Male circumcision is a common surgical procedure, but few epidemiological studies have reported frequency of adverse events, most commonly bleeding and infection. Our review shows that serious adverse events are rare, but there is wide variation in reported frequencies of adverse events following circumcision. This is likely to be due to several factors directly associated with complications such as age at circumcision, training and expertise of the provider, the sterility of the conditions under which the procedure is undertaken and the indication (medical/cultural) for circumcision. In addition, there is variation due to methodological issues such as duration of follow-up, epidemiological study design, and definition of complications.
In general, complications (reported by parents) occur least frequently among neonates and infants than among older boys, with the majority of prospective studies in neonates and infants finding no serious complications, and relatively few other adverse events, which were minor and treatable. The prospective studies in older boys also found virtually no serious adverse events, but a higher frequency of complications (up to 14%) even when conducted by trained providers in sterile settings [47
]. The lower frequency of complications among neonates and infants is likely to be attributable to the simpler nature of the procedure in this age group, and the healing capability in the newborn. Further, a major advantage of neonatal circumcision is that suturing is not usually necessary, whereas it is commonly needed for circumcisions in the post-neonatal period. This advantage is illustrated by the US study in which no complications were seen among 98 boys circumcised in the first month of life, but 30% of boys aged 3-8.5 months had significant postoperative bleeding [24
]. There are alternatives to suturing, either with the disposable clamps, or with alternatives such as cynoacrylate glue [44
] and further research in this area is needed.
Several studies stress the importance of careful training and experience of the provider, and the sterility of the setting. This was most clearly noted in a Nigerian study [27
] in which 24% of boys had reported complications (including retention of excess residual foreskin), but only 1.6% of those circumcised at the public (University Teaching) hospital by medical doctors. Similarly, two case-control studies from Israel have found that UTI are 3-4 times more likely to occur following circumcised by a traditional, rather than medical provider [57
]. However, as noted in our review, neonatal circumcision following traditional circumcision in Israel has low complication rates overall [9
]. A further example is the study from the Comoros Islands which reported results of an exercise in which specific training had been given to surgical aids and nurses to perform circumcisions. The proportion of boys with complications (2.3%) was reported to be a great improvement on that by traditional non-medically trained providers [43
]. The high frequency of adverse events following circumcision by untrained providers in non-sterile settings is striking in two studies of traditional circumcision which found alarmingly high prevalence of around 80% [54
]. Notably, in one of these, the self-reported frequency was much lower, illustrating the under-ascertainment that can occur in retrospective studies. Mass circumcisions are particularly risky, even when undertaken in the hospital. For example, the Turkish study of 700 children circumcised during a 5 day period recorded a complication frequency of 8%, likely due to the difficulty in providing sufficient sterile equipment and conditions [46
]. The reason for surgery can also influence the risk of adverse events as seen in the studies of child circumcision where more complications were generally seen if circumcision was conducted for medical rather than religious reasons.
Our systematic review was restricted to circumcision complications among boys aged 12 years or under. However, there are several published studies of circumcision complications among adolescent and adult men (Table ) and these indicate a generally higher frequency of complications than seen in neonates, infants and children. In the three RCTs of circumcision in adult men, complications were observed in 2-7% of HIV-negative men [14
], and in 6-8% of HIV positive men [14
]. The most detailed observational study was conducted among the Babukusu ethnic group in western Kenya. Of 562 adolescents circumcision by a medical provider (or reported as such), 18% had a complication, as did 35% of boys circumcised traditionally [60
]. A sub-study in the same population directly observed 24 boys undergoing medical and traditional circumcision respectively and found that of those circumcised medically, only one boy had no adverse events, and 3 permanent adverse sequalae were reported, including one very serious life-threatening case by a 'medical' practitioner who was later found to have no documented medical qualifications [60
]. Among the 12 directly observed traditional circumcisions, complications were seen in 10 boys (83%), and 4 (33%) were judged to have permanent adverse sequelae. None had fully healed by 30 days post-operation. Detailed examination showed that traditional circumcision was also associated with slower healing, more swelling, laceration and keloid scarring [60
]. These results show that under non-sterile conditions, adolescent and adult circumcision can frequently be associated with severe complications. Other case-series of circumcision complications among adolescents and young men also report severe morbidity and mortality [63
]. Reported complications tend to be more common in this age group than for neonates and infants, even when circumcision is conducted under the 'gold standard' conditions such as in the RCTs.
Frequency of complications in studies of adolescent and adult circumcision
A major challenge in our review was to standardise the definition of complications. For example, Okeke et al [10
] report complications in 20% of boys, of which half were excessive residual foreskin - an adverse event but arguably not a medical risk. We excluded these cases where possible. Similarly, the paper by Gee et al [38
] cites a total of 110 complications out of 5521 (2.0%) but states that only 14 complications (0.2%) were considered 'really significant' (one life-threatening hemorrhage, 4 systemic infections, 8 circumcisions of infants with hypospadias and one complete denudation of the penile shaft). The other complications included bleeding, infection, circumcision of hypospadiasis, and a Plastibell ring that was too tight. The problem of defining complications is also highlighted in the early (1961-1962) study from Canada in which moderate or severe complications (bleeding, infection, meatal ulcer, meatal stenosis and phimosis) were seen in 15 infants (15%) but a further 68 infants had mild bleeding, meatal ulcers or infection [29
]. Complication risks in this study have previously been reported as 55% [4
], which includes any bleeding, including oozing. A further example is the Australian study [69
] which reported complications in 8% of boys, which included several cases of mild bleeding which either ceased spontaneously or with simple management such as digital pressure. We have attempted to report 'severe' or 'serious' adverse events as a separate outcome, but data on this is often limited and it would be useful to produce a standard classification of mild, moderate and severe complications following circumcision so that in future studies may be more easily comparable. Other limitations related to the design of the epidemiological studies. The length of follow-up varies between, and within, studies, and may affect the estimated frequency of complications. For this reason we tend not to term the frequency as a 'risk'. It is also possible that the lower frequencies of complications in prospective studies are due to improved procedures by practitioners or improved hygiene by patients as a result of participating in the study. Finally, a number of studies are small and the estimates of frequency of complications will be correspondingly imprecise.
We excluded one study of circumcision among patients with inherited bleeding disorders [20
] as we were interested in complications in general populations. In this study, of 71 patients diagnosed from 1961-1996, 52% had a record of post circumcision bleeding. In many settings, boys are not asked about a family history of bleeding disorders and this can potentially lead to severe circumcision-related complications.