Volume 17, Number 9: Pages 739-740,
September 1962.

Cancer of the Cervix in Reference to
Circumcision and Marital History

Elizabeth Stern, M.D., and Peter M. Neely, Ph.D.

THE STUDY OF THE distribution of cancer in populations has led to useful information regarding the epidemiology of cancer and to the formulation of hypotheses on the etiology of cancer. For example, the incidence of cancer of the cervix amongst Jewish women is very low compared to non-Jewish women. Because of the universal custom of circumcision of Jewish males, the low incidence was attributed to this practice. An etiological agent for cancer of the cervix was thought to be associated with a factor related to the presence of the foreskin in males. This hypothesis received support in that Moslems are reported to have a lower incidence of cancer of the cervix than do women of other religious affiliations in India where circumcision is routine for Moslem males.1 There has also been experimental induction of cancer of the cervix in mice by direct application to the cervix of smegma collected from male volunteers.2 Indirect evidence for this hypothesis is offered by the report of Gagnon3 that cancer of the cervix does not occur in celibate women. However, this finding has been refuted to some extent by the report of Towne4 who found a lowered but definite incidence of cases of cancer of the cervix in women living in religious seclusion. Discussion of low incidence in celibate women brings up the question of nulliparity and other factors related to marriage and family life, which will be discussed later.

Since the recommendation had been made that circumcision should be used as a preventive measure against cancer of the cervix.1 we sought further confirmation of this hypothesis. An almost ideal population was that of the well women attending a cancer detection facility, where the population was split almost equally between women whose husbands were circumcised and those whose husbands were not. The discovery rate for cancer of the cervix among non-Jewish women whose marital partners were circumcised was no different from the rate among non-Jewish women with non-circumcised husbands.5 Further, the use of a sheath contraceptive by the marital partner, which has an effect equivalent to circumcision in that the cervix is protected from contact with smegma, was found not to be associated with rate differences for cancer of the cervix. In this population, among religious groups, there was no difference between Catholic and Protestant women; for Jewish women this study confirmed the finding of the low rate of cancer of the cervix in this group.

It was then decided to study a larger number of factors which seemed important in determining cervical cancer by a method of statistical analysis which examined these factors simultaneously, making allowance for the effect of one variable on another. For this reason we can expect greater validity for those variables found to be statistically significant. The variables used were: age; race; religious affiliation; use of a sheath contraceptive; age at marriage; number of marital events defined as the sum of marriages, divorces, and widowhoods; whether or not the individual was currently married; and the circumcision status of the husband. (If a woman had had more than one husband, the status was considered non-circumcision if any husband was not-circumcised.) These data for the well population and for those found to have cancer were analyzed using the statistical method of multiple regression analysis.5

The variable, “number of marital events,” was found to be most useful in selecting the women with cancer of the cervix; the variable “religion” (Jewish), was almost as important as marital events when Jewish women were included in the analysis; the variable, “circumcision” never appeared as a primary variable. (For details see Stern and Dixon.5) These findings were substantiated in a subsequent study using a more extended list of variables including data on parity, occupation, family history, ages at menarche and menopause, and “estrogen index” as estimated by squamous cell differentiation of vaginal smears.6

The Papanicolaou technique was used as screening method for the detection of cancer of the cervix in his presumably well population. This technique provides information essential for exploring the relationships among those found to be normal and those with abnormalities separated into the following groups: dysplasia of the cervix,* carcinoma in situ, and the invasive stages of cancer of the cervix.7

When statistical analysis was applied to the separate groups, the findings suggested a two-stage hypothesis for carcinogenesis in which different factors were associated with the initiation of cancer of the cervix than were associated with the promotion of one stage to another. For example, the variables, “currently married,” “age at marriage,” and “estrogen index,” were most useful as discriminators in separating cases with dysplasia from cases with in situ or invasive cancer. The variable, “age,” perhaps indicating the passage of time was most important in separating in situ from invasive cancer.


The finding of an important statistical association does not, however, constitute evidence for an etiological factor. Statistical associations do supply leads for further investigation, particularly if there is biological plausibility. In the case of cancer of the cervix, on one factor has been observed to confer immunity to this disease, and, although, marked differences in rates do exist in some specialized groups of women, some common denominator must be sought.

We propose the hypothesis, based on the observation that women with cancer of the cervix are more likely to have a history of multiple marital events, and taking all other findings into consideration, that women with cancer of the cervix have a personality makeup suggesting emotional instability (which expresses itself in some women by multiple divorces), and that this may be due to or responsible for an underlying hormonal inbalance.

We hope that other workers will critically review the present findings as well as carry out new studies. In any case, further studies must use analytical methods which help see the forest of basic factors rather than the trees of specific related findings.


1. Wynder, E.: Circumcision as a preventive factor against cancer of the cervix. Proceedings of the third national cancer conference. 603-607, 1956.

2. Pratt-Thomas, H. R.., and others: The carcinogenic effect of human smegma: an experimental study, preliminary report., Cancer 9:671-680, July-Aug., 1956.

3. Gagnon, F.: Contribution to the study of the etiology and prevention of cancer of the cervix of the uterus. Am J. Obst. & Gynec. 60:516-522, Sept, 1950.

4. Towne, J.: Carcinoma of the cervix in nulliparous and celibate women, Am J. Obst. & Gynec. 69:606-613, March 1955.

5. Stern, E., and Dixon, W.J.: Cancer of the cervix–a biometric approach to etiology. Cancer 14:153-160, Jan.-Feb., 1961.

6. Stern, E; Dixon, W. J.., and Neely, P.: Cancer of the cervix–further investigations using the biometric approach. Unpublished data.

7. Stern, J.: Rate, stage, and patient age in cervical cancer; analysis of age specific discover rates for atypical hyperplasia, in situ cancer, and invasive cancer in a well population. Cancer 12:933-937. Sept.-Oct., 1959

*Currently dysplasia of the cervix is not considered a stage of cancer. Recent studies show that dysplasias constitute a high risk group for cancer. This is not inconsistent with the view that dysplasia is a very early stage of cancer perhaps in part reversible.

Dr. Stern is Lecturer in Pathology and Chief of the Cytology Laboratory, Department of Pathology at the University of California Medical Center, Los Angeles.
       Dr. Neely is a Fellow in the School of Public Health at the University of California, Los Angeles.

(File revised 15 November 2006)