Circumcision devices versus standard surgical techniques in adolescent and adult male circumcisions

Abstract

Background

Medical circumcisions are among the most common surgical procedures performed in males. The usual indications are phimosis (inability to completely retract the foreskin and expose the glans due to a congenital or acquired constriction of the prepuce), paraphimosis (when the foreskin is not pulled back over the glans after retraction resulting in a tight constricting band which causes swelling of the distal penis and acute discomfort), balanoposthitis (erythema and edema of the prepuce and glans) and balanitis (inflammation is confined to the glans; the foreskin is usually non‐retractile). Circumcision devices have been developed to shorten the operative time, simplify techniques, and improve safety and cosmetic outcomes. The devices generally aim to crush the foreskin while simultaneously creating hemostasis, the foreskin is then excised or allowed to slough off. Their use is supposedly safer and easier to replicate than the standard dissection techniques. There are at least 20 devices for male circumcision on the market, yet their effectiveness has not been reviewed to date.

Objectives

To assess the effects of device‐based circumcisions compared with standard surgical techniques in adolescent and adult males (10 years old and above).

Search methods

We performed a comprehensive search with no restrictions to the language of publication or publication status. We searched the Cochrane Library, MEDLINE (PubMed), Embase, Web of Science, trials registries, grey literature sources and conference proceedings up to 16 April 2020.

Selection criteria

We included randomized controlled trials of device‐based circumcisions (crush or ligature circumcision devices) compared to standard surgical dissection‐based circumcision conducted by health professionals in a medical setting.

Data collection and analysis

At least two review authors independently assessed study eligibility and extracted data from the included studies. We classified adverse events into serious, moderate or mild. We reported study results as risk ratios (RR) or mean differences (MD) using 95% confidence intervals (CI) and a random‐effects model. We used the GRADE approach to evaluate the overall certainty of the evidence for each outcome.

Main results

Eighteen trials met the inclusion criteria. Trials were conducted in China, South Africa, Kenya and Zambia, Mozambique, Rwanda, Uganda and Zimbabwe.

Primary outcomes

Serious adverse events: there were no serious adverse events in either treatment arm (11 trials, 3472 participants).

Moderate adverse events: there may be a slight increase in moderate adverse events when devices are used compared to standard surgical techniques (RR 1.31, 95% CI 0.55 to 3.10; I²= 68%; 10 trials, 3370 participants; low‐certainty evidence); this corresponds to 8 more (ranging from 15 fewer to 84 more) moderate adverse events per 1000 participants. We downgraded the certainty of the evidence for study limitations and imprecision.

Secondary outcomes

Mild adverse events: we are uncertain about the difference in mild adverse events between groups when devices are used compared to standard surgical techniques (RR 1.09, 95% CI 0.44 to 2.72; I² = 91%; 10 trials, 3370 participants; very low‐certainty evidence). We downgraded the certainty of the evidence for study limitations, imprecision and unexplained inconsistency.

Operative time: operative time is probably about 17 minutes shorter when using a device rather than standard surgical techniques, which constitutes a clinically meaningful decrease in a procedure (MD –17.26 minutes, 95% CI –19.96 to –14.57; I² = 99%; 14 trials, 4812 participants; moderate‐certainty evidence). We downgraded the certainty of the evidence for serious study limitations. The standard surgical technique generally takes about 24 minutes.

There may be less postoperative pain during the first 24 hours when circumcision devices are used compared to standard surgical techniques (measured using a visual analog scale [VAS]; MD 1.30 cm lower, 95% CI 2.37 lower to 0.22 lower; I² = 99%; 9 trials, 3022 participants; low‐certainty evidence). We downgraded the certainty of the evidence for study limitations and unexplained heterogeneity. There may be little or no difference in postoperative pain experienced during the first seven days when compared with standard surgical techniques (measured using a VAS; MD 0.11 cm higher, 95% CI 0.89 lower to 1.11 higher; I² = 94%; 4 trials, 1430 participants; low‐certainty evidence). We downgraded the certainty of the evidence for study limitations and unexplained inconsistency. A higher score on the VAS indicates greater pain.

Participants may slightly prefer circumcision devices compared to standard surgical techniques (RR 1.19, 95% CI 1.04 to 1.37; I² = 97%; 15 trials, 4501 participants; low‐certainty evidence). We downgraded the certainty of the evidence for study limitations and unexplained inconsistency. We recorded satisfaction as a dichotomous outcome. Higher rates reflected greater satisfaction.

Authors' conclusions

We found that there were no serious adverse events reported when using a circumcision device compared to standard surgical techniques, but they may slightly increase moderate adverse effects, and it is unclear whether there is a difference in mild adverse effects. Use of circumcision devices probably reduces the time of the procedure by about 17 minutes, a clinically meaningful time saving. For patients, use of the circumcision device may result in lower pain scores during the first 24 hours and patients may be slightly more satisfied with it compared with standard surgical techniques. Clinicians, patients and policymakers can use these results in conjunction with their own contextual factors to inform the approach that best suits their healthcare settings. High‐quality trials evaluating this intervention are needed to provide further certainty regarding the rates of adverse effects and postoperative pain of using devices compared to standard approaches.

Plain language summary

Are circumcision devices safer or faster when compared to standard surgical circumcisions for males older than 10 years?

Background

Male circumcisions have been performed for many centuries and are one of the most common surgical procedures in males. Medical doctors usually perform circumcisions by removing the foreskin found on the penis in a surgical procedure. Some doctors use circumcision devices specifically designed for medical male circumcisions. It is believed that they may save time, and be simpler and safer circumcision methods. But, it is unclear from research evidence whether males circumcised with these devices have better health outcomes.

Study characteristics

After a wide search of the literature, we found 18 clinical trials including 5246 males. We looked at information from trials that compared surgical and device‐based circumcision procedures. We compared the complications (negative experiences) that patients had after the circumcision; the amount of time a healthcare provider took to do the circumcision; patient's pain during the day after the procedure and one week afterwards; and the amount of satisfaction patients felt with the procedure.

Key results

There was probably little to no difference in serious complications (negative effects) such as admission to hospital or permanent damage to the penis in any trial whether men had a standard surgical circumcision or used a circumcision device.

There may be slightly more moderate complications that require additional treatment such as stitches or antibiotics for those circumcised with devices, but further evidence may help us understand this better.

We are unsure whether or not there is a difference in mild complications such as minor bleeding that requires added treatment, but further evidence may help us understand this better.

The average duration for the standard surgical procedure is about 24 minutes (ranging from 15 minutes to 31 minutes), compared to the use of circumcision devices that has an average duration of about 7 minutes (ranging from 3 minutes to 13 minutes). Therefore, the review found that using a circumcision device saved about 17 minutes compared to standard surgical circumcision.

There may be less pain during the first 24 hours in patients who were circumcised with the device compared to standard surgical circumcision. There may be little or no difference in pain in the first seven days after circumcision for patients who were circumcised with the device compared to those circumcised with standard surgical methods.

We found that patients were slightly more satisfied when circumcised with a device compared to those that received a standard surgical circumcision procedure.

Conclusions

Overall, our review found that circumcision devices may have slightly more complications than standard surgery, such as requiring antibiotics or stitches. We also found that procedures done with circumcision devices probably take less time to complete. Patients may feel less pain in the first 24 hours after a circumcision procedure when a device is used compared to the standard surgery. Patients may slightly prefer the use of a device compared to standard surgery. When deciding whether to use a device, these results should be kept in mind alongside consideration of the local context such as costs and access to trained healthcare workers in different healthcare settings. Further trials can help us to understand the benefits and harms with more certainty, particularly about complications and postoperative pain.

Summary of findings

Summary of findings 1. Circumcision devices compared to standard surgical techniques in adolescent and adult male circumcisions.

Circumcision devices compared to standard surgical techniques in adolescent and adult male circumcisions
Patient or population: adolescent and adult male circumcisions Setting: outpatient Intervention: circumcision devices Comparison: standard surgical techniques
Outcomes	№ of participants (studies)	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Anticipated absolute effects* (95% CI)		Comments
				Risk with standard surgical techniques	Risk difference with circumcision devices
Serious adverse events	3472 (11 RCTs)	⊕⊕⊕⊝ Moderatea,b	Not pooled	Study population		No serious adverse events reported. There is probably little to no difference between circumcision devices and standard surgical approaches.
				Not pooled	Not pooled
Moderate adverse events	3370 (10 RCTs)	⊕⊕⊝⊝ Lowa,c	RR 1.31 (0.55 to 3.10)	Study population		There is may be a slight increase in moderate adverse effects when using circumcision devices compared to surgical techniques.
				27 per 1000	8 more per 1000 (12 fewer to 57 more)
Mild adverse events	3370 (10 RCTs)	⊕⊝⊝⊝ Verylowa,c,d	RR 1.09 (0.44 to 2.72)	Study population		We are uncertain whether devices or surgery are different with respect to mild adverse effects.
				114 per 1000	10 more per 1000 (64 fewer to 195 more)
Operative time (minutes)	4812 (14 RCTs)	⊕⊕⊕⊝ Moderatea,e	—	The mean operative time (minutes) was 0	MD 17.26 lower (19.96 lower to 14.57 lower)	Use of circumcision devices probably reduces operative time by about 17 minutes compared to standard surgical techniques which take on average 24 minutes.
Pain during the first 24 hours (VAS means)	3022 (9 RCTs)	⊕⊕⊝⊝ Lowa,d	—	The mean pain during the first 24 hours (VAS means) was 0	MD 1.3 lower (2.37 lower to 0.22 lower)f	There may be less pain during the first 24 hours after procedure when using circumcision devices compared to standard surgical techniques.
Pain during the first 7 days (VAS means)	1430 (4 RCTs)	⊕⊕⊝⊝ Lowa,d	—	The mean pain during the first 7 days (VAS means) was 0	MD 0.11 higher (0.89 lower to 1.11 higher)f	There may be little to no difference between circumcision devices and standard surgical approaches for pain during the first 7 days.
Participant satisfaction	4501 (15 RCTs)	⊕⊕⊝⊝ Lowa,d	RR 1.19 (1.04 to 1.37)	Study population		Participants may slightly prefer the device compared to standard surgical techniques.
				751 per 1000	143 more per 1000 (30 more to 278 more)g
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio VAS: visual analogue scale.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level for serious study limitations: poor reporting on sequence generation and allocation concealment, and detection bias due to lack of blinding of outcome assessors.
^bThere were no reports of serious adverse events in any of the 11 trials reporting this outcome for either comparison. Thus, the best estimate of the risk ratio would thus be one as there is probably no difference regardless of method for circumcision.
^cDowngraded one level for serious imprecision: the confidence intervals were wide including appreciable benefit and harm with low numbers of events in each arm.
^dDowngraded one level for serious inconsistency: there was considerable unexplained heterogeneity.
^eNot downgraded for inconsistency: despite statistical heterogeneity, there was a consistent finding of reduced operation time in the intervention group of approximately 10 minutes or more compared to standard surgical methods.
^fA higher score on the VAS indicates greater pain. Pain was considered clinically significant if there was a minimally clinical important difference of 1.0 cm on the 10‐cm VAS.
^gParticipant satisfaction (cosmesis) was recorded as a dichotomous outcome. We extracted participants' satisfaction with their circumcision using a device and standard surgical circumcision. Trials measured this using three different approaches. The first approach was a two‐point scale, participants indicated that they were 'satisfied' or 'unsatisfied' with their procedure. The second was a three‐point scale, participants indicated that they were 'very satisfied,' 'satisfied' or 'not satisfied.' We combined 'very satisfied' and 'satisfied.' The third was a five‐point scale, participants indicated that they were 'extremely dissatisfied,' 'dissatisfied,' 'neutral,' 'satisfied' or 'extremely satisfied.' We combined responses 'satisfied' or 'extremely satisfied.' Higher rates reflected greater satisfaction.

Background

Description of the condition

Circumcisions have been recorded in Egypt as early as 2300 BC (Auvert 2005; WHO/UNAIDS 2007), and are among the most common surgical procedures performed in males (Malone 2007). The usual indications for surgical or dissection technique‐based circumcision are the following foreskin conditions. Phimosis (inability to completely retract the foreskin and expose the glans due to a congenital or acquired constriction of the prepuce) (WHO/UNAIDS 2007), which causes swelling, including difficult painful erections, candidiasis and numerous sexually transmissible infections (Morris 2017). Paraphimosis (when the foreskin is not pulled back over the glans after retraction resulting in a tight constricting band that causes swelling of the distal penis and acute discomfort) (Malone 2007). Balanoposthitis (erythema and edema of the prepuce and glans) and balanitis (inflammation is confined to the glans; the foreskin is usually non‐retractile) (Malone 2007). Circumcisions are also performed for hygiene, personal, cultural, ritual or religious reasons (in Jewish, Muslim and traditional African cultures) (WHO/UNAIDS 2007), and decrease the risk of sexually transmitted infections (STIs) and HIV transmission (Siegfried 2009). Importantly, voluntary medical male circumcision (VMMC) is a key World Health Organization (WHO) HIV preventive intervention (WHO/UNAIDS 2011). The results of three randomized controlled trials (RCTs) conducted in Africa showed that VMMC decreased the risk of female‐to‐male sexual transmission by approximately 60% (Auvert 2005; Bailey 2007; Gray 2007). Globally, the prevalence of circumcision among males 15 years or older is estimated to be 36.7% (95% confidence interval [CI] 31.4 to 42.0) (Morris 2016). In Southern Africa, prevalence varies: 15.1% in Botswana, 8.2% in Swaziland, 12.8% in Zambia and 9.2% in Zimbabwe, with higher prevalence in Namibia (25.5%), Malawi (21.6%), South Africa (44.7%) and Lesotho (52.0%) (Morris 2016). The Joint United Nations Programme on HIV and AIDS (UNAIDS) and WHO have prioritized 16 countries in southern and East Africa for scale‐up of VMMC as well as the roll‐out of routine medical circumcision for newborn boys (WHO/UNAIDS 2013). HIV prevention programs in these countries performed over 22 million circumcisions between 2008 and 2018 (UNAIDS 2019).

Description of the intervention

The basis of circumcision devices (irrespective of the individual type of device) is crushing of the foreskin at the proposed tissue apposition line and simultaneously obtaining hemostasis. The foreskin is then excised or allowed to slough off by ischemic necrosis. The crushed apposed edges can then be suture reinforced, glued or are sometimes left to heal (Alanis 2004). There are at least 20 identified devices for male circumcisions (Bakare 2008; WHO 2012). The three commonly used devices are the Gomco clamp, the Mogen clamp and the Plastibell (WHO/UNAIDS 2007). The WHO has prequalified two adult devices: PrePex in 2013 and Shang Ring (SR) in 2015 (WHO 2019). Circumcision devices can be further classified as ligature devices (i.e. they allow the foreskin to slough off by ischemic necrosis with no suturing apposition needed) or crush devices (i.e. they provide crushing hemostasis and simultaneous apposition, the foreskin is excised and edges are suture re‐enforced) (WHO 2013a). The dissection techniques involve using sharp dissection, cautery or ligation of bleeding vessels and suturing to apposed edges. The types of dissection techniques are the traditional forceps‐guided technique, the dorsal slit technique and the inner ring‐outer ring (sleeve) techniques.

Adverse events may include bleeding, hematoma, wound infection, wound disruption and penile injury. The WHO's Framework for Clinical Evaluation of Devices for Adult Male Circumcision provides standardized definitions for grading adverse events as mild, moderate or severe (WHO 2012; WHO 2013a). In brief, adverse events are categorized as mild if they require little or no intervention (e.g. mild wound disruption or slight bleeding), moderate if they require active treatment (e.g. antibiotics or suturing) or severe if they require transfusion or hospitalization or result in permanent damage (Millard 2013).

How the intervention might work

Circumcision devices have been developed to shorten the operative time, simplify techniques, and improve safety and cosmetic outcomes (Peng 2008). Device‐based techniques generally provide protection to the glans, reliably circumcise adequate foreskin and provide crush hemostasis. This technique is supposedly safer and easier to replicate than the standard dissection techniques (Bakare 2008). Device‐based techniques also allow for task‐shifting, as nurses and other non‐physician healthcare providers may safely perform them, thus allowing for rapid scale‐up of VMMC for HIV prevention in resource‐constrained settings (Ridzon 2016; WHO 2013b). It is important to note that the use of device‐based techniques is precluded in men with penile anatomic abnormalities, chronic paraphimosis and active genital infection (Mutabazi 2012; Peng 2008). The shorter procedure duration and other advantages described would then be appropriate for men with normal penile anatomy, seeking circumcision for hygiene, personal, religious reasons or to prevent HIV and other STIs.

Why it is important to do this review

For policymakers, there are several factors that form part of the requirements that need to be taken into consideration when introducing a circumcision device to low‐resource settings. First, ease of use (with a short procedure time), easy and practical removal, and it should be suitable for mid‐level providers to use. Second, it should be low cost or affordably priced, thus, having a cost advantage over standard surgical procedures. Third, it needs to meet regulatory and marketing criteria that support high‐quality clinical data on its safety and effectiveness, and preferably is used in age groups relevant for the country intending to use it (WHO 2012). Currently, research indicates that available circumcision devices have the potential to reduce the complexity and duration of the male circumcision procedure; however, the high number of circumcisions performed can be demanding on both human and financial resources. One study reported a median duration of 30 minutes which ranged from 18 minutes to 63 minutes (Krieger 2005). Therefore, an effective, safe, inexpensive and easy‐to‐use device would assist in easing any burden (Bakare 2008; Millard 2013; WHO 2012). With more types of devices being manufactured, albeit with the same mechanism, it is important to categorically compare the efficacy of circumcision devices with the dissection technique. To date, the WHO Medical Circumcision Technical Advisory Group has published guidelines on the use of PrePex and SR devices with recommendations made based on comparative and non‐comparative studies (WHO 2013a; WHO 2014). This Cochrane Review compared standard surgical techniques with circumcision devices regarding benefits and harms and followed the methodologic standards of a Cochrane Review, together with the application of GRADE and generation of a 'Summary of findings' table.

Objectives

To assess the effects of device‐based circumcisions compared with standard surgical techniques in adolescent and adult males (10 years old and above).

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs regardless of their publication status or the language of publication.

Types of participants

We included trials that included males, ages 10 years and older (adolescent and adult males), as they are the target of massive circumcision campaigns in HIV prevention initiatives using devices. Since neonates and infants are almost exclusively circumcised using devices, and for religious reasons, we excluded them from this review. The Characteristics of included studies table describes detailed specific exclusion criteria for both dissection and device‐based circumcisions.

Types of interventions

Experimental interventions

Device‐based circumcision (crush or ligature circumcision devices): crush devices such as the SR, circular stapler, Gomco clamp and Unicirc provide crushing hemostasis through tight compression of the foreskin between hard surfaces and simultaneous apposition, the foreskin is excised and edges are suture reinforced. Injection of local anesthesia is required for pain control as the device crushes the foreskin on placement, and tissue is excised immediately after device placement. In contrast, with ligature devices such as PrePex, there is slow compression of the foreskin between an elastic ring and a hard surface that allows the foreskin to slough off by ischemic necrosis with no suturing apposition needed. Such devices can be used without injected local anesthetic (WHO 2013a).

Comparator interventions

Dissection‐based circumcision, any recognized dissection technique (forceps‐guided method, dorsal slit method or the sleeve method) conducted by health professionals in a medical setting.

Types of outcome measures

Based on perceived patient‐importance, we distinguished between primary and secondary outcomes. 

Primary outcomes

• Serious adverse events defined as events requiring a blood transfusion, hospitalization or resulting in permanent damage (e.g. penile injury occurring within the intraoperative and early postoperative period [30 days]).

• Moderate adverse events defined as events requiring active treatment such as suturing, antibiotics, and surgical hemostasis within the intraoperative and early postoperative period (30 days).

Secondary outcomes

• Mild adverse events require little or no intervention (e.g. slight wound disruption, minor bleeding and occur within the intraoperative and early postoperative period [30 days]).

• Operative time measured in minutes. An operative time longer than 10 minutes was considered a clinically important difference from a surgical perspective. In three trials, we combined two reported subgroups into a single group (Lv 2014; Miao 2015; Sokal 2014). We referred to the Cochrane Handbook for Systematic Reviews of Interventions Section 7.7.3.8 on combining groups to guide us and used the formulae in Table 7.7a to combine the numbers into a single sample size, mean and standard deviation (SD) for each group. Lv 2014 and Miao 2015 reported mean operative times separately for both circumcision device groups and the surgical group. Sokal 2014 had two trial sites, one in Kenya and one in Zambia. The operative times were reported separately for both groups located at both trial locations. Thus, we combined the means, SDs and sample sizes in the circumcision device groups and for the surgical groups from Kenya and Zambia.

• Postoperative pain measured using the visual analog scale (VAS) centimeter (cm) scale. Minimal clinically important difference (MCID) for pain in circumcisions is undefined in the literature; however, it has been defined for other types of postoperative pain where a change of 10 mm in the 100 mm VAS is considered the MCID (Myles 2017). All included trials in our review used a 10‐point VAS where '0' corresponded to 'no pain at all' and '10' to 'worst pain imaginable' (WHO 2013a). In the absence of definitive data on the MCID of pain after circumcision using VAS, we considered an MCID of 1.0 cm on a 10‐cm VAS scale as clinically significant for this review. A higher score on the VAS indicated greater pain. Pain was evaluated in four categories: pain during the first 24 hours, pain during the first seven days, pain during erection and pain during device removal. These time points were considered most comparable, allowing for comparisons between circumcision devices and standard surgical circumcision. Two trials had three interventions arms consisting of two crush circumcision device arms and a surgical arm (Lv 2014; Miao 2015). To analyze the outcome pain during the first 24 hours for Miao 2015 and pain during the first seven days for Lv 2014, we combined the two circumcision device arms into a single group. We referred to the Cochrane Handbook for Systematic Reviews of Interventions Section 7.7.3.8 on combining groups to guide us and used the formulae in Table 7.7a to combine the numbers into a single sample size, mean and SD for each group.

• Participant satisfaction (cosmesis) recorded as a dichotomous outcome. We extracted participants' satisfaction with their circumcision using circumcision devices and standard surgical circumcision. Trials measured this using three different approaches. The first approach was a two‐point scale, participants indicated that they were 'satisfied' or 'unsatisfied' with their procedure. The second was a three‐point scale, participants indicated that they were 'very satisfied,' 'satisfied' or 'not satisfied.' We combined 'very satisfied' and 'satisfied.' The third was a five‐point scale, participants indicated that they were 'extremely dissatisfied,' 'dissatisfied,' 'neutral,' 'satisfied' or 'extremely satisfied.' We combined responses 'satisfied' or 'extremely satisfied.'

• Inadequate removal of the foreskin.

Method and timing of outcome measurement

See 'Definitions and timing of outcome measurement' in Appendix 1.

Search methods for identification of studies

We performed a comprehensive literature search with no restrictions to the language of publication or publication status. We repeated the search three months before the anticipated publication of the review.

Electronic searches

We conducted electronic searches for RCTs using the key search terms for circumcision, device and individual device names (e.g. Gomco, Mogen, Plastibell) (Appendix 2). We excluded studies of neonates and infants, and used the Cochrane RCT maximizing filter (Lefebvre 2011).

We searched the following databases, using the key search terms with added Boolean operators.

• The Cochrane Library: to 16 April 2020:

  • Cochrane Database of Systematic Reviews (CDSR);

  • Cochrane Central Register of Controlled Trials (CENTRAL);

  • Database of Abstracts of Reviews of Effects (DARE);

  • Health Technology Assessment Database (HTA).

• MEDLINE (PubMed): to 16 April 2020.

• Embase was searched on 16 August 2016, but was not searched on 16 April 2020 as the license was not renewed.

• Web of Science (WOS) Core Collection was searched on 16 April 2020.

We also searched the following for planned, ongoing or not yet published studies:

• ClinicalTrials.gov (www.clinicaltrials.gov/): to 16 April 2020;

• WHO International Clinical Trials Registry Platform (ICTRP) search portal (apps.who.int/trialsearch/), a meta‐register of studies with links to numerous other trials registers. The search was conducted on 4 May 2018 and repeated on 16 April 2020.

We provided the search strategies for the Cochrane Library, MEDLINE, Embase, Web of Science and ClinicalTrials.gov in Appendix 2.

Searching other resources

We tried to identify other potentially eligible trials or ancillary publications by searching the reference lists of retrieved included trials, reviews, meta‐analyses and health technology assessment reports. We attempted to contact the authors of trials for trial data (Lagarde 2009; Kanyago 2013), but were unsuccessful with Lagarde 2009. We searched abstract proceedings of relevant meetings such as the American Urological Association, the European Association of Urology and the Society of Sexual Medicine from 2013 to 2021 for relevant studies.

Data collection and analysis

We used reference management software (e.g. EndNote) to identify and remove potential duplicate studies. Three review authors (MZS, SE, AH) independently scanned the abstract and title of remaining records to determine which studies were potentially eligible.

Selection of studies

Three review authors (MZS, SE, AH) read the full texts of all potentially relevant records, mapped records to trials, and classified studies as included studies, excluded studies or ongoing studies in accordance with the criteria for each as provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). We resolved any discrepancies through consensus or with input from a fourth review author (TK). If the resolution of disagreement was not possible, we planned to designate the study as 'awaiting classification' and we would have contacted the study authors for clarification, but this did not happen. However, we found 12 non‐English language articles that required feedback on whether they met our eligibility criteria. Where possible we contacted trial authors to obtain full texts and to review the eligibility of their studies for this review; none responded to our requests. We then contacted Chinese language‐speaking colleagues via Cochrane Task Exchange (taskexchange.cochrane.org/), who found the published full‐text articles and assessed their eligibility based on title and abstract and, where necessary, assessed eligibility based on full‐text assessment. We then documented reasons for all excluded studies in the Characteristics of excluded studies table. We presented an adapted PRISMA flow diagram reporting the process of study selection (Figure 1) (Liberati 2009).

1.

Study flow diagram.

Data extraction and management

We developed a pilot‐tested data extraction form.

For studies that fulfilled the inclusion criteria, three review authors (MZS, SE, AH) independently extracted the following information, which is provided in the Characteristics of included studies table.

• Study design.

• Study dates (if dates were not available then this was reported as such).

• Study setting and country.

• Participant inclusion and exclusion criteria.

• Participant details and baseline demographics.

• Number of participants by study and by study arm.

• Details of relevant experimental and comparator interventions such as device type, name and method of dissection.

• Definitions of relevant outcomes, method and timing of outcome measurement, and any relevant subgroups.

• Study funding sources.

• Declarations of interest by primary trial authors.

Dealing with duplicate and companion publications

When there were duplicate publications, companion documents or multiple reports of a primary study, we maximized the yield of information by mapping all publications to unique studies and collated all available data in one data extraction form. We used the complete dataset aggregated across all known publications and noted one of the studies as the primary reference. We collated the reports of the same study so that each study, rather than each report, was the unit of interest for the review, and such studies had a single identifier with multiple references.

We extracted outcome data that were relevant to this Cochrane Review to calculate summary statistics and measure variance. We obtained the number of events and totals for populations from 2 × 2 tables for dichotomous outcomes. For continuous outcomes, we obtained means and SDs.

We settled disagreements through discussion. If necessary, we consulted the fourth review author (TK). We provided information, including trial identifier, from potentially relevant ongoing studies in the Characteristics of ongoing studies table. We attempted to contact authors of included studies to obtain missing data required for our analysis or to clarify ambiguous outcomes reported.

Assessment of risk of bias in included studies

Three review authors (MZS, SE, AH) independently assessed the risk of bias of each included study. We resolved disagreements by consensus or by consulting a fourth review author (TK).

We assessed risk of bias using the Cochrane 'Risk of bias' assessment tool (Higgins 2011b). The tool recommends that the following domains are assessed.

• Random sequence generation (selection bias).

• Allocation concealment (selection bias).

• Blinding of participants and personnel (performance bias).

• Blinding of outcome assessment (detection bias).

• Incomplete outcome data (attrition bias).

• Selective reporting (reporting bias).

• Other sources of bias.

The 'Risk of bias' domains were judged at 'low risk,' 'high risk' or 'unclear risk' and individual bias items were described according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). We produced a 'Risk of bias' graph (Figure 2) and summary figure (Figure 3) to illustrate these findings.

2.

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

3.

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

We evaluated the risk of bias separately for performance bias (blinding of participants and personnel) and detection bias (blinding of outcome assessment). We grouped outcomes according to whether they were measured subjectively or objectively when reporting findings in the 'Risk of bias' tables. We evaluated attrition bias (incomplete outcome data) for all outcomes.

For risk of bias assessment, we defined the following endpoints as subjective outcomes.

• Serious adverse events.

• Moderate adverse events.

• Mild adverse events.

• Pain during the first 24 hours (VAS means).

• Pain during the first seven days (VAS means).

• Participant satisfaction (cosmesis).

• Inadequate foreskin removal.

We defined the following endpoint as an objective outcome.

• Operative times.

Measures of treatment effect

We reported dichotomous data as risk ratios (RRs) with 95% CIs. We assessed the dichotomous outcomes based on the number of events and the number of people in the comparison and intervention groups. We reported continuous data as mean differences (MDs) with 95% CIs unless different studies used different measures to assess the same outcome, in which case we expressed data as standardized mean differences (SMDs) with 95% CIs.

Unit of analysis issues

The unit of analysis was the individual participant. We did not identify eligible trials using cross‐over or cluster‐RCT designs.

Dealing with missing data

Missing data were obtained from the study authors in one trial (Kanyago 2013). We performed an intention‐to‐treat (ITT) analyses. Attrition rates, e.g. dropouts, losses to follow‐up and withdrawals were investigated and reasons for missing data were critically appraised.

Assessment of heterogeneity

In the event of substantial unexplained heterogeneity, we reported a narrative description of results.

We assessed heterogeneity (inconsistency) through visual inspection of the forest plots to assess the amount of overlap of CIs and the I² statistic (which quantifies inconsistency across studies) to identify the impact of heterogeneity on the meta‐analysis (Higgins 2002; Higgins 2003). The I² statistic was interpreted as follows.

• 0% to 40%: may not be important.

• 30% to 60%: may indicate moderate heterogeneity.

• 50% to 90%: may indicate substantial heterogeneity.

• 75% to 100%: considerable heterogeneity.

When there was heterogeneity, we examined individual study and subgroup characteristics for possible reasons.

Assessment of reporting biases

Where possible, we assessed study protocols for selective outcome reporting.

Where there were 10 trials that investigated a particular outcome, we used funnel plots to assess small‐study effects. Several explanations can be offered for the asymmetry of a funnel plot, including true heterogeneity of effect with respect to trial size, poor methodologic design (and hence bias of small trials), chance and publication bias.

Data synthesis

As specified in the protocol, we proceeded to conduct meta‐analyses where the trials were methodologically similar and the population, interventions and controls homogeneous enough to combine their data. We synthesized data using a random‐effects meta‐analysis. We performed statistical analyses according to the statistical guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011c). For dichotomous outcomes, we used the Mantel‐Haenszel method; for continuous outcomes, we used the inverse variance method. We used Review Manager 5 software to perform the analyses (Review Manager 2014).

Subgroup analysis and investigation of heterogeneity

We expected the following characteristic to introduce clinical heterogeneity, and, therefore, we carried out a subgroup analysis to investigate interactions.

• Type of circumcision device (crush devices versus ligature devices).

We used the test for subgroup differences in Review Manager 5 to compare subgroup analyses when there was a sufficient number of included studies (Review Manager 2014).

Sensitivity analysis

We performed a sensitivity analysis to assess the influence of the following factors (when applicable) on effect sizes.

• Restricting the analysis by considering risk of bias, by excluding studies at 'high risk' or 'unclear risk.'

• Restricting the analysis to very long or extensive studies to establish the extent to which they dominate the results.

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach to present the overall quality of the evidence for each outcome, which takes into account five criteria related to internal validity (risk of bias, inconsistency, imprecision, publication bias) and external validity (directness of results) (Guyatt 2008). For each comparison, three review authors (MZS, SE, AH) independently rated the certainty of the evidence for each outcome as 'high,' 'moderate,' 'low' or 'very low' using GRADEpro GDT (GRADEpro GDT). We resolved any discrepancies by consensus, or by discussion with a fourth review author (TK). For the comparison: device versus surgical circumcision, we presented a summary of the evidence for the main outcomes in 'Summary of findings' table. This summary provides key information about the best estimate of the magnitude of the effect in relative terms and absolute differences for each relevant comparison of alternative management strategies; the number of participants and studies that address each main outcome; and the rating of the overall confidence in effect estimates for each outcome (Guyatt 2011; Schünemann 2011). If meta‐analysis was not possible, we provided a narrative summary.

We reported the following outcomes in the 'Summary of findings' table.

• Serious adverse events.

• Moderate adverse events.

• Mild adverse events.

• Operative time (minutes).

• Pain during the first 24 hours (VAS means in centimeters).

• Pain during the first seven days (VAS means in centimeters).

• Participant satisfaction (cosmesis).

Results

Description of studies

For a detailed description of the included studies see the Characteristics of included studies table. Studies that did not satisfy our selection criteria are described in the Characteristics of excluded studies table.

Results of the search

We identified 596 records from electronic and supplementary searches and two additional records through reference lists. After removing duplicates, three review authors (MZS and SE, AH) screened 535 titles and abstracts and excluded 507 records. Thus, there were 28 articles in the full‐text screening stage. We excluded nine trials with reasons and one trial is ongoing. We included 18 trials in the final review. The flow of trials identified to be included in the review is summarized in a flow chart (Figure 1).

Included studies

The 18 included trials are described in detail in the Characteristics of included studies table. All included trials were randomized at the individual participant level. There were 10 trials conducted in China (Huo 2015; Jin 2015; Li 2010; Li 2014; Lv 2014; Miao 2015; Pang 2015; Ren 2014; Wang 2014; Wang 2017), three in South Africa (Lagarde 2009; Millard 2014; Shenje 2016), two multi‐country trials in Kenya and Zambia (Sokal 2014) and in Malawi and Zimbabwe (Tshimanga 2016), one trial in Mozambique (Millard 2013), one in Rwanda (Mutabazi 2012), and one in Uganda (Kanyago 2013). Twelve trials stated that the interventions were conducted in hospital settings (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Miao 2015; Mutabazi 2012; Pang 2015; Ren 2014; Wang 2014; Wang 2017), and six trials were conducted in outpatient settings (Lagarde 2009; Millard 2013; Millard 2014; Shenje 2016; Sokal 2014; Tshimanga 2016). Most trials had two arms, except Lv 2014 and Miao 2015, which had three arms comparing two intervention arms both consisting of crush devices to a standard surgical technique.

Participants

The 18 studies included 5246 participants. The included sample sizes ranged from 56 participants (Pang 2015) to 942 (Lv 2014). Nine trials included males older than 18 years (Jin 2015; Lagarde 2009; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Sokal 2014; Tshimanga 2016; Wang 2017), five trials included males older than 12 years (Kanyago 2013; Li 2010; Pang 2015; Shenje 2016; Wang 2014). We considered the two trials that included males older than seven years to be eligible (Miao 2015; Ren 2014); with participants in the Miao 2015 trial having a mean age of 22.5 (SD 5.7) years with ages ranging from nine years to 56 years. Ren 2014 trial participants had a mean age of 24 years ranging from seven years to 56 years. Both trials had mean ages higher than 10 years including wide age ranges. We assumed that these trials likely did not include many participants younger than 10 years. Furthermore, every attempt was made to contact trial authors, we are still awaiting a response. Finally, the age range for two trials was not clearly described, but the mean age was above 26 years (Huo 2015; Li 2014). Twelve trials included participants with genital conditions such as redundant prepuce, phimosis and scarred frenulum (Huo 2015; Jin 2015; Li 2010; Lv 2014; Miao 2015; Millard 2013; Millard 2014; Pang 2015; Ren 2014; Sokal 2014; Wang 2014; Wang 2017). The remaining six trials included healthy male participants (Kanyago 2013; Lagarde 2009; Li 2014; Mutabazi 2012; Shenje 2016; Tshimanga 2016).

Interventions

The types of devices compared to standard surgical procedures can be broadly classified into two categories: 16 trials used crush‐based devices (Huo 2015; Jin 2015; Kanyago 2013; Lagarde 2009; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Millard 2014; Pang 2015; Ren 2014; Shenje 2016; Sokal 2014; Wang 2014; Wang 2017), and two trials used ligature‐based devices (Mutabazi 2012; Tshimanga 2016). Two trials trained doctors and nurses to conduct the procedures (Sokal 2014; Tshimanga 2016), 10 trials trained only doctors to the procedures (Huo 2015; Jin 2015; Kanyago 2013; Lagarde 2009; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Wang 2014), and six trials did not specify the providers (Li 2010; Li 2014; Miao 2015; Pang 2015; Ren 2014; Wang 2017; Wang 2017). Seven trials stated that the healthcare professional conducting procedures were trained or had experience in using the tested circumcision devices (Huo 2015; Lagarde 2009; Lv 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016). One trial clearly stated that the surgeons had no prior experience in using the experimental circumcision device (Kanyago 2013). It was unclear in the remaining 10 trials whether the healthcare professionals had prior experience or had received the relevant training in using the tested circumcision devices (Jin 2015; Li 2010; Li 2014; Miao 2015; Millard 2013; Millard 2014; Pang 2015; Ren 2014; Wang 2014; Wang 2017).

Outcomes

Lagarde 2009 did not clearly report their outcomes. We contacted the authors of this trial to request trial data, with no response. Thus, the outcomes were not included in the analysis in this review. Six trials were written in Chinese; two first‐language Chinese researchers translated these (Huo 2015; Li 2010; Li 2014; Miao 2015; Pang 2015; Ren 2014). Both researchers indicated that the adverse event results were not clearly described. Thus, we extracted only operative time, pain and patient satisfaction from the publications. We attempted to obtain clarity from the contact authors, however, we are yet to receive a response.

None of the participants in the included trials reported serious adverse events, thus no meta‐analysis could be performed. Ten trials including 3370 participants reported moderate adverse events (Jin 2015; Kanyago 2013; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014). One trial did not clearly report moderate adverse events for us to include into the analysis; we contacted the authors to clarify their findings but received no response (Wang 2017). Ten trials including 3370 participants reported mild adverse events (Jin 2015; Kanyago 2013; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014). Wang 2017 did not clearly report mild adverse events for us to include into the analysis; we contacted the authors for clarity but received no response. Twelve trials reported pain during the first 24 hours by VAS using centimeters as the unit of measurement (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Miao 2015; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Wang 2017). However, only nine of these trials reported mean pain scores (Huo 2015; Jin 2015; Kanyago 2013; Li 2014; Li 2010; Miao 2015; Millard 2013; Millard 2014; Mutabazi 2012). These were thus included in the analyses. A further four trials reported VAS scores during the first seven days after circumcision (Kanyago 2013; Lv 2014; Millard 2013; Millard 2014). Two trials reported pain during erection (Mutabazi 2012; Sokal 2014).

Seventeen trials reported operative time (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Millard 2014; Mutabazi 2012; Pang 2015; Ren 2014; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017); of these, 14 trials reported their operative times as a mean value and were included in the analyses (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Mutabazi 2012; Pang 2015; Ren 2014; Sokal 2014; Tshimanga 2016; Wang 2014). Fifteen trials measured the proportion of participant's satisfaction with the circumcision (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Pang 2015; Ren 2014; Shenje 2016; Tshimanga 2016; Wang 2014; Wang 2017).

Excluded studies

We excluded nine studies, see Characteristics of excluded studies table. Five studies had an ineligible study design (Awori 2016; Feldblum 2014; Lei 2015; Liang 2012; Yu 2014), two studies compared two application methods of the same device (Cheng 2012; Feldblum 2016), one study compared two surgical circumcision methods (Decastro 2010), and one study was a retracted article previously published in another journal (Yue 2012).

Studies awaiting classification

There are no studies awaiting classification.

Ongoing trials

We found one ongoing trial that has completed recruiting participants (ChiCTR‐TRC‐13003428; see Characteristics of ongoing studies table).

Risk of bias in included studies

The judgments for the risk of bias of the 18 included studies are summarized in Figure 2 and Figure 3.

Allocation

Random sequence generation

Ten trials reported adequate random sequence generation (Huo 2015; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017). Seven trials did not report clearly on the method for sequence generation and were at unclear risk of bias (Jin 2015; Kanyago 2013; Lagarde 2009; Li 2014; Miao 2015; Pang 2015; Ren 2014). One trial randomized participants according to day/date of clinic visit and was at high risk of bias (Li 2010).

Allocation concealment

Five trials had adequate allocation concealment (Kanyago 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014). Twelve trials had unclear reporting of allocation concealment (Huo 2015; Lagarde 2009; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Pang 2015; Ren 2014; Tshimanga 2016; Wang 2014; Wang 2017). For Jin 2015, the baseline characteristics of participants were not similar across the study groups: there were more participants with phimosis in the surgical group compared to the device group. Therefore, we judged this study at high risk of selection bias.

Blinding

Blinding of participants and personnel

The nature of the surgical intervention precludes the blinding of participants and personnel. Therefore, all studies were at unclear risk of performance bias.

Blinding of outcome assessment

We differentiated between subjective and objective outcomes for which blinding of outcome assessors would be relevant. Subjective outcomes were serious, moderate and mild adverse events; pain and participant satisfaction. We judged all trials at high risk of detection bias for the subjective outcomes. Assessment of operative time was a more objective outcome; despite the lack of blinding with respect to surgical interventions, we judged all trials measuring this outcome at low risk of bias. One trial did not report on operative time (Lagarde 2009).

Incomplete outcome data

Sixteen trials reported low levels of attrition, less than 10% (Huo 2015; Jin 2015; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Millard 2014; Mutabazi 2012; Pang 2015; Ren 2014; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017). In addition where attrition was equal between groups and well described, we judged that domain as low risk of bias. Kanyago 2013 had differential attrition, with 25% losses to follow‐up without reasons in the intervention group and was, therefore, at high risk of bias; authors assumed that participants in the surgical group who did not return had high satisfaction and no minor or major complications. In Lagarde 2009, there was also higher and differential attrition in the intervention group, which we judged at high risk of bias for this domain.

Selective reporting

Six trials had study protocols or registry records (Kanyago 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016). Five of them reported outcomes according to those prespecified and were, therefore, at low risk of bias. Tshimanga 2016 reported pain, which was not specified in the registry record; this resulted in high risk of selective outcome reporting. We did not find protocols for the remaining 12 trials. Two trials seemed as though pain was meant to have been reported but was not (high risk of bias; Pang 2015; Ren 2014). The remaining 10 trials had unclear risk of selective outcome reporting (Huo 2015; Jin 2015; Lagarde 2009; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Wang 2014; Wang 2017).

Other potential sources of bias

One trial had a potentially high risk of bias due to early stopping and the trial not meeting its planned sample size for the primary outcome (power) (Lagarde 2009).

Effects of interventions

See: Table 1

See: Table 1 for circumcision devices versus standard surgical techniques.

Twelve trials reported adverse events (Jin 2015; Kanyago 2013; Lagarde 2009; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017). Results from one trial were not clearly reported and thus not analyzed (Lagarde 2009); we attempted to obtain trial data from the authors. See Table 2 for a description of adverse events from the trials whose data on adverse events could not be used in the analyses (Huo 2015; Lagarde 2009; Li 2010; Li 2014; Miao 2015; Pang 2015; Ren 2014). For a description of the methods and timing for outcome assessment, refer to Appendix 1.

1. Adverse events.

Study ID	Device type	Group	n	Incision edema, n (%)	Hematoma, n (%)	Ecchymosis, n (%)	Infection, n (%)	Reoperation, n (%)	Deformity, n (%)	Wound dehiscence, n (%)	Bleeding, n (%)	Device‐relateda, n (%)
Huo 2015	Crush	DCSD	120	Mild: 80 (66.7%) Moderate: 34 (28.3%) Severe: 6 (5.0%)	2 (1.7%)	25 (20.8%)	3 (2.5%)	—	—	—	—	—
		Surgical	60	Mild: 25 (41.7%) Moderate: 27 (45.0%) Severe: 8 (13.3%)	4 (3.3%)	5 (8.3%)	3 (5.0%)	3 (5.0%)	—	—	—	—
				P = 0.04	P = 0.186	P < 0.01	P = 0.660	P = 0.064	—	—	—	—
Lagarde 2009	Crush	Tara KLamp	35	3 (8.6%)	—	—	5 (14.2%)	—	—	—	1 (2.9%)	7 (20.0%)
		Surgical	34	0 (0%)	—	—	0 (0%)	—	—	—	0 (0%)	N/A
Li 2010	Crush	Shang Ring	402	85 (21.1%)	—	—	3 (0.78%)	—	—	7 (1.7%)	2 (0.5%)	—
		Surgical	322	115 (35.7)	—	—	39 (12.1%)	—	—	5 (1.6%)	40 (12.4%)	—
Li 2014	Crushb	DCSD	129	10 (7.8%)	—	—	1 (0.8%)	—	—	4 (3.1%)	6 (4.7%)	4 (3.1%)
		Surgical	120	5 (4.2%)	—	—	1 (0.8%)	—	—	1 (0.8%)	1 (0.8%)	2 (1.7%)
Miao 2015	Crush/ligaturec	DCSD	92	2 (2.2%)	2 (2.2%)	—	2 (2.2%)	—	—	—	—	—
		Foreskin cerclage	92	9 (9.8%)	1 (1.1%)	—	3 (3.3%)	—	—	—	—	—
		Surgical	92	2 (2.2%)	1 (1.1%)	—	3 (3.3%)	—	—	—	—	—
Pang 2015	Crushd	DCSD	28	1 (3.6%)	—	—	—	—	—	—	—	—
		Surgical	28	6 (21.4%)	—	—	2 (7.1%)	—	3 (10.7%)	—	—	—
Ren 2014	Crush	DCSD	136	Overall complications reported, i.e. bleeding, hematoma, infection, wound dehiscence, wound edema: 9.6% (13/136) in DCSD group and 17.9% (15/84) in the surgical group; P < 0.05.
		Surgical	84

DCSD: disposable circumcision disposable device; n: number of participants; N/A: not applicable.
^aPoor lacing, suture incarceration, device remained in tissue for too long, etc.
^bThe incidence of complications was higher in the DCSD group versus the foreskin cerclage group (P = 0.24).
^cThe incidence of complications was significantly lower in the DCSD group versus the foreskin cerclage group (P < 0.05).
^dThe incidence of complications in the stapler group was significantly lower than the surgical group (P < 0.05).

Primary outcomes

Serious adverse events

Eleven trials reported serious adverse effects (Jin 2015; Kanyago 2013; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017). There was moderate‐certainty evidence that there is probably little or no difference in serious adverse events comparing devices with standard surgical techniques for circumcision. There were no serious adverse events reported in either group (3472 participants, 0 events). We downgraded the evidence one level for methodologic limitations due to potential risk of bias for poor reporting on sequence generation and allocation concealment in several trials, and lack of blinding of the outcome assessors. None of the 11 trials reported serious adverse events, thus we were unable to calculate an effect size estimate for this outcome.

Moderate adverse events

Ten trials reported moderate adverse effects (Jin 2015; Kanyago 2013; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014). There may be slightly more moderate adverse events when devices are used compared to standard surgical techniques (RR 1.31, 95% CI 0.55 to 3.10; I²= 68%; 3370 participants; low‐certainty evidence; Figure 4). This represents eight more moderate adverse events per 1000 people circumcised with a device (95% CI from 12 fewer to 57 more). We downgraded the certainty of the evidence due to serious risk of bias and imprecision.

4.

Forest plot of comparison: 1 Circumcision devices versus standard surgical techniques, outcome: 1.2 Moderate adverse events.

Secondary outcomes

Mild adverse events

Ten trials reported mild adverse events (Jin 2015; Kanyago 2013; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014).

There is uncertainty about the difference in mild adverse events between groups when devices were used compared to standard surgical techniques (RR 1.09, 95% CI 0.44 to 2.72; I² = 91%; 10 trials, 3370 participants; very low‐certainty evidence; Figure 5). We downgraded the certainty of the evidence for serious risk of bias, unexplained heterogeneity and imprecision.

5.

Forest plot of comparison: 1 Circumcision devices versus standard surgical techniques, outcome: 1.3 Mild adverse events.

Operative time

Fourteen trials reported operative time (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Miao 2015; Millard 2013; Mutabazi 2012; Pang 2015; Ren 2014; Sokal 2014; Tshimanga 2016; Wang 2014).

There is probably a clinically relevant 17 minute reduction in operative time when using a device compared to standard surgical techniques (MD –17.26 minutes, 95% CI –19.96 to –14.57; I² = 99%; 14 trials, 4812 participants; moderate‐certainty evidence; Figure 6). We downgraded the certainty of the evidence one level for serious risk of bias. Despite the high statistical inconsistency (I² = 99%), all times indicated a reduction in procedure duration when a device was used compared to the standard surgical technique; therefore, we elected not to downgrade for inconsistency. Furthermore, we considered an operative time less than 10 minutes to be clinically important.

6.

Forest plot of comparison: 1 Circumcision devices versus standard surgical techniques, outcome: 1.4 Operative time (minutes).

Postoperative pain

See Table 3 for narrative reports of pain from each trial.

2. Pain as reported in individual trials.

Study ID	Device type	Pain outcome parameter	Pain during first 24 hours	Pain during the first 7 days	Pain on device removal	Pain during erection
Huo 2015	Crush	VAS (mean, SD)	Mean score: 1.9 (SD 1.3) with device vs 5.2 (SD 1.7) with surgery; P < 0.01	Not reported	Not applicable	Not reported
Jin 2015	Crush	VAS (mean, SD)	Mean intraoperative pain score: 0.8 (SD 0.5) with device vs 2.4 (SD 0.8) with surgery; P < 0.001 Mean 1‐hour postoperative pain scores: 4.0 (SD 0.9) with device vs 5.8 (SD 1.0) with surgery; P < 0.001	Not reported	Not applicable	Not reported
Kanyago 2013	Crush	VAS (median, IQR)	Median pain score: 1/10 (IQR 0–4) with device vs 3/10 (IQR 2–5) with surgery; P = 0.01	Pain‐free at 3 days: 69.9% with device vs 62.7% with surgery; RR 1.11, 95% CI 0.87 to 1.43; P = 0.5. All participants in both groups were pain free at 7 days	Not reported	Not reported
Lagarde 2009	Crush	VAS (mean pain score [0–10])	Not reported	As‐treated analyses results: mean pain score at postoperative day 3: 9.5 with TK vs 6.1 with FG; P = 0.003	Not applicable	Not reported
Li 2010	Crush	VAS (mean, SD)	Mean intraoperative pain score: 0.21 (SD 0.65) with SR vs 3.72 (SD 1.45) with surgery Mean pain score at 24 hours: 1.83 (SD 0.86) with SR vs 5.27 (SD 1.74) with surgery	Not reported	Not applicable	Not reported
Li 2014	Crush	VAS (mean, SD)	Mean intraoperative pain score: 0.81 (SD 0.81) with for DCSD vs 2.42 (SD 1.15) with surgery; P < 0.05 Mean pain score at 24 hours: 1.84 (SD 1.02) with DCSD vs 4.99 (SD 1.36) with surgery; P < 0.05	Not reported	Not applicable	Not reported
Lv 2014	Crush	VAS (mean, SD)	Mean intraoperative pain score: 1.9 (SD 1.3) with DCSD vs 5.8 (SD 2.1) with SR vs 6.2 (2.2) with surgery. Pain scores using DCSD was significantly lower than both SR and surgical circumcision; P < 0.001	Pain scores at 1 week: 2.7 (SD 0.9) with DCSD vs 6.4 (SD 2.0) with SR vs 3.3 (SD 0.8) with surgery; P < 0.001	Not reported for SR. Not applicable for DCSD	Not reported
Miao 2015	Crush	VAS (mean, SD)	Mean intraoperative pain score: 1.37 (SD 0.68) with DCSD vs 1.20 (SD 0.79) with foreskin cerclage vs 3.06 (SD 0.75) with surgery. Pain scores using foreskin cerclage were significantly lower than surgical circumcision; P < 0.05 Mean pain score at 24 hours: 1.85 (SD 0.63) with DCSD vs 3.18 (SD 0.82) with foreskin cerclage vs 1.82 (SD 0.75) with surgery	Not reported	Not applicable	Not reported
Millard 2013	Crush	VAS (mean, SD)	Mean pain score: 3.3 (SD 2.7) with device vs 3.3 (SD 2.4) with conventional surgery; P = 0.91	Mean score at 48 hours: 1.8 (SD 1.6) with device vs 2.5 (SD 1.9) with conventional surgery; P = 0.008	Not applicable	Not reported
Millard 2014	Crush	VAS (mean, SD)	Mean pain score: 4.2 (SD 2.7) with device vs 3.1 (SD 2.4) with conventional surgery; P = 0.01	Mean score reported at 48 hours: 0.7 (SD 1.6) with device vs 1.2 (SD 2) with conventional surgery; P = 0.04	Not applicable	Not reported
Mutabazi 2012	Ligature	VAS (mean, SD)	During placement: mean pain levels: 0.8 (SD 1.2) with device vs 2.4 (SD 0.9) with surgery Mean pain score at 1 hour: 5.6 (SD 1.8) with device vs 3.8 (SD 2.1) with surgery Mean pain score at 2 hours: 4.1 (SD 1.9) with device vs 2.3 (SD 1.6) with surgery Mean pain score at 3 hours: 2.8 (SD 1.8) with device vs 1.4 (SD 1.5) with surgery Mean pain score at 4 hours: 1.9 (SD 1.5) with device vs 0.9 (SD 1.3) with surgery Mean pain score at 8 hours: 1.1 (SD 1.6) with device vs 0.7 (SD 1.2) with surgery Mean pain score at 16 hours: 0.3 (SD 0.8) with device vs 0.9 (SD 1.5) with surgery Statistical difference between groups were not reported	Not reported	Mean pain levels were 4.7 (SD 1.9) during device removal vs 1.4 (SD 1.6) 1 hour immediately after removal. Statistical significance not reported	Mean pain levels during erection at day 3: 2.5 (SD 1.8) with device vs 5.2 (SD 2.2) with surgical methods. Mean pain score during erection at day 9: 1.9 (SD 1.5) with device. The pain scores during erection for the surgical group was significant higher compared with the PrePex group (P<0.0001).
Pang 2015	Crush	Not reported	Not reported	Not reported	Not applicable	Not reported
Ren 2014	Crush	Not reported	Not reported	Not reported	Not applicable	Not reported
Shenje 2016	Crush	VAS (Median, IQR)	Median intraoperative pain score: 1 (IQR 0.5–2) with device vs 1 (IQR 0 to 2) with surgery Median immediate postoperative pain score: 1 (IQR 0.5–2) with device vs 1 (IQR 0–5) with surgery	Not reported	Not applicable	Not reported
Sokal 2014	Crush	VAS (mean, SD) and VAS (median, IQR)	Immediately after procedure and 1 hour: Kenya: pain scores: 0.2 with device vs 1.2 with surgery (P < 0.001), but was the same 1 hour after surgery (combined mean 3.8 [SD 1.8]). Zambia: pain scores: 0.6 with device vs 0.2 with surgery (P = 0.046), but were similar 1 hour after surgery (combined mean 3.4 [SD 1.9])	At 2 days: similar pain scores in both groups with median VAS scores of 1.0	Not reported	SR had worse pain during erections: 3.5 (SD 1.9) with SR vs 2.3 (SD 1.7) with surgery (P < 0.001). This required early removal in 2 cases
Tshimanga 2016	Ligature	VAS (categories 0, 2, 4, 6, 8) with percentages of participants experiencing pain at these levels	Periprocedurally: all participants in device groups had VAS score of 0, while over 90% in surgical group had a VAS score 4–6. After procedure: device group had VAS score of 0 up to 3 hours while surgical group had VAS score 0–2. No statistical differences were reported	111/160 (69%) participants experienced pain (VAS 0–2) at end of the first week with the device vs 14/80 (17.5%) participants with surgery	Transient pain (VAS 2–6) reported in 93.6% of device group. Pain after removal reported in 66%	Pain during erection (VAS 2–4): 81/160 (51%) participants with device vs 57/80 (71%) participants with surgery
Wang 2014	Crush	Categorically: minor (no/slight pain), moderate (tolerable pain) and severe (intolerable pain requiring medication or intervention with acetaminophen) on days 1, 3, 7 and 14; with numbers per category	53 minor pain, 4 moderate pain and 1 severe pain with devices vs 54 minor pain, 6 moderate pain and 0 severe pain	Day 3	—	—
Wang 2017	Crush	VAS (the units of measurement, i.e. mean/median, were not specified)	Pain scores: 6 hours after operation: 5.13 (SD 1.14) with disposable stitching device vs 7.16 (SD 1.23) with surgery; P < 0.0001 12 hours after operation: 3.16 (SD 1.13) with disposable stitching device vs 5.33 (SD 1.25) with surgery; P < 0.0001 24 hours after operation: 1.85 (SD 1.14) with disposable stitching device vs 3.58 (SD 1.26) with surgery; P < 0.0001 Postoperative VAS score improvements in the device group were significantly higher than those of the surgery group at the time points specified	Not reported	Not applicable	Not reported

CI: confidence interval; DCSD: disposable circumcision disposable device; FG: forceps guided; IQR: interquartile range; n: number; RR: risk ratio; SD: standard deviation; SR: Shang Ring; TK: Tara KLamp; VAS: visual analog scale

Pain during the first 24 hours

Nine trials reported pain during the first 24 hours after circumcision (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Miao 2015; Millard 2013; Millard 2014; Mutabazi 2012); all used a VAS to evaluate pain severity.

There may be less pain if circumcision devices are used compared to standard surgical techniques for postoperative pain during the first 24 hours (MD –1.30 cm, 95% CI –2.37 to –0.22; I² = 99%; 3022 participants; low‐certainty evidence; Figure 7). We downgraded the certainty of the evidence for serious methodologic limitations and serious unexplained heterogeneity. Lv 2014 reported this outcome but we did not include the data in this analysis because they measured pain intraoperatively.

7.

Forest plot of comparison: 1 Circumcision devices versus standard surgical techniques, outcome: 1.5 Pain during the first 24 hours (VAS means).

Pain during the first seven days

Four trials reported pain during the first seven days after circumcision by mean and SDs using a VAS (Kanyago 2013; Lv 2014; Millard 2013; Millard 2014).

There may be little or no difference in postoperative pain during the first seven days when compared with standard surgical techniques (MD 0.11 cm, 95% CI –0.89 to 1.11; I² = 94%; 1430 participants; low‐certainty evidence). We downgraded the certainty of the evidence for serious methodologic limitations, and serious unexplained heterogeneity.

Two trials did not report outcomes during this period (Jin 2015; Mutabazi 2012), and five others did not use VAS scores or used other summary statistics (Shenje 2016; Sokal 2014; Tshimanga 2016; Wang 2014; Wang 2017).

Pain during erection

Three trials reported pain during erection (Mutabazi 2012; Sokal 2014; Tshimanga 2016), but only two trials could be included in the meta‐analysis (Mutabazi 2012; Sokal 2014); Tshimanga 2016 reported the number and percentages of participants experiencing a given pain level at categories of the VAS score (0, 2, 4, 6, 8).

There is uncertainty about pain during erection in those who were circumcised with a device compared to standard surgical techniques (MD –0.74 cm, 95% CI –4.57 to 3.08; I² = 99%; 2 trials, 626 participants; very low‐certainty evidence). We downgraded the certainty of the evidence for study limitations (lack of blinding), inconsistency and imprecision (wide CIs due to few studies reporting this outcome, which may be due to selective outcome reporting).

Pain during device removal

This   outcome only applies to participants where surgeons used circumcision devices. Two trials reported pain during device removal (Mutabazi 2012; Tshimanga 2016). One reported moderate transient pain levels (VAS score: 2 cm to 6 cm) in 93.6% of the participants (Tshimanga 2016). One recorded a mean VAS of 4.7 cm during removal (Mutabazi 2012).

Participant satisfaction (cosmesis)

Participant satisfaction (cosmesis) was a dichotomous outcome. We extracted participants' satisfaction with their circumcision using a device or standard surgery. Trials measured this using three different approaches. The first approach was a two‐point scale, participants indicated that they were 'satisfied' or 'unsatisfied' with their procedure. The second was a three‐point scale, participants indicated that they were 'very satisfied,' 'satisfied' or 'not satisfied.' We combined 'very satisfied' and 'satisfied.' The third was a five‐point scale, participants indicated that they were 'extremely dissatisfied,' 'dissatisfied,' 'neutral,' 'satisfied' or 'extremely satisfied.' We combined responses 'satisfied' and 'extremely satisfied.' Higher rates reflected greater satisfaction. See Table 4 with narrative reports of participant satisfaction from each trial.

3. Participant satisfaction (cosmesis).

Study ID	Country	Comparison	Device type	Number of participants	Findings
Huo 2015	China	DCSD vs dorsal slit circumcision	Crush	DCSD: 120; dorsal slit 0	Satisfied with the cosmetic penile appearance: 77.5% (93/120) participants with DCSD vs 51.7% (31/60) with dorsal slit Moderately satisfied with the cosmetic penile appearance: 17.5% (21/120) with DCSD vs 33.3% (20/60) with dorsal slit; P = 0.01 Not satisfied with the cosmetic penile appearance: 5.0% (6/120) with DCSD vs 15.0% (9/60) with dorsal slit
Jin 2015	China	Circular stapler vs dorsal slit circumcision with electric scalpel	Crush	Total: 879; stapler: 441; dorsal slit: 438	Satisfied with the cosmesis and results of procedure: 91.2% (402/441) with stapler vs 90.2% (395/438) with dorsal slit; P = 0.619
Kanyago 2013	Uganda	SR vs FG circumcision	Crush	Total: 138; SR: 73; FG: 65	High patient satisfaction: 74.0% (54/73) with SR vs 60.0% (39/65) with FG; RR 1.38, 95% CI 0.94 to 2.02; P = 0.10
Lagarde 2009	South Africa	TK vs conventional FG circumcision	Crush	As‐treated analyses, total: 29; TK: 19; FG: 19	Satisfied with penile appearance: 84.2% (16/19) with TK vs 100% (19/19) with FG; P = 0.056
Li 2010	China	SR vs dorsal slit circumcision	Crush	SR: 402; dorsal slit: 322	Satisfied with cosmetic penile appearance: 99.5% (400/402) with SR vs 69.6% (224/322) with dorsal slit
Li 2014	China	DCSD vs conventional circumcision	Crush	DCSD: 129; conventional: 120	Satisfied with cosmetic penile appearance: 98.4% (127/129) with DCSD vs 90.8% (109/120) with conventional; P = 0.01
Lv 2014	China	DCSD vs SR vs electrosurgical knife and suture circumcision	Crush	Total: 942; DCSD: 314; SR: 314; electrosurgical knife and suture: 214	Extremely satisfied overall: 30.9% (97/314) with DCSD vs 18.5% (58/314) with SR vs 14.3% (45/314) with electrosurgical knife and suture; P < 0.005 Extremely satisfied with penile appearance: 29.3% (92/314) with DCSD vs 24.8% (78/314) with SR vs 10.5% (33/314) with electrosurgical knife and suture; P < 0.05
Miao 2015	China	DCSD vs foreskin cerclage vs conventional circumcision	Crush	DCSD: 92; foreskin cerclage: 92; conventional: 92	Not reported
Millard 2013	Mozambique	Gomco/tissue adhesive vs dorsal slit circumcision	Crush	Total: 200: Gomco/tissue adhesive: 100; dorsal slit: 100	Very satisfied with cosmetic results after procedure: 58.9% (56/95) with Gomco/adhesive vs 70.7% (70/99) with dorsal slit; P = 0.09 Satisfied with cosmetic results after procedure: 41.1% (39/95) with Gomco/adhesive vs 29.3% (29/99) with dorsal slit; P = 0.09
Millard 2014	South Africa	Unicirc and tissue adhesive vs dorsal slit circumcision	Crush	Total: 150; Unicirc/adhesive: 100; dorsal slit: 50	Very satisfied with cosmetic results after procedure: 79.6% (78/98) with Unicirc/adhesive vs 80.0% (40/50) with dorsal slit Satisfied with cosmetic results after procedure: 15.3% (15/98) with Unicirc/adhesive vs 16.0% (8/50) with dorsal slit Not satisfied with cosmetic results after procedure: 5.1% (5/98) with Unicirc/adhesive vs 4% (2/50) with dorsal slit; P = 0.9
Mutabazi 2012	Rwanda	PrePex vs dorsal slit circumcision	Ligature	Total: 226; PrePex: 150; dorsal slit: 76	Satisfied with cosmetic results after procedure: 99.0% (91/92) with PrePex vs 100% (55/55) with dorsal slit; no significant difference
Pang 2015	China	DCSD vs dorsal slit circumcision	Crush	DCSD: 28; dorsal slit: 28	Satisfied with the cosmetic penile appearance: 96.4% (27/28) with DCSD vs 71.4% (20/28) with dorsal slit
Ren 2014	China	DCSD vs dorsal slit circumcision	Crush	DCSD: 136; dorsal slit: 84	Satisfied with cosmetic penile appearance: 97.8% (133/136) with DCSD vs 88.1% (74/84) with dorsal slit
Shenje 2016	South Africa	Unicirc vs FG/dorsal slit circumcision	Crush	Total: 75; Unicirc: 50; FG/dorsal slit: 25	Very satisfied with result: 97.7% (43/44) with Unicirc vs 95.7% (22/23) with FG/dorsal slit Satisfied with result: 2.3% (1/44) with Unicirc vs 4.3% (1/23) with FG/dorsal slit No significant difference in satisfaction
Sokal 2014	Kenya and Zambia	SR vs FG/dorsal slit circumcision	Crush	Total: 400; SR: 200; FG/dorsal slit: 200	Very satisfied with cosmetic appearance at 60 days in Kenya: 95.7% with SR vs 85.9% with FG/dorsal slit; P = 0.02 Very satisfied with cosmetic appearance at 60 days in Zambia: 96.8% with SR vs 71.3% with FG/dorsal slit; P < 0.01
Tshimanga 2016	Zimbabwe	PrePex vs FG circumcision	Ligature	Total 240; PrePex: 160; FG: 80	'Very' or 'extremely' satisfied with procedure and cosmetic results at 2‐week postprocedure interview: 71.2% (74/104) with PrePex vs 76.7% (23/30) with FG; P > 0.05 'Very' or 'extremely' satisfied with procedure and cosmetic results at 90‐day postprocedure interview: 88.0% (97/110) with PrePex vs 94.1% (48/51) with FG; P > 0.05 Satisfied with procedure and cosmetic results at 90‐day postprocedure interview: 99.1% (109/110) with PrePex vs 98.0% (50/51) with FG; P > 0.05
Wang 2014	China	DCSD vs standard surgical sleeve dissection	Crush	Total: 120; DCSD: 60; standard sleeve: 60	Full satisfaction with penile appearance: 96.6% (56/58) with DCSD vs 95% (57/60) with standard sleeve; P > 0.05 Moderately satisfied with penile appearance: 3.4% (2/58) with DCSD vs 5.0% (3/60) with standard sleeve; P > 0.05
Wang 2017	China	DCSD vs traditional surgical circumcision	Crush	Total: 102; DCSD: 51; traditional surgery: 51	Satisfaction with the incision aesthetic: 98.0% (50/51) with DCSD vs 76.5% (39/51) with traditional surgery; P < 0.0001

CI: confidence interval; DCSD: disposable circumcision disposable device; FG: forceps guided; SR: Shang Ring; TK: Tara KLamp; RR: risk ratio.

Fifteen trials reported participant satisfaction (Huo 2015; Jin 2015; Kanyago 2013; Li 2010; Li 2014; Lv 2014; Millard 2013; Millard 2014; Mutabazi 2012; Pang 2015; Ren 2014; Shenje 2016; Tshimanga 2016; Wang 2014; Wang 2017).

Participants may slightly prefer circumcision devices compared to standard surgical techniques for participant satisfaction (RR 1.19, 95% CI 1.04 to 1.37; I² = 97%; 4501 participants; low‐certainty evidence; Figure 8). The absolute effect is 143 more participants per 1000 population were satisfied with circumcision devices (95% CI from 30 more to 278 more). We downgraded the certainty of the evidence for serious risk of bias and unexplained inconsistency.

8.

Forest plot of comparison: 1 Circumcision devices versus standard surgical techniques, outcome: 1.8 Participant satisfaction.

Inadequate removal of foreskin

None of the studies reported inadequate removal of the foreskin.

Subgroup analyses

We attempted to do preplanned subgroup analyses for the use crush‐based versus ligature‐based devices for all outcomes.

Serious adverse events

No effect size could be calculated for either group and there was therefore no suggestion of a subgroup effect (Analysis 1.1).

1.1. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 1: Serious adverse events

Moderate adverse events

Subgroup analyses strongly favored ligature‐based devices (RR 0.17, 95% CI 0.02 to 1.60) over crush‐based devices (RR 1.60, 95% CI 0.67 to 3.80) (Analysis 1.2). The test for interaction did not meet statistical significance (P = 0.07).

1.2. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 2: Moderate adverse events

Mild adverse events

Effects size for both subgroups were similar (Analysis 1.3). 

1.3. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 3: Mild adverse events

Operative time

Operative time favored crush‐based devices (MD –18.31 minutes, 95% CI –21.05 to –15.57) over ligature‐based devices (MD –11.07 minutes, 95% CI –13.52 to ‐8.62) (Analysis 1.4). The test for interaction was significant (P = 0.0001) supporting the notion that this may not be a chance finding. 

1.4. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 4: Operative time (minutes)

Pain 

Pain during the first 24 hours

Pain during the first 24 hours after circumcision favored crush‐based devices (MD –1.38, 95% CI –2.56 to –0.20) over ligature‐based devices (MD –0.60, 95% CI –0.96 to –0.24) (Analysis 1.5), but the test for interaction did not meet statistical significance (P = 0.21).

1.5. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 5: Pain during the first 24 hours (VAS means)

Pain during the first seven days

We were unable to conduct any subgroup analyses. 

Pain during erection

We were unable to conduct any subgroup analyses. 

Participant satisfaction

Effects size for both subgroups were similar (Analysis 1.8).

1.8. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 8: Participant satisfaction

Sensitivity analyses

We were unable to conduct any predefined sensitivity analyses since all studies were similar with regard to overall risk of bias and length of follow‐up.

Discussion

Summary of main results

See Table 1.

Male medical circumcision conducted in health facilities by trained healthcare providers is considered a relatively safe procedure that has been practiced for many years (WHO 2012). We conducted a comprehensive search for trials of circumcision devices compared to standard surgical procedures for boys older than 10 years and men to evaluate the differences in adverse events, pain, operative time and participant satisfaction. We included 18 trials, with 5246 participants, the majority were older than 10 years. The trials were conducted in lower‐ and middle‐income countries including China (one trial), South Africa (three trials), Rwanda (one trial), Kenya (one trial), Zimbabwe (one trial), Malawi (one trial), Mozambique (one trial) and Zambia (one trial). The devices we evaluated fell into one of two types, either the crush device or ligature devices (WHO 2012).

In our review, we found that when a circumcision device is used compared to a standard surgical technique, there are probably no more serious adverse events such as hospitalization or permanent damage (moderate‐certainty evidence). There may be slightly more moderate adverse events that require an intervention such as suturing or antibiotics (low‐certainty evidence). We are uncertain whether there is a difference in mild adverse events that require little intervention such as minor bleeding (very low‐certainty evidence). The certainty of the evidence for adverse events was downgraded due to low event rates and, therefore, imprecision and possible outcome assessment bias.

From the included studies, the mean duration for the standard surgical procedure was 24 minutes (ranging from 15 minutes to 31 minutes). The procedure is reduced by about 17 minutes (95% CI –20 to –15; moderate‐certainty evidence) when a device is used rather than the standard surgical method. For the outcome pain, there may be less pain postoperatively within the first 24 hours when a circumcision device was used (low‐certainty evidence). Furthermore, there may be little or no difference in postoperative pain during the first seven days when compared with standard surgical techniques (very low‐certainty evidence). Participant satisfaction may be slightly greater when a circumcision device is used compared to standard surgical techniques (low‐certainty evidence).

Overall completeness and applicability of evidence

In our review process, we identified 18 trials addressing the review question. Most trials used crush circumcision devices rather than ligature devices. It is of note that the trials were conducted exclusively in China and countries in sub‐Saharan Africa, which may affect the applicability of our findings to other high‐income settings. However, there is no reason why these devices would work differently in such a setting, except that the health system and training of professionals who perform these procedures may differ from resource‐constrained settings. Low‐ to middle‐income countries generally have fewer specialist doctors trained to perform surgical circumcisions. All the included trials comprised medically trained doctors with varying levels of experience performing the device‐based circumcisions. Two trials also included nurses who carried out the device‐based procedures, suggesting that this task could potentially be shifted to nurses thus saving on costs in the long term. Insights into the level of health practitioner required for these procedures and feasibility of task shifting or sharing would require further evaluation in trials/VMMC programs where this is being initiated.

We attempted to obtain clarity from the contact authors. Consequently, we summarized the reported adverse events in Table 2.

Another area for research not covered by this review would be circumcisions conducted in line with cultural norms in which traditional healers may perform the procedure. Therefore, it is not clear how these crush and ligature devices would be implemented outside the health system, within more traditional settings. This was outside of the scope of the review, which considered procedures done by health professionals, but may have important implications for implementation across other settings.

Our main primary and secondary outcomes were adverse events. We extracted data from 11 trials for adverse events. Six included studies were reported in Chinese and required translation (Huo 2015; Li 2010; Li 2014; Miao 2015; Pang 2015; Ren 2014), and we were unable to extract adverse event data from these trials. Further, one South African trial did not report adverse events in enough detail for us to extract data (Lagarde 2009). Overall, we found that there were no serious adverse events reported in any of the trials of 5246 participants. There were also very low rates of moderate and minor adverse events reported. In many respects this is reassuring, as these adverse outcomes may not be common. However, most trials were powered to consider the outcome operative time and, therefore, would have been underpowered to answer these clinically relevant questions about harms. The result was that the certainty of our evidence ranged from moderate to very low and further suitably powered trials would provide insight into adverse events rates.

The included studies had differing eligibility criteria, as some included participants with genital conditions, whereas others only included healthy participants. There were also two studies that included males under 10 years old (Miao 2015; Ren 2014). The circumcision device mechanisms and surgical procedures also differed across studies. Also, device‐based techniques for circumcisions may not be appropriate in men with certain penile anatomic abnormalities such as a tight phimosis. Postoperative wound/infection control and pain management were not clearly described, which may have skewed the results for these outcomes. Furthermore, the adverse events were generally similarly reported, except for the six Chinese language studies. Continuous outcomes such as pain and operative time were reported as medians instead of means in some studies making the data difficult to incorporate in meta‐analyses.

We did not compare different circumcision devices, as this was out of the scope of this review. Furthermore, we did not include an assessment of resource use, costs or cost‐effectiveness. Economic data are important for decision‐makers to guide decisions on whether circumcision devices work better but they also need to decide on whether approving new methods may lead to a more efficient use of resources. This is especially important in resource‐limited settings where the included trials of this review were conducted; and the WHO has promoted the use of circumcision devices as a means to reduce costs (WHO 2013a). There is a knowledge gap and a systematic review is needed to assess resources and costs such as the device itself, delivery methods, product registration, introduction and staff training.

Quality of the evidence

We graded the certainty of the evidence base using the GRADE approach. We found that the level of evidence ranged from very low to moderate. The following reasons for downgrading are provided.

• Study limitations: selection bias was frequently unclearly reported in most studies, raising concern as to whether participants with genital conditions or poorer baseline values may have been allocated to the surgical group. Furthermore, performance bias was a concern as studies were either unclear or did not blind participants and personnel. In particular, participants may have been treated differently during the trials. Personnel may have provided better pain management during and after procedures, or postoperative hygiene care may have been emphasized in one group more than the other. Similarly, detection bias was a concern because it was unclear whether subjective outcome assessors were blinded. Hence, we judged the certainty of the evidence to be low or very low for the subjective outcomes. Naturally, we acknowledge the difficulty in blinding personnel conducting the procedures, as well as participants and outcome assessors in such trials.

• Inconsistency: there was considerable unexplained heterogeneity prompting us to downgrade the certainty of the evidence for serious inconsistency. We made an exception for 'operative time,' which was not downgraded despite statistical heterogeneity because there was a consistent reduction in time of at least 10 minutes in the intervention group compared to the standard surgical group.

• Imprecision: we downgraded analyses with wide CIs, including appreciable benefit, and harm with low numbers of events in each arm were downgraded.

• Publication bias: there was no reason to downgrade for any outcomes when assessing the potential for publication bias using the funnel plots.

• Selective reporting bias: trials with a protocol registration were evaluated for selective reporting.

Potential biases in the review process

We adhered to rigorous Cochrane standards to minimize bias in the process of conducting and reporting this review. There is a broad range of names given to circumcision devices in this field, which could lead to overlooked trials during our search. Special effort was made to conduct an exhaustive search which included hand‐searching of reference lists and conference abstract proceedings. Additionally, our search may not have identified 'negative' studies, and articles written in the Chinese language. A further limitation of this review was the inability to search Embase during the second search in May 2018 as the license had expired. Furthermore, CINAHL was not searched, although this was planned in the protocol. Last, the WHO has classified these devices into three categories; clamp, ligature and elastic collar compression. When writing the protocol, this third category of devices was categorized as a ligature‐based device, and thus we did not incorporate this as a search term in our search strategy. Furthermore, in August 2019, Professor Millard noted in an email correspondence that the PrePex circumcision device used in two trials (Mutabazi 2012; Tshimanga 2016) had been withdrawn (Millard 2019). This may necessitate a potential re‐evaluation of our findings.

There was variable reporting of pain between studies and it is likely that there was different intra‐ and postoperative pain management between intervention groups of trials and between trials. Furthermore, not all studies evaluated pain at the same time points and time points chosen in the current review evaluating pain as a quantitative outcome, allow for comparisons between devices and standard surgical circumcision (WHO 2013a). For circumcision devices, we reported pain during device placement and removal, as well as in situ use, in the narrative additional tables section of this review (Table 3). Also, pain protocols varied between the studies. Not all studies reported on how pain was managed in the perioperative period and on the discharge of participants. These factors potentially lead to considerable heterogeneity in these outcomes. Overall, using the GRADE approach, we found that the pain evidence was of very low certainty due to imprecision and study limitations and further trials are likely to provide insight.

Agreements and disagreements with other studies or reviews

This systematic review included only RCTs. It applied rigorous Cochrane methodology with GRADE evaluation to rate the certainty of the evidence. We compared a wide range of circumcision device types including in situ and disposable devices with standard surgical techniques. The search identified four systematic reviews and meta‐analyses (Cao 2015; Fan 2016; Huang 2017; Huo 2017). Findings of these reviews can be found in Appendix 3. Of these four systematic reviews, two included network meta‐analysis (Fan 2016; Huang 2017). Unlike our study, these four reviews did not combine all devices when comparing against standard surgical techniques. Furthermore, they did not include a certainty of evidence rating but did highlight the methodological limitations of their included studies. Three systematic reviews assessed adverse events (Cao 2015; Huang 2017; Huo 2017), but did not classify them according to the WHO's Framework for Clinical Evaluation of Devices for Adult Male Circumcision, which provides standardized definitions for grading adverse events as mild, moderate or severe (WHO 2012; WHO 2013a). Three reviews were not in agreement with our results and they reported that participants were less likely to have adverse events when circumcised with a device compared to standard surgical techniques. However, Huang 2017 compared disposable circumcision suture device (DCSD) and SR devices separately to surgical techniques and reported that participants in the SR group had higher odds of developing adverse events than those in the standard surgical technique group.

Three systematic reviews reported postoperative pain (Fan 2016; Huang 2017; Huo 2017). One review did not specify when pain was measured (Fan 2016). The other two reviews measured pain within 24 hours (Huang 2017; Huo 2017). Two reviews reported that participants circumcised with devices experienced less pain (Fan 2016; Huo 2017). Similar to our findings, they found statistically significant differences while Huang 2017 reported that participants experienced pain similarly across the three groups. However, our certainty of evidence for pain was low. Participants may have less pain from a circumcision device within 24 hours and during the first seven days there may be little or no difference between the two groups.

Shorter operative time with circumcision devices was a consistent finding across all reviews and meta‐analyses, including ours (Cao 2015; Fan 2016; Huang 2017; Huo 2017). Notably, reviews by Cao 2015; Huang 2017; and Huo 2017 reported a decrease in SMDs and MDs of more than 15 minutes. This may be related to standardized procedures making devices such as SR and PrePex easy to use. Four reviews measured participant satisfaction for penile appearance (Cao 2015; Fan 2016; Huang 2017; Huo 2017). Similarly to our results, these reviews indicated that participants in the device groups were more satisfied than those in the standard surgical groups.

Circumcision devices are minimally invasive and do not require cautery or suturing for wound closure compared to standard surgical circumcision (Barone 2014). With DCSD, there is simultaneous mechanical cutting and suturing making the procedure simple and quick (Huo 2017). Regarding the other outcomes evaluated, differences between our review and the four reviews (Cao 2015; Fan 2016; Huang 2017; Huo 2017) may be due to heterogeneous study populations, differences in how devices were categorized in our study (crush versus ligature) as mechanisms of action, need for injectable or topical anesthesia, device placement and removal, time in situ and wound closure techniques may differ between devices within these generic categories. There were differences in reporting of outcomes such as follow‐up times, pain scores, protocols for pain management, participant satisfaction rating and adverse events or complications (Fan 2016). Also, qualifications of staff performing both device‐based and surgical circumcisions were not the same across studies included in the current review (specialist surgeons/urologists versus general practitioners versus medical officers and nurses). This may influence operative outcomes due to differences in surgical skill and expertise among these types of healthcare providers.

Device‐based circumcision procedures in comparison to standard surgical techniques are consistently reported to be significantly shorter to conduct, and easier to learn and execute, thus, minimizing surgical skills and requirements such as injected anesthesia and suturing. This allows for task‐shifting to other mid‐level cadres of staff (nurses and medical officers) and has the potential for rapid scale‐up of VMMC programs for HIV prevention in resource‐constrained settings (Barone 2014; Ridzon 2016). However, the use of device‐based techniques, does not eliminate the need for surgical VMMC services as some patients may not be eligible for a device or would prefer a standard surgical circumcision. Abnormal foreskin anatomy and complications related to device use such as slippage or displacement, although rare, may occur, and require surgical intervention. Thus, healthcare providers involved in VMMC services should be aware of the limitations in their skills and expertise and know when to refer to a more qualified clinician should a complication occur (WHO 2013b). In summary, findings from our systematic review support findings from other reviews related to shorter operative time with use of circumcision devices. In comparison to the other reviews described, our review compared a wide range of circumcision devices with standard surgical techniques, used an up‐to‐date, comprehensive, reproducible search strategy and a formal assessment of the certainty of the evidence.

Authors' conclusions

Implications for practice.

We found no differences in serious adverse events when circumcision was performed using a device, and there may be slightly more moderate adverse effects, compared to standard surgical techniques. Importantly, circumcision devices probably reduce operative time by around 17 minutes. Patients may feel less postoperative pain within the first 24 hours and may slightly prefer the use of a device rather than standard surgical procedures. Therefore, circumcision devices are an option for use in clinical practice, and, as they represent a simpler procedure, may enable task‐shifting to different healthcare workers. However, the results of our systematic review should be considered in conjunction with other contextual factors such as cost; patient preferences and values; and access to trained, skilled healthcare workers and health care in some settings.

Implications for research.

Considering our findings, we have highlighted the following research priorities.

• The certainty of the evidence was low or very low for five of the seven outcomes. Thus, there is uncertainty for most of our findings. Future trials should be better powered for pain outcomes and should describe or improve (or both) their sequence generation and allocation concealment methods. Blinding of participants and personnel may not be feasible in surgical trials, thus performance bias remains an issue to consider. However, future trials need to find ways to blind outcome assessors.

• We found that not all trials reported adverse events in a standardized way as per the World Health Organization Framework for Clinical Evaluation of Devices for Adult Male Circumcision making it challenging for review authors to classify these outcomes appropriately (WHO 2012).

• There is a need to evaluate outcomes in routine healthcare settings, explore other issues related to patient acceptability such as discomfort and odor for in situ devices, period of sexual abstinence following device placement, and how in situ devices affect occupational and daily living activities (WHO 2013a; WHO 2013b).

• Feasibility and program costs associated with the use of circumcision devices need further evaluation.

History

Protocol first published: Issue 6, 2016
Review first published: Issue 3, 2021

Notes

We based parts of the Methods section and Appendix 1 of this Cochrane review on a standard template developed by the Cochrane Metabolic and Endocrine Disorders Group, which the Cochrane Urology Group modified and adapted for use.

Appendices

Appendix 1. Definitions and timing of outcome measurement

Outcome	Definition	Timing
Serious adverse events	Requirement of transfusion, hospitalization or resultant permanent damage, e.g. penile injury	Within intraoperative and early postoperative period (30 days)
Moderate adverse events	Requirement of active treatment, e.g. suturing, antibiotics, surgical hemostasis	Within intraoperative and early postoperative period (30 days)
Mild adverse events	Requirement of little or no intervention, e.g. slight wound disruption, minor bleeding	Within intraoperative and early postoperative period (30 days)
Operative time	Time from first clamp on foreskin until dressing placed measured in minutes	Intraoperative
Pain assessment	Visual analog scale (0–10)	Within intraoperative and early postoperative period (30 days)
Participant satisfaction (cosmesis)	Defined as satisfactory or unsatisfactory to participant	Early postoperative period (30 days)
Inadequate removal of foreskin	Assess for clearly uncovered glans with an adequately covered penile shaft and a healed paracoronal wound (Al‐Rahman 1999)	Within intraoperative and early postoperative period (30 days)

Appendix 2. Electronic search strategies

The Cochrane Library

#1 circumcis* OR [mh "Circumcision, male"] #2 Device* OR [mh "Surgical Instruments"] OR Clamp* #3 Gomco #4 Mogen #5 Plastibell* #6 Accucirc #7 Alisklamp #8 Ali’s clamp #9 Ali’s Klamp #10 Ismail Clamp #11 PrePex #12 Shang ring #13 Tara Klamp #14 Unicirc #15 Smartclamp #16 Smartklamp #17 infant* OR neonat* OR [mh "infant, newborn"] #18 #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 #19 #1 AND #2 #20 #1 AND #18 #21 (#19 OR #20) NOT #17

MEDLINE (PubMed)

#1 "Circumcision, male" [Mesh Terms] OR circumcis* #2 Device* OR "Surgical Instruments" [Mesh terms] OR Clamp* #3 Gomco #4 Mogen #5 Plastibell* #6 Accucirc #7 Alisklamp #8 Ali’s clamp #9 Ali’s Klamp #10 Ismail Clamp #11 PrePex #12 Shang ring #13 Tara Klamp #14 Unicirc #15 Smartclamp #16 Smartklamp #17 Neonat* OR "infant, newborn"[Mesh Terms] OR infant* #18 #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 #19 #1 AND #2 #20 #1 AND #18 #21 (#19 OR #20) NOT #17

Embase

#1 'circumcision'/exp OR circumcis*:ti,ab #2 Device*:ti,ab OR 'clamp'/exp OR Clamp*:ti,ab #3 Gomco:ab,ti #4 Mogen:ab,ti #5 'plastibell'/exp #6 Accucirc:ab,ti #7 Alisklamp:ab,ti #8 (Ali* next/1 clamp):ab,ti #9 (Ali* next/1 Klamp):ab,ti #10 (Ismail next/1 Clamp):ab,ti #11 PrePex:ab,ti #12 (Shang next/1 ring):ab,ti #13 (Tara next/1 Klamp):ab,ti #14 Unicirc:ab,ti #15 Smartclamp:ab,ti #16 Smartklamp:ab,ti #17 #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 #18 'circumcision device'/exp #19 'newborn period'/exp OR Neonat* OR 'infancy'/exp OR infant* #20 #1 AND #2 #21 #1 AND #17 #22 (#18 OR #20 OR #21) NOT #19

Web of Science Core Collection

#1 ALL=(male circumcision ) OR TI=(circumci*) OR ALL=(circumci*) Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #2 ALL=(gomco OR mogen OR plastibell OR accucirc OR alisklamp OR (ali's klamp) OR (ali's clamp) OR (ismail clamp) OR prepex OR (shang ring) OR shangring OR (tara klamp) OR unicirc OR smartclamp OR (smart clamp) OR (smart klamp) OR quill) Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #3 ALL=((surgical instruments) OR (surgical instrument*) OR clamp* OR device*) Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #4 #2 AND #1 Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #5 #3 AND #1 Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #6 #5 OR #4 Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #7 (All=((randomized controlled trial) OR (controlled clinical trial)) OR TI=randomized OR TI=placebo OR ALL=(drug therapy) OR TI=randomly OR TI=trial OR TI=groups ) Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years #8 #7 AND #6 Indexes=SCI‐EXPANDED, SSCI, A&HCI, ESCI Timespan=All years

ClinicalTrials.gov

gomco OR mogen OR plastibell OR accucircOR alisklamp OR ali'sklamp OR ali'sclamp OR ismail clamp OR prepex OR shang ring OR shangring OR tara klamp OR unicirc OR smartclamp OR smart clamp OR smart klamp OR quill OR device | Interventional Studies | circumcision | Studies with Male Participants

World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP)

Search for "circumcisions"

Appendix 3. Systematic reviews of circumcision devices

Author	Number of included RCTs (number of participants)	Devices used	Outcomes reported (compared to surgical circumcision technique)
			Adverse events	Operative time	Postoperative time pain during first 24 hours	Postoperative appearance or satisfaction
Cao 2015	8 (3314)	SR vs standard surgical circumcision	RR 0.54, 95% CI 0.39 to 0.74; P < 0.001	MD –17.44, 95% CI –21.61 to –13.27; P < 0.001	Intraoperative pain score: MD –3.13, 95% CI 3.79 to 2.47; P < 0.001	RR 1.29, 95% CI 1.07 to 1.56; P = 0.007
Fan 2016	10 (4649) (pair‐wise and network meta‐analysis conducted)	CDD such as DCSD and Unicirc vs ISD, e.g. SR and PrePex vs surgical circumcision technique	CDD: RR 1.00, 97.5% CI 0.30 to 3.39 ISD: RR 0.77, 97.5% CI 0.47 to 1.26	CDD: SMD –4.33, 97.5% CI –6.43 to –2.23 ISD: SMD –5.72, 97.5% CI –7.11 to –4.33	CDD: SMD –1.38, 97.5% CI –2.28 to –0.48 ISD: SMD –0.95, 97.5% CI –3.09 to 1.19 (no specific time point mentioned for measuring postoperative pain)	CDD: RR 1.12, 97.5% CI 0.96 to 1.30 ISD: RR 1.17, 97.5% CI 1.02 to 1.35
Huang 2017	18 (6179) (pair‐wise and network meta‐analyses conducted)	DCSD vs SR vs surgical circumcision technique	DCSD: OR 0.37, 95% CrI 0.19 to 0.71 SR: OR 1.27, 95% CrI 0.71 to 2.31	DCSD: SMD –20.60, 95% CrI –23.38 to –17.83 SR: SMD –19.14, 95% CrI –21.86 to –16.50	DCSD: SMD –2.01, 95% CrI –4.57 to 0.60 SR: SMD –1.64, 95% CrI –3.77 to 0.54	DCSD: OR 11.42, 95% CrI 3.60 to 37.68 SR: OR 3.85, 95% CrI 1.29 to 12.79
Huo 2017	9 (1898)	DCSD vs surgical circumcision technique	DCSD: OR 0.60, 95% CI 0.41 to 0.87; P = 0.008	DCSD: SMD –21.44, 95% CI –25.08 to –17.79; P < 0.00001	DCSD: SMD –2.36, 95% CI –2.50 to –2.22; P < 0.00001	DCSD: OR 8.77, 95% CI 5.90 to 13.02; P < 0.00001
CDD: circular disposable devices; CI: confidence interval; CrI: credible interval; DCSD: disposable circumcision suture device; ISD: in situ device; MD: mean difference; OR: odds ratio; RCT: randomized controlled trial; RR: risk ratio; SMD: standardized mean difference; SR: Shang Ring.

Data and analyses

Comparison 1. Circumcision devices versus standard surgical techniques.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Serious adverse events	11	3472	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
1.1.1 Crush devices	9	3006	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
1.1.2 Ligature devices	2	466	Risk Ratio (M‐H, Random, 95% CI)	Not estimable
1.2 Moderate adverse events	10	3370	Risk Ratio (M‐H, Random, 95% CI)	1.31 [0.55, 3.10]
1.2.1 Crush devices	8	2904	Risk Ratio (M‐H, Random, 95% CI)	1.60 [0.67, 3.80]
1.2.2 Ligature devices	2	466	Risk Ratio (M‐H, Random, 95% CI)	0.17 [0.02, 1.60]
1.3 Mild adverse events	10	3370	Risk Ratio (M‐H, Random, 95% CI)	1.09 [0.44, 2.72]
1.3.1 Crush devices	8	2904	Risk Ratio (M‐H, Random, 95% CI)	1.21 [0.43, 3.37]
1.3.2 Ligature devices	2	466	Risk Ratio (M‐H, Random, 95% CI)	0.57 [0.08, 3.87]
1.4 Operative time (minutes)	14	4812	Mean Difference (IV, Random, 95% CI)	‐17.26 [‐19.96, ‐14.57]
1.4.1 Crush devices	12	4346	Mean Difference (IV, Random, 95% CI)	‐18.31 [‐21.05, ‐15.57]
1.4.2 Ligature devices	2	466	Mean Difference (IV, Random, 95% CI)	‐11.07 [‐13.52, ‐8.62]
1.5 Pain during the first 24 hours (VAS means)	9	3022	Mean Difference (IV, Random, 95% CI)	‐1.30 [‐2.37, ‐0.22]
1.5.1 Crush devices	8	2796	Mean Difference (IV, Random, 95% CI)	‐1.38 [‐2.56, ‐0.20]
1.5.2 Ligature devices	1	226	Mean Difference (IV, Random, 95% CI)	‐0.60 [‐0.96, ‐0.24]
1.6 Pain during the first 7 days (VAS means)	4	1430	Mean Difference (IV, Random, 95% CI)	0.11 [‐0.89, 1.11]
1.7 Pain during erection (VAS means)	2	626	Mean Difference (IV, Random, 95% CI)	‐0.74 [‐4.57, 3.08]
1.8 Participant satisfaction	15	4501	Risk Ratio (M‐H, Random, 95% CI)	1.19 [1.04, 1.37]
1.8.1 Crush devices	13	4035	Risk Ratio (M‐H, Random, 95% CI)	1.20 [1.03, 1.40]
1.8.2 Ligature devices	2	466	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.57, 2.28]

1.6. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 6: Pain during the first 7 days (VAS means)

1.7. Analysis.

Comparison 1: Circumcision devices versus standard surgical techniques, Outcome 7: Pain during erection (VAS means)

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Huo 2015.

Study characteristics
Methods	Design: prospective RCT conducted in China. Safety and efficacy of a new male circumcision technique involving a circular stapler investigated. 180 participants were prospectively enrolled from Department of Urology, Guangxi Zhuang Autonomous Region People's Hospital. Ethical approval: The People's Hospital of Guangxi Zhuang Autonomous Region Ethics Committee Consent: participants provided written informed consent. Duration of study: enrollment December 2013 to August 2014
Participants	Inclusion criteria: adult men with redundant prepuce or phimosis, who had the indication of circumcision (per 2014 Guidelines for the Diagnosis and Treatment of Urological Diseases in China). For the selection criteria, see "Guide to the Diagnosis and Treatment of Urology in China" edited by Na Yanqun. (那彦群, 叶章群, 孙颖浩, 等主编. 中国泌尿外科疾病诊断治疗指南. 第 1 版. 北京: 人民卫生出版社, 2014. 406–7). Exclusion criteria: urinary tract infection or balanoposthitis; severe adhesion between the inner foreskin layer and the glans, or short frenum and coagulation disorder; diabetes mellitus or other lying diseases that influenced the effect of surgery. Baseline characteristics: comparable between groups. No statistically significant difference existed between age, phimosis and redundant prepuce of 2 groups.  Circumcision device group mean age was 30.4 (SD 9.1) years, traditional method mean age was 31.7 (SD 11.7) years
Interventions	Providers: experienced urologists. Intervention group (stapler): 120 participants. The penis was prepared and disinfected using povidone‐iodine. Local anesthesia performed with 1% lidocaine. Appropriate size of stapler device (Jiangxi Yuanshenglanghe Medical Equipment Company, China) ascertained by measuring diameter of penis and procedure conducted according to a defined protocol. Dressing change and evaluation on wound healing was performed on the 1st, 4th and 7th day after surgery. Follow‐up:  pain assessed 24 hours post‐procedure. Foreskin oedema assessed on day 7, satisfaction assessed on day 3 and 30 post‐procedure. Control group (conventional): 60 participants. Dorsal slit circumcision procedure. Follow‐up:  pain assessed 24 hours post‐procedure. Foreskin oedema assessed on day 7, satisfaction assessed on day 3 and 30 post‐procedure.
Outcomes	Primary outcomes: operative time (minutes) pain score (24 hours after surgery) intraoperative blood loss volume (mL) incision edema (7 days after surgery): distinguished from other complications and summarized by 3 grades postoperative complications/adverse events (infection, hematoma, ecchymosis, reoperation) (24 hours postintervention) wound healing time (days) cosmetic penile appearance (1 month after surgery)
Funding sources	Not stated
Declarations of interest	Not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used a random number table.
Allocation concealment (selection bias)	Unclear risk	No details provided.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions received.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been affected by lack of blinding; rated at high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding is unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	Unclear risk	No study protocol; unclear whether there may have been selective outcome reporting.
Other bias	Low risk	No other biases identified.

Jin 2015.

Study characteristics
Methods	Design: prospective RCT conducted in China. Safety and efficacy of a new male circumcision technique involving the use of a circular stapler was investigated. 879 participants were prospectively enrolled from 4 urologic departments: The First Affiliated Hospital, Beilun People's Hospital, Yuyao People's Hospital, and Xiangshan First People's Hospital. Ethical approval: the Zhejiang University Ethics Committee Consent: participants provided written informed consent. Duration of study: enrollment March 2013 to March 2014
Participants	Inclusion criteria: men aged 18–70 years with a redundant prepuce or phimosis. Exclusion criteria: acute infection of the genitalia (acute posthitis or balanitis), thickened prepuce secondary to chronic inflammation, severe foreskin adhesion, or other contraindications to male circumcision such as a concealed penis or a sexually transmitted disease. Baseline characteristics: comparable between groups. However, an increased number of participants in the conventional group had phimosis, which was statistically significant between the 2 groups. Circumcision device group mean age was 26.9 (SD 9.1) years, conventional group mean age was 25.3 (SD 7.1) years.
Interventions	Providers: trained urologists. Intervention group (stapler): 441 participants. Appropriate size of stapler device ascertained by measuring penis just below glans. The penis was disinfected using povidone‐iodine. Dorsal penile nerve block and circumferential block performed with 1% lidocaine, and procedure conducted according to a defined protocol. Follow‐up: 56 participants in stapler group were followed up for 1 year. Control group (conventional): 438 participants. Male circumcision performed using dorsal slit technique with an electric scalpel (Sanli Medical Equipment Company, China) in accordance with WHO guidance manual. Participants in both groups were counselled to avoid sexual intercourse and masturbation for about 4 weeks after the procedure. Follow‐up: 1 day; 1, 2 and 3 weeks; and 1 and 3 months after surgery;
Outcomes	Primary outcomes: operative time (minutes) pain score (intraoperative and 1 hour after surgery) blood loss volume (mL) postoperative complications (bleeding, wound dehiscence, severe edema, infection) healing time (days) satisfaction treatment costs (US dollars) Definitions for each were clearly outlined in the paper.
Funding sources	Qianjiang Talent Program of Science and Technology Agency of Zhejiang Province (#2013R10045)
Declarations of interest	None stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information provided to determine how a random sequence was generated. The only evidence to suggest randomization was in the title of the paper.
Allocation concealment (selection bias)	High risk	No information provided about allocation concealment. There was a difference in baseline characteristics where the standard surgical care group had more participants with phimosis (11.2% surgical care group vs 6.8% in stapler group; P < 0.05), which could account for poorer outcomes in this group compared to the stapler group.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention; unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Blinding not mentioned in the trial report and was assumed unlikely. Despite the insufficient information on blinding and its potential impact, outcome assessment for subjective outcomes may have been affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	A lack of blinding is unlikely to have resulted in detection bias for the outcome 'time measurement,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 441 participants randomized to the device group and 438 to the conventional group were reported. The reported attrition was < 10% and equal in both groups; low risk of attrition bias.
Selective reporting (reporting bias)	Unclear risk	No study protocol and, therefore, unclear whether there may have been selective outcome reporting. There was no clinical trials registration number in the published report.
Other bias	Low risk	No other biases identified.

Kanyago 2013.

Study characteristics
Methods	Design: effectiveness RCT conducted in southwestern Uganda. Compared SR with FG male circumcision. 138 adult male students (≥ 15 years) attending a local university and undergoing male circumcision recruited from the surgical outpatient department. Ethical approval: Institutional Review Board Committee at Mbarara University of Science and Technology. Study registered with ClinicalTrials.gov (NCT01757938) Consent: participants provided written informed consent. Those who presented for circumcision but who did not consent to participate in the trial were offered standard FG adult male circumcision. Duration of study: January–May 2011
Participants	Inclusion criteria: adult male students attending Mbarara University (aged ≥ 15 years) who consented to participate in the trial. Exclusion criteria: self‐reported HIV infection, chronic paraphimosis, genital ulcers, penile carcinoma, filariasis, xerotica obliterans, balanitis, glans‐prepuce adhesions, frenula scar tissue, or any urethral anatomical abnormality such as hypospadias or epispadias. Baseline characteristics: median age of participants: 22 (interquartile range 21–23) years.
Interventions	Providers: single study surgeon performed all procedures; however, he had no prior experience with the SR procedure. Intervention group (SR): 73 participants. Appropriate size of SR ascertained. Penis disinfected using povidone‐iodine solution. Dorsal penile nerve block and circumferential block performed with 1% lidocaine. Follow‐up: assessed 1 hour after procedure, and on the 3rd, 7th, 14th and 21st postoperative days Control group (FG): 65 participants. Male circumcision performed using a standard technique. Follow‐up: not reported for control group
Outcomes	Primary outcomes: procedure time pain scores time to resumption of normal activity minor and major complications patient satisfaction time to complete healing Definitions for each were clearly outlined in the paper.
Funding sources	1 author, Mark J Siedner, received salary support from the Fogarty International Clinical Research Scholars and Fellows Program at Vanderbilt University (R24 TW007988), and the NIH (T32 AI007433).
Declarations of interest	SR devices supplied by an independent third‐party with no affiliation to Wuhu Santa Medical Devices Technology Co Ltd, China, who was not involved in the design, interpretation or writing of study.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Process for generating the randomization sequence not described; unclear risk of bias.
Allocation concealment (selection bias)	Low risk	Allocation concealment adequately reported; low risk of bias. Quote: "Consenting participants selected an opaque envelope from a box for randomisation to SR or FG groups."
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention; unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	No blinding of outcome assessment was done or was possible for the outcomes that were considered subjective. The presence of the non‐surgical intervention, an SR, until removal would increase the potential risk of bias in these subjective outcomes assessments.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	The lack of blinding was unlikely to have resulted in a detection bias for this objective outcome.
Incomplete outcome data (attrition bias) All outcomes	High risk	There was differential attrition in the 2 groups: 25% attrition in surgical group and 0% attrition in device group. Investigators suggested that the loss to follow‐up in the surgical group likely means that there were no adverse events to report (participants with adverse outcomes are likely to present to the clinic). However, the differential loss to follow‐up without clear reporting on the reasons for losses may have introduced a risk of bias. High risk of bias.
Selective reporting (reporting bias)	Low risk	Study protocol available on ClinicalTrials.gov (NCT01757938). All study's prespecified outcomes were reported.
Other bias	Low risk	No other biases identified.

Lagarde 2009.

Study characteristics
Methods	Design: RCT conducted in South Africa. Compared the TK device with FG male circumcision. 69 participants of the control group of a male circumcision RCT conducted in South Africa were recruited. Ethical approval: University of the Witwatersrand Human Research Ethics Committee (Medical). Adverse events forms were transmitted to a Data Safety Monitoring Board. Consent: participants provided informed consent. Duration of study: September–November 2004
Participants	Inclusion criteria: uncircumcised males from the control group of a male circumcision RCT with no contraindication to circumcision, of good general health with normal physical and genital condition and consenting to participate in trial. They had to agree to avoid sexual contacts (except with condom protection) during the 6 weeks following circumcision, consent to a medical visit 6 weeks after circumcision and consent to reporting of any adverse events. Exclusion criteria: contraindications to circumcision, abnormal physical and genital conditions and not consenting to participate in the trial. Baseline characteristics: groups comparable with no statistically significant differences between them for the listed characteristics. Circumcision device group mean age was 21.5 , interquartile range (IQR) 20 ‐ 23.25 years , traditional method mean age was 22, IQR (21 ‐ 24) years.
Interventions	Providers: 3 GPs who had extensive experience with FG circumcision method conducted both procedures. Intervention group (TK): 35 participants. Appropriate size TK ascertained. Penis disinfected using povidone‐iodine solution. Dorsal penile nerve block and circumferential block performed with 1% lidocaine. Foreskin pulled slightly forward over rim of an inner tube and positioned inside an outer ring. 2 plastic arms locked into place to force 2 surfaces (inner tube and outer clamping ring) into tight contact with foreskin trapped between them. If foreskin was tight, a dorsal slit was required to gain access for ring to be applied over glans. Then foreskin trapped forward of clamping device was cut away. Device was intended to remain on the penis for 7–10 days until removal or it fell off with the necrotized foreskin. Control group (FG): 34 participants. Foreskin was pulled outwards in front of the glans, and forceps clamped across it, parallel to the corona of the glans and immediately in front of the glans. The scalpel was run across the face of the foreskin, and absorbable sutures used to close the cut edges. Excess bleeding controlled with ligature, direct pressure or cautery. Sutured area was covered with sterile paraffin tulle gras, sterile gauze and paper tape; dressing was removed 24–48 hours after surgery by the GP who performed it. Follow‐up: For all groups, 3 days after surgery and participants advised to return to center 6 weeks after surgery for a genital exam and completion of a short questionnaire. Adverse events recorded using a standardized form.
Outcomes	Primary outcomes: number and nature of adverse events penile infection delay in wound healing excessive or insufficient skin removed erectile dysfunction pain score bleeding within 2 weeks following the procedure lesions to the penis, swelling or hematoma within 2 weeks following the procedure, any problem when urinating satisfaction with penis appearance
Funding sources	Agence Nationale de Recherche sur le Sida et le Hépatites, Institut Nationale de la Santé et de la Recherche Médicale (Paris) and National Institute for Communicable Diseases (NICD) (Johannesburg). Quote: "INSERM and the NICD signed collaboration and licensing agreements with a private South African company for providing all sterilised instruments and consumables for performing a single male circumcision according to the FG method."
Declarations of interest	Funders had no role in study design, data collection and analysis, decision to publish or manuscript preparation.
Notes	All participants who attended the center for their last visit of the male circumcision RCT follow‐up received a ZAR 150 (EUR 20) payment whether or not they participated in the study. Participants were asked to return to the center 6 weeks after surgery for a genital exam and completion of a short questionnaire. Participants who attended this postcircumcision visit were paid ZAR 40.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Process for development of the random sequence not described; unclear risk of bias.
Allocation concealment (selection bias)	Unclear risk	Some information about allocation concealment was provided. Quote: "For randomisation, each participant chose an envelope containing the group name from a basket of 10 envelopes. After each choice, a new envelope taken sequentially from a set of envelopes prepared in such a way that each set of 10 envelopes contained the same number of Usual and TK tokens, was added to the basket." Comment: certain details that would inform whether the allocation could be tampered with were not provided. For example it was not reported whether envelopes were sealed or opaque; unclear risk of bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention; unclear whether this may have impacted the performance of either group. Quote: "Neither GPs, participants nor investigators were blinded to the randomisation group. At interview, the nurse was not aware of the method used but on examination could conclude which technique was used."
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	Unclear risk	Outcome assessment was not blinded. Quote: "…nurse who performed the interview and the clinical examination was blinded to the intervention group, but was obviously able to identify the group from seeing the scar." As many of the outcomes were subjective, this may have introduced a high risk of detection bias.
Incomplete outcome data (attrition bias) All outcomes	High risk	There was greater attrition in the TK group (20% with TK vs 10% with surgery). 4 participants crossed over to the standard surgical group. 19/35 participants randomized to the TK group were analyzed for all the outcomes; 7 participants did not visit the GP after circumcision and were excluded from the analyses. 29/34 participants randomized to the FG group were analyzed for all outcomes; 6 did not visit the GP in the FG group and were excluded from further analyses. High risk of attrition bias.
Selective reporting (reporting bias)	Unclear risk	No study protocol and, therefore, it was unclear whether there may have been selective outcome reporting. No mention of clinical trials database registration.
Other bias	High risk	There was potential for bias due to early stopping of the trial and the trial not meeting its planned sample size for the primary outcome (power). The planned number of participants of 400 in each group was not achieved because of trial interruption owing to a high number of adverse events in the TK group. Quote: "The protocol was designed to randomize participants to three groups, including men circumcised with a single‐use sterile surgery kit, but because of delay in its availability, the study started with only two (the TK vs. FG method with re‐usable instruments). The postcircumcision visit was originally planned exactly 6 weeks after surgery, but only 3 participants made the visit 1–3 days before 6 weeks. Median and mean intervals between circumcision and visit were more than 6 weeks. The trial investigators informed the DSMB and stopped the trial." Comment: the TK supplier attributed the high number of adverse events to the inexperience of GPs using the device and an experienced practitioner was sent to provide additional training to the GPs.

Li 2010.

Study characteristics
Methods	Design: prospective RCT conducted in China. The safety and efficacy of SR circumcision compared to conventional surgical procedures was investigated. 724 participants prospectively enrolled from the Department of Urology in Shaanxi Second People's Hospital. Ethical approval: not described Consent: participants provided written informed consent. Duration of study: enrollment March 2008 to August 2009
Participants	Inclusion criteria: males who wished to undergo circumcision. Exclusion criteria: not described. Baseline characteristics: participant characteristics comparable between groups. Circumcision device group, there were  282 participants in the 16 ‐ 35 years age group, 98 participants in the 36 ‐ 55 age group, 22 in the 56 ‐ 69 age group. For the traditional methods group, there were 219 participants in the 16 ‐ 35 years age group, 87 participants in the 36 ‐ 55 age group, 16 in the 56 ‐ 69 age group.
Interventions	Providers: not described in detail. Intervention group (SR): 402 participants (single or double date of visit). Appropriate size of SR was ascertained. Penis was disinfected using povidone‐iodine solution. Dorsal penile nerve block and circumferential block performed with 1% lidocaine. Follow up: Pain assessed at 24 hours post‐procedure Control group (conventional): 322 participants. A circumcision laser incision was used.   Follow up: Pain assessed at 24 hours post‐procedure
Outcomes	Postoperative observation items: operation time (minutes) postoperative pain scores (24 hours after surgery) using VAS postoperative complications (bleeding, wound dehiscence, edema, infection, healing time) postoperative appearance satisfaction postoperative sexual function cost
Funding sources	Unclear
Declarations of interest	Unclear
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Participants were randomized according to day/date of clinic visit.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions they had received.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding was unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	Unclear risk	No study protocol available and, therefore, it was unclear whether there may have been selective outcome reporting.
Other bias	Low risk	No other biases identified.

Li 2014.

Study characteristics
Methods	Design: prospective RCT conducted in China. The safety and efficacy of a new male circumcision technique involving the use of a circular stapler was investigated. 249 participants prospectively enrolled from Department of Urology, Second Affiliated Hospital of Wenzhou Medical University. Ethical approval: not stated Consent: participants provided informed consent. Duration of study: May 2013 to December 2013
Participants	Inclusion criteria: men with redundant prepuce or phimosis, who intended to have circumcision. Exclusion criteria: acute urinary tract infection, occult penis, severe foreskin penis head adhesion, too short foreskin and long foreskin, history of neurologic or psychiatric disorders, leukemia, bleeding disorders and uncontrolled diabetes. Baseline characteristics: participant characteristics comparable between groups in age, phimosis and redundant prepuce. Mean age of both groups combined was 26.94 (SD 6.25) years.
Interventions	Providers: not described in detail. Intervention group (stapler): 129 participants. Appropriate size of stapler device (Jiangxi Yuanshenglanghe Medical Equipment Company, China) ascertained and procedure performed according to defined protocol. Follow up: pain assessed at 24 hours,  3rd and 30th day post‐procedure. Control group (conventional): 120 participants. Dorsal slit circumcision procedure. Follow up: pain assessed at 24 hours,  3rd and 30th day post‐procedure.
Outcomes	Primary outcomes: operation time (minutes) pain score using VAS (intraoperative and 24 hours after surgery) blood loss volume (mL) postoperative complications (infection, edema, bleeding, wound dehiscence) wound healing time (days) penile appearance satisfaction (3 days and 1 month after surgery) total treatment cost (Chinese Yuan)
Funding sources	Not stated
Declarations of interest	Not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details given.
Allocation concealment (selection bias)	Unclear risk	No details given.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding was unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	Unclear risk	No study protocol and, therefore, unclear whether there may have been selective outcome reporting.
Other bias	Low risk	No other biases identified.

Lv 2014.

Study characteristics
Methods	Design: prospective RCT conducted at 5 different hospitals in China. Compared conventional circumcision vs SR vs DCSD. 942 participants from 5 hospitals. Ethical approval: ethics committee (number: 2012‐Research‐52) Consent: participants provided written informed consent. Duration of study: October 2012 to May 2013
Participants	Inclusion criteria: men requiring circumcision for a redundant prepuce or phimosis. Exclusion criteria: ages < 18 years, dense adhesions between glans and foreskin and obese men with partly buried penis. Baseline characteristics: mean age across groups: 31.5 (SD 5.4; range 18–58) years. No significant age differences between groups (P > 0.05). No other comments were made regarding other characteristics evaluated across groups. Mean age of both groups combined was 31.5 (SD 5.4) years,  age ranging between 18 ‐ 58 years.
Interventions	Providers: 1 experienced surgeon who had performed conventional and SR circumcisions and was trained to use the devices before the study began was chosen from each of the 5 hospitals. Intervention group 1 (DCSD): 314 participants. 5% lidocaine cream was used for anesthesia. The device had bell‐shaped glans pedestal, suture staple, ring‐shaped blade, handle and shell. The glans was covered by the U‐shaped glans rest with the edge of the U‐shape at the level of the coronary sulcus. After the foreskin was wrapped around the rod, the rod was inserted into the center hole of the circumcision device. The application knob was tightened, the safe buckle was removed and the rod was pushed down to trigger the circumcision device. After triggering the application knob when the ring‐shaped blade hidden in the shell was pushed out with the staples. The blade cut the foreskin instantly, while simultaneously placing staples, by tightening the application knob at the bottom for 3–5 seconds and releasing it. Intervention group 2 (SR): 314 participants. 2% lidocaine injection (10 mL) for dorsal penile nerve block and a suitable sized device used. Control group (conventional): 314 participants. 2% lidocaine injection (10 mL) for dorsal penile nerve block and a traditional electrosurgical knife and absorbable suture used. Follow‐up: 1 and 2 weeks, and 1 month after for all groups
Outcomes	Primary outcomes: postoperative complications (hematoma, edema, disruption of incision, incision infection) healing time intraoperative and postoperative pain intraoperative blood loss operation time patient satisfaction Definitions for each were clearly outlined in the paper.
Funding sources	Jiangxi Yuansheng Lang He Medical Instrument Co., Ltd. for the equipment and technique provided.
Declarations of interest	Authors declared no competing interests.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table used to develop random sequence.
Allocation concealment (selection bias)	Unclear risk	No information about the process for ensuring allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention; unclear whether this may have impacted the performance of either group. No information provided regarding blinding of participants and personnel.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Outcome assessment was unblinded and would most likely result in detection bias for the subjective outcomes. The researchers attempted to standardize outcome assessment for adverse effects, as it stated in the paper. Quote: "we defined assessment standards for incision healing: the incision is closed and covered by skin, no rupture under moderate exercise; with time, the healed incision can withstand certain stretching force and pressure, the pigmentation is relieved and is close to the color of normal skin; return to the intact normal skin function." Comment: despite this, as outcome detection still had an element of subjectivity; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Lack of blinding was unlikely to affect the outcomes assessment of operative time, which we considered an objective outcome.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants randomized were included in the analysis of all outcomes.
Selective reporting (reporting bias)	Unclear risk	No study protocol. Therefore, unclear whether there may have been selective outcome reporting. No mention of clinical trials database registration.
Other bias	Low risk	No other biases identified.

Miao 2015.

Study characteristics
Methods	Design: prospective RCT conducted in China. Clinical effects of circumcision stapler, circumcision cerclage and traditional circumcision in the treatment of phimosis and redundant prepuce. 276 participants prospectively enrolled from Department of Urology, Zhangjiagang First People's Hospital. Ethical approval: First Hospital of Zhangjiagang Affiliated to Suzhou University Ethics Committee Consent: participants provided informed consent. Duration of study: January 2012 to January 2014
Participants	Inclusion criteria: Males with redundant prepuce or phimosis. Exclusion criteria: people with acute inflammation, edema or deformity in the surgical site. Baseline characteristics: participant characteristics comparable between groups in age, phimosis and redundant prepuce. Mean age 22.5 (SD 5.7) years, ranging from nine to 56 years.
Interventions	Providers: not clearly described. Intervention group 1 (stapler): 92 participants. Appropriate size of stapler device (Jiangxi Yuanshenglanghe Medical Equipment Company, China) ascertained by measuring diameter of penis using a specialized measurement card. Dorsal penile nerve block with 1% lidocaine (5–10 mL) and procedure performed according to a defined protocol. Intervention group 2 (cerclage): 92 participants. Disposable circumcision ring pack produced by Guangzhou Nanfeng Biotech Company. Hemostatic forceps used to clamp the inner and outer plates of both sides of foreskin and side bands of foreskin, double‐stretch elastic ligature used to evenly ligate the foreskin in groove of collar, excess foreskin removed and wound disinfected. Follow up: 3 months for both intervention groups Control group (conventional): 92 participants. Dorsal slit circumcision procedure. Follow up: 3 months
Outcomes	Primary outcomes: operation time (minutes) pain score using VAS (intraoperative and 24 hours after surgery) blood loss volume (mL) postoperative complications (edema, hematoma, deformity and infection)
Funding sources	Not stated
Declarations of interest	Not stated
Notes	Information on the proportion of the total number of participants younger than 10 years was not available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details given.
Allocation concealment (selection bias)	Unclear risk	No details given.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions they had received.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding was unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	Unclear risk	No study protocol; therefore, it was unclear whether there may have been selective outcome reporting.
Other bias	Low risk	No other biases identified.

Millard 2013.

Study characteristics
Methods	Design: single‐center non‐blinded RCT conducted in Mozambique. Compared open surgical circumcision with suturing and Gomco instrument plus tissue adhesive. 200 participants recruited via a poster at the medical school in Beira, Mozambique. Ethical approval: University of Pittsburgh Institutional Review Board and the National Committee of Bioethics of the Mozambique Ministry of Health Consent: participants provided informed consent. Duration of study: 30 October 2012 to 2 March 2013
Participants	Inclusion criteria: healthy uncircumcised men ages > 18 years, including men with a scarred frenulum. Exclusion criteria: concurrent illness, history of bleeding disorder, previous reaction to local anesthetic, infection, penile abnormality and phimosis. Baseline characteristics: groups comparable with no statistically significant differences between them for the listed characteristics. Circumcision device group, there were  49 participants in the 18 ‐ 20 years age group, 27 participants in the 21 ‐ 24 age group, 24 in the > 25 age group. For the traditional methods group, there were 43 participants in the 18 ‐ 20 years age group, 26 participants in the 21 ‐ 24 age group, 31 in the > 25 age group. Participants received HIV prevention counseling. HIV testing offered, but was not a study prerequisite. Participants advised to abstain from sexual intercourse until wound was completely healed and for ≥ 4 weeks after circumcision.
Interventions	Providers: 2 doctors performed each circumcision: a faculty member – highly experienced in circumcision – and a recent medical graduate with basic surgical skills. As junior team members became more skilled, they performed more of the procedure with assistance from the senior doctor. Intervention group (Gomco circumcision clamp with cyanoacrylate skin adhesive): 100 participants. Local anesthesia (2% lidocaine as a subcutaneous ring block at base of penis) provided as per the WHO Manual for Male Circumcision under LocalAnaesthesia. Gomco circumcision clamp with cyanoacrylate skin adhesive. Gomco clamp is a metal, sterilizable instrument, available in sizes from infant to adult. 4 diameters were available: 2.6 cm, 2.9 cm, 3.2 cm and 3.5 cm. Clamp was applied to penis and after 5 minutes the foreskin was excised with a surgical scalpel. Instrument was then removed and the apposed skin–mucosal edges sealed with high viscosity 2‐octyl cyanoacrylate skin adhesive. Wound was covered with an adherent tape or absorbent gauze. Control group (open surgical technique): 100 participants. Local anesthesia (2% lidocaine as a subcutaneous ring block at the base of penis) given and dorsal slit technique used as described in the WHO Manual for Male Circumcision under LocalAnaesthesia. After suturing, wound was covered with an absorbent gauze dressing. Follow‐up: 2 and 7 days, 2 and 4 weeks. Those who were not completely healed by 4 weeks had an additional 6‐week follow‐up visit for all groups.
Outcomes	Primary outcome: intraoperative time Secondary outcomes: doctor‐described ease in performing the technique operative and postoperative complications (anesthetic complications, bleeding, hematoma, infection, wound disruption, problems with urination, subsequent procedures conducted to correct complications and occupational exposure to blood and body fluids) postoperative pain time to healing patient satisfaction cosmetic result cost of expendable materials Outcome definitions were provided in the paper
Funding sources	Funded by the Catholic University of Mozambique and by a Fogarty International Center, National Institutes of Health grant (no. 3 D43 TW01038) to the University of Pittsburgh. No funding bodies played any role in the design, writing or decision to publish the manuscript.
Declarations of interest	None stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization using a random number table in blocks of 10 to ensure balanced numbers in each arm.
Allocation concealment (selection bias)	Unclear risk	Although methods of concealment of allocation groups were described, such as using sealed opaque envelopes, the block randomization method increases predictability of allocation groups in a non‐blinded trial; unclear risk of bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention; unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	This study was not blinded. The subjective outcomes were self‐reported and could have been influenced by the lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Low risk of bias as more objective outcomes are unlikely to be influenced by lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Low attrition (< 5%) in each group; low risk of attrition bias.
Selective reporting (reporting bias)	Unclear risk	No study protocol and it was, therefore, unclear whether there may have been be selective outcome reporting. There was no mention of clinical trials database registration.
Other bias	Low risk	No other biases identified.

Millard 2014.

Study characteristics
Methods	Design: single‐center non‐blinded RCT conducted in South Africa. Compared open surgical circumcision with suturing and the Unicirc disposable instrument plus tissue adhesive. 150 participants recruited via posters at 2 affiliated primary healthcare clinics in Cape Town, South Africa. Ethical approval: South African Medical Association's Ethics Committee (SAMAREC). ClinicalTrials.gov identifier: NCT01877408 Consent: participants provided informed consent. Duration of study: 21 June to 8 August 2013
Participants	Inclusion criteria: healthy uncircumcised men ages > 18 years, including men with partial phimosis without adhesions and with scarring of the frenulum. Exclusion criteria: concurrent illness, history of bleeding disorder, past reaction to local anesthetic, infection, penile abnormality and phimosis. Baseline characteristics: groups comparable with no statistically significant differences between them for the listed characteristics. Circumcision device group, there were  21 participants in the 18 ‐ 25 years age group, 49 participants in the 26 ‐ 35 age group, 30 in the > 36 age group. For the traditional methods group, there were 12 participants in the 18 ‐ 25 years age group, 21 participants in the 26 ‐ 35 age group, 17 in the > 36 age group Participants received HIV prevention counseling. HIV testing offered, but was not a study prerequisite. Participants advised to abstain from sexual intercourse until wound was completely healed and for ≥ 4 weeks after circumcision.
Interventions	Providers: 4 generalist doctors, assisted by registered nurses. Intervention group (Unicirc disposable instrument plus tissue adhesive): 100 participants. 2% lidocaine with marcaine local anesthetic used as subcutaneous ring block at base of penis as per the WHO Manual for Male Circumcision under Local Anaesthesia. Unicirc with cyanoacrylate skin adhesive. Unicirc is a plastic and metal single‐use disposable instrument. The instruments were gas sterilized in sealed packages. Applied the Unicirc to foreskin, and waited 5 minutes before excising the foreskin with a surgical scalpel. Then removed the instrument and sealed the apposed skin–mucosal edges with cyanoacrylate skin adhesive. Used 4 different Unicirc sizes: 2.6 cm, 2.9 cm, 3.2 cm and 3.5 cm. Covered wound with an adherent tape (Hypafix) and absorbent gauze. Control group (open surgical technique): 50 participants. 2% lidocaine with marcaine local anesthesia used as a subcutaneous ring block at the base of the penis as per the WHO Manual for Male Circumcision under Local Anaesthesia. Dorsal slit technique, sleeve technique or FG technique performed according to doctor preference as described in the WHO Manual for Male Circumcision under LocalAnaesthesia. After suturing, the wound was covered with an absorbent gauze dressing. Follow‐up: For all group, 2nd and 7th day, 2nd and 4th week. Those who were not completely healed by 4 weeks had an additional 6‐week follow‐up visit.
Outcomes	Primary outcome: intraoperative time Secondary outcomes: doctor‐described ease in performing the technique operative and postoperative complications (anesthetic complications, bleeding, hematoma, infection, wound disruption, problems with urination, subsequent procedures conducted to correct complications and occupational exposure to blood and body fluids) postoperative pain time to healing patient satisfaction cosmetic result cost of expendable materials Outcome definitions are provided in the paper
Funding sources	None stated
Declarations of interest	Supported by Simunye Primary Health Care. No funding bodies played any role in the design, writing or decision to publish the manuscript
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization sequence generated using a random number table and an investigator who was not involved in the surgeries allocated participants in a 2:1 ratio in blocks of 15 using a random number table.
Allocation concealment (selection bias)	Low risk	Methods were used to conceal allocation groups as stated. Quote: "Slips of paper with the group assignment were folded and placed in sealed, opaque envelopes. Each envelope was opened only at the time of surgery." Comment: 1 disadvantage of block randomization is that the allocation of participants may be predictable in an unblinded trial; however, we rated allocation concealment at low risk of bias as baseline characteristics were similar between groups and did not reveal selection bias in group allocation.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Lack of blinding was judged to confer a high risk of performance bias for these subjective outcomes.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	No mention of blinded outcome assessors. The lack of blinding was unlikely to affect the assessment of the objectively defined outcomes, such as operative time, which was the primary outcome.
Incomplete outcome data (attrition bias) All outcomes	Low risk	There was a very low attrition rate, and no difference between groups; low risk of attrition bias.
Selective reporting (reporting bias)	Low risk	The study outline was found on ClinicalTrials.gov (NCT01877408) and all outcomes were reported; low risk of selective reporting bias.
Other bias	Low risk	No other biases identified.

Mutabazi 2012.

Study characteristics
Methods	Design: single‐center unblinded RCT conducted in Rwanda. Compared PrePex device to surgical circumcision (dorsal‐slit technique). 226 men randomized and allocated according to WHO allocation recommendations (2:1) = (PrePex circumcision: surgical circumcision). Ethical approval: Rwanda National Ethics Committee on 13 January 2011, approval number N0 001/RNEC/2011 Consent: participants provided written informed consent. Duration of study: 17 February 2011 to 8 April 2011
Participants	Inclusion criteria: men ages 21–54 years who were willing to undergo both surgical and PrePex device circumcisions. Exclusion criteria: any contraindication to surgical circumcision (anatomical abnormalities, genital ulcer disease, previous circumcision etc.), HIV or unwilling to consent to an HIV test, and mental illness with unreliable follow‐up. Baseline characteristics: no difference in baseline characteristics. Mean age: 26 years. All were HIV negative.  Circumcision device group mean age was 26 (SD 5.17) years, traditional method mean age was 24 (SD 4.10) years
Interventions	Providers: all surgical circumcisions conducted by surgeons experienced in circumcision. All non‐surgical circumcision procedures were performed by surgeons who participated in prior safety and efficacy studies using PrePex in Rwanda. Intervention group (PrePex circumcision device): 150 participants. Clean skin preparation, anesthetic cream (lidocaine and prilocaine) followed by application of the PrePex device, dry gauze dressing applied and the device removed 7 days later. Follow‐up: assessed 16 hour after procedure, and on the 3rd, 9th, 14th, 21st, 28th, 35th, 42nd, 49th, 56th, 63rd postoperative days Control group (conventional [dorsal slit] surgical circumcision): 76 participants. Sterile skin preparation, injection base penile ring block, dorsal‐slit technique followed by absorbable skin closure and dry dressings. Follow‐up: 7th postoperative day
Outcomes	Primary outcomes: total male circumcision time (skin to skin) Secondary outcomes: procedure preparation time clinical adverse event rates surgical procedure‐related and device‐related incident rates expected adverse effects pain at key time points and during erection on days 3 and 9 loss of productivity patient satisfaction time to complete healing
Funding sources	Government of Rwanda. PrePex devices were provided by Circ MedTech Limited (Tortola, British Virgin Islands).
Declarations of interest	Disclosed no conflicts of interest.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The process for randomization was clearly described. Quote: "A randomized number table was created by the study PI using the RAND algorithm in Excel (Microsoft Corp, Redmond, WA) in which each randomized number was distributed between 1.00000 and 0.00000. To achieve an allocation ratio of 2:1, each random number below 0.33333 was assigned to the surgical arm and each number higher than 0.333334 was assigned to the PrePex arm." Comment: low risk of selection bias.
Allocation concealment (selection bias)	Low risk	The process for allocation concealment was clearly described. Quote: "Study numbers, randomised numbers and their respective treatment arms were then placed in envelopes that were validated to be non‐transparent. This was done by the principal study investigator who was not involved in the enrolment and treatment allocation. It was also done before the enrolment process." Comment: low risk of selection bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	There were no independent outcome assessors. The lack of blinding was judged to potentially affect the assessment of these subjective outcomes. In addition, the trial design allowed for differential follow‐up schedules for the intervention group (PrePex) within the first visit at 3 days compared to the surgical group who had follow‐up at 7 days. This may introduce some bias as the adverse effects related to wound healing may be expected to be different. High risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	As the operative time was considered an objective outcome, the lack of blinding was not considered a risk of detection bias. In addition, the trial put in place additional measures to minimize biased reporting. Quote: "Time intervals were measured using a stopwatch; as a backup, a video of each procedure was recorded."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition for the main outcomes was very low and considered a low risk of attrition bias. It should be noted that there was high attrition (35%) for the outcome 'participant satisfaction' which required a questionnaire to be completed.
Selective reporting (reporting bias)	Low risk	Study registered at ClinicalTrials.gov (NCT01284088) and the prespecified outcomes align with those reported in the published trial.
Other bias	Low risk	No other biases identified.

Pang 2015.

Study characteristics
Methods	Design: prospective RCT conducted in China. The safety and efficacy of a circular stapler compared with conventional surgery. 56 participants prospectively enrolled from the Department of Urology at Huainan Chaoyang Hospital. Ethical approval: not stated Consent: participants provided written informed consent. Duration of study: not stated
Participants	Inclusion criteria: men with redundant prepuce or phimosis. Exclusion criteria: other diseases such as urinary tract infection, leukemia and bleeding disease. Baseline characteristics: participant characteristics comparable between groups in age, phimosis and redundant prepuce. Mean age 21.3 (SD 2.5) years, ranging from 12 to 61 years.
Interventions	Providers: not clearly described. Intervention group (stapler): 28 participants. Penis prepared and disinfected using povidone‐iodine and procedure performed according to a defined protocol. Follow‐up: Not reported Control group (conventional surgical): 60 participants. Sterile skin preparation, injection base penile ring block, dorsal slit circumcision procedure performed. Follow‐up: Not reported
Outcomes	Primary outcomes: operative time (minutes) pain score (24 hours after surgery) blood loss volume (mL) postoperative complications (edema, infection, deformity) healing time (days) satisfaction treatment costs
Funding sources	Not stated
Declarations of interest	Not stated
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details given.
Allocation concealment (selection bias)	Unclear risk	No details given.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions they had received.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding was unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	High risk	No study protocol and, therefore, it was unclear whether there may have been selective outcome reporting. In addition, the pain outcome was not reported.
Other bias	Low risk	No other biases identified.

Ren 2014.

Study characteristics
Methods	Design: prospective RCT conducted in China. The safety and efficacy of a circular stapler compared with conventional surgery. 220 participants prospectively enrolled from Department of Urology, Xuecheng People's Hospital, Zaozhuang City, Shandong Province. Ethical approval: not stated Consent: not stated Duration of study: from January 2013 to March 2014
Participants	Inclusion criteria: Males with redundant prepuce or phimosis. Exclusion criteria: other diseases such as urinary tract infection, leukemia and bleeding disease. Baseline characteristics: not stated. 84 cases in the control group, including 18 cases of phimosis and 66 cases of redundant foreskin. Intervention group (device group) included 136 cases, including 27 cases of phimosis and 109 cases of redundant foreskin. Mean age 24 years, ranging from seven to 56 years.
Interventions	Providers: not clearly described. Intervention group (stapler): 136 participants. Penis prepared and disinfected using povidone‐iodine. Dorsal penile nerve block with 1% lidocaine and procedure performed according to a defined protocol. Follow up: satisfaction assessed day 30 post‐procedure Control group (conventional surgery): 84 participants. Sterile skin preparation, injection base penile ring block, dorsal slit circumcision procedure was performed. Follow up: satisfaction assessed day 30 post‐procedure
Outcomes	Primary outcomes: operative time (minutes) pain score (intraoperative and 24 hours after surgery) blood loss volume (mL) postoperative complications (hemorrhage, hematoma, wound disruption, edema, infection) healing time (days) satisfaction (30 days after surgery) Definitions for each are clearly outlined in the paper
Funding sources	Not stated
Declarations of interest	Not stated
Notes	Information on the proportion of the total number of participants younger than 10 years was not available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details given.
Allocation concealment (selection bias)	Unclear risk	No details given.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Not described. Participants were likely aware of interventions they had received.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Not described. Providers were likely aware of interventions; outcome assessment for subjective outcomes may have been be affected by lack of blinding; high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Not described. A lack of blinding was unlikely to have resulted in detection bias for the outcome 'operation time,' which we considered objective.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data or missing data had limited effect on results.
Selective reporting (reporting bias)	High risk	No study protocol and, therefore, it was unclear whether there may have been selective outcome reporting. In addition, the pain outcome was not reported.
Other bias	Low risk	No other biases identified.

Shenje 2016.

Study characteristics
Methods	Design: single‐center non‐blinded, parallel RCT conducted in South Africa. Compared Unicerc devise vs surgery. 75 men randomized in 2:1 (Unicirc device: surgical circumcision) allocation ratio in balanced blocks of 15. Ethical approval: South African Medical Association's Ethics Committee (SAMAREC) Consent: participants provided written informed consent. Duration of study: 15 July to 7 August 2015
Participants	Inclusion criteria: healthy uncircumcised men ages ≥ 16 years (age of consent in South Africa). Exclusion criteria: current illness, bleeding disorder, reaction to local anesthetic, infection or penile abnormality potentially complicating the circumcision. Baseline characteristics: there were no differences in baseline characteristics between groups. Median age was 37.5 years for the Unicirc group and 34 years for the surgical group. Most were married or in a relationship. Most were of Christian faith. Most cited reasons for seeking circumcision as being improved hygiene and to reduce HIV infection. Although HIV testing was performed as per hospital policy, it was NOT an exclusion criterion for the trial. Investigators were unaware of the HIV status.
Interventions	Providers: same 2 doctors for both open surgical and device circumcisions. Intervention group (Unicirc with cyanoacrylate tissue adhesive): 50 participants. Skin preparation with povidone iodine, topical anesthetic cream (lidocaine and prilocaine) application, device size selection by means of a sizing plate. The device was then applied for 5 minutes, followed by excision with a scalpel. The skin–mucosa edges were then sealed with cyanoacrylate glue (Dema‐Flex QS). Dressings were dry sterile gauze with Hypafix adhesive tape. Control group (open surgical technique): 25 participants. Standard FG circumcision technique under local anesthesia (as above). Follow‐up: assessed at day 7, and week 4 and 6 for all groups.
Outcomes	Primary outcome: intraoperative duration Secondary outcome: pain (intraoperative and postoperative using VAS) adverse events (intraoperative and postoperative – classified according to the WHO Framework) wound disruption healing at 4 weeks patient satisfaction (5‐point Likert scale)
Funding sources	Simunye Primary Health Care. The funder had no role in study design, data collection or analysis, decision to publish or preparation of the manuscript.
Declarations of interest	Declared as no competing or conflicts of interest.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The sequence was generated using a random number table; low risk of selection bias. Quote: "An investigator who was not involved in the surgeries allocated participants in a 2:1 ratio using a random number table; block randomisation (in blocks of 15) was used to ensure exact 2:1 allocation."
Allocation concealment (selection bias)	Low risk	Allocation concealed using opaque sealed envelopes. A disadvantage of block randomization used in this trial was that the allocation of participants may be predictable in an unblinded trial; however, we rated allocation concealment at low risk of bias as baseline characteristics were relatively similar between groups and did not indicate selection bias in group allocation.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	This study was not blinded. Despite attempts to provide specific descriptions of the outcomes for assessment, there was a risk of detection bias for the subjective outcomes such as adverse effects. High risk for detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	As the operative time was considered an objective outcome, the lack of blinding was not considered a risk of detection bias.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall, attrition was low and follow‐up good; low risk of attrition bias.
Selective reporting (reporting bias)	Low risk	Study protocol in ClinicalTrials.gov (NCT02443792). The prespecified outcomes were reported; low risk of selective outcome reporting.
Other bias	Low risk	No other biases identified.

Sokal 2014.

Study characteristics
Methods	Design: parallel‐group open‐label RCT with 1‐to‐1 allocations conducted in 2 sites in Kenya. Designed as a superiority trial to prove superiority of Shang Ring over conventional circumcision. 400 men randomized. Ethical approval: FHI 360, the Kenya Medical Research Institute, and the University of Zambia. The Zambia Ministry of Health also granted permission to conduct the study Consent: participants provided informed consent. Duration of study: March–June 2011
Participants	Inclusion criteria: healthy uncircumcised, HIV‐negative men ages 18–54 years. Exclusion criteria: active genital infection, previous circumcision, an anatomic abnormality or another condition that contraindicated elective surgery under local anesthesia (e.g. bleeding diathesis, lidocaine allergy). Baseline characteristics: in both the Kenyan and Zambian groups, there were no differences in baseline characteristics between groups. The median age was 19 years in Kenya and 22.5 years in Zambia.
Interventions	Providers: 2 physicians and 2 nurses (experienced in conventional circumcisions and trained on SR) performed the procedures. Intervention group (SR): 200 participants. Local anesthetic (dorsal penile nerve and ring blocks injected with 1% lidocaine) and application of SR as per previous description in the literature. No specific mention of skin preparation techniques. Control group (conventional circumcision): 200 participants. Local anesthetic technique as above and conventional circumcision as per previous description in the literature. FG technique in Kenya and dorsal‐slit method in Zambia. Follow‐up: assessed on the 2nd, 7th, 14th, 21st, 28th, 35th, 42nd, 60th postoperative days for all groups.
Outcomes	Primary outcomes: safety assessed using the Public Services International/WHO Adverse Event Action Guide to classify adverse events as mild, moderate and severe moderate wound dehiscence: mucocutaneous gap of > 1 cm between wound edges pain and acceptability outcomes were assessed using a VAS ease of use and provider preferences measured duration of procedure (minutes) surgical difficulties encountered
Funding sources	Bill and Melinda Gates Foundation to FHI 360 (Grant OPPGH5166)
Declarations of interest	No conflicts of interest to disclose.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequences generated using permuted blocks with randomly chosen block size; low risk of selection bias.
Allocation concealment (selection bias)	Low risk	Allocation concealment was done using sequentially numbered, opaque, sealed, tamper‐evident envelopes. The unblinded nature of the trial coupled with block randomization could make allocations predictable; however, this was mitigated by use of random block size. Low risk of selection bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	The outcome assessment was unblinded and would result in a possibility of detection bias for the subjective outcomes such as adverse effects.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	As the operative time was an objective outcome, the lack of blinding was not considered a risk of detection bias.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition was clearly described in the published trial and was low for both groups.
Selective reporting (reporting bias)	Low risk	Study protocol available in ClinicalTrials.gov (NCT01300910) and the planned outcomes matched those reported in the trial; low risk of selective outcome reporting.
Other bias	Low risk	No other biases identified.

Tshimanga 2016.

Study characteristics
Methods	Design: randomized, open‐label RCT conducted in Zimbabwe and Malawi. Compared the performance of the PrePex device to FG surgical circumcision. 240 men randomized. Ethical approval: Medical Research Council of Zimbabwe's (MRCZ) and Battelle's Institutional Review Board approved this study (MRCZ: approval 5 September 2011, MRCZ/A/1628; Battelle: approval October 2011, IRB000284). All interviewers and nurses attended study protocol training sessions, as well as a good clinical practice (GCP) courses offered jointly by the Research Support Centre, College of Medicine of Malawi and MRCZ Consent: participants provided written formal consent and confirmed their understanding of their participation. Duration of study: 21 November 2011 to 18 January 2012
Participants	Inclusion criteria: males aged ≥ 18 years, uncircumcised and agreement to be circumcised, HIV‐negative, ability to understand study procedures and requirements, and freely give informed consent for participation in this study, agreement to abstain from sexual intercourse and directly rubbing circumcised area (up to 70 days after procedure) and until the end of the follow‐up, and to return to the healthcare facility for follow‐up visits (or as instructed). Exclusion criteria: active genital infection, anatomic abnormality, or other condition, which in the opinion of the investigator prevented the participant from undergoing a circumcision. Men who had phimosis, paraphimosis, warts under the prepuce, torn or tight frenulum, narrow prepuce, hypospadias, epispadias, known bleeding/coagulation abnormality, uncontrolled diabetes, did not agree to anonymous video and photographs of the procedure and follow‐up visits, HIV‐positive, refusal to take HIV test and ages < 18 years. Baseline characteristics: no differences in baseline characteristics between groups. Mean age: 29.1 (SD 9.0) years for the PrePex group and 27.6 (SD 7.6) years for surgical group (difference not statistically significant). 49.2% were single and 73% were employed.
Interventions	Providers: doctors and nurses trained by PrePex Masters from Rwanda. Intervention group (PrePex): 160 participants. Clean and non‐sterile preparation followed by dermal anesthetic cream (5% lidocaine). The device was then applied. Follow‐up: 7 days for device removal, 2 days later for dressing removal and weekly thereafter for 6 weeks. Control group (surgical circumcision): 80 participants. Surgical circumcisions were done under local anesthesia using the FG method consistent with the national guidelines adapted from the WHO Manual for Male Circumcision under Local Anesthesia. Follow‐up: assessed on the 3rd, 7th, 14th, 21st, 42nd postoperative days
Outcomes	Primary outcomes: procedure operative time Secondary outcomes: pain using VAS safety – adverse events (bleeding; hematoma, edema, infection, device‐related adverse events) satisfaction (participant and partner filled out questionnaire using a 5‐point scale) time to complete healing based on treating physician's objective wound assessment using specified parameters difficulty during procedure based on reports cost of procedure
Funding sources	United Nations Population Fund
Declarations of interest	The authors declared that they had no competing interests.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization sequence generated using Excel; low risk of selection bias.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment in the trial used numbered marbles picked out of a box. Participants picked 1 numbered marble from the box, which was then set aside. The participant was assigned that number. The nurse verified which study arm the participant was assigned to according to the randomization list. If a participant opted out of the assigned arm, he was excluded from the study. His marble was put back in the box. It is possible that this approach could result in changes to allocation. Unclear risk of selection bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	There was no mention of independent outcome assessment. The lack of blinding of subjective outcome assessment could confer a high risk of detection bias. In addition, each arm received different follow‐up visit days which could also result in differential assessment of the outcomes. High potential risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	As the operative time was an objective outcome, the lack of blinding was not considered a risk of detection bias.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall there was low attrition between groups with no differences in numbers lost to follow‐up; low risk of attrition bias.
Selective reporting (reporting bias)	High risk	Study outline in ClinicalTrials.gov (NCT01956370). The primary outcome was unchanged; however, reporting on pain was not explicitly mentioned in the registry record. Therefore, indicated a potential reporting bias.
Other bias	Low risk	No other biases identified.

Wang 2014.

Study characteristics
Methods	Design: randomized, multicenter pilot RCT conducted in China with non‐inferiority design. Compared a novel disposable device to conventional dissection technique (control group). 120 men randomized. Ethical approval: Medical Ethics Committee of the Jiangxi Province People's Hospital (approval No. 2011‐clinical inspection 20) with the authorization of the Medical Ethics Committee of the Jiangxi University of Traditional Chinese Medicine Consent: participants provided informed signed consent. Duration of study: 20 October 2011 to 12 February 2012
Participants	Inclusion criteria: males ages > 16 years; redundant prepuce (glans penis covered completely or partially by prepuce under the relaxed condition, and a glans penis that was uncovered with the coronary sulcus exposed by force) or phimosis; and willing to sign informed consent for study participation. Exclusion criteria: penis size or shape incompatible with the tested device; malformations of the glans penis or prepuce, including wide‐range adhesion between glans penis and prepuce; infection and edema of the glans penis or prepuce; suspected penis carcinoma; systemic diseases, such as bleeding tendency, hypoproteinemia, or serious cardiovascular disease; other surgical contraindications; participation in another clinical trial of medical devices or drugs within the past 30 days; and refusal to comply with study protocols. Baseline characteristics: no differences in baseline characteristics between groups. Mean age: 26.6 years ranging from 17 ‐ 67 years with similar number of penile conditions in both groups.
Interventions	Providers: specialist urologists for both groups. Intervention group (novel circumcision device): 60 participants. Lidocaine 1% injection. No mention of skin preparation techniques. Device applied and method executed according to specifications. Foreskin removed. Staples mostly fell off spontaneously by 2 weeks. Dry gauze dressings applied. Control group (surgical circumcision): 60 participants. Standard sleeve dissection technique with blade or scissors. Follow‐up: assessed on the 1st, 3rd, 7th, and 14th postoperative days for all groups
Outcomes	Primary outcomes: surgical success rate (not defined in article, but inferred to be completion of an acceptable circumcision until wound healing) Secondary outcomes: intraoperative bleeding surgery duration pain healing satisfaction
Funding sources	Stated as "Departmental sources"
Declarations of interest	Authors declared no competing interests existed.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Sequence generated using a random table method generated by computer; low risk of selection bias.
Allocation concealment (selection bias)	Unclear risk	Process for treatment allocation concealment not reported; unclear risk of bias.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	No independent outcome assessors and this could have resulted in a high risk of detection bias for the subjective outcomes.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Although unblinded, low risk for detection bias for objective outcomes such as operative time.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Very low attrition in both groups and no expected attrition bias.
Selective reporting (reporting bias)	Unclear risk	No study protocol and it was, therefore, unclear whether there may have been selective outcome reporting. No mention of study registration on a clinical trials database.
Other bias	Low risk	No other biases identified.

Wang 2017.

Study characteristics
Methods	Design: randomized, open‐label RCT conducted in China. Compared the performance of the DCSD (produced by Jiangxi Langhe Medical Instrument Co., Ltd., Jiangxi, China) to FG surgical circumcision. 102 men randomized. Ethical approval: Ethics Committee of Shandong Provincial Hospital Consent: participants provided signed written informed consent Duration of study: June 2013 to December 2014
Participants	Inclusion criteria: men with typical phimosis or redundant prepuce who needed operative treatment; normal development of penis; signed informed consent. Exclusion criteria: penis dysplasia and genital tract malformation; premature ejaculation caused by spermatocystitis, urethritis and prostatitis; without re‐examination and follow‐up visit according to study requirement. Baseline characteristics: no differences in baseline characteristics between groups. Mean age: 28.86 (SD 7.69) in device group and 29.35 (SD 8.18) in surgical group. Similar number of penile conditions in both groups (38 cases of redundant prepuce in device group and 35 cases in surgical group. 13 cases of phimosis in device group and 16 cases in surgical group).
Interventions	Providers: Department of Plastic and Aesthetic Surgery. Authors did not mention the qualifications of staff who provided the procedures. Intervention group (DCSD device): 51 participants. DCSD instrument applied. The 2nd, 6th and 10th migration spots of inner and outer plates of prepuce were raised by hemostatic forceps. 30‐degree angle towards the dorsal part between penis axis and the bar axis of campaniform glans penis base was kept. Ribbon or silk thread used for fixation on rod and to take safety catch out. Application knob was pressed for 10 seconds, to lightly spin and quit the campaniform glans penis base to stop the bleeding. Pressure released after 20 seconds to check the suture situation, with pressure dressing to the incision. Control group (surgical circumcision): 51 participants. Traditional method. Hemostatic forceps clamped the middle frenum of the prepuce flank area, and were pulled for cutting the bilateral prepuce tissue along the parallel site (with a parallel distance of 5 mm) of coronary sulcus. The prepuce frenum tissue was kept at approximately 10 mm. Electrocoagulation applied to stop the bleeding and the incision sutured after detection of active hemorrhage. Follow‐up: assessed at 6 months for all groups
Outcomes	Primary outcomes: operation time intraoperative bleeding volume incision healing time incision infection incision edema pain using VAS incision aesthetic satisfaction sexual function and psychologic states
Funding sources	Not reported
Declarations of interest	Not reported
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization sequence generated using computer software; low risk of selection bias.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment was not reported and, therefore, it was unclear whether there was any possible bias introduced. There were no systematic differences in baseline characteristics.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding of participants and personnel was not possible with this surgical intervention and it was, therefore, unclear whether this may have impacted the performance of either group.
Blinding of outcome assessment (detection bias) Subjective outcomes (severe adverse events, moderate adverse events, minor adverse events, pain, patient satisfaction)	High risk	Given that many of the outcomes were subjective, the lack of blinding may have resulted in a high risk of detection bias.
Blinding of outcome assessment (detection bias) Objective outcomes (operative time)	Low risk	Although unblinded, there was low risk for detection bias for objective outcomes such as operative time.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No attrition; low risk of attrition bias.
Selective reporting (reporting bias)	Unclear risk	No study protocol and it was, therefore, unclear whether there may have been selective outcome reporting. There was no mention of study registration on a clinical trials database.
Other bias	Low risk	No other biases identified.

DCSD: disposable circumcision suture device; FG: forceps guided; GP: general practitioner; RCT: randomized controlled trial; SD: standard deviation; SR: Shang Ring; TK: Tara KLamp; VAS: visual analog scale; WHO: World Health Organization.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Awori 2016	Ineligible study design.
Cheng 2012	Comparison between 2 application methods of 1 device.
Decastro 2010	Comparison between 2 surgical methods of circumcision.
Feldblum 2014	Ineligible study design.
Feldblum 2016	Comparison between 2 application methods of 1 device.
Lei 2015	Ineligible study design.
Liang 2012	Ineligible study design.
Yu 2014	Ineligible study design.
Yue 2012	Article retracted, was published elsewhere the year before.

Characteristics of ongoing studies [ordered by study ID]

ChiCTR‐TRC‐13003428.

Study name	New circumcision standardized application and reproductive health effects, a large multi‐center clinical trial
Methods	Randomized parallel‐group trial
Participants	Inclusion criteria: males ages 18–60 years with redundant prepuce or phimosis with no abnormalities on blood, coagulation or urine routine exam. They should be willing to participate in study and follow study procedures, understand the research and sign an informed consent. Exclusion criteria: foreskin/glans infection, edema, hypospadias, priapism, epispadias, penile torsion, concealed penis, webbed penis, small penis, urethra hypoplasia, hermaphroditism, suspected penile cancer; systemic disease (bleeding tendency, severe cardiovascular disease, diabetes); skin or connective tissue diseases that may affect wound healing.
Interventions	Intervention group: circumcision using the disposable circumcision stapler Control group: traditional surgical circumcision
Outcomes	Primary outcomes: operation time amount of bleeding postoperative pain scores wound healing time incidence of complications appearance satisfaction rate evaluation of sexual satisfaction Secondary outcomes: reproductive health indicators (HIV, human papillomavirus, herpes simplex virus, syphilis)
Starting date	1 January 2014
Contact information	dch0313@163.com; brianzg86@163.com
Notes	Primary sponsor: The First Affiliated Hospital of Sun Yat‐Sen University Awaiting a response from authors to obtain data. Contact made using the email addresses provided in the Chinese Clinical Trial Registry.

Differences between protocol and review

This review is based on a published protocol (Shaik 2016), with differences as described here.

We did not search CINAHL as we considered it an irrelevant database for this review.

Embase access not possible in 2020 as the license was not renewed.

Contributions of authors

AH: screened titles and abstracts, assessed full‐text eligibility, performed data extraction, analyzed results and wrote the final manuscript.

SE: wrote the protocol, developed the search strategy, screened titles and abstracts, assessed full‐text eligibility, performed data extraction, analyzed results and drafted the manuscript.

MZS: conceptualized the project, wrote the protocol, developed the search strategy, screened titles and abstracts, assessed full‐text eligibility, performed data extraction and drafted the manuscript.

TK: reviewed and approved the protocol and manuscript drafts.

Sources of support

Internal sources

• N/A, South Africa

  N/A

External sources

• Department for International Development, UK

  Project number 300342‐104

Declarations of interest

AH: none.

MZS: none.

SE: none.

TK: none.

New