Objective: To assess the prevalence of circumcision across various social groups and examine the health and sexual outcomes of circumcision.
Design: An analysis of data from the National Health and Social Life Survey.
Participants: A national probability sample of 1410 American men aged 18 to 59 years at the time of the survey. In addition, an oversample of black and Hispanic minority groups is included in comparative analyses.
Main Outcome Measures: The contraction of sexually transmitted diseases, the experience of sexual dysfunction, and experience with a series of sexual practices.
Results: We find no significant differences between circumcised and uncircumcised men in their likelihood of contracting sexually transmitted diseases. However uncircumcised men appear slightly more likely to to experience sexual dysfunctions, especially later in life. Finally, we find that circumcised men engage in a more elaborated set of sexual practices. This pattern differs across ethnic groups, suggesting the influence of social factors.
Conclusions: The National Health and Social Life Survey evidence indicates a slight benefit of circumcision but a negligible association with most outcomes. These findings inform existing debates on the utility of circumcision. The considerable impact of circumcision status on sexual practice represents a new finding that should further enrich such discussion. Our results support the view that physicians and parents be informed of the potential benefits and risks before circumcising newborns.
Numerous recent studies have attempted to assess the value of neonatal circumcision. Several have determined that the procedure has positive effects. For example an association has been found between circumcision and lower rates of urinary tract infections in infancy[1,2], as well as lower rates of certain sexually transmitted diseases (STD)[3,4], As a result of these and other findings, the 1989 American Academy of Pediatrics (AAP) Task Force on Circumcision shifted its previous position, acknowledging that circumcision has certain potential medical benefits that must be weighed against its risks.
Male satisfaction has also been debated. Some believe that circumcision reduces male sensitivity and coital enjoyment while others argue that circumcision may afford greater ejaculatory control[7,8]. Masters and Johnson reported no clinically significant difference in the tactile sensitivity of the glans, More recent reports suggest the sensitivity of the circumcised glans may in fact be reduced[10,11]. Such claims of reduced sexual satisfaction for circumcised men have spurred a significant movement against the circumcision of infants and the reversing of circumcision in adult men. A technique of uncircumcising has even been introduced. Nevertheless, little consensus exists regarding the role of the foreskin in sexual performance and satisfaction.
The present study attempts to shed light on the circumcision debate by exploiting the National Health and Social Life Survey (NHSLS), a unique data source on the sexual, attitudinal, and health-related experiences of circumcised and uncircumcised Americans. Our analysis of these data proceeds as follows. First, we describe the prevalence of circumcision in the NHSLS sample. Beyond limited records on historical circumcision rates, very little is known regarding how the practice is distributed across various groups and strata of Americans. Second, we illustrate how the NHSLS data speak to current debates regarding the effect of neonatal circumcision on outcomes such as STDs and sexual dysfunctions. Finally, we present results indicating significant differences between circumcised and uncircumcised men in terms of their sexual practices.
The NHSLS conducted in 1992, is a nationally representative probability sample of 1511 men and 1981 women between the ages of 18 and 59 years living in households throughout the United States. It covers about 97% percent of the population in this age group - roughly 150 million Americans. It excludes people living in group quarters such as barracks, college dormitories, and prisons as well as those who do not know English well enough to be interviewed. There is an oversample of African Americans (n=458) and Hispanics (n=267). The sample completion rate was greater than 79%. Checks with other high quality samples (eg, Census Bureau's Current Population Survey) suggest that the NHSLS succeeded in getting a truly representative sample of the population. Each person was surveyed in person by experienced interviewers, who watched respondents on various social attributes, for an interview averaging 90 minutes in duration. Extensive discussion of the sampling design and evaluation of sample and data quality can be found in Laumann et al.
A respondent's circumcision status was ascertained by asking him whether he was circumcised. He was not asked, however, if the procedure was performed as a newborn or later in life. Experience with STDs was measured by asking respondents if they had ever been told by a doctor that they had a specified list of such diseases, which were identified by medical names as well as vernacular names (eg, gonorrhea, clap, or drip). Experience of sexual dysfunction was ascertained with the following question:
Sometimes people go through periods in which they are not interesting in sex or are having trouble achieving sexual gratification. I have just a few questions about whether you have experienced this in the past twelve months. During the past twelve months, has there been a period of several months or more when you ...
This question was asked regarding a series of sexual dysfunctions ranging from the inability to climax to lacking interest in sex. Respondents were also asked regarding their engagement in various sexual practices. Lifetime experience of a series of partnered behavior was ascertained by asking whether they had engaged in various sexual acts during their lifetimes. Respondents were also asked to describe the frequency with which they masturbated on a 10-point scale ranging from "never" to "every day."
We used a series of univariate and multivariate tests to chart the distribution of circumcision and examine its effects. In assessing the prevalence of circumcision for various social groups, we preformed multiple logistic regression of group membership on circumcision status and calculated adjusted odds rations (ORs) that reflected the odds of being circumcised for group members relative to all others in the sample. In addition, we examined how such differences have changed over time by repeating the analyses within 3 broad age cohorts.
In assessing the impact of circumcision on the contraction of STDs and susceptibility to sexual dysfunction, we performed a series of 2-tailed t tests to uncover significant differences between circumcised and uncircumcised men across a wide array of related outcomes. Next, we estimated logistic equations for each of the STDs and dysfunctions of circumcision status and a series of control variables. These factors included the number of lifetime sexual partners; education; race/ethnicity; religion, nativity; residence in urban; suburban, or rural areas; a 7-point scale indicating how liberal or conservative were the respondents sexual attitudes; and the respondent's age. Net of these controls, we calculated adjusted ORs for the odds of having the various STDs and dysfunctions for circumcised relative to uncircumcised respondents. Finally, we repeated these comparisons within categories of a critical third variable - for STDs, the number of lifetime sexual partners reported by the respondents, and for sexual dysfunction, the respondent's age.
We conducted similar analyses to assess the association between circumcision status and various sexual practices. We performed logistic regression to assess this association with the respondent's masturbation experience. In addition, we repeated this analysis as a logistic regression with a critical cutoff point, masturbating at least once a month, as the dependent variable. As results from this analysis matched that from the ordered logit, we present the logistic regression results to afford comparability with the other analyses conducted. In all of the models of sexual practices, we controlled for the same factors listed above. However, we excluded the number of lifetime sexual partners, a variable that is confounded with the tendency to engage in various practices. When this variable is included as a regressor, results for masturbation remain significant and findings regarding heterosexual oral sex weaken somewhat. In addition, we repeated the analyses for each of the 3 major ethnic groups featured in the survey, whites, blacks, and Hispanics. All analyses were performed using Stata version 4.0.14
Prevalence of Circumcision
Since the age range of males participating in the NHSLS was 18 to 59 years, estimates of the prevalence of neonatal circumcision can be calculated for the years 1933 to 1974. As shown in the Figure, the steady increase in circumcision rates among respondents during much of this period reflects the increase identified by other investigators. The proportion of newborns that were circumcised reached 80% in the years after World War II and climaxed in the mid 1960s. This rise mirrors the increasing prevalence of hospital births. In addition, a slight decline in the proportion of newborns circumcised occurred in the last years covered by the survey, about the time when the medical establishment began to question the value of routine neonatal circumcision. Overall 77% of the 1284 US-born men surveyed in the main NHSLS sample were circumcised, compared with 42% of the non-US born men.
Table 1 analyzes circumcision rates for religious, racial, and socioeconomic groupings of US-born respondents. We see that differences in circumcision rates across groups are not great. As may be expected, Jewish men report the largest proportion circumcised.
|All Men (N=1369)||Born 1933-1947 (N=358)||Born 1947-1962 (N=653)||Born 1963-1973 (N=488)|
|Predictor Variable||Circumcised, No. (%)||Adjusted OR (95% CI)+||Circumcised, No. (%)||Adjusted OR (95% CI)+||Circumcised, No. (%)||Adjusted OR (95% CI)+||Circumcised, No. (%)||Adjusted OR (95% CI)+|
|No religion||178 (80)||Referent||28 (71)||Referent||77 (82)||Referent||73 (82)||Referent|
|Mainline Protestant||301 (82)||1.24 (0.78-1.28)||85 (64)||0.88 (0.37-2.09)||143 (87)||1.83 (0.88-3.79)||73 (93)||1.73 (0.69-4.34)|
|Catholic||324 (79)||1.18 (0.91-1.47)||80 (68)||1.19 (0.50-2.83)||150 (84)||1.53 (0.74-3.17)||94 (80)||1.03 (0.48-2.21)|
|Jewish||25 (98)||... (...)||... (...)||... (...)||... (...)||... (...)||... (...)||... (...)|
|White||1067 (81)||Referent||251 (65)||Referent||478 (86)||Referent||332 (86)||Referent|
|Black||189 (65)||0.58 (0.40-0.65)||66 (49)||0.63 (0.31-1.29)||89 (64)||0.37 (0.21-0.65)||58 (76)||0.62 (0.30-1.27)|
|Hispanic||90 (54)||0.34(0.26-0.44)||22 (35)||0.38 (0.17-0.86)||31 (61)||0.15 (0.10-0.22)||36 (58)||0.21 (0.10-0.46)|
|Less than high school||372 (62)||Referent||142 (52)||Referent||153 (68)||Referent||78 (68)||Referent|
|High school graduate||538 (84)||2.56 (1.90-3.46)||107 (69)||2.17 (1.30-3.62)||264 (89)||2.78 (1.66-4.84)||168 (84)||1.48 (0.68-3.21)|
|Some college||167 (84)||2.72 (2.15-3.47)||21 (57)||1.12 (0.48-2.80)||72 (93)||4.82 (1.90-11.71)||74 (84)||1.86 (0.87-3.96)|
|College graduate||168 (87)||2.83 (2.25-3.81)||22 (82)||3.06 (1.13-8.32)||58 (81)||1.74 (1.14-2.39)||88 (94)||4.07 (1.54-10.72)|
* Source: National Health and Social Life Survey
+ Estimated ratio of odds of being circumcised for members of the specified group to odds for reference group. Derived from logistic regression model performed on all US-born respondents. This model includes all predictor variables as well as controls for residences in rural, suburban, or urban areas. OR indicates odds ratio; and CI, confidence interval. Ellipses indicate 20 cases or fewer. # Asians and American Indians were also included in the full model but are excluded here because they represent too few cases for reliable results.
However, given the small number of Jews surveyed in the NHSLS, this does not result in a significantly adjusted OR. Mainline Protestants (including, eg, Methodists, Presbyterians, Lutherans, and Episcopalians) are somewhat more likely than conservative Protestants to be circumcised (see Laumann et al (pp 146-147) for rationale and details on coding of religious groups). However, as indicated by the insignificant adjusted OR, this difference disappears when race is controlled. Indeed, the predominant pattern is one of high circumcision rates for all major American religious groups.
Differences in circumcision levels across racial and ethnic groups are more revealing. In particular, whites are considerably more likely to be circumcised than are blacks or Hispanics (81% vs 65% or 54%). These differences remain significant when other variables are controlled. Net of these factors, the odds of a black being circumcised are roughly half (95% confidence interval[CI], 0.40-0.85) that of whites; the odds for Hispanics are about one third (95% CI, 0.26-0.44) that of whites. There has been some convergence in circumcision rates for the 3 groups across cohorts, though differences among groups have persisted (Table 1).
Circumcision rates vary significantly by the level of education obtained by a respondent's mother. The critical break occurred between respondents whose mothers did and did not earn a high school diploma. While 62% of respondents whose mothers did not finish high school were circumcised, the rate for all other respondents varied from 84% to 87%. These differences remained significant when other factors were controlled. Net of such factors, the odds of being circumcised for respondents whose mothers earned at least a high school diploma was about 2.5 times (95% CI, 1.9-3.8) that of those whose mothers were less educated. This discrepancy appeared to be narrowing in more recent cohorts. Among the youngest group of respondents, only those whose mothers had finished college displayed a significantly higher circumcision rate than those whose mothers had not earned a high school diploma.
Sexually Transmitted Diseases
Table 2 compares the incidence of various STDs reported by circumcised and uncircumcised respondents. Note that in this and subsequent tables, circumcision status serves as an independent variable rather than a dependent variable. Dependent variables are listed in the first column of these tables.
|All men||1-4 Lifetime Partners||5-20 Lifetime Partners||21+ Lifetime Partners|
|Rate per 1000||Rate per 1000||Rate per 1000||Rate per 1000|
|Criterion Variable||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+|
|Gonorrhea||94.7||104.8||1.42 (0.79-2.56)||22.1||20.4||2.22 (0.25-18.77)||108.4||143.9||0.79 (0.38-1.82)||237.4||196.4||3.26 (0.96-11.11)|
|Syphilis||9.3||10.8||2.14 (0.41-11.30)||2.5||0.0||... (...)||9.8||7.3||0.77 (0.07-8.87)||28.2||0.0||... (...)|
|Chlamydia||25.1#||0.0||... (...)||7.4||0.0||... (...)||32.2||0.0||... (...)||58.5||0.0||... (...)|
|Nongonococcal urethritis||24.4||19.1||1.39 (0.43-4.09)||2.5||6.8||... (...)||27.2||28.1||1.17 (0.20-6.93)||74.3||56.8||1.53 (0.27-8.78)|
|All bacteria||123.9||112.9||1.61 (0.94-2.78)||29.5||20.4||2.48 (0.35-17.13)||157.8||151.5||0.99 (0.50-1.94)||316.4||232.1||2.88 (1.03-8.03)|
|Herpes||14.9||8.1||1.18 (0.23-8.94)||2.5||0.0||... (...)||9.9||15.5||... (...)||62.1||17.9||1.27 (0.12-12.95)|
|Hepatitis||24.2||24.4||1.22 (0.43-3.46)||14.6||13.7||1.53 (0.19-12.38)||32.1||16.3||1.50 (0.30-7.42)||28.2||72.7||0.04 (0.00-2.59)|
|Human immunodeficiency virus||1.9||5.4||... (...)||0.0||6.8||... (...)||2.5||7.8||... (...)||5.7||0.0||... (...)|
|All viral||35.3||32.9||1.30 (0.52-3.22)||17.2||13.7||1.35 (0.19-9.67)||32.0||30.3||1.84 (0.36-9.51)||90.4||89.3||0.70 (0.15-3.24)|
|No. of respondents||1449||1118||554||449||598||444||233||190|
* Source: National Health and Social Life Survey. C indicates circumcised; and UC, uncircumcised.
+ Estimated ratio of odds of circumcised men having experienced the specified sexually transmitted disease relative to odds for uncircumcised men. Derived from logistic regression model performed in which experience of the sexually transmitted disease is the dependent variable and independent variables, in addition to circumcision status, include ethnicity, religion, residence in rural, suburban, or urban areas, education and, in models for "all men," the number of sexual partners. OR indicates odds ration; and CI, confidence interval. Ellipses indicate unstable estimates.
# Indicates Ps.05 t test for difference between circumcised and uncircumcised men.
Several instructive features of the data presented in Table 2 deserve attention. First, circumcision status does not appear to lower the likelihood of contracting an STD. Rather, the opposite pattern holds. Circumcised men were slightly more likely to have had both a bacterial and a viral STD in their lifetime. While these differences are not statistically significant, they do not lend support to the thesis that circumcision helps prevent the contraction of STDs. Indeed, for chlamydia, the difference between circumcised men and uncircumcised men is quite large. While 26 of 1033 circumcised men had contracted chlamydia in their lifetime, non of the 353 uncircumcised men reported having had it.
Table 2 shows a marked decrease in the experience of STDs as the number of partners increases. The small, nonsignificant tendency for circumcised men to contract STDs at greater rates appeared for each category of sexual experience. In addition, contraction of bacterial STDs occurred at a significantly greater rate for men who have had more than 20 sex partners in their lifetimes. Among circumcised men with such a sexual background, the odds of contracting a bacterial STD are estimated at 2.88 that for uncircumcised men. While significant, the exact size of the difference is difficult to establish as indicated by the wide confidence interval (95% CI, 1.03-9.03). Note as well that this difference is driven largely by differential contraction of gonorrhea.
While some difficulties, such as experiencing pain during sex, were rare, Table 3 demonstrates that sexual dysfunction is a relatively common event for American men. Approximately 45% of both circumcised and uncircumcised men experience at least one of these dysfunctions in the year prior to the survey. In addition, the data suggest a slight tendency for such dysfunctions to plague uncircumcised men. When all age groups are considered, almost every dysfunction is slightly more common among men who have not been circumcised. In particular, the likelihood of having difficulty in maintaining an erection is significantly lower for circumcised men When other factors are considered, the difference in the odds of experiencing this dysfunction for circumcised men is significantly less than for circumcised men (OR, 0.66), but only at the 0.07 level (95% CI, 0.42-1.03).
|All Men||Age, 18-29 y||Age, 30-44 y||Age, 45-59 y|
|Comparison, %||Comparison, %||Comparison, %||Comparison, %|
|Criterion Variable||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+|
|Lacked interest in sex||15||17||1.04 (0.71-1.53)||14||15||1.39 (0.62-3.13)||16||12||1.59 (0.79-3.20)||14||23||0.52 (0.27-1.02)|
|Unable to ejaculate||8||10||1.05 (0.64-1.72)||8||9||0.58 (0.25-1.82)||9||5||1.86 (0.78-4.54)||8||16||0.70 (0.30-1.22)|
|Ejaculated prematurely||28||91||0.87 (0.64-1.18)||29||27||1.00 (0.54-1.84)||29||28||1.17 (0.88-2.00)||25##||37||0.59 (0.34-1.03)|
|Experienced pain during sex||3||3||1.14 (0.49-2.68)||4||4||0.93 (0.22-3.94)||3||1||... (...)||2||6||1.48 (0.29-9.04)|
|Was anxious about performance||17||16||0.87 (0.60-1.27)||23||15||1.35 (0.62-2.94)||16||17||1.05 (0.55-2.00)||13##||22||0.50 (0.22-0.95)|
|Had trouble achieving/maintaining erection||10##||17||0.66 (0.42-1.03)||7||9||0.85 (0.30-2.44)||10||7||1.28 (0.53-3.11)||13++||29||0.39 (0.16-0.77)|
|Had any dysfunction||43||48||0.85 (0.64-1.13)||47||46||0.88 (0.57-1.72)||44||39||1.33 (0.81-2.18)||40++||58||0.47 (0.28-0.79)|
|No. of respondents||1221||1203||378||399||686||519||340||285|
* Source: National Health and Social Life Survey, C indicates circumcised; and UC, uncircumcised.
+ Estimated ratio of odds of circumcised men having experienced the specified dysfunction relative to odds for uncircumcised men. Derived from logistic regression model performed in which experience of the dysfunction is the dependent variable and independent variables, in addition to circumcision status, include number of sexual partners, religion, residence in rural, suburban, or urban areas, education and, in models for "all men," ethnicity. OR indicates odds ratio; and CI, confidence interval. Ellipses indicate unstable estimates.
++ Indicates Ps.05 t test for difference between circumcised and uncircumcised men.
## Indicates Ps.10 t test for difference between circumcised and uncircumcised men.
Significant differences are more prominent when we examine the association between sexual dysfunction and circumcision status within age cohorts. While there appears to be little difference between circumcision status and sexual dysfunction for the two younger cohorts, the association is quite strong for the oldest group of respondents. (Table 3). Of the 7 sexual dysfunctions considered, uncircumcised older men were more likely to experience everyone of these difficulties than are their uncircumcised peers. Three of these differences are statistically significant even in the presence of controls. The odds of a circumcised man experiencing anxiety about his performance were approximately half that for his uncircumcised peers. (95% CI, 0.22-0.95); the odds of a member of the former group having difficulty achieving or maintaining an erection are about 0.40 that for the latter group (95% CI, 0.16-0.77). Overall, the odds that a circumcised man of the oldest cohort experienced sexual dysfunction was 0.48 that for uncircumcised men of the same age group (95% CI, 0.28-0.79). Thus, while, circumcised men did not generally appear to experience greater rates of sexual dysfunction, older circumcised men did display lower levels of dysfunction.
As shown in Table 4, NHSLS data indicate that circumcised men engage in a somewhat more elaborated set of sexual practices than do men who are not circumcised. For each of the practices examined, lifetime experience of various forms of oral and anal sex and masturbation frequency in the past year, circumcised men engaged in these behaviors at greater rates. The difference between circumcised men was greatest for masturbation - ironically, a practice that circumcision was once thought to limit. A total of 47% of circumcised men reported masturbating at least once a month vs 34% for their uncircumcised peers. This difference remains significant even when various demographic factors are controlled. The odds of a circumcised man masturbating at least once a month are estimated at 1.4 (95% CI, 1.04-1.89) that for uncircumcised men. In terms of lifetime sexual experience, the greatest differences occurred for heterosexual oral sex. In models with controls, circumcision status was associated with active heterosexual oral sex with a probability of insignificance of 0.07 (OR, 1.37; 95%CI, 0.97-1.92) and with passive heterosexual oral sex with a probability of insignificance of 0.08 (OR, 1.36; 95%CI, 0.96-1.93).
|All Men||Age, 18-29 y||Age, 30-44 y||Age, 45-59 y|
|Comparison, %||Comparison, %||Comparison, %||Comparison, %|
|Criterion Variable||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+||C||UC||Adjusted OR (95% CI)+|
|Active heterosexual oral sex||80##||65||1.37 (0.97-1.92)||84##||73||1.57 (1.04-2.37)||52||47||0.96 (0.40-2.28)||75||67||0.96 (0.70-1.32)|
|Passive heterosexual oral sex||81##||61||1.36 (0.95-1.93)||84##||73||1.57 (1.04-2.38)||71##||57||1.38 (0.56-3.37)||71||75||0.26 (0.05-1.25)|
|Active homosexual oral sex||5||2||2.15 (0.85-5.42)||6||3||2.22 (0.73-6.74)||3||3||... (...)||10##||2||... (...)|
|Passive homosexual oral sex||8||6||1.58 (0.85-2.96)||9||6||1.88 (0.73-6.74)||6||8||... (...)||10||5||... (...)|
|Homosexual anal intercourse||27||23||0.80 (0.56-1.14)||27||22||0.69 (0.45-1.06)||24||22||1.16 (0.40-3.38)||35||36||1.10 (0.35-3.48)|
|Masturbation >1/mo in past year||47##||34||1.40 (1.04-2.69)||50##||34||1.76 (1.24-2.51)||27||28||0.92 (0.35-2.39)||44||38||0.55 (0.22-1.95)|
|No. of respondents||1404||1143||1048||903||190||125||119||74|
The association between circumcision status and the various sexual practices exhibited differences across ethnic groups (Table 4). While circumcised men of all 3 ethnic groups tended to engage in more elaborated set of sexual practices, this was less true of blacks and Hispanics. For white men, the difference between being circumcised and being uncircumcised was quite stark. Circumcised men exhibited a greater likelihood of experiencing each of the various practices. In particular, the estimated ratio of the odds of masturbating at least once a month for circumcised men was 1.76 that for uncircumcised men (95% CI, 1.24-2.51). The adjusted OR for both forms of heterosexual oral sex were significant as well. These associations were less consistent for blacks and Hispanics. While circumcised Hispanic men did masturbate at a slightly higher rate than did their uncircumcised peers, no such difference appeared for blacks. Similarly, while circumcised black men exhibited a greater tendency to engage in passive oral sex than did uncircumcised black men (this difference was insignificant when other factors are controlled), the reverse was true among Hispanic men. Thus, while circumcision status appeared to be significantly linked with a higher likelihood to engage in various sexual practices, this applied mostly to whites and considerably less for blacks and Hispanics.
The United States stands apart from the rest of the world for its high rates of neonatal circumcision. Nevertheless, medical research on the topic has generated an ambiguous set of results regarding the impact of circumcision on the lives of men. As a result, rhetoric has reached a fever pitch as each side of the debate appeals to divergent criteria to make its case. Recognizing the merit of each position, the AAP has counseled that parents be fully informed of the risks and benefits of the procedure before deciding to have their son circumcised.
Our analysis of the NHSLS furnishes information that should be useful in such decision-making processes. We ascertained the prevalence of circumcision among American men as well as its impact on sexually transmitted diseases, sexual functioning, and sexual practice and preferences. Each of these areas generated noteworthy findings.
With respect to prevalence, we demonstrated that circumcision rates are greatest among whites and better educated respondents and that Americans of various religions do not display significantly different rates. The latter fact illustrates the unique cultural status maintained by circumcision in the United States. While circumcision has been employed as a religious marker in other Western societies, it has clearly lost such an association in America. American religious organizations have never strongly objected to the circumcision of its members, as they have instead deferred to medical authority on this subject.
Several factors may account for differential rates across ethnic groups. First, as race is associated with socioeconomic differences between Americans, these differences reflect the greater tendency for middle-class parents to desire circumcision for their sons. Similarly, blacks and Hispanics are concentrated is such regions as the South or Southwest where circumcision is less prevalent. However, the fact that differences in circumcision rates remain significant when region and class are controlled suggests that the various racial groups may have different preferences for circumcision. Members of groups for whom circumcision is less common may avoid circumcision for their sons so a sense of shared physical appearance is retained. Indeed, a recent study revealed that such social considerations typically outweigh various medical issues in determining parents' circumcision decisions.
As with race, differences among socioeconomic groupings may result from a differential likelihood of being born in a hospital. However, this difference remained salient even in later periods where hospital births were virtually universal. Thus, it again appears likely that significant social variation exists in the preference for and acceptance for circumcision. Better-educated parents, who are more likely to be exposed to the prevailing scientific wisdom favoring circumcision and to be exposed to significant social pressure to conform to this wisdom, circumcised their sons at greater levels than less educated parents.
In sum, we see that just as significant social institutions have played critical roles in the propagation of neonatal circumcision in the United States, the practice spread and persisted differentially across social groups. In contrast to societies where circumcision has indicated religious or cultic difference, the popularity of circumcision in 20th century United States clearly reflects social distinctions that are salient in contemporary society. Those social groups that had the education and cultural affinity for following the recommendations of medical and state authorities adopted circumcision while those groups who maintained a more distant relationship to such institutions did not embrace the practice to a comparable degree.
Note that these results should be treated with some caution when applied to rates of neonatal circumcision. The NHSLS respondents were asked whether they were circumcised, but not if the procedure was performed as a newborn or later in life. Given the rarity of circumcision on nonnewborns and the match of our data with existing records on neonatal circumcision, we feel that it is reasonable to assume that different rates across social groups reflect differential experience at time of birth. Note that this assumption is not necessary for examining the association between circumcision status and various outcomes. In such cases, the procedure must only have been performed before the outcome in question. A second source of ambiguity concerns the accuracy of respondents' reports of their circumcision status. Research from the 1950s and on adolescents in the 1980s suggests that up to a third of men may not know their circumcision status. While we have no means of independently verifying respondents' reports, we control for education in all of our models, a variable that should significantly impact knowledge of circumcision status.
Sexually transmitted diseases
With respect to STDs, we found no evidence of a prophylactic role for circumcision and a slight tendency in the opposite direction. Indeed, the absence of a foreskin was significantly associated with contraction of bacterial STDs among men who have had many partners in their lifetimes. These results suggest a reexamination of the prevailing wisdom regarding the prophylactic effect of circumcision. While circumcision may have an impact that was not picked up by the NHSLS data, it seems unlikely to justify the claims made by those who base their support for widespread circumcision on it.
Several cautions apply to these findings, however. The NHSLS self-reports of STD contraction were almost certainly underreports, because respondents may have lacked knowledge that they had the STD if it was asymptomatic or never diagnosed, or if they did not understand the diagnosis. Moreover, given the social stigma associated with having STDs, some respondents may have been reluctant to report that they had such a condition. Considerable effort was expended in minimizing underreporting by devising an interview protocol that gave a maximum sense of privacy and confidentiality, that persuaded the respondent of full disclosure for public health reasons, and provided memory aids to facilitate respondent recall. Even if these procedures were especially effective, and we think they were, we must still acknowledge that the self-reports understate the incidence of STDs to a substantial but still unknown extent. In addition, there may have been systematic biases in underreporting related to particular attributes of respondents. In particular, we might expect better educated people to be more knowledgeable about disease labels than less well educated people. By contrast, however, better educated persons may be more sensitive to the social stigma associated with having an STD. Controlling for education in the various models helped but probably did not fullly redress these issues.
Results may also have been affected by the possibility that some respondents did not know whether they had been circumcised. If one assumes that respondents were more likely to say that they were circumcised (ie, the default answer is to give an affirmative answer to any question) and that the same people report higher levels of STDs, this may account for any association found. While we feel that such distortions are small, it is worth reiterating that the principal result of the presented analysis is that no discernible differences in STD experience can be found to distinguish circumcised and uncircumcised men.
The NHSLS data suggest a benefit of circumcision with respect to sexual dysfunction. Circumcised men were slightly less likely than those who had not been circumcised to experience various sexual difficulties. This difference was significant among the oldest age group. Interestingly, this age group presents the best test for establishing differences between circumcised and uncircumcised men. as it contains comparable numbers of both. It would seem then that NHSLS data supply some support for those who see circumcision as performing a positive role in promoting healthy sexual behavior. This appears most true for older men. While there may be many sources for this relationship, it is possible that the association between masturbation frequency and circumcision status (discussed below) provides a clue. If older men require more direct stimulation to function sexually, men for whom masturbation is part of their sexual script may be better able to adapt sexually as they age. Clearly such reasoning must remain speculative until further research is performed.
Our findings regarding sexual practice pose the greatest challenge for future research. NHSLS results reveal a clear pattern in that circumcised men report a more highly elaborated set of sexual practices. In particular, the association between circumcision status and masturbation frequency was quite strong. Similar results, at a somewhat weaker level, occurred for heterosexual oral sex. These results escape easy interpretation. Certainly, they cast doubt on the Victorian-era notion that circumcision reduces the urge to masturbate.
While we do not wish to push speculation too far, differences in the association between circumcision status and sexual practice across ethnic groups suggest that cultural, rather than physiological forces may be responsible. In particular, the presented results may reflect attitudes regarding the cultural acceptability of the uncircumcised penis. Note that the association of circumcision with experience of sexual practices is weakest among ethnic minorities for whom circumcision is less prevalent. Among whites by contrast, uncircumcised men are relatively uncommon. A consequence of this may be that a certain stigma is attached to the uncircumcised penis by the white population. If the uncircumcised penis assumes a somewhat negative cultural association among whites, this may lead uncircumcised men to engage in a somewhat less elaborated set of sexual practices.
While NHSLS results do not lead clear support to either side of the circumcision debate, they make a significant contribution to our knowledge regarding the potential risks and benefits of circumcision. In addition to documenting the prevalence of circumcision across various social groups, we have discovered that circumcision provides no discernible prophylactic benefit and may in fact increase the likelihood of STD contraction; that circumcised men have a slightly lessened risk of experiencing sexual dysfunction, especially among older men; and that circumcised men displayed a greater rates of experience of various sexual practices. While evidence regarding STD experience contributes to ongoing debates, our results concerning sexual dysfunction suggest the need for continued research that should further aid parents in weighing the benefits and risks of circumcising their sons.
- Wiswell T, Smith FR, Bass JW. Decreased incidence of urinary tract infections in circumcised male infants. Pediatrics. 1985;75:901-903.
- Herzog L. Urinary tract infection and circumcision: a case control study. Am J Dis Child 1989;143:343-350.
- Parker S, Stewart AJ. Wren MN, et al. Circumcision and sexually transmissible disease. Med J Aust. 1968;2:288-290.
- Task Force on Circumcision. Report of the task force on circumcision. Pediatrics 1989; 84;388-391.
- Committee on Fetus and Newborn. Report of the ad hoc task force on circumcision. Pediatrics 1975; 56:610-611.
- Masi CM. Circumcision: The Medical Issues. Chicago, Ill. University of Chicago School of Social Services Administration 1995. Working paper.
- Morgan W. Penile Plunder. Med J Aust 1967;1:1102-1103.
- Burger R. Guthrie T. Why circumcision? Pediatrics. 1974;54:362.
- Masters W, Johnson V. Human Sexual Response. Boston Mass: Little Brown & Co. 1966.
- Preston EN. Whither the foreskin? A consideration of routine circumcision. JAMA. 1970; 213:1853-1858.
- Cleary T. Kohl S. Overwhelming infection with group B beta-hemolytic streptococcus associated with circumcision. Pediatrics 1979; 64:301-303.
- Goodwin W. Uncircumcision: a technique for plastic reconstruction of the prepuce after circumcision. J Urol. 1990;144:1203-1205.
- Laumann EO, Michael RT, Gagnon JH, et al. The Social Organization of Sexuality: Sexual Practices in the United States. Chicago, Ill: University of Chicago Press; 1994 chapt 2, appendixes A, B.
- Stata Corporation. STATA Version 4.0. College Station, Tex: Stata Corp; 1994.
- Brown MS, Brown CA. Circumcision decision: prominence of social concerns. Pediatrics. 1987; 80:215-219.
- Graves, EJ. Detailed diagnoses and procedures: National Hospital Discharge Survey, 1993, from the National Center for Health Statistics. Vital Health Stat 13. 1995;122:128.
- Lilienfeld AM, Graham S. Validity of determining circumcision status by questionaire as related to epidemiological studies of cancer of the cervix. J Natl Cancer Inst. 1958;21:713-720.
- Schlossberger NM, Turner RA, Irwin CE Jr. Early adolescent knowledge and attitudes about circumcision: methods and implications for research. J Adolesc Health. 1991; 12:293-297.