Types of studies

Randomised controlled trials (RCTs) of sulodexide were eligible for inclusion.

Types of participants

We included studies involving people of any age, gender, and in any care setting, described as having a venous leg ulcer.

Types of interventions

We included RCTs which evaluated oral or parenteral administration of sulodexide (any strength and over any period of administration) compared with placebo or any other drug therapy (such as pentoxifylline, flavonoids, aspirin), with or without compression therapy. Any co‐interventions (e.g. other drug therapy or compression therapy) had to be used in a standardised way across all arms in a trial.

Types of outcome measures

Electronic searches

We searched the following electronic databases to identify relevant RCTs:

• The Cochrane Wounds Specialised Register (searched 01 July 2015);

• The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, 2015, Issue 6);

• Ovid MEDLINE (1946 to 01 July 2015);

• Ovid MEDLINE (In‐Process & Other Non‐Indexed Citations) (searched 01 July 2015);

• Ovid EMBASE (1974 to 01 July 2015);

• EBSCO CINAHL (1982 to 01 July 2015);

• Chinese Biomedical Literature Database (CBM) (1978 to 13 August 2015);

• China National Knowledge Infrastructure Database (CNKI) (1979 to 13 August 2015);

• Wan Fang Database (1986 to 13 August 2015);

• VIP Database (1989 to 13 August 2015).

The search strategies used for CENTRAL, MEDLINE, EMBASE and CINAHL can be found in Appendix 1, Appendix 2, Appendix 3, and Appendix 4. The search strategies was modified and translated appropriately for each Chinese database search (Appendix 5; Appendix 6; Appendix 7; Appendix 8). We combined the Ovid MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity‐ and precision‐maximising version (2008 revision) (Lefebvre 2011). We combined the EMBASE search with the Ovid EMBASE filter terms developed by the UK Cochrane Centre (Lefebvre 2011). We combined the CINAHL search with the trial filter terms developed by the Scottish Intercollegiate Guidelines Network (SIGN 2015). There were no restrictions based on date of publication, language or study setting.

We also searched the following clinical studies registries for any ongoing studies:

• ClinicalTrials.gov (http://clinicaltrials.gov/);

• International Standard Randomised Controlled Trial Number Register (ISRCTN) Registry (http://www.isrctn.com/);

• WHO International Clinical Trials Registry Platform (ICTRP) (http://www.who.int/ictrp/en/),

• Chinese Clinical Trial Registry (ChiCTR) (http://www.chictr.org.cn/searchprojen.aspx).

Searching other resources

We searched the references listed in relevant trial reports and reviews to identify any further relevant RCTs.

Selection of studies

We used EndNote X7 software (EndNote 2014) to merge retrieved reports of each database and to remove duplicate records of the same study. Two review authors (BW, JL) assessed titles and abstracts of studies independently to exclude obviously irrelevant reports. We retrieved full‐text copies of all potentially eligible reports, and reviewed them in the light of the inclusion criteria. Two review authors (BW, JL) made final decisions on the inclusions by cross‐checking the results, and a third review author (MY) was consulted when there were any disagreements. Where we identified multiple reports of the same trial, we extracted the maximum amount of data from the multiple reports and identified one report as the primary reference. We set out a study flow diagram as recommended by the PRISMA statement to illustrate the process of screening and selecting studies for inclusion in the review (Figure 1) (Liberati 2009).

Figure 1

---

---

PRISMA flow diagram of literature screening

Data extraction and management

We extracted study data based on the data extraction sheet provided by Cochrane Wounds, which included the following information: general data (authors, publication year and contact information), baseline data (participants' number, age and gender), risk of bias assessment information (randomisation, allocation concealment, blinding and incomplete outcome data), interventions, duration of follow up, outcome measures and results. We planned to contact the study authors for more information, if necessary.

Independently, two review authors (BW, JL) extracted and managed data from all included trials, and discussed any disagreements. A third review author (MY) acted as arbiter when BW and JL failed to reach an agreement.

Assessment of risk of bias in included studies

We assessed risk of bias of included studies by the methods recommended in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Two review authors (BW, MY) independently evaluated the following seven items for each study: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other potential sources of bias. We judged risk of bias for each item as 'Low risk', 'High risk' and 'Unclear risk' following the assessment criteria recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Appendix 9). Finally, we presented our assessment of risk of bias in a 'Risk of bias' graph (Figure 2) and 'Risk of bias' summary (Figure 3).

Figure 2

---

---

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

---

---

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

Measures of treatment effect

For dichotomous outcomes, we presented the risk ratio (RR) with 95% confidence interval (CI) for each individual study. For continuous outcomes, we used the mean difference (MD) with 95% CI for each individual study. For time‐to‐event data, we planned to report estimates of time to complete ulcer healing and to plot hazard ratio (HR) estimates with 95% CI when available from study reports (Deeks 2011), however, HR was not reported in Niglio 1991 or Scondotto 1999 and the mean time to complete ulcer healing was presented in Scondotto 1999.

Unit of analysis issues

We considered individual participants as the unit of analysis. We intended to present a narrative summary if the unit of randomisation or analysis of included study was ulcer or limb. We planned to re‐analyse any cluster‐randomised studies identified by calculating the effective sample sizes with the intra‐cluster coefficient (ICC) estimated externally from similar studies (Deeks 2011), however, we did not include any cluster‐randomised studies in this review. The outcome of 'proportion of ulcers completely healed' was reported at different time points; we firstly performed an overall analysis by pooling all the included studies, then conducted subgroup analyses at 30 days', 60 days' and 90 days' follow up.

Dealing with missing data

We attempted to contact the original study authors when essential data were missing. We tried to contact authors of the included RCT published as an abstract only (Niglio 1991) by email, unfortunately, we got no replies. We assumed the missing participants had both positive and negative outcomes (e.g. missing participants assumed for ulcers healed and ulcers not healed), and undertook an analysis based on these assumptions (best case/worst case scenario) respectively, performing a sensitivity analysis.

Assessment of heterogeneity

Firstly, if clinical diversity existed between the trials (e.g. ulcers of different severity, different types of interventions, or different durations of follow up), we planned not to combine data from these trials and to present a narrative summary. Secondly, for clinically homogeneous (similar) studies, we intended to perform a Chi² test, with P values less than 0.1 indicating significant statistical heterogeneity. In order to quantify heterogeneity not due to chance, we used the I² statistic (Higgins 2003). A rough guide for the interpretation of I² was as follows: 0% to 40% represented heterogeneity that might not be important; 30% to 60% might represent moderate heterogeneity; 50% to 90% might represent substantial heterogeneity; 75% to 100% represented considerable heterogeneity (Deeks 2011).

Assessment of reporting biases

We performed a comprehensive search for eligible RCTs to minimise reporting bias. We planned to use funnel plots to assess publication bias (Sterne 2011) when more than 10 studies were included; however, there were insufficient studies to conduct this analysis. To evaluate selective reporting of outcomes, we planned to compare the study protocols with the final study reports, however, none of the protocols was available. Then we compared the 'Methods' section of the published studies with the 'Results' section to identify any outcomes that were measured but not reported.

Data synthesis

We analysed the data using Review Manager (RevMan) 5.3 software provided by the Cochrane Collaboration (RevMan 2014). We used a fixed‐effect model for meta‐analysis in the absence of clinical, methodological and statistical heterogeneity. If the I² statistic was greater than zero, we also applied a random‐effects model to see whether the conclusions differed, and any difference was noted. If pooling was not possible or appropriate, we intended to present a narrative summary (Deeks 2011).

Subgroup analysis and investigation of heterogeneity

If clinical heterogeneity was investigated and there were sufficient RCTs included, we planned to conduct subgroup analyses as follows.

• Different types of interventions, determined by the type of control group (placebo or other therapies) or combination of compression therapy (with or without compression).

• Different durations of follow up.

There were sufficient RCTs included to allow us to undertake the second planned subgroup analysis for the primary outcome of proportion of ulcers completely healed.

Sensitivity analysis

We intended to perform a sensitivity analysis for the primary outcomes to investigate the robustness of our findings. We planned to conduct a sensitivity analysis by comparing meta‐analysis results of:

• RCTs with low risk of bias (adequate sequence generation, adequate allocation concealment and an adequate method of outcome assessor blinding) compared with all included RCTs.

• the assumption that missing participants had a positive outcome versus a negative outcome.

We were able to undertake the second planned sensitivity analysis for one of the studies comparing sulodexide to placebo (Coccheri 2002).