Dietary Patterns, Supplement Use, and the Risk of Symptomatic Benign Prostatic Hyperplasia: Results From the Prostate Cancer Prevention Trial

Abstract

This study examined dietary risk factors for incident benign prostatic hyperplasia (BPH) in 4,770 Prostate Cancer Prevention
Trial (1994–2003) placebo-arm participants who were free of BPH at baseline. BPH was assessed over 7 years and was defined
as medical or surgical treatment or repeated elevation (>14) on the International Prostate Symptom Score questionnaire. Diet,
alcohol, and supplement use were assessed by use of a food frequency questionnaire. There were 876 incident BPH cases (33.6/1,000
person-years). The hazard ratios for the contrasts of the highest to lowest quintiles increased 31% for total fat and 27%
for polyunsaturated fat and decreased 15% for protein (all ptrend < 0.05). The risk was significantly lower in high consumers of alcoholic beverages (0 vs. ≥2/day: hazard ratio (HR) = 0.67)
and vegetables (<1 vs. ≥4/day: HR = 0.68) and higher in daily (vs. <1/week) consumers of red meat (HR = 1.38). There were
no associations of supplemental antioxidants with risk, and there was weak evidence for associations of lycopene, zinc, and
supplemental vitamin D with reduced risk. A diet low in fat and red meat and high in protein and vegetables, as well as regular
alcohol consumption, may reduce the risk of symptomatic BPH.

Full text

Original Contribution
Dietary Patterns, Supplement Use, and the Risk of Symptomatic Benign Prostatic
Hyperplasia: Results from the Prostate Cancer Prevention Trial
Alan R. Kristal
1,2
, Kathryn B. Arnold
1
, Jeannette M. Schenk
1,2
, Marian L. Neuhouser
1
, Phyllis
Goodman
1
, David F. Penson
3
, and Ian M. Thompson
4
1
Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA.
2
Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA.
3
Department of Urology, Keck School of Medicine, University of Southern California, Los Angeles, CA.
4
Department of Urology, University of Texas Health Sciences Center, San Antonio, TX.
Received for publication July 6, 2007; accepted for publication December 11, 2007.
This study examined dietary risk factors for incident benign prostatic hyperplasia (BPH) in 4,770 Prostate Cancer
Prevention Trial (1994–2003) placebo-arm participants who were free of BPH at baseline. BPH was assessed over
7 years and was defined as medical or surgical treatment or repeated elevation (>14) on the International Prostate
Symptom Score questionnaire. Diet, alcohol, and supplement use were assessed by use of a food frequency
questionnaire. There were 876 incident BPH cases (33.6/1,000 person-years). The hazard ratios for the contrasts
of the highest to lowest quintiles increased 31% for total fat and 27% for polyunsaturated fat and decreased 15% for
protein (all p
trend
< 0.05). The risk was significantly lower in high consumers of alcoholic beverages (0 vs. 2/day:
hazard ratio (HR) ¼ 0.67) and vegetables (<1 vs. 4/day: HR ¼ 0.68) and higher in daily (vs. <1/week) consumers
of red meat (HR ¼ 1.38). There were no associations of supplemental antioxidants with risk, and there was weak
evidence for associations of lycopene, zinc, and supplemental vitamin D with reduced risk. A diet low in fat and red
meat and high in protein and vegetables, as well as regular alcohol consumption, may reduce the risk of symp-
tomatic BPH.
alcohol drinking; diet; dietary supplements; prostatic hyperplasia
Abbreviations: BPH, benign prostatic hyperplasia; CI, confidence interval; DHA, docosahexaenoic acid; EPA, eicosapentaenoic
acid; FFQ, food frequency questionnaire; HR, hazard ratio; IPSS, International Prostate Symptom Score; PCPT, Prostate
Cancer Prevention Trial.
Benign prostatic hyperplasia (BPH) is one of the most
common medical conditions in older men. Estimates of
BPH prevalence range from 40 percent to 50 percent at 50
years of age to as high as 80 percent for men aged 70 years
(1, 2). Both the high prevalence of BPH and the associated
costs of medical care (approximately 4 billion dollars per
year in the United States) strongly motivate research to
better understand the causes of BPH and identify modifiable
risk factors to prevent or delay the disease (3). The current
literature on BPH risk factors is quite limited. Most reports
have been based on case series, cross-sectional associations,
or hospital-based case-control studies, and few studies have
examined the risk of incident BPH using case definitions
that reflect current medical practice and validated symptom
questionnaires. The only well-established modifiable risk
factor for BPH is obesity and, in particular, abdominal obe-
sity (4–7). Of the studies examining diet, only two have
examined dietary risk factors for incident BPH (8, 9). Large
studies of diet and BPH incidence are neede d to clarify how
diet may affect BPH risk.
Correspondence to Dr. Alan R. Kristal, Cancer Prevention Program, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North,
M4-B402, P.O. Box 19024, Seattle, WA 98109-1024 (e-mail: akristal@fhcrc.org).
925 Am J Epidemiol 2008;167:925–934
American Journal of Epidemiology
ª The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health.
All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.
Vol. 167, No. 8
DOI: 10.10 93/aje/kwm389
Advance Access publication February 7, 2008
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

Symptomatic BPH is caused by two components: enlarge-
ment of the prostate and heightened tone in prostate smooth
muscle, both of which can obstruct urinary flow. Although
the pathogenesis of BPH is not well understood, age-related
changes in hormonal and other growth-regulatory factors are
the likely cause of cellular proliferation (10). Thus, dietary
patterns that alter the hormonal milieu, such as a high-fat
diet, or other regulator factors, such as insulin-like growth
factors, could conceivably affect BPH risk. Prostate smooth
muscle tone is controlled by the sympathetic nervous sys-
tem, which is directly affected by many diet-related factors
including energy intake, hyper- and hypoglycemia, and obe-
sity (11, 12). BPH may also be caused or exacerbated by
chronic inflammation and subsequent oxidative damage
(13), and thus dietary factors such as x-3 fatty acids, poly-
unsaturated fats, and antioxidants may also affect risk.
Clearly, there are many mechanisms whereby dietary pat-
terns could affect the risk of symptomatic BPH.
Here, we give results of a prospective cohort study exam-
ining the 7-year incidence of symptomatic BPH among men
participating in the Prostate Cancer Prevention Trial (PCPT).
Data from the PCPT include rigorous assessment of both the
symptoms and treatment of BPH, as well as extensive in-
formation on diet and other lifestyle factors that may affect
BPH risk. This report examines whether dietary patterns,
supplement use, and alcohol consumption affect the risk
of incident, symptomatic BPH in a population of healthy
men aged 55 years or older.
MATERIALS AND METHODS
Data are from placebo arm participants in the PCPT,
a randomized, placebo-controlled trial testing whether finas-
teride, a 5a-reductase inhibitor, could reduce the 7-year
period prevalence of prostate cancer. Details regarding study
designand participant characteristics have been describedpre-
viously (14). Briefly, 18,880 men aged 55 years or older with
normal digital rectal examinations and prostate-specific
antigen levels of 3 ng/ml or below, as well as no history of
prostate cancer, severe BPH (defined as an International
Prostate Symptom Score (IPSS) of 19 or lower), or clinically
significant coexisting conditions (judged by the clinic physi-
cian to affect survival until or eligibility for the end-of-study
biopsy at 7 years postrandomization), were randomized to
receive finasteride (5 mg/day) or placebo. During the PCPT,
a prostate biopsy was recommended for participants with an
abnormal digital rectal examination or a prostate-specific
antigen level of 4.0 ng/ml or greater; all men were request ed
to undergo biopsy at 7 years postrandomization.
Data collection
Extensive data are available on the demographic and life-
style characteristics of PCPT participants. Details regarding
age, race/ethnicity, education, and history of smoking were
collected at baseline by self-administered questionnaires.
The level of physical activity was assessed by use of a
six-item questionnaire and categorized as ‘‘sedentary,’’
‘‘light,’’ ‘‘moderate,’’ and ‘‘very active’’ (15). Height and
weight were measured at the baseline clinic visit.
At 1 year postrandomization, men completed a 15-page
diet and supplement questionnaire, and clinic staff measured
height, weight, and body circumferences as part of an an-
cillary study protocol. Diet was assessed with a food fre-
quency questionnaire (FFQ) developed specifically for this
population of older men. This FFQ consisted of questions on
the usual consumption over the past year of 99 foods or food
groups and nine beverages, along with 13 questions on food
preparation and purchasing and three questions on usual
consumption of fruits, vegetables, and fried foods. Algo-
rithms for analysis of this questionnaire are available
at http://www.fhcrc.org/science/shared_resources/nutrition/
ffq/tech_doc.pdf. The dietary supplement questionnaire and
its analysis have been described in detail previously (16). In
brief, participants reported usual pills per day for multivita-
mins and antioxidant mixtures and both pills per day and
dose for b-carotene, vitamin C, vitamin E, calcium, and
zinc. In addition, partic ipants reported whether they used
stress-type multivitamins, vitamin D, fish oil, B-complex,
iron, vitamin A, selenium, or niacin at least three times
a week. We conducted an inter- and intramethod reliability
study in 150 randomly sel ected men, to compare nutrients
calculated from the initial FFQ, from six 24-hour recalls
collected over the following year and from an additional
FFQ comple ted after all 24-hour recalls had been collected.
Based on the 128 men who completed at least five 24-hour
recalls, correlations between the first FFQ and 24-hour re-
calls (adjusted for energy and deattenuated for measur ement
error in the 24-hour recalls) were as follows: total fat, 0.71;
polyunsaturated fat, 0.66; monounsaturated fat, 0.66; satu-
rated fat, 0.75; alcohol, 0.84; carbohydrate, 0.70; lycopene,
0.58; b-carotene, 0.58; vitamin D, 0.57; docosahexaenoic acid
(DHA) plus eicosapentaenoic acid (EPA), 0.87; vitamin C,
0.62; calcium, 0.62; vitamin C, 0.40; and zinc, 0.87. Corre-
lations between repeat FFQs were above 0.60 for all nutrients
with the exception of 0.54 for EPA plus DHA.
Extensive medical data, including physician diagnosis of
and tre atment for BPH, prostatitis, diabetes, cardiovascular
disease, and cancer, were collected at the baseline clinic
visit, each annual and 6-month clinic visit, and every 3- and
9-month phone contact between scheduled clinic visits.
At recruitment, randomization, and each annual follow-up
clinic visit, participants completed the seven-item IPSS
self-administered questionnaire (17).
Incident BPH was defined as either the first report of
treatment or the second report of an IPSS of 15 or higher.
Treatments included use of a-blockers, finasteride, or any
surgical intervention (transurethral prostatectomy, balloon
dilation, or laser prostatectomy). Men who repor ted a physi-
cian diagnosis of BPH alone, in the absence of symptoms or
treatment, were not included as cases in this analysis. We
did complete analyses that included men reporting a physi-
cian diagnosis only as BPH cases and report the few differ-
ences in the Results section below.
Statistical methods
All analyses were based on the time between randomiza-
tion and the estimated time of inciden t BPH. For cases
defined by the IPSS, we assigned incidence time as the
926 Kristal et al.
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

midpoint between the second elevated IPSS and preceding
IPSS (most often the previous year). For cases defined by
treatment, which was assessed every 3 months, we assigned
incidence time as the midpoint between the two annual
visits when BPH treatment was first reported. Follow-up
was censored at the last reported IPSS or at the time of
prostate cancer diagnosis. We calculated simple incidence
rates as the annual incidence per 1,000 person-years of ob-
servation. We used Cox proportional hazards models to cal-
culate the associations of diet with the relative hazards of
BPH; p < 0.05 was considered statistically significant, and
p values are reported in the text rounded to the third decimal
place.
This study is based on 9,457 placebo arm participants,
from whom we excluded 33.8 percent with a history of BPH
at baseline (542 for surgery, 12 for medication use, 1,090 for
an IPSS of 15 or higher at either the recruitment or random-
ization visit, and 1,548 who reported a previous BPH di-
agnosis). We excluded an additional 502 men who were
missing at least half of the expected number of postrandom-
ization IPSS values, leaving 5,763 p articipants. We then
excluded 993 men with missing data (404 missing food
frequency questionnaires, 281 with unreliable dietary data
(energy <800 or >5,000 kcal), and 308 missing complete
anthropometry data), leaving 4,770 men for analyses. Most
men missing diet and anthropometry measures were en-
rolled at study sites that chose not to participate in the spe-
cial dietary and anthropometry assessment protocols.
RESULTS
There were 876 incident BPH cases during the 7 years of
follow-up, corresponding to an incidence rate o f 33.6 per
1,000 person-years. Most BPH endpoints were based on
medical treatment (52 percent) or elevated IPSS (41 percent),
and only 7 percent were based on surgery. Table 1 gives
distributions of participants’ baseline demographic and
health-related characteristics, as well as the unadjusted in-
cidence rates for BPH stratified by these characteristics. The
mean age of participants was 62.6 (standard deviation: 5.5)
years and ranged from 54 years to 86 years. Only 24 percent
of men were normal weight (body mass index: <25 kg/m
2
),
and 21 percent had a waist/hip ratio of 1.0 or greater. BPH
incidence rates increased with increasing age, body mass
index, and waist/hip ratio, and they were higher in African
Americans and Hispanics compared with Caucasians.
TABLE 1. Demographic and health-related characteristics of the study sample and their
associations with the incidence of symptomatic benign prostatic hyperplasia, Prostate
Cancer Prevention Trial placebo arm, 1994–2003
Sample Incident benign prostatic hyperplasia
No. %
Events
(no.)
Person-years
(no.)
Rate/1,000
person-years
Total sample 4,770 100 876 26,079 33.6
Age (years)
54–59 1,681 35.2 242 9,472 25.5
60–64 1,489 31.2 250 8,144 30.7
65 1,600 34.5 384 8,463 45.4
Race/ethnicity
African American 153 3.2 37 770 48.0
Hispanic 98 2.1 22 529 41.6
Caucasian 4,460 93.5 808 24,486 33.0
Other 59 1.2 9 293 30.7
Current smoker
Yes 381 8.0 73 2,022 36.1
No 4,382 92.0 803 24,047 33.4
Waist/hip ratio
<0.95 2,065 43.3 356 11,458 31.1
0.95–0.99 1,693 35.5 307 9,229 33.3
1.00–1.04 834 17.5 174 4,467 39.0
1.05 178 3.7 39 925 42.2
Body mass index (kg/m
2
)
<25 1,150 24.1 196 6,450 30.4
25–29 2,486 52.1 452 13,589 33.3
30–34 879 18.4 177 4,661 38.0
35 255 5.3 51 1,379 37.0
Diet, Alcohol, and Risk of Benign Prostatic Hyperplasia 927
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

Table 2 gives the adjusted hazard ratios for BPH associ-
ated with energy and macronutrient intake. There was no
association of energy intake with BPH risk. For each mac-
ronutrient, we give results from two statistical models, la-
beled ‘‘percent energy’’ and ‘‘total energy.’’ Percent energy
models examine the percentage of energy from each mac-
ronutrient (for alcohol, models used categorized drinks per
day), use a linear term for total energy as a covariate, and
can be interpreted as the effect of substituting energy from
each specific macronutrient for other macronutrients. Total
energy models examine energy from each macronutrient
(for alcohol, models used categorized drinks per day), use
a linear term for energy from all other macronutrients as
a covariate, and can be interpreted as the effect of increasing
energy from a specific macronutrient while keeping the en-
ergy from each other macronutrient constant. In percent
energy models, there were statistically significant increases
in BPH risk assoc iated with high percent ages of energy from
total and polyunsaturated fats and significant decreases in
risk associated with a high percentage of energy from pro-
tein and number of alcoholic drinks per day (all p
trend
<
0.05). Comparing men in the highest and lowest quintiles,
risk was 31 percent higher for total fat (p ¼ 0.018), 27
percent higher for polyunsaturated fat (p ¼ 0.025), and 15
percent lower for protein (p ¼ 0.134); compared with less
than 1 drink/month, consuming two or more drinks/day was
associated with a 33 percent (p < 0.001) reduction in risk.
Overall, this pattern of findings was similar in the total
energy models. We also fit a model with linear terms for
energy from fat, carbohydrate, and protein, plus alcohol
categorized by drinks per week, and found significant asso-
ciations with fat (4.5 percent increase per 100 kcal, p ¼
0.003), protein (5.1 percent decrease per 100 kcal, p ¼
0.008), and alcohol (compar ing <1/month with 2/day:
30 percent reduction, p ¼ 0.002). Finally, we examined
whether the finding of increased risk associated with total
fat was attributable to saturated, monounsaturated, or poly-
unsaturated fats specifically. We fit a set of percent energy
models that controlled for total fat and a single total energy
model that included energy from each type of fat; however,
in all of these models, there were no significant associations
of any specific type of fat with risk (data not shown).
Table 3 gives associations of micronutrients with BPH
risk. We give results for nutrients for which we hypothesized
an association with BPH risk because of antioxidant, anti-
inflammatory, or growth-regulatory properties. We report
results for diet alone and for ‘‘total’’ (diet plus supplements)
where appropriate. Results for dietary vitamin E and sele-
nium are not reported because, on the basis of very poor
correlations between FFQ-based dietary intake of these nu-
trients and serum concentrations (18–23), we believe they
cannot be assessed using an FFQ. Both dietary and total zinc
were associated with reduced BPH risk. Compared with
men in the lowest quintile of total zinc intake, those in the
highest quintile had a 32 percent (p ¼ 0.002) lower BPH
risk. Total but not dietary vitamin D was associated with
reduced risk. Compared with men in the lowest quintile of
total vitamin D intake, thos e in the highest quintile had an
18 percent reduced BPH risk (p
trend
¼ 0.032). There was
a suggestive but not statistically significant 18 percent
reduction in BPH risk associated with high lycopene intake
(p
trend
¼ 0.056). This association was modestly stronger
when physician diagnosis of BPH was included as an end-
point (quintile 1 vs. quinti le 5: hazard ratio (HR) ¼ 0.79, 95
percent confidence interval (CI): 0.63, 0.98; p
trend
¼ 0.023).
Neither vitamin C, calcium, nonlycopene carotenoids, nor
long-chain x-3 fatty acids were associated with risk.
Table 4 gives associations of dietary supplement use with
BPH risk. Supplement use is categorized as low, corre-
sponding to no or infrequent use of a supplement, medium,
corresponding to the amounts generally obtained from multi-
vitamins, and high, corresponding to amounts that are gen-
erally only possible from using high-dose single
supplements. The exceptions were EPA plus DHA and sin-
gle vitamin D supplements, for which we had data only on
whether they were used at least three times per week. Thus,
EPA plus DHA was coded as 0 or 0.5 g/day, and vitamin D
from a single supplement was coded as 0 or 10 lg/day.
Because the vitamin D content of multivitamins is also 10
lg, only men who used both multivitamins and single vita-
min D supplements could be in the high-dose vitamin D
category. There were no associations of supplement use with
BPH risk, with the exception of a trend for decreasing BPH
risk with increasing dose of supplemental vitamin D (p
trend
¼
0.048). We also examined results excluding multivitamin
users, because regular multivitamin users have, by defini-
tion, at least moderate intakes of many micronutrients. In
these analyses (not shown), there were no associations of
any supplement with BPH risk. In particular, the hazar d
ratio contrasting users with nonusers of vitamin D supple-
ments was 1.00 (95 percent CI: 0.65, 1.52).
Table 5 gives associa tions of food groups often associated
with prostate health. Compared with men eating red meat
less than o nce per week, men eating red meat at least daily
had a 38 percent increas ed BPH risk (p ¼ 0.044) and, com-
pared with men eating fewer than one serving of vegetables
per day, men eating four or more servings had a 32 percent
decreased BPH risk (p ¼ 0.011). There were no clear dose-
response effects for either red meat or vegetables; however,
the largest associations were in the contrasts between ex-
treme quintiles. In analyses adding physician-diagnosed
BPH as an endpoint, the association with red mean was at-
tenuated and no longer statistically significant (<1 vs. 4
servings/day: HR ¼ 1.30, 95 percent CI: 0.97, 1.75), while
there was a significant dose response for vegetables with
reduced risk (p
trend
¼ 0.023). Neither cruciferous vegeta-
bles, fruit, nor dairy products were associated with BPH
risk.
DISCUSSION
In this large prospective study, we found that diets high in
total fat were associated with increased risk of symptomatic
BPH and that diets high in protein and alcohol were asso-
ciated with decreased risk. In analyses of foods, high vege-
table consumption was associated with lower risk, and high
red meat consumption was associated with increased risk.
There were no associations of antioxidant nutrients, includ-
ing supplemental vitamin E and selenium or total vitamin C,
928 Kristal et al.
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

TABLE 2. Association of energy and macronutrient intake with risk of total incident symptomatic benign prostatic hyperplasia,
Prostate Cancer Prevention Trial placebo arm, 1994–2003
Macronutrient intake
p
trend
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Energy
Total energy (kcal) <1,442 1,442–1,849 1,850–2,272 2,273–2,831 2,832
HR* (95% CI*)y Referent 0.91 (0.74, 1.12) 0.86 (0.70, 1.07) 0.86 (0.70, 1.06) 1.05 (0.86, 1.29) 0.800
Cases/men (nos.) 186/954 172/948 160/955 162/955 196/958
Total fat
Percent energy <25.7 25.7–30.9 31.0–34.9 35.0–38.1 38.2
HR (95% CI)y Referent 1.21 (0.98, 1.51) 1.19 (0.95, 1.48) 1.22 (0.98, 1.53) 1.31 (1.05, 1.63) 0.034
Cases/men (nos.) 147/966 183/947 175/952 179/953 192/952
Total energy (kcal) <410 410–567 568–745 746–989 990
HR (95% CI)z Referent 1.06 (0.86, 1.32) 1.08 (0.86, 1.36) 1.10 (0.86, 1.41) 1.60 (1.19, 2.15) 0.014
Cases/men (nos.) 169/953 174/955 171/961 158/950 204/951
Saturated fat
Percent energy <8.0 8.0–9.7 9.8–11.2 11.3–12.9 13
HR (95% CI)y Referent 0.93 (0.75, 1.16) 0.96 (0.77, 1.19) 1.09 (0.88, 1.34) 1.08 (0.87, 1.33) 0.214
Cases/men (nos.) 171/956 160/960 171/954 186/943 188/957
Total energy (kcal) <128 128–181 182–240 241–326 327
HR (95% CI)z Referent 0.98 (0.79, 1.22) 0.84 (0.66, 1.08) 0.86 (0.65, 1.13) 1.02 (0.70, 1.50) 0.416
Cases/men (nos.) 178/954 177/946 162/963 160/956 199/951
Monounsaturated fat
Percent energy <9.6 9.6–11.7 11.8–13.4 13.5–15.2 15.3
HR (95% CI)y Referent 1.09 (0.88, 1.36) 1.03 (0.82, 1.28) 1.30 (1.05, 1.61) 1.14 (0.92, 1.42) 0.074
Cases/men (nos.) 155/962 179/953 157/938 206/958 179/959
Total energy (kcal) <153 153–216 217–285 286–381 382
HR (95% CI)z Referent 0.99 (0.79, 1.24) 0.93 (0.72, 1.21) 0.87 (0.63, 1.19) 1.02 (0.64, 1.62) 0.426
Cases/men (nos.) 171/953 177/956 173/957 157/950 198/954
Polyunsaturated fat
Percent energy <5.1 5.1–6.1 6.2–7.1 7.2–8.3 8.4
HR (95% CI)y Referent 1.16 (0.94, 1.43) 0.96 (0.77, 1.20) 1.15 (0.93, 1.43) 1.27 (1.03, 1.57) 0.043
Cases/men (nos.) 159/966 180/950 155/954 181/952 201/948
Total energy (kcal) <85 85–116 117–151 152–206 207
HR (95% CI)z Referent 1.08 (0.87, 1.34) 0.93 (0.73, 1.19) 0.96 (0.73, 1.26) 1.13 (0.79, 1.61) 0.949
Cases/men (nos.) 168/954 184/959 160/953 165/951 199/953
Carbohydrates
Percent energy <41.5 41.5–46.0 46.1–50.5 50.6–56.1 56.2
HR (95% CI)y Referent 0.96 (0.78, 1.19) 0.94 (0.76, 1.17) 0.94 (0.76, 1.16) 1.07 (0.87, 1.32) 0.647
Cases/men (nos.) 175/960 177/955 168/944 169/952 187/959
Total energy (kcal) <690 690–900 901–1,100 1,101–1,372 1,373
HR (95% CI)z Referent 0.87 (0.70, 1.09) 1.00 (0.80, 1.25) 0.95 (0.74, 1.22) 1.16 (0.86, 1.55) 0.347
Cases/men (nos.) 187/955 158/952 174/945 162/953 195/965
Protein
Percent energy <14.6 14.6–16.1 16.2–17.6 17.7–19.4 19.5
HR (95% CI)y Referent 1.02 (0.83, 1.25) 0.92 (0.75, 1.13) 0.85 (0.69, 1.05) 0.85 (0.69, 1.05) 0.037
Cases/men (nos.) 187/958 190/949 180/957 158/954 161/952
Total energy (kcal) <236 236–311 312–385 386–489 490
HR (95% CI)z Referent 0.76 (0.61, 0.94) 0.68 (0.53, 0.86) 0.67 (0.51, 0.87) 0.67 (0.47, 0.95) <0.001
Cases/men (nos.) 196/952 171/952 157/953 167/959 185/954
Alcohol (drinks) <1/month 1–3/month 1–6/week 7–13/week 14/week
HR (95% CI)y Referent 0.74 (0.58, 0.96) 0.85 (0.72, 1.01) 0.83 (0.68, 1.01) 0.67 (0.45, 0.84) <0.001
Cases/men (nos.) 324/1,540 72/449 240/1,333 136/738 104/710
HR (95% CI)z Referent 0.74 (0.58, 0.96) 0.86 (0.72, 1.01) 0.83 (0.68, 1.02) 0.68 (0.57, 0.85) 0.001
Cases/men (nos.) 324/1,540 72/449 240/1,333 136/738 104/710
* HR, hazard ratio; CI, confidence interval.
y Controlled for age, race/ethnicity, waist/hip ratio, and total energy (equivalent to the percent energy model).
z Controlled for age, race/ethnicity, waist/hip ratio, and energy from other macronutrients (equivalent to the total energy model).
Diet, Alcohol, and Risk of Benign Prostatic Hyperplasia 929
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

TABLE 3. Association of energy and micronutrient intake with risk of total incident symptomatic benign prostatic hyperplasia,
Prostate Cancer Prevention Trial placebo arm, 1994–2003
Micronutrient intake
p
trend
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Vitamin C (mg/day)
Diet <69.8 69.8–104.1 104.2–142.0 142.1–193.9 194.0
HR* (95% CI*)y Referent 1.05 (0.85, 1.30) 0.93 (0.75, 1.16) 0.98 (0.79, 1.23) 1.02 (0.81, 1.29) 0.899
Cases/men (nos.) 172/955 183/964 162/937 178/958 181/956
Total 104.4 104.5–169.1 169.2–297.1 297.2–711.0 711.1
HR (95% CI)y Referent 0.92 (0.75, 1.13) 0.73 (0.59, 0.91) 0.91 (0.74, 1.12) 0.92 (0.74, 1.14) 0.477
Cases/men (nos.) 192/968 182/956 154/950 177/945 171/951
Zinc (mg/day)
Diet <8.8 8.8–11.7 11.8–14.6 14.7–18.8 18.9
HR (95% CI)y Referent 0.77 (0.62, 0.95) 0.68 (0.54, 0.86) 0.69 (0.53, 0.90) 0.69 (0.50, 0.96) 0.018
Cases/men (nos.) 201/952 167/953 159/959 170/956 179/950
Total <11.5 11.5–17.5 17.6–24.9 25.0–33.2 33.3
HR (95% CI)y Referent 0.70 (0.56, 0.87) 0.81 (0.66, 1.01) 0.79 (0.63, 0.98) 0.68 (0.54, 0.87) 0.026
Cases/men (nos.) 206/951 159/963 181/963 173/943 157/950
Carotenoids (excluding
lycopene) (lg/day)
Diet <4,748 4,748–7,088 7,089–9,724 9,725–14,056 14,057
HR (95% CI)y Referent 0.91 (0.73, 1.12) 0.88 (0.71, 1.09) 0.94 (0.75, 1.16) 0.89 (0.71, 1.12) 0.455
Cases/men (nos.) 181/953 172/960 168/957 182/951 173/949
Total <6,496 6,496–10,499 10,500–15,499 15,500–28,499 28,500
HR (95% CI)y Referent 0.89 (0.72, 1.10) 0.79 (0.64, 0.98) 0.95 (0.77, 1.17) 0.88 (0.71, 1.09) 0.437
Cases/men (nos.) 181/912 164/906 169/996 196/996 166/960
Lycopene (lg/day)
Diet <3,540 3,540–5,616 5,617–8,175 8,176–12,588 12,589
HR (95% CI)y Referent 1.01 (0.82, 1.24) 0.96 (0.77, 1.19) 0.90 (0.72, 1.13) 0.82 (0.65, 1.03) 0.056
Cases/men (nos.) 180/958 190/963 174/945 169/946 163/958
Calcium (mg/day)
Diet <542 542–733 734–940 941–1,237 1,238
HR (95% CI)y Referent 1.02 (0.82, 1.26) 0.93 (0.73, 1.17) 0.87 (0.67, 1.12) 0.95 (0.71, 1.26) 0.393
Cases/men (nos.) 174/958 179/953 169/944 166/951 188/964
Total <617 617–842 843–1,085 1,086–1,445 1,446
HR (95% CI)y Referent 0.95 (0.76, 1.17) 0.86 (0.68, 1.08) 0.88 (0.69, 1.12) 0.96 (0.74, 1.25) 0.647
Cases/men (nos.) 182/955 174/963 159/941 176/962 185/949
Vitamin D (lg/day)
Diet <2.7 2.7–3.9 4.0–5.2 5.3–7.3 7.4
HR (95% CI)y Referent 0.99 (0.80, 1.22) 0.92 (0.73, 1.14) 0.93 (0.74, 1.18) 0.96 (0.75, 1.24) 0.656
Cases/men (nos.) 171/951 178/954 170/951 172/963 185/951
Total <3.6 3.6–5.9 6.0–11.7 11.8–15.5 15.6
HR (95% CI)y Referent 0.90 (0.73, 1.10) 0.88 (0.71, 1.09) 0.77 (0.62, 0.96) 0.82 (0.66, 1.03) 0.032
Cases/men (nos.) 191/954 182/960 178/953 156/951 169/952
EPA* plus DHA* (g/day)
Diet <0.05 0.05–0.10 0.11–0.16 0.17–0.27 0.28
HR (95% CI)y Referent 0.84 (0.68, 1.03) 0.99 (0.81, 1.21) 0.80 (0.65, 1.00) 0.83 (0.67, 1.04) 0.117
Cases/men (nos.) 190/957 167/949 194/955 158/951 167/958
Total <0.05 0.05–0.10 0.11–0.18 0.19–0.34 0.35
HR (95% CI)y Referent 0.85 (0.69, 1.05) 0.96 (0.78, 1.18) 0.85 (0.68, 1.05) 0.95 (0.77, 1.18) 0.672
Cases/men (nos.) 187/955 166/950 182/956 162/952 179/957
* HR, hazard ratio; CI, confidence interval; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid.
y Controlled for age, race/ethnicity, waist/hip ratio, and total energy.
930 Kristal et al.
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

with risk. Dietary but not supplemental zinc was associated
with reduced risk, and use of vitamin D supplements was
associated with reduced risk. Finally, there was a suggestion
that high intake of lycopene, but not other carotenoids, was
associated with reduced risk.
Before discussing the consistency of our findings with
those in the published literature, it is important to note that
research on dietary patterns and BPH is very limited. Most
reports are from small case-control studies in which cases
were men undergoing surgical treatment (24–27) or from
cross-sectional studies examining associations of lower uri-
nary tract symptoms with current diet (28, 29) or serum
micronutrient concentrations (30). Two studies used a longi-
tudinal design to examine true BPH incidence, using either
surgery (8) or the combination of medical and surgical treat-
ment plus the development of severe lower urinary tract
symptoms (9) as BPH endpoints. For dietary assessment,
two studies used a FFQ (9, 25, 31), one used serum micro-
nutrients (30), and the rest collected limited information on
specific foods or food groups. Given these differences in
study design, BPH endpoints, and dietary assessment meth-
ods, inconsistencies in findings across studies are expected.
Our finding that total fat was associated with increased
BPH risk, with no evidence that associa tions were specific to
type of fat, was in part consistent with the two previous
studies that have examined macronutrients and BPH risk.
The large study by Suzuki et al. (31) reported modest in-
creases in the 6-year period prevalence of BPH associated
with high intakes of energy, animal protein, polyunsaturated
fat, and long-chain x-3 fatty acids. Lagiou et al. (25), in
a very small case-control study, reported a nonsignificant
increased risk associated with high intake of polyunsatu-
rated fat. Our finding that regular alcohol consumption
was associated with reduced risk was consistent with find-
ings from many studies that have examined this question (8,
26, 29, 32, 33) and probably due to the effects of alcohol on
the production and metabolism of testosterone (34). We did
examine whether the alcohol finding could be attributed to
avoiding beverages to reduce symptoms; however, BPH in-
cidence was not assoc iated with consumption of either tea or
TABLE 4. Association of dietary supplement use with the risk of total incident
symptomatic benign prostatic hyperplasia, Prostate Cancer Prevention Trial placebo arm,
1994–2003
Supplement use
p
trend
Low Medium High
Multivitamin (pills/week) <1 1–5 6
HR* (95% CI*)y Referent 0.86 (0.65, 1.14) 0.92 (0.80, 1.06) 0.224
Cases/men (nos.) 515/2,686 55/336 306/1,748
Vitamin C (mg/day) <60 60–250 >250
HR (95% CI)y Referent 0.84 (0.70, 1.00) 1.01 (0.87, 1.17) 0.944
Cases/men (nos.) 409/2,135 178/1,082 289/1,553
Vitamin E (mg/day) <8 8–30 >30
HR (95% CI)y Referent 0.94 (0.79, 1.12) 0.95 (0.82, 1.11) 0.486
Cases/men (nos.) 426/2,241 176/983 274/1,546
Calcium (mg/day) <150 150–199 200
HR (95% CI)y Referent 0.82 (0.69, 0.97) 1.05 (0.87, 1.26) 0.713
Cases/men (nos.) 548/2,879 181/1,141 147/750
Zinc (mg/day) <15 15–23 >23
HR (95% CI)y Referent 0.93 (0.80, 1.08) 0.92 (0.73, 1.16) 0.310
Cases/men (nos.) 523/2,760 271/1,520 82/490
EPA* plus DHA* (g/day) 0 0.5
HR (95% CI)y Referent 1.23 (0.97, 1.57) 0.085
Cases/men (nos.) 802/4,415 74/355
Selenium (lg/day) <10 10–30 >30
HR (95% CI)y Referent 0.83 (0.71, 0.97) 1.01 (0.83, 1.25) 0.336
Cases/men (nos.) 528/2,725 238/1,444 110/601
Vitamin D (lg/day) <2.5 2.5–10 11
HR (95% CI)y Referent 0.88 (0.77, 1.02) 0.82 (0.60, 1.11) 0.047
Cases/men (nos.) 528/2,721 305/1,763 43/286
* HR, hazard ratio; CI, confidence interval; EPA, eicosapentaenoic acid; DHA, docosahex-
aenoic acid.
y Controlled for age, race/ethnicity, and waist/hip ratio.
Diet, Alcohol, and Risk of Benign Prostatic Hyperplasia 931
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

coffee. We did not confirm the previous finding of an in-
creased risk associated with long-chain x-3 fatty acids, and
our finding of a decreased risk associated with high protein
intake is novel and requires replication.
Both our study and previously published studies do not
support an association of antioxidant nutrients with BPH
risk. In the only other longitudinal study, Rohrmann et al.
(9) reported decreased risks for 8-year BPH period preva-
lence associated with high intakes of vitamin C from foods
(but not supplements) and individual carotenoids (but not
lycopene and b-carotene) and no associations with to-
copherols. They also found no significant associations of anti-
oxidants with BPH incidence. Lagiou et al. (25) reported no
significant associations with antioxidants, although there
were nonsignificant reduced risks associated with high in-
takes of vitamins C and E. Rohrmann et al. (30) reported no
significant cross-sectional associations of serum antioxi-
dants with lower urinary tract symptoms, although men in
the lowest quintiles of a-tocopherol, lycopene, and selenium
were approximately twice as likely to report symptoms as
were men with higher serum antioxidant levels. We judge it
unlikely that dietary antioxidants play an important role in
preventing symptomatic BPH.
Results from previous studies of specific foods and food
groups are generally consistent with our finding that diets
high in vegetables are associated with lower BPH risk. The
largest studies reported an 11 percent reduction in BPH
prevalence when comparing the lowest and highe st quintiles
of vegetable intake (9) and a 30 percent reduced risk of
symptoms when comparing less than daily with daily con-
sumption of fresh vegetables (28). There is also some sup-
port of our finding that high intake of red meat increases risk
(28, 31).
Our findings on dietary zinc and supplemental vitamin D
are difficult to interpret. The correlation between dietary
zinc and protein, controlled for energy, was 0.52; in models
that included dietary zinc and protein, both associations
were attenuated and not statistically significant and, in mod-
els that included total zinc and protein, only protein was
significant (p ¼ 0.017). Combined with the observation that
supplemental zinc was not associated with risk, it is possible
that the dietary zinc finding is the result of collinearity with
protein. It is also possible that there is a threshold above
which zinc has no additional effect on risk. In analyses
cross-classifying men by dietary and supplemental zinc in-
take using quintile 1 of dietary zinc without supplements as
the comparison group, the hazard ratios were 0.69 (95 per-
cent CI: 0.52, 0.91) for quintile 1 with supplements, 0.60 (95
percent CI: 0.48, 0.77) for quintiles 2–5 without supple-
ments, and 0.61 (95 percent CI: 0.48, 0.78) for quintiles
2–5 with supplements. Vitamin D supplementation was as-
sociated with reduced risk, but our calculated dose of total
vitamin D is imprecise, and the association was observed
only among men who used both multivitamins and single
vitamin D supplements. Larger studies with more detailed
data on supplement use, including frequency, dose, and du-
ration, will be needed to further address whether high-dose
vitamin D supplementation is associated with BPH risk.
Many of the dietary factors that we found to be associated
with BPH risk can affect both steroid hormone concentra-
tions and the sympathetic nervous system. The dietary pat-
tern characterized by low fat, moderate alcohol, and high
vegetables is associated with less obesity (15), lower serum
estrogens and androgens, and higher sex hormone binding
globulin (34, 35) and probably also les s sympathetic ner-
vous stimulation (36). It is possible that these physiologic
TABLE 5. Association of food use with the risk of incident symptomatic benign prostatic hyperplasia, Prostate Cancer Prevention
Trial placebo arm, 1994–2003
Frequency of consumption (servings) p
trend
Red meat (servings/week) <1 1.0–2.9 3.0–4.9 5.0–6.9 7
HR* (95% CI*)y Referent 1.26 (0.97, 1.63) 1.01 (0.77, 1.33) 0.99 (0.74, 1.33) 1.38 (1.01, 1.88) 0.557
Cases/men (nos.) 71/438 291/1,462 202/1,196 145/889 166/784
Dairy (servings/day) <1 1.0–1.9 2.0–2.9 3
HR (95% CI)y Referent 1.02 (0.87, 1.19) 1.00 (0.79, 1.26) 0.94 (0.72, 1.24) 0.762
Cases/men (nos.) 377/2,116 321/1,703 102/535 75/415
Vegetables (servings/day) <1 1.0–1.9 2.0–2.9 3.0–3.9 4
HR (95% CI)y Referent 0.76 (0.62, 0.93) 0.85 (0.69, 1.06) 0.79 (0.60, 1.03) 0.68 (0.50, 0.92) 0.095
Cases/men (nos.) 136/646 303/1,771 257/1,331 104/574 75/447
Cruciferous vegetables
(servings/week) <1 1.0–1.9 2.0–2.9 3.0–3.9 4
HR (95% CI)y Referent 0.90 (0.77, 1.07) 1.12 (0.90, 1.39) 0.94 (0.69, 1.28) 0.88 (0.66, 1.16) 0.592
Cases/men (nos.) 464/2,443 207/1,218 101/491 46/264 57/353
Fruit (servings/day) <1 1.0–1.9 2.0–2.9 3
HR (95% CI)y Referent 0.83 (0.71, 0.97) 0.87 (0.71, 1.07) 1.02 (0.82, 1.27) 0.776
Cases/men (nos.) 330/1,717 291/1,726 132/732 122/594
* HR, hazard ratio; CI, confidence interval.
y Controlled for age, race/ethnicity, waist/hip ratio, and total energy.
932 Kristal et al.
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

effects moderate both the hormonally regulated prostate
growth and heightened smooth muscle tone that cause BPH.
Planned future analyses, based on assays of serum steroi d
hormone, cytokine, and adipokine concentrations, may pro-
vide insight into whether these mechanisms underlie asso-
ciations of diet with BPH risk.
There are several strengths to this study, including the
longitudinal design, the large sample size, the standardized
and frequent assessments of BPH treatment and symptoms,
and the use of incident rather than prevalent BPH as a study
endpoint. There are also several limitations. First, all infor-
mation used to define BPH endpoi nts was based on self-
report, including medical and surgical treatment and lower
urinary tract symptoms. We used specific interviewer-
administered probes to collect treatment information every
3 months during the duration of the trial, but we did not have
access to medical records to validate responses. We also
used a standardized and well-validated, self-administered
questionnaire to collect symptom data annually, but report-
ing of symptoms is highly subjective and may differ across
racial, ethnic, and socioeconomic groups. Second, we used
data from a FFQ to assess diet. Although we did not analyze
nutrients that are very poorly ascertained by FFQ, we and
many others are concerned by recent studies suggesting that
FFQs correlated poorly with unbiased biomarkers of diet
and perform poorly compared with dietary records (37).
Finally, our assessment of supplement use was incomplete,
and in particular we lacked frequency and dose information
for selenium, vitamin D, and fish oil.
In conclusion, we found evidence that a dietary pattern
high in vegetables and protein, moderate in alcohol, and low
in fat and red meat may protect men from developing symp-
tomatic BPH. We found no evidence that antioxidant nu-
trients, from either supplements or food, were assoc iated
with reduced BPH risk, nor did we find evidence that con-
sumption of long-chain x-3 fatty acids, zinc, or calcium was
associated with reduced risk. Although confirmatory studies
are needed, it is possible that dietary modification could be
useful for preventing BPH and the management of BPH
symptoms.
ACKNOWLEDGMENTS
This work was supported by National Institutes of Health
grants DK63303, CA37429, CA18964, and CA108964.
Conflict of interest: none declared.
REFERENCES
1. Kirby RS. The natural history of benign prostatic hyperplasia:
what have we learned in the last decade? Urology 2000;
56:3–6.
2. Platz EA, Smit E, Curhan GC, et al. Prevalence of and racial/
ethnic variation in lower urinary tract symptoms and non-
cancer prostate surgery in U.S. men. Urology 2002;59:877–83.
3. Kortt MA, Bootman JL. The economics of benign prostatic
hyperplasia treatment: a literature review. Clin Ther 1996;18:
1227–41.
4. Giovannucci E, Rimm EB, Chute CG, et al. Obesity and
benign prostatic hyperplasia. Am J Epidemiol 1994;140:
989–1002.
5. Chokkalingam AP, Pollak M, Filmore CM, et al. Insulin-like
growth factors and prostate cancer: a population-based case
control study in China. Cancer Epidemiol Biomarkers Prev
2001;10:421–8.
6. Dahle SE, Chokkalingam AP, Gao YT, et al. Body size and
serum levels of insulin and leptin in relation to the risk of
benign prostatic hyperplasia. J Urol 2002;168:599–604.
7. Kristal AR, Arnold KB, Schenk JM, et al. Race/ethnicity,
obesity, health related behaviors and the risk of symptomatic
benign prostatic hyperplasia: results from the Prostate Cancer
Prevention Trial. J Urol 2007;177:1395–400.
8. Chyou PH, Nomura AM, Stemmermann GN, et al. A pro-
spective study of alcohol, diet, and other lifestyle factors in
relation to obstructive uropathy. Prostate 1993;22:253–64.
9. Rohrmann S, Giovannucci E, Willett WC, et al. Fruit and
vegetable consumption, intake of micronutrients, and benign
prostatic hyperplasia in US men. Am J Clin Nutr 2007;85:
523–9.
10. Foster CS. Pathology of benign prostatic hyperplasia. Prostate
2000;9:4–14.
11. Troisi RJ, Weiss ST, Parker DR, et al. Relation of obesity and
diet to sympathetic nervous system activity. Hypertension
1991;17:669–77.
12. Eikelis N, Schlaich M, Aggarwal A, et al. Interactions between
leptin and the human sympathetic nervous system. Hyperten-
sion 2003;41:1072–9.
13. Kramer G, Marberger M. Could inflammation be a key com-
ponent in the progression of benign prostatic hyperplasia?
Curr Opin Urol 2006;16:25–9.
14. Thompson IM, Goodman PJ, Tangen CM, et al. The influence
of finasteride on the development of prostate cancer. N Engl J
Med 2003;349:215–24.
15. Satia-Abouta J, Patterson RE, Schiller RN, et al. Energy from
fat is associated with obesity in U.S. men: results from the
Prostate Cancer Prevention Trial. Prev Med 2002;34:493–501.
16. Neuhouser ML, Kristal AR, Patterson RE, et al. Dietary sup-
plement use in the Prostate Cancer Prevention Trial: implica-
tions for prevention trials. Nutr Cancer 2001;39:12–18.
17. Barry MJ, Floyd JF, O’Leary MP, et al. The American
Urological Association symptom index for benign prostatic
hyperplasia. J Urol 1992;148:1549–57.
18. Stryker WS, Kaplan LA, Stein EA, et al. The relation of diet,
cigarette smoking, and alcohol consumption to plasma beta-
carotene and alpha-tocopherol levels. Am J Epidemiol 1988;
127:283–96.
19. Levander OA. The need for measures of selenium status. J Am
Coll Toxicol 1986;5:37–44.
20. Hunter DJ, Morris JS, Chute CG, et al. Predictors of selenium
concentration in human toenails. Am J Epidemiol 1990;132:
114–22.
21. Dixon LB, Subar AF, Weideroff L, et al. Carotenoid and
tocopherol estimates from the NCI Diet History Questionnaire
are valid compared with multiple recalls and serum bio-
markers. J Nutr 2006;136:3054–61.
22. Brunner E, Stallone D, Janeja M, et al. Dietary assessment in
Whitehall II: comparison of 7 d diet diary and food-frequency
questionnaire and validity against biomarkers. Br J Nutr 2001;
86:405–14.
23. Talegawkar SA, Johnson EJ, Carithers T, et al. Total
a-tocopherol intakes are associated with serum a-tocopherol
concentrations in African American adults. J Nutr 2007;137:
2297–303.
Diet, Alcohol, and Risk of Benign Prostatic Hyperplasia 933
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from

24. Araki H, Watanabe H, Mishina T, et al. High-risk group for
benign prostatic hypertrophy. Prostate 1983;4:253–64.
25. Lagiou P, Wuu J, Trichopoulou A, et al. Diet and benign
prostatic hyperplasia: a study in Greece. Urology 1999;54:
284–90.
26. Morrison AS. Risk factors for surgery for prostatic hypertro-
phy. Am J Epidemiol 1992;135:974–80.
27. Signorello LB, Tzonou A, Lagiou P, et al. The epidemiology of
benign prostatic hyperplasia: a study in Greece. BJU Int
1999;84:286–91.
28. Koskima
¨
ki J. Association of dietary elements and lower
urinary tract symptoms. Scand J Urol Nephrol 2000;34:
46–50.
29. Gass R. Benign prostatic hyperplasia: the opposite effects of
alcohol and coffee intake. Br J Urol 2002;90:649–54.
30. Rohrmann S, Smit E, Giovannucci E, et al. Association be-
tween serum concentrations of micronutrients and lower
urinary tract symptoms in older men in the Third National
Health and Nutrition Examination Survey. Urology 2004;64:
504–9.
31. Suzuki S, Platz EA, Kawachi I, et al. Intakes of energy and
macronutrients and the risk of benign prostatic hyperplasia.
Am J Clin Nutr 2002;75:689–97.
32. Platz EA, Rimm EB, Kawachi I, et al. Alcohol consumption,
cigarette smoking, and risk of benign prostatic hyperplasia.
Am J Epidemiol 1999;149:106–15.
33. Kang D, Andriole GL, van de Vooren RC, et al. Risk behav-
iours and benign prostatic hyperplasia. BJU Int 2004;93:1241–5.
34. Gordon GG, Altman K, Southren AL, et al. Effect of alcohol
(ethanol) administration on sex-hormone metabolism in nor-
mal men. N Engl J Med 1976;295:793–7.
35. Dorgan JF, Judd JT, Longcope C, et al. Effects of dietary fat
and fiber on plasma and urine androgens and estrogens in men:
a controlled feeding. Am J Clin Nutr 1996;64:850–5.
36. Landsberg L. Feast or famine: the sympathetic nervous system
response to nutrient intake. Cell Mol Neurobiol 2006;26:
495–506.
37. Kristal AR, Peters U, Potter JD. Is it time to abandon the food
frequency questionnaire? Cancer Epidemiol Biomarkers Prev
2005;14:2826–8.
934 Kristal et al.
Am J Epidemiol 2008;167:925–934
by guest on June 2, 2013http://aje.oxfordjournals.org/Downloaded from