OBJECTIVES. Previous research suggests that male circumcision may be a protective factor against the acquisition of sexually transmitted infections; however, studies examining this question have produced mixed results. The aim of this study was to examine the association between circumcision status and sexually transmitted infection risk using a longitudinal birth cohort study.
METHODS. Data were gathered as part of the Christchurch Health and Development Study, a 25-year longitudinal study of a birth cohort of New Zealand children. Information was obtained on: (1) the circumcision status of males in the cohort before 15 years old, (2) measures of self-reported sexually transmitted infection from ages 18 to 25 years, and (3) childhood, family, and related covariate factors.
RESULTS. Being uncircumcised had a statistically significant bivariate association with self-reported sexually transmitted infection. Adjustment for potentially confounding factors, including number of sexual partners and unprotected sex, as well as background and family factors related to circumcision, did not reduce the association between circumcision status and reports of sexually transmitted infection. Estimates of the population-attributable risk suggested that universal neonatal circumcision would have reduced rates of sexually transmitted infection in this cohort by 48.2%.
CONCLUSIONS. These findings suggest that uncircumcised males are at greater risk of acquiring sexually transmitted infection than circumcised males. Male circumcision may reduce the risk of sexually transmitted infection acquisition and transmission by up to one half, suggesting substantial benefits accruing from routine neonatal circumcision.
I'm Dr Sambola a Traditional Herbal and a Spell Caster.
I Cure Different Types Of Human Infections, Diseases Using Natural Roots And Herbs. My major areas of concentration are chronic infection itching, STDs, STIs Gonorrhea, body obor/mouth odor, Gonorrhea, Staphylococcus and Toilet infection. my whatsapp +2348085930654 or email: sambolaherbalfoundation11@gmail.com
https://sambolaherbalfoundation.wordpress.com
Dear Editors:
The study by Fergusson et al. in the November issue of Pediatrics was interesting.[1] While it is unclear why a study of the risk of sexually transmitted infections (STIs) in men aged 18 to 25 years would be of interest to pediatricians, the final article lacked the polish that one would expect of an article published in Pediatrics. It is readily apparent that the article did not receive the scrutiny it should have from those who reviewed it. Consequently, a reader may be more likely to dismiss the findings of the study because of the poor quality of the manuscript. A list of deficiencies, most of which could be easily addressed by the authors, which should have been identified by the peer-reviewers, follows.
1. The results in the Abstract should have included the absolute risk reduction (ARR), the odds ratios with confidence intervals, and the number needed to treat (NNT).
2. The Conclusion section of the Abstract states: “Male circumcision may reduce the risk of sexually transmitted infection acquisition and transmission by up to one half … .” Since the study only measured the acquisition of STIs and not the transmission of STIs, inclusion of a reference of the transmission of STIs is inappropriate.
3. Likewise, the phrase “substantial benefits” is used several times in the article. This rings of hyperbole, especially since the authors failed to include a calculation of the NNT, which should temper the authors’ rhetoric.
4. Of the ten references citing the debate and controversy surround the issue of infant male circumcision, seven espouse a strongly pro-circumcision view, two are neutral, and only one presents a clear message that infant circumcision is wrong. This sampling of citations suggests there is little controversy for the authors.
5. The authors mention a paucity of prospective studies, yet they fail to mention a prospective cohort study from their own country, New Zealand.[2] This study had markedly different findings than those published in this article. The authors need to contrast their findings with the other New Zealand cohort study.
6. Circumcision status was based on parent report at 4 months of age and hospital reports. Parental report is a potential source of bias that the authors fail to acknowledge. The authors fail to report in how many subjects they relied on parental report to assign the circumcision status or the degree of concurrence between parental report and hospital records.
7. It is not clear whether subjects were asked the number of STIs since the previous survey. If so, this number is not reported.
8. There is no mention of whether anyone in the cohort reported STIs prior to 18 years of age.
9. It is not clear how many of the cohort who completed the survey at 21 years of age were lost to follow-up by 25 years of age. It is extremely unlikely that this number is zero. If subjects lost to follow-up at the 25-year-old evaluation were not included in the analysis, many would consider this unethical.
10. The authors used the wrong statistical model. The study measured events in a population over a time-span. Data collected in this manner should be assessed using Poisson regression, which is specifically designed for information collected in this fashion. A repeated measures (mixed marginal) model using dichotomous outcomes used by the authors is inappropriate, and in this situation had inherent instabilities.[3] In the 18- to 20-year -old time span, only two circumcised men reported having had an STI. In the 21- to 25-year-old span, five circumcised men reported having had an STI. Logistic regression models are unstable and unreliable with cell sizes smaller than ten. There is further instability when multivariate analysis is performed.
There are two options: using exact logistic regression (for which the software for a repeated measures with random effects model is not available) or Poisson regression. If exact logistic regression is used the results are ExOR=2.65, 95%CI=0.58-24.64 for the first time span, ExOR=2.64, 95% CI=0.98-8.90 for the second time span, and ExOR=2.43, 95%CI=1.03-6.62 for the total duration of the study. The last result is much lower from what the authors reported, thus confirming the unreliability of using a parametric model with small cell sizes.
Poisson regression, using scant information provided in the article, gives the following results: RR=2.40, 95%CI=1.07-5.39, which is also much lower than incorrect method employed by the authors. One of the advantages of Poisson regression is that information provided by subjects who supplied information during only one of the two time spans can be included in the analysis.
11. The software used to calculate the statistics is not mentioned.
12. Reporting only the PAR can be misleading. With this type of study it is expected that the ARR and NNT be reported. The authors failed to do so. For the first time span the ARR is 2.06%, which translates into a NNT of 48.7. For the second time span the ARR is 4.86%, which translates into a NNT of 20.6. If both time spans are combined the ARR is 5.80%, which translates into a NNT of 17.2.
With the NNT, a rough estimate of the cost-effectiveness of infant male circumcision can be performed. With a NNT of 17.2 and the cost of infant circumcision at roughly $162, it would cost $2793 to prevent one STI. But the $162 spent to circumcision the infant is not available for the 20 some years as the male grows up, so the dollar amount needs to be discounted. If one assumes a 3% discount, the cost per STI is $5513. If a 5% discount is used, the cost per STI is $8579. The cost of treating an STI is usually less than $200.
13. The authors allude to an association between circumcision status and sexual practices, yet they do not include their findings. This needs to be disclosed since this finding is both interesting and important.
14. In the multivariate analysis, both the number of sexual partners and circumcision status are included in the model. Since the authors allude to an association between number of sexual partners and not having a foreskin, an assessment of the colinearity of the two factors is needed to determine whether both factors can be safely included in the same model. Colinearity destabilizes regression models that can yield unreliable results. The degree of colinearity between these factors can be measured using linear regression and should have been reported.
15. The authors are missing 20% of their subjects. Consequently, their standard errors need to be adjusted accordingly. The Yates method indicates that the standard errors need to be increased by a factor of 1.12. This would change the first time span 95% CI to 0.50-14.4, the second time span 95% CI to 0.88-7.71, and the pooled CI to 1.06-6.73. For the Poisson regression, adjusting for missing data changes the 95% CI to 0.98-5.92. For estimate using exact logistic regression, the 95% CI is 0.93-7.44.
16. Table 2 should use 95% CI instead of or in addition to p-values.
17. In the Discussion section, the authors state, “These estimates are consistent with those presented in other research that has suggested that circumcision may lead to substantial reductions in rates of STI infection.” The medical literature entirely contradicts this statement. The study question for Fergusson and associates was whether circumcision status is associated to overall risk of acquiring any STI versus no STI. The authors failed to report on the studies that ask the same question raised by their study. To date there have been eleven studies that are able to answer the study question raised by Fergusson and associates.[2, 4-13] In none of these studies nor in any of the subgroups of these studies has an association between circumcision and a decreased the risk of acquiring an STI been found. To the contrary, several of the studies have found circumcised men to be a statistically significant increased risk for acquiring an STI.[11-13] It is unclear why the references made in Introduction and the Discussion sections provided by the authors fail to compare their study to studies with a similar research question.
This lapse in basic scholarship is most apparent in the failure to compare the data from this cohort study to the only other cohort study that has reached publication. The Dunedin Multidisciplinary Health and Development Study found serologic evidence of herpes simplex virus type 2 in 7.4% of 26-year - old men, with no difference by circumcision status (OR=0.98, 95% CI=0.47-2.04). This compares to 0.8% of men reporting a genital herpes in infection in the Christchurch study. Likewise, 18.2% of circumcised men and 17.5% of noncircumcised men in the Dunedin study reported a STI in the past 5 years (OR=0.95, 95%CI=0.67-1.58).[2] In Christchurch, only 6.6% of men reported a STI.
18. Mention of the studies linking HIV infection to circumcision status is inappropriate in the discussion of this article since this study did not evaluate HIV infection.
19. In the Discussion section, the authors state, “A study of circumcision status and STI in a US sample (with a circumcision rate of ~75%) estimated that circumcision could reduce rates of STI in males by about a third.36” This misrepresents the findings of the study cited.[6] When the number of subjects with any form of STI are compared to the number of subjects without an STI, there was a trend toward noncircumcised men being at lower risk (OR=0.82, 95%CI=0.64-1.04). This is the opposite of what the authors state.
20. The authors cite a reduction of penile inflammation in middle childhood, but fail to mention that the same study[14] and subsequent study[15,16] have found circumcised boys three-years-old and younger to be at greater risk of penile inflammation. Why would the lead author misrepresent the results of his own study?
21. In 1997, Edgar Schoen criticized Laumann et al. by stating: “Relying on the memories of men ranging in age from 18 to 59 years on this sensitive topic is not justified.”[17] This study also relies on the memories of men, but the authors correctly point out that unless memory is affected by circumcision status, then this should not be a source of bias.
The publication of this study in its present form represents a failure of the peer-review system. Even if those called upon to review this article share the bias of the authors, they still have a duty to the authors, editors, and readers to help the authors provide the best article possible. The resultant sloppiness and clear bias of the final manuscript reflects poorly on everyone involved and may lead to reader to place value on the findings than the finding deserve.
References:
1. Fergusson DM, Boden JM, Horwood LJ. Circumcision status and risk of sexually transmitted infection in young adult males: an analysis of a longitudinal birth cohort. Pediatrics 2006; 118: 1971-7.
2. Dickson N, van Roode T, Paul C. Herpes simplex virus type 2 status at age 26 is not related to early circumcision in a birth cohort. Sex Transm Dis 2005; 32: 517-9.
3. Selvin S. Practical biostatical methods. Belmont, CA: Duxbury Press; 1995.
4. Aynaud O, Piron D, Bijaoui G, Casanova JM. Developmental factors of urethral human papillomavirus lesions: correlation with circumcision. BJU Int 1999; 84: 57-60.
5. Dave SS, Fenton KA, Mercer CH, Erens B, Wellings K, Johnson AM. Male circumcision in Britain: findings from a national probability sample survey. Sex Transm Infect 2003; 79: 499-500.
6. Diseker RA 3rd, Peterman TA, Kamb ML, et al. Circumcision and STD in the United States: cross sectional and cohort analyses. Sex Transm Infect 2000; 76: 474-9.
7. Laumann EO, Masi CM, Zuckerman EW. Circumcision in the United States: prevalence, prophylactic effects, and sexual practice. JAMA 1997; 277: 1052-7.
8. Parker SW, Stewart AJ, Wren MN, Gollow MM, Straton JA. Circumcision and sexually transmissible disease. Med J Aust 1983; 2: 288-90.
9. Schrek R, Lenowitz H. Etiologic factors in carcinoma of penis. Cancer Research 1947; 7: 180-7.
10. Richters J, Smith AMA, de Visser RO, Grulich AE, Rissel CE. Circumcision in Australia: prevalence and effects on sexual health. Int J STD AIDS 2006; 17: 547-54.
11. Seed J, Allen S, Mertens T, et al. Male circumcision, sexually transmitted disease, and risk of HIV. J Acquir Immune Defic Syndr Hum Retrovirol 1995; 8: 83-90.
12. Taylor PK, Rodin P. Herpes genitalis and circumcision. Br J Vener Dis 1975; 51: 274-7.
13. Urassa M, Todd J, Boerma JT, Hayes R, Isingo R. Male circumcision and susceptibility to HIV infection among men in Tanzania. AIDS 1997; 11: 73- 80.
14. Fergusson DM, Lawton JM, Shannon FT. Neonatal circumcision and penile problems: an 8-year longitudinal study. Pediatrics 1988; 81: 537-41.
15. Van Howe RS. Variability in penile appearance and penile findings: a prospective study. Br J Urol 1997; 80: 776-82.
16. Van Howe RS. Neonatal circumcision and penile inflammation in young boys. Clin Pediatr (Phila) In press.
17. Schoen EJ. Advantages and disadvantages of neonatal circumcision. JAMA 1997; 278: 201.
Conflict of Interest:
None declared
In our recent article (1) we found that males who were not circumcised as children in the CHDS had an increased risk of contracting sexually transmitted infection (STI) as adults (OR = 3.12). We noted that this association was similar to those found in studies of the association between circumcision status and more severe forms of STI (including HIV, syphilis, and gonorrhea). Recent correspondence to the journal has highlighted the fact that our findings are not consistent with cross- sectional studies of the linkages between circumcision and the more common forms of STI (including Chlamydia, genital warts, genital herpes, gonorrhea, and non-specific urethral infections). Of particular relevance is a recent Australian survey of 10,000 male respondents (2), and the preliminary findings from the Dunedin study (3). These discrepancies with our findings are too large to be disregarded, and we are of the view that it would be premature to use our findings to promote the view that circumcision reduces risks of less severe forms of STI, until further research clarifying this issue is conducted.
We have checked and rechecked our findings to ensure that no error has occurred in data analysis or data preparation. We have found no such errors, and our findings suggest a relatively strong association between circumcision status and STI. It is notable that this association was replicated using data gathered at two ages (ages 21 and 25), and we are unaware of any previous study that has reported repeated measures longitudinal data. Below we examine some of the possible explanations the discrepancy between our findings and previous research.
1) Sample Size: Two correspondents (4, 5) suggest that our findings may be explained away as an artefact of (relatively) small sample size. This view reflects the common opinion that small sample size may produce misleading results. That view is incorrect in general, since it can be readily shown that the effects of small sample size are to bias results towards the null hypothesis, thus increasing risks of type II errors (6). Sample size does not influence risk of type I statistical errors. Put another way, the effects of small sample size are to reduce study precision, but they does not impugn study validity. We would note that throughout our report we report confidence intervals on all statistics to describe the precision of our findings.
2) Measurement Error : Young (4) suggests our conclusions may be due to measurement errors. As we note in our report, there is no reason to assume that these errors were systematic, such that accuracy of report varied with circumcision status. Under conditions of non-differential error in reporting, it is well known that the effects of measurement error are to bias results towards the null hypothesis, leading to an under- estimation of effect size (7, 8). To explore this we have re-estimated odds ratios for our study assuming a non-differential 50% under-reporting rate. This shows that under this scenario the true odds ratio would be 3.50, compared to our observed estimate of 3.19. Young also argues that measurement error could be differential since “many circumcised men report avoidance of doctors, for example”. This is no evidence of this in our sample; a comparison of the rates of medical attendance amongst circumcised and non circumcised males over the periods 17-18 years, 20-21 years, and 24-25 years shows almost identical rates of attendance (799 visits per 100 circumcised vs. 797 visits per 100 uncircumcised; p >.90). As would be expected from this, our results do not change if we include medical attendance as a covariate in our analysis.
3) Perinatal confounding: Glass (9) proposes the novel hypothesis that our results may be confounded by perinatal status, since those males who were circumcised may be a more robust and viable group. While this argument draws a long bow, it is nonetheless worth testing. To address this we have extended our analysis to include a range of perinatal measures collected during the course of our study including: fetal distress; whether the infant was resuscitated; admission to NICU; diagnoses of cardiac, genitourinary, or orthopedic anomalies; being injured during birth; diagnosis of respiratory distress syndrome or other neonatal illness; five minute apgar score; birth weight, length, and head circumference; and time to onset of respiration. With the exception of a small association with birth weight, these factors were not correlated with circumcision status, and their inclusion as covariates in the analysis did not change our findings.
The above considerations clearly suggest that it would be premature to dismiss our findings on the grounds of sample size, measurement error, or perinatal confounding. In turn, this conclusion points to the need for further research to clarify the issues raised by our study. Two lines of research are needed. First, in 2007 we intend to revisit the CHDS cohort at age 30. This will give us an opportunity to gather further data on STI, and will make it possible to determine whether our findings are internally consistent and replicable. However, there is also a need for further external validation of our results. It is our view that well designed case-control research comparing a case series of those with STI with a random control series offers the best prospect for testing the external validity of our results.
Finally, we would note that Blair (10) claims that we propose routine circumcision to prevent adult disease. Nowhere in the paper did we make such a proposal. What we did say was that on the basis of all of the available evidence (not just our study), the case for routine circumcision was more finely balanced than admitted by critics of circumcision. This statement is a long way from proposing routine circumcision. To make our position on this matter clear, our views coincide closely with those of the American Academy of Pediatrics (11), that the evidence as it stands currently is not sufficiently compelling to advocate routine circumcision, and that this should be a parental decision.
1. Fergusson DM, Boden JM, Horwood LJ. Circumcision status and risk of sexually transmitted infection: An analysis of a longitudinal birth cohort Pediatrics 2006;118(5):1971-1977.
2. Richters J, Smith AMA, de Visser RO, Grulich AE, Rissel CE. Circumcision in Australia: prevalence and effects on sexual health. International Journal of STD and AIDS 2006;17:547-554.
3. Dickson N, van Roode T, Paul C. Herpes simplex virus type 2 status at age 26 is not related to early circumcision in a birth cohort. Sexually Transmitted Diseases 2005;32(8):517-519.
4. Young HP. Christchurch result is anomalous [Electronic Version]. Pediatrics Post-Publication Peer Review; 2006. Retrieved 20 November 2006 from http://pediatrics.aappublications.org/cgi/eletters/118/5/1971.
5. Sykes J. Christchurch study is discordant with other studies [Electronic Version]. Pediatrics: Post-Publication Peer Review; 2006. Retrieved 20 November 2006 from http://pediatrics.aappublications.org/cgi/eletters/118/5/1971.
6. Hays WL. Statistics. 4th ed. Sydney: Holt, Rinehart, & Winston; 1988.
7. Goldberg J. The effects of misclassification on the bias in the difference between two proportions and the relative odds in the fourfold table. Journal of the American Statistical Association 1975;70(561-567).
8. Kuha J, Skinner C, Palmgren J. Misclassification error. In: Armitage P, Colton T, editors. Encyclopedia of Biostatistics. Chichester: Wiley; 1998. p. 2615-2621.
9. Glass M. A confounding factor in the Christchurch circumcision studies [Electronic Version]. Pediatrics Post-Publication Peer Review; 2006. Retrieved 20 November 2006 from http://pediatrics.aappublications.org/cgi/eletters/118/5/1971.
10. Blair R. Surrogate consent is inadequate for non-therapeutic excision of tissue [Electronic Version]. Pediatrics Post-Publication Peer Review; 2006. Retrieved 20 November 2006 from http://pediatrics.aappublications.org/cgi/eletters/118/5/1971.
11. Circumcision policy statement. American Academy of Pediatrics. Task Force on Circumcision. Pediatrics 1999;103(3):686-93.
Conflict of Interest:
None declared
Dear Editor:
Fergusson and associates propose neonatal circumcision as a routine procedure to prevent adult disease.[1] This approach raises an ethical issue. Some believe that non-essential childhood surgery should be postponed until the patient can decide for himself about prevention of adult disease.
Winberg and Bollgren, in a comment about infant circumcision, suggest:
The Committee on Bioethics of the American Academy of Pediatrics takes a somewhat similar stance:
Deferral of the circumcision decision now has judicial approval. A court, which ruled in a circumcision case, recently determined that a child should be protected from circumcision until he becomes 18 and can decide for himself.[4]
The ethical principle at work is patient autonomy. Surrogate consent ethically is insufficient to protect patient autonomy in cases of irreversible non-therapeutic excision of healthy tissue. Patient autonomy can be protected only by postponing elective non-therapeutic circumcision until the child can decide for himself.
Respectfully submitted
Robert M. Blair
References
[1] Fergusson DM, Boden JM, Horwood LJ. Circumcision status and risk of sexually transmitted infection in young adult males: an analysis of a longitudinal birth cohort Pediatrics 2006;118(5):1971-1917.
[2] Bollgren I, Winberg J. Letter. Acta Paediatr Scand 1991;80:575-577.
[3] Committee on Bioethics, American Academy of Pediatrics. Informed consent, parental permission, and assent in pediatric practice. Pediatrics 1995;95(2):314-317.
[4] Johnson CK. Judge takes father's side in circumcision feud. Chicago Sun-Times, Tuesday, October 24, 2006.
Conflict of Interest:
None declared
To the Editor—
Fergusson et al. report that no HIV infection was found in either circumcised or uncircumcised males.1 From this, one might argue that both circumcision and uncircumcision are 100 percent protective against HIV infection! In reality, this illustrates the bizarre results that small numbers can produce in scientific studies.
Several large cross-sectional studies have been published with quite different results. A summary of their findings along with the findings of Fergusson et al. is presented in table one.
Social Life Survey
Sexual Attitudes and Lifestyles
and Relationships
and Development Study
The findings of the retrospective Fergusson et al. longitudinal study are sharply discordant with several published cross-sectional studies with much larger numbers of participants. The small numbers in the longitudinal study may have produced false results. This discordance suggests that careful study of these data is needed before they can be accepted.
References
Conflict of Interest:
None declared
To the editors, Pediatrics:
Dave et al.1 "used data from the 2000 British National Survey of Sexual Attitudes and Lifestyles (Natsal 2000) - a large-scale, stratified, probability sample survey" with, apparently, 913 circumcised and 4833 intact subjects, and found "no significant associations between circumcision and being diagnosed with any one of the seven specific STIs."
Richters et al.2 used "The results from a telephone survey in 2001-02 of a probability sample of Australian households including 10,173 men aged 16-59 (response rate 69.4%) ... to assess the prevalence of circumcision across social groups in Australia and examine lifetime history of sexually transmissible infection (STI)...." They found that "After correction for age, circumcision was unrelated to reporting STI, but appeared to protect against penile candidiasis."
Thomas et al.3 made "a case-control study of male HIV infected U.S. military personnel (n= 232) recruited from 7 military medical centers and male U.S. Navy controls (n=516) from a general aircraft carrier population." They found that "After adjustment for demographic and behavioral risk factors lack of circumcision was not found to be a risk factor for HIV (OR = 0.9; 95% CI: 0.51, 1.7) or STI (OR = 1.08; 95% CI 0.52, 2.26)."
Laumann et al.4 used the National Health and Social Life Survey conducted in 1992, "a nationally representative probability sample of 1511 men and 1981 women between the ages of 18 and 59 years." They found "no significant differences between circumcised and uncircumcised men in their likelihood of contracting sexually transmitted diseases."
Now Fergusson et al.5 claim to find there is a correlation between circumcision and a reduced rate of STI.
How to explain this anomaly? One significant difference is the small number of infected men in the Christchurch study - 44 in total - and the small proportion of circumcised men - 30% - compared to the others, resulting in only seven circumcised men being infected altogether. Small numbers are more susceptible to random variations than larger - a single coin toss will not deliver 50:50 heads and tails, even with an unloaded penny. If only two more of the men who contracted STIs were circumcised, the association would vanish in the younger age group, four to five more when they were older. Only sexually active men are at any risk of STI, so the effective sample size is smaller than it appears. Sophisticated analyses of significance are inappropriate for these small numbers.
Contrary to Fergusson et al.'s assertions, the correlations in the two age-ranges did not reach statistical significance: in both cases, the 95% CI included 1.0 - no association between the variables. When the data were pooled, the lower bound was still only 1.17 - a very small departure from no significance.
Fergusson et al. say they found that the circumcised men were more sexually active and less likely to use condoms than the intact men. They argue that this increases the significance of their findings, but this is not at all foregone, it is not clear what it does signify, and they give no figures. They dismiss without evidence any possible effect of being circumcised on the accuracy of reporting of STIs (many circumcised men report avoidance of doctors, for example), and make no estimate of the accuracy of reporting of circumcision status. Their medical records do not count circumcisions performed outside hospitals.
They say, "The unadjusted associations between circumcision status and reports of Chlamydia were similar to those for overall STI but were not statistically significant because of the small number of cases of Chlamydia (n = 22)" - implying that they would have been statistically significant if there had been more cases. Laumann et al. 4 had more cases. They found that "While 26 of 1033 circumcised men had contracted chlamydia in their lifetime, none of the 353 uncircumcised men reported having had it." (With zero cases on one side, the odds ratio is indeterminate, but if circumcision and Chlamydia were unrelated, one would expect 19 cases among the circumcised and 7 among the intact - the chance of 26 and none is less than one in a thousand.) Fergusson et al. refer to this as "the extent to which circumcision may play a protective role in reducing risks of Chlamydia." and as one of "[a] number of previous studies in this area [that] have failed to find evidence of a protective effect of circumcision in reducing risks of Chlamydia." In fact, Laumann et al. seemed to find that if circumcision has any effect, it is the reverse of protective.
Much has been made of the relative reduction (48%) Fergusson et al. claimed circumcision to cause in the (small) incidence of STI, but one obvious figure lacking from their paper is the Number Needed to Treat (NNT). It turns out that nearly 49 babies would have to have their foreskins cut off to protect one promiscuous 18-21 year old from an STI (which can be readily treated at much less cost than 49 circumcisions), and nearly 21 babies to protect one 21-25 year old (assuming the study will map accurately on to the wider world). The other 20 men might well feel aggrieved.
That the 48% figure has been quoted around the world in support of routine infant circumcision6,7 (and not the 21, to dismiss it) is to say the least, unfortunate.
Hugh Young
REFERENCES
1. Dave SS, Fenton KA, Mercer CH, Erens B, Wellings K, Johnson AM. Male circumcision in Britain: findings from a national probability sample survey Sex Transm Infect 2003;79:499-500
2. Richters J, Smith AM, de Visser RO, Grulich AE, Rissel CE Circumcision in Australia: prevalence and effects on sexual health. Int J STD AIDS, August 1, 2006; 17(8): 547-54.
3. Thomas AG, Bakhireva LN, Brodine SK, Shaffer RA Prevalence of male circumcision and its association with HIV and sexually transmitted infections in a U.S. navy population Abstract submitted for The XV International AIDS Conference, 2004 Abstract no. TuPeC4861
4. Laumann EO, Masi CM, Zuckerman EW. Circumcision in the United States. Prevalence, prophylactic effects, and sexual practice. JAMA. 1997;277:1052–1057
5. Fergusson DM, Boden JM, Horwood JL Circumcision Status and Risk of Sexually Transmitted Infection in Young Adult Males: An Analysis of a Longitudinal Birth Cohort Pediatrics 2006;118;1971-1977 DOI: 10.1542/peds.2006-1175
6. "Circumcision cuts STI risk" Health24 http://www.health24.com/news/Man/1-926,38249.asp
7. "Circumcision appears to reduce STD risk" Medscape/Reuters http://www.medscape.com/viewarticle/547242
I declare that I have no financial or other competing interests.
Conflict of Interest:
None declared
Dear Editor,
First, I wish to thank Professor Fergusson for sending me a pre- publication copy of his paper on circumcision and breastfeeding. I believe that when data from that paper is combined with “Circumcision Status and Risk of Sexually Transmitted Infection in Young Adult Males: An Analysis of a Longitudinal Birth Cohort (Pediatrics 2006; 118: 1971 –1977), it reveals a confounding factor in the studies. Professor Fergusson’s results were so striking at one point that he rejected them. There is a way to account for his findings, including the result that he rejected, but it does mean confronting a confounding factor that would be inherent in any study of this kind.
The studies did not find any difference between the circumcised and uncircumcised men in such measures as birth weight and social class. However, there is evidence in the data that the circumcised group were different. In Professor Fergusson’s paper on breastfeeding the circumcised group had significantly fewer consultations for lower respiratory tract infections in the first two years of life. Also, while other differences were not statistically significant, they were mostly in favour of the circumcised group and one, the difference in gastro-intestinal troubles, was somewhere between P= 0.1 and P= 0.2.
His paper on circumcision and sexually transmitted infections found that the circumcised group had more sex, more unprotected sex and a greater number of sexual partners but despite this, they had fewer sexually transmitted infections.
Professor Fergusson rejected the correlation of circumcision with a lower rate of lower respiratory tract infections as a statistical anomaly. Why? Could any factor account for this and other findings?
I believe that one factor has not been addressed. The Australasian policy on circumcision states that an “absolute contraindication” for infant circumcision is “sick and unstable infants” <http:// www.racp.edu.au/hpu/paed/circumcision/complications.htm>. Similarly, the American policy states: “Newborn circumcision is an elective procedure to be performed at the request of the parents, on newborn boys who are physiologically and clinically stable.” <http://www.cirp.org/library/statements/aap-acog2002/>
These policies help ensure that infants who are circumcised are strong enough to withstand the procedure. In this context, circumcision is evidence that the child was healthy and vigorous at birth. It is plausible that a group of newborns chosen for health and vigour may have significantly different health outcomes from a group containing both healthy and not so healthy neonates! The differences that persist into childhood and adulthood may be a reflection of this selection factor. This does not, of course, preclude the possibility that circumcision also had an effect.
Why consider this factor? Professor Fergusson’s hypothesis doesn’t account for all the differences he found in the data, so other explanations should be considered. Even if it would seem unlikely that this selection factor would have an effect, its impact should be checked.
One way of doing this might be to crosscheck the results with the APGAR scores for the men in the study with circumcision and with their later health status. The APGAR test results for these men, if available, could give valuable clues on the correlation of health in infancy with later measures of health and fitness.
Yours sincerely,
Michael Glass
Conflict of Interest:
None declared