eResults. Analyses Maximizing Specificity of Data (eTable 2 and eTable 3)
eTable 1.ICD-9 Codes, CPT Codes, and Risk Windows of 41 Possible Male Circumcision Adverse Events Used in Initial Screening Analysis
eTable 2. Incidence Rates, Incidence Risk Ratio, and Incidence Rate Difference of Adverse Events Between Circumcised and Uncircumcised Newborn Males (Charge Data Master 2 [CDM2] and Outpatient Data With Exact Dates for Procedures, United States, 2005-2010)
eTable 3. Counts, Incidence Rates, Incidence Risk Ratio, and Incidence Rate Difference of Adverse Events Between Males Circumcised at 3 Different Age Groups (Charge Data Master 2 [CDM2] and Outpatient Data With Exact Dates for Procedures, United States, 2005-2010)
El Bcheraoui C, Zhang X, Cooper CS, Rose CE, Kilmarx PH, Chen RT. Rates of Adverse Events Associated With Male Circumcision in US Medical Settings, 2001 to 2010. JAMA Pediatr. 2014;168(7):625–634. doi:10.1001/jamapediatrics.2013.5414
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Approximately 1.4 million male circumcisions (MCs) are performed annually in US medical settings. However, population-based estimates of MC-associated adverse events (AEs) are lacking.
To estimate the incidence rate of MC-associated AEs and to assess whether AE rates differed by age at circumcision.
We selected 41 possible MC AEs based on a literature review and on medical billing codes. We estimated a likely risk window for the incidence calculation for each MC AE based on pathogenesis. We used 2001 to 2010 data from SDI Health, a large administrative claims data set, to conduct a retrospective cohort study.
Setting and Participants
SDI Health provided administrative claims data from inpatient and outpatient US medical settings.
Main Outcomes and Measures
For each AE, we calculated the incidence per million MCs. We compared the incidence risk ratio and the incidence rate difference for circumcised vs uncircumcised newborn males and for males circumcised at younger than 1 year, age 1 to 9 years, or 10 years or older. An AE was considered probably related to MC if the incidence risk ratio significantly exceeded 1 at P < .05 or occurred only in circumcised males.
Records were available for 1 400 920 circumcised males, 93.3% as newborns. Of 41 possible MC AEs, 16 (39.0%) were probable. The incidence of total MC AEs was slightly less than 0.5%. Rates of potentially serious MC AEs ranged from 0.76 (95% CI, 0.10-5.43) per million MCs for stricture of male genital organs to 703.23 (95% CI, 659.22-750.18) per million MCs for repair of incomplete circumcision. Compared with boys circumcised at younger than 1 year, the incidences of probable AEs were approximately 20-fold and 10-fold greater for males circumcised at age 1 to 9 years and at 10 years or older, respectively.
Conclusions and Relevance
Male circumcision had a low incidence of AEs overall, especially if the procedure was performed during the first year of life, but rose 10-fold to 20-fold when performed after infancy.
The American Academy of Pediatrics updated its guidance on male circumcision (MC) in 2012 to state that “the procedure’s benefits justify access to this procedure for families who choose it.”1(p585) Whether MC should be considered an important public health intervention in the United States and other developed countries based on the results of 3 randomized controlled trials2- 4 showing its human immunodeficiency virus protective effect has been debated.5- 8 A key aspect of this debate is the rate of adverse events (AEs), especially serious ones, attributable to MC for males circumcised as infants and for males undergoing voluntary circumcision as adults or adolescents.
Numerous studies9- 21 have reported on MC mild to severe AEs, ranging from 0.0008% to 3.6% in infants and from 0.9% to 8.8% in adults. However, most of these studies were based on small samples, a single clinical site or state, cross-sectional data, or nonrepresentative cohorts.22 While the review by Weiss et al22 and a case series23 describing the experience of one pediatric urologist conducting Gomco circumcision in 150 neonates and infants suggested generally higher rates of AEs with older age at MC, none to our knowledge have compared rates of AEs across all age groups at MC (from neonates to adults) in the same study. To provide stakeholders with better population-based information on the risk of MC AEs, we used a large administrative claims data set (1) to estimate the incidence rate of MC-associated AEs via comparison of the incidence risk ratio (IRR) and the incidence rate difference (IRD) of AEs between circumcised and uncircumcised newborn males and (2) to compare the IRR and IRD of AEs associated with MC across age groups (<1 year, 1-9 years, and ≥10 years).
This research was determined to be exempt from institutional review board evaluation because it entailed secondary analysis of administrative data procured from SDI Health (http://sdihealth.com/portal/site/imshealth) without personal identifiable human participants. SDI Health originally collected these data from the processing of US health care insurance reimbursement claims. The SDI Health data include International Classification of Diseases, Ninth Revision (ICD-9) and Current Procedural Terminology (CPT) codes and are available about 2 months after clinical visits. SDI Health creates a unique anonymous identifier for each patient, enabling individuals to be followed up longitudinally.
The Charge Data Master (CDM) is SDI Health’s inpatient data set. It gathers data from a 20% convenience sample of all inpatient encounters of short-stay, acute care, and nonfederal hospitals from 48 states and Washington, DC, representing approximately 120 million unique hospitalized patients. The CDM hospitals are located in all US regions (25% in the East, 12% in the North, 45% in the South, and 16% in the West; data on regions are unspecified for 2%). Of these, 85% are urban hospitals, and 36% are teaching hospitals, with a wide variability of bed size (median size, 200-299 beds). Of patients seen at these hospitals, about 10% are covered by Medicaid and 30% by Medicare, and the remainder are covered by third-party payers. The CDM data are formed by 2 data sets, the CDM1 and the CDM2. The CDM1 (approximately 80% of the CDM) has been available since 2001 and is updated monthly; only the month of a diagnosis or procedure is provided, with the date of discharge defaulted to the first day of the discharge month. The CDM2 (approximately 20% of the CDM) has been available since 2005 and is updated weekly; unlike the CDM1, the exact discharge date of a diagnosis or a procedure is available. SDI Health also collects data from more than 870 000 unique outpatient medical providers, including the exact dates of diagnoses and procedures. For this study, we used CDM data available through February 2010.
Possible MC AEs for this analysis were identified from (1) a review of PubMed using the search terms circumcision and adverse events and (2) the ICD-9 and CPT manuals for conditions that are not necessarily due to but could be related to MC. Our search yielded 41 possible MC AEs, which we classified into 10 clinical syndromic groups (eTable 1 in the Supplement).
For each of 41 possible MC AEs, one of the coauthors who is a board-certified pediatric urologist (C.S.C.) a priori defined the likely risk window in days based on pathogenesis (eTable 1 in the Supplement). The possible AEs were further classified by C.S.C. as potentially serious (italicized in eTable 1 in the Supplement) or not based on clinical judgment and assuming a worst-case scenario. We edited the CDM MC data set by (1) removing circumcised males who had an MC date before their birth date and (2) reclassifying as circumcised newborn males who did not have an MC record but had an MC-specific AE (CPT code 54162 or 54163).
We performed a retrospective cohort study using log binomial regression modeling (SAS 9.2; SAS Institute Inc) to ascertain the risk associated with MC. We first calculated the incidence of each AE during its risk window per million circumcised (and separately for uncircumcised) newborn boys using the discharge date of circumcision (or birth for uncircumcised) for the beginning of the risk window. We then calculated the IRR, IRD, and their respective 95% CIs between the circumcised and uncircumcised groups.24
To minimize potential confusion on causal relationships in this exploratory study, the AEs and person-times outside of the risk window in circumcised males were not considered for the analysis (instead of being included in a group unexposed to medical procedures, as done in another risk window safety study25). An AE was considered probably related to MC if the IRR significantly exceeded 1 at P < .05 or occurred only in circumcised newborn males. Multiple comparisons were not adjusted for in our analysis because almost all the significant associations found were at P < .001, which is less than any typically used correction factor (eg, Bonferroni adjustment). To estimate the total incidence of AEs associated with MC, we calculated the IRD between the incidences of probable AEs in circumcised vs uncircumcised newborns using unduplicated counts of boys who had 1 or more AEs in each group divided by the number of circumcised and uncircumcised newborns, respectively. For some syndromic groups, the risk window was not equal for all AEs. To obtain the total for the syndromic group in this case, all conditions were followed up for the longest risk window in the group. The IRR and 95% CI were then generated.
We assessed whether rates of probable MC AEs differed in the following 3 age groups: males circumcised at younger than 1 year (reference group), age 1 to 9 years, or 10 years or older. The age groups’ cutoff points separated infants from children before puberty and older males. We used the same statistical approach as above to calculate the incidence per million MCs (PMMCs), IRR, IRD, and 95% CIs.
To better detect rare MC AEs, we first conducted an analysis using all available data, including the CDM1, CDM2, and outpatient data sets. Because the CDM1 day of discharge was defaulted to the first day of the month, all AE risk windows of less than 28 days were reset to 28 days, the shortest risk window that could possibly be tracked and the closest to a complete month, in this analysis (Table 1 and Table 2). We then conducted a second analysis maximizing specificity of date by using only the CDM2 and the outpatient data, the 2 data sets with exact dates for each procedure needed for the exact risk window analysis (eResults, eTable 2, and eTable 3 in the Supplement).
From 2001 to 2010, a total of 1 400 920 MC reimbursement claims for males of all ages were submitted from US hospital settings and available to SDI Health (CDM1, CDM2, and outpatient data). Forty-seven males (0.003%) had an MC dated before their birth date, and these records were removed from the analysis. Of all newborn boys, 346 (0.015%) had an MC-specific AE but did not have an MC record. These were reclassified as circumcised newborn males.
Data were available for 2 339 760 newborn male births. Among these, 1 306 812 (55.9%) were linked to a circumcision record. Of the initial 41 possible MC AEs, 16 (39.0%) met the criteria for probable MC AEs (underlined in Table 1). Six probable MC AEs occurred only in circumcised newborns and not in uncircumcised newborns (amputation of penis, partial; replantation of penis; lysis or excision of penile postcircumcision adhesions; repair of incomplete circumcision; stricture of male genital organs; and suture of artery). Among 16 probable AEs, 10 were also classified as potentially serious.
There were 4924 newborns with 1 or more probable AEs. Of these, 4059 were circumcised, 865 were not. In total, 5385 and 1100 AEs were recorded among circumcised newborns and uncircumcised newborns, respectively. Of the 4924 total, 4523 (91.9%) were cared for in a hospital setting and 401 (8.1%) in an outpatient setting. The estimated incidence of probable AEs associated with MC was less than 1% for crude data (4.059/1 306 812 = 0.31% [95% CI, 0.30%-0.32%]) or with adjustment for the background rate (4059/1 306 812 − 865/1 032 948) = 0.23% (95% CI, 0.21%-0.24%).
The IRDs for potentially serious probable AEs ranged from a low of 0.76 (95% CI, 0.10-5.43) persons with stricture of male genital organs PMMCs to a high of 703.23 (95% CI, 659.22-750.18) persons with repair of incomplete circumcision PMMCs. The most common probable MC AE was division of penile adhesions, with an IRD of 199.69 (95% CI, 153.92-245.66).
Nine AEs were significantly less likely to occur in circumcised infants compared with uncircumcised infants at P < .05. Circumcised newborn males had a higher risk for wounds, correctional procedures, inflammation, and bleeding compared with uncircumcised newborn males but had a lower risk for surgical procedures, penile disorders and gangrene, pneumothorax, and infections. Among the extremely rare but serious AEs occurring only among circumcised newborns (but once or none among uncircumcised newborns), we found 0 cases of complete amputation of penis, 1 case of stricture of male genital organs, 3 cases of partial amputation of penis, 4 cases of replantation of penis, and 16 cases of suture of artery.
Of 1 400 920 circumcised males, 1 335 180 (95.3%) were circumcised during infancy (<1 year of age) (Table 2). Another 28 197 boys (2.0%) were circumcised at age 1 to 9 years, and 37 543 (2.7%) males were circumcised at 10 years or older (of whom 8590 [22.9%] were aged 10-18 years). The incidences of probable AEs varied by age group, namely, 0.40% (95% CI, 0.39%-0.41%) among boys circumcised during infancy, 9.06% (95% CI, 8.73%-9.40%) among boys circumcised at age 1 to 9 years, and 5.31% (95% CI, 5.09%-5.55%) among males circumcised at 10 years or older. The incidences of probable AEs were approximately 20-fold and 10-fold greater for males circumcised at age 1 to 9 years and at 10 years or older, respectively.
Except for the comparisons in which no AE cases occurred in one or both of the older age groups, the IRR of each of the other studied AE comparisons significantly exceeded 1, and the IRD exceeded 100 PMMCs (except for suture of artery) when MC was performed after the first year of life. The highest IRR among boys circumcised at age 1 to 9 years was for division of penile adhesions (IRR, 67.64; 95% CI, 61.98-73.81). The highest IRR among males circumcised at 10 years or older was for other inflammatory disorders of penis (IRR, 112.06; 95% CI, 93.88-133.75). While these are not explicitly defined in the ICD-9 manual, they can be skin conditions such as infections, cellulitis, abscess, boil, carbuncle, or cavernitis.
We studied the AE outcomes after approximately 1.4 million MCs in the United States, about 10-fold more than the largest prior studies.9,10 Using a broad definition of 41 possible MC AEs to search a large medical administrative database and then restricting to 16 probable MC AEs with significantly elevated rates in predefined risk windows or occurring only in circumcised persons, we estimated the incidence of AEs associated with newborn MC in medical settings adjusted for the background rate to be less than 0.5% (0.30% for the more specific CDM2 data set). Overall, the most common probable MC AEs were related to correctional procedures (approximately 2000 PMMCs) and bleeding (approximately 1000 PMMCs). Our findings were largely similar irrespective of whether the month-specific or date-specific data sets were used and were consistent with earlier US studies9,10 given differences in methods.
Our findings also suggest that many AEs, such as penile reconstruction, pneumothorax, and infections, occur less frequently in circumcised males, perhaps due to a healthy infant bias: those newborns who undergo MC are more likely to be healthier (and without such disorders) compared with their uncircumcised counterparts. This type of selection bias is commonly seen in nonrandomized observational investigations of outcomes after medical procedures26,27and resulted in the observed lower rate of AEs among circumcised males.
We found that the incidence of MC AEs was 10-fold to 20-fold higher when performed among older age groups compared with infancy. These findings are consistent with earlier studies22,23 and may provide for the first time a direct measure of the relative difference in AE rates by age at MC. Recent data on MC AEs from a clinical trial28 in Kenya that included males 12 years or older showed similar high rates of AEs for this age group. In a study29 of infant MC AEs in Kenya, an increased risk for AEs was found if MC was performed in the second month of life compared with the first month. The indications for MC among older age groups in the United States may be more medical in nature (eg, infections or adhesions) than the cultural or religious basis in most routine healthy newborns; however, future studies will need to carefully adjust for this potential source of confounding.
The incidence of penile amputation was highest (0.17%; 95% CI, 0.13%-0.21%) among males circumcised at 10 years or older. In total, the absolute number of penile amputations in our database was 71. Most penile amputations captured in our data set (45 of 71) were recorded using ICD-9 code 643.0, which does not differentiate complete from partial penile amputation. Of 71 recorded penile amputations, 3 were coded as complete, namely, 1 among boys circumcised in infancy and 2 among males circumcised at 10 years or older. Wiswell and Geschke9 reported an absence of total penile amputations during 5 years in a study of MC AEs among newborns from US Army hospital settings. Consistent with these findings, our data captured less than 1 total penile amputation PMMCs, suggesting that most penile amputations recorded using ICD-9 codes and captured in our data set were likely to be partial. Without access to primary medical records, we can only speculate that the 4 patients who had penile amputation in the uncircumcised population likely were miscoded or circumcised at nonmedical settings, or they may have been patients undergoing operative intervention for severe genital anomalies. Other studies30,31 have reported on the success of treatment, including replantation, in the case of penile amputation. We could not study mortality potentially related to MC because deaths in general are not captured in health care reimbursement claims databases, such as SDI Health. In the earlier review, Wiswell and Geschke9 reported 3 deaths due to MC from 1954 to 1989 (approximately 0.08 deaths from neonatal MC in the United States per year).
Our study has several potential limitations. First, most of our data (approximately 80%) assign a discharge date of the first day of the month for the medical record. Hence, if an AE has a risk window of less than 28 days and falls in the same month of the MC, it will be counted even if it occurs outside of the risk window. Also, in the case in which an AE has a risk window of less than 28 days and is encountered during the month following MC, it will be missed. The first scenario tends to overcount some AEs, while the second scenario tends to undercount some others. However, limiting our analysis to data with exact discharge dates, our findings remained almost unchanged (eResults, eTable 2, and eTable 3 in the Supplement). At the same time, some of the males circumcised within the last year of our data might have experienced an AE within a risk window outside of the available data. This may have decreased our overall rate of AEs by a small fraction.
Second, if an AE occurred on the same day as MC, it is impossible to determine whether the AE occurred before or after the circumcision. Indeed, certain AEs can also be an indication for MC. Hence, our reported rate might be inflated in case some AEs were diagnosed on the same day as MC or before MC.
Third, our data may not be generalizable to the entire US population because they are derived from a convenience sample. However, the large volume of administrative SDI Health data used in this study (approximately 20% of US hospital discharges and >870 000 unique outpatient medical providers) strengthens our findings. A 2011 study32 showed that the trends in neonatal MC in SDI Health data were virtually identical to those of 2 nationally representative data sets, further supporting the validity of our findings, at least for newborn boys.
Fourth, our data were collected for billing purposes only. If a circumcision or an AE was not covered by a third-party payer, it would be missing from this analysis. Also, some circumcisions might occur in nonmedical settings, such as religious MC, but a resulting AE might require medical intervention and be captured as occurring among uncircumcised newborns. Indeed, some uncircumcised newborn males in our data had an MC-specific AE. However, these did not exceed 0.01% of all newborns, and the incidence of AEs in our analysis was in the range of those from previous US publications.9,10 Therefore, while the true rate may be lower or higher than our estimates, billing records should capture most MC procedures.
Fifth, MC can occur concurrently with other operative procedures for reasons of anesthesia convenience. The AE that might result in these cases may be confounded by other health conditions of the patient. Future studies overcoming these limitations and examining other databases to confirm our findings are needed to better estimate specific AE rates attributable only to MC.
Our data suggest that the rate of AEs associated with newborn circumcision is less than 0.5%. Most important, the incidence of AEs increased substantially when MC occurred after the first year of life. Given the current debate about whether MC should be delayed from infancy to adulthood for autonomy reasons,33 our results are timely and can help physicians counsel parents about circumcising their sons.
Accepted for Publication: November 22, 2013.
Corresponding Author: Charbel El Bcheraoui, PhD, Institute for Health Metrics and Evaluation, University of Washington, 2301 Fifth Ave, Ste 600, Seattle, WA 98121 (email@example.com).
Published Online: May 12, 2014. doi:10.1001/jamapediatrics.2013.5414.
Author Contributions: Drs El Bcheraoui and Zhang had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: El Bcheraoui, Cooper, Kilmarx, Chen.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: El Bcheraoui.
Statistical analysis: Zhang, Rose.
Administrative, technical, or material support: All authors.
Conflict of Interest Disclosures: None reported.
Disclaimer: The findings and conclusions in this study are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Additional Contributions: Farid Khan, MPH, Laurel Edelman, BS, Joel Greenspan, MD, and Ed Burleigh, MBA, from SDI Health answered our inquiries about the data. Kris Greiner, BA, from the Division of Pediatric Urology, The University of Iowa, formatted the manuscript to adhere to the style of JAMA Pediatrics. Sanjyot Shinde, PhD, Deborah Gust, PhD, and Charles LeBaron, MD, from the Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, reviewed the manuscript and provided valuable comments.