Cancer Causes Control (2011) 22:3339

DOI 10.1007/s10552-010-9665-8

ORIGINAL PAPER

Soft drinks, sweetened beverages and risk of pancreatic cancer

Silvano Gallus Federica Turati Alessandra Tavani
Jerry Polesel Renato Talamini Silvia Franceschi
Carlo La Vecchia

Received: 5 August 2010 / Accepted: 7 October 2010 / Published online: 28 October 2010
Springer Science+Business Media B.V. 2010

Abstract Soft drinks usually contain sugar and caffeine

that might influence pancreatic carcinogenesis. We considered the association between carbonated drink consumption
and pancreatic cancer risk in an Italian casecontrol study
conducted in 19912008 on 326 pancreatic cancer cases and
652 matched controls. We also combined the results from all
the studies on soft drinks or sweetened beverages and pancreatic cancer published before June 2010, using a metaanalytic approach. In the casecontrol study, compared with
non-drinkers, the multivariate odds ratio was 1.02 (95%
confidence interval, CI, 0.721.44) for carbonated drink
consumers and 0.89 (95% CI 0.531.50) for regular consumers (at least one drink/day). Besides our study, from the
literature search, we identified 4 other casecontrol (1,919

cases) and 6 cohort studies (2,367 cases). The pooled relative

risks (RR) for soft drink consumers vs. non-consumers were
0.97 (95% CI 0.811.16) for casecontrol, 1.05 (95% CI
0.941.17) for cohort, and 1.02 (95% CI 0.931.12) for all
studies. The pooled RRs for heavy drinkers were 1.08 (95%
CI 0.731.60) for casecontrol, 1.21 (95% CI 0.901.63) for
cohort, and 1.16 (95% CI 0.931.45) for all studies. In conclusion, soft drink consumption is not materially related to
pancreatic cancer risk.
Keywords Soft drink
Pancreatic cancer
Casecontrol
study
Meta-analysis
Risk factor

Introduction
S. Gallus
F. Turati
A. Tavani
C. La Vecchia
Dipartimento di Epidemiologia, Istituto di Ricerche
Farmacologiche Mario Negri, 20156 Milan, Italy
F. Turati
C. La Vecchia
Dipartimento di Medicina del Lavoro Clinica del Lavoro Luigi
Devoto, Sezione di Statistica Medica e Biometria Giulio
A. Maccacaro, Universita` degli Studi di Milano,
20133 Milan, Italy
J. Polesel
R. Talamini
Unita` di Epidemiologia e Biostatistica, IRCCS Centro
di Riferimento Oncologico, 33081 Aviano, PN, Italy
S. Franceschi
Infections and Cancer Epidemiology Group, International
Agency for Cancer Research, 69372 Lyon CEDEX 08, France
S. Gallus (&)
Unit of Epidemiology for Clinical Research, Department
of Epidemiology, Istituto di Ricerche Farmacologiche
Mario Negri, Via La Masa, 19, 20156 Milan, Italy
e-mail: silvano.gallus@marionegri.it

Soft drinks usually contain sugar and may contain caffeine.

Sugars play a role in the development of obesity and
diabetes and have been shown to influence the secretion
of insulin and insulin-related growth factors [1]. Insulin
acts as a growth promoter and mitogen in the pancreas [2],
and high insulin concentrations and abnormal glucose
metabolism have been involved in the aetiology of
pancreatic cancer [35]. The association between sugars or
carbohydrates and pancreatic cancer risk has been investigated in numerous studies, with most investigations
reporting no association [6]. The few studies reporting
positive associations with intake of fructose, glucose and/or
sucrose [711], however, found little association with
added sugar and soft drinks.
Caffeine is a component of several soft drinks. In the
early 1980s, coffee (a major source of caffeine) was related
to pancreatic cancer risk in a hospital-based casecontrol
study, where the comparison group included patients with
chronic diseases, including other cancers [12]. Subsequent

123

34

studies did not consistently confirm the relation [13], thus

the issue remains open to discussion [14].
The relation between soft drink consumption and pancreatic cancer risk has been considered in several epidemiological studies [15], mostly finding a lack of association
[8, 11, 1619]. However, recently a cohort study from
Singapore found an about twofold increased risk of pancreatic cancer in drinkers of C2 soft drinks/week compared
with non-drinkers [15].
We investigated the role of soft drinks on pancreatic
carcinogenesis using an Italian casecontrol study, and
quantitatively combined in a systematic meta-analysis all
published data on soft drinks and pancreatic cancer risk
from casecontrol and prospective cohort studies.

Cancer Causes Control (2011) 22:3339

(categorical, \50/5059/6069/ C 60 years), and adjusted

for year of interview (categorical, \1998/ C 1998), education (categorical, \7/711/ C 12 years), body mass index
(BMI; categorical, \ 20/2024.9/2529.9/ C 30 kg/m2),
tobacco smoking (categorical, never smokers/ex-smokers/
smokers of \15 cigarettes per day/smokers of C15
cigarettes per day), alcohol drinking (categorical, 0/ \8/
814/ C 15 drinks per week), total energy intake (categorical, \1993.07/1993.072611.27/ C 2611.28 kcal per
day), family history of pancreatic cancer (categorical,
yes/no), and diabetes (categorical, yes/no).
To test for heterogeneity between strata, the difference
between the -2*log(likelihood) of the models with and
without the interaction terms was compared with the v2
distribution with the same number of degrees of freedom as
the number of the interaction terms.

Materials and methods

Meta-analysis
Italian casecontrol study
Between 1991 and 2008, we conducted in northern Italy a
multicentre casecontrol study on pancreatic cancer [20].
Cases were 326 patients (174 men and 152 women, median
age 63 years) with incident histologically confirmed pancreatic cancer, admitted to major teaching and general
hospitals in the province of Pordenone and in the greater
Milan area. Controls were 652 patients (frequency matched
1:2 by study centre, sex, and age) admitted to the same
hospitals of cases for acute, non-neoplastic conditions,
unrelated to known risk factors for pancreatic cancer or
long-term diet modifications. They were hospitalized for
traumatic (31%) or other orthopaedic disorders (31%),
acute surgical conditions (28%), and miscellaneous other
illness (10%). Participation rate was over 95% for both
cases and controls. Centrally trained interviewers used a
structured questionnaire to collect information on sociodemographic and anthropometric characteristics, selected
lifestyle habits, physical activity, personal medical history,
and family history of cancer. Usual diet during the 2 years
prior to the diagnosis was assessed through a validated
food-frequency questionnaire (FFQ) with satisfactory
reproducibility [21] and validity [22]. A question asked to
report the average weekly frequency of consumption of
carbonated drinks (1 drink = 150 ml). No specification of
type of carbonated drink (e.g. regular or diet) was available. Intakes lower than once a week, but at least once a
month, were defined as occasional consumptions and coded
as 0.5 per week.
Odds ratios (OR) and the corresponding 95% confidence intervals (CI) of pancreatic cancer were estimated
using multiple logistic regression models conditioned on
study centre (categorical, Milan/Pordenone), sex and age

123

We performed a Medline search of the literature from

1966 to June 2010, using the string (pancreas OR
pancreatic) AND cancer AND risk AND (soda OR sugar
beverage OR soft drinks OR sweetened OR carbonated), following the Meta-analysis Of Observational
Studies in Epidemiology (MOOSE) guidelines [23]. We
also checked the reference list of the publications
retrieved. We included in the meta-analyses all the studies
providing data on soft drinks, sweetened beverages, and
pancreatic cancer risk. For studies providing data for
various different types of sweetened beverages, we considered total intake for the most frequent one. No study
was excluded a priori for weakness of design or data
quality. A study providing data on diet soda only was
excluded [24].
We considered two analyses: drinkers vs. nondrinkers and heavy drinkers vs. non-drinkers. Nondrinkers (the reference category for both the analyses)
were defined as those with the lowest soft drink consumption
in each study (non-drinkers for most studies). For each study,
heavy drinkers were defined as those in the highest category
of soft drink consumption, in the studies providing RR estimates for at least two categories of soft drink consumption.
In studies reporting risk estimates for more than one
category of soft drink consumption only, we used the
fixed-effects model to collapse the estimates of the exposure
categories, obtaining the study-specific risk estimate for
drinkers vs. non-drinkers.
Summary measures were computed using randomeffects models that consider both within-study and betweenstudy variations. Statistical heterogeneity among studies
was assessed using the v2 tests (results were defined as
heterogeneous for p \ 0.10).

Cancer Causes Control (2011) 22:3339

35

Results
In the present dataset, compared with non-drinkers, the
multivariate OR for carbonated drink consumers was 1.02
(95% CI 0.721.44) (Table 1); it was 1.09 (95% CI
0.721.65) for \ 1 drink/day (59 cases and 118 controls)
and 0.89 (95% CI 0.531.50) for C1 drink/day (30 cases
and 77 controls). These estimates did not materially change
after further allowance for physical activity. The OR for
drinkers vs. non-drinkers was 0.60 among subjects aged
\60 years and 1.35 among those aged C60 years, with a
significant heterogeneity between strata (p = 0.008). No
significant heterogeneity was found among strata of sex,
BMI, and energy intake. Excluding diabetics, the OR for
carbonated drink consumers was 0.93 (95% CI, 0.641.35).
From the MEDLINE search, we identified 4 other case
control studies [8, 16, 17, 25], for a total of 1,919 cases
including those from the present study, and 6 cohort studies
[7, 11, 15, 18, 19, 26], for a total of 2,367 cases. Study
characteristics, including details on soft drink definition
and categories of exposure, are reported in Table 2. Using
a meta-analytic approach, the pooled RRs for soft drink
consumers vs. non-drinkers were 0.97 (95% CI 0.811.16,
p for heterogeneity = 0.209) for casecontrol, 1.05 (95%
CI 0.941.17, p for heterogeneity = 0.173) for cohort, and
1.02 (95% CI 0.931.12, p for heterogeneity = 0.143) for

all studies (Fig. 1). The corresponding RRs for heavy

drinkers (defined as those in the highest category of soft
drink consumption in each study) vs. non-drinkers were
1.08 (95% CI 0.731.60, based on 3 studies and 210
exposed cases) for casecontrol, 1.21 (95% CI 0.901.63,
based on 5 studies and 325 exposed cases) for cohort, and
1.16 (95% CI 0.93-1.45, p for heterogeneity = 0.010) for
all studies combined.

Discussion
Our casecontrol study shows a lack of association
between carbonated drink consumption and pancreatic
cancer risk, even for regular drinkers. These results were
consistent in strata of sex, BMI, and total energy intake.
Younger subjects, more likely those with higher consumption, showed a lower relative risk.
Most carbonated drinks provide considerable amounts
(up to 3040 g per drink) of sugars and some of them
caffeine. Thus, the present results support the lack of
association of pancreatic cancer risk with sugar and
caffeine intake.
With reference to potential selection bias, we identified
cases and controls in the major teaching and general hospitals and excluded from the comparison group any patient

Table 1 Distribution of 326 pancreatic cancer cases and 652 controls, and corresponding odds ratios (OR) and 95% confidence intervals (CI),
according to carbonated drink consumption, overall and in strata of selected covariates
Carbonated drink consumptiona

p for heterogeneity

No

Yes

Cases/controls

Cases/controls

OR (95% CI)b

236/457

89/195

1.02 (0.721.44)

Males

123/247

51/101

0.85 (0.501.46)

Females

113/210

38/94

1.14 (0.711.84)

\60 years

88/153

33/89

0.60 (0.331.08)

C60 years

148/304

56/106

1.35 (0.862.12)

\25 kg/m2

108/199

32/68

1.02 (0.561.84)

C25 kg/m2

128/258

57/127

0.99 (0.621.56)

\2245.91 kcal/day

110/251

26/75

0.93 (0.511.68)

C2245.91 kcal/day

126/206

63/120

1.11 (0.691.78)

Total
Sex

0.405

Age
0.008

Body mass index

0.890

Total energy intakec

0.677

Italy, 19912008
a

The sum does not add up to the total because of one missing value among cases

Estimated from logistic regression models, conditioned on study centre, sex and age, and adjusted for year of interview, education, body mass
index, tobacco smoking, alcohol drinking, total energy intake, family history of pancreatic cancer, and history of diabetes. No consumption is the
reference category

Strata are based on the median of total energy intake computed among all controls

123

123

Canada

USA

Italy

Anderson et al. [17]

Chan et al. [8]

Gallus et al. (present

work)

Multiethnic cohort
study

Cohort of Swedish Men

Nothlings et al. [11]
USA

Swedish mammography cohort;

Sweden

8 years

19931996

7.2 years (mean)

19972005

19862000
(HPFS)

Health professionals
follow-up study
(HPFS)

20 years
19802000
(NHS)

Larsson et al. [7]

326

532

422

149

490

434

131

379

162,150 PR

563,430 PY

77,797 PR

C1 drink/day vs.
\1 drink/month

Ever vs. never

C1 vs. \1 drink/
day

Soft drinking,
drinkers vs nondrinkers

Carbonated drink/juice

Carbonated beverages

Non-diet soda

Soft drinks (carbonated

soft drinks and noncarbonated sweetened
drinks)

Drinkers vs. nondrinkers

Drinkers vs. nondrinkers

C1 vs. \1 drink/
month

[1 drink/week vs.
none or
occasional

Drinkers vs. nondrinkers

All sweetened beverages Drinkers vs. nondrinkers

Caffeinated soft drinks

Caffeinated soft drinks

Carbonated beverages

Sugar-sweetened soft
2,240,547 PY drinks

138,158 PR

3,158 PR

652 hc

1,701 pc

312 pc

363 pc

490 pc

No. of No. and type Soft drink definition

cases of controls/
size of cohort

19842002
25a
13.8 years for
men, 14.8 for
women (mean)

19912008

19951999

20032007

19841987

19761981

Years of study/
duration of
follow-up

Nurses health study

(NHS);

Schernhammer et al. [18] USA

Khan et al. [26]

Japan

USA

Lyon et al. [16]

Cohort studies

USA

Country

Mack et al. [25]

Case-control studies

First author

Table 2 Casecontrol and cohort studies on the association between soft drink consumption and pancreatic cancer risk

Women: age, health status, health

education, health screening and
smoking

Men: age, smoking

Age, sex, centre, year of interview,

education, BMI, smoking, alcohol,
energy intake, diabetes, family history
of pancreatic cancer

Age, sex, race, education, BMI,

smoking, physical activity, energy
intake, diabetes, red meat, white meat,
vegetable and fruit, eggs, fish, dairy,
whole grain, refined grain, sweets,
sugar-type carbonated beverages,
sugar-free carbonated beverages

Age, sex, education, BMI, smoking,

family history of pancreatic cancer,
fruit, alcohol, allergies

Age, sex, residence

Age, sex, race, neighbourhood

Adjustment factors

C75.7 g*1000/
(Kcal*day) vs.
none

C2 drinks/day vs.
none

Age, sex, time on study, race-ethnicity,

smoking, energy intake, red meat and
processed meat, BMI, family history
of pancreatic cancer

Age, sex, education, smoking, BMI,

alcohol, energy intake

[3 drinks/week vs. Age, sex, follow-up cycles, smoking,

\1 drink/month
non-vigorous physical activity, energy
intake, diet soft drinks, diabetes

C1 drink/day vs.
none

C1 drinks/day vs.
none

C5 vs. \1 drink/
day

Soft drinking,
heavy drinkers vs
non-drinkers

36
Cancer Causes Control (2011) 22:3339

National Institutes of HealthAmerican Association of Retired Persons Diet and Health Study

Deaths for pancreatic cancer

Hc, hospital controls; pc, population controls; BMI, body mass index; PR, person at risk; PY, person years

The Singapore
Chinese Health Study

Mueller et al. [15]

China

14 years

140

60,524 PR

Soft drinks

Drinkers vs. nondrinkers

C2 drinks/week vs. Age, sex, ethnicity, year of interview,

none
education, smoking, physical activity,
alcohol, added sugar and candy, juice,
energy intake

37

648,387 PY

Age, sex, race, education, BMI, smoking,

alcohol, physical activity, energyadjusted red meat, energy-adjusted
folate, total energy intake, diet
beverages
[234.3 grams/day
vs. none
3,521,088 PY
7.2 years (mean)
NIH-AARP Diet and
Healt Studyb

Bao et al. [19]

USA

1995/96-2003

1,258 487,922 PR

Sugar-sweetened
beverages (soft drinks
and regular fruit
drinks)

Drinkers vs. nondrinkers

Adjustment factors
First author

Table 2 continued

Country

Years of study/
duration of
follow-up

No. of No. and type Soft drink definition

cases of controls/
size of cohort

Soft drinking,
drinkers vs nondrinkers

Soft drinking,
heavy drinkers vs
non-drinkers

Cancer Causes Control (2011) 22:3339

admitted to hospital for any chronic disease and for conditions related to alcohol drinking, tobacco smoking, or
major modifications in diet. The strengths of our study
include the use of a validated [22] and satisfactorily
reproducible FFQ (the correlation coefficient for the
reproducibility of information on soft drinks was 0.61)
[21], the almost complete participation rate ([ 95%), the
same interview setting for cases and controls, and the
availability of a large number of covariates for multivariate
analyses, including a measure of total energy intake.
Our risk estimates are consistent with the overall evidence
from published studies, assessed in this study using a metaanalytic approach. Single studies were conducted in various
countries with different population baseline characteristics,
and estimates were derived allowing for different adjustment
factors. Moreover, the original articles considered different
types of soft drinks, such as carbonated beverages, caffeinated soft drinks, sugar-sweetened beverages, and non-diet
soda, sometimes including low-calorie sweetener drinks,
non-carbonated sweetened drinks, and juices. Various types
of soft drinks have a different composition in terms of sugars,
other nutrients and hence calories, vitamins, minerals, food
components, and other micro-nutrients, which may influence
pancreatic cancer risk [27]. Available data, however, are
inadequate to distinguish between various types of soft
drinks. Furthermore, the measures considered were different
(glass = about 150200 ml, or can = 330 ml). However,
no heterogeneity among studies was found for drinkers vs.
non-drinkers. Moreover, in studies [8, 24] where soft drinks
were grouped with different criteria (non-diet/diet soft
drinks, carbonated/non-carbonated soft drinks), the results
were similar [8], and no increased risk was found also for
sugar-free soft drinks [8, 24]. The significant heterogeneity
found when considering heavy vs. non-drinkers may be due
to the vast variety of categorization of heavy drinkers,
ranging from C2 drinks per week [15] to C5 drinks per day
[25], and to other correlates to heavy consumption.
Publication bias is also possible, with selective reporting
of significant findings. We did not search for unpublished
data or abstracts, given the difficulties in their interpretation. However, no significant asymmetry was present in the
funnel plot, and the Beggs and Eggers tests indicated that
publication bias is unlikely to have materially influenced
our results [28]. Moreover, the ORs of the largest case
control study [8] were 0.81 (95% CI 0.660.99) for
drinkers vs. non/occasional drinkers and 1.00 (95% CI
0.731.36) for the highest category of consumption (C1
drink/day, based on 166 cases) compared with non/occasional drinkers (111 cases). The largest cohort study [19]
found a RR of 0.95 (95% CI 0.851.04) for drinkers vs.
non/occasional drinkers and a RR of 0.83 (0.671.03) for
the highest quintile of consumption (151 cases) compared
with the lowest one (non/occasional drinkers, 574 cases).

123

38

Cancer Causes Control (2011) 22:3339

Study

Cancer Cases

RR

95% CI

drinkers/non drinkers
Casecontrol studies
Mack et al, 1986

70/420

0.90

0.651.25

Lyon et al, 1992

85/58

1.31

0.891.93

*/*

1.24

0.662.33

Chan et al, 2009

415/111

0.81

0.660.99

Gallus et al, 2010^

89/236

1.02

0.721.44

0.97

0.811.16

Anderson et al, 2009

All casecontrol studies

Cohort studies
Khan et al, 2004

*/*

0.20

0.021.82

174/205

1.07

0.891.29

65/66

1.26

0.941.69

Nothlings et al, 2007

266/168

1.08

0.921.26

Bao et al., 2008

684/574

0.94

0.851.04

Mueller et al, 2010

30/110

0.841.84

All cohort studies

1.24
1.05

All studies

1.02

0.931.12

Schernhammer et al, 2005

Larsson et al, 2006

0.941.17

* not available
^ present work

0.5

Relative risk

Fig. 1 Summary relative risks (RRs) of pancreatic cancer for drinkers versus non-drinkers, from casecontrol, cohort studies and overall

In conclusion, our casecontrol study and a meta-analysis

of all published studies support the absence of association
between soft drink consumption and pancreatic cancer risk.
Acknowledgments This work was supported by the Italian Association for Cancer Research (AIRC), Milan, Italy.
Conflict of interest
conflicts of interest.

The authors declare they do not have any

References
1. Gnagnarella P, Gandini S, La Vecchia C, Maisonneuve P (2008)
Glycemic index, glycemic load, and cancer risk: a meta-analysis.
Am J Clin Nutr 87:17931801
2. Fisher WE, Boros LG, Schirmer WJ (1996) Insulin promotes
pancreatic cancer: evidence for endocrine influence on exocrine
pancreatic tumors. J Surg Res 63:310313
3. Batty GD, Shipley MJ, Marmot M, Smith GD (2004) Diabetes
status and post-load plasma glucose concentration in relation to
site-specific cancer mortality: findings from the original Whitehall study. Cancer Causes Control 15:873881
4. Huxley R, Ansary-Moghaddam A, Berrington de Gonzalez A,
Barzi F, Woodward M (2005) Type-II diabetes and pancreatic
cancer: a meta-analysis of 36 studies. Br J Cancer 92:20762083

123

5. Rossi M, Lipworth L, Polesel J et al (2010) Dietary glycemic

index and glycemic load and risk of pancreatic cancer: a casecontrol study. Ann Epidemiol 20:460465
6. Patel AV, McCullough ML, Pavluck AL, Jacobs EJ, Thun MJ,
Calle EE (2007) Glycemic load, glycemic index, and carbohydrate intake in relation to pancreatic cancer risk in a large US
cohort. Cancer Causes Control 18:287294
7. Larsson SC, Bergkvist L, Wolk A (2006) Consumption of sugar
and sugar-sweetened foods and the risk of pancreatic cancer in a
prospective study. Am J Clin Nutr 84:11711176
8. Chan JM, Wang F, Holly EA (2009) Sweets, sweetened beverages, and risk of pancreatic cancer in a large population-based
case-control study. Cancer Causes Control 20:835846
9. Michaud DS, Liu S, Giovannucci E, Willett WC, Colditz GA,
Fuchs CS (2002) Dietary sugar, glycemic load, and pancreatic
cancer risk in a prospective study. J Natl Cancer Inst 94:
12931300
10. Jiao L, Flood A, Subar AF, Hollenbeck AR, Schatzkin A,
Stolzenberg-Solomon R (2009) Glycemic index, carbohydrates,
glycemic load, and the risk of pancreatic cancer in a prospective
cohort study. Cancer Epidemiol Biomarkers Prev 18:11441151
11. Nothlings U, Murphy SP, Wilkens LR, Henderson BE, Kolonel
LN (2007) Dietary glycemic load, added sugars, and carbohydrates as risk factors for pancreatic cancer: the multiethnic cohort
study. Am J Clin Nutr 86:14951501
12. MacMahon B, Yen S, Trichopoulos D, Warren K, Nardi G (1981)
Coffee and cancer of the pancreas. N Engl J Med 304:630633
13. La Vecchia C, Tavani A (2007) Coffee and cancer risk: an
update. Eur J Cancer Prev 16:385389

Cancer Causes Control (2011) 22:3339

14. Nilsson LM, Johansson I, Lenner P, Lindahl B, Van Guelpen B
(2010) Consumption of filtered and boiled coffee and the risk of
incident cancer: a prospective cohort study. Cancer Causes
Control 21:15331544
15. Mueller NT, Odegaard A, Anderson K et al (2010) Soft drink and
juice consumption and risk of pancreatic cancer: the singapore
chinese health study. Cancer Epidemiol Biomarkers Prev
19:447455
16. Lyon JL, Mahoney AW, French TK, Moser R Jr (1992) Coffee
consumption and the risk of cancer of the exocrine pancreas: a
case-control study in a low-risk population. Epidemiology
3:164170
17. Anderson LN, Cotterchio M, Gallinger S (2009) Lifestyle, dietary, and medical history factors associated with pancreatic
cancer risk in Ontario, Canada. Cancer Causes Control 20:
825834
18. Schernhammer ES, Hu FB, Giovannucci E et al (2005) Sugarsweetened soft drink consumption and risk of pancreatic cancer in
two prospective cohorts. Cancer Epidemiol Biomarkers Prev
14:20982105
19. Bao Y, Stolzenberg-Solomon R, Jiao L et al (2008) Added sugar
and sugar-sweetened foods and beverages and the risk of pancreatic cancer in the National Institutes of Health-AARP Diet and
Health study. Am J Clin Nutr 88:431440
20. Polesel J, Talamini R, Negri E et al (2009) Dietary habits and risk
of pancreatic cancer: an Italian case-control study. Cancer Causes
Control 21:493500

39
21. Franceschi S, Negri E, Salvini S et al (1993) Reproducibility of
an Italian food frequency questionnaire for cancer studies: results
for specific food items. Eur J Cancer 29A:22982305
22. Decarli A, Franceschi S, Ferraroni M et al (1996) Validation of a
food-frequency questionnaire to assess dietary intakes in cancer
studies in Italy. Results for specific nutrients. Ann Epidemiol
6:110118
23. Stroup DF, Berlin JA, Morton SC et al (2000) Meta-analysis of
observational studies in epidemiology: a proposal for reporting.
Meta-analysis of observational studies in epidemiology
(MOOSE) group. JAMA 283:20082012
24. Gold EB, Gordis L, Diener MD et al (1985) Diet and other risk
factors for cancer of the pancreas. Cancer 55:460467
25. Mack TM, Yu MC, Hanisch R, Henderson BE (1986) Pancreas
cancer and smoking, beverage consumption, and past medical
history. J Natl Cancer Inst 76:4960
26. Khan MM, Goto R, Kobayashi K et al (2004) Dietary habits and
cancer mortality among middle aged and older Japanese living in
Hokkaido, Japan by cancer site and sex. Asian Pac J Cancer Prev
5:5865
27. World Cancer Research Fund and American Institute for Cancer
Research. Food, Nutrition, Physical Activity and the Prevention
of Cancer: a Global Perspective. Washington, DC: AICR; 2007
28. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in
meta-analysis detected by a simple, graphical test. BMJ 315:
629634

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.