| |
Ejaculation Frequency and Subsequent Risk of Prostate Cancer - "high ejaculation frequency was related to decreased risk of total prostate cancer."
|
| |
| |
Michael F. Leitzmann, MD; Elizabeth A. Platz, ScD; Meir J. Stampfer, MD; Walter C. Willett, MD; Edward Giovannucci, MD
JAMA. 2004;291:1578-1586.
Sexual activity has been hypothesized to play a role in the development of prostate cancer, but epidemiological data are virtually limited to case-control studies, which may be prone to bias because recall among individuals with prostate cancer could be distorted as a consequence of prostate malignancy or ongoing therapy.
"In this prospective cohort study among predominantly white men, higher ejaculation frequency was not related to increased risk of prostate cancer. Our results suggest that high ejaculation frequency possibly may be associated with a lower risk of total and organ-confined prostate cancer. These associations were not explained by potential risk factors for prostate cancer, such as age, family history of prostate cancer, history of syphilis or gonorrhea, smoking, and diet. Although we cannot exclude a possibly greater risk of advanced prostate cancer with higher recent ejaculation frequency, we did not observe a higher risk of advanced prostate cancer for high ejaculation frequency earlier in life."
"sexual activity is a complex physiological function, which may relate to prostate cancer risk through several nonandrogenic pathways. For example, frequency of ejaculations may modulate prostate carcinogenesis by altering the composition of prostatic fluid. Frequent ejaculations may decrease the intraprostatic concentration of xenobiotic compounds and chemical carcinogens, which readily accumulate in prostatic fluid.44-45 Frequent ejaculations may also reduce the development of intraluminal prostatic crystalloids,46 which have been associated with prostate cancer in some,47-48 but not all pathology studies.49 Because seminal plasma locally reduces host responsiveness50-52 (possibly by factors produced by the prostate gland53), retained prostatic fluid may diminish intraprostatic immune surveillance against tumor cells.
A more speculative possibility linking increased ejaculation frequency with decreased prostate cancer risk is that ejaculation is accompanied by a release of psychological tension during the emission phase,54 which may lower central sympathetic nervous activity when repeated frequently. Prostate epithelial cell division is stimulated by the release of growth factors from adjacent stromal cells that are heavily innervated with 1 adrenergic receptors.55-56
In summary, our results among predominantly white men suggest that ejaculation frequency is not related to increased risk of prostate cancer. High ejaculation frequency may possibly be associated with a lower risk of total and organ-confined prostate cancer. It is unlikely that reverse causation, differences in prostate cancer screening behavior, or confounding are entirely responsible for the observed results. Mechanisms other than the link between androgenicity and ejaculation frequency should be evaluated as potential etiological factors underlying the inverse association between ejaculation frequency and prostate cancer."
Context- Sexual activity has been hypothesized to play a role in the development of prostate cancer, but epidemiological data are virtually limited to case-control studies, which may be prone to bias because recall among individuals with prostate cancer could be distorted as a consequence of prostate malignancy or ongoing therapy.
Objective- To examine the association between ejaculation frequency, which includes sexual intercourse, nocturnal emission, and masturbation and risk of prostate cancer.
Design, Setting, and Participants- Prospective study using follow-up data from the Health Professionals Follow-up Study (February 1, 1992, through January 31, 2000) of 29 342 US men aged 46 to 81 years, who provided information on history of ejaculation frequency on a self-administered questionnaire in 1992 and responded to follow-up questionnaires every 2 years to 2000. Ejaculation frequency was assessed by asking participants to report the average number of ejaculations they had per month during the ages of 20 to 29 years, 40 to 49 years, and during the past year (1991).
Main Outcome Measure- Incidence of total prostate cancer.
Results -
During 222 426 person-years of follow-up, there were 1449 new cases of total prostate cancer, 953 organ-confined cases, and 147 advanced cases of prostate cancer. Most categories of ejaculation frequency were unrelated to risk of prostate cancer. However, high ejaculation frequency was related to decreased risk of total prostate cancer. The multivariate relative risks for men reporting 21 or more ejaculations per month compared with men reporting 4 to 7 ejaculations per month at ages 20 to 29 years were 0.89 (95% confidence interval [CI], 0.73-1.10); ages 40 to 49 years, 0.68 (95% CI, 0.53-0.86); previous year, 0.49 (95% CI, 0.27-0.88); and averaged across a lifetime, 0.67 (95% CI, 0.51-0.89). Similar associations were observed for organ-confined prostate cancer. Ejaculation frequency was not statistically significantly associated with risk of advanced prostate cancer.
Conclusions Our results suggest that ejaculation frequency is not related to increased risk of prostate cancer.
Author Affiliations: Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, Md (Dr Leitzmann); Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Md (Dr Platz); Departments of Epidemiology and Nutrition, Harvard School of Public Health, Boston, Mass (Drs Stampfer, Willett, and Giovannucci); and Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women's Hospital, Boston, Mass (Drs Stampfer, Willett, and Giovannucci).
INTRODUCTION
Sexual activity is hypothesized to affect prostate carcinogenesis through numerous etiologic pathways. One of the most commonly postulated mechanisms implicates increased sexual activity as an indicator of higher androgenic activity and thus a marker for a high-risk population.1 Another mechanism proposes that sexual activity represents a marker for opportunity for exposure to infectious agents, although no sexually transmitted infection has been consistently implicated in prostate cancer development.2
An alternative hypothesis suggests that a reduced ejaculatory output in otherwise normal men is an etiologic risk factor for prostate cancer. That proposition is based on the theory that infrequent ejaculation increases the risk of prostate cancer because of retained carcinogenic secretions in the prostatic acini.3 A further hypothesis implicates repression of sexuality as a risk factor for prostate cancer and is derived from reports of greater sexual drive coupled with deprived sexual activity4 and greater interest in more sexual intercourse than experienced5 among prostate cancer cases compared with controls.
In the United States, 38% of married persons aged 60 years or older reportedly engage in sexual activity between 1 and 4 times per month, and 14% indicate being sexually active at least 5 times per month.6 Although the libido declines with age, sexual activity is common among 70-, 80-, and even 90-year-old men.7 Given that sexual activity is common, including in older men,6-7 and that prostate cancer risk is high,8 any association between these factors would have clinical and public health relevance. A recent meta-analysis9 reported an increased risk of prostate cancer with greater sexual activity (odds ratio, 1.2; 95% confidence interval [CI], 1.1-1.3 for an increase in sexual activity of 3 times per week).
Epidemiological data on sexual activity and prostate cancer are almost entirely limited to case-control studies,2, 9-10 which may be particularly prone to methodological bias because information on prediagnosis sexual activity is collected after the diagnosis of cancer. Sexual function may diminish after the diagnosis of prostate cancer and its treatment,11 and recall of past levels of sexual activity among individuals with prostate cancer could be distorted as a consequence of prostate malignancy or ongoing therapy.
Prospective data on self-reported sexual activity and prostate cancer are restricted to 2 investigations. These studies12-13 considered age at first marriage, marital status, and number of children as measures of sexual activity and found no association between these factors and prostate cancer. Thus, the relationship between sexual activity and risk of prostate cancer is not clear.
To help resolve this issue, we prospectively examined the association between ejaculation frequency and risk of prostate cancer in the Health Professionals Follow-up Study, a large cohort of middle-aged US men. We focused on ejaculation frequency, capturing sexual intercourse, nocturnal emission, and masturbation. Thus, our exposure definition encompasses a wide range of variability in exposure to sexual activity. In addition, we reasoned that by examining ejaculation frequency, the effects of sexual function per se would be more readily distinguishable from effects related to exposure to sexually transmitted agents, which we do not consider in this article.
COMMENT
In this prospective cohort study among predominantly white men, higher ejaculation frequency was not related to increased risk of prostate cancer. Our results suggest that high ejaculation frequency possibly may be associated with a lower risk of total and organ-confined prostate cancer. These associations were not explained by potential risk factors for prostate cancer, such as age, family history of prostate cancer, history of syphilis or gonorrhea, smoking, and diet. Although we cannot exclude a possibly greater risk of advanced prostate cancer with higher recent ejaculation frequency, we did not observe a higher risk of advanced prostate cancer for high ejaculation frequency earlier in life.
Although each of several analytic approaches indicated that high ejaculation frequency was related to decreased risk of total and organ-confined prostate cancer, there are several plausible alternative explanations for our results. We were concerned about the possibility that the observed inverse relationships were due to avoidance of ejaculation among men with early symptoms related to prostate cancer. However, diminished ejaculation frequency as a preclinical consequence of prostate cancer would be expected to be more pronounced among men with advanced prostate cancer than among men with organ-confined prostate cancer, a circumstance that was not supported by our data. In addition, our findings were essentially unaltered when we excluded cases diagnosed in the early years of follow-up. Hence, our results suggest that reverse causation may have accounted for very little, if any of the observed inverse association between high ejaculation frequency and total and organ-confined prostate cancer risk.
A further potential explanation for our results is that men with high ejaculation frequency may wish to preserve their sexual function and, thus, undergo less screening tests for prostate cancer, leading to less diagnosis of organ-confined prostate cancer among these men. The fact that men in the highest category of ejaculation frequency underwent slightly fewer PSA screening tests and less prostate biopsies than most men with lower ejaculation frequencies suggests the possibility of modest detection bias. In contrast to this possible explanation, the inverse relationship with total and organ-confined prostate cancer persisted when the analysis was restricted to men with the opportunity to have prostate cancer detected by PSA. Thus, decreased prostate cancer detection among men with greater ejaculation frequency is unlikely to entirely account for our results.
Because factors such as diet, smoking, physical activity, and the quality of personal relationships are strong determinants of sexual function,15-18 a further potential concern was the possibility that the apparent beneficial effect of greater ejaculation frequency on risk for total and organ-confined prostate cancer was due to the existence of a healthy lifestyle related both to ejaculation frequency and to prostate cancer. We observed similar results before and after controlling for a broad range of lifestyle and dietary factors potentially related to prostate cancer risk. In addition, the fact that the age-adjusted and multivariate-adjusted RRs were almost identical makes it unlikely that an unconsidered factor that correlates with these lifestyle and dietary factors could produce such strong confounding. Thus, our results are probably not due to confounding by purported lifestyle or dietary risk factors for prostate cancer.
Our results are not likely to be explained by differential measurement error in our assessment of ejaculation frequency between cases and noncases. Notwithstanding, self-reported ejaculation frequency may have contained some inaccuracy because of its sensitive nature and the need for individuals to recall ejaculation frequency in the distant past. It is possible that the oldest men in our cohort (men aged 81 years in 1992) may not have accurately recalled their average monthly ejaculation frequency from ages 20 to 29 years. Moreover, recall of past levels of ejaculation frequency may have been more accurate among men who had the highest ejaculation frequencies than among those with lower ejaculation frequencies. Reported ejaculation frequency rates among men in our study are largely consistent with survey data on sexual activity among US adults.6, 19 Furthermore, our mailed questionnaire on ejaculation frequency was identifiable only by study identification number and not by name, which may have resulted in a more accurate report than in-person interviews. Because these data were collected prior to the occurrence of prostate cancer, the accuracy of reported ejaculation frequency should not differ between men with and without subsequent prostate cancer. Thus, error in the measurement of ejaculation frequency would tend to dampen the results, but would not produce an inverse association. In addition, reporting of other lifestyle factors in this cohort of health professionals has been found to be reasonably accurate.20-21
We noted a suggestive decrease in risk of total and organ-confined prostate cancer among men in the lowest category of ejaculation frequency across a lifetime. Whether that finding was due to lower androgenicity among these men remains unknown. The apparent decrease in risk of total and organ-confined prostate cancer among men with a low ejaculation frequency was not due to low prevalence of sexually transmitted infections among these men because adjustment for history of syphilis or gonorrhea did not alter the results.
We only evaluated ejaculation frequency during adulthood, but not during adolescence. The peripubertal period may be of etiologic significance with respect to prostate carcinogenesis because prostate epithelial cell differentiation occurs at this critical period.22 If ejaculation frequency during puberty was most important for prostate carcinogenesis, measuring adult ejaculation frequency would fail to capture the relevant period of exposure. However, our findings suggest that ejaculation frequency during mid and late adulthood rather than in early adulthood are etiologically relevant periods for influencing prostate tumors. Because the inverse relation was observed for organ-confined cases but not advanced cases, sexual activity may be hypothesized only to affect slow-growing, early stage prostate cancers. Our results are generalizable to white US men aged 46 years or older.
Previous investigations on reported ejaculation frequencies or sexual intercourse and prostate cancer are limited to studies of retrospective design and results are mixed. Nine studies observed a statistically significant1, 23-27 or nonsignificant28-30 positive association; 3 studies27, 31-32 reported no association; 7 studies found a statistically significant4-5,10, 33 or nonsignificant34-36 inverse relationship; and 1 study37 found a U-shaped relationship.
Nine4, 24-25,27, 30-32,35-36 of the aforementioned studies found little or no variation in prostate cancer risk according to sexual activity during different ages. However, 1 study35 observed a nonsignificant inverse association between sexual activity before the age of 30 years and prostate cancer, and no relationship with sexual activity in later life. In contrast, 2 other studies38-39 reported a positive association between frequency of sexual intercourse before ages 50 to 60 years and prostate cancer and an inverse relationship for frequency of sexual intercourse after age 60 years. A recent meta-analysis9 of these studies1, 4-5,23-39 reported RRs for sexual activity at 3 times per week of 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. That meta-analysis9 noted the somewhat inconsistent association between frequency of sexual activity and risk of prostate cancer in previous studies.
Several features distinguish our analysis from previous reports on sexual activity and prostate cancer. First, the prospective study design precluded bias attributable to differential recall of sexual activity by men with and without prostate cancer. Second, we focused on ejaculation frequency rather than on frequency of sexual intercourse, which enhanced exposure variability and allowed us to explore the physiological effects of sexual function per se. Third, our analysis included nearly 50% more cases than the number of cases included in any of the previous studies reporting on sexual activity and prostate cancer. Fourth, our study had data on PSA tests, which allowed us to address the possibility of detection bias. Finally, because we controlled for a wide range of medical, lifestyle, and dietary factors, potential confounding by these factors was likely minimized.
Our finding of no association or a possibly inverse association between high ejaculation frequency and prostate cancer is difficult to reconcile with the commonly proposed concept that androgenic stimulation is related both to enhanced libido and to increased risk of prostate cancer. In some studies, circulating testosterone levels are positively associated with prostate cancer risk.40 Limited evidence shows that circulating levels of testosterone or its major metabolite dihydrotestosterone correlate positively with sexual desire,41 erectile function,42 and frequency of orgasms.43
However, sexual activity is a complex physiological function, which may relate to prostate cancer risk through several nonandrogenic pathways. For example, frequency of ejaculations may modulate prostate carcinogenesis by altering the composition of prostatic fluid. Frequent ejaculations may decrease the intraprostatic concentration of xenobiotic compounds and chemical carcinogens, which readily accumulate in prostatic fluid.44-45 Frequent ejaculations may also reduce the development of intraluminal prostatic crystalloids,46 which have been associated with prostate cancer in some,47-48 but not all pathology studies.49 Because seminal plasma locally reduces host responsiveness50-52 (possibly by factors produced by the prostate gland53), retained prostatic fluid may diminish intraprostatic immune surveillance against tumor cells.
A more speculative possibility linking increased ejaculation frequency with decreased prostate cancer risk is that ejaculation is accompanied by a release of psychological tension during the emission phase,54 which may lower central sympathetic nervous activity when repeated frequently. Prostate epithelial cell division is stimulated by the release of growth factors from adjacent stromal cells that are heavily innervated with 1 adrenergic receptors.55-56
In summary, our results among predominantly white men suggest that ejaculation frequency is not related to increased risk of prostate cancer. High ejaculation frequency may possibly be associated with a lower risk of total and organ-confined prostate cancer. It is unlikely that reverse causation, differences in prostate cancer screening behavior, or confounding are entirely responsible for the observed results. Mechanisms other than the link between androgenicity and ejaculation frequency should be evaluated as potential etiological factors underlying the inverse association between ejaculation frequency and prostate cancer.
RESULTS
During 222 426 person-years of follow-up between 1992 and 2000, there were 1449 cases of total prostate cancer, 953 cases of organ-confined prostate cancer, and 147 cases of advanced prostate cancer. The mean (SD) ejaculations per month at ages 20 to 29 years were 15.1 [6.9]; 40 to 49 years, 11.3 [6.1]; 50 to 59 years, 9.4 [6.1]; and 60 years or older, 5.0 [4.5]. Fifty-eight percent of men reported an ejaculation frequency of more than 3 times per week at ages 20 to 29 years, whereas the proportion of men having more than 3 ejaculations per week decreased to 32% at ages 40 to 49 years, and further declined to 22% at ages 50 to 59 years and to 5% at ages 60 years or older. Ejaculation frequency at ages 20 to 29 years showed positive correlations with ejaculation frequency at ages 40 to 49 years (r = 0.70); 50 to 59 years (r = 0.54); and 60 years or older (r = 0.39); all P<.001. The number of ejaculations at ages 40 to 49 years was positively correlated with those at ages 50 to 59 years (r = 0.81) and 60 years or older (r = 0.53); all P<.001.
Age-standardized lifetime ejaculation frequency was evaluated in relation to various risk factors for prostate cancer to assess the potential for confounding (Table 1). Men with greater lifetime ejaculation frequency tended to be physically more active and were more likely to have a history of syphilis or gonorrhea, prostatitis, and vasectomy than men with lower ejaculation frequency. In addition, men with greater ejaculation frequency were more likely to be currently divorced or separated and consumed more total energy, lycopene, fish, alcohol, supplemental vitamin E, and supplemental zinc. Men in the highest category of average lifetime ejaculation frequency were less likely to have a family history of prostate cancer and a history of surgery for enlarged prostate than men in the lower categories of ejaculation frequency. Men in the highest and lowest categories had a lower prevalence of prostate-specific antigen (PSA) screening relative to men in the intermediate categories of ejaculation frequency.
Ejaculation frequency was examined in relation to risk of total prostate cancer (Table 2). In age- and multivariate-adjusted analyses, most categories of ejaculation frequency were unrelated to risk of total prostate cancer. However, a lower risk was observed in the highest category of ejaculation frequency. The multivariate RR for men reporting 21 or more ejaculations per month compared with men reporting between 4 and 7 ejaculations per month at ages 20 to 29 years was 0.89 (95% CI, 0.73-1.10); 40 to 49 years, 0.68 (95% CI, 0.53-0.86); in the prior year, 0.49 (95% CI, 0.27-0.88); and across a lifetime, 0.67 (95% CI, 0.51-0.89). When the entire range of ejaculation frequency was analyzed as a continuous variable in the multivariate model, each increment of 3 ejaculations per week across a lifetime was associated with a 15% (95% CI, 4%-24%) decrease in risk of total prostate cancer. However, there was a suggestive decreased risk of total prostate cancer observed among men in the lowest category of ejaculation frequency at ages 40 to 49 years and across a lifetime. The multivariate RR for men reporting 3 or less ejaculations per month compared with men reporting between 4 and 7 ejaculations per month across a lifetime was 0.89 (95% CI, 0.69-1.15).
Restricting the outcomes to organ-confined prostate cancer, the relationships were similar to those for total prostate cancer, albeit somewhat stronger (Table 3). Each increment of 3 ejaculations per week across a lifetime was associated with a 19% (95% CI, 7%-30%) decrease in risk of organ-confined prostate cancer. Again, a suggestive decreased risk of organ-confined prostate cancer was observed among men in the lowest category of ejaculation frequency at ages 20 to 29 years, 40 to 49 years, and across a lifetime. The multivariate RR for men reporting 3 or fewer ejaculations per month compared with men reporting between 4 and 7 ejaculations per month across a lifetime was 0.72 (95% CI, 0.51-1.01).
In contrast to the results for total and organ-confined prostate cancer, the intermediate categories of ejaculation frequency at ages 40 to 49 years were associated with a lower risk of advanced prostate cancer (Table 4). High ejaculation frequencies during the previous year and across a lifetime were associated with a suggestive increase in risk of advanced prostate cancer. The multivariate RR for men reporting 21 or more ejaculations per month compared with men reporting between 4 and 7 ejaculations per month across a lifetime was 1.76 (95% CI, 0.81-3.80).
The consistency among ejaculation frequencies at ages 20 to 29 years, 40 to 49 years, and during the previous year was examined. Compared with men who were consistently in the lowest 4 categories of ejaculation frequency, men who were consistently in the highest category of frequency of ejaculations showed a markedly reduced risk of total prostate cancer (RR, 0.25; 95% CI, 0.12-0.54) and organ-confined prostate cancer (RR, 0.15; 95% CI, 0.05-0.47). There was not a sufficient number of cases to examine patterns of ejaculation frequency over time in relation to advanced prostate cancer risk.
The results regarding the relationship of ejaculation frequency with prostate cancer were not materially altered after controlling for history of syphilis or gonorrhea or limiting the study population to men (1) younger than 60 years; (2) 60 years or older; (3) married; (4) without a history of prostatitis, a previous diagnosis of enlarged prostate, or surgery for enlarged prostate; (5) with a PSA test by 2000. The results also were not altered when either the first 2 or 4 years of follow-up were excluded and the 1992 reporting of ejaculation frequency was related to incidence of prostate cancer from 1994 to 2000 or 1996 to 2000, respectively (Table 5).
The associations between ejaculation frequency and risk of prostate cancer were not modified by subgroups defined by current body mass index, family history of prostate cancer, history of smoking, history of prostatitis, history of syphilis or gonorrhea, or marital status (all P for interaction >.05).
METHODS
Study Population
The Health Professionals Follow-up Study was initiated in 1986 when 51 529 US predominantly white male health professionals aged 40 to 75 years responded to a questionnaire concerning their medical history and known or suspected risk factors for cancer and other chronic diseases. Subsequently, follow-up questionnaires were mailed every 2 years to the entire cohort to update information on potential risk factors and to identify newly diagnosed illnesses. In 1992, the questionnaire included an assessment of frequency of ejaculation. Men were included if they did not have a diagnosis of cancer (6098 excluded); answered the 1992 questionnaire (8478 excluded); answered the question on ejaculation frequency (6862 excluded); and provided adequate dietary data (749 excluded). Men who had nonmelanoma skin cancer were allowed in the study. The analytic cohort consisted of 29 342 men, who were followed up to 2000. Participants provided informed consent. The Health Professionals Follow-up Study was approved by the human subjects committee of the Harvard School of Public Health, Boston, Mass.
Assessment of Ejaculation Frequency
Ejaculation frequency was assessed in 1992 by asking participants to report the average number of ejaculations they had per month during the ages of 20 to 29 years, 40 to 49 years, and during the past year (1991). There were 6 possible response categories for each age period (none, 1-3, 4-7, 8-12, 13-20, and 21 ejaculations per month). We collapsed the none and 1 to 3 ejaculation categories because of insufficient numbers of participants in the none category.
Case Ascertainment
On each follow-up questionnaire, we asked participants whether they had received a diagnosis of prostate cancer during the prior 2 years. For men who reported a diagnosis of prostate cancer (or next of kin for decedents), we requested permission to obtain medical records and pathological reports to confirm the diagnosis and obtain further details. Most deaths in the cohort were ascertained through reports from family members and searches of the National Death Index. A study investigator who was unaware of questionnaire data used the information received from any procedures or tests conducted during the initial diagnosis, including staging prostatectomy and bone scans, to stage prostate cancer. The major end point was total prostate cancer incidence. We also considered organ-confined prostate cancer and advanced cancer as separate groups. The latter was defined as cancer extending regionally to the seminal vesicle or other adjacent organs, metastasis to pelvic lymph nodes or distant organs (usually bone) at the time of diagnosis, or that was fatal by the end of follow-up (C2 or D or fatal; T3b or T4 or N1 or M1, or fatal). Because stage T1a lesions are typically indolent and are especially prone to detection bias, we excluded these (3% of the total) from our primary analyses.
Data Analysis
Person-time of follow-up for each participant was calculated from the date of return of the 1992 questionnaire to the date of prostate cancer, death, or the end of the study period in 2000. The relative risk (RR) was calculated as the incidence rate of prostate cancer among men in a specific category of ejaculation frequency divided by the rate among a common reference group with adjustment for age. We used the category of 4 to 7 ejaculations per month as the common reference group to achieve meaningful comparisons between increasingly extreme ejaculation frequencies and to ensure stability of the RR estimates. Multivariate RRs were computed using the Cox proportional hazards model.14 The basic multivariate model included covariates that were previously observed to be associated with risk of total or advanced prostate cancer or that have been associated with risk in the literature. Covariates were racial group; family history of prostate cancer; history of vasectomy; body mass index at age 21 years, which was calculated as weight in kilograms divided by square height in meters; height; pack-years of smoking in the previous decade; history of type 2 diabetes; vigorous physical activity; and intake of total energy, calcium, fructose, supplemental vitamin E, supplemental zinc, red meat, tomato-based foods, fish, a-linolenic acid, and alcohol. To account for lack of proportionality in the hazards across follow-up, we fit separate baseline hazards for groups defined by age and calendar period. Because the relationships between age-specific ejaculation frequencies and prostate cancer could potentially confound each other, we entered terms representing ejaculation frequency at ages 20 to 29 years, 40 to 49 years, and during the year prior to the 1992 questionnaire simultaneously in our models. We also considered ejaculation frequency across a lifetime, which was calculated as the average of the ejaculation frequencies at ages 20 to 29 years, 40 to 49 years, and during the prior year. All reported P values are 2-tailed and P<.05 was considered significant. All analyses were performed using SAS statistical software (release 8.02, SAS Institute Inc, Cary, NC).
|
|
| |
| |
|
|
|
