SEARCH

SEARCH BY CITATION

Keywords:

  • breast cancer;
  • colorectal cancer;
  • endometrial cancer;
  • prostate cancer;
  • kidney cancer;
  • gallbladder cancer;
  • body mass index;
  • risk factors

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

There is growing evidence that excess body weight increases the risk of cancer at several sites, including kidney, endometrium, colon, prostate, gallbladder and breast in post-menopausal women. The proportion of all cancers attributable to overweight has, however, never been systematically estimated. We reviewed the epidemiological literature and quantitatively summarised, by meta-analysis, the relationship between excess weight and the risk of developing cancer at the 6 sites listed above. Estimates were then combined with sex-specific estimates of the prevalence of overweight [body mass index (BMI) 25–29 kg/m2] and obesity (BMI ≥30 kg/m2) in each country in the European Union to obtain the proportion of cancers attributable to excess weight. Overall, excess body mass accounts for 5% of all cancers in the European Union, 3% in men and 6% in women, corresponding to 27,000 male and 45,000 female cancer cases yearly. The attributable proportion varied, in men, between 2.1% for Greece and 4.9% for Germany and, in women, between 3.9% for Denmark and 8.8% for Spain. The highest attributable proportions were obtained for cancers of the endometrium (39%), kidney (25% in both sexes) and gallbladder (25% in men and 24% in women). The largest number of attributable cases was for colon cancer (21,500 annual cases), followed by endometrium (14,000 cases) and breast (12,800 cases). Some 36,000 cases could be avoided by halving the prevalence of overweight and obese people in Europe. © 2001 Wiley-Liss, Inc.

Since a large prospective study1 showed the first evidence that overweight and obesity increase the risk of mortality from certain cancers, numerous other studies linking excess weight to cancer at several sites have been reported. The evidence for an association is now strong for cancers of the endometrium,2 kidney3 and the post-menopausal breast,4 as well as colon,5 gallbladder6 and prostate.7 Cancers of the thyroid8, 9 and gastric cardia10–12 as well as adenocarcinoma of the esophagus13 have been associated with overweight and obesity.

Notwithstanding the growing evidence that overweight is causally related to cancer of several sites, there has been no systematic attempt to estimate the total proportion of all cancers attributable to this potentially avoidable but increasingly more common cause. We have reviewed the epidemiological literature and quantitatively summarised the relationship between excess weight and the risk of developing cancer. We performed a meta-analysis for the 6 cancer sites for which the evidence is most consistent and estimated the proportion of cancer cases attributable to overweight or obesity that occurred in the countries of the European Union.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Search methods

We identified studies published between 1966 and 1997 (we began data abstraction in 1998) through the MEDLINE (National Library of Medicine, Washington, D.C.) database. Since heterogeneity of study design, methods of exposure assessment and data analysis are common causes of inconsistency between observational studies, we chose, a priori, desirable properties for design, study size and period of publication and, whenever possible, included only studies with these characteristics. Throughout, we refer to studies having these characteristics as “eligible”.

Besides a MEDLINE search, we also systematically examined the list of references in the identified articles. The literature review and meta-analysis were confined to cancers of the breast, colon, endometrium, prostate, kidney and gallbladder. We included prospective studies with at least 100 observed cases and population-based case-control studies with at least 200 cases. These restrictions were feasible for cancers of the breast, colon, prostate and endometrium. However, most studies of kidney cancer were relatively small, particularly those on women; therefore, we included cohort studies with at least 50 and case-control studies with at least 100 female cases. Due to the limited number of epidemiological studies on gallbladder cancer, we included all published data. When several articles were published from the same study, we used the most recent report or the one providing the most detailed information.

Design of meta-analysis

For each of the cancer sites considered, we performed a co-variance analysis of the log relative risk on body mass, with terms for study (categories), body mass index (BMI, continuous) and their interaction. We checked the assumption of linearity of the relationship by testing the statistical significance of adding a squared term of BMI. In only one case was this close to statistical significance (colon cancer), and this case is discussed in the relevant section of Results.

Once we had accepted the log-linear dose-response, we estimated a relationship for each of the selected studies according to the co-variance–adjusted approach.14 Individual slopes were then combined by weighted average, using the inverse of their variances as weights.15 Fixed- (FE) and random- (RE) effects 95% confidence intervals (CIs) were calculated for the slope of the common regression slope. The dose-response curve thus obtained provided estimates of the relative risk (RR) associated with standard definitions of normal (reference category), overweight and obese people. These were then combined with country-specific estimates of the population prevalence to obtain the proportion of site-specific cases attributable to excess weight, according to Cole and MacMahon's16 classical formula. In addition to RR and its variance, estimation of the pooled slope required the distribution of cases and controls/person-years in the exposure categories.

Data abstraction and coding

As a measure of excess weight, we used BMI, calculated as weight divided by height squared (kg/m2). This index, recommended in population studies,17–19 estimates the level of fatness independent of height in a general adult population. For each of the associations considered, we listed the major potential confounders. We then abstracted and coded the following information: study design, country, measure of fatness adopted, intervals of body mass and corresponding adjusted RRs (rate ratios, odd ratios, standardised morbidity/mortality ratios, hereinafter denoted as RR), their 95% CIs, number of cases, number of controls/person-years and confounders considered. The variance of the ln RRs was derived from the 95% CIs by the following formula:

  • equation image

Since body mass is measured on a continuous scale, exposure intervals vary from study to study depending on the actual range observed. We summarised closed intervals with their mid-point and open intervals with +10% of their cut-point.15, 20 We used in the quantitative analyses all study results expressing weight as BMI and including all of the following items: extremes of the exposure intervals adopted, corresponding RRs and their variances or 95% CIs. Of these, only those also providing the distribution of cases and controls/person-years in the exposure categories could be used to obtain the co-variance–adjusted estimates of the slopes.

For each cancer site, 3 meta-analyses were performed. The first analysis (model A) included all eligible studies suitable for meta-analysis, the second (model B) was restricted to studies with incident cases as outcome and the third (model C) was further restricted to studies accounting for major confounders. In most cases, use of model C entailed a major loss of information. Model B was therefore chosen to predict the RRs used in calculating the attributable risk when the results from model C were similar.

Attributable fractions of cancer were estimated for overweight (25 < BMI < 30 kg/m,2 BMI = 27) and obese (BMI ≥ 30 kg/m,2 BMI = 32) persons compared to those of normal weight (BMI 20–25 kg/m,2 BMI = 22) according to the WHO guidelines for a healthy physical status.21, 22 These were applied to the number of incident cases, by country and sex, estimated for 1995 by the International Agency for Research on Cancer.23 The incidence of gallbladder cancer, not available from this source, was obtained from Cancer Incidence in Five Continents24 as national figures for Sweden, Norway, Finland, Denmark and the Netherlands or as age- and sex-specific mean rates in the case of country registries. Rates for Portugal and Greece, where no cancer registry is active, were obtained by applying the average European mortality/incidence ratios of gallbladder cancer based on Cancer Incidence in Five Continents to national mortality from the WHO mortality data bank.25

Prevalence of obesity in Europe

Since data on the prevalence of overweight and obesity among men and women at the country level could not always be identified, the most informative estimates were obtained from the WHO MONICA26 and CINDI27 studies. These sources also have the advantage of complying with a common protocol of data collection, thus improving the comparability of the data for the different countries. In the MONICA project, cross-sectional data were collected between 1983 and 1986. Trained personnel measured height and weight in research centres. Each centre studied between 1,046 and 3,563 men and women, aged between 35 and 64 years. The MONICA data are population-based but not necessarily representative of the whole country. Data were available by sex but not in subgroups of the study age group (34–64 years). The CINDI programme collected data between 1982 and 1987 on behavioural, anthropometric and socio-economic characteristics of 24 countries including Austria, Northern Ireland and Portugal, with the aim of improving community health. Height and weight were measured in 1,287 to 2,361 men and women aged 25 to 64 years. Neither of the 2 projects included centres in the Netherlands, for which we used data obtained from a random sample of 5,000 men and women aged 20–59 years, collected each year from 1993 to 1996 (data not shown). For Greece, the only published data28, 29 were highly selected by age, an important determinant of excess weight,30 or social class. We therefore adopted the mean of the values in the other southern European countries, i.e., Portugal, Spain and Italy.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Prevalence of overweight and obesity

Estimates of the prevalence of overweight and obese subpopulations by sex and country are shown in Table I. In both men and women, excess weight (BMI >25 kg/m2) was slightly more common in southern Europe than in northern countries. The mean prevalence was 61% for men and 52% for women in southern Europe compared with 59% for men and 47% for women in the north. Greater geographic variation was observed in women (35% to 68%) than in men (55% to 70%). The prevalence of obesity (BMI ≥30 kg/m2) varied between 9% (Spain) and 18% (Finland) in men, while that among women varied between 10% (Denmark) and 24% (Spain). With few exceptions (Denmark and the Netherlands), obesity was more prevalent in women, while overweight was more prevalent in men.

Table I. Prevalence of Overweight (BMI 25.1–30 kg/m2) and Obese (BMI ≥ 30 kg/m2) Men and Women in the European Union, by Country
CountryMenWomen
OverweightObeseOverweightObese
  • 1

    WHO CINDI study.27 Men and women 20–59 years old in 1982–1987.

  • 2

    WHO MONICA study.26 Men and women 35–64 years old in 1983–1986.

  • 3

    Mean of Italy, Portugal and Spain.

  • 4

    Seidell, personal communication (1998). Men and women 20–59 years 1993–1996.

Austria148122917
Belgium249153620
Denmark244112510
Finland250183820
France249123017
Germany253173520
Greece350134022
Ireland147113217
Italy248153621
Luxembourg245143318
The Netherlands445113111
Portugal145143921
Spain, Catalonia25794424
Sweden245102912
United Kingdom246113615
European Union50133519

Breast cancer

Pre-menopausal women.

Seventeen studies on pre-menopausal breast cancer were eligible for our review.31–47

Only 9 studies could be used in the meta-analysis. They are summarised in Table IIA. The summary analysis showed that obesity was inversely associated with pre-menopausal breast cancer, with a RR of 0.98 per unit of increase in BMI (95% CI 0.97–0.99) (Table IIA). Limiting the analysis to studies with only incident cases, which meant omitting only 1 study,39 did not change the point estimate. A further restriction to studies adjusted for age, reproductive factors and alcohol or diet42, 45, 46 only changed the confidence limits.

Table II. A Summary of Selected Studies of Obesity and Premenopausal Breast Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryConfounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; RF, reproductive factors; ADP, alcohol, diet, physical exercise.

36CCPI306CanadaA0.980.94–1.02
37CCPI634DenmarkA1.010.98–1.04
39CCPI, P751USAA, RF1.000.97–1.03
41CCPI420Sweden, NorwayA, RF1.010.96–1.06
42CCPI270CanadaA, ADP, RF0.980.94–1.02
32CohI137NorwayA0.910.87–0.96
45CCPI1,588USAA, ADP, RF0.970.95–0.99
46CCPI1,427USAADP, RF0.980.97–1.00
34CohI1,000USAA, RF0.980.97–1.00
Summary analysis (number of studies)
 A (9)0.980.97–0.99
 B (8)0.980.97–0.99
 C (3)0.980.96–0.99
Table II. B Summary of Selected Studies of Obesity and Post-Menopausal Breast Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryConfounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; RF, reproductive factors; ADP, age, diet, physical exercise.

36CCPI517CanadaA1.000.98–1.01
37CCPI489DenmarkA1.010.97–1.04
52CohI215 >50: 30 ≤ 50:185USAA1.061.01–1.11
39CCPI, P1,017USAA, RF1.031.01–1.06
54CohI221Iowa, USAA1.010.97–1.04
42CCPI329CanadaA, ADP, RF1.031.00–1.06
32CohI99NorwayA0.970.92–1.03
55CCPI439USAA, RF1.051.02–1.08
57CCPI, P306SwedenNone1.020.98–1.05
60CC in CohI626NetherlandsA, ADP, RF1.000.98–1.03
46CCPI4,921USAADP, RF1.031.02–1.04
34CohI1,517USAA, RF1.011.00–1.02
61CCPI2,704SwedenA, RF1.041.03–1.06
Summary analysis (number of studies)
 A (13)1.021.02–1.03
 B (12)1.021.02–1.03
 C (3)1.031.02–1.04
Table II. C Summary of Selected Studies of Obesity and Colon Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryAge (years)SexConfounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, Cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; ADP, alcohol, diet, physical exercise; RF, reproductive factors; TA, tobacco, alcohol; FH, family history of disease; HRT, hormone replacement therapy; AD, analgesic drug use.

68CCPI163Sweden40–80MA, sex1.081.03–1.12
189F
69CCPI388AustraliaAll casesMA, ADP1.030.98–1.08
327F0.980.95–1.02
71CohI, D302USAMA, RF1.051.01–1.09
73CohD611USA≥30MADP1.021.00–1.05
539F1.010.99–1.04
75CohI203USA40–75M1.041.00–1.08
77CohI393USA30–55FAP, FH, TA, HRT, AD1.031.00–1.05
Summary analysis (number of studies)
 A (8)1.031.02–1.04
 B (5)1.031.01–1.05
 C idemidemidem
Table II. D Summary of Selected Studies of Obesity and Prostate Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryAge (years)Confounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, Cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; BMI21, BMI at age 21.

83CohI174Japanese men, Hawaii, USA46–68A1.040.99–1.09
84CohI180USA≥25A1.020.97–1.07
88CCPI207CanadaA1.020.96–1.08
90CohI2,368SwedenA1.011.00–1.03
91CohI1,338USA40–75A, BMI 211.000.98–1.02
92CCPI325UK<75A1.040.99–1.10
Summary analysis (number of studies)
 A (6)1.011.00–1.02
 B idemidemidem
 C idemidemidem
Table II. E Summary of Selected Studies of Obesity and Endometrial Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryAge (years)Confounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, Cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; RF, reproductive factors; OC, oral contraceptive use.

97CCPI400USA20–74A, OC, RF1.101.07–1.14
99CCPI268China18–74A, RF1.091.03–1.16
100CCPI376USA20–74A, RF1.061.02–1.10
101CCPI232USA40–85A, RF1.141.08–1.20
Summary analysis (number of studies)
 A (4)1.101.07–1.12
 B (4)idemidem
 C (1)1.101.07–1.14
Table II. F Summary of Selected Studies of Obesity and Kidney Cancer and Results of Dose-Response Meta-Analyses
ReferenceStudy1CasesCountryAge (years)SexConfounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; KS, kidney stones; T, tobacco; TC, tobacco, coffee.

104CCPP313USA30–64MA, KS, T1.030.99–1.07
182F1.101.03–1.16
105CCPI209USA≥40MA1.101.05–1.16
105F1.030.96–1.09
107CCPI310Australia20–79MA1.071.01–1.13
179F1.030.98–1.08
109CCPIM:282Canada25–69MA, T1.030.98–1.09
F:181F1.121.05–1.20
110CCPIM:1,050Multinational20–79MA, T1.061.03–1.09
F:682F1.071.04–1.09
111CohDM:212USA≥30F+MA1.061.02–1.10
F:1231.091.04–1.13
112CohIF:62USA55–69FA1.121.03–1.21
Summary analysis (number of studies)
 A (13)1.061.05–1.34
 B (7)1.061.05–1.08
 C (4)1.061.05–1.08
Table II. G Summary of Studies of Obesity and Gallbladder Cancer and Results of Dose-Response Meta-Analysis
ReferenceStudy1CasesCountryAge (years)SexConfounders3Estimated dose-response per unit increase in BMI
Type2NumberCoefficient95% CI
  • 1

    CCH, case-control hospital-based; CCP, case-control population-based; Coh, cohort.

  • 2

    D, dead cases; I, incident cases.

  • 3

    A, age; TA, tobacco, alcohol; SES, socio-economic status.

115CCHIM, 13Bolivia, MexicoAll casesM, 15%; F, 85%A, sex1.090.90–1.33
F, 71
116CCPIM, 44Australia, Canada, Netherlands, PolandAll casesMA, Sex, TA, SES0.990.87–1.11
F, 145F1.081.01–1.15
Summary analysis: 2 studies, 3 observations1.061.00–1.12
 Model A = B

Because excess weight entails other major adverse effects on health,48 no attributable fractions for pre-menopausal breast cancer were estimated.

Menopausal women.

Twenty-seven studies were eligible,31–40, 42, 44, 46, 47, 49–61 conducted in North America (n = 14), Europe (n = 9), China (n = 2) and Israel (n = 1), with 1 multi-national study. Four studies included a small number of pre-menopausal cases, representing a maximum of 21% of cases.52, 53, 57, 58

A statistically significant positive association between obesity and breast cancer in post-menopausal women was reported in 7 of the 13 cohort studies,31, 33, 35, 42, 52, 56, 59 while no significant association was detected in the remaining 6.32, 34, 53, 54, 58, 60 Among the population-based case-control studies, 7 showed a positive association with obesity39, 40, 47, 49–51, 55, 61 and 4 showed no association.37, 38, 46, 57 We considered age, age at menarche, parity and other reproductive factors to be the major potential confounders. Twelve studies accounted for the main reproductive factors34, 35, 39, 40, 42, 46, 47, 49, 55, 59–61 and 3 for alcohol, diet and physical exercise.42, 46, 60

Thirteen studies contributed to the meta-analysis (Table IIB). We estimated a 2% increase in risk per unit increase in BMI (FE 95% CI 1.02–1.03, RE 95% CI 0.69–1.52) (Table IIB). Limiting the analysis to studies with only incident cases, which meant excluding 1 study,39 did not affect the estimates. Further limitations to studies adjusted for age, reproductive factors and alcohol or diet42, 46, 60 marginally affected the point estimate and reduced heterogeneity between studies (RR 1.03 per unit of increase in BMI; FE 95% CI 1.02–1.04, RE 95% CI 0.75–1.27).

The estimates predict that the excess risk is 12% for overweight and 25% for obese women (Table III). Table IV shows the proportion of cases attributable to excess weight in Europe. In total, 8.5% of breast-cancer cases in women aged 50 years and over and 6.6% for all ages could be attributed to overweight and obesity. Table V shows the numbers of new cases attributable to excess weight. In all, 13,000 cases of breast cancer could be avoided each year in the European Union if overweight and obesity were eliminated.

Table III. RR Associated with Overweight and Obesity Predicted by Common Dose-Response Slopes, by Cancer Site
Cancer siteRR for overweight1vs. normal weightRR for obese2vs. normal weight
  • 1

    Obese, BMI ≥ 30.

  • 2

    Overweight, 25 ≤ BMI < 30.

Breast in post-menopausal women1.121.25
Colon1.151.33
Endometrium1.592.52
Prostate1.061.12
Kidney1.361.84
Gallbladder1.341.78
Table IV. Proportion of Cancer Cases Attributable to Overweight and Obesity in the Countries of the European Union, by Cancer Site
CountryBreast womenColonEndometrium womenProstate menKidneyGallbladderTotal
MenWomenMenWomenMenWomenMenWomen
Austria7.410.59.535.14.223.921.922.620.73.76.0
Belgium8.811.611.240.84.625.925.924.824.42.96.7
Denmark5.49.76.826.03.922.116.020.915.12.93.9
Finland9.112.611.641.65.028.426.426.824.94.47.2
France7.610.79.635.64.324.222.222.821.03.16.1
Germany8.812.711.240.45.128.525.627.024.14.96.8
Greece10.011.112.444.24.425.128.223.726.62.15.9
Ireland7.810.19.936.44.022.922.821.721.53.15.0
Italy9.111.411.641.74.625.926.524.525.03.17.5
Luxembourg8.110.710.337.84.324.523.823.122.42.66.7
Netherlands6.39.87.929.83.922.418.521.217.43.14.8
Portugal9.510.712.042.94.324.527.323.125.82.97.1
Spain10.710.813.547.34.424.130.522.828.82.68.8
Sweden6.39.58.030.03.821.718.620.517.53.85.2
United Kingdom7.810.09.836.14.022.722.721.421.42.74.9
European Union8.511.110.739.24.425.524.524.823.73.46.4
Table V. Number of New Cancer Cases Attributable to Overweight and Obesity in the Countries of the European Union Around 1995, by Cancer Site
CountryBreast womenColonEndometrium womenProstate menKidneyGallbladderTotal
MenWomenMenWomenMenWomenMenWomen
Austria260240200290110130906080530930
Belgium43030030042019017012040707001,340
Denmark1401601101705080502020320490
Finland210120120260120110905040400730
France1,8401,7601,4401,6808509004102103603,7205,730
Germany3,3402,9503,0202,9201,3902,0801,2101,5201,2307,94011,720
Greece30017017025070140702030400820
Ireland909070804030201020180280
Italy2,1801,8701,6703,5506201,3406704405404,2708,600
Luxembourg1010102010<10<10<10<102050
The Netherlands48040033042025023013090809701,440
Portugal280290260430901006030605101,080
Spain1,0809901,0501,6603204702802305702,0204,640
Sweden29024020034021014090901006901,020
United Kingdom1,9501,5601,4901,7606807004401802503,1205,890
European Union12,87011,15010,46014,2304,9906,6403,7403,0103,45025,79044,750

Colon cancer

Nineteen studies on excess weight and colon cancer were eligible.1, 58, 62–78 Of the 12 prospective studies, 4 analysed cancers of the colon and rectum as 1 entity.1, 63, 69, 76

Most cohort studies indicated a positive association between overweight and colon cancer;1, 58, 63, 66, 70, 71, 75, 76, 77 this was somewhat less consistent among women and was confined to older men in a Japanese cohort.62 One prospective study showed no association in either sex.64 A positive relationship was supported by most population-based case-control studies,65, 67, 68, 74, 76, 78 generally with a higher RR but often with borderline significance. One case-control study showed no association.69 We considered age, sex, family history of colon cancer, ethnicity, social class, physical exercise and diet as potential confounders. Most studies accounted for age and 1 other of these confounders.63, 67–69, 71–78

Only 6 of the studies reviewed could be included in the meta-analyses68, 69, 71, 73, 75, 77 (Table IIC). When results were available by sex, these were abstracted separately and considered as independent observations in our meta-analysis.69, 73 Since the meta-analyses showed no significant heterogeneity by sex, only the combined results are reported. We found a positive association between excess weight and colon cancer, the average RR being 1.03 (95% CI 1.02–1.04) per unit increase in BMI. Restriction of the analysis to studies having incident cases as outcome68, 69, 75, 77 did not affect the estimate (FE 95% CI 1.01–1.05, RE 95% CI 1.00–1.06). All of these studies accounted for age and at least 1 additional factor (alcohol, diet or physical exercise) so that model C coincided with model B.

The dose-response relationship estimated by model B corresponded to a 15% increase in the risk of developing colon cancer for an overweight person compared with a person having a normal weight and a 33% increase in risk for an obese person (Table III). The proportion of colon cancers attributable to excess weight among Europeans is 11% in either sex (Table IV). Excess weight accounts for about 11,000 new cases of colon cancer in men and 10,000 in women every year (Table V).

The statistical test on the log-linear assumption of the dose-response was not significant in model A (p = 0.67 based on 11 observations) but close to the conventional 5% cut-point for models B (p = 0.07, n = 8) and C (p = 0.11, n = 6). The second-degree model under restriction B would predict RRs of 1.30 and 1.54 for overweight and obese, respectively, corresponding to twice the excess risk of the log-linear model. If we accepted this model, attributable fractions and number of attributable cases for colon cancer would be approximately twice those reported.

Prostate cancer

Seventeen studies of obesity and prostate cancer were eligible for review;1, 58, 79–92 all were conducted in North America or Europe. Two1, 58 of the 9 cohort studies reported a significant association between excess weight and prostate cancer, while the others found no association.79, 80, 83, 84, 87, 90, 91 A positive association was supported by 2 of the population-based case-control studies,85, 89 but the majority showed no relationship with obesity.80, 81, 82, 86, 88, 92 We considered age, family history of prostate cancer, ethnicity, social class, diet, physical activity and occupational exposures to be the most important potential confounders for the association between obesity and prostate cancer. Most of the cohort82, 84, 90 and case-control81, 82, 88, 92 studies accounted only for age.

Only 6 of the studies qualified for meta-analysis, 4 cohort studies, all with incident cases,83, 84, 90, 91 and 2 population-based case-control studies88, 92 (Table IID). All studies accounted for age but none of the other potential confounders. Models A, B and C were therefore identical. Elevated BMI was positively associated with prostate cancer but with borderline statistical significance. The RR was 1.01 per unit of increase in BMI (FE 95% CI 1.00–1.02, RE 95% CI 0.80–1.28) (Table IID).

The estimated RR corresponded to a 6% increase in risk of prostate cancer for an overweight man compared with a normal weight and to a 12% increase in risk for an obese man (Table III). Based on the estimated RR, the proportion of prostate cancers attributable to overweight and obesity among European men is 4% (Table IV). This corresponds to 5,000 new cases per year (Table V).

Endometrial cancer

Fourteen studies on endometrial cancer were included in the review;1, 33, 58, 93–103 1 was conducted in China and the others in North America or Europe. All but 196 of the cohort studies1, 33, 58, 102 and all population-based case-control studies93, 95, 97, 99, 100, 101, 103 showed a positive association between endometrial cancer and excess weight. Associations were stronger with overweight and obesity late in life98 and for women with metastatic disease.95 Besides BMI, measurements of waist-to-hip ratio and of skinfold have been used in studies of endometrial cancer,99, 100, 102, 103 also indicating a positive association. We considered age, social class, parity, use of oral contraceptives and hormone replacement therapy as the main potential confounders. Half of the selected studies adjusted only for age33, 58, 93, 94 or for no confounders.1, 95, 102 Three studies adjusted for all of the major confounders.97, 98, 103

Only 4 studies provided enough information to be included in the meta-analysis97, 99–101 (Table IIE). The average increase in RR was 1.10 per unit of increase in BMI (95% CI 1.07–1.12) (Table IIE). All of these studies included incident cases, and models A and B were therefore identical. Only 1 study qualified for analysis C.97 Restricting the analysis to this single study affected only the CI (RR 1.10, 95% CI 1.07–1.14).

The estimate obtained in analysis A/B implied an increase in risk by 59% for overweight women compared with normal weight and 152% for obese women (Table III). The proportion of endometrial cancers due to obesity in European women was estimated at 39% (Table IV), equivalent to 14,000 new cases per year (Table V).

Kidney cancer

We found 11 studies on kidney cancer eligible for review.1, 58, 104–112 All were conducted in affluent countries. All but 2108, 112 included men and women. The case-control study was limited to cases of renal-cell cancer, the type responsible for 80% to 90% of all adult kidney neoplasms,107 while 2 of the cohort studies included renal-cell and renal pelvic cancers.1, 58 A positive association with obesity was reported in all but 2 studies;1, 106 this was somewhat stronger in women than in men.58, 104, 105, 107–111 We considered age and smoking to be the major potential confounders in the association between obesity and kidney cancer. All but 2 studies1, 106 adjusted for age, but only 4 adjusted for smoking.104, 106, 109, 110

In 2 of the studies, BMI was calculated as kg/m1.5 for women.107, 110 To include these studies in the meta-analysis, we recalculated BMI (kg/m2) by dividing the estimate (kg/m1.5) by the square root of height, using a value of 1.64 m, based on the mean height for women in North America, the Netherlands and Sweden. Of the 7 studies included in the meta-analysis104, 105, 107, 109–112 (Table IIF), all but 1 provided separate results for men and women (contributing 13 degrees of freedom for the estimation of RR). Overweight and obesity were positively associated with kidney cancer (RR = 1.06 per unit of increase in BMI, 95% CI 1.05–1.34) (Table IIF). Restricting the analyses to studies based only on incident cases105, 107, 109, 110, 112 and further to studies that adjusted for age and smoking109, 110 changed the estimates only marginally. Models B and C gave the same estimates for the log-linear relationships between BMI and risk of kidney cancer, though fewer studies contributed to the second. We did not detect heterogeneity by sex: the results of model B in men and women estimated separately were 1.06 (95% CI 1.03–1.08) and 1.07 (95% CI 1.05–1.09), respectively.

Estimated RR corresponds to an increase in risk of 36% for an overweight person compared with one of normal weight and an increase of 84% for an obese person (Table III). In the European Union, about 25% of male cases and 24% of female cases of kidney cancer can be attributed to excess weight (Table IV). This corresponds to over 10,000 new cases of kidney cancer annually (Table V).

Gallbladder cancer

We found so few epidemiological studies of gallbladder cancer that eligibility criteria could not be applied. The 6 studies1, 58, 113–116 are summarised in Table IIG. The results of these studies are conflicting. An excess of deaths from cancer of the gallbladder or biliary ducts was observed in women weighing 10% more than the average population, but no excess risk was observed in men.1 Furthermore, no association was found for either sex in a Danish cohort study.58 All but 1113 of the case-control studies indicated an increased risk of gallbladder cancer with obesity. However, the association was limited to women in the large multi-centre study of Zatonski et al..116 This was also the only study that controlled for the potential risk factors age, alcohol drinking, tobacco smoking and socio-demographic characteristics.

Only 1 study115 provided sufficient information to estimate a dose-response relationship, and this was conducted in South America. The only data missing in the largest study116 were the cut-points corresponding to the BMI quartiles that define exposure levels. We tested the extent to which different assumptions on the distribution of the BMI, within reasonable limits, would affect the estimated slope and found variations on the order of 1/1,000. We therefore included this study in our calculations, assuming as cut-points those given in the study of kidney cancer by Mellemgaard et al.,110 which was conducted with similar methodology in the same populations. Data for men and women were included as independent observations. We found a positive association (RR = 1.06 per unit of increase in BMI, 95% CI 1.00–1.12). The results are presented in Table IIG.

The estimated association predicted 34% and 78% increases in risk for overweight and obese vs. normal subjects, respectively (Table III). As a result, 24% of these rare cancers are attributable to excess body mass in the European Union (Table IV), i.e., 6,000 new cases per year (Table V).

All sites combined

The strongest relationships between excess weight and cancer risk were for cancers of the endometrium, kidney and gallbladder (Table III), giving correspondingly high attributable proportions: 2 of 5 endometrial cancer cases and 1 of 4 kidney and gallbladder cancers (Table IV). These are, however, relatively uncommon sites and, thus, add little to the total burden of disease (Table V). The largest number of attributable cases is for colon cancer, with 21,500 cases annually due to overweight and obesity, or 1.5% of the 1.5 million annual new cases in the European Union. Overall, excess body mass accounts for 5% of all cancers, 3% in men and 6% in women (Table VI), corresponding to 27,000 male and 45,000 female cancer cases yearly. Overweight accounts for a slightly higher proportion of cases among men, while cancer cases due to obesity are more common among women.

Table VI. Proportion of Cancer Cases Attributable to Overweight and Obesity in the European Union, by Cancer Site
 MenWomen
Overweight1Obese2Overweight1Obese2
  • 1

    Obese, BMI ≥ 30.

  • 2

    Overweight, 25 ≤ BMI < 30.

Breast4.14.5
Colon6.94.25.05.7
Endometrium17.222.0
Prostate2.91.6
Kidney15.210.311.113.4
Gallbladder14.710.110.713.0
All cancer sites2.11.32.93.5

The greatest attributable proportion of cases, for women, is estimated in Spain and, for men, in Germany.

A clear gradient from north to south is observed for the male to female ratio of attributable cases. High body mass accounts for 3 times more cases in women than in men in Spain, Greece and Portugal, while in northern Europe, the ratio is ≤2, mainly because of the higher occurrence of tobacco-related cancers in men in northern Europe.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Our analysis indicates that each year about 34,800 new cases of cancer in the European Union are related to obesity and a further 37,000 cases to overweight. Most likely, these are conservative estimates if the association of BMI with several other cancer sites is confirmed. Our findings have considerable public health relevance since they suggest that it is possible to prevent an appreciable proportion of cancer cases by maintaining a healthy body weight. In terms of number of cases, reduction of body mass would have its greatest effects on endometrial, breast and colon cancers.

We quantified the proportion of cancers attributable to excess body mass for selected cancer sites in the European Union, following a systematic approach. For this purpose, we implicitly assumed that the associations considered were causal. The concept of causality in observational sciences has been thoroughly discussed in the epidemiological literature.117–119 The assumption of causality is established based not only on epidemiological evidence but also on information from other domains, such as animal experiments and molecular biology.120

Walter121, 122 discussed the concept and interpretation of the attributable fraction. For a multifactorial disease such as cancer, the impact of a causal factor in determining the cancer burden in a given population might be modified by the prevalence of co-factors. Any intervention to reduce excess weight would entail the modification of nutritional habits towards a different composition of the habitual diet to reduced energy intake (increased consumption of vegetables at the expense of sources of protein, fat and carbohydrates) and increased physical exercise. The latter 2 may play independent roles in the causation of the same cancers, and the overall impact of interventions to reduce excess weight might therefore be greater than that estimated in our work.

We carried out these meta-analyses to obtain quantitative estimates of the strength of the associations considered; their statistical significance was, therefore, not a critical aspect of the evaluation. We tried to limit the impact of biases potentially affecting observational studies by selecting all those which satisfy a priori characteristics of study design so as to mimic the situation of pooled analyses of experimental studies. By selecting large studies we meant to reduce “publication” bias (failure to include negative studies never published because of their outcome). These take advantage of external funding and are therefore bound to report their results. The (a priori) limitations which we imposed in defining studies eligible for the analysis, therefore, were aimed at improving the validity of our results.

Conversely, an unwanted limitation is that not all published studies provide sufficient information to permit their inclusion in our analyses. This is a limitation always faced in meta-analyses relying only on material as published. For each cancer site, several of the eligible studies could not be included in the meta-analysis due to at least 1 of the following conditions: use of measures of obesity other than BMI, missing RR or 95% CI. The effect of this forced selection on the point estimates of the slopes is unpredictable but certainly not biased by a priori hypotheses. One can instead affirm that their 95% CIs do not reflect the amount of information published and would have been narrower if all information could be exploited for the quantitative analyses. We estimated the RR in 3 models, with gradually increasing restrictions for inclusion criteria. For all of the 6 cancer sites studied, the RRs and 95% CIs thus obtained were virtually the same.

The cut-points of BMI which we adopted to define the population at risk are those recommended by WHO. These are based on evidence that mortality (all causes) increases significantly with BMI above 25 kg/m,2 which applies to several diseases common in Western countries.19, 22 Since the ultimate objective of this kind of exercise is to provide a basis for public health priorities, the definition of the “at-risk” population should be consistent with general guidelines.

We used non-national prevalence data obtained from 2 consistent sources26, 27 to improve comparability of the data from the different countries considered, possibly at the expense of representativeness. National estimates for the relevant period have been published for 3 countries. In the age range of the MONICA study, the prevalence of obesity in Germany in 1990123 was similar to the figures we used. In the United Kingdom, national estimates of population excess weight (BMI >25 kg/m2) in 1980 were greater than the MONICA figures by 7% and 11% in men and women, respectively.124 In Swedish men, national estimates from 1988 to 1989125 were 13% lower than the figures we used. There were no comparable data for Swedish women.

Overall, our results are in agreement with fractions attributable to obesity computed for the Nordic countries in a previous study.126 Exceptions are estimates of endometrial and prostate cancers, which are lower in our results, and the attributable proportions of gallbladder cancers in women, which are higher. The differences for prostate cancer are explained by different RR values (1.12 in our study vs. 2.0). Other differences are therefore due to differences in the estimated prevalence of obesity. Our estimate of menopausal breast cancer attributable to overweight is in agreement with an estimate obtained from direct observation.127

We aimed at quantifying the magnitude of the problem of overweight as a basis of the potential of interventions to reduce cancer burden. Over 70,000 cancer cases in the European Union may be attributed to overweight and obesity, corresponding to 5% of all cancer cases. It is probably unrealistic to aim at eliminating completely this factor in Europe, where approximately 50% of the adult population is either overweight or obese; however, a substantial number of cancer cases could be avoided by halving that prevalence. Because excess weight is on the increase in most European countries,128 adverse effects might therefore be even more pronounced in the future. Translation of the present findings into preventive strategies should become an important public health priority.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We are grateful to Dr J.C. Seidell, for providing data on the prevalence of excess weight in the Netherlands, and Dr J. Cheney and Ms B. Geoffre for editing the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Lew EA, Garfinkel L. 1979. Variations in mortality by weight among 750,000 men and women. J Chronic Dis 1979;32: 563576.
  • 2
    Hill HA, Austin H. Nutrition and endometrial cancer. Cancer Causes Control 1996;7: 1932.
  • 3
    Wolk A, Lindblad P, Adami HO. Nutrition and renal cell cancer. Cancer Causes Control 1996;7: 518.
  • 4
    Hunter DJ, Willett WC. Nutrition and breast cancer. Cancer Causes Control 1996;7: 5668.
  • 5
    Potter JD. Nutrition and colorectal cancer. Cancer Causes Control 1996;7: 127146.
  • 6
    Fraumeni JF, Devesa SS, McLaughlin JK, Stanford J. Biliary tract cancer. In: SchottenfeldD, FraumeniJFJr. Cancer epidemiology and prevention. New York: Oxford University Press, 1996.
  • 7
    Kolonel LN. Nutrition and prostate cancer. Cancer Causes Control 1996;7: 8394.
  • 8
    Kolonel LN, Hankin JH, Wilkens LR, Fukunaga FH, Hinds MW. An epidemiologic study of thyroid cancer in Hawaii. Cancer Causes Control 1990;1: 223234.
  • 9
    Ron E, Kleinerman RA, Boice J, Livolsi VA, Flannery JT, Fraumeni JF Jr. A population-based case-control study of thyroid cancer. J Natl Cancer Inst 1987;79: 112.
  • 10
    Chow WH, Blot WJ, Vaughan TL, et al. Body mass index and risk of adenocarcinomas of the esophagus and gastric cardia. J Natl Cancer Inst 1998;90: 150155.
  • 11
    Ji BT, Chow WH, Yang G, et al. Body mass index and the risk of cancers of the gastric cardia and distal stomach in Shanghai, China. Cancer Epidemiol Biomarkers Prev 1997;6: 481485.
  • 12
    Hansson LE, Baron J, Nyren O, et al. Early-life risk indicators of gastric cancer. A population-based case-control study in Sweden. Int J Cancer 1994;57: 3237.
  • 13
    Lagergren J, Bergstrom R, Nyren O. Association between body mass and adenocarcinoma of the esophagus and gastric cardia. Ann Intern Med 1999;130: 883890.
  • 14
    Greenland P, Longnecker MP. Methods for trend estimation from summarised dose-response data, with application to meta-analysis. Am J Epidemiol 1992;135: 301309.
  • 15
    Berlin JA, Longnecker MP, Greenland S. Meta-analysis of epidemiologic dose-response data. Epidemiology 1993;4: 218228.
  • 16
    Cole P, MacMahon B. Attributable risk percent in case-control studies. Br J Prev Soc Med 1971;25: 242244.
  • 17
    Burton BT, Foster WR, Hirsch J, Van Itallie TB. Health implications of obesity: an NIH Consensus Development Conference [published erratum appears in Int J Obes 1986;10:79]. Int J Obes 1985;9: 155170.
  • 18
    Royal College of Physicians. A report of the Royal College of Physicians, Obesity. J R Coll Physicians Lond 1983;17: 565.
  • 19
    Willett WC, Dietz WH, Colditz GA. Guidelines for healthy weight. N Engl J Med 1999;341: 427434.
  • 20
    Ursin G, Longnecker MP, Haile RW, Greenland S. A meta-analysis of body mass index and risk of pre-menopausal breast cancer. Epidemiology 1995;6: 137141.
  • 21
    FAO/WHO/UNO. Energy and protein requirements; report of a Joint Expert Consultation. WHO Technical report 724. Geneva: World Health Organization, 1985.
  • 22
    World Health Organisation Expert Committee on Physical Status. The use and interpretation of anthropometry. World Health Organisation Technical Report 854. Geneva: WHO, 1995.
  • 23
    Ferlay J, Bray F, Sankila R, Parkin DM. EUCAN: cancer incidence, mortality and prevalence in the European Union 1995, version 2.0. IARC CancerBase 4. Lyon: IARC Press, 1999.
  • 24
    Parkin DM, Whelan SL, Ferlay J, Raymond L, Young J. Cancer incidence in five continents, Volume VII, IARC Scientific Publication 143. Lyon: IARC, 1997.
  • 25
    World Health Organisation. Causes of death statistics and vital rates, civil registration systems and alternative sources of information. In: World Health Statistics Annual 1995. Geneva: WHO, 1997.
  • 26
    World Health Organisation MONICA Project. A worldwide monitoring system for cardiovascular diseases: cardiovascular mortality and risk factors in selected communities. In: World Health Statistics Annual 1987. Geneva: WHO, 1989: 2759.
  • 1
    MorgensternW, TsechkovskiMS, NusselE, SchettlerG (eds.) CINDI, Countrywide Integrated Noncommunicable Diseases Intervention Programme: baseline evaluation. Berlin: WHO Regional Office for Europe and Springer-Verlag, 1991.
  • 28
    Kafatos A, Kouroumalis I, Vlachonikolis I, Theodorou C, Lababdarios D. Coronary-heart-disease risk-factor status of the Cretan urban population in the 1980s. Am J Clin Nutr 1991;54: 591598.
  • 29
    Mamalakis G, Kafatos A. Prevalence of obesity in Greece. Int J Obes Relat Metab Disord 1996;20: 488492.
  • 30
    Seidell JC. Prevalence of obesity in Europe. Bibl Nutr Dieta 1989;44: 17.
  • 31
    Tretli S. Height and weight in relation to breast cancer morbidity and mortality. A prospective study of 570,000 women in Norway. Int J Cancer 1989;44: 2330.
  • 32
    Vatten LJ, Kvinnsland S. Body mass index and risk of breast cancer. A prospective study of 23,826 Norwegian women. Int J Cancer 1990;45: 440444.
  • 33
    Tornberg SA, Carstensen JM. Relationship between Quetelet's index and cancer of breast and female genital tract in 47,000 women followed for 25 years. Br J Cancer 1994;69: 358361.
  • 34
    Huang Z, Hankinson SE, Colditz GA, et al. Dual effects of weight and weight gain on breast cancer risk. JAMA 1997;278: 14071411.
  • 35
    Yong LC, Brown CC, Schatzkin A, Schairer C. Prospective study of relative weight and risk of breast cancer: the Breast Cancer Detection Demonstration Project follow-up study, 1979 to 1987–1989. Am J Epidemiol 1996;143: 985995.
  • 36
    Hislop TG, Coldman AJ, Elwood JM, Brauer G, Kan L. Childhood and recent eating patterns and risk of breast cancer. Cancer Detect Prev 1986;9: 4758.
  • 37
    Ewertz M. Influence of non-contraceptive exogenous and endogenous sex hormones on breast cancer risk in Denmark. Int J Cancer 1988;42: 832838.
  • 38
    Tao SC, Yu MC, Ross RK, Xiu KW. Risk factors for breast cancer in Chinese women of Beijing. Int J Cancer 1988;42: 495498.
  • 39
    Swanson CA, Brinton LA, Taylor PR, Licitra LM, Ziegler RG, Schairer C. Body size and breast cancer risk assessed in women participating in the Breast Cancer Detection Demonstration Project. Am J Epidemiol 1989;130: 11331141.
  • 40
    Hsieh CC, Trichopoulos D, Katsouyanni K, Yuasa S. Age at menarche, age at menopause, height and obesity as risk factors for breast cancer: associations and interactions in an international case-control study. Int J Cancer 1990;46: 796800.
  • 41
    Lund E, Adami HO, Bergstrom R, Meirik O. Anthropometric measures and breast cancer in young women. Cancer Causes Control 1990;1: 16972.
  • 42
    Rosenberg L, Palmer JR, Miller DR, Clarke EA, Shapiro S. A case-control study of alcoholic beverage consumption and breast cancer. Am J Epidemiol 1990;131: 614.
  • 43
    Wang QS, Ross RK, Yu MC, Ning JP, Henderson BE, Kimm HT, A case-control study of breast cancer in Tianjin, China. Cancer Epidemiol Biomarkers Prev 1992;1: 435439.
  • 44
    Ziegler RG, Hoover RN, Nomura AM, et al. Relative weight, weight change, height, and breast cancer risk in Asian-American women. J Natl Cancer Inst 1996;88: 650660.
  • 45
    Swanson CA, Coates RJ, Schoenberg JB, Malone KE, Gammon MD, Stanford JL, et al. Body size and breast cancer risk among women under age 45 years. Am J Epidemiol 1996;143: 698706.
  • 46
    Trentham Dietz A, Newcomb PA, Storer BE, Longnecker MP, Baron J, Greenberg ER, et al. Body size and risk of breast cancer. Am J Epidemiol 1997;145: 10111019.
  • 47
    Chu SY, Lee NC, Wingo PA, Senie RT, Greenberg RS, Peterson HB. The relationship between body mass and breast cancer among women enrolled in the Cancer and Steroid Hormone Study. J Clin Epidemiol 1991;44: 11971206.
  • 48
    Stevens J, Cai J, Pamuk ER, et al. The effect of age on the association between body-mass index and mortality. N Engl J Med 1998;338: 17.
  • 49
    Lubin F, Ruder AM, Wax Y, Modan B. Overweight and changes in weight throughout adult life in breast cancer etiology. A case-control study. Am J Epidemiol 1985;122: 579588.
  • 50
    Kolonel LN, Nomura AM, Lee J, et al. Anthropometric indicators of breast cancer risk in post-menopausal women in Hawaii. Nutr Cancer 1986;8: 247256.
  • 51
    Yuan JM, Yu MC, Ross RK, Gao YT, Henderson BE. Risk factors for breast cancer in Chinese women in Shanghai. Cancer Res 1988;48: 19491953.
  • 52
    Mills PK, Beeson WL, Phillips RL, Fraser GE. Dietary habits and breast cancer incidence among Seventh-day Adventists. Cancer 1989;64: 582590.
  • 53
    Ballard Barbash R, Schatzkin A, Carter CL, Kannel WB, Kreger BE, D'agostino RB, et al. Body fat distribution and breast cancer in the Framingham Study. J Natl Cancer Inst 1990;82: 286290.
  • 54
    Folsom AR, Kaye SA, Prineas RJ, Potter JD, Gapstur SM, Wallace RB. Increased incidence of carcinoma of the breast associated with abdominal adiposity in post-menopausal women. Am J Epidemiol 1990;131: 794803.
  • 55
    Graham S, Hellmann R, Marshall J, et al. Nutritional epidemiology of post-menopausal breast cancer in western New York. Am J Epidemiol 1999;134: 552566.
  • 56
    den Tonkelaar I, Seidell JC, Collette HJ, De Waard F. A prospective study on obesity and subcutaneous fat patterning in relation to breast cancer in post-menopausal women participating in the DOM project. Br J Cancer 1994;69: 352357.
  • 57
    Holmberg L, Ohlander EM, Byers T, et al. Diet and breast cancer risk. Results from a population-based, case-control study in Sweden. Arch Intern Med 1994;154: 18051811.
  • 58
    Møller H, Mellemgaard A, Lindvig K, Olsen JH. Obesity and cancer risk: a Danish record-linkage study. Eur J Cancer 1994;30A: 344350.
  • 59
    Freni SC, Eberhardt MS, Turturro A, Hine RJ. Anthropometric measures and metabolic rate in association with risk of breast cancer (United States). Cancer Causes Control 1996;7: 358365.
  • 60
    van den Brandt PA, Dirx MJ, Ronckers CM, van den Hoogen P, Goldbohm RA. Height, weight change, and post-menopausal breast cancer risk: the Netherlands Cohort Study. Cancer Causes Control 1997;8: 3947.
  • 61
    Magnusson C, Baron J, Persson I, Wolk A, Bergstrom R, Trichopoulos D, et al. Body size in different periods of life and breast cancer risk in post-menopausal women. Int J Cancer 1998;76: 2934.
  • 62
    Nomura A, Heilbrun LK, Stemmermann GN. Body mass index as a predictor of cancer in men. J Natl Cancer Inst 1985;74: 319323.
  • 63
    Phillips RL, Snowdon DA. Dietary relationships with fatal colorectal cancer among Seventh-day Adventists. J Natl Cancer Inst 1985;74: 307317.
  • 64
    Whittemore AS, Paffenbarger RS Jr, Anderson K, Lee JE. Early precursors of site-specific cancers in college men and women. J Natl Cancer Inst 1985;74: 4351.
  • 65
    Graham S, Marshall J, Haughey B, Mittelman A, Swanson M, Zielezny M, et al. Dietary epidemiology of cancer of the colon in western New York. Am J Epidemiol 1988;128: 490503.
  • 66
    Waaler HT. Hazard of obesity—the Norwegian experience. Acta Med Scand Suppl 1988;723: 1721.
  • 67
    West DW, Slattery ML, Robison LM, Schuman KL, Ford MH, Mahoney AW, et al. Dietary intake and colon cancer: sex- and anatomic site-specific associations. Am J Epidemiol 1989;130: 883894.
  • 68
    Gerhardsson de Verdier M, Hagman U, Steineck G, Rieger A, Norell SE. Diet, body mass and colorectal cancer: a case-referent study in Stockholm. Int J Cancer 1990;46: 832838.
  • 69
    Kune GA, Kune S, Watson LF. Body weight and physical activity as predictors of colorectal cancer risk. Nutr Cancer 1990;13: 917.
  • 70
    Kreger BE, Anderson KM, Schatzkin A, Splansky GL. Serum cholesterol level, body mass index, and the risk of colon cancer. The Framingham Study. Cancer 1992;70: 10381043.
  • 71
    Lee IM, Paffenbarger RS Jr. Quetelet's index and risk of colon cancer in college alumni. J Natl Cancer Inst 1992;84: 13261331.
  • 72
    Le Marchand L, Wilkens LR, Mi MP. Obesity in youth and middle age and risk of colorectal cancer in men. Cancer Causes Control 1992;3: 349354.
  • 73
    Thun MJ, Calle EE, Namboodiri MM, Flanders WD, Coates RJ, Byers T, et al. Risk factors for fatal colon cancer in a large prospective study. J Natl Cancer Inst 1992;84: 14911500.
  • 74
    Dietz AT, Newcomb PA, Marcus PM, Storer BE. The association of body size and large bowel cancer risk in Wisconsin (United States) women. Cancer Causes Control 1995;6: 3036.
  • 75
    Giovannucci E, Ascherio A, Rimm EB, Colditz GA, Stampfer MJ, Willett WC. Physical activity, obesity, and risk for colon cancer and adenoma in men. Ann Intern Med 1995;122: 327334.
  • 76
    Le Marchand L, Wilkens LR, Kolonel LN, Hankin JH, Lyu LC. Associations of sedentary lifestyle, obesity, smoking, alcohol use, and diabetes with the risk of colorectal cancer. Cancer Res 1997;57: 47874794.
  • 77
    Martinez ME, Giovannucci E, Spiegelman D, Hunter DJ, Willett WC, Colditz GA. Leisure-time physical activity, body size, and colon cancer in women. Nurses' Health Study Research Group. J Natl Cancer Inst 1997;89: 948955.
  • 78
    Slattery ML, Potter J, Caan B, Edwards S, Coates A, Ma KN, et al. Energy balance and colon cancer—beyond physical activity. Cancer Res 1997;57: 7580.
  • 79
    Greenwald P, Damon A, Kirmss V, Polan AK. Physical and demographic features of men before developing cancer of the prostate. J Natl Cancer Inst 1974;53: 341346.
  • 80
    Whittemore AS, Kolonel LN, Wu AH, et al. Prostate cancer in relation to diet, physical activity, and body size in blacks, whites, and Asians in the United States and Canada. J Natl Cancer Inst 1995;87: 652661.
  • 81
    Ross RK, Shimizu H, Paganini Hill A, Honda G, Henderson BE. Case-control studies of prostate cancer in blacks and whites in southern California. J Natl Cancer Inst 1987;78: 869874.
  • 82
    Kolonel LN, Yoshizawa CN, Hankin JH. Diet and prostatic cancer: a case-control study in Hawaii. Am J Epidemiol 1988;127: 9991012.
  • 83
    Severson RK, Grove JS, Nomura AM, Stemmermann GN. Body mass and prostatic cancer: a prospective study. BMJ 1988;297: 713715.
  • 84
    Mills PK, Beeson WL, Phillips RL, Fraser GE. Cohort study of diet, lifestyle, and prostate cancer in Adventist men. Cancer 1989;64: 598604.
  • 85
    Fincham SM, Hill GB, Hanson J, Wijayasinghe C. Epidemiology of prostatic cancer: a case-control study. Prostate 1990;17: 189206.
  • 86
    West DW, Slattery ML, Robison LM, French TK, Mahoney AW. Adult dietary intake and prostate cancer risk in Utah: a case-control study with special emphasis on aggressive tumors. Cancer Causes Control 1991;2: 8594.
  • 87
    Le Marchand L, Kolonel LN, Wilkens LR, Myers BC, Hirohata T. Animal fat consumption and prostate cancer: a prospective study in Hawaii. Epidemiology 1994;5: 276282.
  • 88
    Rohan TE, Howe GR, Burch JD, Jain M. Dietary factors and risk of prostate cancer: a case-control study in Ontario, Canada. Cancer Causes Control 1995;6: 145154.
  • 89
    Gronberg H, Damber L, Damber JE. Total food consumption and body mass index in relation to prostate cancer risk: a case-control study in Sweden with prospectively collected exposure data. J Urol 1996;155: 969974.
  • 90
    Andersson SO, Wolk A, Bergstrom R, Adami HO, Engholm G, Englund A, et al. Body size and prostate cancer: a 20-year follow-up study among 135006 Swedish construction workers. J Natl Cancer Inst 1997;89: 385389.
  • 91
    Giovannucci E, Rimm EB, Stampfer MJ, Colditz GA, Willett WC. Height, body weight, and risk of prostate cancer. Cancer Epidemiol Biomarkers Prev 1997;6: 557563.
  • 92
    Key TJ, Silcocks PB, Davey GK, Appleby PN, Bishop DT. A case-control study of diet and prostate cancer. Br J Cancer 1997;76: 678687.
  • 93
    Elwood JM, Cole P, Rothman KJ, Kaplan SD. Epidemiology of endometrial cancer. J Natl Cancer Inst 1977;59: 10551060.
  • 94
    Jensen H. Relationship of premorbid state of nutrition to endometrial carcinoma. Acta Obstet Gynecol Scand 1986;65: 301306.
  • 95
    Tretli S, Magnus K. Height and weight in relation to uterine corpus cancer morbidity and mortality. A follow-up study of 570,000 women in Norway. Int J Cancer 1990;46: 165172.
  • 96
    Le Marchand L, Wilkens LR, Mi MP. Early-age body size, adult weight gain and endometrial cancer risk. Int J Cancer 1991;48: 807811.
  • 97
    Brinton LA, Berman ML, Mortel R, Twiggs LB, Barrett RJ, Wilbanks GD, et al. Reproductive, menstrual, and medical risk factors for endometrial cancer: results from a case-control study. Am J Obstet Gynecol 1992;167: 13171325.
  • 98
    Levi F, La Vecchia C, Negri E, Parazzini F, Franceschi S. Body mass at different ages and subsequent endometrial cancer risk. Int J Cancer 1992;50: 567571.
  • 99
    Shu XO, Brinton LA, Zheng W, Swanson CA, Hatch MC, Gao YT, et al. Relation of obesity and body fat distribution to endometrial cancer in Shanghai, China. Cancer Res 1992;52: 38653870.
  • 100
    Swanson CA, Potischman N, Wilbanks GD, Twiggs LB, Mortel R, Berman ML, et al. Relation of endometrial cancer risk to past and contemporary body size and body fat distribution. Cancer Epidemiol Biomarkers Prev 1993;2: 321327.
  • 101
    Olson SH, Trevisan M, Marshall JR, Graham S, Zielezny M, Vena JE, et al. Body mass index, weight gain, and risk of endometrial cancer. Nutr Cancer 1995;23: 141149.
  • 102
    De Waard F, De Ridder CM, Baanders-Van-Halewijn EA, Slotboom BJ. Endometrial cancer in a cohort screened for breast cancer. Eur J Cancer Prev 1996;5: 99104.
  • 103
    Goodman MT, Hankin JH, Wilkens LR, Lyu LC, McDuffie K, Liu LQ, et al. Diet, body size, physical activity, and the risk of endometrial cancer. Cancer Res 1997;57: 50775085.
  • 104
    McLaughlin JK, Mandel JS, Blot WJ, Schuman LM, Mehl ES, Fraumeni JF Jr. A population-based case-control study of renal cell carcinoma. J Natl Cancer Inst 1984;72: 275284.
  • 105
    Asal NR, Risser DR, Kadamani S, Geyer JR, Lee ET, Cherng N. 1988. Risk factors in renal cell carcinoma: I. Methodology, demographics, tobacco, beverage use, and obesity. Cancer Detect Prev 1988;11: 359377.
  • 106
    Partanen T, Heikkila P, Hernberg S, Kauppinen T, Moneta G, Ojajarvi A. Renal cell cancer and occupational exposure to chemical agents. Scand J Work Environ Health 1991;17: 231239.
  • 107
    McCredie M, Stewart JH. 1992. Risk factors for kidney cancer in New South Wales, Australia. II. Urologic disease, hypertension, obesity, and hormonal factors. Cancer Causes Control 1992;3: 323331.
  • 108
    Finkle WD, McLaughlin JK, Rasgon SA, Yeoh HH, Low JE. Increased risk of renal cell cancer among women using diuretics in the United States. Cancer Causes Control 1993;4: 555558.
  • 109
    Kreiger N, Marrett LD, Dodds L, Hilditch S, Darlington GA. Risk factors for renal cell carcinoma: results of a population-based case-control study. Cancer Causes Control 1993;4: 101110.
  • 110
    Mellemgaard A, Lindblad P, Schlehofer B, et al. 1995. International renal-cell cancer study. III. Role of weight, height, physical activity, and use of amphetamines. Int J Cancer 1995;60: 350354.
  • 111
    Heath CW Jr, Lally CA, Calle EE, McLaughlin JK, Thun MJ. Hypertension, diuretics, and antihypertensive medications as possible risk factors for renal cell cancer. Am J Epidemiol 1997;145: 607613.
  • 112
    Prineas RJ, Folsom AR, Zhang ZM, Sellers TA, Potter J. Nutrition and other risk factors for renal cell carcinoma in post-menopausal women. Epidemiology 1997;8: 3136.
  • 113
    Kato K, Akai S, Tominaga S, Kato I. A case-control study of biliary tract cancer in Niigata Prefecture, Japan. Jpn J Cancer Res 1989;80: 932938.
  • 114
    Moerman CJ, Bueno De Mesquita H, Smeets FW, Runia S. Consumption of foods and micronutrients and the risk of cancer of the biliary tract. Prev Med 1995;24: 591602.
  • 115
    Strom BL, Soloway RD, Rios Dalenz JL, Rodriguez Martinez HA, West SL, Kinman JL, et al. Risk factors for gallbladder cancer. An international collaborative case-control study. Cancer 1995;76: 17471756.
  • 116
    Zatonski WA, Lowenfels AB, Boyle P, et al. Epidemiologic aspects of gallbladder cancer: a case-control study of the SEARCH Program of the International Agency for Research on Cancer. J Natl Cancer Inst 1997;89: 11321138.
  • 117
    Evans AS. Causation and disease: a chronological journey. Am J Epidemiol 1978;108: 249258.
  • 118
    Rothman JK, Poole C. Causation and causal inference. In: SchottenfeldD, FraumeniJF. Cancer epidemiology and prevention. New York: Oxford University Press, 1996. pp 310.
  • 2
    MacMahon B, Trichopoulos D. Epidemiology: principles and methods, 2nd edition. Little, Brown, Boston, 1996.
  • 120
    IARC. Hormonal contraception and post-menopausal hormonal therapy [preamble]. IARC monographs on the evaluation of carcinogenic risks to humans, Volume 72. Lyon: IARC, 1999.
  • 121
    Walter SD. The estimation and interpretation of attributable risk in health research. Biometrics 1976;32: 829849.
  • 122
    Walter SD. Prevention for multifactorial diseases. Am J Epidemiol. 1980;112: 409416.
  • 123
    Hoffmeister H, Mensink GB, Stolzenberg H. National trends in risk factors for cardiovascular disease in Germany. Prev Med 1994;23: 197205.
  • 124
    Millar WJ, Stephens T. The prevalence of overweight and obesity in Britain, Canada, and United States [published erratum appears in Am J Public Health 1987;77:818]. Am J Public Health 1987;77: 3841.
  • 125
    Kuskowska Wolk A, Bergstrom R. Trends in body mass index and prevalence of obesity in Swedish men 1980–89. J Epidemiol Community Health 1993;47: 103108.
  • 126
    Winther JF, Dreyer L, Overvad K, et al. Diet, obesity & physical activity. In: Olsen JH, Andersen A, Dreyer L, Pukkala E, Tryggvadottir L, Gerhardsson de Verdier M. Avoidable cancers in the Nordic countries. APMIS 1997; 105(Suppl 76): 100119.
  • 127
    Mezzetti M, La Vecchia C, Decarli A, Boyle P, Talamini R, Franceschi S. Population attributable risk for breast cancer: diet, nutrition, and physical exercise. J Natl Cancer Inst 1998;90: 389394.
  • 128
    Seidell JC, Verschuren WM, Kromhout D. Prevalence and trends of obesity in the Netherlands 1987–1991. Int J Obes Relat Metab Disord 1995;19: 924927.