Radwan 2008.

Methods	Randomised controlled trial. Pre‐specified sample size calculation reported. Overall validity of results: high risk of bias.
Participants	56 participants. Surgery group: n=27, 18 males, 16 with right side involved, 17 with dominant side involved, mean age 40 years (range 22 to 59 years), duration of symptoms 18 months (range 6 to 60 months). Extracorporal shock wave therapy (ESWT) group: n=29, 15 males, 18 with right elbow involved, 19 with dominant side involved, mean age 40 years (range 23 to 60 years), mean duration of symptoms 17 months (range 6 to 48 months). Inclusion criteria: lateral epicondylitis diagnosed by pain induced by 2 or more diagnostic tests (palpation of the lateral epicondyle; Thomsen test‐ resisted wrist extension; chair test‐ patient attempts to lift a 3.5 kg chair with shoulder flexed to 60 degrees and elbow extended); and failure of at least 6 months of conservative treatment including NSAIDs, corticosteroid injections, physical therapy, exercise program and elbow brace. Exclusion criteria: younger than 18 years; pregnant; local infection; malignancy; elbow arthritis; generalised polyarthritis; ipsilateral shoulder dysfunction; neurological abnormalities; radial nerve entrapment; cardiac arrhythmia or pacemaker; received corticosteroid injection in the previous 6 weeks.
Interventions	Group 1, 'Operative group' (27 participants): percutaneous tenotomy, release of the common extensor origin performed through a transverse incision of 1‐2cm just distal to the lateral epicondyle, under general anaesthesia and tourniquet control; the common extensor origin was divided leaving a 1cm gap. The incision was left open to drain; bulky dressing and posterior plaster splint applied for one week. Group 2, 'ESWT Group' (29 participants): ESWT given under conscious sedation anaesthesia, repetitive high energy shock waves delivered by the electro hydraulic method; each patient received 100 graded shocks (14‐18 kV; 0.12‐0.22 J/mm2) initially to assess the effectiveness of anaesthesia, then 1400 shocks at 18 kV (0.22 mJ/m2) for a total of 1500 shocks at 4 shocks/s, and total energy delivered of 324.25J.
Outcomes	1) Pain VAS 0 (no pain) to 100 (maximum pain) at baseline, 3, 6, 12 and 52 weeks: at night; at rest; on pressure; on Thomsen test; and chair test. 2) Grip strength compared with normal side, on 4 point categorical scale at baseline, 3, 6, 12 and 52 weeks: 1 (equal strength on both sides), 2 (up to 25% reduction), 3 (up to 50% reduction), 4 (up to 75% reduction). 3) Participant assessment of residual pain compared with baseline using Roles and Maudsley criteria at 52 weeks: Excellent ‐ no pain, full movement, full activity; Good ‐ occasional discomfort, full movement, full activity; Acceptable‐ some discomfort after prolonged activities; Poor ‐ pain limiting activity. 4) Proportion of participants with treatment success at 52 weeks: defined as 'excellent' or 'good score' based on Roles and Maudsley criteria (in 3, above).
Sources of funding	None reported
Notes	The trial authors reported median and interquartile ranges for continuous outcomes measured using 100mm VAS scale (night pain, rest pain, pain on pressure, Thomsen test, Chair test). For the purposes of data extraction into RevMan, we assumed the median approximates the mean, and calculated the standard deviation assuming the width of the interquartile range is approximately 1.35 standard deviations. We made an ad hoc decision to present outcomes in RevMan at the last time point only (12 months), due to lack of differences between treatments at all time points.
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	Quote: 'Patients were randomised into two groups' Comment: method used to generate the allocation sequence not reported
Allocation concealment?	Unclear risk	Quote: 'Closed envelope technique' Comment: insufficient information reported to determine whether envelopes were sequentially numbered and opaque
Blinding? All outcomes	Unclear risk	Comment: blinding not reported, but unlikely, as it is not possible to blind participants and study personnel to treatment. It is also unlikely that outcome assessors were blinded to treatment; but it is uncertain if this would bias one treatment over the other
Incomplete outcome data addressed? All outcomes	Low risk	Comment 6/56 participants (3 in each group) did not complete the 12 month follow‐up; these losses were unlikely to bias outcomes
Free of selective reporting?	High risk	Adverse events reported for ESWT group; not reported for surgery group
Free of other bias?	Low risk

ECRL: extensor carpi radialis longus 
 ECRB: extensor carpi radialis brevis 
 DASH score: American Academy of Orthopedic Surgeons Disability of Arm, Shoulder and Hand score 
 PIN: posterior interosseous nerve 
 ECRB: extensor carpi radialis brevis 
 NSAIDs: non‐steroidal anti‐inflammatory drugs