Watson 2011.

Methods	Randomised controlled trial in a variety of settings in UK and Ireland.
Participants	337 people with hard to heal venous leg ulcers (defined as more than 6 months' duration or area greater than 5cm2 or both); ABPI ≥0.8. Variety of settings in UK and Ireland (community nursing services, hospital outpatients clinics). US group: n = 168; Standard care group: n = 169
Interventions	US group: low dose (0.5 W/cm2) ultrasound at 1 MHz with a pulsed pattern of 1:4 applied to the periulcer skin (using a water‐based contact gel) once a week for up to 12 weeks plus standard care. US was applied for a period of 5‐10 minutes per treatment to the reference ulcer; the actual time being determined by a protocol based on ulcer area. Standard care group: simple low adherent dressing and high compression (4‐layer bandaging), reduced compression or no compression according to the clinician's assessment of the level of pressure tolerated by the participant
Outcomes	Time to healing of the reference ulcer Cost effectiveness Proportion of participants with healed ulcers at 3, 6, 12 months Percentage and absolute change in ulcer size HRQoL and adverse events
Notes	Maximal duration of follow‐up was of 12 months.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participants were randomised equally between the two trial arms: ultrasound plus standard care and standard care alone. Randomisation was carried out using varying block sizes of four and six participants ... The computerised randomisation system was checked periodically during the trial following standard operating procedures."
Allocation concealment (selection bias)	Low risk	Quote: "To maintain allocation concealment the generation of the randomisation sequence and subsequent treatment allocation were performed by an independent, secure, remote, telephone randomisation service (York Trials Unit)."
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Owing to the nature of the intervention, it was not possible to conceal the treatment allocation from either the patient or the nurse" This lack of blinding leaves the study susceptible to performance bias.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote for 12‐week outcome: "The reference ulcer was the largest ulcer on either leg (as assessed at the time of trial entry). The date of healing was recorded by the research nurses on the Ulcer Healed Form and the photographs of the reference ulcer were assessed independently by two people blind to treatment group. Any disagreements were resolved by discussion or referral to a third blinded assessor. The primary outcome was calculated using the date of healing as decided by the blind assessors. If the blinded assessors did not agree on a healing date, then the date as recorded on the Ulcer Healed Form was used." This blinded, remote adjudication of healing reduced the risk of detection bias. It is difficult to accurately judge the risk of bias in this scenario because an unmasked research nurse took a photograph. However, the blinded adjudication gives some reassurance that the risk of detection bias is low. 12‐month outcome: unclear. Quote: "The number of leg ulcers that had completely healed by 12 months was based on nurse‐reported data and not on blinded photographs ..."
Incomplete outcome data (attrition bias) Ulcer healing	Low risk	Quote: "Time to healing was derived as the number of days between randomisation and the first date that healing was confirmed. Patients who withdrew unhealed from the trial or died prior to healing were treated as censored in the analysis. Their time to censoring was derived using the date of trial exit, the date of their last ulcer assessment or the date of trial closure." Participants who completed the full 12‐month follow‐up without their reference ulcer healing were treated as censored and their time to censoring was calculated as 12 months (365 days). Quote: "All randomised participants were included in the analysis and numbers of full withdrawals were low (only 10 patients ceased contributing data on the primary endpoint)." Analysis was by intention to treat.
Selective reporting (reporting bias)	Low risk	A full protocol was available and the published trial followed the protocol. Amendments to the original protocol were detailed and justified.
Other bias	Low risk	No other serious bias.