We conducted a prospective trial of neonatal male circumcision in 3 Lusaka sites: the University Teaching Hospital (UTH), and two public-sector primary health care clinics. Neonatal vitamin K is not routinely used in our setting and was not administered to participants in our trial.

Prior to participant enrollment, we developed an NMC training package consisting of didactic lectures, practice on models of neonatal genitalia, and clinical practice. Nine physicians from the UTH were trained to serve as NMC supervisors for provider trainees enrolling in the study. During the study, these supervisors evaluated the competence of the trainees at set intervals, examined infants at review visits, and assessed adverse events.

Provider trainees, including physicians, clinical officers, nurse midwives, and registered nurses. After completing informed consent to participate in the study of NMC provider training, providers were trained in groups consisting of 2-to-4 trainees depending on the availability of trainees. The individual methods were taught sequentially and in blocks. If a supervisor determined that a trainee was not competent in a given method after 10 procedures, then the provider would perform that particular method until competency and before moving on to another method. To ensure variation in the sequence of the three methods among trainees (assuming sequence could influence their competence and preferences among methods), the order of NMC methods for each provider was pre-determined by an independent statistician. (Table 1) Providers were blinded to the sequence. Blinding the postoperative evaluations was not possible since the Plastibell technique involves a small plastic bell remaining on the penis for up to a week and the other two techniques did not

The supervisors evaluated trainees using standard competency checklists (tailored to the procedure type being evaluated) after every five procedures. The trainees completed pre-and post-training questionnaires, to assess their knowledge, attitudes and previous experience with male circumcision, as well as their NMC method preferences (post-training only).

Parents of newborn boys were recruited for the study by peer educators in the post-natal wards of the three NMC sites and other public clinics. One or both parents (if available) or legal guardians of eligible infants completed an informed consent process prior to their infant’s circumcision. All healthy male infants of gestational age >37 weeks at birth, aged 0–28 days, and between 2,500 grams and 5,000 grams were considered eligible to be in the study. Infants with urethral or penile shaft abnormality, local infection, any current illness, or family history of bleeding disorder were excluded. Testing of mothers and infants for HIV was not offered in the study. However all pregnant mothers are routinely tested for HIV under the Ministry of Health policy guidelines.

Participating infants received a dorsal nerve block using 0.8ml of 1% lignocaine for local anesthesia, and were given up to one ml of an oral glucose solution, administered using a pacifier for analgesia during the procedure. Infants were then circumcised using one of the three NMC methods being trained (Mogen, Gomco, or Plastibell). The method performed on a given infant was determined based on which method was being practiced by the attending provider trainee at the time of the procedure, as previously described. For each circumcision performed, information such as provider, location, type of procedure used, procedure start and end time, and individual infant characteristics as well as parental characteristics were recorded on standard clinic forms.

Study staff performed a detailed physical exam on all infants at one week and six weeks following the procedure using standardized follow-up forms to guide their examinations. At infants’ six weeks visits, parents completed a satisfaction survey during which they were asked to rate their level of satisfaction with the procedure outcome using a scale between 0 and 100, where 0 = not at all satisfied, and 100 = completely satisfied. Study participants who missed their scheduled visits were followed up at home.

Any observed complication was referred to study staff and recorded on an adverse events form. We categorized adverse events as mild, moderate, or severe based on criteria previously defined by the WHO(17). Mild bleeding episodes were defined as requiring pressure to control, moderate events required placement of a suture, and severe events were those requiring transfusion. We defined mild infections as requiring local cleaning, a moderate one as needing application of topical antibiotic ointment, and severe ones as requiring systemic antibiotics. A surgical revision was always classified as a severe cosmetic adverse event.

Descriptive statistics were used to summarize provider characteristics, parental and infant characteristics. Differences in provider, parent and infant characteristics between the three neonatal male circumcision methods were compared using the nonparametic Kruskal-Wallis test for continuous variables and the Pearson Chi-square test for categorical variables. Providers ranked the three methods according to preference and ease of use, and the method with the highest rank was tallied. Mean parent satisfaction scores and procedure times were compared across the three circumcision methods, first using a one–way ANOVA model. Pairwise differences were evaluated using the Tukey’s HSD test to control for multiple comparisons.

The classification of the adverse event status for each participant was conducted in two ways. First, participants with missing follow up data were treated as missing at random and removed from the denominator of all calculations. A separate sensitivity analysis classifying these missing patients as having an adverse event provided a conservative estimate of the upper bound for the percentage of adverse events. Adverse events were also tabulated by infant characteristics (such as HIV exposure status, age at procedure, birth and procedure weight) and provider characteristics, including provider type. Although the study was not powered to detect any specific differences in adverse events between the three circumcision methods, a safety analysis was performed for the descriptive comparison of the methods. Data was summarized in terms of percentage of infants experiencing an adverse event at each follow-up visit. Separate comparisons between categorizations of adverse events and between each circumcision method were made using Fisher’s exact test. All statistical analyses were performed using SAS version 9.1.3 (SAS Institute Inc., Cary, North Carolina).

Our sample size was predetermined and based on convenience. Our goal was to train 15–20 providers, who would perform at least 10 circumcisions in each of three methods, for a total of approximately 600 circumcisions (200 per method). Enrollment of infants ended once all healthcare providers had completed at least 30 procedures and reached competency. Some providers completed more than 30 procedures; all procedures were included in the analysis.

The study protocol and informed consent documents underwent continuing ethical review from the relevant Institutional Review Boards at the University of Zambia, the University of Alabama at Birmingham, and the US Centers for Disease Control and Prevention. (ClinicalTrials.gov registration: NCT01115335.)

Between October 27, 2009 and March 24, 2011, we enrolled 17 providers into our study, each of whom performed at least 30 circumcisions (10 of each method). 661 infants were enrolled and 21 (3.2%) were excluded due to ineligibility after enrolling. Infants were excluded from the study if they were older than 28 days, had an infection or illness at the time of the procedure, or if they had pustules or blisters on their body. These infants did not undergo circumcision. (Figure 1) The total number of circumcisions performed was 640. Six of the 17 providers (35%) were male, 5 were doctors, 3 were clinical officers, and 9 were nurses or midwives. Of the doctors, 1 was a surgeon. None of the providers had previous experience performing neonatal circumcision, but 5 had experience with adult circumcision. 6 providers were trained in Plastibell first, 6 providers performed Mogen first, and 5 providers were trained in Gomco methods first.

635/640 (99%) had a one week follow up visit and 630/640 (98%) were seen at six weeks post-procedure. The median age of the infants at time of circumcision was 11 days (IQR 7–18 days). Approximately 11% were less than 48 hours at the time of the procedure. The median birthweight was 3.2 kg (IQR: 2.9–3.5 kg) and the median weight at the time of the procedure was 3.6 kg (IQR: 3.2–4.0 kg). 23.3% (149/624) were HIV exposed, and 17.2% were delivered by cesarean section. None of these infant characteristics differed by circumcision method. (Table 1) Parental characteristics are described in Table 3. The median age of infant’s mothers was 28 years (IQR:24–32 years) and 34 years (IQR: 30–38 years) for the fathers. The majority of mothers reported being married (92.6%). 70% had a household income greater than $200 US Dollars/month. More than half of all fathers (51.0%) were circumcised. Overall 10.7% of mothers were Muslim and 11.3% of fathers were Muslim. Infants circumcised by the Mogen group were more likely to have a Muslim mother (p=0.001), and have a Muslim father (p<0.001).

For the 613 mothers and 64 fathers who completed a satisfaction survey, mean satisfaction score for mothers and fathers was 96.4 (SD 7.5) and 95.9 (SD 8.2), respectively, based on a scale from 0 (very dissatisfied) to 100 (very satisfied). Parental satisfaction scores did not differ significantly by method and no pair-wise differences between any two methods was found.

11/17 providers (65%) preferred the Mogen clamp to the Gomco or the Plastibell. (Table 3) The majority of providers (15/17, 88%) also found that the Mogen clamp was the easiest to perform. (Table 3) The mean duration of the procedures differed significantly by method and was 11.5 minutes (SD: +/− 6.7) for the Mogen clamp, 13.9 minutes for the Gomco (SD=+/−6.3), and 11.9 minutes for the Plastibell (SD: +/−4.0)(p<0.001). Tukey-adjusted pairwise comparisons indicated that the duration of the Gomco procedure was 2.4 minutes longer (95%CI: 1.1–3.8 minutes) compared to the Mogen and 2.0 minutes longer (95%CI:0.7–3.4 minutes) compared to the Plastibell procedures. The mean number of procedures to competency was 10.3 (SD:+/−3.3) for Gomco, 10.3 (SD:+/−3.7) for Plastibell and 8.9 (SD:+/−2.9) for the Mogen clamp. All providers were competent in each circumcision method by fifteen procedures. In general, nurses took longer to train than the other providers but this was not statistically significant.

Treating participants with missing follow-up data as missing at random, the overall adverse event rate was 31/630 (4.9%, 95% CI: 3.4%–6.9%). When patients missing a follow-up visit were classified as having an adverse event, the overall adverse event rate was 41/640 (6.4%, 95% CI: 4.6% − 8.6%). The 31 documented adverse events included 5 (16%) which were mild adverse events, 10 (32%) which were moderate events, and 16 (52%) which were severe events. All of the mild events were bleeding events requiring pressure to stop. The ten moderate complications included bleeding needing either Surgicel or a suture to attenuate. The sixteen severe complications included 11 cosmetic issues requiring revisions, one retained Plastibell that required removal in the theatre, three infections requiring intravenous antibiotics and hospitalization, and one transfusion for bleeding in a patient who was delayed on the way to the tertiary hospital. The overall adverse event rate did not differ by method (p=0.609). There were more cosmetic complications with the Mogen clamp compared to the other methods. (Table 4)