Meta-Analysis of Measures of Sexual Activity and Prostate Cancer : Epidemiology

Secondary Logo

Journal Logo


Meta-Analysis of Measures of Sexual Activity and Prostate Cancer

Dennis, Leslie K.1 and; Dawson, Deborah V.2

Author Information



Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.

The etiology of prostate cancer is unclear, despite its being the most common cancer among men in the United States. 1,2 Meta-analyses of potential risk factors may guide the ongoing process of understanding etiology through consideration of candidate risk factors and new etiologic theories. Such analyses can evaluate strength of evidence, identify important covariates, and examine subgroups that individual studies lack the power to assess. The consideration of sexual activity includes a constellation of risk factors that may be associated with prostate cancer. Specifics of sexual activity may correlate with an infectious agent or may be related to a hormonal effect. Although the precise mechanism through which sexual activity is related to prostate cancer may not be understood, determining the strength and consistency of the association may help to determine which men are at high risk.

The overall aim of this study was to examine both the strength and the consistency of the observed associations between aspects of sexual activity and prostate cancer. The review article by Key 3 provides some information on sexual behavior, although it did not include all previous studies. Another drawback of this earlier review is that for continuous variables Key 3 compared the lowest and the highest levels as dichotomous variables, rather than using a dose-response approach; as the author suggested, the relative risk (RR) estimates for continuous variables are not strictly comparable among studies because of the various cutpoints used. 3 We included early studies not included by Key, 3 as well as more recently published studies, including six studies not published in English. In addition, the analyses presented here are more sophisticated. We conducted a detailed meta-analysis examining RR estimates for prostate cancer associated with various aspects of sexual activity, including: multiple sexual partners, multiple marriages, age of initiation of sexual activity, age at first marriage, frequency of sexual activity, number of children, and history of sexually transmitted infections (STIs).

Materials and Methods

Literature Search

The studies considered in this meta-analysis were analytic studies that measured some aspect of sexual activity in relation to prostate cancer. We used the following search strategy to identify relevant studies. First, we searched the MEDLINE database for articles published from 1966 through the end of September 2000. MeSH headings, key words, and text words searched included: prostatic neoplasms and prostate cancer, sex behavior, sexual activity, multiple partners, multiple sexual partners, multiple marriages, marriage, first intercourse, first marriage, age at intercourse, age at marriage, sexually transmitted disease, venereal disease, gonorrhea, syphilis, number of children, fertility, coitus, first coitus, and prostitution. Of 205 articles found through MEDLINE, we reviewed 40 in detail; the remaining articles included 15 reviews, 83 articles related to treatment, and 67 containing other biological discussions. We identified no additional articles through CANCERLIT, Current Contents, or Biology Abstracts.

We gathered the 40 articles found through literature searches and checked their references for additional relevant studies. We also searched for recent publications by authors who presented preliminary reports. In an attempt to identify “unpublished” RR estimates for prostate cancer and sexual activity, we examined additional articles on other risk factors for prostate cancer that may have reported sexual activity as a possible confounder. An additional 13 articles were identified using these methods. Of the 53 articles reviewed in detail, several were determined to have no relevant data and others were found to have used the same study populations. Of the relevant 44 articles, eight publications appeared to be on populations already reported, leaving 36 independent studies.


For each study and level of a factor, the natural log of the relative risk estimate (lnRR) and its variance were needed. The variances were calculated on the basis of the reported confidence intervals, if available. 4 Otherwise, we calculated the variance of the lnRR from reported data. 4 For studies that stated that no association was found but did not report a RR, we assumed an estimated RR of 1.0. We estimated corresponding variances on the basis of the number of subjects, assuming an average exposure rate calculated from the other studies.

Two independent reviewers (L. K. D. and D. V. D.) abstracted the data. Inconsistencies were re-reviewed until agreement was achieved. We sought third-party resolution of disagreements when needed. Hospital-based studies included studies with hospital, clinic, or unidentified control groups. We considered studies with neighborhood controls, community controls, and general population-based controls to be population-based. For articles that reported both population-based and hospital-based controls, only the RRs based on the population-based controls were used in this meta-analysis. When available, covariate-adjusted RRs were used in the meta-analysis; most studies adjusted for age.

Statistical Analysis

For dichotomous factors, we report both fixed-effects and random-effects models for estimation of pooled relative risk estimates. 5 Fixed-effects models lead to inferences about the particular studies that have been assembled, whereas random-effects models assume that the study samples were drawn from a larger universe of possible studies, leading to inferences about all studies in the hypothetical population of studies. 5 It is unclear whether either method is appropriate when the studies are heterogeneous. Statistical tests of homogeneity 4,6 were performed to assess the consistency of associations. Studies with the greatest contribution to the heterogeneity were identified.

For factors with multiple ordinal categories, we used a fixed-effects dose-response method to evaluate possible linear relations. 6 This method combines different levels of exposure in a linear regression of the lnRR, adjusting for correlated measures within studies. We used the median of the category range for each reported RR. Rather than arbitrarily assigning the lowest or highest category in a study, we used data from other studies that reported more information to assign an exposure score. For “fewer than 6 partners,” we assigned the value of 2 partners; for a range of “0–10,” 3 partners; for “3 or more,” 10 partners; for “6 or more” 12.5 partners; and for “11 or more,” 17 partners. When nonlinear effects were suspected, we examined quadratic models. In the quadratic model, the statistical method could not be applied to factors with a nonzero or inconsistent reference category (eg, age at first intercourse). 6

We stratified data by study design and type of control subjects.


Table 1 describes the 36 studies from 44 published articles included in the meta-analysis. 7–49 When duplicate publications exist, Table 1 describes which publication for each study was used in the meta-analyses for each measurement of sexual activity. The 36 studies included three cohort studies, 13 population-based case-control studies, and 20 non-population-based studies. The six publications not in English were translated to obtain the relevant information. 11,14,23,29,36,37

Table 1:
Studies Included in Meta-Analysis of Aspects of Sexual Activity, by Study Design and Year of Publication

Eighteen studies reported information on the number of sexual partners; however, two reported only the number of premarital partners. 34,46 No cohort studies examined number of sexual partners. Data from the 16 case-control studies support a modest linear increase in the estimated RR (Table 2). The association is similar among population-based and hospital-based case-control studies with wider CI, among the hospital-based studies (Figure 1). A quadratic model did not explain the data better than the linear model did. Of the 16 studies reporting number of sexual partners, a small hospital-based study that reported the highest RR (by Steele et al48) was responsible for the heterogeneity. When this study was excluded, the overall association with prostate cancer for an increase of 20 sexual partners was reduced from RR = 1.17 to RR = 1.16 (95% CI = 1.04–1.29; N = 15; heterogeneity P = 0.36).

Table 2:
Pooled Relative Risk Estimates Based on a Linear Dose-Response Model for Prostate Cancer with Number of Sexual Partners, Number of Marriages, Age at First Intercourse, and Age at First Marriage, by Study Design
Estimated relative risk and 95% confidence interval for an increase of five sexual partners based on a fixed-effects dose-response model for each study by study design and year published, along with pooled relative risk estimates.

We examined multiple marriages after excluding never-married men. The ten studies suggest no linear association with multiple marriages relative to a single marriage (Table 2). For multiple marriages, the heterogeneity was due to a recent study by Hayes et al. 31 This study had the greatest decreased risk of prostate cancer among men who had been married more than once. When this study was removed, the association increased to RR = 1.11 for three marriages compared with one marriage (95% CI = 0.96–1.29; N = 9; heterogeneity P = 0.19).

The data suggest that ever-married men were at slightly higher risk than never-married men (RR = 1.17; 95% CI = 0.98–1.40; N = 20; heterogeneity P = 0.24) based on a fixed-effects model (not shown).

Table 2 displays the RR for age at first intercourse and for age at first marriage for a 5-year decrease in age based on a linear model. Both of these analyses excluded two studies that reported only mean or median ages for cases and controls. The lack of an association may imply no association with prostate cancer or reflect data without comparable categories. The heterogeneity seen in Table 2 for age at first intercourse and for age at first marriage is due, in each case, to one study that reported a significant increase with age rather than a decrease. When the study by Rotkin 43 was removed from analysis of age at first intercourse, the RR for a 5-year decrease in age increased to 1.09 (95% CI = 1.02–1.15; N = 11; heterogeneity P = 0.23). When the study by LaVecchia et al30 was removed from analysis of age at first marriage, the RR for a 5-year decrease in age increased to 1.09 (95% CI = 1.03–1.16; N = 11; heterogeneity P = 0.48).

Among the 18 case-control studies that examined frequency of sexual activity, six studies were not used because they reported data without comparable categories. 21,24,34,38,39,43 For the remaining 12 studies, the RR for prostate cancer appears to increase with increasing frequency of sexual activity (Table 3). The association is strongest among hospital-based studies (Figure 2). A quadratic model did not explain the data better than the linear model. The heterogeneity among studies reporting frequency of sexual activity was largely due to a study by Du et al, 11 which found a fivefold increase in men who had sex more than three times per week compared with those engaged in sexual activity less frequently. The heterogeneity was also affected by a Russian study 36 that had poorly described study methods. However, removing these studies had little effect on the pooled results, with an RR of 1.18 for three times per week (95% CI = 1.08–1.28; N = 10; heterogeneity P = 0.38).

Table 3:
Pooled Relative Risk Estimates Based on a Linear Dose-Response Model for Prostate Cancer with Frequency of Sexual Activity and Fertility (Number of Children), by Study Design
Estimated relative risk and 95% confidence interval for frequency of sexual activity for an increase of one time per week based on a fixed-effects dose-response model for each study by study design and year published, along with pooled relative risk estimates.

Five studies reported age-specific frequencies of sexual activity: two studies had data on activity for the third, fifth, and seventh decades of life 10,22; one study provided data for the third and fifth decades of life 35; the fourth study considered only the seventh decade 24; and the fifth study reported activity for ages greater than 37 years (approximately the fifth decade) and in the past 5–7 years (approximately the seventh decade). 36 Compared with the overall RR of 1.20 (N = 12) for an average lifetime frequency of sexual activity of three times per week, the RR for three times per week was 1.14 (95% CI = 0.98–1.31) during the third decade of life, 1.24 (95% CI = 1.05–1.46) during the fifth decade, and 0.68 (95% CI = 0.51–0.91) during the seventh decade. Little difference was seen when the fifth study 36 was excluded (fifth decade RR = 1.20 and seventh decade RR = 0.64). The average lifetime frequency of sexual activity among studies reporting age-specific data (RR = 1.20) was similar to that for all 12 studies, therefore not indicating an age-specific reporting bias by studies with positive findings.

Table 3 also details a lack of an association seen with increasing number of children, based on the 18 heterogenous studies. Among these studies, 14% of cases had no children compared with 19% of controls. Two studies affected the heterogeneity for this association. Du et al11 and Graham et al45 each reported a twofold increased risk of prostate cancer in men who had more than three children as compared with those who had fewer. Removing these studies from analyses reduced the association to RR = 1.02 for four children (95% CI = 0.92–1.12; N = 16; heterogeneity P = 0.09).

Twenty-three studies reported on sexually transmitted infections, venereal disease, gonorrhea, or syphilis, with a weighted average STI history of 19% in cases and 12% in controls. Among these studies, there was one cohort study of syphilis patients. An additional study reporting only on genital warts 50 and four studies reporting only on herpes titers 51–54 are not included here. There appears to be an association with the presence of any STI based on 17 studies (Figure 3). Table 4 describes the RRs by study design, with a slightly higher risk seen among the seven studies with population-based controls (RR = 1.51). Seven of the 23 studies reported heterogeneous RRs for syphilis. The heterogeneity (P = 0.003) was largely due to the New York syphilis cohort study, which appeared to underestimate cancer cases as a result of a lack of inclusion of cancers diagnosed outside New York. Excluding this study from analyses, the RR for prostate cancer and syphilis is 2.30 (Table 4). Thirteen of the 23 studies reported risk of prostate cancer for gonorrhea, 5 of which did not report other STIs. The risk for gonorrhea (RR = 1.36) is not as strong as that for the presence of any STI (Table 4). Studies of STIs appeared to be homogeneous, with little differences between fixed-effects and random-effects models. Slightly higher RRs were seen for early publications (1971–1988) compared with recent publications (1990–2000). Two studies 19,32 also reported an increased risk of prostate cancer among men whose partners reported STIs, for a pooled risk of 2.06 (95% CI = 1.02–4.19). Five studies reported an association for prostate cancer with men who had visited a prostitute, suggesting a slight increased risk (RR = 1.19; 95% CI = 1.01–1.41).

Estimated relative risk and 95% confidence interval for history of sexually transmitted infections based on a random-effects model for each study, by study design and year published, along with pooled relative risk estimates.
Table 4:
Pooled Relative Risk Estimates for Prostate Cancer and Sexually Transmitted Infections (STIs), by Study Design*

Publication bias is always a concern for meta-analyses. To estimate the number of unpublished studies needed to nullify significant findings to nonsignificance (P > 0.05), we assumed a null effect for unpublished studies and arbitrarily used a variance equivalent to that seen in the study by Honda et al. 17 For STIs, an additional 95 such studies would be needed. More than 200 studies would be needed to nullify the results seen for frequency of sexual activity and 29 to nullify results for number of sexual partners.


This meta-analysis found an association for prostate cancer with some measures of sexual activity, including sexually transmitted infections. The increased RR for prostate cancer seen with STIs suggests a possible infectious component in the development of prostate cancer. The effects of infections on prostate cancer are not clearly understood. Inconsistent findings have been seen for the risk of prostate cancer related to infections other than STIs, including epididymitis, hepatitis, cirrhosis, urethritis, trichomoniasis, cytomegalic virus, and Epstein-Barr virus. 13,20,22,26,33,49

These data are supportive of a role of STIs in the etiology of prostate cancer. Consistent associations that are not likely to be explained by publication bias were seen for history of any STI (RR = 1.4), gonorrhea (RR = 1.4), and syphilis (RR = 2.3). The association between prostate cancer and STIs is further supported by an increased RR seen for men whose partners reported STIs (RR = 2.1), men who visited prostitutes (RR = 1.2), men reporting more than 30 sexual partners (RR = 1.3), and men reporting extramarital affairs (RR = 2.2). The 12% of controls reporting STIs in these studies is equivalent to that reported among men 50–59 years of age for a nationwide survey of sexual activity conducted in 1992. 55,56 Based on these data, 30.6% of prostate cancers among men with STIs are attributable to the STI. Assuming causality, the corresponding population attributable risk percentage is 5.8% of all prostate cancers having STI-related etiology.

The higher RR for prostate cancer and STIs seen among population-based studies rather than hospital-based studies could reflect differential misclassification caused by reporting bias, where population-based controls are less likely to report socially undesirable factors such as STIs regardless of their history. Differences seen by study design may instead reflect a high percentage of hospital controls who actually have a history of STIs. Although case-control studies are prone to recall or reporting bias, which may affect reporting of STIs, the recent study by Hayes et al10 found a higher prevalence of antibodies to Treponema pallidum among prostate cancer cases than in controls, suggesting that recall or reporting bias does not entirely explain the relation. Population-based and hospital-based case-control studies tend to have different advantages and biases; however, both study designs showed an increased association between prostate cancer and STIs.

This meta-analysis also found an association for prostate cancer and frequency of sexual activity. The increased RRs seen with increased sexual frequency, particularly before 60 years of age, suggest a possible link with sexual hormone levels. The association between prostate cancer and frequency of sexual activity is not likely to be explained by reporting bias or publication bias. These data suggest that 16.7% of prostate cancer among men with frequency of sexual activity three times or more per week may be attributable to the sexual activity.

Data from a national survey 55,56 suggest that Americans are more likely to be sexually active or engage in sexual activity if they are married. This would not support a positive correlation between STIs or number of sexual partners and frequency of sexual activity, suggesting multiple mechanisms in the etiology of prostate cancer related to sexual activity.

A mechanism to explain the relation between sexual frequency and prostate cancer is less well understood than that for STIs and prostate cancer. Substantial evidence indicates that hormones play a major role in the etiology of several cancers. 57 Among some men, frequency of sexual activity may be related to hormone levels. It is thought that neoplasms may occur in response to excessive hormonal stimulation of an organ of which the normal function and growth are controlled by hormones. 57 Androgens are required for the growth, maintenance, and functional activities of the prostate gland. 58 Reports suggest that men whose testes have been removed or never developed are not at risk for prostate cancer, 24,59,60 supporting evidence of a hormonal effect on prostate cancer development. Although underlying hormonal factors may contribute to the relation of sexual activity and prostate cancer, human sexual practices are complex and may not directly correlate with hormone levels.


1. Greenlee RT, Hill-Harman MB, Murray T, Thun M. Cancer statistics, 2001. CA Cancer J Clin 2001; 51: 15–36.
2. American Cancer Society. Cancer Facts and Figures 1998. Atlanta: American Cancer Society, 1998.
3. Key T. Risk factors for prostate cancer. Cancer Surv 1995; 23: 63–77.
4. Woolf B. On estimating the relation between blood group and disease. Ann Hum Genet 1955; 19: 251–253.
5. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188.
6. Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. Am J Epidemiol 1992; 135: 1301–1309.
7. Michalek AM, Mahoney MC, McLaughlin CC, Murphy D, Metzger BB. Historical and contemporary correlates of syphilis and cancer. Int J Epidemiol 1994; 23: 381–385.
8. Mills PK, Beeson L, Phillips RL, Fraser GE. Cohort study of diet, lifestyle, and prostate cancer in Adventist men. Cancer 1989; 64: 598–604.
9. Severson RK, Nomura AMY, Grove JS, Stemmermann GN. A prospective study of demographics, diet, and prostate cancer among men of Japanese ancestry in Hawaii. Cancer Res 1989; 49: 1857–1860.
10. Hayes RB, Pottern LM, Strickler H, et al. Sexual behaviour, STDs and risks for prostate cancer. Br J Cancer 2000; 82: 718–725.
11. Du SF, Shi LY, He SP. A case-control study of prostate cancer (in Chinese). Zhonghua Liu Xing Bing Xue Za Zhi 1996; 17: 343–345.
12. Andersson S-O, Baron J, Bergstrom R, Lindgren C, Wolk A, Adami H-O. Lifestyle factors and prostate cancer risk: a case-control study in Sweden. Cancer Epidemiol 1996; 5: 509–513.
13. Hiatt RA, Armstrong MA, Klatsky AL, Sidney S. Alcohol consumption, smoking, and other risk factors and prostate cancer in a large health plan cohort in California (United States). Cancer Causes Control 1994; 5: 66–72.
14. Nakata S, Imai K, Yamanaka H. Study of risk factors for prostatic cancer (Japanese). Acta Urol Jpn 1993; 39: 1017–1024.
15. Fincham SM, Hill GB, Hanson J, Wijayasinghe C. Epidemiology of prostate cancer: a case-control study. Prostate 1990; 17: 189–206.
16. Hill GB, Fincham SM, Wijayasinghe C, Haronga CL, Hendin MM. Sex ratio of offspring of patients with prostatic cancer. Can Med Assoc J 1985; 133: 567–571.
17. Honda GD, Bernstein L, Ross RK, Greenland S, Gerkins V, Henderson BE. Vasectomy, cigarette smoking, and age at first sexual intercourse as risk for prostate cancer in middle-aged men. Br J Cancer 1988; 57: 326–331.
18. Kolonel LN, Yoshizawa CN, Hankin JH. Diet and prostatic cancer: a case-control study in Hawaii. Am J Epidemiol 1988; 127: 999–1012.
19. Mandel JS, Schuman LM. Sexual factors and prostatic cancer: results from a case control study. J Gerontol 1987; 42: 259–264.
20. Schuman LM, Mandel J, Blackard C, Bauer H, Scarlett J, McHugh R. Epidemiologic study of prostatic cancer: preliminary report. Cancer Treat Rep 1977; 61: 181–186.
21. Ross RK, Shimizu H, Paganini-Hill A, Honda G, Henderson BE. Case-control studies of prostate cancer in blacks and whites in Southern California. J Natl Cancer Inst 1987; 78: 869–874.
22. Mishina T, Watanabe H, Araki H, Nakao M. Epidemiological study of prostate cancer by matched-pair analysis. Prostate 1985; 6: 423–436.
23. Mishina T, Watanabe H, Araki H, et al. High-risk group for prostatic cancer by matched pair analysis (in Japanese). Nippon Hinyokika Gakkai Zasshi 1981; 72: 1256–1279.
24. Ross RK, Paganini-Hill A, Henderson BE. The etiology of prostate cancer: what does the epidemiology suggest? Prostate 1983; 4: 333–344.
25. Greenwald P, Damon A, Kirmse V, Polan AK. Physical and demographic features of men before developing cancer of the prostate. J Natl Cancer Inst 1974; 53: 341–346.
26. Hsieh C, Thanos A, Mitropoulos D, Deliveliotis C, Mantzoros CS, Trichopolous D. Risk factors for prostate cancer: a case-control study in Greece. Int J Cancer 1999; 80: 699–703.
27. Ewings P, Bowie C. A case-control study of cancer of the prostate in Somerset and east Devon. Br J Cancer 1996; 74: 661–666.
28. Ilic M, Vlajinac H, Marikovic J. Case-control study of risk factors for prostate cancer. Br J Cancer 1996; 74: 1682–1686.
29. Wei Q, Tang X, Yang Y, Zhan Y, Yin H. Risk factors of prostate cancer: a matched case-control study (in Chinese). Hua Xi Yi Ke Da Xue Xue Bao 1994; 25: 87–90.
30. LaVecchia C, Franceschi S, Talamini R, Negri E, Boyle P, D’Avanzo B. Marital status, indicators of sexual activity and prostate cancer. J Epidemiol Commun Health 1993; 47: 450–453.
31. Hayes RB, de Jong FH, Raatgever J. Physical characteristics and factors related to sexual development and behavior and the risk of prostate cancer. Eur J Cancer Prev 1992; 1: 239–245.
32. Oishi K, Okada K, Yoshida O, et al. A case control study of prostatic cancer in Kyoto, Japan: sexual risk factors. Prostate 1990; 17: 269–279.
33. Checkoway H, DiFerdinando G, Hulka BS, Mickey DD. Medical, life-style, and occupational risk factors for prostate cancer. Prostate 1987; 10: 79–88.
34. Banerjee AK. Carcinoma of prostate and sexual activity. Urology 1986; 28: 159.
35. Lees RM, Steele R, Wardle D. Arsenic, syphilis and cancer of the prostate. J Epidemiol Commun Health 1985; 39: 227–230.
36. Chaklin AV, Plotnikov SV. Importance of various factors in the occurrence of prostatic cancer (in Russian). Urol Nefrol (Mosk) 1984; 4: 46–51.
37. Malatinsky E, Zajac R, Jancar M. The sexual life of patients with adenoma and carcinoma of the prostate (in Slovak). Bratisl Lek Listy 1984; 82: 1380–1384.
38. Baker LH, Mebust WK, Chin TD, Chapman AL, Hinthorn D, Towle D. The relationship of herpesvirus to carcinoma of the prostate. J Urol 1981; 125: 370–374.
39. Jackson MA, Kovi J, Heshmat MY, Jones GW, Rao MS, Ahjuwalin BS. Factors involved in the high incidence of prostatic cancer among American blacks. Prog Clin Biol Res 1981; 53: 111–132.
40. Jackson MA, Ahluwalia BS, Attah EB, et al. Characterization of prostatic carcinoma among blacks: a preliminary report. Cancer Chemother Rep 1975; 59: 3–15.
41. Heshmat MY, Kovi J, Herson J, Jones GW, Jackson MA. Epidemiologic association between gonorrhea and prostatic carcinoma. Urology 1975; 6: 457–460.
42. Niijima T, Koiso K. Incidence of prostatic cancer in Japan and Asia. Scand J Urol Nephrol Suppl 1980; 55: 17–21.
43. Rotkin ID. Studies in the epidemiology of prostatic cancer: expanded sampling. Cancer Treat Rep 1977; 61: 173–180.
44. Armenian HK, Lilienfeld AM, Diamond EL, Bross ID. Epidemiologic characteristics of patients with prostatic neoplasms. Am J Epidemiol 1975; 102: 47–54.
45. Graham S, Haughey B, Marshall J, et al. Diet in the epidemiology of carcinoma of the prostate gland. J Natl Cancer Inst 1983; 70: 687–692.
46. Krain LS. Some epidemiologic variables in prostatic carcinoma in California. Prev Med 1974; 3: 154–159.
47. Krain LS. Epidemiologic variables in prostatic cancer. Geriatrics 1973; 28: 93–98.
48. Steele R, Lees REM, Kraus AS, Rao C. Sexual factors in the epidemiology of cancer of the prostate. J Chron Dis 1971; 24: 29–37.
49. Wynder EL, Mabuchi K, Whitmore WF. Epidemiology of cancer of the prostate. Cancer 1971; 28: 344–360.
50. Newell GR, Fueger JJ, Spitz MR, Babaian RJ. A case control study of prostate cancer. Am J Epidemiol 1989; 130: 395–398.
51. Herbert JT, Brikoff JD, Feorino PM, Caldwell G. Herpes simplex virus type 2 and cancer of the prostate. J Urol 1976; 116: 611–612.
52. Luleci G, Sakizli M, Gunlap A, Erkan I, Remzi D. Herpes simplex type 2 neutralization antibodies in patients with cancers of urinary bladder, prostate, and cervix. J Surg Oncol 1981; 16: 327–331.
53. Shah KV, Daniel RW, Stone KR, Elliott AY. Investigation of human urogenital tract tumors for papovavirus etiology: brief communication. J Natl Cancer Inst 1978; 60: 579–582.
54. Haid M, Sharon N. Immunofluorescent evidence of prior herpes simplex virus type-2 infection in prostate carcinoma. Urology 1984; 24: 623–625.
55. Michael RT, Gagnon JH, Laumann EO, Kolata G. Sex in America: A Definitive Study. Boston: Warner Books, 1994.
56. Laumann EO, Gagnon JH, Michael RT, Michaels S. The Social Organization of Sexuality: Sexual Practices in the United States: The Complete Findings from America’s Most Comprehensive Survey of Sexual Behavior. Chicago: The University of Chicago Press, 1994.
57. Bernstein L, Henderson BE. Exogenous hormones. In: Schottenfeld D, Fraumeni JF Jr, eds. Cancer Epidemiology and Prevention. 3rd ed. New York: Oxford University Press, 1996.
58. Nomura AMY, Kolonel LN. Prostate cancer: a current perspective. Epidemiol Rev 1991; 13: 200–227.
59. Perez CA, Fair WR, Ihde DC. Carcinoma of the Prostate. In: Devita VT, Hellmann S, Rosenberg SA, eds. Cancer: Practices and Principles of Oncology. Philadelphia: JB Lippincott Co, 1989.
60. Wilson JD. Recent studies on the mechanism of action of testosterone. N Engl J Med 1972; 287: 1284–1291.

first intercourse; gonorrhea; marriage; meta-analysis; multiple sexual partners; prostatic neoplasms; sexual activity; sexually transmitted disease; syphilis

Supplemental Digital Content

© 2002 Lippincott Williams & Wilkins, Inc.