Assessment of clinical effectiveness

Elizabeth Thurgar; Samantha Barton; Charlotta Karner; Steven J Edwards

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Thurgar E, Barton S, Karner C, et al. Clinical effectiveness and cost-effectiveness of interventions for the treatment of anogenital warts: systematic review and economic evaluation. Southampton (UK): NIHR Journals Library; 2016 Mar. (Health Technology Assessment, No. 20.24.)

Cover of Clinical effectiveness and cost-effectiveness of interventions for the treatment of anogenital warts: systematic review and economic evaluation

Clinical effectiveness and cost-effectiveness of interventions for the treatment of anogenital warts: systematic review and economic evaluation.

Show details

Contents

< Prev Next >

Chapter 3Assessment of clinical effectiveness

Methods for reviewing effectiveness

Evidence on the clinical effectiveness of interventions to treat AGWs was identified by conducting a systematic review of the published research literature. The review was undertaken following the general principles published by the Centre for Reviews and Dissemination⁸⁴ and the Cochrane Collaboration.⁸⁵ The protocol for the systematic review is registered on PROSPERO database (registration number CRD42013005457).⁸⁶

Identification of studies

To identify relevant studies, multiple electronic databases were searched:

Ovid MEDLINE In-Process & Other Non-Indexed Citations and Ovid MEDLINE
Ovid EMBASE
The Cochrane Library [specifically Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials (CENTRAL), Database of Abstracts of Reviews of Effects and Health Technology Assessment (HTA) database]
Web of Science.

Search strategies were designed to include medical subject headings (MeSH) and text terms for AGWs, including ‘condyloma acuminata’ (the medical term for AGWs). To maximise the number of potentially relevant studies retrieved, no MeSH or text terms were included for interventions of interest. Based on the results of the scoping search and clinical expert advice, it was anticipated that few RCTs meeting the eligibility criteria would be identified, despite the number of studies retrieved. Therefore, searches were simultaneously carried out for prospective observational studies (matched control studies, case series and case–control studies).

Search filters designed to retrieve reports by study design were identified through the InterTASC Information Specialists’ Sub-Group search filter resource.⁸⁷ Filters developed and validated by the Scottish Intercollegiate Guidelines Network were used to identify RCTs in MEDLINE and EMBASE.⁸⁸ Filters devised by Clinical Evidence (a collection of systematic overviews covering various conditions) were chosen to retrieve potentially relevant observational studies from MEDLINE and EMBASE.⁸⁹ Search terms for AGWs were tailored to the database searched.

Bibliographies of previous overviews, guidelines and retrieved articles were manually reviewed for additional studies. Clinical trial registries (World Health Organization International Clinical Trials Registry Platform and ClinicalTrials.gov) were also searched to identify planned, ongoing and finalised clinical trials of interest. The website of the US FDA [see www.fda.gov/ (accessed 16 December 2015)] was also searched to identify unpublished data. In addition, clinical experts were contacted with a request for information on any additional studies of which they had knowledge.

No language restriction was applied to the searches. With the exception of Web of Science, electronic databases were searched from inception, with the initial search carried out on 30 August 2013. Search parameters for Web of Science were limited to a search period of 2000 to 1 September 2014, with study type restricted to article, meeting abstract, proceedings paper and corrections. Search results were uploaded into Reference Manager version 11.0 (Thomson ResearchSoft, San Francisco, CA, USA) and deduplicated. Update searches were carried out on 22 April and 1 September 2014. Full details of the search strategies are presented in Appendix 1.

Two researchers (SB and one of ET or CK) independently screened the titles and abstracts returned by the search strategy according to prespecified eligibility criteria (see Table 2). In cases in which consensus could not be achieved, the full texts of potentially relevant studies were ordered. During abstract appraisal, to facilitate discussion of whether or not sufficient evidence had been identified to restrict inclusion of study type to RCTs, potentially relevant studies were categorised as RCT, observational study or systematic review. Two reviewers (SB and CK) independently assessed full publications for inclusion, with studies classified as RCTs evaluated first. Discrepancies were resolved by discussion, with involvement of a third reviewer (SEd) if consensus could not be reached. After appraisal of full-text publications, the number of RCTs identified as eligible for inclusion in the review of clinical effectiveness and AEs led to the decision to limit reporting to RCTs.

Inclusion and exclusion criteria

Eligibility criteria for the review of clinical effectiveness were as specified in the decision problem (summarised in Table 2). The review included only RCTs, with systematic reviews and non-randomised studies excluded. The interventions of interest were topical treatments and ablative techniques, either alone or in combination. RCTs were included if the treatments were evaluated in a population with AGWs and compared with each other, placebo or no treatment. Studies were excluded if none of the outcomes of interest was reported.

Data abstraction

Because of the large number of RCTs identified, in the first instance two reviewers (SB and Victoria Wakefield) independently extracted data from only 10 studies onto a standardised data extraction form; a function of the initial 10 extractions was to pilot the suitability of the data extraction form. Subsequently, one reviewer (various) extracted data from the remaining studies onto a modified data extraction form, with validation of the data by a second reviewer (SB). Information extracted included details on study design and methodology, the baseline characteristics of the population and data on outcomes of interest, both clinical effectiveness outcomes and AEs. Discrepancies were resolved by discussion, with involvement of a third reviewer (SEd) when necessary. During data extraction, if the reviewer(s) identified areas with limited reporting (e.g. aspects of trial conduct) or discrepancies in reporting within the publication (e.g. in event rate), authors were contacted with a request for clarification. If a study was reported as a conference proceeding or an abstract only, study authors were contacted with a request for further details. Studies reporting data on an outcome of interest but for which insufficient methodological details were available to allow full critical appraisal of study quality, even after contact with authors, were included in sensitivity analyses (additional detail provided in Results). Data extraction forms for the included studies are provided in Appendix 2.

Critical appraisal strategy

Two reviewers independently assessed the quality of the clinical effectiveness studies. Discrepancies were resolved by discussion, with involvement of a third reviewer when necessary. Study quality was assessed according to recommendations of the Centre for Reviews and Dissemination⁸⁴ and the Cochrane Handbook for Systematic Reviews of Interventions.⁸⁵ Study quality was recorded using the Cochrane risk of bias tool⁸⁵^,⁹⁰ and was incorporated into the data extraction form (see Appendix 2).

Outcome-specific risk of bias was determined for the outcomes for which data were extracted.⁹⁰ The three bias assessment categories used were low, unclear and high. A study was deemed to be at low risk of bias when all key domains were associated with low risk of bias, at an unclear risk of bias when one or more key domains had an unclear risk of bias and at a high risk of bias when one or more key domains was thought to be at a high risk of bias.

Methods of data synthesis

Details of the clinical effectiveness results and quality assessment for each included study are presented in structured tables (see Appendix 2) and an overall assessment of study quality is provided as a narrative summary (see Quality assessment). The possible effects of study quality on the clinical effectiveness data and review findings are discussed where relevant.

Standard pair-wise meta-analysis was performed, where possible, to evaluate clinical effectiveness and was based on intention-to-treat analysis. Intention-to-treat analysis was defined as people being analysed in the treatment group to which they were allocated at randomisation, irrespective of whether they changed treatment, withdrew or were lost to follow-up. Dichotomous outcome data were meta-analysed using Mantel–Haenszel odds ratios (ORs) with 95% confidence intervals (CIs) and a fixed-effects model; meta-analysis with a random-effects model was carried out as a sensitivity analysis. To facilitate comparison and interpretation of estimates of effect across studies, when data from a single trial were available for a comparison of interest, and if appropriate, the trial data were analysed and presented as for meta-analysed data. Missing data were imputed and were analysed as a treatment failure for all outcomes (i.e. for complete clearance, people lost to follow-up were considered not to have achieved clearance and, for recurrence, people lost to follow-up were considered to have recurred).

Meta-analysis was carried out using Review Manager version 5.3 (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark). Inconsistency among studies included in the meta-analysis was assessed using the I² test and the level for statistically significant heterogeneity was set at p < 0.10. Levels of inconsistency were defined as follows: low level, I² of 0–25%; moderate level, I² of 26–50%; and high level, I² of > 50%. In the presence of statistically significant heterogeneity (p < 0.10), possible sources were investigated, including differences in study populations, methods or interventions. The low number of studies included in each meta-analysis precluded the evaluation of publication bias and/or small study effects.

Additionally, the comparative clinical effectiveness of interventions was investigated using MTCs. The methods used for MTC followed the guidance described in the National Institute for Health and Care Excellence Decision Support Unit’s Technical Support Documents for Evidence Synthesis.⁹¹^,⁹² MTCs were conducted using a Bayesian Markov chain Monte Carlo simulation in WinBUGS (version 1.4.3; MRC Biostatistics Unit, Cambridge, UK). The following were implemented for each analysis:

uniform priors (also called ‘uninformed’ or ‘flat’ priors) were used
all outcomes were considered independent
to ensure convergence on the posterior distribution:
- results for all clinical effectiveness outcomes analysed were based on 50,000 iterations after a ‘burn in’ of 150,000 iterations
- results for all safety outcomes analysed had a ‘burn in’ of 30,000 iterations, with results based on 100,000 iterations
the OR was used as the summary effect estimate for all outcomes
a weighted average of the baseline treatment was used in combination with the relative treatment effect to calculate probabilities for all treatments included in the analysis of that outcome
any results taken forward into the economic model used the posterior sampling to retain the correlation between parameter estimates caused by their joint estimation from a single data set.

When a random-effects model was deemed the best fit, the extent of the between-study heterogeneity was investigated by evaluating the posterior mean of tau-squared.

The potential limitations of the MTC, together with the associated influence on the generated estimates of effect, are discussed in the strengths and limitations of the report (see Chapter 5).

Sensitivity analyses were carried out for the outcomes of complete clearance (at the end of treatment and at later time points) and recurrence of AGWs in both standard meta-analysis and the MTC. Sensitivity analyses included studies or individual outcomes deemed to be at an overall high risk of bias, together with studies in people with comorbid HIV infection and a cluster of differentiation 4+ cell count of < 200 cells/mm³.

As a consequence of the limited reporting of baseline characteristics in the included studies (discussed in greater detail in Quantity and quality of research available), planned subgroup analyses were not carried out. Planned analyses included:

soft, moist, non-keratinised AGWs
dry, keratinised AGWs
number of AGWs [to be grouped as single, few (two–five) or multiple (six or more)]
site of AGWs
no previous treatment for AGWs (‘first attack’ patients)
recurrent AGWs (return of AGW after a complete response to treatment)
persistent AGWs (treatment is continued for > 6 months)
immune status (immunosuppressed vs. not immunosuppressed).

Results

Quantity and quality of research available

Searches of electronic databases retrieved 4231 records (post deduplication) that were of possible relevance to the review (Figure 2). Manual searching identified one additional reference, giving a total of 4232 records screened for inclusion in the review. Full publications for 155 references were ordered. Of these, 13 publications were unobtainable.⁹³^–¹⁰⁵ Five studies published in Chinese were identified for which translations could not be obtained within the time frame of the project.¹⁰⁶^–¹¹⁰ An evaluation of RCTs published in Chinese journals found that most studies described as randomised were not truly random.¹¹¹ Based on this report and translations obtained for this project for other retrieved studies published in Chinese, the authors of the project consider that most of the studies in Chinese are unlikely to be random and omission of these studies is unlikely to have influenced the assessment of clinical effectiveness. Of the 137 full articles evaluated, 70 publications describing 60 studies (full publications⁶⁰^–⁶⁸^,¹¹²^–¹⁶²) were relevant to the review. Citation details for conference abstracts related to full publications are provided only when additional information was available in the abstract. A list of publications screened but subsequently excluded (with reasons for exclusion) from the review is available in Appendix 3.

FIGURE 2

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Summaries of the studies included in the review are presented by treatments evaluated (Table 3) and by key characteristics of studies (Table 4).

TABLE 3

Interventions evaluated within the studies included in the review

TABLE 4

Summary of studies included in the systematic review of clinical effectiveness

Study characteristics

Population

Inclusion criteria varied considerably across identified studies, with some studies reporting clinical diagnosis of AGWs as the sole criterion for eligibility (see Table 4). The size of studies was also wide-ranging, with the largest study randomising 450 people¹⁵¹ and the smallest randomising 12 people.¹²⁰

Few identified studies reported comprehensive baseline characteristics for the enrolled study population. Key characteristics of AGWs that potentially influence treatment choice and effectiveness include size, number and location of lesions, together with wart type (keratinised vs. non-keratinised). Of those studies reporting baseline characteristics, most studies included a mixed population in terms of AGW characteristics. AGW morphology was rarely specified as an inclusion criterion and only one study⁶¹ was identified that restricted inclusion by AGW type, specifying that AGWs should be non-keratinised. Most studies evaluated the treatment of external AGWs. One study¹⁵⁵ was identified that evaluated treatment of intra-anal AGWs. Duration of AGWs and line of treatment also varied across studies. Some studies focused on people who were treatment naive or restricted inclusion to those who had not received treatment in a set time period before enrolment, typically the preceding 3 months. As well as varying across studies, duration of disease varied considerably within studies, for example ranging from 1 to 300 weeks in the study by Godley et al.¹⁵⁴

Several studies specified a minimum age for eligibility of 18 years and the reported mean age of people enrolled in studies ranged from 20 to 35 years. Although many studies included both genders, 18⁶¹^,⁶⁶^,⁶⁷^,¹¹⁵^,¹²³^,¹²⁷^,¹²⁹^,¹³³^,¹³⁶^,¹⁴⁰^,¹⁴²^,¹⁴³^,¹⁴⁵^,¹⁴⁷^,¹⁵⁰^,¹⁵⁴^,¹⁶⁰^,¹⁶² and eight⁶²^,¹²²^,¹²⁴^,¹²⁶^,¹³⁰^,¹³⁴^,¹³⁷^,¹⁶¹ studies focused on men and women respectively. Three studies evaluated treatments in people with HIV infection.¹¹⁹^,¹²⁰^,¹²⁸

The generally limited reporting of baseline characteristics precludes discussion of the extent of clinical heterogeneity across studies. Given the limited information available, clinical experts were consulted about the potential disparity across studies in populations enrolled. Experts fed back that they considered the population across the studies likely to be representative of people with AGWs and who present to GUM clinics. Thus, the project team considered it appropriate to carry out a MTC.

Interventions and comparators

Studies evaluating all interventions of interest were identified (see Table 4). Topical treatments evaluated were the self-applied imiquimod 5% cream and podophyllotoxin, and the clinician-applied TCAA and podophyllin 20–25%. Various concentrations (e.g. podophyllotoxin 0.5% cream and 0.15% cream), formulations (e.g. podophyllotoxin 0.5% cream and podophyllotoxin 0.5% solution) and application schedules of podophyllotoxin were evaluated across studies. Additionally, the duration of treatment with imiquimod 5% cream varied across studies. The licence for imiquimod 5% cream indicates that treatment can be given for up to 16 weeks and recommends an application schedule of three times per week on non-consecutive days.⁵⁸ Studies were identified evaluating application of imiquimod 5% for 4, 8, 12 and 16 weeks¹³⁰^,¹³³ and at application schedules of once daily, twice daily and three times daily.¹²⁹^,¹³⁴

After consultation with clinical experts on the potential influence of treatment duration, dose and formulation (relates to ease of application) on clinical effectiveness, the following assumptions were made for the purposes of carrying out meta-analysis, both direct and indirect:

Imiquimod 5% cream of any schedule (e.g. three times a week vs. once daily vs. twice daily) applied for 12 weeks is equivalent in clinical effectiveness to imiquimod 5% cream applied for 16 weeks; schedules of < 12 weeks’ duration were excluded from the analysis.
Clinician-applied podophyllin 20% and clinician-applied podophyllin 25% are clinically equivalent.
Different formulations of podophyllotoxin are of potentially sufficiently dissimilar clinical effectiveness to warrant analysis by preparation.
With the exception of podophyllotoxin 0.25% and 0.3%, doses of podophyllotoxin are potentially of sufficiently dissimilar clinical effectiveness to warrant analysis by dose.

Outcomes

The outcomes of interest to this review and reported in the included studies are listed in Table 5. No study reported data on the outcomes of relief of symptoms, malignancy or QoL. Although most studies presented results on the primary outcome of complete clearance of AGWs at the end of treatment, few studies presented data on complete clearance at subsequent time points. The definition of recurrence of AGWs differed slightly across studies. Most studies defined recurrence as the appearance of AGWs at a site previously cleared of AGWs, whereas some included appearance of AGWs at sites additional to those initially cleared.

TABLE 5

Summary of outcomes reported by RCTs included in the review

Quality assessment

No study was deemed to be at an overall low risk of bias, with the largest proportion of studies categorised as having an overall unclear risk of bias predominantly because of the limited reporting in the full publications (Table 6). In an attempt to supplement the information available, study authors were contacted with requests for additional detail on trial methodology. Only two authors replied by the prespecified deadline. Given that it is over 10 years since publication of most of the identified studies, the low response rate was to be expected.

TABLE 6

Summary of the risk-of-bias assessments of RCTs included in the review

All studies were described as randomised but details on methods used to generate the randomisation sequence were rarely reported. Assessment of clearance and recurrence of AGWs was subjective and thus at risk of bias. Most studies involving topical applications were described as double blinded but, as for randomisation, the full publications provided little information on methods implemented to initially conceal allocation and to subsequently maintain masking of treatment from clinicians and participants. Additionally, for most studies described as double blind, it was unclear whether or not the outcome assessor was the treating clinician and, if not, if the outcome assessor was masked to treatment. Some studies evaluating topical treatments were described as open label in design and thus were categorised as being at high risk of bias for most outcomes reported (see Table 6). Differences in setting could make implementation of masking problematic in studies evaluating self-applied against clinician-applied treatments and topical against ablative therapies. However, masking in these studies could be achieved using sham treatments.

Follow-up at the end of treatment was generally high across studies, with several studies categorised as being at low risk of attrition bias for the outcomes evaluated. However, follow-up at later time points to evaluate recurrence was variable, with high rates of loss to follow-up reported in several studies. Authors of some studies suggested that the low rate of return for further assessments could be attributed to treatment success, that is, complete clearance of AGWs without recurrence had been achieved and people felt that they needed no additional treatment or monitoring.

Of the domains relating to study characteristics, selective reporting was the domain most frequently determined to be at a high risk of bias (see Table 6). In these cases, results for the primary clinical effectiveness outcomes for this project (complete clearance and recurrence) were either not reported, despite being listed in the publication as a primary outcome, or not reported in a way that facilitated incorporation of the data into meta-analysis.

Assessment of effectiveness

Complete clearance at the end of treatment

The comparative clinical effectiveness of achieving complete clearance at the end of treatment was evaluated through a MTC and standard pair-wise meta-analysis. It should be noted that, as a consequence of variation in the duration of treatment for active topical interventions, the duration of placebo treatment or no treatment differs across studies. Given the consistently small proportion of people achieving complete clearance without treatment in the included studies, the project team considers variation in duration of placebo treatment to have a minimal impact on relative estimates of comparative clinical effectiveness.

The primary network generated included 22 studies⁶⁰^,⁶¹^,⁶⁶^,¹¹²^,¹¹⁵^,¹¹⁶^,¹¹⁸^,¹²²^,¹²³^,¹²⁶^,¹²⁷^,¹³²^,¹³⁸^,¹⁴⁰^,¹⁴⁵^,¹⁴⁶^,¹⁴⁸^,¹⁴⁹^,¹⁵³^,¹⁵⁴^,¹⁵⁶^,¹⁵⁹ and provided information on 15 treatments; a list of the studies informing the MTC for complete clearance at the end of treatment is presented in Table 5. A prespecified sensitivity analysis included studies that (1) were deemed to be at high risk of bias, (2) enrolled people with AGWs who were seropositive for HIV infection and (3) were reported only as conference abstracts. The sensitivity analysis incorporated an additional 17 studies⁶³^,⁶⁴^,⁶⁷^,⁶⁸^,¹¹⁴^,¹¹⁹^,¹²¹^,¹²⁴^,¹²⁵^,¹²⁸^,¹³¹^,¹³³^,¹³⁶^,¹³⁷^,¹⁴²^,¹⁴³^,¹⁵¹ and provided information on four extra treatments. The networks of evidence are presented in Figure 3. Although no longer recommended as a treatment for AGWs, podophyllin 20–25% (clinician applied) was chosen as the baseline treatment for the MTCs because of the comparatively large number of studies available for analysis.

FIGURE 3

Network diagrams of different interventions for complete clearance. The nodes represent the interventions. The thickness of the lines represents the number of studies informing the direct comparison. (a) Primary analysis; and (b) sensitivity analysis. (more...)

For the primary and sensitivity MTCs, analysis of model fit identified the random-effects model to be the best-fitting model in each case (Table 7). Additionally, the total residual deviance of the random-effects model in each analysis was closer to the number of data points included in the analysis (see Table 7). In the primary analysis, the model was a good fit for the data, with a residual deviance close to the number of unconstrained data points. There was evidence of heterogeneity in treatment effects across studies in the primary and sensitivity analyses (see Table 7). Codes for fixed- and random-effects models implemented in MTC analyses are supplied in Appendix 4.

TABLE 7

Summary of the MTC model characteristics

Results from the MTCs indicate that ablative techniques, particularly CO₂ laser therapy, have higher probabilities of being the best treatment than most of the topical treatments (summarised in Table 8). CO₂ laser therapy is associated with the largest probability of achieving complete clearance at the end of treatment, with probabilities of 97.1% [95% credible interval (CrI) 84.8% to 99.9%] and 96.9% (95% CrI 81.6% to 99.9%) in the primary and sensitivity analyses, respectively. The probabilities of complete clearance at the end of treatment for the primary and sensitivity analyses are summarised in Figures 4 and 5, respectively. By contrast, placebo was associated with a probability of clearance of only 7.6% (95% CrI 1.1% to 20.9%) in the primary analysis and 7.1% (95% CrI 1.7% to 17.7%) in the sensitivity analysis, with the duration of treatment with placebo ranging from 2 weeks to 16 weeks. Of the topical treatments, podophyllotoxin 0.5% solution and podophyllotoxin 0.3% solution were associated with the highest probabilities of completely clearing AGWs by the end of treatment (see Figures 4 and 5). In the primary analysis, podophyllotoxin 0.5% solution had a marginally higher probability of achieving the outcome than podophyllotoxin 0.3% solution (see Figure 4), which was the reverse of the results generated by the sensitivity analysis (see Figure 5). Of the topical treatments, podophyllin (clinician applied), imiquimod 5% cream and podophyllotoxin creams were associated with no or a low probability of being the best treatment (see Table 8).

TABLE 8

Probability of each treatment being the best treatment for achieving complete clearance at the end of treatment

FIGURE 4

Probability of complete clearance at the end of treatment: primary analysis. The square represents the mean probability of complete clearance for a particular treatment. The bar lines indicate the 95% CrI around the estimate of effect.

FIGURE 5

Probability of complete clearance at the end of treatment: sensitivity analysis. The square represents the mean probability of complete clearance for a particular treatment. The bar lines indicate the 95% CrI around the estimate of effect.

When compared with placebo or no treatment, in both the MTC primary and sensitivity analyses, all treatments evaluated were associated with a statistically significant improvement in complete clearance at the end of treatment (Figure 6) (full results of the MTC are presented in Appendix 5). Of the treatments evaluated, CO₂ laser therapy was associated with the largest improvement over placebo or no treatment for this outcome, with an OR of 6533 (95% CrI 65.49 to 25,760) in the primary analysis. The results of the MTC are in agreement with findings from standard pairwise meta-analyses, in which all interventions analysed were found to be statistically significantly more effective than placebo at effecting complete clearance at the end of treatment (Figure 7). It should be noted that a high level of statistical heterogeneity was present in the pairwise analysis of imiquimod 5% cream (I² = 53% in the primary analysis) and podophyllotoxin 0.5% cream (I² = 61% in the primary analysis) compared with placebo. Forest plots for individual pairwise meta-analyses are available in Appendix 6.

FIGURE 6

Forest plots presenting ORs and accompanying 95% CrIs for complete clearance of AGWs at the end of treatment for the comparison between the active interventions and placebo generated from the MTC. (a) Primary analysis; and (b) sensitivity analysis.

FIGURE 7

Forest plots presenting ORs and accompanying 95% CIs for complete clearance of AGWs at the end of treatment for the comparison between the active interventions and placebo generated from standard pairwise meta-analysis.

When compared with podophyllin 20–25%, the direction of the effect estimate favoured all other active interventions evaluated in primary and sensitivity analyses. However, most differences between treatments did not reach statistical significance (Figure 8). Of the treatments evaluated, CO₂ laser therapy (OR 104.6, 95% CrI 3.35 to 505.2) and podophyllotoxin 0.5% solution (OR 11.65, 95% CrI 2.65 to 38.50) were statistically significantly more effective than podophyllin 20–25% at completely clearing AGWs at the end of treatment (primary and sensitivity analyses). Additionally, sensitivity analyses identified electrotherapy as statistically significantly more effective than podophyllin 20–25% at achieving the outcome (OR 32.72, 95% CrI 3.15 to 150.80). Results from the MTC are predominantly in agreement with findings from standard pairwise meta-analysis. Of the direct evidence available, with the exception of TCAA and podophyllotoxin 0.15% cream, all treatments were associated with a higher probability than podophyllin 20–25% of completely clearing AGWs at the end of treatment (Figure 9). Forest plots for individual pairwise meta-analyses are available in Appendix 6.

FIGURE 8

Forest plots presenting ORs and accompanying 95% CrIs for complete clearance of AGWs at the end of treatment for the comparison between the interventions and podophyllin 20–25% generated from the MTC. (a) Primary analysis; and (b) sensitivity (more...)

FIGURE 9

Forest plots presenting ORs and accompanying 95% CIs for complete clearance of AGWs at the end of treatment for the comparison between the interventions and podophyllin 20–25% generated from standard pairwise meta-analysis.

In MTC analyses, there was no statistically significant difference between most treatments for complete clearance of AGWs at the end of treatment. Of those differences that reached statistical significance, most of the comparisons involved CO₂ laser therapy or podophyllotoxin 0.5% or 0.3% solution. CO₂ laser therapy was found to be significantly more effective than imiquimod 5% cream, TCAA and cryotherapy and the combinations of TCAA plus podophyllin and cryotherapy plus podophyllin. Analysis of direct evidence available for the comparison between CO₂ laser therapy and cryotherapy indicated a statistically significant difference between treatments that favoured CO₂ laser therapy (OR 22.08, 95 CI 7.37 to 66.16).

The MTC analysis indicated that podophyllotoxin 0.5% and 0.3% solution were both statistically significantly more effective at effecting complete clearance of AGWs than imiquimod 5% cream. However, analysis of direct evidence found no statistically significant difference between podophyllotoxin 0.5% solution and imiquimod 5% cream for this outcome (OR 1.50, 95% CI 0.47 to 4.76). Results from the MTC also demonstrated that podophyllotoxin 0.5% solution was associated with statistically significant improvements in complete clearance compared with podophyllotoxin 0.5% cream, podophyllotoxin 0.3% cream and TCAA. Full results of the MTC are presented in Appendix 5. A summary of the estimates of effect generated by the MTC and standard pairwise meta-analysis is presented in Table 9.

TABLE 9

Comparison of estimates of effect generated by the MTC and standard pairwise meta-analysis

The project team acknowledges the considerable uncertainty around the results, as evidenced by the wide CrIs (see Figures 6 and 8).

Complete clearance at another time point

Few identified studies reported clinical effectiveness data for complete clearance without recurrence at time points after cessation of treatment. Additionally, some studies reported data on this outcome only for people achieving complete clearance at the end of treatment rather than the full study population. Complete clearance without recurrence is distinct from recurrence as the former outcome accounts for people who clear within a few days of completion of treatment. Here, results for complete clearance evaluated at least 1 month after the end of treatment are reported from those studies aiming to observe all those randomised. Of the 60 studies included in the review, seven studies (based on the clinical assumptions outlined in Quantity and quality of research available) presented results on complete clearance at ≥ 1 month.¹¹⁴^,¹²⁹^,¹³³^,¹³⁴^,¹⁵⁰^,¹⁵⁹^,¹⁶⁰ Three studies were not suitable for the MTC as they evaluated various dosing schedules of the same treatment.¹²⁹^,¹³⁴^,¹⁵⁰ Only two¹¹⁴^,¹⁵⁹ of the four remaining studies were judged to be of unclear or low risk of bias, with the remaining studies deemed to be at high risk of bias because they were reported only as a conference abstract or enrolled people with HIV infection. Thus, a MTC was feasible only for the preplanned sensitivity analysis.

No study in the MTC evaluated podophyllin 20–25% and so cryotherapy was chosen as the baseline treatment because of the larger number of studies available. The fixed-effects model was the best-fitting model, with a deviance information criterion (DIC) of 47.9 (the DIC for the random-effects model was 48.1). The fixed-effects model was a good fit for the data, with a total residual deviance close to the number of data points analysed (residual deviance of 8.2 compared with eight unconstrained data points analysed).

Five interventions were indirectly compared in the MTC:

placebo or no treatment
imiquimod 5% cream (three times a week, patient applied)
cryotherapy
electrotherapy
cryotherapy plus podophyllotoxin 0.15% cream.

Of the five interventions analysed, electrotherapy was associated with the highest probability of achieving complete clearance without recurrence 3–6 months after the end of treatment (65.5%, 95% CrI 40.0% to 86.2%; Figure 10).

FIGURE 10

Probability of complete clearance at another time point by treatment: sensitivity analysis.

Results of the MTC indicate that the four active interventions are more effective than placebo at improving complete clearance without recurrence, based on median OR (Table 10). Mean OR estimates generated for comparisons with placebo/no treatment were unstable and fell outside the 95% CrI. The instability is likely due to the low event rate for placebo informing the analysis (1 event on 48 patients at risk), which leads to extreme values of the ratio of the odds and produces distributions that are highly skewed.

TABLE 10

Results of the MTC for complete clearance without recurrence

Results from the individual studies informing the MTC are in line with the findings of the MTC (Table 11).

TABLE 11

Results of direct evidence for complete clearance without recurrence

Of the three studies not included in the MTC, Fife et al.¹²⁹ and Trofatter et al.¹³⁴ compared various dosing schedules of imiquimod 5% cream and Simmons¹⁵⁰ assessed different doses of clinician-applied podophyllin. Meta-analysis of the studies by Fife et al.¹²⁹ and Trofatter et al.¹³⁴ found no statistically significant differences between imiquimod 5% cream applied three times weekly, once daily or twice daily (all possible comparisons) (forest plot presented in Appendix 7).

Recurrence

Based on advice from clinical experts, for the analysis of clinical effectiveness, recurrence of AGWs has been analysed by period of follow-up. Durations of follow-up assessed are from 3 months up to, but not including, 6 months (hereafter referred to as < 6 months) and ≥ 6 months (maximum reported follow-up of 12 months). A separate analysis of recurrence carried out to inform the cost-effectivness analysis included all relevant studies and encompassed recurrence from 3 months onwards. The results of this analysis are not discussed in this section but are presented in Appendix 8.

Analyses of recurrence are based on all people reported to have achieved complete clearance of AGWs at the end of treatment. In cases in which fewer people were followed up than cleared their AGWs, a worst-case scenario was implemented and people lost to follow-up were assumed to have undergone recurrence of AGWs.

Many study groups receiving placebo treatment or no treatment included no people with completely cleared lesions at the end of treatment and, consequently, no person could experience recurrence. As such, it was not possible to include placebo treatment in the analysis of recurrence, which restricted the network.

Recurrence at < 6 months

Applying the clinical assumptions outlined earlier identified four studies¹²⁷^,¹⁴⁰^,¹⁴⁵^,¹⁴⁸ to inform the primary analysis of recurrence occurring at < 6 months, which facilitated indirect comparison of:

podophyllin 20–25%
podophyllotoxin 0.5% solution
podophyllotoxin 0.25% solution
TCAA
TCAA plus podophyllin 20–25%.

The random- and fixed-effects models were similar in terms of goodness of fit (DIC 44.4 vs. 44.4, respectively) and had the same residual deviance (8.3 vs. 8.3, respectively), which was close to the number of unconstrained data points in the analysis (eight data points). However, because of the possibility of clinical heterogeneity in the populations of the trials combined in the network, the random-effects model was preferred.

There were no statistically significant differences between any comparisons for recurrence at < 6 months; results from the MTC are presented in Appendix 9. TCAA was associated with the lowest probability of recurrence (23.4%, 95% CrI 1.5% to 76.6%). By contrast, podophyllotoxin 0.25% solution had the highest probability of recurrence (66.9%, 95% CrI 5.2% to 99.5%). The probability of recurrence for all treatments is presented in Figure 11.

FIGURE 11

Probability of recurrence at < 6 months by treatment: primary analysis.

Sensitivity analysis of recurrence at < 6 months included an additional six studies⁶⁴^,⁶⁷^,¹³⁶^,¹³⁷^,¹⁴²^,¹⁵¹ and six other interventions:

podophyllotoxin 0.5% cream
podophyllotoxin 0.3% cream
podophyllotoxin 0.15% cream
podophyllin 0.5% solution (patient applied)
cryotherapy
electrotherapy.

The probability of recurrence for all treatments included in the sensitivity analysis is presented in Figure 12. Of the 11 interventions analysed, TCAA had the lowest probability of recurrence (24.7%, 95% CrI 8.7% to 46.9%). In marked contrast, podophyllotoxin 0.25% solution (83.3%, 95% CrI 49.5% to 98.5%) and podophyllin 0.5% solution (75.9%, 95% CrI 22.0% to 99.5%) had the highest probabilities of recurrence. The comparatively high probability of recurrence of the two topical treatments is reflected in the results of the MTC. As in the primary analysis, for most comparisons, the differences between interventions did not reach statistical significance. However, six comparisons were statistically significant, with four of these involving podophyllotoxin 0.3% solution and with the direction of effect favouring the comparator (podophyllotoxin 0.5% solution, TCAA, cryotherapy and electrotherapy). Additionally, TCAA was significantly more effective than podophyllin 20–25% (OR 0.35, 95% CrI 0.09 to 0.88) and podophyllotoxin 0.15% cream (OR 0.30, 95% CrI 0.06 to 0.90) at reducing recurrence at < 6 months. Full results of the MTC are presented in Appendix 9.

FIGURE 12

Probability of recurrence at < 6 months by treatment: sensitivity analysis.

Recurrence at ≥ 6 months

Four studies¹³²^,¹³⁸^,¹⁴⁶^,¹⁵⁵ reported recurrence at ≥ 6 months and informed the primary MTC, three¹³²^,¹³⁸^,¹⁴⁶ of which were included in the MTC. No additional studies were identified for inclusion in a sensitivity analysis.

The random- and fixed-effects models had a similar goodness of fit (32.8 vs. 32.8, respectively). However, the fixed-effects model had a slightly lower residual deviance than the random-effects model (6.1 vs. 6.2, respectively), which is close to the number of unconstrained data points in the analysis (six data points).

The network generated evaluated four interventions:

podophyllin 20–25%
imiquimod 5% cream
podophyllotoxin 0.5% solution
surgical excision.

Of the comparisons evaluated, only two differences were statistically significant. Surgical excision was found to be statistically significantly more effective than podophyllin 20–25% (OR 0.16, 95% CrI 0.03 to 0.43) and podophyllotoxin 0.5% solution (OR 0.14, 95% CrI 0.02 to 0.50) at reducing recurrence at ≥ 6 months. Full results of the MTC are presented in Appendix 9. Surgical excision was also associated with the lowest probability of recurrence of the four treatments (15.4%, 95% CrI 4.7% to 33.5%; summarised in Table 12).

TABLE 12

Probability of recurrence at ≥ 6 months by treatment

An additional study not included in the MTC¹⁵⁵ investigated whether or not argon plasma coagulation in combination with imiquimod 5% cream was more effective than argon plasma coagulation alone in the treatment of intra-anal AGWs. Viazis et al.¹⁵⁵ found no statistically significant difference between the two treatments for recurrence of AGWs at a mean follow up of 6 months [5/22 people recurred with combination treatment vs. 8/23 people recurred with monotherapy; OR 0.55 (calculated by project authors), 95% CI 0.15 to 2.06].

Time to complete clearance of anogenital warts

No identified study reported time to complete clearance in a format that facilitated inclusion in a standard pairwise meta-analysis or MTC. Six studies¹¹⁶^–¹¹⁸^,¹³²^,¹³⁴^,¹⁵⁵ presented results for time to complete clearance, but only one study¹⁵⁵ reported an accompanying measure of variance and statistical evaluation.

Five studies evaluated imiquimod 5% cream (conventional regimen) compared with placebo,¹¹⁶^–¹¹⁸ podophyllin 20%¹³² or alternative application schedules.¹³⁴ The median time to complete clearance for imiquimod 5% cream (conventional regimen) varied across the studies, ranging from 7 to 12 weeks.¹¹⁶^–¹¹⁸^,¹³²^,¹³⁴ Once daily and twice daily application of imiquimod 5% cream were associated with a median time to complete clearance of 6 and 8 weeks, respectively.¹³⁴ Padhiar et al.¹³² found a median time to complete clearance of 4.85 weeks for podophyllin 20%. Time to complete clearance for placebo ranged from 10 to 12 weeks.¹¹⁶^–¹¹⁸

Viazis et al.¹⁵⁵ found that adding imiquimod 5% cream to argon plasma coagulation significantly reduced the time to complete clearance [62.5 days, standard error (SE) 5.4 days, with argon plasma coagulation plus imiquimod 5% cream vs. 91.2 days, SE 6.4 days, with argon plasma coagulation alone; p = 0.0016].

Volume of clearance of anogenital warts

Presented analyses are based on the subgroups reduction in baseline AGW volume by < 50% and reduction in baseline AGW volume by ≥ 50% at the end of treatment. Results for clearance by ≥ 50% exclude those achieving complete AGW clearance, which is evaluated as a separate outcome of interest. Applying the clinical assumptions outlined earlier led to analysis of data from three studies judged to be of unclear or low risk of bias reporting < 50% clearance of AGW volume¹¹²^,¹²¹^,¹³² and six studies reporting ≥ 50% clearance of AGW volume.⁶¹^,¹¹²^,¹¹⁶^,¹¹⁸^,¹²¹^,¹³² Twelve other studies reporting volume of AGW clearance were not suitable for inclusion in the MTC as they were judged to be at a high risk of bias, evaluated treatments not included in the analysis or did not present data in a format suitable for inclusion in a meta-analysis¹¹⁴^,¹¹⁷^,¹¹⁹^,¹²⁰^,¹²⁴^,¹²⁵^,¹³⁴^,¹⁴³^,¹⁵⁰^,¹⁵⁹^,¹⁶¹^,¹⁶² (see Table 5).

Clearance of anogenital warts of < 50%

Arican et al.¹¹² and Snoeck et al.¹²¹ compared imiquimod 5% cream and cidofovir 1% gel, respectively, with placebo. The third study¹³² compared imiquimod 5% cream with podophyllin 20–25%.

The fixed-effects model was a slightly better fit than the random-effects model (DIC 28.6 vs. 28.7, respectively). The residual deviance of the fixed-effects model was also similar to the number of unconstrained data points analysed (residual deviance of 6.4 compared with six data points, respectively).

The MTC using the fixed-effects model indicated that the effect estimate for achieving < 50% clearance of AGWs was statistically significantly lower with imiquimod 5% cream than with placebo and cidofovir 1.0% gel (Table 13). For the comparison of imiquimod 5% cream and podophyllin 20–25%, the estimate of effect approached 1 and the difference between the treatments did not reach statistical significance.

TABLE 13

Results from the MTC for clearance of < 50% of the volume of AGWs

Clearance of anogenital warts of ≥ 50%

Five interventions were evaluated in the MTC:

placebo or no treatment
imiquimod 5% cream
podophyllotoxin 0.5% solution
podophyllin 20–25%
cidofovir 1% gel.

The fixed-effects model was a better fit than the random-effects model (DIC 64.2 vs. 65.3, respectively). The residual deviance of the fixed-effects model was also similar to the number of unconstrained data points in the analysis (11.7 vs. 12, respectively).

Based on the fixed-effects model, podophyllotoxin 0.5% solution was found to be significantly more effective than imiquimod 5% cream and placebo at reducing the volume of AGWs by ≥ 50% compared with the baseline volume (Table 14). No other difference between treatments achieved statistical significance.

TABLE 14

Results from the MTC for clearance of ≥ 50% of the volume of AGWs

Appearance of new anogenital warts during treatment

Ten identified studies referred to recording the appearance of new AGWs during treatment.⁶¹^,⁶²^,¹¹²^,¹¹⁶^–¹¹⁹^,¹²¹^,¹²⁹^,¹³⁸ Of the 10 studies, three were deemed to be at an unclear or low risk of bias, provided data in a format that could be incorporated into a meta-analysis and evaluated interventions meeting the assumptions outlined earlier.⁶¹^,¹¹⁶^,¹¹⁸ As per the protocol, sensitivity analysis for this outcome was not planned.

The network generated evaluated imiquimod 5% cream, podophyllotoxin 0.5% solution and placebo or no treatment. Placebo was chosen as the baseline treatment because a placebo or no treatment group was involved in all studies in the MTC. The random-effects model was the best-fitting model, with a DIC of 38.9; the DIC for the fixed-effects model was 41.1. The random-effects model was a good fit for the data, with a total residual deviance close to the number of data points analysed (residual deviance of 6.3 compared with six data points analysed).

No statistically significant differences were found between any comparisons for the probability of developing new AGWs during treatment (Table 15). In comparison with placebo or no treatment, the effect estimate favoured imiquimod 5% cream but not podophyllotoxin 0.5% solution (i.e. placebo was favoured over podophyllotoxin 0.5% solution; see Table 15).

TABLE 15

Results of the MTC for the development of new AGWs during treatment

Imiquimod 5% cream was associated with the lowest probability of new AGWs developing during treatment:

imiquimod 5% cream: 30.4% (95% CrI 6.7% to 68.5%)
podophyllotoxin 0.5% solution: 45.4% (95% CrI 5.8% to 91.1%)
placebo or no treatment: 49.8% (95% CrI 40.8% to 58.7%).

Adverse effects

Given the large number of studies identified, the decision was taken to restrict the comparison and reporting of AEs. After consultation with clinical experts, the project team focused on AEs highlighted as causing discomfort to the patient or being difficult to treat should they occur. The potential for a MTC was investigated for ulceration, blistering, erythema, oedema and itching. Sensitivity analyses for AEs were not planned and reporting of AEs is limited to data from studies deemed to be at low or unclear risk of bias. AEs data were extracted in full from individual studies and are presented in the data abstraction forms in Appendix 2.

Ulceration

Abdullah et al.¹⁵³ and Godley et al.¹⁵⁴ compared TCAA with cryotherapy and reported on the occurrence of ulceration. In both studies, a larger proportion of people in the TCAA group than in the cryotherapy group experienced ulceration. Standard pairwise meta-analysis using the fixed-effects model found that TCAA was associated with a significantly higher risk of ulceration than cryotherapy (OR 0.22, 95% CI 0.10 to 0.46; p < 0.0001); the forest plot is presented in Figure 13. However, the level of heterogeneity in the analysis was high (I² = 68%). Using a random-effects model as a sensitivity analysis generated a non-statistically significant difference between treatments (OR 0.13, 95% CI 0.01 to 1.62; p = 0.11).

FIGURE 13

Forest plot for the comparison between cryotherapy and TCAA for the outcome of ulceration. df, degrees of freedom; M–H, Mantel–Haenszel.

Blistering

In a comparison of cryotherapy and CO₂ laser therapy, Azizjalali et al.¹⁵⁶ found a low occurrence of blistering in each treatment group, with two and no people in the cryotherapy and CO₂ laser therapy groups, respectively, experiencing blistering (80 people were randomised to each group).

Erythema

Seven RCTs informed the MTC of erythema,⁶⁰^,⁶¹^,¹¹²^,¹¹⁶^,¹¹⁸^,¹²³^,¹³² generating a network incorporating five interventions:

placebo or no treatment
imiquimod 5% cream
podophyllin 20–25%
podophyllotoxin 0.5% solution
podophyllotoxin 0.5% cream.

The fixed-effects model was found to be a better fit than the random-effects model (DIC 75.0 vs. 76.2, respectively), with a similar residual deviance to the number of unconstrained data points (14.8 vs. 15, respectively).

Using the fixed-effects model with imiquimod 5% cream as baseline, the MTC found that all active interventions were associated with a statistically significant increase in risk of erythema compared with placebo (Table 16). However, no statistically significant differences in erythema were identified across the comparisons of active interventions (see Table 16).

TABLE 16

Results of the MTC for erythema

Of the five interventions compared, podophyllotoxin 0.5% solution was associated with the highest probability of occurrence of erythema:

placebo or no treatment: 19.1% (CrI 13.1% to 26.3%)
imiquimod 5% cream: 60.3% (95% CrI 52.6% to 67.7%)
podophyllin 20–25%: 44.5% (95% CrI 24.3% to 66.1%)
podophyllotoxin 0.5% solution: 66.8% (95% CrI 31.8% to 92.0%)
podophyllotoxin 0.5% cream: 50.9% (95% CrI 21.2% to 81.7%).

Oedema

The studies by Beutner et al.,¹¹⁶ Edwards et al.¹¹⁸ and Padhiar et al.¹³² were included in the MTC of oedema, facilitating comparison of imiquimod 5% cream, podophyllin 20–25% and placebo or no treatment.

The best-fitting model was the fixed-effects model, with a DIC of 31.5 compared with 32.3 for the random-effects model. Additionally, the fixed-effects model was a good fit for the data (residual deviance of 5.1 compared with six unconstrained data points in the analysis).

Imiquimod 5% cream and podophyllin 20–25% both statistically significantly increased the risk of oedema compared with placebo:

imiquimod 5% cream: OR 0.05 (95% CrI 0.01 to 0.13; OR < 1 favours placebo)
podophyllin 20–25%: OR 316.2 (95% CrI 39.86 to 1304; OR > 1 favours placebo).

Additionally, podophyllin 20–25% was found to statistically significantly increase the risk of oedema compared with imiquimod 5% cream (OR 12.39, 95% CrI 2.74 to 40.21; OR > 1 favours imiquimod 5% cream).

Itching

Three studies informed the MTC of itching,⁶¹^,¹¹⁶^,¹³² enabling comparison of imiquimod 5% cream, podophyllin 20–25%, podophyllotoxin 0.5% solution and placebo or no treatment.

The random- and fixed-effects models were similar in terms of goodness of fit (DIC 35.2 vs. 35.2, respectively) and had the same residual deviance (6.1 vs. 6.1, respectively), which was close to the number of unconstrained data points in the analysis (six data points). However, because of the possibility of clinical heterogeneity in the populations of the trials combined in the network, the random-effects model was preferred.

There was no significant difference in the risk of itching between any of the active treatments compared with placebo (Table 17). Similarly, there was no significant difference between any of the active treatments compared with one another for this adverse event.

TABLE 17

Results of the MTC for itching

Of the active interventions, podophyllin 20–25% had the highest probability of being associated with itching (51.2%, 95% CrI 6.8% to 93.7%). Imiquimod 5% cream and podophyllotoxin 0.5% solution had similar probabilities of being associated with itching, at 30.7% (95% CrI 21.5% to 40.6%) and 39.9% (95% CrI 0.1% to 96.5%), respectively.

Summary of evidence synthesis

Analysis by MTC indicated that, in line with the conclusions outlined in European guidelines,⁶ ablative techniques, and in particular CO₂ laser therapy, are generally associated with higher probabilities of complete clearance of AGWs at the end of treatment. There was considerable disparity in the probability of achieving complete clearance between podophyllotoxin 0.5% solution and imiquimod 5% cream, which are the mainstays of topical treatment. Podophyllotoxin 0.5% solution had a 92.6% (CrI 81.8% to 98.4%) probability of completely clearing lesions compared with 56.1% (CrI 20.3% to 85.0%) for imiquimod 5% cream. However, the wide CrIs indicate that there is considerable uncertainty associated with the results and the findings should be interpreted with caution.

In the primary MTC, there was no statistically significant difference between most of the treatments evaluated for complete clearance of AGWs at the end of treatment. Of those differences that reached statistical significance, most of the comparisons involved CO₂ laser therapy or podophyllotoxin 0.5% or 0.3% solution.

CO₂ laser therapy was found to be significantly more effective than:

imiquimod 5% cream: OR 247.0 (95% CrI 3.03 to 1087; OR > 1 favours CO₂ laser therapy)
TCAA: OR 86.15 (95% CrI 4.05 to 415.3; OR > 1 favours CO₂ laser therapy)
cryotherapy: OR 44.61 (95% CrI 3.30 to 201.7; OR > 1 favours CO₂ laser therapy)
TCAA plus podophyllin: OR 0.13 (95% CrI 0.003 to 0.59; OR < 1 favours CO₂ laser therapy)
cryotherapy plus podophyllin: OR 0.22 (95% CrI 0.004 to 0.94; OR < 1 favours CO₂ laser therapy).

Podophyllotoxin 0.5% solution was associated with statistically significant improvements in complete clearance at the end of treatment compared with:

podophyllotoxin 0.5% cream: OR 0.30 (95% CrI 0.04 to 0.99; OR < 1 favours podophyllotoxin 0.5% solution)
podophyllotoxin 0.3% cream: OR 0.19 (95% CrI 0.007 to 0.874; OR < 1 favours podophyllotoxin 0.5% solution)
TCAA: OR 0.17 (95% CrI 0.02 to 0.63; OR < 1 favours podophyllotoxin 0.5% solution).

Limited reporting of data in available publications for other outcomes of interest led to restricted networks involving few interventions.

Copyright © Queen’s Printer and Controller of HMSO 2016. This work was produced by Thurgar et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK355292

Contents