Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2007 May 24.
Published in final edited form as: Hum Reprod. 2004 Jun 24;19(8):1746–1754. doi: 10.1093/humrep/deh309

Risk of uterine leiomyomata in relation to tobacco, alcohol and caffeine consumption in the Black Women’s Health Study

Lauren AWise 1,2,7, Julie RPalmer 2, Bernard LHarlow 1,3, Donna Spiegelman 1,4, Elizabeth AStewart 5, Lucile LAdams-Campbell 6, Lynn Rosenberg 2
PMCID: PMC1876785  NIHMSID: NIHMS19327  PMID: 15218005

Abstract

BACKGROUND

Tobacco, alcohol and caffeine consumption may influence risk of uterine leiomyomata via changes in ovarian function or hormone metabolism.

METHODS

We prospectively assessed the relation of these exposures to risk of self-reported uterine leiomyomata in the Black Women’s Health Study. From 1997 to 2001, we followed 21 885 premenopausal women with intact uteri and no prior myoma diagnosis. Cox regression models were used to estimate incidence rate ratios (IRRs) and 95% confidence intervals (CIs).

RESULTS

During 73 426 person-years of follow-up, 2177 incident cases of uterine leiomyomata confirmed by ultrasound (n = 1920) or hysterectomy (n = 257) were reported. Cigarette smoking was not associated with risk of uterine leiomyomata. Risk was positively associated with years of alcohol consumption and current consumption of alcohol, particularly beer. Relative to non-drinkers, multivariate IRRs for beer consumption of <1, 1–6 and 7+ drinks/week were 1.11 (95% CI 0.98–1.27), 1.18 (95% CI 1.00–1.40) and 1.57 (95% CI 1.17–2.11), respectively. Heavy coffee and caffeine consumption were not associated with risk overall, but IRRs were increased among women aged <35 years.

CONCLUSIONS

In US black women, risk of uterine leiomyomata was positively associated with current consumption of alcohol, particularly beer. Cigarette smoking and caffeine consumption were unrelated to risk overall.

Keywords: African-Americans, female, premenopausal, risk factors, uterine neoplasms

Introduction

Uterine leiomyomata (fibroids) are the most common tumours of the female reproductive tract and the leading indication for hysterectomy among US women (Farquhar and Steiner, 2002). Black women are more likely than white women to be diagnosed with uterine leiomyomata (Brett et al., 1997; Marshall et al., 1997; Baird et al., 2003), develop them at earlier ages (Baird et al., 2003), and have more numerous and symptomatic tumours at the time of diagnosis (Kjerulff et al., 1996). The observation that uterine leiomyomatas develop during the reproductive years and regress after menopause suggests that ovarian hormones play a critical role in their aetiology.

Tobacco, alcohol and caffeine consumption are modifiable risk factors that may affect endogenous levels of hormones via changes in ovarian function or alterations in hormone metabolism. Smoking is associated with lower serum and urinary estrogen levels in some (MacMahon et al., 1982; Westhoff et al., 1996), but not all (Longcope and Johnston, 1988; Zumoff et al., 1990; Daniel et al., 1992), studies. Alcohol consumption is associated with higher endogenous levels of estradiol (E2) and estrone (Katsouyami et al., 1991; Reichman et al., 1993; Hankinson et al., 1995), but other studies show no such associations (Cauley et al., 1989; London et al., 1991; Dorgan et al., 1994; Newcomb et al., 1995). Coffee and caffeine consumption are associated with increased levels of early follicular phase E2, independent of alcohol or tobacco use (Lucero et al., 2001), and may enhance sex steroid production (Leonard et al., 1987).

Many epidemiological studies have shown an inverse relationship between current cigarette smoking and risk of uterine leiomyomata (Ross et al., 1986; Romieu et al., 1991; Lumbiganon et al., 1996; Parazzini et al., 1996; Faerstein et al., 2001), but no association was found in the Nurses’ Health Study II (Marshall et al., 1998), a US prospective cohort study. The latter study found a positive association for current alcohol consumption (Marshall et al., 1997). A case–control study of Italian women (Chiaffarino et al., 1999) found no association with alcohol, tea or coffee consumption. Since previous study populations comprised mostly white women (Schwartz and Marshall, 2000), little is known about the impact of these factors in black women. With evidence that premenopausal black women have higher ovarian hormone levels than white women (Woods et al., 1996; Haiman et al., 2002), risk factors for uterine leiomyomata may differ by race. The present study prospectively examines the association of tobacco, alcohol and caffeine consumption with risk of uterine leiomyomata in a large cohort of premenopausal US black women. Results on reproductive and hormonal risk factors for uterine leiomyomata in this cohort have been published previously (Wise et al., 2004).

Patients and methods

Study population

The Black Women’s Health Study (BWHS) is an ongoing prospective cohort study, established in 1995, when ~59 000 US black women aged 21–69 years were enrolled through questionnaires mailed mainly to subscribers of Essence magazine (Rosenberg et al., 1995). The baseline questionnaire elicited information on demographic and behavioural characteristics, reproductive history, health-care utilization, and medical conditions. The cohort is followed every 2 years by postal questionnaire and >80% of the original cohort completed a questionnaire in each follow-up cycle.

Follow-up for the present analysis began in 1997, the start of the second questionnaire cycle, because data on method of confirmation for uterine leiomyomata were first obtained in 1999. Of the 53 322 women who completed the 1997 questionnaire, we restricted the sample to premenopausal women with intact uteri (n = 36 618) and excluded women who reported a diagnosis of leiomyomata before 1997 (n = 10 455), women who did not complete any follow-up questionnaires in 1999 or 2001 (n = 2193), ‘cases’ who provided no information about year of diagnosis (n = 99) or confirmation type (n = 208), and women with incomplete data on smoking, alcohol or caffeine (n = 1015). After these exclusions, 21 885 women remained and were followed for incidence of leiomyomata in the subsequent 4 years.

Assessment of outcome

Because studies limited to histologically confirmed cases of uterine leiomyomata may spuriously identify risk factors associated with large tumour size, symptoms or treatment preference (Schwartz et al., 2000), our outcome definition included confirmation by ultrasound or hysterectomy. This is the same definition as employed by the Nurses’ Health Study II, a prospective study with similar methodology to the BWHS (Marshall et al., 1997). Ultrasound has high sensitivity (99%) and specificity (91%) relative to histological evidence (Loutradis et al., 1990; Dueholm et al., 2002).

On the 1999 and 2001 follow-up questionnaires, women were asked if they had been diagnosed with ‘fibroids’ in the previous 2-year interval and, if ‘yes’, the calendar year in which they were first diagnosed and whether their diagnosis was confirmed by ‘pelvic exam’ and/or by ‘ultrasound/hysterectomy’. A diagnosis was considered ‘hysterectomy-confirmed’ if the woman reported hysterectomy on the same questionnaire.

Incident cases were defined as women who self-reported a first diagnosis of ‘fibroids’ confirmed by ultrasound or hysterectomy. Women with diagnoses confirmed only by pelvic exam (n = 394) were treated as non-cases in primary analyses, because their diagnoses may have represented other pathology (Loutradis et al., 1990). Since their diagnosis may have influenced a change in lifestyle factors, their exposure information was not updated beyond the time of diagnosis.

We assessed the accuracy of self-reported uterine leiomyomata in a random sample of 248 ultrasound- or hysterectomy-confirmed cases. These women were mailed supplemental questionnaires regarding their method of confirmation, symptoms and treatment, and were asked for permission to review their medical records. We obtained medical records for 123 of the 128 women who gave us permission and confirmed the self-report in 118 (96%). The proportion of cases reporting on the supplemental questionnaire the presence of symptoms at the time of diagnosis (71%) was similar among those who did (n = 128) and did not (n = 50) release their medical record (73% versus 60%).

Assesment of exposures and covariates

Data on smoking history were first obtained in 1995 and were updated in 1997 and 1999. Ever-smokers were asked about the age at which they started to smoke regularly, the number of cigarettes they smoked and the cumulative number of years they had smoked. Former smokers were asked how many years ago they quit. Current exposure to environmental tobacco smoke was measured in 1997 and was defined as being ‘in the same room with a smoker for at least 1 h a day for 12 consecutive months or more, at home or in the workplace’.

Alcohol intake was derived mainly from the 1995 questionnaire, because this was the only questionnaire on which we asked about type of alcoholic beverage. Current drinkers were asked to report the average frequency of beer [12 oz (360 ml) can or bottle], wine [4 oz (120 ml) glass] and liquor [1 oz (30 ml) shot] consumption during the previous year. We computed the average daily intake of each beverage by multiplying the frequency of consumption by the alcoholic content in the specified portion size; per drink, beer contains 12.8 g of alcohol, wine 11.0 g and liquor 14.0 g (Department of Agriculture, 1982). Total alcohol intake was the sum of values for all three beverages. In 1997 and 1999, participants were asked about frequency of alcohol consumption.

The 1995 questionnaire included a 68-item Block NCI food frequency questionnaire to assess the consumption of specified foods during the previous year, including caffeinated items (coffee, decaffeinated coffee, tea, soft drinks and chocolate candy), with frequencies ranging from ‘never or <1 per month’ to ‘6 + per day’, and portion sizes of ‘small’, ‘medium’ [8 oz (250 ml) for coffee or tea, 12 oz (360 ml) for soft drinks and 1 oz (31 ml) for chocolate candy] or ‘large’ (Block et al., 1986). This questionnaire was validated in 408 cohort members against three non-consecutive 24-h recalls and a 3-day food diary (Kumanyika et al., 2003). We calculated a summary caffeine score for each subject based on estimates that there are 137 mg per cup of coffee, 47 mg per cup of tea, 46 mg per 12 oz soft drink, 7 mg per serving of chocolate candy and 5 mg per cup of decaffeinated coffee (Willett et al., 1987). These quantities were then multiplied by portion size.

Age at menarche and education were ascertained on the baseline survey. Data on parity, age at each birth, oral contraceptive (OC) use and body mass index (BMI) were obtained on the baseline and follow-up surveys, and were treated as time-dependent variables in the analysis, as was smoking.

Data analysis

Each participant contributed person-time from March 1997 until the diagnosis of uterine leiomyomata, menopause, death, loss to follow-up or end of follow-up (March 2001), whichever came first. Analyses were carried out using SAS statistical software (SAS Institute, 2002). We used multivariate Cox regression to estimate incidence rate ratios (IRRs) and 95% confidence intervals (CIs) for tobacco, alcohol and caffeine consumption (Cox and Oakes, 1984). To control for age, calendar time and any two-way interactions between these two time scales, we stratified our analyses jointly by age in 1-year intervals at the start of follow-up and calendar year of the questionnaire cycle (Hertzmark and Spiegelman, 2001). The Anderson–Gill data structure was used to update time-varying covariates (Therneau, 1997) and exact methods were used to handle tied event times (Kalbfleisch and Prentice, 1980).

A covariate was included in multivariate analyses if the literature supported its role as a risk factor or if adding it to a model containing all other predictors of uterine leiomyomata changed the IRR by 10% or more (Greenland, 1989). Based on these criteria, we adjusted for age at menarche, parity, age at first birth, years since last birth, OC use, education and BMI, and mutually adjusted for smoking, alcohol and caffeine.

Departures from the proportional hazards assumption (i.e. effect modification by age and time) were tested by the likelihood ratio test comparing models with and without cross-product terms for exposures (in their categorical form) with age (<35 versus 35 + years) and time period (1997–1999 versus 1999–2001). In addition, we conducted stratified analyses and computed likelihood ratio tests to evaluate effect modification by education, BMI, parity and OC use, risk factors by which BWHS participants may differ from the general population of black women. Finally, we used Robins’ methods of ‘inverse probability of censoring weighting’ (Robins et al., 2000) to evaluate the impact of differential loss to follow-up. This method constructs a regression model that weights women who were not lost to follow-up more heavily to account for those who were, given the same exposure and covariate history.

Results

At the start of follow-up, the prevalence of current and former smoking in the cohort was 14% and 13%, respectively (Table I). Relative to never smokers, current smokers consumed more alcohol and caffeine, and were older, less educated and less likely to report a recent Pap smear. Current alcohol drinkers (25%) were older than never drinkers, and were more likely to smoke, consume caffeine and have greater energy intake. Median caffeine consumption was 64 mg/day, equivalent to half a cup of caffeinated coffee (data not shown). Heavy consumers of caffeine [500 + mg/day (5%)] were older than light consumers, and were more likely to smoke, consume alcohol, be less educated, and have higher BMI and energy intake.

Table I.

Characteristics of 21 885 women according to tobacco, alcohol and caffeine consumption at the start of follow-up: the Black Women’s Health Study, 1997a

Characteristic Cigarette smoking
Alcohol consumption
Caffeine consumption (mg/day)
Current Former Never Current Former Never <50 150–299 500 +
Number studied 3055 2945 15 885 5367 2684 13 834 9199 3150 1106
Mean of characteristic
 Pack-years of cigarette smoking 9.1 5.3 – 3.9 3.9 1.3 1.5 2.9 5.3
 Current alcohol consumption (drinks/week) (1995)
  Beer 1.5 0.9 0.2 2.0 – – 0.4 0.6 0.8
  Wine 0.7 0.8 0.3 1.7 – – 0.3 0.5 0.7
  Liquor 1.2 0.6 0.2 1.6 – – 0.3 0.5 0.6
 Current alcohol from beer, wine, liquor (g/week) (1995) 43.9 28.5 9.6 67.1 – – 11.7 20.8 27.2
 Caffeine consumption (mg/day) (1995) 212 176 112 179 145 115 22 204 752
 Caffeinated coffee (cups/day) (1995) 1.0 0.8 0.4 0.8 0.6 0.4 0.1 0.6 4.2
 Energy intake (kilocalories/day) (1995) 1638 1636 1512 1637 1642 1495 1334 1838 1918
 Body mass index (kg/m2) 28.2 29.1 27.8 28.0 29.3 27.7 27.4 28.7 29.4
 Education (years) (1995) 14.2 14.7 15.1 14.7 14.6 15.1 15.0 14.8 14.7
 Age at menarche (years) 12.4 12.3 12.3 12.4 12.3 12.3 12.4 12.3 12.2
Percentage with characteristic
 Age (years)
  <30 15.3 10.9 33.9 22.7 22.2 31.5 34.3 22.1 12.9
  30–39 48.0 43.9 45.0 46.6 43.4 45.2 43.8 46.7 42.8
  40 + 36.7 45.2 21.1 30.7 34.4 23.3 21.8 31.2 44.3
 Current exposure to environmental tobacco smoke 67.1 18.7 13.7 31.7 27.8 16.5 16.1 26.7 35.8
 Parous 62.6 63.2 54.4 55.6 64.5 55.2 54.9 57.9 58.7
 Current oral contraceptive use 15.4 21.8 24.7 24.0 19.6 23.6 24.3 21.7 19.4
 Pap smear within past 2 years 87.4 91.0 91.0 91.0 89.2 90.5 91.3 89.2 86.7
Open in a new tab
a

Data from 1997 (start of follow-up) unless otherwise noted. Means and percentages (with exception of age) are standardized to the age distribution of women free of uterine leiomyomata at the start of follow-up. Percentages may not add to 100% due to rounding errors.

During 73 426 person-years of follow-up, 2177 new cases of uterine leiomyomata confirmed by ultrasound (n = 1920) or hysterectomy (n = 257) were reported. There was little evidence of an association between smoking and uterine leiomyomata (Table II). The IRR was slightly reduced for current smokers, but there was no trend across pack-years of smoking, number of cigarettes smoked per day, years of smoking or age first smoked. None of the IRRs varied appreciably by age or by exposure to environmental tobacco smoke (data not shown). To assess differences according to method of confirmation, we repeated our analyses among hysterectomy-confirmed cases only. In this case group, the IRR was 0.65 (95% CI 0.46–0.94) for former smoking and 0.92 (95% CI 0.66–1.29) for current smoking. Results for all other smoking variables were similar to those presented in Table II (data not shown).

Table II.

Risk of ultrasound- or hysterectomy-confirmed uterine leiomyomata in relation to cigarette smoking: the Black Women’s Health Study, 1997–2001

Person-years Cases Age-adjusted IRRa Multivariate IRR (95% CI)b
Smoking status
 Never smoker 53 799 1549 1.00c 1.00c
 Former smoker 9927 329 0.97 0.95 (0.84–1.08)
 Current smoker 9700 299 0.94 0.88 (0.77–1.01)
Pack-years of smokingd
 Never smoker 53 799 1549 1.00c 1.00c
 <2.0 4884 146 0.98 0.94 (0.79–1.12)
 2.0–5.9 4461 144 1.00 0.97 (0.81–1.16)
 6.0–12.4 4679 148 0.91 0.88 (0.73–1.04)
 12.5 + 4716 169 0.99 0.93 (0.79–1.11)
Cigarettes per day
 Non-smoker 63 726 1878 1.00c 1.00c
 <10 2722 94 1.11 1.06 (0.86–1.30)
 10–19 4922 139 0.86 0.82 (0.68–0.98)
 20 + 2045 66 0.94 0.87 (0.78–1.12)
Years of smokingd
 Never smoker 53 799 1549 1.00c 1.00c
 <10 7947 235 0.98 0.94 (0.82–1.09)
 10–19 6661 222 0.97 0.93 (0.80–1.08)
 20 + 4267 152 0.95 0.89 (0.74–1.06)
Years since quittinge
 Never smoker 53 799 1549 1.00c 1.00c
 <5 13 091 395 0.92 0.87 (0.78–0.99)
 5–9 2710 98 1.10 1.08 (0.88–1.34)
 10 + 3662 129 0.98 0.95 (0.79–1.15)
Age at first smokef
 Never smoker 53 799 1549 1.00c 1.00c
 <16 4178 138 1.00 0.94 (0.78–1.13)
 16–17 4282 123 0.86 0.82 (0.68–1.00)
 18–20 6216 197 0.93 0.90 (0.77–1.06)
 21 + 4671 162 1.06 1.02 (0.86–1.20)
Open in a new tab
a

Adjusts for age at start of questionnaire cycle (1-year intervals) and time period (1997–1999 versus 1999–2001).

b

Adjusts for age, time period, age at menarche (years), parity (ever versus never), age at first birth (years), years since last birth (four-knot restricted cubic spline), use of oral contraceptives (current, past, never), education (≤12, 13–15, 16, 17 + ), smoking (never, former, current), caffeine intake (ordinal) and BMI (kg/m2, four-knot restricted cubic spline).

c

Reference group.

d

Omits 21 cases and 887 person-years among ever smokers with missing data.

e

Omits six cases and 164 person-years among ever smokers with missing data.

f

Omits eight cases and 280 person-years among ever smokers with missing data.

There was a small positive association between current alcohol consumption and uterine leiomyomata (Table III). Among ever drinkers, risk was positively associated with years of alcohol consumption. In models that controlled for other sources of alcohol, the IRR increased monotonically with increasing current consumption of beer, but not wine or liquor. The elevated risk among beer drinkers remained when the alcohol variables were assessed in their continuous form: for each additional beer consumed per week, holding other sources of alcohol constant, the multivariate IRR was 1.02 (95% CI 1.01–1.04). The association for beer differed significantly from liquor (P = 0.02), but not wine (P = 0.10) (data not shown). Additional control for variables that distinguished beer drinkers from other drinkers—e.g. occupation and marital status—did not alter these findings. Results were not notably different when we stratified by age, when we repeated analyses among hysterectomy-confirmed cases only, when we updated status or frequency of consumption in 1997 and 1999, or when we used a cumulative average measure for alcohol consumption (data not shown).

Table III.

Risk of ultrasound- or hysterectomy-confirmed uterine leiomyomata in relation to alcohol consumption in 1995: the Black Women’s Health Study, 1997–2001

Person-years Cases Age-adjusted IRRa Multivariate IRR (95% CI)b
Alcohol consumption status
 Never drinker 46 835 1319 1.00c 1.00c
 Former drinker 8804 263 0.99 1.05 (0.91–1.20)
 Current drinker 17 786 595 1.12 1.13 (1.01–1.25)
Current alcohol consumption, drinks per week
 None 55 639 1582 1.00c 1.00c
 1–6 14 207 462 1.09 1.09 (0.98–1.21)
 7–13 2322 86 1.21 1.21 (0.97–1.51)
 14 + 1257 47 1.22 1.23 (0.92–1.66)
Current alcohol consumption, grams per day
 None 55 639 1582 1.00c 1.00c
 <5 7200 223 1.03 1.03 (0.90–1.19)
 5–14 7760 265 1.16 1.15 (1.00–1.31)
 15–24 1590 60 1.22 1.22 (0.94–1.59)
 25 + 1236 47 1.24 1.25 (0.93–1.68)
Years of alcohol consumptiond
 Never drinker 46 835 1319 1.00c 1.00c
 <10 years 7034 189 1.00 1.03 (0.92–1.17)
 10–19 years 7402 260 1.10 1.12 (0.98–1.27)
 20 + years 3237 144 1.27 1.32 (1.11–1.58)
Type of alcohole
 Beer, drinks per week
  None 55 639 1582 1.00c 1.00c,e
  <1 11 042 359 1.08 1.11 (0.98–1.27)
  1–6 5559 183 1.12 1.18 (1.00–1.40)
  7 + 1185 53 1.44 1.57 (1.17–2.11)
 Wine, drinks per week
  None 55 639 1582 1.00c 1.00c,e
  <1 9520 329 1.15 1.16 (1.01–1.33)
  1–6 7657 244 1.07 1.05 (0.91–1.21)
  7 + 609 22 1.19 1.15 (0.74–1.78)
 Liquor, drinks per week
  None 55 639 1582 1.00c 1.00c,e
  <1 11 587 378 1.08 1.08 (0.95–1.23)
  1–6 5441 194 1.21 1.18 (1.00–1.39)
  7 + 758 23 1.00 0.95 (0.61–1.48)
Open in a new tab
a

Adjusts for age at start of questionnaire cycle (1-year intervals) and time period (1997–1999 versus 1999–2001).

b

Adjusts for age, time period, age at menarche (years), parity (ever versus never), age at first birth (years), years since last birth (four knot restricted cubic spline), use of oral contraceptives (current, past, never), education (≤12, 13–15, 16, 17 + ), smoking (never, former, current), caffeine intake (ordinal) and BMI (kg/m2, four knot restricted cubic spline).

c

Reference group.

d

Omits two cases and 113 person-years among current drinkers with missing data.

e

Multivariate models further adjust for other sources of alcohol.

No association was detected for coffee or caffeine consumption overall (Table IV). Since the caffeine data indicated a departure from the proportional hazards assumption, we presented these data within strata of age. The IRRs were significantly elevated in the highest category of coffee and caffeine consumption among women <35 years of age, but there was no evidence of a monotonic dose–response relation for either of these measures. When analyses were confined to hysterectomy-confirmed cases only, findings remained similar overall; small numbers of cases precluded us from assessing differences according to age.

Table IV.

Risk of ultrasound- or hysterectomy-confirmed uterine leiomyomata in relation to caffeine consumption in 1995, according to age at start of questionnaire cycle: the Black Women’s Health Study, 1997–2001

Age <35 years
Age 35 + years
All women
P-value, test for age interaction
Cases Person- years Age-adjusted IRRa Multivariate IRR (95% CI)b Cases Person-years Age-adjusted IRRa Multivariate IRR (95% CI)b Multivariate IRR (95% CI)b
Coffee (caffeinated), cups/day 0.04
 Almost never 598 17 592 1.00c 1.00c 751 21 106 1.00c 1.00c 1.00c
 <1 134 5197 1.13 1.11 (0.92–1.34) 240 6262 1.07 1.04 (0.90–1.20) 1.06 (0.95–1.20)
 1 61 2200 1.15 1.11 (0.85–1.45) 165 4230 1.08 1.04 (0.88–1.23) 1.07 (0.92–1.23)
 2 26 1104 0.98 0.89 (0.60–1.32) 108 2745 1.10 1.06 (0.86–1.30) 1.03 (0.86–1.24)
 3 or more 31 768 1.67 1.53 (1.06–2.22) 63 2223 0.80 0.77 (0.59–1.00) 0.93 (0.75–1.15)
Coffee (decaffeinated), cups/day 0.69
 Almost never 774 34 032 1.00c 1.00c 1108 30 663 1.00c 1.00c
 <1 56 2073 1.12 1.12 (0.86–1.48) 128 3613 0.97 0.96 (0.80–1.15) 1.01 (0.87–1.18)
 1 or more 20 756 1.05 1.07 (0.69–1.68) 91 2290 1.12 1.11 (0.89–1.38) 1.11 (0.92–1.35)
Tea, cups/day 0.75
 Almost never 372 17 571 1.00c 1.00c 631 17 698 1.00c 1.00c 1.00c
 <1 313 13 112 1.12 1.10 (0.94–1.27) 438 12 026 1.02 1.01 (0.89–1.14) 1.05 (0.95–1.15)
 1 88 3354 1.19 1.16 (0.93–1.47) 133 3446 1.10 1.09 (0.90–1.31) 1.12 (0.97–1.30)
 2 34 1352 1.16 1.09 (0.76–1.54) 54 1590 0.95 0.94 (0.71–1.24) 0.99 (0.80–1.23)
 3 or more 43 1471 1.36 1.27 (0.93–1.75) 71 1806 1.11 1.12 (0.88–1.43) 1.18 (0.97–1.43)
Soft drinks, cans/day 0.66
 Almost never 350 15 651 1.00c 1.00c 657 17 812 1.00c 1.00c 1.00c
 <1 333 13 947 1.09 1.16 (1.00–1.35) 431 12 190 0.96 0.97 (0.86–1.10) 1.04 (0.95–1.14)
 1 84 3380 1.14 1.25 (0.98–1.59) 122 3276 1.00 1.04 (0.86–1.27) 1.11 (0.96–1.30)
 2 45 2107 0.96 1.07 (0.78–1.46) 62 1666 1.03 1.04 (0.80–1.35) 1.04 (0.85–1.27)
 3 or more 38 1776 1.00 1.14 (0.81–1.60) 55 1620 0.92 0.93 (0.70–1.22) 0.99 (0.80–1.23)
Caffeine, mg/day 0.01
 <50 359 17 799 1.00c 1.00c 470 13 422 1.00c 1.00c 1.00c
 50–149 300 11 382 1.27 1.27 (1.09–1.49) 408 10 974 1.06 1.05 (0.92–1.20) 1.14 (1.03–1.26)
 150–299 107 4528 1.12 1.13 (0.91–1.41) 235 5964 1.12 1.11 (0.94–1.30) 1.12 (0.99–1.28)
 300–499 42 2085 0.94 0.92 (0.67–1.28) 135 3689 1.03 1.00 (0.82–1.22) 1.00 (0.85–1.18)
 500 + 42 1068 1.79 1.70 (1.23–2.36) 79 2516 0.90 0.87 (0.68–1.11) 1.07 (0.88–1.30)
Open in a new tab

‘Almost never’ = <0.14 servings per day.

a

Adjusts for age at start of questionnaire cycle (1-year intervals) and time period (1997–1999 versus 1999–2001).

b

Adjusts for age, time period, age at menarche (years), parity (ever versus never), age at first birth (years), years since last birth (four knot restricted cubic spline), use of oral contraceptives (current, past, never), education (≤12, 13–15, 16, 17 + ), smoking (never, former, current), current alcohol consumption (<1, 1–6, 7 + per week) and BMI (kg/m2, four knot restricted cubic spline). Estimates for decaffeinated coffee, tea and soft drinks are additionally adjusted for caffeinated coffee (cups per day).

c

Reference group.

Effect modification was not observed by BMI, education, parity or OC use on any of the associations of interest, nor was there any evidence that smoking modified the IRRs for alcohol or caffeine consumption. Because current smokers, alcohol drinkers and heavy consumers of caffeine were less likely to report a recent Pap smear (Table I), a marker of pelvic exam, we restricted analyses to the 90% of women who reported this practice in the prior 2 years. None of the results changed materially. Results were also unchanged when cases confirmed by pelvic examination only (n = 394) were included as part of the outcome definition, censored at the time of diagnosis or excluded from the analysis.

Discussion

The BWHS is the first prospective cohort study to examine risk factors for uterine leiomyomata in a large population of US black women. Although uterine leiomyomata represent a major public health problem among black women of reproductive age, most epidemiological studies have been conducted in white women. In the BWHS, risk was positively associated with current consumption of alcohol, particularly beer. Cigarette smoking and caffeine consumption were unrelated to risk overall.

While some studies show an association between smoking and lower endogenous estrogen levels (MacMahon et al., 1982; Westhoff et al., 1996), others show no such association (Longcope and Johnston, 1988; Zumoff et al., 1990; Daniel et al., 1992). Tobacco components may inhibit aromatase, an enzyme that synthesizes estrogen in granulosa cells (Barbieri et al., 1986), or shift E2 metabolism toward 2-hydroxylation pathways, thereby decreasing estrogen bioavailability (Michnovicz et al., 1986; Bradlow, 1994).

In the epidemiological studies that show an inverse association between current smoking and uterine leiomyomata, decreases in risk range from 20% to 50% (Ross et al., 1986; Romieu et al., 1991; Lumbiganon et al., 1996; Parazzini et al., 1996; Faerstein et al., 2001). The null association for cigarette smoking in the BWHS is consistent with the Nurses’ Health Study II (Marshall et al., 1998). The counteracting effects of tobacco smoke components could explain the lack of association. While smoking may have an anti-estrogen effect on the endogenous hormonal milieu (MacMahon et al., 1982; Westhoff et al., 1996), components of cigarette smoke (e.g. dioxin) may also exert estrogen-related effects on the uterus that could promote cell proliferation (Ohtake et al., 2003).

High levels of estrogens, prolactin and insulin-like growth factor (IGF) may promote growth of uterine leiomyomata (Mora et al., 1995; Andersen, 1996). Several studies have shown associations between alcohol intake and high levels of plasma or urinary E2 (Katsouyami et al., 1991; Reichman et al., 1993; Hankinson et al., 1995), estrone, androstenedione, IGF-I and prolactin (Soyka et al., 1991; Singletary and Gapstur, 2001), as well as lower levels of FSH (Singletary and Gapstur, 2001), while others have shown no such associations (Cauley et al., 1989; London et al., 1991; Dorgan et al., 1994; Newcomb et al., 1995).

In the present study, risk of uterine leiomyomata was positively associated with years of alcohol consumption and current consumption of alcohol, particularly beer. The BWHS is the first study to examine risk in relation to type of alcoholic beverage. The Nurses’ Health Study II also found a positive association for current alcohol consumption, but did not report effect estimates (Marshall et al., 1997), while an Italian case–control study found no association (Chiaffarino et al., 1999). In the latter study, subjects were interviewed after the diagnosis of uterine leiomyomata, and no data on portion size were elicited.

In the BWHS, there appeared to be a stronger association for beer consumption than for wine or liquor consumption. This finding requires confirmation by future studies. It is possible that beer exerts a different effect than other types of alcohol on hormone-dependent neoplasms. In the Italian case–control study, wine accounted for >90% of the alcohol consumed (Chiaffarino et al., 1999), which could explain the lack of association between alcohol and uterine leiomyomata in that study. A meta-analysis of cohort studies showed that beer drinkers have the largest relative risk of breast cancer, although the beverage-specific risks were not significantly different (Smith-Warner et al., 1998). Moreover, a recent study showed that the phytoestrogen in beer, 8-prenylnarigenin, stimulates the growth of MCF-7/6 breast cancer cell lines in vitro, and may mimic the effects of 17β-E2 (Rong et al., 2001).

Few studies have examined the effect of caffeine on ovarian hormones. In a cross-sectional study, coffee and total caffeine consumption were associated with increased levels of early follicular phase E2, independent of alcohol or tobacco use (Lucero et al., 2001). By inhibiting phosphodiesterase, caffeine may decrease clearance of cAMP and enhance steroid production (Leonard et al., 1987). Moreover, in high doses, caffeine can induce stress-like effects in the pituitary–adrenal axis (Spiller, 1998), which could raise risk of uterine leiomyomata via increased secretion of prolactin (Reichlin, 1988; Mora et al., 1995; Andersen, 1996).

The one previous study to examine caffeine consumption as a risk factor for uterine leiomyomata was a case–control study, and no association was found (Chiaffarino et al., 1999). In the BWHS, no overall association was observed for coffee or caffeine consumption, but an increased risk was found for heavy consumption in younger women. This latter finding should be interpreted with caution, because there is no evidence that caffeine metabolism differs by age in humans (Arnaud, 1993). It is also possible that unmeasured factors related both to heavy consumption of caffeine and risk of leiomyomata confounded the results, or that reverse causality played a role (e.g. subclinical cases could have experienced anemia and fatigue, leading to heavy caffeine consumption). There was no evidence that differential screening by age and caffeine consumption accounted for the observed effect modification. Pap smear screening levels were similar for both younger (88%) and older (85%) heavy caffeine consumers, and within age groups, the heavy consumers were consistently less likely than the light consumers to report a recent Pap smear.

BWHS participants were not systematically screened for uterine leiomyomata. Owing to the high cumulative incidence of these tumours and their tendency to be asymptomatic (Baird et al., 2003), true cases may have been misclassified as non-cases. If misclassification was unrelated to tobacco, alcohol or caffeine consumption, the IRRs would have been biased towards the null, unless there was perfect specificity, in which case the IRR would be unbiased (Greenland, 1998). Since we were able to confirm 96% of self-reported cases in the BWHS validation study, specificity of outcome classification was high. In contrast, if exposure status influenced the likelihood of detection, over- or underestimation of the IRR could have occurred. Since restriction of the analytic sample to women with a recent Pap smear yielded similar results, detection bias is unlikely to account for our findings. Moreover, 71% of validation study cases had symptoms prior to their initial diagnosis, suggesting that the proportion of incidentally detected cases was low.

One strength of the present study is that exposure data were collected before the diagnosis and confirmation of uterine leiomyomata, thereby avoiding recall bias. In addition, low loss to follow-up in the BWHS reduced the potential for selection bias. While mean length of follow-up was similar for current and never smokers, current and never alcohol drinkers, and heavy and light caffeine consumers, significant differences existed for educational attainment (40 versus 43 months of follow-up for <13 versus 17 + years of education; P < 0.001). As education was inversely associated with smoking and alcohol consumption, and positively associated with risk of uterine leiomyomata, differential loss to follow-up may have produced a downward bias of the IRRs. However, use of Robins’ methods (Robins et al., 2000) did not change the results appreciably.

Prevalence estimates for smoking and drinking in the BWHS were slightly lower than those documented in nationwide representative studies of black women (US Department of Health and Human Services, 2001; Centers for Disease Control, 2002). Because we did not find effect modification by any factor, with the exception of age, BWHS findings should be generalizable to other black women.

At present, few modifiable risk factors for uterine leiomyomata have been identified. Observations from the present study regarding smoking and alcohol consumption are consistent with another large US prospective study (Marshall et al., 1998). Future studies are needed to evaluate whether the BWHS findings for beer, coffee and caffeine consumption can be replicated. As the associations are modest, and the prevalence of these risk factors low, they are unlikely to explain a large fraction of the disease burden among US black women.

Acknowledgments

The authors gratefully acknowledge the study participants and staff of the Black Women’s Health Study. This work was supported by National Cancer Institute grant CA56420.

References

  1. Andersen J. Growth factors and cytokines in uterine leiomyomas. Semin Reprod Endocrinol. 1996;14:269–282. doi: 10.1055/s-2007-1016336. [DOI] [PubMed] [Google Scholar]
  2. Arnaud MJ. Metabolism of caffeine and other components of coffee. In: Garattini S, editor. Caffeine, Coffee and Health. Raven Press; New York, NY, USA: 1993. pp. 43–95. [Google Scholar]
  3. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. High cumulative incidence of uterine leiomyoma in black and white women: ultrasound evidence. Am J Obstet Gynecol. 2003;188:100–107. doi: 10.1067/mob.2003.99. [DOI] [PubMed] [Google Scholar]
  4. Barbieri RL, McShane PM, Ryan KJ. Constituents of cigarette smoke inhibit human granulosa cell aromatase. Fertil Steril. 1986;46:232–236. [PubMed] [Google Scholar]
  5. Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J, Gardner L. A data-based approach to diet questionnaire design and testing. Am J Epidemiol. 1986;124:453–469. doi: 10.1093/oxfordjournals.aje.a114416. [DOI] [PubMed] [Google Scholar]
  6. Bradlow L. Variations in estrogen metabolism. In: Snow R, Hall P, editors. Steroid Contraceptives and Women’s Response. Plenum Press; New York, NY, USA: 1994. pp. 171–178. [Google Scholar]
  7. Brett KM, Marsh JV, Madans JH. Epidemiology of hysterectomy in the United States: demographic and reproductive factors in a nationally representative sample. J Womens Health. 1997;6:309–316. doi: 10.1089/jwh.1997.6.309. [DOI] [PubMed] [Google Scholar]
  8. Cauley JA, Gutai JP, Kuller LH, LeDonne D, Powell JG. The epidemiology of serum sex hormones in postmenopausal women. Am J Epidemiol. 1989;129:1120–1131. doi: 10.1093/oxfordjournals.aje.a115234. [DOI] [PubMed] [Google Scholar]
  9. Centers for Disease Control. Alcohol use among women of childbearing age—United States 1991–1999. MMWR Morb Mortal Wkly Rep. 2002;51:273–276. [PubMed] [Google Scholar]
  10. Chiaffarino F, Parazzini F, La Vecchia C, Chatenoud L, Di Cintio E, Marsico S. Diet and uterine myomas. Obstet Gynecol. 1999;94:395–398. doi: 10.1016/s0029-7844(99)00305-1. [DOI] [PubMed] [Google Scholar]
  11. Cox DR, Oakes D. Analysis of Survival Data. Chapman Hall; London, UK: 1984. [Google Scholar]
  12. Daniel M, Martin AD, Faiman C. Sex hormones and adipose tissue distribution in premenopausal cigarette smokers. Int J Obes Relat Metab Disord. 1992;16:245–254. [PubMed] [Google Scholar]
  13. Department of Agriculture. Provisional Table on the Nutrient Content of Beverages. Human Nutrition Information Service; Washington, DC, USA: 1982. [Google Scholar]
  14. Dorgan JF, Reichman ME, Judd JT, Brown C, Longcope C, Schatzkin A, Campbell WS, Franz C, Kahle L, Taylor PR. The relation of reported alcohol ingestion to plasma levels of estrogens and androgens in premenopausal women (Maryland United States) Cancer Causes Control. 1994;5:53–60. doi: 10.1007/BF01830726. [DOI] [PubMed] [Google Scholar]
  15. Dueholm M, Lundorf E, Hansen ES, Ledertoug S, Olesen F. Accuracy of magnetic resonance imaging and transvaginal ultrasonography in the diagnosis, mapping, and measurement of uterine myomas. Am J Obstet Gynecol. 2002;186:409–415. doi: 10.1067/mob.2002.121725. [DOI] [PubMed] [Google Scholar]
  16. Faerstein E, Szklo M, Rosenshein N. Risk factors for uterine leiomyoma: a practice-based case-control study. I. African–American heritage, reproductive history, body size, and smoking. Am J Epidemiol. 2001;153:1–10. doi: 10.1093/aje/153.1.1. [DOI] [PubMed] [Google Scholar]
  17. Farquhar CM, Steiner CA. Hysterectomy rates in the United States 1990–1997. Obstet Gynecol. 2002;99:229–234. doi: 10.1016/s0029-7844(01)01723-9. [DOI] [PubMed] [Google Scholar]
  18. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health. 1989;79:340–349. doi: 10.2105/ajph.79.3.340. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Greenland S. Basic methods for sensitivity analysis and external adjustment. In: Rothman K, Greenland S, editors. Modern Epidemiology. Lippincott-Raven; Philadelphia, PA, USA: 1998. pp. 350–352. [Google Scholar]
  20. Haiman CA, Pike MC, Bernstein L, Jaque SV, Stanczyk FZ, Afghani A, Peters RK, Wan P, Shames L. Ethnic differences in ovulatory function in nulliparous women. Br J Cancer. 2002;86:367–371. doi: 10.1038/sj.bjc.6600098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hankinson SE, Willett WC, Manson JE, Hunter DJ, Colditz GA, Stampfer MJ, Longcope C, Speizer FE. Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst. 1995;87:1297–1302. doi: 10.1093/jnci/87.17.1297. [DOI] [PubMed] [Google Scholar]
  22. Hertzmark E, Spiegelman D. The SAS MPHREG Macro. Channing Laboratory; Boston, MA, USA: 2001. [Google Scholar]
  23. Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data. John Wiley & Sons, Inc.; New York, NY, USA: 1980. [Google Scholar]
  24. Katsouyami K, Boyle P, Trichopoulos D. Diet and urine estrogens among postmenopausal women. Oncology. 1991;48:490–494. doi: 10.1159/000226987. [DOI] [PubMed] [Google Scholar]
  25. Kjerulff KH, Langenberg P, Seidman JD, Stolley PD, Guzinski GM. Uterine leiomyomas: racial differences in severity, symptoms, and age at diagnosis. J Reprod Med. 1996;41:483–490. [PubMed] [Google Scholar]
  26. Kumanyika SK, Mauger D, Mitchell DC, Phillips B, Smiciklas-Wright H, Palmer JR. Relative validity of food frequency questionnaire nutrient estimates in the Black Women’s Health Study. Ann Epidemiol. 2003;13:111–118. doi: 10.1016/s1047-2797(02)00253-3. [DOI] [PubMed] [Google Scholar]
  27. Leonard TK, Watson RR, Mohs ME. The effects of caffeine on various systems: a review. J Am Diet Assoc. 1987;87:1048–1053. [PubMed] [Google Scholar]
  28. London S, Willett W, Longcope C, McKinlay S. Alcohol and other dietary factors in relation to serum hormone concentrations in women at climacteric. Am J Clin Nutrition. 1991;53:166–171. doi: 10.1093/ajcn/53.1.166. [DOI] [PubMed] [Google Scholar]
  29. Longcope C, Johnston CC. Androgen and estrogen dynamics in pre- and postmenopausal women: a comparison between smokers and nonsmokers. J Clin Endocrinol Metab. 1988;67:379–383. doi: 10.1210/jcem-67-2-379. [DOI] [PubMed] [Google Scholar]
  30. Loutradis D, Antsaklis A, Creatsas G, Hatzakis A, Kanakas N, Gougoulakis A, Michalas S, Aravantinos D. The validity of gynecological ultrasonography. Gynecol Obstet Invest. 1990;29:47–50. doi: 10.1159/000293299. [DOI] [PubMed] [Google Scholar]
  31. Lucero J, Harlow BL, Barbieri RL, Sluss P, Cramer DW. Early follicular phase hormone levels in relation to patterns of alcohol, tobacco, and coffee use. Fertil Steril. 2001;76:723–729. doi: 10.1016/s0015-0282(01)02005-2. [DOI] [PubMed] [Google Scholar]
  32. Lumbiganon P, Rugpao S, Phandhu-fung S, Laopaiboon M, Vudhikamraksa N, Werawatakul Y. Protective effect of depot-medroxyprogesterone acetate on surgically treated uterine leiomyomas: a multicentre case–control study. Br J Obstet Gynaecol. 1996;103:909–914. doi: 10.1111/j.1471-0528.1996.tb09911.x. [DOI] [PubMed] [Google Scholar]
  33. MacMahon B, Trichopoulos D, Cole P, Brown J. Cigarette smoking and urinary estrogens. N Engl J Med. 1982;307:1062–1065. doi: 10.1056/NEJM198210213071707. [DOI] [PubMed] [Google Scholar]
  34. Marshall LM, Spiegelman D, Barbieri RL, Goldman MB, Manson JE, Colditz GA, Willett WC, Hunter DJ. Variation in the incidence of uterine leiomyoma among premenopausal women by age and race. Obstet Gynecol. 1997;90:967–973. doi: 10.1016/s0029-7844(97)00534-6. [DOI] [PubMed] [Google Scholar]
  35. Marshall LM, Spiegelman D, Manson JE, Goldman MB, Barbieri RL, Stampfer MJ, Willett WC, Hunter DJ. Risk of uterine leiomyomata among premenopausal women in relation to body size and cigarette smoking. Epidemiology. 1998;9:511–517. [PubMed] [Google Scholar]
  36. Michnovicz JJ, Hershcopf RJ, Nagunuma H, Bradlow HL, Fishman J. Increased 2-hydroxylation of estradiol as a possible mechanism for the anti-estrogenic effect of cigarette smoking. N Engl J Med. 1986;315:1305–1309. doi: 10.1056/NEJM198611203152101. [DOI] [PubMed] [Google Scholar]
  37. Mora S, Diehl T, Stewart EA. Prolactin is an autocrine growth regulator for human myometrial and leiomyoma cells. J Soc Gynecol Invest. 1995;2:396. doi: 10.1159/000010154. [DOI] [PubMed] [Google Scholar]
  38. Newcomb PA, Klein R, Klein BE, Haffner S, Mares-Perlman J, Cruickshanks KJ, Marcus PM. Association of dietary and life-style factors with sex hormones in postmenopausal women. Epidemiology. 1995;6:318–321. doi: 10.1097/00001648-199505000-00022. [DOI] [PubMed] [Google Scholar]
  39. Ohtake F, Takeyama K, Matsumoto T, Kitagawa H, Yamamoto Y, Nohara K, Tohyama C, Krust A, Mimura J, Chambon P, et al. Modulation of oestrogen receptor signalling by association with the activated dioxin receptor. Nature. 2003;423:545–550. doi: 10.1038/nature01606. [DOI] [PubMed] [Google Scholar]
  40. Parazzini F, Negri E, La Vecchia C, Rabaiotti M, Luchini L, Villa A, Fedele L. Uterine myomas and smoking. Results from an Italian study. J Reprod Med. 1996;41:316–320. [PubMed] [Google Scholar]
  41. Reichlin S. Prolactin and growth hormone secretion in stress. Adv Exp Med Biol. 1988;245:353–376. doi: 10.1007/978-1-4899-2064-5_28. [DOI] [PubMed] [Google Scholar]
  42. Reichman ME, Judd JT, Longcope C, Schatzkin A, Clevidence BA, Nair PP, Campbell WS, Taylor PR. Effects of alcohol consumption on plasma and urinary hormone concentrations in premenopausal women. J Natl Cancer Inst. 1993;85:722–727. doi: 10.1093/jnci/85.9.722. [DOI] [PubMed] [Google Scholar]
  43. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11:550–560. doi: 10.1097/00001648-200009000-00011. [DOI] [PubMed] [Google Scholar]
  44. Romieu I, Walker AM, Jick S. Determinants of uterine fibroids. Post Mark Surveill. 1991;5:119–133. [Google Scholar]
  45. Rong H, Boterberg T, Maubach J, Stove C, Depypere H, Van Slambrouck S, Serreyn R, De Keukeleire D, Mareel M, Bracke M. 8-Prenylnaringenin, the phytoestrogen in hops and beer, upregulates the function of E-cadherin/catenin complex in human mammary carcinoma cells. Eur J Cell Biol. 2001;80:580–585. doi: 10.1078/0171-9335-00190. [DOI] [PubMed] [Google Scholar]
  46. Rosenberg L, Adams-Campbell LL, Palmer JR. The Black Women’s Health Study: a follow-up study for causes and preventions of illness. J Am Med Womens Assoc. 1995;50:56–58. [PubMed] [Google Scholar]
  47. Ross RK, Pike MC, Vessey MP, Bull D, Yeates D, Casagrande JT. Risk factors for uterine fibroids: reduced risk associated with oral contraceptives. Br Med J Clin Res Ed. 1986;293:359–362. doi: 10.1136/bmj.293.6543.359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. SAS Institute. SAS/STAT User’s Guide. SAS Institute; Cary, NC, USA: 2002. version 8.02. [Google Scholar]
  49. Schwartz SM, Marshall LM. Uterine leiomyomata. In: Goldman MB, Hatch MC, editors. Women and Health. Academic Press; San Diego, CA, USA: 2000. pp. 240–252. [Google Scholar]
  50. Schwartz SM, Marshall LM, Baird DD. Epidemiologic contributions to understanding the etiology of uterine leiomyomata. Environ Health Perspect. 2000;108:821–827. doi: 10.1289/ehp.00108s5821. [DOI] [PubMed] [Google Scholar]
  51. Singletary KW, Gapstur SM. Alcohol and breast cancer: review of epidemiologic and experimental evidence and potential mechanisms. JAMA. 2001;286:2143–2151. doi: 10.1001/jama.286.17.2143. [DOI] [PubMed] [Google Scholar]
  52. Smith-Warner SA, Spiegelman D, Yaun SS, van den Brant PA, Folsom AR, Goldbohm RA, Graham S, Holmberg L, Howe GR, Marshall JR, et al. Alcohol and breast cancer in women: a pooled analysis of cohort studies. JAMA. 1998;279:535–540. doi: 10.1001/jama.279.7.535. [DOI] [PubMed] [Google Scholar]
  53. Soyka M, Gorig E, Naber D. Serum prolactin increase induced by ethanol: a dose-dependent effect not related to stress. Psychoneuroendocrinology. 1991;16:441–446. doi: 10.1016/0306-4530(91)90009-i. [DOI] [PubMed] [Google Scholar]
  54. Spiller GA. Basic metabolism and physiological effects of the methylxanthines. In: Spiller GA, editor. Caffeine. CRC Press; Boca Raton, FL, USA: 1998. pp. 225–231. [Google Scholar]
  55. Therneau TM. Extending the Cox model. In: Lin DY, Fleming TR, editors. Proceedings of the First Seattle Symposium in Biostatistics: Survival Analysis. Springer Verlag; New York: 1997. pp. 51–84. [Google Scholar]
  56. US Department of Health and Human Services. Women and Smoking: A Report of the Surgeon General, US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. 2001. [Google Scholar]
  57. Westhoff C, Gentile G, Lee J, Zacur H, Helbig D. Am J Epidemiol. Vol. 144. US Department of Health and Human Services; Rockville MD: 1996. Predictors of ovarian steroid secretion in reproductive-aged women; pp. 381–388. [DOI] [PubMed] [Google Scholar]
  58. Willett W, Reynolds RD, Cottrell-Hoehner S, Sampson L, Browne ML. Validation of a semi-quantitative food frequency questionnaire: comparison with a 1-year diet record. J Am Diet Assoc. 1987;87:43–47. [PubMed] [Google Scholar]
  59. Wise LA, Palmer JR, Harlow BL, Spiegelman D, Stewart EA, Adams-Campbell LL, Rosenberg L. Reproductive factors, hormonal contraception and risk of uterine leiomyomata in African–American women: a prospective study. Am J Epidemiol. 2004;159:113–123. doi: 10.1093/aje/kwh016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Woods MN, Barnett JB, Spiegelman D, Trail N, Hertzmark E, Longcope C, Gorbach SL. Hormone levels during dietary changes in premenopausal African–American women. J Natl Cancer Inst. 1996;88:1369–1374. doi: 10.1093/jnci/88.19.1369. [DOI] [PubMed] [Google Scholar]
  61. Zumoff B, Miller L, Levitt CD, Miller EH, Kalin M, Denman H, Jandorek R, Rosenfeld RS. The effect of smoking on serum progesterone, estradiol, and luteinizing hormone levels over a menstrual cycle in normal women. Steroids. 1990;55:507–511. doi: 10.1016/0039-128x(90)90089-t. [DOI] [PubMed] [Google Scholar]

RESOURCES