Navigate to figure in ReviewOpen in new tab

Analysis 1.6

Comparison 1: Autologous blood or PRP injection versus placebo injection, Outcome 6: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 2.1

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 1: Pain relief ≥ 50%

Navigate to figure in ReviewOpen in new tab

Analysis 2.2

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 2: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 2.3

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 3: Function (various scales)

Navigate to figure in ReviewOpen in new tab

Analysis 2.4

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 4: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 2.5

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 5: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 2.6

Comparison 2: Autologous blood or PRP injection versus glucocorticoid injection, Outcome 6: Grip strength

Navigate to figure in ReviewOpen in new tab

Analysis 3.1

Comparison 3: PRP and dry needling versus dry needling alone, Outcome 1: Pain

Navigate to figure in ReviewOpen in new tab

Analysis 3.2

Comparison 3: PRP and dry needling versus dry needling alone, Outcome 2: Function

Navigate to figure in ReviewOpen in new tab

Analysis 3.3

Comparison 3: PRP and dry needling versus dry needling alone, Outcome 3: Withdrawal due to adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 3.4

Comparison 3: PRP and dry needling versus dry needling alone, Outcome 4: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 4.1

Comparison 4: PRP versus autologous blood, Outcome 1: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 4.2

Comparison 4: PRP versus autologous blood, Outcome 2: Function (various scales)

Navigate to figure in ReviewOpen in new tab

Analysis 4.3

Comparison 4: PRP versus autologous blood, Outcome 3: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 4.4

Comparison 4: PRP versus autologous blood, Outcome 4: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 5.1

Comparison 5: Autologous blood versus ESWT, Outcome 1: Pain relief > 50%

Navigate to figure in ReviewOpen in new tab

Analysis 5.2

Comparison 5: Autologous blood versus ESWT, Outcome 2: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 5.3

Comparison 5: Autologous blood versus ESWT, Outcome 3: Function (various scales)

Navigate to figure in ReviewOpen in new tab

Analysis 5.4

Comparison 5: Autologous blood versus ESWT, Outcome 4: Grip strength

Navigate to figure in ReviewOpen in new tab

Analysis 5.5

Comparison 5: Autologous blood versus ESWT, Outcome 5: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 6.1

Comparison 6: PRP versus surgery, Outcome 1: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 6.2

Comparison 6: PRP versus surgery, Outcome 2: Function

Navigate to figure in ReviewOpen in new tab

Analysis 6.3

Comparison 6: PRP versus surgery, Outcome 3: Grip strength

Navigate to figure in ReviewOpen in new tab

Analysis 7.1

Comparison 7: Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise, Outcome 1: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 7.2

Comparison 7: Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise, Outcome 2: Mean function

Navigate to figure in ReviewOpen in new tab

Analysis 7.3

Comparison 7: Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise, Outcome 3: Hand grip strength

Navigate to figure in ReviewOpen in new tab

Analysis 7.4

Comparison 7: Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise, Outcome 4: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 8.1

Comparison 8: PRP versus laser applications, Outcome 1: Pain

Navigate to figure in ReviewOpen in new tab

Analysis 8.2

Comparison 8: PRP versus laser applications, Outcome 2: Function

Navigate to figure in ReviewOpen in new tab

Analysis 8.3

Comparison 8: PRP versus laser applications, Outcome 3: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 8.4

Comparison 8: PRP versus laser applications, Outcome 4: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 9.1

Comparison 9: Autologous blood versus polidocanol injection, Outcome 1: Function

Navigate to figure in ReviewOpen in new tab

Analysis 9.2

Comparison 9: Autologous blood versus polidocanol injection, Outcome 2: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 10.1

Comparison 10: Sensitivity analysis (mean pain and function at 3 months), Outcome 1: Pain at 3 months (low vs high or unclear risk of selection bias)

Navigate to figure in ReviewOpen in new tab

Analysis 10.2

Comparison 10: Sensitivity analysis (mean pain and function at 3 months), Outcome 2: Function at 3 months (low vs unclear or high selection bias)

Navigate to figure in ReviewOpen in new tab

Analysis 10.3

Comparison 10: Sensitivity analysis (mean pain and function at 3 months), Outcome 3: Pain at 3 months (adequate vs inadequate participant blinding)

Navigate to figure in ReviewOpen in new tab

Analysis 10.4

Comparison 10: Sensitivity analysis (mean pain and function at 3 months), Outcome 4: Function at 3 months (adequate vs inadequate participant blinding)

Navigate to figure in ReviewOpen in new tab

Analysis 11.1

Comparison 11: Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months, Outcome 1: Mean pain

Navigate to figure in ReviewOpen in new tab

Analysis 11.2

Comparison 11: Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months, Outcome 2: Function

Navigate to figure in ReviewOpen in new tab

Analysis 11.3

Comparison 11: Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months, Outcome 3: Treatment success

Navigate to figure in ReviewOpen in new tab

Analysis 11.4

Comparison 11: Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months, Outcome 4: Adverse events

Navigate to figure in ReviewOpen in new tab

Analysis 12.1

Comparison 12: Subgroup PRP versus autologous blood at 3 months, Outcome 1: Mean pain at 3 months

Navigate to figure in ReviewOpen in new tab

Analysis 12.2

Comparison 12: Subgroup PRP versus autologous blood at 3 months, Outcome 2: Mean function at 3 months

Navigate to figure in ReviewOpen in new tab

Analysis 12.3

Comparison 12: Subgroup PRP versus autologous blood at 3 months, Outcome 3: Withdrawals due to adverse events

Navigate to figure in ReviewOpen in new tab

Summary of findings 1. Autologous blood or PRP versus placebo at 3 months' follow‐up

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№. of participants (studies)	Quality of the evidence (GRADE)	Comments
Autologous blood or PRP versus placebo at 3 months' follow‐up
Patient or population: lateral elbow pain Setting: outpatient Intervention: autologous blood or PRP injection Comparison: placebo
Outcomes	Risk with placebo	Risk with autologous blood or PRP injection	Relative effect (95% CI)	№. of participants (studies)	Quality of the evidence (GRADE)	Comments
Pain (VAS, PRTEE) translated to 0 to 10, where 0 is no pain Follow‐up: 3 months	Mean pain in the placebo group was 3.7 points^a	Mean pain was 0.16 points better (0.60 better to 0.29 worse)	‐	523 participants (8 studies)	⊕⊕⊕⊝ Moderate^b	PRP probably provides little to no benefit for pain. Absolute benefit 1.6% better (6% better to 3% worse); relative benefit 2.3% better (9% better to 4% worse).^c Not clinically significant
Function (PRTEE, DASH, MMCPIE, Roles‐Maudsley), translated to 0 to 100, where 0 is best function, or no disability Follow‐up: 3 months	Mean function in placebo was 27.5 points^d	Mean function was 1.86 points better (4.97 better to 1.25 worse)	‐	502 participants (8 studies)	⊕⊕⊕⊝ Moderate^d	PRP probably provides little to no benefit for function. Absolute benefit 1.9% better (5% better to 1% worse); relative benefit 4% (11% better to 3% worse).^e Not clinically significant
Treatment success (> 25% improvement in pain or function) Follow‐up: 3 months	650/1000	670/1000 (582 to 765)	RR 1.0 (0.83 to 1.19)	372 participants (4 studies)	⊕⊝⊝⊝ Very low^b,e,f	We are uncertain whether PRP provides better treatment success. Absolute benefit 0% higher (11.1% lower to 12.4% higher); relative benefit 0% higher (17% lower to 19% higher)
Health‐related quality of life Not measured	See comment	See comment	‐	(0 studies)	See comment	Not measured in any of the included studies
Pain relief ≥ 30% or ≥ 50% Not measured at 3 months	See comment	See comment	‐	(0 studies)	See comment	Not reported in any of the included studies at 3 months
Withdrawal due to adverse events	77/1000	24/1000 (2 to 225)	RR 0.32 (0.03 to 2.92)	80 participants (1 study)	⊕⊝⊝⊝ Very low^b,g	We are uncertain whether PRP results in more people withdrawing due to adverse events. Absolute change 5.2% less (7.5% less to 14.8% more); relative change 68% less (97% less to 192% more)
Adverse events (pain and swelling at injection site and limitation of elbow movement following injection) Follow‐up: 12 months	168/1000	192/1000 (128 to 290)	RR 1.14 (0.76 to 1.72)	425 participants (5 studies)	⊕⊕⊝⊝ Low^b,f	PRP may not increase the number of people reporting adverse events. Absolute change 2.4% more (4% less to 12% more); relative change 14% more (24% less to 72% more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; DASH: Disabilities of the Arm, Shoulder and Hand; MMCPIE: Modified Mayo Clinic Performance Index for Elbow; OR: odds ratio; PRP: platelet‐rich plasma; PRTEE: Patient‐Rated Tennis Elbow Evaluation; RR: risk ratio; VAS: visual analogue scale.
GRADE Working Group grades of evidence. High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aMedian pain value from placebo groups in the included studies (excluding Mishra 2014, which reported percentage improvement). ^bDowngraded one level for risk of bias in the included studies. ^cRelative changes calculated relative to baseline in control group (i.e. mean difference divided by mean at baseline in the placebo group) (from Montalvan 2015 ‐ value for pain was 7 points on a 0 to 10 scale; for function from Krogh 2013 ‐ value was 47 points on a 0 to 100 scale). Absolute change calculated as mean difference divided by scale of the instrument, expressed as percentage. ^dMedian function from placebo groups at 3 months' follow‐up. ^eDowngraded one level for indirectness, as none of the studies measured global participant‐reported success directly but measured pain or function improvement cutoff values. ^fDowngraded one level for imprecision due to 95% CIs including both effect and no effect. ^gDowngraded evidence by two levels because of a small number of events leading to very wide confidence intervals, which overlap relative risk estimates of 0.75 and 1.25.

Summary of findings 1. Autologous blood or PRP versus placebo at 3 months' follow‐up

Navigate to table in Review

Table 1. Outcome reporting bias In trials (ORBIT) matrix

Study ID	Participant‐reported pain relief ≥ 30%	Pain	Function or disability	Treatment success	Health‐related quality of life	Withdrawal due to adverse events	Adverse events
Arik 2014	?	Full	Full	Full	?	?	Full
Behera 2015	?	Full	Full	?	?	Full	Full
Branson 2016	Not measured	Not measured	Full	Full	Not measured	?	Full
Creaney 2011	?	?	Full	Full	?	Full	?
Dojode 2012	?	Full	?	Full	?	Full	Full
Gautam 2015	?	Partial	Partial	?	?	Full	?
Gedik 2016	?	?	Full	Full	?	Full	?
Gosens 2011	Not measured	Full	Full	Full	Not measured	Full	Full
Gupta 2019	?	Full	Full	Full	?	?	Full
Jindal 2013	?	Full	?	Full	?	Full	?
Kazemi 2010	Not measured	Full	Full	Not measured	Not measured	Full	Full
Krogh 2013	Not measured	Full	Full	Not measured	Not measured	Full	Full
Lebiedziński 2015	?	?	Full	Full	?	Full	Full
Lim 2017	?	Partial	Partial	Full	?	Full	Full
Linnanmäki 2020	Not measured	Full	Full	Not measured	Not measured	Not measured	Full
Martin 2019	Not measured	Full	Full	Full	Not measured	Full	Full
Martínez‐Montiel 2015	?	Full	Full	?	?	Full	?
Merolla 2017	?	Partial	Partial	Measured	?	Full	?
Mishra 2014	Full	Partial	Partial	Full	?	Full	Full
Montalvan 2015	Not measured	Full	Full	Not measured	Not measured	Full	Full
Omar 2012	?	Full	Full	?	?	Full	?
Ozturan 2010	Full	Full	Full	?	?	Full	Full
Palacio 2016	?	?	Full	Full	?	?	?
Raeissadat 2014	?	Full	Full	Full	?	Full	?
Schoffl 2017	?	?	Full	?	?	?	?
Stenhouse 2013	?	Full	Full	Full	?	Full	Full
Tetschke 2015	?	Full	Full	Full	?	Full	?
Thanasas 2011	?	Full	Full	?	?	Full	Full
Wolf 2011	?	Full	Full	?	?	Full	?
Watts 2020	not measured	Full	Full	Not measured	Not measured	Not measured	Full
Yadav 2015	?	Partial	Partial	?	?	Full	?
Yerlikaya 2018	?	Full	Measured	?	?	Full	Measured
'Full': sufficient data for inclusion in a meta‐analysis were reported (e.g. mean, standard deviation, sample size per group for continuous outcomes). 'Partial': insufficient data for inclusion in a meta‐analysis were reported (e.g. means only, with no measures of variance). 'Measured': outcome was measured but no outcome data were reported. 'Not measured': outcome was not measured by trialists. '?': unclear whether the outcome was measured or not (as a trial protocol was unavailable).

Table 1. Outcome reporting bias In trials (ORBIT) matrix

Navigate to table in Review

Comparison 1. Autologous blood or PRP injection versus placebo injection

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Pain relief ≥ 30% or ≥ 50% Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1.1 6 months	1	119	Risk Ratio (M‐H, Random, 95% CI)	1.36 [1.08, 1.72]
1.2 Mean pain (VAS 0 to 10, PRTEE) Show forest plot	8		Mean Difference (IV, Random, 95% CI)	Subtotals only

1.2.1 to 3 weeks	1	19	Mean Difference (IV, Random, 95% CI)	2.00 [‐0.65, 4.65]
1.2.2 > 3 weeks to 6 weeks	7	570	Mean Difference (IV, Random, 95% CI)	0.26 [‐0.14, 0.65]
1.2.3 > 6 weeks to 3 months	8	523	Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.60, 0.29]
1.2.4 > 3 months to 6 months	7	387	Mean Difference (IV, Random, 95% CI)	‐0.45 [‐1.49, 0.59]
1.2.5 > 6 months to 12 months	5	241	Mean Difference (IV, Random, 95% CI)	‐0.69 [‐1.78, 0.39]
1.3 Function (DASH, MMCPIE, Roles‐Maudsley) Show forest plot	9		Mean Difference (IV, Random, 95% CI)	Subtotals only

1.3.1 Up to 3 weeks	1	19	Mean Difference (IV, Random, 95% CI)	12.00 [‐5.33, 29.33]
1.3.2 > 3 weeks to 6 weeks	6	473	Mean Difference (IV, Random, 95% CI)	1.30 [‐1.64, 4.25]
1.3.3 > 6 weeks to 3 months	8	502	Mean Difference (IV, Random, 95% CI)	‐1.86 [‐4.97, 1.25]
1.3.4 > 3 months to 6 months	7	379	Mean Difference (IV, Random, 95% CI)	‐1.15 [‐8.62, 6.31]
1.3.5 > 6 months to 12 months	4	203	Mean Difference (IV, Random, 95% CI)	‐5.81 [‐16.66, 5.05]
1.4 Treatment success (> 25% improvement in pain or function) Show forest plot	4		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.4.1 > 6 weeks to 3 months	4	372	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.83, 1.19]
1.5 Withdrawal due to AEs Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.5.1 Total	1	80	Risk Ratio (M‐H, Random, 95% CI)	0.32 [0.03, 2.92]
1.6 Adverse events Show forest plot	5		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.6.1 Total	5	425	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.76, 1.72]

Comparison 1. Autologous blood or PRP injection versus placebo injection

Navigate to table in Review

Comparison 2. Autologous blood or PRP injection versus glucocorticoid injection

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Pain relief ≥ 50% Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

2.1.1 6 weeks	1	38	Risk Ratio (M‐H, Random, 95% CI)	0.19 [0.07, 0.53]
2.1.2 1 year	1	38	Risk Ratio (M‐H, Random, 95% CI)	1.67 [1.03, 2.71]
2.2 Mean pain Show forest plot	13		Mean Difference (IV, Random, 95% CI)	Subtotals only

2.2.1 Up to 3 weeks	5	280	Mean Difference (IV, Random, 95% CI)	2.06 [0.67, 3.45]
2.2.2 > 3 weeks to 6 weeks	13	707	Mean Difference (IV, Random, 95% CI)	0.99 [0.21, 1.77]
2.2.3 > 6 weeks to 3 months	11	627	Mean Difference (IV, Random, 95% CI)	‐1.15 [‐1.71, ‐0.59]
2.2.4 > 3 months to 6 months	8	427	Mean Difference (IV, Random, 95% CI)	‐1.55 [‐2.21, ‐0.90]
2.2.5 > 6 months to 1 year	4	258	Mean Difference (IV, Random, 95% CI)	‐1.59 [‐2.22, ‐0.97]
2.2.6 > 1 year	1	100	Mean Difference (IV, Random, 95% CI)	‐2.11 [‐3.19, ‐1.03]
2.3 Function (various scales) Show forest plot	14		Mean Difference (IV, Random, 95% CI)	Subtotals only

2.3.1 Up to 3 weeks	3	170	Mean Difference (IV, Random, 95% CI)	16.46 [‐4.23, 37.15]
2.3.2 > 3 weeks to 6 weeks	13	724	Mean Difference (IV, Random, 95% CI)	6.11 [1.79, 10.44]
2.3.3 > 6 weeks to 3 months	12	635	Mean Difference (IV, Random, 95% CI)	‐10.19 [‐14.16, ‐6.21]
2.3.4 > 3 months to 6 months	7	374	Mean Difference (IV, Random, 95% CI)	‐5.07 [‐12.66, 2.52]
2.3.5 > 6 months to 1 year	4	317	Mean Difference (IV, Random, 95% CI)	‐8.94 [‐12.09, ‐5.78]
2.3.6 > 1 year	1	100	Mean Difference (IV, Random, 95% CI)	‐18.90 [‐28.27, ‐9.53]
2.4 Treatment success Show forest plot	7		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

2.4.1 Up to 6 weeks	5	317	Risk Ratio (M‐H, Random, 95% CI)	0.26 [0.07, 0.95]
2.4.2 > 6 weeks to 3 months	3	188	Risk Ratio (M‐H, Random, 95% CI)	1.56 [1.08, 2.26]
2.4.3 > 3 months to 6 months	3	187	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.23, 4.44]
2.4.4 > 6 months to 1 year	3	300	Risk Ratio (M‐H, Random, 95% CI)	1.61 [0.55, 4.75]
2.4.5 > 1 year	2	199	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.31, 3.16]
2.5 Adverse events Show forest plot	5	317	Risk Ratio (M‐H, Random, 95% CI)	1.64 [0.65, 4.12]

2.6 Grip strength Show forest plot	6		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.6.1 Up to 3 weeks	3	170	Std. Mean Difference (IV, Random, 95% CI)	‐0.52 [‐0.87, ‐0.16]
2.6.2 > 3 weeks to 6 weeks	6	348	Std. Mean Difference (IV, Random, 95% CI)	‐0.26 [‐0.68, 0.16]
2.6.3 > 6 weeks to 3 months	6	348	Std. Mean Difference (IV, Random, 95% CI)	0.56 [0.19, 0.93]
2.6.4 > 3 months to 6 months	2	68	Std. Mean Difference (IV, Random, 95% CI)	0.35 [‐0.13, 0.83]
2.6.5 > 6 months to 1 year	2	118	Std. Mean Difference (IV, Random, 95% CI)	0.66 [0.29, 1.03]

Comparison 2. Autologous blood or PRP injection versus glucocorticoid injection

Navigate to table in Review

Comparison 3. PRP and dry needling versus dry needling alone

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Pain Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1.1 > 6 weeks to 3 months	1	28	Mean Difference (IV, Random, 95% CI)	‐0.14 [‐2.13, 1.85]
3.1.2 > 3 months to 6 months	1	28	Mean Difference (IV, Random, 95% CI)	‐0.35 [‐2.88, 2.18]
3.2 Function Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.2.1 > 3 weeks to 6 weeks	1	36	Mean Difference (IV, Random, 95% CI)	9.60 [‐2.49, 21.69]
3.2.2 > 6 weeks to 3 months	1	28	Mean Difference (IV, Random, 95% CI)	2.80 [‐16.88, 22.48]
3.2.3 > 3 months to 6 months	1	28	Mean Difference (IV, Random, 95% CI)	5.70 [‐14.36, 25.76]
3.2.4 > 6 months to 12 months	1	36	Mean Difference (IV, Random, 95% CI)	4.30 [‐9.70, 18.30]
3.3 Withdrawal due to adverse events Show forest plot	1	28	Risk Ratio (M‐H, Fixed, 95% CI)	1.73 [0.18, 16.99]

3.4 Adverse events Show forest plot	1	28	Risk Ratio (M‐H, Fixed, 95% CI)	1.73 [0.18, 16.99]

Comparison 3. PRP and dry needling versus dry needling alone

Navigate to table in Review

Comparison 4. PRP versus autologous blood

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Mean pain Show forest plot	3		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1.1 > 3 weeks to 6 weeks	3	169	Mean Difference (IV, Random, 95% CI)	‐0.24 [‐1.21, 0.73]
4.1.2 > 6 weeks to 3 months	3	169	Mean Difference (IV, Random, 95% CI)	‐0.40 [‐1.11, 0.30]
4.1.3 > 3 months to 6 months	3	169	Mean Difference (IV, Random, 95% CI)	‐0.28 [‐1.04, 0.48]
4.1.4 > 6 months to 12 months	2	141	Mean Difference (IV, Random, 95% CI)	0.05 [‐1.12, 1.22]
4.2 Function (various scales) Show forest plot	4		Mean Difference (IV, Random, 95% CI)	Subtotals only

4.2.1 > 3 weeks to 6 weeks	4	276	Mean Difference (IV, Random, 95% CI)	‐3.44 [‐6.60, ‐0.28]
4.2.2 > 6 weeks to 3 months	4	292	Mean Difference (IV, Random, 95% CI)	‐3.25 [‐6.33, ‐0.17]
4.2.3 > 3 months to 6 months	4	297	Mean Difference (IV, Random, 95% CI)	‐2.83 [‐6.02, 0.37]
4.2.4 > 6 months to 12 months	2	140	Mean Difference (IV, Random, 95% CI)	‐0.71 [‐8.53, 7.11]
4.3 Treatment success Show forest plot	2	191	Risk Ratio (M‐H, Random, 95% CI)	1.03 [0.77, 1.37]

4.4 Adverse events Show forest plot	2	139	Risk Ratio (M‐H, Random, 95% CI)	2.25 [0.90, 5.62]

Comparison 4. PRP versus autologous blood

Navigate to table in Review

Comparison 5. Autologous blood versus ESWT

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Pain relief > 50% Show forest plot	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only

5.1.1 6 weeks	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.38 [0.12, 1.21]
5.1.2 1 year	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.68, 1.16]
5.2 Mean pain Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.2.1 > 3 weeks to 6 weeks	1	37	Mean Difference (IV, Random, 95% CI)	0.63 [‐0.28, 1.54]
5.2.2 > 6 weeks to 3 months	1	37	Mean Difference (IV, Random, 95% CI)	0.29 [‐0.75, 1.33]
5.2.3 > 3 months to 6 months	1	37	Mean Difference (IV, Random, 95% CI)	0.23 [‐0.78, 1.24]
5.2.4 > 6 months to 1 year	1	37	Mean Difference (IV, Random, 95% CI)	0.23 [‐0.61, 1.07]
5.3 Function (various scales) Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.3.1 > 3 weeks to 6 weeks	1	37	Mean Difference (IV, Random, 95% CI)	3.80 [‐1.56, 9.16]
5.3.2 > 6 weeks to 3 months	1	37	Mean Difference (IV, Random, 95% CI)	1.40 [‐5.82, 8.62]
5.3.3 > 3 months to 6 months	1	37	Mean Difference (IV, Random, 95% CI)	1.50 [‐4.17, 7.17]
5.3.4 > 6 months to 1 year	1	37	Mean Difference (IV, Random, 95% CI)	‐0.90 [‐5.98, 4.18]
5.4 Grip strength Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.4.1 > 3 weeks to 6 weeks	1	37	Mean Difference (IV, Random, 95% CI)	0.40 [‐4.90, 5.70]
5.4.2 > 6 weeks to 3 months	1	37	Mean Difference (IV, Random, 95% CI)	1.10 [‐2.84, 5.04]
5.4.3 > 3 months to 6 months	1	37	Mean Difference (IV, Random, 95% CI)	0.30 [‐3.37, 3.97]
5.4.4 > 6 months to 1 year	1	37	Mean Difference (IV, Random, 95% CI)	‐2.30 [‐5.73, 1.13]
5.5 Adverse events Show forest plot	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.25 [0.08, 0.75]

Comparison 5. Autologous blood versus ESWT

Navigate to table in Review

Comparison 6. PRP versus surgery

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Mean pain Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

6.1.1 6 weeks	1	56	Mean Difference (IV, Random, 95% CI)	0.80 [‐0.38, 1.98]
6.1.2 3 months	2	153	Mean Difference (IV, Random, 95% CI)	‐0.14 [‐1.40, 1.12]
6.1.3 6 months	2	159	Mean Difference (IV, Random, 95% CI)	0.14 [‐0.91, 1.20]
6.1.4 12 months	2	153	Mean Difference (IV, Random, 95% CI)	0.39 [‐1.86, 2.64]
6.1.5 24 months	1	101	Mean Difference (IV, Random, 95% CI)	5.00 [4.02, 5.98]
6.2 Function Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

6.2.1 6 weeks	1	56	Mean Difference (IV, Random, 95% CI)	7.00 [‐5.94, 19.94]
6.2.2 3 months	2	153	Mean Difference (IV, Random, 95% CI)	‐0.59 [‐19.63, 18.45]
6.2.3 6 months	2	159	Mean Difference (IV, Random, 95% CI)	1.36 [‐15.92, 18.63]
6.2.4 12 months	2	153	Mean Difference (IV, Random, 95% CI)	1.53 [‐13.27, 16.33]
6.2.5 24 months	1	101	Mean Difference (IV, Random, 95% CI)	48.00 [40.20, 55.80]
6.3 Grip strength Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

6.3.1 3 months	1	101	Mean Difference (IV, Random, 95% CI)	1.00 [‐0.99, 2.99]
6.3.2 6 months	1	101	Mean Difference (IV, Random, 95% CI)	‐26.80 [‐29.03, ‐24.57]
6.3.3 12 months	1	101	Mean Difference (IV, Random, 95% CI)	‐23.70 [‐25.59, ‐21.81]
6.3.4 24 months	1	101	Mean Difference (IV, Random, 95% CI)	‐25.60 [‐27.31, ‐23.89]

Comparison 6. PRP versus surgery

Navigate to table in Review

Comparison 7. Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Mean pain Show forest plot	1	120	Mean Difference (IV, Random, 95% CI)	‐1.14 [‐1.86, ‐0.42]

7.1.1 > 3 weeks to 6 weeks	1	120	Mean Difference (IV, Random, 95% CI)	‐1.14 [‐1.86, ‐0.42]
7.2 Mean function Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

7.2.1 > 3 weeks to 6 weeks	1	105	Mean Difference (IV, Random, 95% CI)	‐7.81 [‐12.71, ‐2.91]
7.2.2 3 months	1	45	Mean Difference (IV, Random, 95% CI)	1.60 [‐2.19, 5.39]
7.2.3 6 months	1	45	Mean Difference (IV, Random, 95% CI)	2.46 [‐0.41, 5.33]
7.3 Hand grip strength Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

7.3.1 3 months	1	45	Mean Difference (IV, Random, 95% CI)	‐2.20 [‐7.10, 2.70]
7.3.2 6 months	1	45	Mean Difference (IV, Random, 95% CI)	‐3.00 [‐8.85, 2.85]
7.4 Treatment success Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

7.4.1 3 months	1	45	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.79, 1.08]
7.4.2 6 months	1	45	Risk Ratio (M‐H, Random, 95% CI)	0.99 [0.87, 1.12]

Comparison 7. Autologous blood plus tennis elbow strap and exercise versus tennis elbow strap and exercise

Navigate to table in Review

Comparison 8. PRP versus laser applications

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 Pain Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

8.1.1 >6 weeks to 3 months	1	56	Mean Difference (IV, Random, 95% CI)	‐1.00 [‐2.13, 0.13]
8.1.2 >3 months to 6 months	1	56	Mean Difference (IV, Random, 95% CI)	‐0.90 [‐1.90, 0.10]
8.1.3 >6 months to 12 months	1	56	Mean Difference (IV, Random, 95% CI)	‐0.90 [‐2.03, 0.23]
8.2 Function Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

8.2.1 3 month	1	56	Mean Difference (IV, Random, 95% CI)	‐9.10 [‐20.03, 1.83]
8.2.2 6 month	1	56	Mean Difference (IV, Random, 95% CI)	‐2.50 [‐13.22, 8.22]
8.2.3 12 month	1	56	Mean Difference (IV, Random, 95% CI)	‐8.50 [‐19.32, 2.32]
8.3 Treatment success Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

8.4 Adverse events Show forest plot	1	56	Risk Ratio (M‐H, Random, 95% CI)	Not estimable

Comparison 8. PRP versus laser applications

Navigate to table in Review

Comparison 9. Autologous blood versus polidocanol injection

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 Function Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

9.1.1 6 weeks	1	30	Mean Difference (IV, Random, 95% CI)	4.40 [‐10.76, 19.56]
9.1.2 3 months	1	30	Mean Difference (IV, Random, 95% CI)	‐2.10 [‐16.78, 12.58]
9.1.3 6 months	1	30	Mean Difference (IV, Random, 95% CI)	0.50 [‐15.21, 16.21]
9.2 Treatment success Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

9.2.1 6 weeks	1	30	Risk Ratio (M‐H, Random, 95% CI)	1.71 [0.33, 8.83]
9.2.2 3 months	1	30	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.37, 2.45]
9.2.3 6 months	1	30	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.50, 1.25]

Comparison 9. Autologous blood versus polidocanol injection

Navigate to table in Review

Comparison 10. Sensitivity analysis (mean pain and function at 3 months)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 Pain at 3 months (low vs high or unclear risk of selection bias) Show forest plot	8	523	Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.60, 0.29]

10.1.1 Adequate allocation concealment	4	166	Mean Difference (IV, Random, 95% CI)	0.40 [‐0.27, 1.08]
10.1.2 Unclear or inadequate	4	357	Mean Difference (IV, Random, 95% CI)	‐0.53 [‐1.08, 0.02]
10.2 Function at 3 months (low vs unclear or high selection bias) Show forest plot	8	502	Mean Difference (IV, Random, 95% CI)	‐1.86 [‐4.97, 1.25]

10.2.1 Adequate	5	235	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐4.80, 4.78]
10.2.2 Unclear or inadequate	3	267	Mean Difference (IV, Random, 95% CI)	‐2.93 [‐7.51, 1.65]
10.3 Pain at 3 months (adequate vs inadequate participant blinding) Show forest plot	8	523	Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.60, 0.29]

10.3.1 Adequate	7	498	Mean Difference (IV, Random, 95% CI)	0.00 [‐0.47, 0.47]
10.3.2 Unclear or inadequate	1	25	Mean Difference (IV, Random, 95% CI)	‐0.90 [‐1.91, 0.11]
10.4 Function at 3 months (adequate vs inadequate participant blinding) Show forest plot	8	502	Mean Difference (IV, Random, 95% CI)	‐1.86 [‐4.97, 1.25]

10.4.1 Adequate	6	437	Mean Difference (IV, Random, 95% CI)	‐0.23 [‐3.74, 3.29]
10.4.2 Unclear or inadequate	2	65	Mean Difference (IV, Random, 95% CI)	‐5.84 [‐10.98, ‐0.70]

Comparison 10. Sensitivity analysis (mean pain and function at 3 months)

Navigate to table in Review

Comparison 11. Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
11.1 Mean pain Show forest plot	6	485	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.56, 0.26]

11.1.1 Leukocyte rich	3	292	Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.71, 0.30]
11.1.2 Leukocyte poor	4	193	Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.80, 0.66]
11.2 Function Show forest plot	6	404	Mean Difference (IV, Random, 95% CI)	‐1.90 [‐5.61, 1.82]

11.2.1 Leukocyte rich	3	272	Mean Difference (IV, Random, 95% CI)	‐2.34 [‐6.91, 2.23]
11.2.2 Leukocyte poor	3	132	Mean Difference (IV, Random, 95% CI)	‐0.09 [‐8.36, 8.18]
11.3 Treatment success Show forest plot	4	382	Risk Ratio (M‐H, Random, 95% CI)	0.90 [0.65, 1.24]

11.3.1 Leukocyte rich	2	275	Risk Ratio (M‐H, Random, 95% CI)	1.03 [0.67, 1.59]
11.3.2 Leucocyte poor	2	107	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.53, 1.06]
11.4 Adverse events Show forest plot	5	425	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.76, 1.72]

11.4.1 Leucocyte rich	2	270	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.71, 1.84]
11.4.2 Leucocyte poor	3	155	Risk Ratio (M‐H, Random, 95% CI)	1.15 [0.53, 2.51]

Comparison 11. Subgroup leukocyte‐rich vs leukocyte‐poor PRP at 3 months

Navigate to table in Review

Comparison 12. Subgroup PRP versus autologous blood at 3 months

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
12.1 Mean pain at 3 months Show forest plot	8		Mean Difference (IV, Random, 95% CI)	Subtotals only

12.1.1 PRP vs placebo	7	534	Mean Difference (IV, Random, 95% CI)	‐0.19 [‐0.63, 0.25]
12.1.2 AB vs placebo	2	98	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐1.40, 1.15]
12.2 Mean function at 3 months Show forest plot	8	581	Mean Difference (IV, Random, 95% CI)	‐1.89 [‐4.60, 0.83]

12.2.1 PRP vs placebo	7	483	Mean Difference (IV, Random, 95% CI)	‐2.24 [‐5.30, 0.82]
12.2.2 AB vs placebo	2	98	Mean Difference (IV, Random, 95% CI)	0.50 [‐6.56, 7.55]
12.3 Withdrawals due to adverse events Show forest plot	7	499	Risk Ratio (M‐H, Random, 95% CI)	0.32 [0.03, 2.92]

12.3.1 PRP vs placebo	6	480	Risk Ratio (M‐H, Random, 95% CI)	0.32 [0.03, 2.92]
12.3.2 AB vs placebo	1	19	Risk Ratio (M‐H, Random, 95% CI)	Not estimable

Comparison 12. Subgroup PRP versus autologous blood at 3 months

Navigate to table in Review

Cochrane Review language

Website language

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

PICOs

PICOs

Population

Intervention

Comparison

Outcome

Plain language summary

Autologous blood or PRP injection for lateral elbow pain

Visual summary

Authors' conclusions

Implications for practice

Implications for research

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Major outcomes

Minor outcomes

Timing of outcome assessment

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Interpreting results and reaching conclusions

Results

Description of studies

Results of the search

Included studies

Study design and setting

Participant characteristics

Interventions

Outcomes

Major outcomes

Minor outcomes

Excluded studies

Risk of bias in included studies

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Autologous blood or PRP injection versus placebo

Benefits

Participant‐reported pain relief (≥ 30% or ≥ 50%)

Mean pain

Function

Treatment success

Health‐related quality of life