Development of a job-task-exposure matrix to assess occupational exposure to disinfectants among US nurses.

OBJECTIVES: Occupational exposure to disinfectants is associated with work-related asthma, especially in healthcare workers. However, little is known about the specific products involved. To evaluate disinfectant exposures, we designed job-exposure (JEM) and job-task-exposure (JTEM) matrices, which are thought to be less prone to differential misclassification bias than self-reported exposure. We then compared the three assessment methods: self-reported exposure, JEM and JTEM. METHODS: Disinfectant use was assessed by an occupational questionnaire in 9073 US female registered nurses without asthma, aged 49-68 years, drawn from the Nurses' Health Study II. A JEM was created based on self-reported frequency of use (1-3, 4-7 days/week) of 7 disinfectants and sprays in 8 nursing jobs. We then created a JTEM combining jobs and disinfection tasks to further reduce misclassification. Exposure was evaluated in 3 classes (low, medium, high) using product-specific cut-offs (eg, <30%, 30-49.9%, ≥50%, respectively, for alcohol); the cut-offs were defined from the distribution of self-reported exposure per job/task. RESULTS: The most frequently reported disinfectants were alcohol (weekly use: 39%), bleach (22%) and sprays (20%). More nurses were classified as highly exposed by JTEM (alcohol 41%, sprays 41%, bleach 34%) than by JEM (21%, 30%, 26%, respectively). Agreement between JEM and JTEM was fair-to-moderate (κ 0.3-0.5) for most disinfectants. JEM and JTEM exposure estimates were heterogeneous in most nursing jobs, except in emergency room and education/administration. CONCLUSIONS: The JTEM may provide more accurate estimates than the JEM, especially for nursing jobs with heterogeneous tasks. Use of the JTEM is likely to reduce exposure misclassification.

Gas cooking is associated with small reductions in lung function in children.

Inconsistent effects of gas cooking on lung function have been reported. In a previous study from Austria, we demonstrated a significant, though small, reduction of lung function parameters in children living in homes with gas stoves. We used a larger international database to check if this finding can be generalised. To study the relative impact of cooking with gas on lung function parameters of primary school children in a wide range of geographical settings, we analysed flow and volume data of approximately 24,000 children (aged 6-12 yrs) from nine countries in Europe and North America. Exposure information was obtained by comparable questionnaires and spirometry according to an American Thoracic Society/European Respiratory Society protocol. Linear regressions were used, controlling for individual risk factors and study area. Heterogeneity between study-specific results and mean effects were estimated using meta-analytical tools. On average, gas cooking reduced lung function parameters. Overall effects were small (-0.1-0.7%) and only significant for forced vital capacity and forced expiratory volume in 1 s. There was some indication that allergic children were more affected by gas cooking. Under current housing conditions, gas cooking is associated with only small reductions in lung function.

A prospective study of incident squamous cell carcinoma of the skin in the nurses' health study.

BACKGROUND: Few epidemiologic studies are available that quantify the magnitude of the risk of squamous cell carcinoma (SCC) of the skin associated with sun exposure and related factors such as skin type. In addition, several studies have found an association between cigarette smoking and SCC. PURPOSE: We prospectively examined the risk of developing SCC in relation to phenotype and the effects of sun exposure, as well as to cigarette smoking and other factors, during 8 years of follow-up in a cohort of 107,900 predominantly white women aged 30-55 years at base line in 1976. METHODS: Questionnaires regarding medical history and health-related variables were sent to Nurses' Health Study participants every 2 years, beginning in 1976. Information on constitutional factors (natural hair color, childhood and adolescent tendency to sunburn and tan, and lifetime number of severe sunburns), lifestyle factors (regular time spent outdoors in the summer and sunscreen use), the state lived in at birth and at ages 15 and 30 years, and cigarette smoking habits were ascertained by questionnaire. A total of 197 women with first-incident, histologically confirmed, invasive SCCs that were diagnosed from 1982 to 1990 were included in this analysis. Multivariate analysis using proportional hazards models was used to calculate the relative risks (RRs) and corresponding 95% confidence intervals (CIs), with adjustment for confounders. RESULTS: The risk of SCC was increased in women living in California (RR = 1.8; 95% CI = 1.3-2.6) and Florida (RR = 2.1; 95% CI = 1.1-3.9) at base line, compared with those living in the northeastern states. This risk was higher for women living in those states at birth and at 15 years of age (RR = 2.5; 95% CI = 1.4-4.4 for California and RR = 3.0; 95% CI = 0.7-1.2 for Florida). Red (RR = 2.0; 95% CI = 1.1-3.7) and light brown (RR = 1.7; 95% CI = 1.2-2.4) hair colors were associated with an increased risk of SCC, compared with dark brown hair. After adjusting for the number of sunburns, women who tended to burn after 2 or more hours of sun exposure as children had a slightly higher risk of SCC than those who never burned (RR = 1.5; 95% CI = 0.9-2.5 for burn and RR = 1.1; 95% CI = 0.6-2.0 for painful burn), although the actual number of severe burns appeared to be a more important factor (RR = 2.4; 95% CI = 1.5-4.0 for six or more burns). Finally, current cigarette smokers showed a 50% increase in the risk of SCC compared with never smokers (RR = 1.5; 95% CI = 1.1-2.1). CONCLUSION: Exposure to the sun leading to sunburn, particularly at early ages, should be avoided to decrease the risk of incident SCC.

Risk factors for breast cancer according to family history of breast cancer. For the Nurses' Health Study Research Group.

BACKGROUND: Family history of breast cancer is an established risk factor for this disease and is used to identify women at higher risk, although the impact of risk factors for breast cancer among women with a family history is not well defined. PURPOSE: Using a modified extended log-incidence Pike model, we prospectively examined the impact of risk factors for breast cancer among women with and without a family history of the disease. METHODS: Data analyzed were obtained prospectively from the Nurses' Health Study. Two thousand two hundred forty-nine incident cases of invasive breast cancer were identified in a cohort of 89,132 women aged 30-55 years in 1976 followed biennially through 1990 (1.1 million person years of follow-up). With the use of proportional hazards models, we evaluated the association between risk factors for breast cancer and risk among women with and those without a family history of the disease. We then fit a modified extended log-incidence Pike model to these data. RESULTS: Among women with a family history of breast cancer, reproductive risk factors has associations that were different from those observed among women without a family history of the disease. In particular, we observed little protection from later age at menarche, no protection from multiple births when compared with nulliparity, nor from early, as compared with later, age at first birth. Fitting these data to a model of breast cancer incidence on the basis of reproductive risk factors, we observed an adverse effect of first pregnancy on risk of breast cancer among women with a family history of breast cancer that was approximately 50% greater in magnitude than among women without a family history. Additional births after the first birth conveyed little protection for women with a family history. History of benign breast disease, past use of oral contraceptives, and use of postmenopausal hormones showed relative risks that did not differ between women with a family history and those without a family history of the disease. CONCLUSIONS: We observed a consistent increase in risk of breast cancer among women with a mother or sister history of the disease that was exacerbated by first pregnancy. Among women with a family history of breast cancer, the adverse effect of pregnancy persisted so that to age 70 years, parous women were at higher risk of breast cancer than nulliparous women. Among women without a family history of the disease, first pregnancy was associated with a smaller increase in risk, and early pregnancy and higher number of births were each associated with reduced breast cancer incidence.

Oral contraceptive use and malignant melanoma.

There was no overall relationship between a prior history of oral contraceptive (OC) use and the development of melanoma among 141 cases of nonfatal malignant melanoma and 2,820 age-matched controls drawn from respondents to a large postal survey of registered U.S. nurses; crude relative risk (RR) was 0.93 and 95% confidence limits (CL) were between 0.64 and 1.36. Adjustment for a number of additional variables did not alter this estimate materially. Duration of OC use and interval since first use were similarly unrelated to the occurrence of melanoma. For women diagnosed before age 40, there was a crude positive association of "ever" use of OC and melanoma (RR = 1.78; 95% CL, 1.11-2.86). However, adjustment for geography and other variables diminished this association and rendered it statistically not significant (RR = 1.43, 95% CL, 0.83-2.46). These data do not support the hypothesis that OC use is an independent risk factor for melanoma.

Use of permanent hair dyes and risk of breast cancer.

Aromatic amines contained in permanent hair dyes can be absorbed percutaneously and are mutagenic and carcinogenic in some laboratory studies. Concern has been raised that use of these dyes may increase the risk of human cancers. Therefore, the present study examined the relationship between permanent hair dye use and incidence of breast cancer among 118,404 U.S. women aged 30-55 years who were followed prospectively for 6 years. Among women who had ever used permanent hair dyes, 353 developed breast cancer during 246,848 person-years of follow-up, while 505 cases occurred during 397,460 person-years among never users (age-adjusted rate ratio = 1.1; 95% confidence interval = 0.9-1.2). Identical rate ratios were observed when women who had ever used hair dyes were subdivided into current and past users. Adjustment for known determinants of breast cancer in multivariate models did not alter these relationships. The risk of breast cancer did not increase with more frequent use, longer duration of use, or interval since first use. On the basis of these data and previous findings, it appears unlikely that the use of permanent hair dyes causes any important increase in risk of breast cancer.

Moles and site-specific risk of nonfamilial cutaneous malignant melanoma in women.

We examined the relationship between self-reported mole counts and cutaneous melanoma with respect to anatomic site in 110 case and 231 control female nurses. Counts of moles on the lower leg were better predictors of melanoma risk than were counts of moles on the arm. The relative risk for the highest quintile of lower leg mole counts versus no lower leg moles was 4.2. Mole counts at each site (arm, thigh, and lower leg) were associated with risk of melanoma of the trunk and lower leg, but none were associated with the risk of melanoma of the upper extremity. The absence of direct site-specificity suggests that mole counts primarily indicate systemic melanoma risk, rather than direct risk from the moles themselves.

Prospective study of talc use and ovarian cancer.

BACKGROUND: Perineal talc use has been associated with an increased risk of ovarian cancer in a number of case-control studies; however, this association remains controversial because of limited supporting biologic evidence and the potential for recall bias or selection bias in case-control studies. In this study, we conducted a prospective analysis of perineal talc use and the risk of ovarian cancer. METHODS: The Nurses' Health Study is a prospective study of 121 700 female registered nurses in the United States who were aged 30-55 years at enrollment in 1976. Talc use was ascertained in 1982 by use of a self-administered questionnaire: after exclusions, 78 630 women formed the cohort for analysis. Three hundred seven epithelial ovarian cancers subsequently diagnosed in this cohort through June 1, 1996, were confirmed by medical record review and met inclusion criteria. Proportional hazards models by use of pooled logistic regression were used to derive relative risks (RRs) and 95% confidence intervals (CIs). RESULTS: In 1982, 40.4% (n = 31 789) of the cohort reported ever using talc, and 14.5% (n = 11 411) reported ever using talc daily. We observed no overall association with ever talc use and epithelial ovarian cancer (multivariate RR = 1.09; 95% CI = 0.86-1.37) and no increase in risk of ovarian cancer with increasing frequency of use. There was a modest elevation in risk for ever talc use and invasive serous ovarian cancer (multivariate RR = 1.40; 95% CI = 1.02-1.91). The risk of epithelial ovarian cancer for talc users was not greater among women who had never had a tubal ligation (multivariate RR = 0.97; 95% CI = 0.71-1.32). CONCLUSION: Our results provide little support for any substantial association between perineal talc use and ovarian cancer risk overall; however, perineal talc use may modestly increase the risk of invasive serous ovarian cancer.

Vitamin A supplementation and plasma retinol levels: a randomized trial among women.

Although dietary intake of vitamin A has little, if any, overall effect on blood retinol in generally well-nourished populations, subgroups may exist that would be responsive to supplementation. The hypothesis that vitamin A supplementation increases blood retinol in apparently well-fed individuals with lower than usual blood levels was tested in female health workers, with relatively low blood retinol values, who were randomly assigned to receive vitamin A (10,000 IU daily) or placebo. After 4 weeks the mean change in plasma retinol was -0.4 micrograms/dl for the group receiving placebo and +4.1 micrograms/dl (an increase of 9% over base-line values) for the group receiving vitamin A (P = .02). The results were similar when the base-line retinol level and several other covariates were considered. Thirteen women who had initially received placebo were then switched to vitamin A for 4 weeks. These women experienced a mean increase of 5.3 micrograms/dl in plasma retinol (P = .04). Responses to vitamin A supplementation tend to be greater among women with lower previous total vitamin A intake, as assessed by questionnaire [Spearman rank correlation coefficient (r) = 0.50; P = .01].

Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women.

BACKGROUND: Alcohol use, height, and postmenopausal adiposity have each been positively associated with postmenopausal breast cancer risk in most epidemiologic studies. The mechanism underlying these associations is unclear, although an effect of these factors on hormone levels has been hypothesized. Few previous studies have evaluated the relationship of either alcohol consumption or height with plasma hormone levels. A positive association between adiposity and plasma estrogen levels in postmenopausal women has been reported consistently. PURPOSE: Using archived frozen plasma samples and corresponding data from participants in the Nurses' Health Study, we determined plasma hormone levels and assessed these levels in relation to alcohol consumption, height, and adiposity among postmenopausal women. METHODS: Blood samples were collected from a subset of participants in the Nurses' Health Study in 1989 and 1990, then stored in liquid nitrogen. Hormone concentrations in 217 archived plasma samples (from healthy postmenopausal women) were analyzed in 1993. Spearman correlation coefficients were calculated to assess the linear association between alcohol consumption during the previous year (mean daily intake in grams per day ascertained from semiquantitative food-frequency questionnaires completed in 1990 or 1991), height, and adiposity (as measured by body mass index [BMI] in kg/m2, with weight reported at time of blood collection), and plasma hormone levels. Two-sided P values were also calculated. RESULTS: After controlling for age, height, smoking status, and BMI, alcohol consumption was positively associated with estrone sulfate concentrations (r = .17; P = .02); no statistically significant association was noted for the other plasma hormones measured. Mean plasma estrone sulfate levels were 159 pg/mL in women who reported no alcohol use versus 211 pg/mL in women consuming 30 g or more of alcohol per day. After adjusting for the other covariates, we observed a strong positive correlation between BMI and plasma estrogens (r ranging from .37 for estrone and estrone sulfate to .63 for bioavailable estradiol, with all P values < or = .01; prolactin was the only hormone unassociated with BMI, r = -.01). Height was unrelated to either plasma estrogens or prolactin. CONCLUSIONS: BMI and alcohol use were positively associated with postmenopausal plasma estrogen and estrone sulfate levels, respectively. IMPLICATIONS: The association of alcohol consumption and postmenopausal obesity with subsequent breast cancer risk might be mediated, at least in part, through an influence on postmenopausal plasma estrogen levels. Additional studies are needed to further quantify the relationship between alcohol consumption and plasma hormone levels and to elucidate the physiologic basis for this association.

A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women.

BACKGROUND: A positive correlation between smoking and colorectal cancer has been reported for men, but not for women, who began smoking in substantial numbers in the late 1940s and early 1950s. We hypothesized that smoking acts as an initiator of colorectal neoplasia and that an association with cancer in women has been missed because of the long period between initiation and diagnosis of cancer. PURPOSE: Our purpose was to assess the association between smoking and risk of colorectal adenoma and colorectal cancer in women and to estimate the minimum induction period between the onset of smoking and cancer diagnosis. METHODS: Our data came from the ongoing Nurses' Health Study, in which current and lifetime smoking histories and other potential confounding factors were assessed by questionnaire at baseline and at 2-year intervals. We documented 586 new cases of colorectal cancer from 1976 to 1990 from a group of 118,334 women and 564 new cases of adenoma among 12,143 women who had a first colonoscopy or sigmoidoscopy between 1980 and 1990. We then assessed the relative risk (RR) of small adenoma, large adenoma (> or = 1 cm), and colorectal cancer in terms of pack-years of smoking. RESULTS: The amount smoked in the prior 20 years was related to the prevalence of small adenomas (multivariate RR = 1.45; 95% confidence interval [CI] = 1.25-1.68; P < .0001) and less strongly with large adenomas (RR = 1.31; 95% CI = 1.17-1.47; P < .0001), adjusting for age, intake of saturated fat, dietary fiber, folate, and alcohol, body mass, family history of colorectal cancer, and pack-years of cigarettes smoked within the prior 20 years. Pack-years of cigarettes smoked more than 20 years in the past was associated with risk of large adenoma (multivariate RR for a 20 pack-year increment = 1.29; 95% CI = 1.01-1.64; P trend = .04), but not of small adenoma (RR = 1.11; 95% CI = 0.93-1.34). Cigarette smoking was unrelated to colorectal cancer until 35 years after smoking began, but then became progressively more strongly related with time. Among women who had started smoking more than 10 cigarettes per day 35-39 years in the past, the RR for cancer was 1.47 (95% CI = 1.07-2.01), progressing to 1.63 (95% CI = 1.14-2.33) after 40-44 years and 2.00 (95% CI = 1.14-3.49) after 45 years. CONCLUSIONS: Because the minimum induction period for colorectal cancer appears to be at least 35 years, an association between smoking and colorectal cancer in women may just now be showing up. IMPLICATIONS: Our findings suggest that cigarette smoking may be a previously unrecognized, preventable cause of colorectal cancer among women.

Geographic variation in breast cancer incidence rates in a cohort of U.S. women.

BACKGROUND: Breast cancer mortality and incidence rates vary by geographic region in the United States. Previous analytic studies have measured mortality, not incidence, and have used regional prevalences to control for geographic variation in risk factors rather than adjusting for risk factors measured at the level of the individual. We prospectively evaluated regional variation in breast cancer incidence rates in the Nurses' Health Study and assessed the influence of breast cancer risk factors measured at the individual level. METHODS: The Nurses' Health Study cohort was established in 1976 when 121700 female nurses aged 30-55 years living in 11 U.S. states were enrolled. These states represent all four regions of the continental United States. We identified 3603 incident cases of invasive breast cancer through 1992 (1794565 person-years of follow-up). We calculated relative risks (RRs) adjusted for age and for age and established risk factors (i.e., multivariate-adjusted analysis), comparing California, the Northeast, and the Midwest with the South. RESULTS: For premenopausal women, there was little evidence of regional variation in breast cancer incidence rates, either in age-adjusted or in multivariate-adjusted analyses. For postmenopausal women in California, age-adjusted risk was modestly elevated (RR = 1.24; 95% confidence interval [CI] = 1.05-1.47); after adjusting for age and for established risk factors, the excess rate in California was attenuated by 25% (RR = 1.18; 95% CI = 1.00-1.40). No excess of breast cancer incidence was observed for postmenopausal women in either the Northeast or the Midwest. CONCLUSIONS: Little regional variation in age-adjusted breast cancer incidence rates was observed, with the exception of a modest excess for postmenopausal women in California. Adjustment for differences in the distribution of established risk factors explained some of the excess risk in California.

Plasma prolactin levels and subsequent risk of breast cancer in postmenopausal women.

BACKGROUND: In animal studies, prolactin has been found to be important for mammary epithelial development and its administration has been shown consistently to increase the rate of mammary tumor formation. Previous epidemiologic studies of prolactin and breast cancer risk in postmenopausal women have been limited in size, and the results have been inconsistent. We conducted a nested case-control study within the prospective Nurses' Health Study cohort to better determine the relationship between plasma prolactin levels and postmenopausal breast cancer risk. METHODS: Blood samples were collected from cohort members during the period from 1989 through 1990. Prolactin levels were measured by use of a microparticle enzyme immunoassay. Included in this analysis were 306 postmenopausal women who were diagnosed with breast cancer after blood donation but before June 1994. One or two postmenopausal control subjects were matched per case subject on the basis of age, postmenopausal hormone use, and time of day and month of blood collection; the study included a total of 448 control subjects. RESULTS: In conditional logistic regression analyses, a significant positive association was observed between plasma level of prolactin and postmenopausal breast cancer risk (highest versus lowest quartile, multivariate relative risk = 2.03; 95% confidence interval = 1.24-3.31; two-sided P for trend = .01). The relationship was independent of plasma sex steroid hormone levels and was similar after excluding case subjects diagnosed in the first 2 years after blood collection. CONCLUSIONS: These prospective data suggest that higher plasma prolactin levels are associated with an increased risk of breast cancer in postmenopausal women.

A case-control study of oral contraceptive use and breast cancer.

Among 989 cases of breast cancer and 9,890 controls selected from a cohort of married, female registered nurses aged 30-55 years, the relative risk (RR) of breast cancer for women who had ever used oral contraceptives (OC) compared with those who had never used them was 1.0, with 95% confidence limits 0.9-1.2. Among OC users, there was no consistent pattern of excess risk with increasing duration; in fact, the few women who had used OC longest (greater than 10 yr) had a slightly lower risk than never-users. Moreover, there was no association between OC use and breast cancer among women with a positive history of breast cancer in the mother or sister or with OC use before their first pregnancy. The only subgroup of women among whom any adverse effect was apparent was current OC users aged 50-55 years (two onsets expected vs. seven observed). This finding is consistent with earlier reports of an increased risk of breast cancer among older OC users; however, it is also likely to reflect, at least to some extent, the play of chance, since at ages 45-49 and in each younger age group fewer cases than expected were observed among current OC users.

Rotating night shifts and risk of breast cancer in women participating in the nurses' health study.

BACKGROUND: Melatonin shows potential oncostatic action, and light exposure during night suppresses melatonin production. There is little information, however, about the direct effect of night work on the risk of cancer. We investigated the effect of night work in breast cancer. METHODS: We examined the relationship between breast cancer and working on rotating night shifts during 10 years of follow-up in 78 562 women from the Nurses' Health Study. Information was ascertained in 1988 about the total number of years during which the nurses had worked rotating night shifts with at least three nights per month. From June 1988 through May 1998, we documented 2441 incident breast cancer cases. Logistic regression models were used to calculate relative risks (RRs) and 95% confidence intervals (CIs), adjusted for confounding variables and breast cancer risk factors. All statistical tests were two-sided. RESULTS: We observed a moderate increase in breast cancer risk among the women who worked 1-14 years or 15-29 years on rotating night shifts (multivariate adjusted RR = 1.08 [95% CI = 0.99 to 1.18] and RR = 1.08 [95% CI = 0.90 to 1.30], respectively). The risk was further increased among women who worked 30 or more years on the night shift (RR = 1.36; 95% CI = 1.04 to 1.78). The test for trend was statistically significant (P =.02). CONCLUSIONS: Women who work on rotating night shifts with at least three nights per month, in addition to days and evenings in that month, appear to have a moderately increased risk of breast cancer after extended periods of working rotating night shifts.

Prospective study of oral contraceptive use and risk of breast cancer in women.

In 1976, 118,273 female nurses 30-55 years of age with no history of cancer completed a questionnaire regarding possible risk factors. By 1986, after 1,137,415 person-years of follow-up, we had documented 1,799 newly diagnosed cases of breast cancer. Compared with the risk of breast cancer for nonusers of oral contraceptives, the multivariate relative risks were 1.07 (95% confidence interval, 0.97-1.19) for all users, 1.06 (95% confidence interval, 0.96-1.18) for past users, and 1.53 (95% confidence interval, 1.06-2.19) for current users--women who used oral contraceptives up to 2 years before diagnosis of breast cancer. We conclude that overall past use of oral contraceptives is not associated with a substantial increase in the risk of breast cancer. Although we did not find women who used oral contraceptives before the first pregnancy to have an increased risk of breast cancer, the number of women who used oral contraceptives for a long duration in early reproductive life was too small to permit firm conclusions regarding the risk in this subgroup.

Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women.

BACKGROUND: A positive relationship has generally been observed between plasma estrogen levels and breast cancer risk in postmenopausal women, but most of these studies have been small and few have evaluated specific estrogen fractions (such as percent bioavailable estradiol). In addition, few studies have evaluated plasma androgen levels in relation to breast cancer risk, and their results have been inconsistent. We prospectively evaluated relationships between sex steroid hormone levels in plasma and risk of breast cancer in postmenopausal women by use of a case-control study nested within the Nurses' Health Study. METHODS: Blood samples were collected during the period from 1989 through 1990. Among postmenopausal women not using hormone replacement therapy at blood collection (n = 11,169 women), 156 women were diagnosed with breast cancer after blood collection but before June 1, 1994. Two control subjects were selected per case subject and matched with respect to age, menopausal status, month and time of day of blood collection, and fasting status at the time of blood collection. RESULTS: From comparisons of highest and lowest (reference) quartiles, we observed statistically significant positive associations with risk of breast cancer for circulating levels of estradiol (multivariate relative risk [RR] = 1.91; 95% confidence interval [CI] = 1.06-3.46), estrone (multivariate RR = 1.96; 95% CI = 1.05-3.65), estrone sulfate (multivariate RR = 2.25; 95% CI = 1.23-4.12), and dehydroepiandrosterone sulfate (multivariate RR = 2.15; 95% CI = 1.11-4.17). We found no substantial associations with percent free or percent bioavailable estradiol, androstenedione, testosterone, or dehydroepiandrosterone. The positive relationships were substantially stronger among women with no previous hormone replacement therapy. CONCLUSION: Our data, in conjunction with past epidemiologic and animal studies, provide strong evidence for a causal relationship between postmenopausal estrogen levels and the risk of breast cancer.

Calcium, vitamin D, and the occurrence of colorectal cancer among women.

BACKGROUND: Despite evidence from animal studies for a protective effect of higher calcium and possibly vitamin D intake against colorectal cancer, epidemiologic studies have been inconclusive. PURPOSE: We investigated the associations between the intake of calcium and vitamin D and the occurrence of colorectal cancer. METHODS: In a prospective study, 89 448 female registered nurses who were free of cancer responded to a mailed, semiquantitative food-frequency questionnaire in 1980; dietary information was updated in 1984 and 1986. Through 1992, 501 incident cases of colorectal cancer (396 colon and 105 rectal cancers) were documented. As measures of exposure, we used nutrient intake in 1980 and also two measures of long-term intake on the basis of the three questionnaires: the average of intakes from the three questionnaires and consistent intakes, which were defined as high if women were in the upper tertile on all questionnaires and low if they were in the lower tertile on all questionnaires. To further characterize long-term intake, we conducted analyses excluding women who reported a change in their consumption of milk (primary source of calcium and vitamin D) in the 10 years prior to 1980. Relative risks (RRs) and 95% confidence intervals (CIs) were calculated using the lowest quintile of intake as a reference. The Mantel extension test was used to evaluate linear trends across the categories of nutrient intake. In multivariate analyses, the trends were tested with use of the medians of the intake as a continuous variable in the logistic model. The P values for the trends were two-sided. RESULTS: On the basis of the data from the 1980 questionnaire alone, the multivariate RR for colorectal cancer for women in the upper versus the lower quintile were 0.80 (95% CI = 0.60-1.07) for dietary calcium, 0.84 (95% CI = 0.63-1.13) for dietary vitamin D (from foods only), and 0.88 (95% CI = 0.66-1.16) for total vitamin D (from foods and supplements). After the exclusion of women who reported a change in their milk intake, the RRs for colorectal cancer for the highest versus the lowest categories of average intake were 0.74 (95% CI = 0.36-1.50) for dietary calcium, 0.72 (95% CI = 0.34-1.54) for dietary vitamin D, and 0.42 (95% CI = 0.19-0.91) for total vitamin D. The corresponding RRs for the consistency analyses were 0.70 (95% CI = 0.35-1.39) for dietary calcium, 0.59 (95% CI = 0.30- 1.16) for dietary vitamin D, and 0.33 (95% CI = 0.16-0.70) for total vitamin D. CONCLUSIONS: These findings do not support a substantial inverse association between calcium intake and risk of colorectal cancer, but an inverse association between intake of total vitamin D and risk of colorectal cancer was suggested. IMPLICATIONS: Available evidence does not warrant an increase in calcium intake to prevent colon cancer, but longer-term studies of both calcium and especially vitamin D in relation to colorectal cancer risk are needed.

Prospective study of fruit and vegetable consumption and risk of lung cancer among men and women.

BACKGROUND: Diets high in fruits and vegetables have been shown to be associated with a lower risk of lung cancer. beta-Carotene was hypothesized to be largely responsible for the apparent protective effect, but this hypothesis was not supported by clinical trials. METHODS: We examined the association between lung cancer risk and fruit and vegetable consumption in 77 283 women in the Nurses' Health Study and 47 778 men in the Health Professionals' Follow-up Study. Diet was assessed with the use of a food-frequency questionnaire that included 15 fruits and 23 vegetables. We used logistic regression models to estimate relative risks (RRs) of lung cancer within each cohort. All statistical tests were two-sided. RESULTS: We documented 519 lung cancer cases among the women and 274 among the men. Total fruit and vegetable consumption was associated with a modestly lower risk of lung cancer among the women but not among the men. The RR for the highest versus lowest quintile of intake was 0.79 (95% confidence interval [CI] = 0.59-1.06) among the women and 1.12 (95% CI = 0.74-1.69) among the men after adjustment for smoking status, quantity of cigarettes smoked per day, time since quitting smoking, and age at initiation of smoking. However, total fruit and vegetable consumption was associated with a lower risk of lung cancer among never smokers in the combined cohorts, although the reduction was not statistically significant (RR = 0.63; 95% CI = 0.35-1.12 in the highest tertile). CONCLUSION: Higher fruit and vegetable intakes were associated with lower risks of lung cancer in women but not in men. It is possible that the inverse association among the women remained confounded by unmeasured smoking characteristics, although fruits and vegetables were protective in both men and women who never smoked.

Prospective study of fruit and vegetable consumption and incidence of colon and rectal cancers.

BACKGROUND: Frequent consumption of fruit and vegetables has been associated with a reduced risk of colorectal cancer in many observational studies. METHODS: We prospectively investigated the association between fruit and vegetable consumption and the incidence of colon and rectal cancers in two large cohorts: the Nurses' Health Study (88 764 women) and the Health Professionals' Follow-up Study (47 325 men). Diet was assessed and cumulatively updated in 1980, 1984, 1986, and 1990 among women and in 1986 and 1990 among men. The incidence of cancer of the colon and rectum was ascertained up to June or January of 1996, respectively. Relative risk (RR) estimates were calculated with the use of pooled logistic regression models accounting for various potential confounders. All statistical tests were two-sided. RESULTS: With a follow-up including 1 743 645 person-years and 937 cases of colon cancer, we found little association of colon cancer incidence with fruit and vegetable consumption. For women and men combined, a difference in fruit and vegetable consumption of one additional serving per day was associated with a covariate-adjusted RR of 1.02 (95% confidence interval [CI] = 0.98-1.05). A difference in vegetable consumption of one additional serving per day was associated with an RR of 1.03 (95% CI = 0.97-1.09). Similar results were obtained for women and men considered separately. A difference in fruit consumption of one additional serving per day was associated with a covariate-adjusted RR for colon cancer of 0.96 (95% CI = 0.89-1.03) among women and 1. 08 (95% CI = 1.00-1.16) among men. For rectal cancer (total, 244 cases), a difference in fruit and vegetable consumption of one additional serving per day was associated with an RR of 1.02 (95% CI = 0.95-1.09) in men and women combined. None of these associations was modified by vitamin supplement use or smoking habits. CONCLUSIONS: Although fruits and vegetables may confer protection against some chronic diseases, their frequent consumption does not appear to confer protection from colon or rectal cancer.