THE CIRCUMCISION REFERENCE LIBRARY
Departamentos de Nefrologia y Epidemiologia Clinica,
Hospital de Pediatriá,
Centro Medico Nacional, Siglo XXI
Instituto Mexicano del Seguro Social,
Mexico, D.F., Mexico.
Neonatal circumcision may decrease the incidence of urinary tract infection (UTI), and so the literature was reviewed to address their methodological limitations of the studies, and to analyze the data in individual and grouped form. A systematic search of the publications on circumcision and UTI was conducted in Index Medicus (1975-1991) and Medline (1988-1991). Six papers were included in the meta analysis because these presented original data obtained from groups of babies. All of the selected articles were considered in individual and grouped form to calculate odds ratio (OR) and confidence interval (Cl) at 95%. The number of patients included in each study ranged from 112 to 219,775. Clustering of the articles enabled us to obtain a total sample number of 221,799 infants. In each article there was a higher risk of UTI in intact patients (OR from 10.82 to 156.42). The global risk obtained from the six studies was of 13.05 with a CI (95%) ranging from 10.86 to 15.70. Intact male infants have low risk of UTI in the first year and the risk may decrease even more with circumcision. This conclusion may not be considered as definitive because of the methodological limitations of the reviewed studies. Recommendation of routine circumcision of all newborns is not justified with the data. Urinary tract infection; neonatal circumcision; meta-analysis.
Solicitud de sobretiros [Reprint requests. - CIRP]:
Dr. Dante Amato Martinez,
Jefatura de Servicios de Investigacion Medica.
Instituto Mexicano del Securo Social,
Bloque "B" Unidad de Congresos, Centro Medico Nacional,
Siglo XXI, Av. Cuauhtemoc 330.
Col. Doctores, C.P. 06720,
Mexico, D.F., Mexico.
In 1975 the Ad Hoc Task Force on Circumcision stated that "There is no absolute medical indication for routine circumcision in the newborn period".1 In 1983 The American Academy of Pediatrics and the American College of Obstetricians and Gynecologists in their joint publication Guidelines to Perinatal Care supported this viewpoint.2 Nevertheless, circumcision continued to be done on the large majority of infants in the United States of America after the above statements were published3. It seems though in the recent years that the frequency of this procedure has somewhat diminished4.
In 1989 The Task Force on Circumcision5 published a different statement in which the original recommendation was substantially changed. The new position stated that "circumcision of the newborn male infant has medical benefits and potential advantages, but there were concerns and risks". The benefits and advantages alluded to are the prevention of phimosis, paraphimosis, and balanoposthitis, a decrease in the incidence of penile cancer in circumcised males7 and of cervico-uterine cancer in their female partners8, a decreased risk of acquiring sexually transmitted disease (STD)9 and the diminished frequency of urinary tract infections among infants10 with possible reduction in the rate of chronic terminal renal failure11,12.
The attended risks are infections, hemorrhages, meatitis with ulcers or stenosis and loss of the glans or the whole penis.
Several reviewers of the available evidence on this issue have stated that in the absence of well-designed prospective trials, the nexus between circumcision and UTI cannot be considered definitive5,14,15.
Meta-analysis is a epidemiological tool which combines the results of different studies, with the intent of obtaining validity about diagnosis, therapeutic efficacy and risk factors16,17. This method initially was only used to evaluate therapeutic efficacy and was exclusively considered in the domain of clinical experiments. Indeed the technique demonstrated great promise in observational studies18,19. The odds ratio is a good estimate of relative risk when the disease of interest is rare. The relative risk is the ratio between two rates of incidence: the incidence of developing disease in those subjects exposed to a attended risk compared to the incidence of developing disease in those subjects not exposed to attended risk20.
Our evaluation of the literature on circumcision and UTI was conducted to demonstrate the methodological limitations of the published studies, to the group the results by meta-analysis and to assess the cost benefits of circumcision.
A systematic search was conducted of the relevant articles listed in Index Medicus from January 1975 to June 1991, as well as data from the years 1988 to 1991 included in the Medline computer search system with infant circumcision and UTI as index items. In the selection of studies methodological characteristics were taken into account. In the selection process the authors were blinded to the final results and conclusions. Unpublished articles were not sought for; neither were abstracts of studies presented at conferences, or conclusions of these; since it was considered that the incorporation of this sort of information would diminish the precision of the data21. Only those studies incorporating a comparison group were included. Descriptive studies were excluded since these do not allow the calculation of the risk factors. Only seven studies with original patient data were found3,4,10,13, 22-24. Of these only six were eligible; one was excluded as it had no comparison group. Two3,4 of the studies were grouped together as the patient population was the same. All the studies were of an observational type in which the cohorts were assessed prospectively as well as retrospectively. As our objective was to avoid of the biased choice of published estimates, it was decided to calculate the actual estimates of risk in confidence intervals using the crude data as the authors presented them. It is important to recognize that none of the studies focused on the risks of the circumcision procedure.
The meta-analysis was divided into two phases: in the first phase a qualitative analysis was done with the intent of evaluating the quality of the eligible studies. In this regard the criteria of quality as proposed by Jenicek were used. In the second phase a quantitative analysis was performed to calculate the risk of UTI in the intact state.
The studies were considered in individual and group form for the calculation of odds ratios and 95% confidence intervals. The method of Miettinen was used. The statistical significance was tested by the method of Mantel and Haenszel25.
Of the studies analyzed, five were retrospective, and one had a prospective design . The number of babies enrolled ranged from 112 and 219,775. The grouping of the studies obtained a sample of 221,799 infant (Fig. 1). In all the individual studies, a risk of having UTI was at least ten times in the intact babies compared to circumcised group. The odds ratio for the individual studies varied from 10.822 to 156.4223 and was statistically significant in all cases; moreover in no instance did the confidence interval exceed the value of 1 (Fig. 1). The global odds ratio, obtained by grouping the results, was 13.05 with a 95% confidence interval ranging from 10.86 to 15.70 (From the methodological point of view, the optimal design for supporting the hypothesis that circumcision in the newborn male infant reduces the frequency of UTI, is a controlled clinical trial. This design will consists of a cohort of newborns being randomly allocated circumcision or not. The infants must be observed for a certain time to register the episodes of UTI. The method of observation must be sufficiently good in order to guarantee that no UTI episode escapes detection. This design, simple in appearance, has two difficult problems in its application. If the Wiswell data are accepted3,4, one may expect a frequency of UTI [in US hospitals] in about 1% of intact infants, and that the frequency of UTI is reduced to about 0.1% with circumcision. The hypothesis is that a surgical intervention will reduce the frequency of an event from 1 to 0.1%, wi accepting the null hypothesis, although in reality this does not exist (type 1 error of 5%), and a probability of rejecting the difference, although in reality this does exist (type 2 error of 10%), it will require a large number of babies i.e., around 1,400 in each group26. This quantity of subjects creates logistical problems (to achieve a great number of interventions and to follow-up patients over a year's time) and the inherent costs. The second factor that affects the feasibility of such a study is the problem of getting the parents of the babies to remain in either group randomly, particularly if it is contrary to their personal preference. These difficult issues of feasibility will make it unlikely that in the near future such a study will be performed. Therefore, the best evidence that can be drawn from the data is that UTI in the young infant is an event of low frequency which can be further reduced (between 11 and 16 times) with circumcision. if one calculates the cost of circumcision in public health institutions; and we assume the cost to be $150; the operation on a group of 1,000 infants would lead to a cost of $150,000. In this cohort, one would expect to avoid nine episodes of UTI, at an approximate cost of $17,000 per episode. One would expect one case of UTI in the circumcised group and two cases of complications resulting from circumcision, such as local infection or hemorrhage.
Figure 1. Author a b c d e Wiswell 1985 2,502 1,919 583 7(0.21%) 24 (4.12%) Wiswell 1986 2,019 1,575 444 0 8 (1.80%) 217,116 175,317 41,799 193(0.11%) 468 (1.12%) Wiswell 1987 219,775 173,663 46,112 151(0.09%) 459( 0.99%) Wiswell 1989 136,086 100,157 35,929 20(0.02%) 88 (0.24%) Herzog 1989 112 52 60 0 36 (32.1%) Crain 1990 139 36 103 4 2.88% 18 (12.9%) a. total number of males in the study; b. number of circumcised infants; c. number of intact babies; d. number of circumcised babies with UTI; e. number of intact babies with UTI. Odds ratio ]i[, confidence interval of 95%( ] ____ ___ [ ) for each of the studies analyzed and then combined for global representation The studies referred to : Wiswell 198527, Wiswell 19863, Wiswell 1987,4 Herzog 1989,25 Wiswell 198913 and Crain 199024. *P
Recently, Chessare27 applied decision analysis and came to the conclusion that in the group of US parents studied, the preference would be continue to be not to circumcise unless the anticipated frequency of UTI during the first year of life was 29%. In this analysis, factors such as parental attitudes to the morbidity associated with circumcision e.g., pain, inflammation, hemorrhage, etc. were taken into account. In the calculation of decision analysis, it was considered that there would be a 7.5% probability of renal sears resulting after an episode of UTI. In conclusion, the decision of whether or not to amputate the newborn foreskin will continue to be based on the personal preferences of the parents, but we believe the issues raised in this article should help them make an informed choice. Within the bounds of hospitalized care, the cost-benefit ratio of circumcision does not justify its routine and large scale application for doubtful benefits. It is essential that a multicenter controlled clinical trial be conducted to scientifically clarify the controversy whether circumcision can reduce the incidence of UTI.
Return to CIRP library