Serra 2015b.

Methods	Control arm of RCT (doxycycline vs control) Setting: single clinical centre Country: Italy Duration of follow‐up: 5 months Treatments: all participants had basic treatment, including vein surgery if appropriate and compression stockings. Randomised to receive a 3‐month course of doxycycline or standard care
Participants	32 participants with VLUs in control group Mean age: 51.3 (SD 7.5) years, range 46‐57 years Sex (M:F): 11:21 Stage of ulcer: class 6 of CEAP (inclusion criteria) Ulcer duration: ≥ 6 weeks (inclusion criterion) Ulcer size at baseline: median (range): 11.7 (3.1‐17.5) cm² Wound infection: no infection within 6 weeks (inclusion criterion) Number of wounds: unclear Inclusion criteria: aged 20‐70 years, VLU (class 6 CEAP), duration at least weeks, ABPI > 0.9 Exclusion criteria: arterial disease, infection within 6 weeks, connective tissue disorder, blood disorder, cancer, gastroenteritis, allergy to tetracyclines
Prognostic factors	MMP‐9 (continuous data). Measurement method ELISA: kit from Amersham Pharmacia Biotech; same approach as Serra 2013 Time of measurement: baseline, 4 weeks, 3 months, 5 months Wound fluid sampling method: not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Selection bias	Unclear risk	Moderate: insufficient information to assess this domain. RCT, so may be less representative
Attrition bias	Low risk	No missing data
Prognostic factor measurement	Unclear risk	Moderate: unclear if ELISA is adequate method and unclear if active or latent protease
Outcome measurement	High risk	Not adjusted for baseline; unclear if blinded
Adjustment factors	High risk	None of the key adjustment factors taken into account in the design or analysis
Analysis and reporting	Unclear risk	Moderate: number of participants healing not reported, so SMD could not be applied. Treatment unlikely to interact with protease levels
Overall risk of bias	High risk