Types of studies

We considered only randomised controlled trials (RCTs) to be eligible for inclusion.

Types of participants

• Patients having groin dissection for metastatic malignant disease, including ‐ but not limited to ‐ malignant melanoma, squamous cell carcinoma of the skin, penis, vulva or anus. The groin dissection could be therapeutic, prophylactic or completion (see Description of the condition).

• People over 18 years of age; no restriction by gender, no restriction on country of origin.

• The setting for this surgery was hospital inpatient care; adverse events that occurred in the outpatient setting were to be assessed if recorded by the studies.

We excluded trials in which participants underwent more extensive surgery, where dissection included higher lymph node groups including iliac, pelvic and para‐aortic nodes. We deemed Cloquet’s node (the most superior inguinal node in the proximal groin) to be the superior limit of dissection for eligible studies.

Types of interventions

We decided to include studies reporting the following comparisons (all references below to the use of a surgical drain(s) allude to those inserted into the groin following inguinal lymph node dissection):

• wound drainage compared with no wound drainage;

• one type of wound drain compared with another type of wound drain;

• different timing of drain removal: to include removal according to fixed‐time and fixed‐volume protocols.

Types of outcome measures

Electronic searches

We searched the following electronic databases to identify reports of relevant RCTs:

• The Cochrane Wounds Group Specialised Register (4 September 2014);

• The Cochrane Central Register of Controlled Trials (CENTRAL ) (The Cochrane Library 2014, Issue 8);

• Ovid MEDLINE (1946 to October Week 5 2013);

• Ovid MEDLINE (In‐Process & Other Non‐Indexed Citations 6 November 2013);

• Ovid Embase (1974 to 2013 November 06)

• EBSCO CINAHL (1982 to 07 November 2013).

We used the following search strategy in the Cochrane Central Register of Controlled Trials (CENTRAL):

#1 MeSH descriptor: [Lymph Node Excision] explode all trees 1033
#2 MeSH descriptor: [Lymph Nodes] explode all trees and with qualifiers: [Surgery ‐ SU] 76
#3 ("inguinal lymph node dissection" or "inguinal lymphadenectomy" or (groin next dissect*) or (groin next surg*)):ti,ab,kw 38
#4 {or #1‐#3}
#5 MeSH descriptor: [Drainage] explode all trees 1986
#6 MeSH descriptor: [Suction] explode all trees 733
#7 drain*:ti,ab,kw 4040
#8 #5 or #6 or #7 4641
#9 #4 and #8 133

The search strategies for Ovid MEDLINE, Ovid EMBASE and EBSCO CINAHL can be found in Appendix 1, Appendix 2 and Appendix 3 respectively. We adapted this strategy to search Ovid MEDLINE, Ovid EMBASE and EBSCO CINAHL. 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 developed by the UK Cochrane Centre (Lefebvre 2011). We combined the CINAHL searches with the trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN 2011). We did not restrict trials with respect to language, date of publication or trial setting.

We searched the following clinical trials registries:

• ClinicalTrials.gov (http://www.clinicaltrials.gov/) (13 April 2014);

• WHO International Clinical Trials Registry Platform (http://apps.who.int/trialsearch/Default.aspx) (13 April 2014);

• EU Clinical Trials Register (https://www.clinicaltrialsregister.eu/) (13 April 2014).

Searching other resources

Two review authors (DT, HS) checked the reference lists of included trials independently for additional eligible trials.

Selection of studies

Two review authors (DT, HS) independently determined the eligibility of each trial identified by analysing the titles and abstracts of all citations found through the search strategy previously described. We retrieved a full‐text copy of each citation that reported a potentially eligible trial, and for records where eligibility could not be determined by the title and abstract alone. Working independently, the two review authors applied the eligibility criteria; any discrepancies were resolved by consensus discussion with the third review author (DF). Where necessary, we sought additional information from the principal investigator of the trial concerned. We justified any exclusion of a potentially eligible trial from the review in the final report.

Data extraction and management

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, the plan is that two review authors (DT, HS) independently will extract data for each included trial using a pre‐designed data extraction form to extract the following data:

• details of the trial (first author, year of publication, journal, publication status, period and country of trial, sources of funding, trial design, sample size);

• patient characteristics (age, sex, type of disease, stage of disease, type of surgery and prior treatment status);

• quality of the trial;

• details of the surgery (PLND, TLND or CLND, number of nodes removed);

• details of the intervention;

• clinical variables related to patient well‐being;

• duration of follow‐up; and

• outcomes.

The third review author (DF) will resolve any discrepancies regarding data extraction, in order to reach consensus. In the presence of multiple reports on n included trial, we will assemble the most complete data‐set feasible for the trial.

Assessment of risk of bias in included studies

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, two review authors (DT, HS) independently will assess each trial for risk of bias using the Cochrane Collaboration's 'Risk of bias' tool (Higgins 2011). Selection bias, performance bias, detection bias, attrition bias and reporting bias will be considered for each individual trial. Where there are unclear risks of bias due to inadequate descriptions within the trial report, we will attempt to contact the trial authors in order to clarify the risk. All efforts made to obtain additional information will be reported in the review. Any discrepancies will be resolved by consensus discussion with the third review author (DF).

Measures of treatment effect

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, we will analyse dichotomous variables (wound complications, wound infection, drain re‐insertion, lymphoedema) as risk ratios (RR) with 95% confidence intervals (CI). We will analyse continuous variables (length of stay, volume of fluid drained, number of aspirations, total aspirate volume, total volume of drainage, patient‐reported outcome measures) as mean differences. Where different scales are used to assess continuous outcomes, we will analyse using standardised mean difference (SMD).

Unit of analysis issues

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, we will use the clinically relevant time points reported in each trial to calculate complication rates. Wound infection, as defined by the CDC criteria, has to occur within 30 days of the procedure, therefore this time point will be used as a cut‐off for this outcome measure (see above). Where cluster‐randomised trial designs are encountered, including, for example, randomisation by surgeon, or by operating list, we will analyse on the basis of allocation, using summary values for each cluster.

If bilateral groin dissections are performed with separate randomisation of each groin, we plan to analyse the results on a 'per groin' rather than a 'per participant' basis, as the outcomes of the two sides are likely to be independent, except for patient‐reported outcome measures, which will not be analysed in these participants, as the effects of the two interventions are not separable. We also plan to conduct a sensitivity analysis on a 'per participant' basis to determine if there is an effect of bilateral groin dissections in the same patient.

Dealing with missing data

We did not identify any trials that could be included in the review. However, in future, if the results of an RCT are published, but information on the outcome of interest is not reported, we will contact the trial authors for the missing information. Where continuous data are not presented as mean and standard deviations we contact the trial authors for the information in this format. All efforts made to obtain additional information will be reported in the completed review. Where possible, we intend to analyse by intention‐to‐treat (Hollis 1999). If participants are allocated to one intervention, but after randomisation underwent a different intervention, they will be analysed according to their randomisation allocation. If the results for dichotomous variables are not reported in some participants, we will base our analysis on both a worst possible outcome (for example, wound infection occurred in all non‐reported cases), and a best possible outcome (for example, wound infection did not occur in all non‐reported cases). Where participants are excluded from the analysis without good cause, we will conduct a sensitivity analysis to determine any effect of attrition bias.

Assessment of heterogeneity

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, we will explore heterogeneity using the Chi² test with significance set at P value 0.10, and measure the quantity of heterogeneity using the I² statistic (Higgins 2002).

Thresholds for the interpretation of the I² statistic can be misleading. A rough guide to the interpretation used is as follows:

• 0% to 40%: might not be important;

• 30% to 60%: may represent moderate heterogeneity;

• 50% to 90%: may represent substantial heterogeneity;

• 75% to 100%: considerable heterogeneity.

When interpreting the I² statistic, we will take factors such as clinical and methodological heterogeneity, along with whether the heterogeneity is in the magnitude of effect or in the direction of effect, into account, particularly where ranges overlap.

Assessment of reporting biases

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, we will compare the reported outcomes with those stated in the methods of the trials, and also those listed in the protocols in clinical trials registries as both primary and secondary outcomes. Where sufficient trials are identified (a minimum of 10), we will assess for publication bias by creating a funnel plot using software within Review Manager 5.3, using visual inspection to detect publication bias.

Data synthesis

We did not identify any trials that could be included in the review. However, in future, if suitable trials are published, we will present a narrative overview of any included trials. Where appropriate, we will present meta‐analyses of outcome data using RevMan 5.3 (RevMan 2014). The decision to pool data in a meta‐analysis will depend on the availability of outcome data and the assessment of between‐trial heterogeneity. Where trials measure the same underlying effect we plan to use a fixed‐effects model, and where this is not the case we plan to use a random‐effects model ‐ if sufficient studies are identified to inform the distribution of effects (i.e. more than five trials). However, we do not intend to pool data where heterogeneity is very high (I² values of 75% or above).

For dichotomous outcomes we plan to present the summary estimate as a risk ratio (RR) with 95% confidence intervals (CI). Where continuous outcomes are measured in the same way across trials, we will present a mean difference (MD) with 95% CI. Where trials measure the same outcome using different methods we will present a standardised mean difference (SMD).

Subgroup analysis and investigation of heterogeneity

If suitable trials are published in future, we will consider the following variables for subgroup analyses.

• Effects of CLND or PLND versus TLND. These subgroups may behave differently, as those patients undergoing a TLND have macroscopic cancer in the lymph node basin, which, theoretically, may require a larger volume of tissue to be removed and create a larger dead space, thereby increasing the potential drainage after inguinal lymph node dissection. By contrast, some authors have suggested that, when compared to TLND, CLND has an increased risk of wound infection owing to two procedures having been performed in the same anatomical location in rapid succession (de Vries 2006).

• Type of cancer, for example melanoma versus vulval cancer, versus anal cancer, versus penile cancer.

• Type of drain used, for example closed suction drainage versus gravity drainage.

Ability to conduct such analyses will also depend on whether the required information is recorded in the trial publications. If data are not included, we plan to contact trial authors to obtain the data.

Sensitivity analysis

If suitable trials are published in future, we plan to undertake a sensitivity analysis that includes, and then excludes, trials at high risk of bias, repeating this analysis to include trials at unclear risk of bias as well. We acknowledge that there is no accepted definition of what constitutes a trial at high risk of bias, therefore, we intended to assess the severity of any biases for each outcome. We intended to assess risk of bias using the Cochrane Collaboration 'Risk of bias' tool (Higgins 2011).