Topically Applied Treatments for External Genital Warts in Nonimmunocompromised Patients

A Systematic Review and Network Meta-analysis

J.M. Jung; C.J. Jung; W.J. Lee; C.H. Won; M.W. Lee; J.H. Choi; S.E. Chang


The British Journal of Dermatology. 2020;183(1):24-36. 

In This Article


We performed this study following the PRISMA guidelines. The study protocol was registered on PROSPERO (CRD42019131710).

Literature Search Strategy

An electronic search of five databases (MEDLINE via PubMed, Embase, Cochrane Central, Web of Science and Scopus) was performed in April 2019 according to the search strategy described in Table S1 (see Supporting Information). We used no filters for language or publication period. Reference lists of selected articles were also examined. Two independent reviewers screened the retrieved reports for eligibility through title, abstract and full-text review. Discrepancies were adjudicated through a third reviewer.

Study Eligibility Criteria

We included all randomized controlled trials (RCTs) that evaluated the safety and efficacy of any topically applied agents for treating external genital warts in nonimmunocompromised patients. No restrictions were made regarding patient sex or race. We excluded studies with intraindividual designs; physically destructive therapies; systemic therapies; patients with genital warts on intravaginal, intra-anal or intraurethral areas; and no outcomes defined in our inclusion criteria.

Data Extraction and Outcomes

Two independent reviewers extracted data, which were checked by a third reviewer. Discrepancies in this process were resolved through expert discussions. The extracted data included the first author's last name; publication year; country of origin; inclusion criteria; exclusion criteria; sample-size determination; baseline demographic data; number of participants randomized; final number of participants assessable; treatment schedules, frequencies and durations; complete clearance; recurrence; adverse events; patients with severe adverse events (SAEs); and patients who were lost to follow-up because of treatment-related side-effects. An SAE was defined as an adverse event measured at the most severe grade in the scale of each study, regardless of the kinds of adverse events. Complete clearance was defined as complete lesion clearance at 8 ± 4 weeks after treatment. Recurrence was defined as presence of any wart at 12 ± 4 weeks after complete clearance. Adverse events included local skin reactions and any treatment-related side-effects. SAE data related to the treatment and the exact causes of patient withdrawals were also collected. Intention-to-treat analysis was performed whenever possible, except for recurrence, for which a per protocol analysis was used.

Risk-of-bias Assessment

We used the Cochrane Collaboration's tool for assessing risk of bias in randomized trials,[14] which comprised seven specific domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other sources of bias. Two authors independently assessed the risk of bias of the included studies. Discrepancies were settled through discussions.

Statistical Analysis

The NMA was performed using the R package netmeta, using a frequentist approach. We employed a random effects model assuming there can be inherent heterogeneity between trials in this study. We presented all outcomes as odds ratios (ORs) with 95% confidence intervals (CIs). The P-score based on frequentist point estimates and standard errors was used to rank different topical treatments.[15] A higher score meant better treatment. Heterogeneity between studies was assessed using Cochran Q-statistics and the I 2-measure from the netmeta statistical package. I 2-values of 25–49% are considered to indicate low, 50–74% moderate and ≥ 75% high levels of heterogeneity. The function was used to assess the global inconsistency in each model. A P-value < 0·05 was considered suggestive of significant inconsistency. The net split function was used to evaluate the consistency between direct and indirect evidence. We also employed net heat plot, a graphical tool for locating inconsistency in NMAs. The stronger the intensity of the colour, the greater the inconsistency between specific direct evidence in the whole network. Publication bias was assessed using comparison-adjusted funnel plots. Subgroup analyses or pairwise meta-analyses were also performed if applicable to explore the data further. P-values < 0·05 indicated a significant difference.