Table 16 Studies examining the relationship between physical activity and breast cancer.

Rockhill et al 1999 [106]

To examine the effect of PA on the risk for BC.

• n = 121,701

PA assessment: Self-reported LTPA, grouped into hr/wk

3,137 cases of BC

Women who engaged in 7 or more hours per week of MVPA had a 20% lower risk of BC. An inverse dose-response relationship existed between PA and BC incidence.

• Sex: Women

• Age: 30-55 yr

RR (95% CI) for BC and LTPA

USA

• 16-yr follow-up

G1 = <1

• Characteristics: Free of BC

G2 = 1.0-1.9

   • G1 = 1.00 (referent)

Prospective cohort

G3 = 2.0-3.9

   • G2 = 0.88 (0.79-0.98)

• The nurses Health Study

G4 = 4.0-6.9

   • G3 = 0.89 (0.81-0.99)

D & B score = 13

G5 = ≥7

   • G4 = 0.85 (0.77-0.94)

Multivariate pooled logistic regression

   • G5 = 0.82 (0.70-0.97)

Trend p = 0.004

Sesso et al 1998 [107]

To examine the association between PA and BC among postmenopausal women.

• n = 1,566

31-yr follow-up

109 cases of BC

There is an inverse relationship between PA and BC in post-menopausal women.

• Sex: Women

USA

• Age: 45.5

PA assessment: Questionnaire at baseline, divided into tertiles (kcal/wk)

RR (95% CI) for BC and PA

• Characteristics: Free of BC

Prospective cohort

   • T1 = 1.00 (referent)

T1 = <500

   • T2 = 0.92 (0.58-1.45)

D & B score = 14

T2 = 500-999

T3 = ≥ 1,000

   • T3 = 0.73 (0.46-1.14)

RR (95% CI), post-menopausal women only

   • T1 = 1.00 (referent)

   • T2 = 0.95 (0.58-1.57)

   • T3 = 0.49 (0.28-0.86)

Dosemeci et al 1993 [278]

To conduct a multiple-site case-control study of 15 cancers to examine associations between PA, SES, and these cancer sites among workers.

• n = 2,643 control group

Cases: obtained from an oncological treatment center from 1979-1984

31 men had BC and 241 women had BC

This study shows the sitting-time index showed an elevated risk of female BC for sedentary jobs without SES adjustment.

• n = 2,127 men and n = 244 women)

• Sex: Men and women

Turkey

• Characteristics: Cases - diagnosed with one of the 15 cancers being examined. Control Group - subjects diagnosed with non-cancers, cancers of the buccal cavity, esophagus, liver, bone, soft tissue, brain, lymphoma and other cancer sites for which there is no suggestion on an association with PA

Adjusted SES OR (95%CI), men

Controls: pulled from the same hospital as the cases

Case control

   • G1 = 1.40 (0.60-3.90)

D & B score = 12

PA assessment: OPA (kJ/min)

   • G2 = 1.10 (0.40-3.10)

G1 = <8

   • G3 = 1.00 (referent)

The slightly elevated risk of male BC was based on a small number and disappeared when the risk was adjusted for SES.

G2 = 8-12

Trend p = 0.34

G3 = >12

Adjusted SES OR (95%CI), women

Gart's method and Mantel's chi-square test

   • G1 = 1.10 (0.60-2.10)

   • G2 = 0.90 (0.50-1.80)

   • G3 = 1.00 (referent)

Trend p = 0.23

Bernstein et al 1994 [305]

To determine whether young women who regularly participate in PA during their reproductive years had a reduced risk of BC.

• n = 1,090 (545 cases; 545 controls)

PA assessment: Questionnaire for overall participation in PA after menarche (h/wk), PA within 10 years after menarche (h/wk), each divided into 5 groups:

Adjusted OR (95% CI) by PA after menarche

PA may substantially reduce a women's lifetime risk of BC.

• Sex: Women

   • G1 = 1.00 (referent)

USA

• Age: ≤ 40 yr

   • G2 = 0.95 (0.64-1.41)

• Characteristics: White women matched for age and parity

Case control

   • G3 = 0.65 (0.45-0.96)

D & B score = 15

G1 = none

   • G4 = 0.80 (0.54-1.17)

G2 = 0.1-0.7

G3 = 0.8-1.6

   • G5 = 0.42 (0.27-0.64)

G4 = 1.7-3.7

G5 = ≥ 3.8

Trend p = 0.0001

Logistic regression

Adjusted OR (95% CI) by PA within 10 years after menarche

   • None = 1.00 (referent)

   • 0.1-1.2 = 0.93 (0.63-1.38)

   • 1.3-2.9 = 0.78 (0.52-1.19)

   • 3.0-5.5 = 0.69 (0.45-1.05)

   • ≥5.6 = 0.70 (0.47-1.06)

Trend p = 0.027

Adjusted OR (95% CI) by PA after menarche, nulliparous women

   • G1 = 1.00 (referent)

   • G2 = 0.81 (0.42-1.57)

   • G3 = 0.65 (0.35-1.21)

   • G4 = 0.94 (0.53-1.67)

   • G5 = 0.73 (0.38-1.41)

Trend p = 0.43

Adjusted OR (95% CI) by PA after menarche, parous women

   • G1 = 1.00 (referent)

   • G2 = 1.06 (0.65-1.74)

   • G3 = 0.65 (0.40-1.06)

   • G4 = 0.70 (0.42-1.18)

   • G5 = 0.38 (0.16-0.50)

Trend p < 0.0001

Bernstein et al 2005 [306]

To examine the relationship between BC risk and lifetime and time- or age-specific measures of LTPA among white and black women.

• n = 9,187 (4,538 cases; 4,649 control)

Cases: histologically confirmed cases of BC

4,538 cases of BC

This study supports an inverse association between PA and BC among black women and among white women.

• Sex: Women

• Age: 35-64

Multivariate adjusted OR (95%CI) annual MET h/wk, White participants

USA

• Ethnicity: White (including Hispanics) or Black

Controls: random-digit dialing methods

• Characteristics: Case Group: histologically confirmed cases of invasive BC

Case control

• Control Group: healthy

   • G1 = 1.00 (referent)

PA assessment:

   • G2 = 0.84 (0.71-0.99)

D & B score = 13

Questionnaire for lifetime PA (MET h/wk), divided into 5 groups

   • G3 = 0.89 (0.75-1.04)

G1 = Inactive

   • G4 = 0.82 (0.69-0.97)

The relationship appears to be similar between black and white women.

G2 = ≤ 2.2

   • G5 = 0.81 (0.69-0.96)

G3 = 2.3-6.6

Trend p = 0.09

G4 = 6.7-15.1

G5 = ≥ 15.2

Unconditional logistic regression modeling

Multivariate adjusted OR (95%CI) annual MET h/wk, Black participants

   • G1 = 1.00 (referent)

   • G2 = 1.11 (0.91-1.35)

   • G3 = 0.83 (0.67-1.03)

   • G4 = 0.79 (0.63-0.99)

   • G5 = 0.77 (0.62-0.95)

Trend p = 0.003

Carpenter et al. 1999 [307]

To examine whether lifetime exercise activity is related to BC risk in post-menopausal women.

• n = 2,027 (1,123 case; 904 control)

Cases: diagnosed with primary invasive or in situ BC

1,123 cases of BC

Strenuous exercise appears to reduce BC risk among post-menopausal women who do not gain sizable amounts of weight during adulthood.

• Sex: Women

Multivariate adjusted OR (95%CI)

USA

• Age: 55-64 yr

• Ethnicity: White (including Hispanic)

Controls: individually matched to each case patient based on birth date and race

   • G1 = 1.00 (referent)

Case control

   • G2 = 0.88 (0.72-1.07)

D & B score = 15

   • G3 = 0.55 (0.37-0.83)

PA assessment: Questionnaire for lifetime PA (MET hr/wk), divided into 3 groups

Trend p = 0.01

G1 = no activity

G2 = 0.1-17.59

G3 = ≥17.6

Conditional logistic regression

Carpenter et al 2003 [308]

To examine the effects of obesity and lifetime exercise patterns on post-menopausal BC risk according to family history.

• n = 3,511 (cases n = 1,883, controls) n = 1,628

PA assessment: Interview for the following PA variables

1,883 cases of BC

Exercise independent of body size seemed to exert a protective effect primarily among women with a negative family history.

• Sex: Women

Adjusted OR (95% CI) by lifetime exercise between menarche and reference date (MET hr/wk)

USA

• Age: 55-72

• Characteristics: Postmenopausal Women

Lifetime exercise between menarche and reference date (MET hr/wk)

Case control

D & B score = 15

G1 = 0

   • G1 = 1.00 (referent)

G2 = 0.1-3.74

   • G2 = 0.85 (0.71-1.03)

G3 = 3.75-8.74

G4 = 8.75-17.59

   • G3 = 0.87 (0.69-1.10)

G5 = ≥17.60

   • G4 = 1.02 (0.79-1.30)

Average exercise activity in 10 years prior to reference date (MET hr/wk)

   • G5 = 0.66 (0.48-0.90)

G1 = 0

Trend p = 0.07

G2 = 0.1-6.9

G3 = 7.0-13.9

Adjusted OR (95% CI) by average exercise activity in 10 years prior to reference date (MET hr/wk)

G4 = 14.0-24.4

G5 = ≥24.5

   • G1 = 1.00 (referent)

   • G2 = 0.93 (0.71-1.22)

   • G3 = 0.92 (0.70-1.19)

   • G4 = 0.86 (0.65-1.11)

   • G5 = 0.75 (0.55-1.02)

Trend p = 0.05

Chang et al 2006 [309]

To address the independent and combined effects of energy intake, BMI, and PA on BC incidence in women.

• n = 27,541

9.3 year follow-up (median 4.9 yr)

764 women developed BC

The study suggests that energy intake, BMI and physical inactivity are each independently and positively associated with BC risk.

• Sex: Women

USA

• Age: 55-74

• Characteristics: no history of any cancer (nonmelanoma skin cancer patients were included in the trial)

PA assessment: Questionnaire for vigorous PA (h/wk), divided into 6 groups

Multivariate adjusted RR (95%CI)

Prospective cohort

• Prostate, Lung, colorectal, and Ovarian Cancer Screening Trial

G1 = 0

G2 = <1

   • G1 = 1.00 (referent)

D & B score = 13

G3 = 1

   • G2 = 0.89 (0.69-1.15)

G4 = 2

   • G3 = 0.96 (0.73-1.26)

G5 = 3

   • G4 = 0.90 (0.70-1.16)

G6 = ≥4

   • G5 = 1.02 (0.79-1.30)

   • G6 = 0.78 (0.61-0.99)

Cox proportional HR

   Trend p = 0.153

Colditz et al 2003 [310]

To evaluate the relationship between PA and risk of pre-menopausal BC by type of activity and within subgroups of adiposity and oral contraceptive use.

• n = 110,468

PA assessment: Self report on 8 activities (walking or hiking, jogging (>10 min mile), running, Biking, racquet sports, lap swimming, calisthenics/aerobics other aerobic activities) to calculate MET scores (MET hrwk), divided into 5 groups:

Total cases diagnosed n = 849

These data among pre-menopausal women suggest that there is no overall association between PA and risk of BC. The findings also suggest that the effect of PA could be substantially modified by the underlying degree of adiposity.

• Sex: Women

USA

• Age: 25-42

• Characteristics: pre-menopausal, no history of cancer other than nonmelanoma skin cancer

Multivariate adjusted RR (95% CI)

Prospective cohort

   • G1 = 1.00 (referent)

D & B score = 12

   • G2 = 1.05 (0.82-1.34)

   • G3 = 0.96 (0.75-1.23)

G1 = <3

   • G4 = 1.05 (0.80-1.37)

G2 = 3-8.9

G3 = 9-17.9

   • G5 = 1.07 (0.84-1.36)

G4 = 18-26.9

G5 = ≥27

Trend p = 0.69

Cox proportional HR

Coogan et al 1997 [311]

To evaluate the effect of OPA on BC risk.

• n = 11,646 (4,863 cases and 6,783 controls).

PA assessment: Telephone interview to estimate OPA, divided into tertiles:

4,863 cases of BC

There was evidence of a graded inverse relationship between the intensity of work related activity and the incidence of BC.

OR (95% CI)

USA

• Sex: Women

T1 = Sedentary

   • T1 = 1.00 (referent)

• Age: <74 yr

T2 = Medium activity jobs

   • T2 = 0.86 (0.77-0.97)

Case control

T3 = Heavy jobs

   • T3 = 0.82 (0.63-1.08)

D & B score = 14

Logistic regression models

Coogan and Aschengrau 1999 [312]

To evaluate the effect of OPA on BC risk.

• n = 903 (233 case; 670 control)

PA assessment: Telephone interview to estimate OPA, divided into tertiles:

233 cases of BC

There was no evidence that holding a job of medium/heavy activity reduced BC.

• Sex: Women

OR (95%CI)

• Age: <50 - 80+

T1 = Exclusively sedentary

   • T1 = 1.00 (referent)

USA

• Ethnicity: White, Black or Other

T2 = Exclusively light

   • T2 = 1.20 (0.70-1.90)

• Characteristics: must have worked outside the home. Cases: All incident cases of BC reported to the Massachusetts Cancer Registry from 1983 to 1986 were eligible

T3 = Exclusively medium or heavy

   • T3 = 0.90 (0.40-1.90)

The study was limited by OPA misclassification and by the lack of information on LTPA.

Case control

Trend p = 0.63

D & B score = 14

Miettinen's test-based method and Fisher's exact method

Dallal et al 2007 [313]

To examine the relationship between LTPA and invasive and in situ BC among women.

• n = 110,599

6.6 yr follow-up

2,649 cases of invasive

The results support a protective role of strenuous long-term exercise activity against invasive and in situ BC and suggest differing effects by hormone receptor status.

• Sex: Women

BC

USA

• Age: 20-79

PA assessment: Self-reported participation in moderate and strenuous activities to estimate annual strenuous physical activity (hr/wk), divided into quintiles

593 cases of in situ BC

• Ethnicity: White, Black, Hispanic, Asian, American Indian or other

Prospective cohort

• Characteristics: California resident at baseline and no history of BC

Multivariate adjusted RR (95% CI) for invasive BC

D & B score = 12

• California Teachers Study cohort

   • Q1 = 1.00 (referent)

   • Q2 = 0.93 (0.85-1.02)

Q1 = 0.00-0.50

Q2 = 0.51-2.00

   • Q3 = 0.88 (0.78-0.99)

Q3 = 2.01-3.50

Q4 = 3.51-5.00

   • Q4 = 1.02 (0.88-1.18)

Q5 = >5

   • Q5 = 0.80 (0.69-0.94)

Cox proportional HR

Trend p = 0.02

Multivariate adjusted RR (95% CI) for in situ BC

   • Q1 = 1.00 (referent)

   • Q2 = 0.96 (0.79-1.17)

   • Q3 = 0.86 (0.67-1.11)

   • Q4 = 0.95 (0.70-1.30)

   • Q5 = 0.69 (0.48-0.98)

Trend p = 0.40

Dirx et al 2001 [314]

To evaluate the relationship between PA and BC risk with specific emphasis on interaction with other aspects of energy balance.

• n = 62,537

7.3 yr follow-up

1,208 cases of incident BC

The current study supports the hypotheses that PA is related inversely to BC risk in postmenopausal women.

• Sex: Women

Netherlands

• Age: 55-69

PA assessment: Questionnaire for total recreational PA (min/day), divided into quartiles

• Characteristics: healthy, postmenopausal

Multivariate adjusted RR (95% CI)

Case study

   • Q1 = 1.00 (referent)

D & B score = 11

Q1 = <30

   • Q2 = 0.84 (0.67-1.07)

Q2 = 30-60

Q3 = 61-90

   • Q3 = 0.78 (0.60-1.00)

Q4 = >90

   • Q4 = 0.76 (0.58-0.99)

Exponentially distributed failure time regression models

Trend p = 0.003

Dorn et al 2003 [315]

To examine the associations between LTPA and OPA across the lifespan and pre- and post-menopausal BC.

• n = 1,550 (740 case; 810 control)

Cases: women diagnosed and histologically confirmed with BC

740 cases of BC

The study supports the hypothesis that strenuous LTPA is associated with a reduced risk of BC risk in both pre- and post menopausal women

USA

• Sex: Women

Multivariate adjusted OR (95%CI), pre- menopausal

• Age: 40-85

Case control

• Characteristics: Case Group -- histologically confirmed incidence of BC. Control Group -- healthy

Controls: randomly selected and frequency matched on age and county with the cases.

   • G1 = 1.00 (referent)

D & B score = 13

   • G2 = 0.94 (0.64-1.38)

   • G3 = 0.73 (0.44-1.22)

PA assessment: Questionnaire for lifetime strenuous PA (hr/yr)

   • G4 = 1.07 (0.57-2.02)

Trend p = 0.82

G1 = 0

G2 = 1-273

G3 = 274-545

Multivariate adjusted OR (95%CI), post-menopausal

G4 = >546

Logistic regression

   • G1 = 1.00 (referent)

   • G2 = 0.85 (0.61-1.19)

   • G3 = 0.73 (0.45-1.17)

   • G4 = 0.78 (0.47-1.29)

Trend p = 0.19

Drake 2001 [316]

To evaluate PA as a predictor of BC and describe BC risk factors in this sample.

• n = 4,520

PA assessment: Self-report of type, intensity, duration and frequency of walking, jogging, biking, stationary biking, swimming, dancing, racket sports, stretching, participating in other exercise, calisthenics, weight-lifting and treadmill exercises, divided into groups

150 incident cases of breast cancer

Increased frequency of a specific PA (jogging) was found to have an important protective role in BC incidence.

• Sex: Women

USA

• Age: 21-86

• Characteristics: no diagnosis of BC at entry

OR (95% CI) for BC and PA

Prospective cohort

D & B score = 11

• Aerobic Center Longitudinal Study

Activity type

   • G1 = 1.32

   • G2 = 1.08

   • G3 = 1.35

Trend p = 0.05

G1 = Aerobic (job, bike, aerobic dance)

G2 = Moderate (golf, walk)

G3 = Weight training

Chi-square

Friedenreich et al 2001 [317]

To examine the type and dose of PA and the time periods in life when PA may be specifically associated with BC risk.

• n = 2,470 (1,233 case; 1,237 control)

Cases: in situ and invasive cases of BC from 1995-1997

1,233 cases of BC

This study provides evidence that lifetime PA reduces risk of post-menopausal BC.

Canada

• Sex: Women

OR (95%CI), pre-menopausal

Case Control

• Age: ≤ 80

Controls: matched to cases on age and place of residence

   • Q1 = 1.00 (referent)

• Characteristics: Case Group - Alberta residents, English speaking, capable of completing an in-person interview. Control Group -- no history of cancer diagnoses excluding nonmelanoma skin cancer

   • Q2 = 1.15 (0.78- 1.70)

D & B score = 13

PA assessment: Questionnaire for lifetime PA (MET hr/wk/yr), divided into quartiles by menopausal status

   • Q3 = 1.15 (0.78- 1.69)

Pre-menopausal

   • Q4 = 1.07 (0.72- 1.61)

Q1 = <86.6

Trend p = 0.50

Q2 = 86.6-108.3

OR (95%CI), post- menopausal

Q3 = 108.3-134.9

   • Q1 = 1.00 (referent)

Q4 = ≥ 134.9

   • Q2 = 0.73 (0.55- 0.98)

Post-menopausal

   • Q3 = 0.75 (0.56- 1.00)

Q1 = <104.8

   • Q4 = 0.70 (0.52- 0.94)

Q2 = 104.8-128.1

Trend p = 0.003

Q3 = 128.1-160.9

Q4 = ≥ 160.9

Logistic regression

Friedenreich et al 2001 [318]

To examine the influence of frequency, duration, and intensity of PA on risk of BC and to compare BC risks associated with self-reported versus assigned intensity of PA.

• n = 2,470 (1,233 case; 1,237 control)

Cases: in situ and invasive cases of BC

1,233 cases of BC

This study found that moderate- intensity activities were the major contributors to the decrease in BC risk found in this study.

Canada

• Sex: Women

Multivariate adjusted OR (95% CI), pre- menopausal

Case control

• Age: ≤ 80

Controls: matched to cases on age and place of residence

   • Q1 = 1.00 (referent)

D & B score = 13

• Characteristics: Case Group -- resident of Alberta, English speaking and able to complete an in-person interview. Control Group -- free of any cancer diagnosis excluding nonmelanoma skin cancer

PA assessment: Questionnaire for lifetime PA questionnaire (MET hr/wk/yr), divided into quartiles

   • Q2 = 1.19 (0.80- 1.76)

Q1 = <28.8

   • Q3 = 1.33 (0.90- 1.96)

Q2 = 28.8-35.4

   • Q4 = 1.07 (0.71- 1.62)

Q3 = 35.4-42.7

Trend p = 0.52

Q4 = ≥ 42.7

Multivariate adjusted OR (95% CI), post- Menopausal

Logistic regression

   • Q1 = 1.00 (referent)

   • Q2 = 0.85 (0.64- 1.14)

   • Q3 = 0.83 (0.62- 1.10)

   • Q4 = 0.69 (0.51- 0.93)

Trend p = 0.006

Friedenreich and Rohan 1995 [319]

To describe the association between LTPA and BC.

• n = 902 (451 case; 451 control)

Cases: first diagnosis of BC in 1982 and 1984

Adjusted OR (95%CI), pre-menopausal

This study found some evidence (of borderline statistical significance) that recreational PA is associated with decreased risk of BC.

Australia

• Sex: Women

   • Q1 = 1.00 (referent)

Case control

• Age: 20-74 yr

Controls: Randomly selected from the electoral roll, matched on date of birth to each case

   • Q2 = 0.77 (0.36- 1.65)

D & B score = 13

• Characteristics: Australian women

PA assessment: Self reported PA (kcal/wk), divided into quartiles

   • Q3 = 0.48 (0.22- 1.03)

Q1 = 0

   • Q4 = 0.60 (0.30- 1.17)

Q2 = 1-2,000

Trend p = 0.09

Q3 = 2000-4000

Adjusted OR (95%CI), post-menopausal

Q4 = >4000

   • Q1 = 1.00 (referent)

Logistic regression models

   • Q2 = 0.74 (0.46- 1.18)

   • Q3 = 0.88 (0.53- 1.48)

   • Q4 = 0.73 (0.44- 1.20)

Trend p = 0.32

Gammon et al 1998 [320]

To examine the association between LTPA and BC among young women.

• n = 3,173 (1,668 case; 1,505 control)

Cases: women diagnosed with BC between 1990-1992

1,668 cases of BC

The study's data does not support the hypothesis of a reduced risk of BC among young women with increased recreational PA in adolescence, young adulthood or during the year prior to the interview, or with the average PA over the three time periods

USA

• Sex: Women

Multivariate adjusted OR (95%CI)

Case control

• Age: <45

Controls: were matched to cases by age group and geographic center

   • Q1 = 1.00 (referent)

D & B score = 13

• Characteristics: Case Group -- diagnosed with invasive or in situ BC. Control Group - healthy

PA assessment: Questionnnaire for recreational PA, for ages 12- 13 yr, age 20 yr and 1 year prior to the interview. Divided into quartiles MET score

   • Q2 = 0.79 (0.63- 0.98)

Q1 = 1.62-18.07

   • Q3 = 0.98 (0.79- 1.22)

Q2 = 18.08-30.00

   • Q4 = 1.01 (0.81- 1.25)

Q3 = 30.01-42.95

Trend p = 0.42

Q4 = 42.96-98.00

Logistic regression

Gilliland et al 2001 [321]

To investigate the relationship of PA with BC risk in Hispanic and non- Hispanic White women

• n = 1,556 (712 case; 844 control)

Cases: diagnosed with BC between 1992-1994

712 cases of BC

Hispanic and non- Hispanic women with high PA during non-OPA were at substantially reduced risk of BC.

USA

• Sex: Women

Case control

• Age: between 35-74 at diagnosis

Controls: matched on ethnicity, age and seven health planning districts

Adjusted OR (95%CI), pre-menopausal Hispanic

D & B score = 13

• Ethnicity: Hispanic and non-Hispanic White

PA assessment: Self-reported non-OPA (MET hr/wk score)

   • G1 = 1.00 (referent)

• Characteristics: Case Group -- diagnosed with in situ or invasive BC and residents of New Mexico at time of diagnosis. Control Group -- healthy

G1 = <25

   • G2 = 1.17 (0.53- 2.55)

G2 = 25-50

   • G3 = 0.49 (0.22- 1.07)

G3 = 50-80

   • G4 = 0.29 (0.12- 0.72)

G4 = ≥ 80

Trend p < 0.001

Logistic regression

Adjusted OR (95%CI), pre-menopausal non- Hispanic

   • G1 = 1.00 (referent)

   • G2 = 1.35 (0.64- 2.85)

   • G3 = 1.44 (0.67- 3.10)

   • G4 = 1.13 (0.49- 2.61)

Trend p = 0.741

Adjusted OR (95%CI), post-menopausal Hispanic

   • G1 = 1.00 (referent)

   • G2 = 0.74 (0.40- 1.36)

   • G3 = 0.37 (0.18- 0.75)

   • G4 = 0.38 (0.18- 0.77)

Trend p = 0.002

Adjusted OR (95%CI), post-menopausal non- Hispanic

   • G1 = 1.00 (referent)

   • G2 = 0.45 (0.26- 0.78)

   • G3 = 0.49 (0.28- 0.86)

   • G4 = 0.45 (0.24- 0.85)

Trend p = 0.019

Hsing et al 1998 [322]

To evaluate the role of selected demographic, lifestyle, and anthropometric factors in the risk for male BC.

• n = 690 (178 case; 512 control)

Cases: selected from 18,733 decedents included in the 1986 NMFS conducted by the US

178 cases of BC

This study suggests that obesity increases the risk of male BC, possibly through hormonal mechanisms, while dietary factors, PA and SES indicators also deserve further investigation.

USA

• Sex: Men

National Center for Health Statistics (NCHS)

Adjusted OR (95%CI)

• Age: 25-74

   • G1 = 1.00 (referent)

Case control

• Ethnicity: Black and White

   • G2 = 0.60 (0.30- 1.10)

D & B score = 12

• Characteristics: Case Group -- deceased. Control Group -- dying (or deceased) of causes other than BC

Controls: selected from male decedents dying of causes other than BC

   • G3 = 1.30 (0.80- 2.00)

PA assessment: Questionnaire (frequency and intensity), divided into groups

G1 = Regular

G2 = Irregular

G3 = Hardly any

Logistic regression analysis

Hu et al 1997 [323]

To study breast cancer focusing on breast-feeing, body weight, and PA as well as reproductive histories on pre- and post- menopausal Japanese women.

• n = 526 (157 case; 369 control)

Cases: Histologically confirmed cases of BC from 1989-1993.

157 cases of BC

Reduced risk of pre- menopausal BC was associated with high EE in PA during teenage years, although the trend was not statistically significant.

Japan

• Sex: Women

Unadjusted RR (95%CI), pre-menopausal

Case control

• Age: 26-75

   • G1 = 1.00 (referent)

D & B score = 13

• Characteristics: Case Group -- histologically confirmed cases of BC and resident of Gifu prefecture at time of diagnosis. Control Group -- no breast disease or hormone- related (ovarian, endometrial and thyroid) cancers

Controls: individuals who had the screening test for BC during the same period

   • G2 = 0.74 (0.38- 1.44)

PA assessment: Questionnaire for TPA (kcal/wk), divided into groups

   • G3 = 1.01 (0.54- 1.87)

G1 = 0

Trend p = 0.876

G2 = 1-649

Unadjusted RR (95%CI), post-menopausal:

G3 = ≥ 650

   • G1 = 1.00 (referent)

   • G2 = 1.53 (0.69- 3.54)

Logistic regression models

   • G3 = 1.39 (0.61- 3.13)

John et al 2003 [324]

To examine BC risk in relation to lifetime histories of MPA and VPA including LTPA, transportation household and outdoor chores, and OPA in a multiethnic population.

• n = 2,870 (1,277 case; 1,593 control)

Cases: diagnosed between 1995-1998

1,277 cases of BC

This study supports previous reports of a reduced risk of BC in physically active women.

USA

• Sex: Women

Multivariate adjusted OR (95%CI), pre- menopausal Latinas

Case control

• Age: 35-79

   • G1 = 1.00 (referent)

D & B score = 12

• Ethnicity: Latina, African-American and White

Controls: randomly selected according race/ethnicity and age distribution of cases

   • G2 = 0.84 (0.49- 1.45)

PA assessment: In-person interview for lifetime PA (hr/wk), divided into groups

   • G3 = 0.73 (0.42- 1.28)

Pre-menopausal

Multivariate adjusted OR (95%CI), pre- menopausal African Americans

G1 = <9.1

   • G1 = 1.00 (referent)

G2 = 9.1-20.7

   • G2 = 1.00 (0.55- 1.84)

G3 = ≥ 20.7

   • G3 = 0.68 (0.35- 1.34)

Post-menopausal

G1 = <9.6

G2 = 9.6-21.6

G3 = ≥ 21.7

Logistic regression modeling

Multivariate adjusted OR (95%CI), pre- menopausal Whites

   • G1 = 1.00 (referent)

   • G2 = 0.82 (0.42- 1.58)

   • G3 = 0.76 (0.36- 1.61)

Multivariate adjusted OR (95%CI), post- menopausal Latinas

   • G1 = 1.00 (referent)

   • G2 = 0.82 (0.55- 1.24)

   • G3 = 0.81 (0.54- 1.22)

Multivariate adjusted OR (95%CI), post- menopausal African Americans

   • G1 = 1.00 (referent)

   • G2 = 0.78 (0.52- 1.17)

   • G3 = 0.71 (0.47- 1.07)

Multivariate adjusted OR (95%CI), post- menopausal Whites

   • G1 = 1.00 (referent)

   • G2 = 0.94 (0.64- 1.37)

   • G3 = 0.91 (0.60- 1.41)

Kruk 2007 [325]

To examine the association between all types of PA and BC risk among Polish women.

• n = 590 (268 case; 322

control)

PA assessment: Questionnaire for lifetime PA (MET hr/wk/yr), divided into groups

268 cases of BC

The results of this study provide evidence of an inverse association between PA and the risk of BC.

Poland

• Sex: Women

G1 = <110

Multivariate adjusted OR (95%CI), pre- menopausal

Case control

• Age: 35-75 yr

G2 = 110-150

   • G1 = 1.00 (referent)

D & B score = 13

• Characteristics: Polish women. Cases: identified from the Szczecin Regional Cancer Registry. Controls: matched on age and place of residence

G3 = >150

   • G2 = 0.45 (0.14- 1.44)

Logistic regression analysis

   • G3 = 0.44 (0.14- 1.37)

Trend p = 0.42

Multivariate adjusted OR (95%CI), post- menopausal

   • G1 = 1.00 (referent)

   • G2 = 0.60 (0.33- 1.09)

   • G3 = 0.31 (0.21- 0.70)

Trend p = 0.002

Kruk 2007 [326]

To examine the relationship between LTPA and BC risk.

• n = 822 (cases n = 257, control n = 565

PA assessment: Questionnaire for LTPA (METs), divided into groups

Adjusted OR (95% CI)

The findings provide further support to the hypothesis that increased LTPA throughout life is associated with a decreased risk of BC.

Poland

• Sex: Women

G1 = Low

   • G1 = 1.00 (referent)

Case control

• Age: 35-93 yr

G2 = Medium

   • G2 = 0.57 (0.36- 0.89)

D & B score = 13

G3 = High

   • G3 = 0.22 (0.14- 0.35)

Trend p < 0.0001

Lahmann et al 2007 [327]

To examine the association of PA with pre- and post-menopausal BC risk.

• n = 218,169

Baseline and 6.4 year follow-up

3,423 cases of BC

Increasing PA reduces BC risk.

Europe (9 countries)

• Sex: Women

Multivariate adjusted HR (95% CI) by TPA, pre- menopausal

Prospective cohort

• Age: 20-80

   • Q1 = 1.00 (referent)

D & B score = 12

• The European Prospective Investigation into Cancer and nutrition study

PA assessment: Interviews and questionnaire for TPA and recreational PA, each divided into quartiles

   • Q2 = 1.02 (0.84- 1.24)

TPA Index

   • Q3 = 0.84 (0.68- 1.04)

Q1 = Inactive

   • Q4 = 1.02 (0.77- 1.36)

Q2 = Moderately inactive

Trend p = 0.267

Q3 = Moderately active

Multivariate adjusted HR (95% CI) by TPA, Post- menopausal

Q4 = Active

   • Q1 = 1.00 (referent)

Recreational PA (MET hr/wk)

   • Q2 = 0.89 (0.79- 1.00)

Q1 = <14

   • Q3 = 0.84 (0.74- 0.96)

Q2 = 14-24

   • Q4 = 0.92 (0.76- 1.12)

Q3 = 25-42

Trend p = 0.06

Q4 = >42

Multivariate adjusted HR (95% CI) by recreational PA, pre-menopausal

Cox proportional index

   • Q1 = 1.00 (referent)

   • Q2 = 0.91 (0.75- 1.10)

   • Q3 = 0.95 (0.78- 1.14)

   • Q4 = 0.94 (0.76- 1.15)

Trend p = 0.580

Multivariate adjusted HR (95% CI) by recreational PA, post-menopausal

   • Q1 = 1.00 (referent)

   • Q2 = 1.05 (0.94- 1.17)

   • Q3 = 0.92 (0.83- 1.03)

   • Q4 = 0.96 (0.85- 1.08)

Trend p = 0.176

Lee et al 2001 [328]

To examine the association between PA and BC risk.

• n = 39,322

Baseline and 4 year follow- up

411 cases of BC

The data suggest that PA during middle age and older is not uniformly associated with decreased BC risk. Among post- menopausal women only, higher levels of PA may decrease the risk of BC.

USA

• Sex: Women

• Age: ≥ 45 yr

Multivariate adjusted RR (95% CI) by PA, all women

Prospective cohort

• Characteristics: Healthy women

PA assessment: Questionnaire

   • Q1 = 1.00 (referent)

D & B score = 12

Women's Health Study

PA (kJ/wk), divided into quartiles

   • Q2 = 1.04 (0.77- 1.40)

Q1 = <840

   • Q3 = 0.86 (0.64- 1.17)

Q2 = 840-2519

   • Q4 = 0.80 (0.58- 1.12)

Q3 = 2520-6299

Trend p = 0.11

Q4 = ≥ 6300

Multivariate adjusted RR (95% CI) by PA, post- menopausal only

VPA (kJ/wk), divided into quintiles

   • Q1 = 1.00 (referent)

Q1 = none

   • Q2 = 0.97 (0.68- 1.39)

Q2 = 1-839

   • Q3 = 0.78 (0.54- 1.12)

Q3 = 840-2099

   • Q4 = 0.67 (0.44- 1.02)

Q4 = 2100-4199

Trend p = 0.03

Q5 = ≥ 4200

Multivariate adjusted RR (95% CI) by VPA, all women

Proportional hazard regression

   • Q1 = 1.00 (referent)

   • Q2 = 1.02 (0.70- 1.48)

   • Q3 = 1.11 (0.78- 1.58)

   • Q4 = 0.97 (0.66- 1.44)

   • Q5 = 0.98 (0.69- 1.40)

Trend p = 0.98

Multivariate adjusted RR (95% CI) by VPA, post- menopausal only

   • Q1 = 1.00 (referent)

   • Q2 = 0.93 (0.57- 1.50)

   • Q3 = 0.91 (0.57- 1.47)

   • Q4 = 0.93 (0.57- 1.50)

   • Q5 = 0.76 (0.47- 1.24)

Trend p = 0.29

Magnusson et al 2005 [329]

To report the relationship between pre-menopausal BC, body fatness at age 10 years and in adulthood, and sports participation during puberty, late adolescence and early adulthood from three related case-control studies.

• n = 3,108 (1,560 cases; 1,548 controls)

PA assessment: Interview for sports participation (h/wk in the following age categories (12-14 yr, 16-18 yr, 20-30 yr, 12-30 yr, around age of diagnosis)

Adjusted RR (95% CI), 12-14 yr

An inverse association between body fatness but not PA at a young age and the risk of BC in pre-menopausal women.

UK

• Sex: Women

   • G1 = 1.00 (referent)

Case control

• Age: Study 1 = 36 yr, study 2 = 36-45 yr, study 3 = 46-54 yr

   • G2 = 1.04 (0.93- 1.17)

D & B score = 13

• Characteristics: White women with no previous malignancy, mental handicap or illness

Sports participation (h/wk)

   • G3 = 1.03 (0.93- 1.14)

G1 = 0-1

Trend p = 0.95

G2 = 2-3

Adjusted RR (95% CI), 16-18 yr

G3 = ≥ 4

   • G1 = 1.00 (referent)

   • G2 = 0.95 (0.83- 1.09)

   • G3 = 0.89 (0.79- 1.02)

Trend p = 0.20

Adjusted RR (95% CI), 20-30 yr

   • G1 = 1.00 (referent)

   • G2 = 0.90 (0.76- 1.08)

   • G3 = 1.01 (0.81- 1.26)

Trend p = 0.73

Adjusted RR (95% CI), 12-30 yr

   • G1 = 1.00 (referent)

   • G2 = 0.99 (0.89- 1.11)

   • G3 = 1.01 (0.88- 1.16)

Trend p = 0.94

Adjusted RR (95% CI), around age of diagnosis

   • G1 = 1.00 (referent)

   • G2 = 0.84 (0.71- 1.00)

   • G3 = 1.06 (0.86- 1.32)

Trend p = 0.82

Malin et al 2005 [330]

To evaluate a pattern of behavioral exposures indicating positive energy balance would be associated with increased BC risk.

• n = 3,015 (1,459 cases; 1,556 control)

PA assessment: Questionnaire for PA (MET hr/d/yr), divided into groups

OR (95% CI)

The study suggests that promotion of behavioral patterns that optimize energy balance maybe a viable option for BC prevention.

China

• Sex: Women

G1 = 0

   • G1 = 1.86 (1.44- 2.41)

Case control

• Age: Mean ~47 yr

G2 = 0.1-2.92

   • G2 = 1.33 (0.96- 1.83)

D & B score = 12

• Characteristics: Residents of urban Shanghai

G3 = >2.92

   • G3 = 1.00 (referent)

Shanghai Breast Cancer Study

Margolis et al 2005 [331]

To study the association between PA and incident invasive BC.

• n = 99,504

Baseline and 9.1 year follow-up

1,166 cases of BC

No evidence of a protective effect of PA on BC risk was found.

Norway/Sweden

• Sex: Women

Prospective cohort

• Age: 30-49 (mean 41 yr)

PA assessment: Questionnaire for PA using a 5 point scale and for competitive PA (years of participation), each divided into groups

Multivariate adjusted RR (95% CI) by PA level, at enrollment

D & B score = 13

• The Norwegian- Swedish Women's Lifestyle and Health Study

PA level (5 point scale)

   • G1 = 1.00 (referent)

G1 = None

   • G2 = 1.35 (0.96- 1.90)

G2 = Low

   • G3 = 1.26 (0.91- 1.74)

G3 = Moderate

   • G4 = 1.19 (0.85- 1.67)

G4 = High

   • G5 = 1.24 (0.85- 1.82)

G5 = Vigorous

Trend p = 0.85

Competitive PA (years)

Multivariate adjusted RR (95% CI) by PA level, at age 30

G1 = None

   • G1 = 1.00 (referent)

G2 = 1-4

   • G2 = 1.03 (0.64- 1.66)

G3 ≥ 5

   • G3 = 1.16 (0.74- 1.81)

   • G4 = 1.06 (0.67- 1.68)

   •G5 = 1.20 (0.77- 1.95)

Trend p = 0.60

Multivariate adjusted RR (95% CI) by PA level, at age 14

   • G1 = 1.00 (referent)

   • G2 = 0.93 (0.62- 1.39)

   • G3 = 0.94 (0.65- 1.35)

   • G4 = 1.07 (0.73- 1.55)

   • G5 = 1.05 (0.72- 1.54)

Trend p = 0.14

Multivariate adjusted RR (95% CI) by years of competitive PA

   • G1 = 1.00 (referent)

   • G2 = 1.21 (0.95- 1.54)

   • G3 = 0.95 (0.75- 1.19)

Trend p = 0.96

McTiernan et al 1996 [332]

To investigate the relationship between LTPA and BC.

• n = 1,029 (cases n = 537, controls n = 492)

PA assessment: Questionnaire (Minnesota LTPA Questionnaire) for LTPA (hr/wk), divided into groups

Adjusted OR (95% CI) by LTPA during adulthood, all ages and menopausal status

The results indicate a weak negative association between PA and risk of BC in middle-aged women.

USA

• Sex: Women

G1 = None

   • G1 = 1.00 (referent)

Case control

• Age: 50-64

G2 = 0.1-1.5

   • G2 = 1.1 (0.7-1.6)

D & B score = 13

G3 = 1.6-2.5

   • G3 = 0.7 (0.4-1.1)

G4 = 2.6-3.5

   • G4 = 0.7 (0.4-1.1)

G5 = 3.6-5.0

   • G5 = 0.6 (0.4-0.9)

G6 = >5

   • G6 = 1.1 (0.7-1.6)

Calculated categories of EE (total time x intensity code)

Trend p = 0.29

Adjusted OR (95% CI) by LTPA during adulthood, aged ≥ 55 yr, post-menopausal only

   • G1 = 1.00 (referent)

   • G2 = 0.8 (0.5-1.3)

   • G3 = 0.5 (0.3-0.9)

   • G4 = 0.6 (0.4-1.1)

   • G5 = 0.4 (0.2-0.8)

   • G6 = 0.8 (0.5-1.3)

Trend p = 0.03

Adjusted OR (95% CI) by category of total EE in adulthood, all ages and menopausal status

   • G1 = 1.00 (referent)

   • G2 = 1.2 (0.8-2.0)

   • G3 = 0.9 (0.6-1.3)

   • G4 = 0.6 (0.4-0.9)

   • G5 = 0.9 (0.6-1.5)

   • G6 = 0.9 (0.6-1.4)

Trend p = 0.25

Adjusted OR (95% CI) by category of total EE in adulthood, aged ≥ 55 yr, post-menopausal only

   • G1 = 1.00 (referent)

   • G2 = 0.8 (0.4-1.4)

   • G3 = 0.7 (0.4-1.2)

   • G4 = 0.5 (0.3-0.8)

   • G5 = 0.8 (0.5-1.3)

   • G6 = 0.6 (0.4-1.0)

Trend p = 0.009

McTiernan et al 2003 [333]

To examine the association between current and past LTPA and incidence of BC in post-menopausal women.

• n = 74,171

Baseline and mean follow-up of 4.7 years

1,780 cases of BC

Increased PA is associated with reduced risk for BC in post-menopausal women.

USA

• Sex: Women

PA assessment: Questionnaire for TPA (MET hr/wk), moderate or strenuous PA (hr/wk) and strenuous PA (hr/wk), each divided into groups

Adjusted RR (95% CI) by TPA

Longer duration provides the most benefit however need not be strenuous.

Prospective cohort

• Age: 50-79

TPA (MET hr/wk)

   • G1 = 1.00 (referent)

D & B score = 13

• Characteristics: Women from the Women's Health Initiative Observational Study

G1 = none

   • G2 = 0.90 (0.77-1.07)

G2 = 0-5.0

   • G3 = 0.82 (0.68-0.97)

G3 = 5.1-10.0

   • G4 = 0.89 (0.76-1.00)

G4 = 10.1-20.0

   • G5 = 0.83 (0.70-0.98)

G5 = 20.1-40

   • G6 = 0.78 (0.62-1.00)

G6 = ≥ 40.0

Trend p = 0.03

Moderate or strenuous PA (hr/wk)

Adjusted RR (95% CI) by TPA,

G1 = none

BMI ≤ 24.13

G2 = ≤ 1

   • G1 = 1.00 (referent)

G3 = 1.1-2.0

   • G2 = 0.78 (0.57-1.10)

G4 = 2.1-3.0

   • G3 = 0.70 (0.51-0.97)

G5 = 3.1-4.0

   • G4 = 0.80 (0.60-1.10)

G6 = 4.1-7.0

   • G5 = 0.68 (0.51-0.92)

G7 = >7.0

   • G6 = 0.63 (0.43-0.93)

Strenuous PA (hr/wk)

Trend p = 0.03

G1 = none

Adjusted RR (95% CI) by TPA, BMI 24.14-28.44

G2 = ≤ 1.0

   • G1 = referent

G3 = 1.1-2.0

   • G2 = 0.72 (0.53-0.98)

G4 = 2.1-4.0

   • G3 = 0.78 (0.57-1.10)

G5 = >4.0

   • G4 = 0.77 (0.58-1.00)

Cox proportional hazard ratio

   • G5 = 0.85 (0.64-1.10)

   • G6 = 0.78 (0.52-1.20)

Trend p = 0.74

Adjusted RR (95% CI) by TPA, BMI >28.44

   • G1 = 1.00 (referent)

   • G2 = 1.10 (0.88-1.50)

   • G3 = 0.90 (0.67-1.20)

   • G4 = 1.00 (0.79-1.30)

   • G5 = 0.89 (0.65-1.20)

   • G6 = 0.94 (0.57-1.60)

Trend p = 0.30

Adjusted RR (95% CI) by current moderate or strenuous PA

• G1 = 1.00 (referent)

   • G2 = 0.92 (0.78-1.10)

   • G3 = 0.91 (0.79-1.10)

   • G4 = 0.94 (0.81-1.10)

   • G5 = 0.99 (0.83-1.20)

   • G6 = 0.91 (0.78-1.10)

   • G7 = 0.79 (0.63-0.99)

Trend p = 0.12

Adjusted RR (95% CI) by current strenuous PA

   • G1 = 1.00 (referent)

   • G2 = 0.94 (0.80-1.10)

   • G3 = 0.95 (0.80-1.10)

   • G4 = 0.93 (0.78-1.10)

   • G5 = 0.91 (0.67-1.20)

Trend p = 0.25

Navarro Silvera et al 2006 [334]

To study the independent and combined associations of VPA, energy consumption and BMI with risk of subsequent BC.

• n = 40,318 in analysis (49,613 prior to exclusion)

Baseline and 16.4 year follow-up

1,673 cases of BC from the 40,318 included in the analysis (2,545 cases from total prior to exclusion)

The results of the study suggest that BC risk may vary according to various combinations of the components of energy balance.

Canada

• Sex: Women

PA assessment: Questionnaire for VPA (min/d), divided into groups

Adjusted HR (95% CI) by VPA

Prospective cohort

• Age: 40-59

G1 = none

   • G1 = 1.00 (referent)

D & B score = 13

• Characteristics: Canadian women with no history of BC

G2 = Any

   • G2 = 0.98 (0.85-1.13)

• National Breast Screening Study (NBSS)

G3 = 0-30

   • G3 = 1.06 (0.88-1.27)

G4 = 30-60

   • G4 = 0.98 (0.83-1.16)

G5 > 60

   • G5 = 0.93 (0.78-1.10)

Cox proportional hazard ratio

Trend p = 0.38

Adjusted HR (95% CI) by VPA, pre-menopausal

   • G1 = 1.00 (referent)

   • G2 = 0.91 (0.75-1.10)

   • G3 = 1.02 (0.80-1.31)

   • G4 = 0.88 (0.70-1.11)

   • G5 = 0.87 (0.68-1.09)

Trend p = 0.23

Adjusted HR (95% CI) by VPA, post- menopausal

   • G1 = 1.00 (referent)

   • G2 = 1.06 (0.87-1.30)

   • G3 = 1.08 (0.81-1.42)

   • G4 = 1.11 (0.87-1.41)

   • G5 = 1.00 (0.78-1.29)

Trend p = 0.96

Patel et al 2003 [335]

To examine the association between various measures of PA and post-menopausal BC risk.

• n = 72,608

Baseline and 5 year follow-up

1,520 cases of breast cancer

The study shows a lower risk of post- menopausal BC is associated with regular PA.

USA

• Sex: Women

PA assessment: Questionnaire for LTPA (METs hr/wk) at various times during life, divided into groups

Adjusted RR (95% CI), LTPA at study entry

Prospective cohort

• Age: 50-74

G1 = none

   • G1 = 0.86 (0.70- 1.04)

D & B score = 14

• Characteristics: Postmenopausal women

G2 = 0.1-6.9

   • G2 = 1.00 (referent)

• The American Cancer Society Cancer Prevention Study II (CPS-II) Nutritional Cohort

G3 = 7.0-17.5

   • G3 = 0.92 (0.81- 1.04)

G4 = 17.6-31.5

   • G4 = 0.94 (0.81- 1.09)

G5 = 31.6-42.0

   • G5 = 0.77 (0.56- 1.06)

G6 = >42.0

   • G6 = 0.71 (0.49- 1.02)

LTPA at 10 years prior to study, calculated MET score and categorized into groups:

Trend p = 0.08 (among active women p = 0.03)

None

Adjusted RR (95% CI), LTPA at age 40 yr

Slight

   • G1 = 1.03 (0.88- 1.21)

Moderate

   • G2 = 1.00 (referent)

Heavy

   • G3 = 1.05 (0.92- 1.20)

Cox proportional hazard ratio

   • G4 = 1.01 (0.87-1.18)

   • G5 = 1.16 (0.92- 1.46)

   • G6 = 0.79 (0.61- 1.03)

Trend p = 0.31 (among active women p = 0.36)

Adjusted RR (95% CI), LTPA at 10 years prior to study entry

   • None = 0.80 (0.51- 1.25)

   • Slight = 1.00 (referent)

   • Moderate = 0.93 (0.83-1.04)

   • Heavy = 0.87 (0.68- 1.13)

Trend p = 0.32 (among active women, trend p = 0.16)

Patel et al 2003 [336]

To evaluate the association between lifetime LTPA and BC risk.

• n = 1,183 (cases n = 616) n = 567, controls n = 616)

PA assessment: Calendar reporting for lifetime exercice activity (MET h/wk), divided into groups

Adjusted OR (95% CI)

The findings suggest that PA may modify the risk of in situ BC particularly in women without a family history of BC.

USA

• Sex: Women

G1 = None

   • G1 = 1.00 (referent)

Case control

• Age: 35-64

G2 = 0.0-3.0

   • G2 = 0.70 (0.48- 1.03)

D & B score = 14

• Characteristics: White and Black women

G3 = 3.0-8.0

   • G3 = 0.65 (0.44- 0.96)

G4 = 8.0-16.0

   • G4 = 0.61 (0.41- 0.92)

G5 = 16.0-32.0

   • G5 = 0.63 (0.40- 0.98)

G6 = >32.0

   • G6 = 0.65 (0.39- 1.08)

Unconditional logistical regressioin

Trend p = 0.27 (among exercisers only p = 0.81)

Rintala et al 2002 [337]

To obtain an estimate of BC incidence in association with self-rated OPA.

• n = 680,000

PA assessment: Self-reported OPA in 5 classes (1=low, 5=high)

17,986 cases of BC

The results support the hypothesis that OPA, if high enough, markedly reduced BC risk.

• Sex: Women

Class 1 = Jobs sitting and light hand tasks

Adjusted RR (95% CI), age 25-39 years

Prospective cohort

• Age: Women born in 1930-1969

Class 2 = Handling of heavier items (conveyor belt)

   • C1+2 = 1.00 (referent)

D & B score = 11

• Characteristics: Finish women

Class 3 = Jobs involving body motion

   • C3 = 0.99 (0.85- 1.17)

Class 4 = Jobs involving walking up stairs or long distances, bending and carrying

   • C4 = 0.90 (0.76- 1.07)

Class 5 = Same as class 4 except heavy tasks were performed for most of the day

   • C5 = 0.68 (0.51- 1.93)

Poisson regression models

Trend

Adjusted RR (95% CI), age 40-54 years

   • C1+2 = 1.00 (referent)

   • C3 = 1.02 (0.94- 1.11)

   • C4 = 0.99 (0.91- 1.09)

   • C5 = 0.84 (0.70- 1.00)

Trend

Adjusted RR (95% CI), age ≥ 55 years

   • C1+2 = 1.00 (referent)

   • C3 = 1.01 (0.96- 1.07)

   • C4 = 1.04 (0.98- 1.11)

   • C5 = 0.82 (0.71- 0.94)

Trend

Rockhill et al 1998 [338]

To examine the association between PA at two different times in life and BC risk.

• n = 372

Baseline and 6 year follow-up

372 cases of BC

The findings do not support a link between PA in late adolescence or in the recent past and BC risk among young adult women.

USA

• Sex: Women

PA assessment: Questionnaire for MVPA (h/wk)

Multivariate adjusted RR (95% CI)

Prospective cohort

• Age: 25-42

G1 = <1

   • G1 = 1.00 (referent)

D & B score = 12

• Characteristics: Nurses

G2 = 1.0-1.9

   • G2 = 1.1 (0.8-1.4)

• The Nurses Health Study

G3 = 2.0-3.9

   • G3 = 1.1 (0.8-1.4)

G4 = 4.0-6.9

   • G4 = 1.0 (0.7-1.4)

G5 = ≥ 7

   • G5 = 1.1 (0.8-1.5)

Logistic regression

Slattery et al 2007 [339]

To evaluate the BC risk associated with TPA and VPA at ages 15, 30 and 50 years and the referent year prior to diagnosis/selection.

• n = 4,850 Non-Hispanic white: n = 3,128 (cases n = 1,527 controls n = 1601); Hispanic American Indian: n = 1,722 (cases n = 798, controls n = 924)

PA assessment: Questionnaire for TPA (activity score) and lifetime VPA (h/wk)

1527 cases of BC (non- Hispanic white; 798 cases of BC (Hispanic American Indian)

The data suggest that PA is important in reducing risk of BC in non-Hispanic white and Hispanic American Indian women.

USA

• Sex: Women

TPA score

Case control

• Age: <50 yr

G1 = 0-3

D & B score = 12

• Characteristics: Non- Hispanic white and Hispanic American Indian

G2 = 4-6

OR (95% CI) by TPA score, non-Hispanic white

G3 = 7-9

   • G1 = 1.00 (referent)

G4 = 10-12

   • G2 = 0.78 (0.52- 1.17)

Lifetime VPA

   • G3 = 0.84 (0.57- 1.22)

G1 = None

   • G4 = 0.70 (0.44- 1.12)

G2 = <1.0

Trend p = 0.26

G3 = 1.0-2.9

OR (95% CI) by TPA score, Hispanic American Indian

G4 = ≥ 3.0

   • G1 = 1.00 (referent)

Multivariable logistic regression

   • G2 = 1.49 (0.98- 2.26)

   • G3 = 1.21 (0.80- 1.84)

   • G4 = 0.97 (0.53- 1.76)

Trend p = 0.90

OR (95% CI) by lifetime VPA, non-Hispanic white

   • G1 = 1.00 (referent)

   • G2 = 0.66 (0.36- 1.23)

   • G3 = 0.73 (0.40- 1.34)

   • G4 = 0.69 (0.37- 1.27)

Trend p = 0.68

OR (95% CI) by lifetime VPA, Hispanic American Indian

   • G1 = 1.00 (referent)

   • G2 = 1.15 (0.67- 1.96)

   • G3 = 1.19 (0.70- 2.03)

   • G4 = 1.09 (0.62- 1.90)

Trend p = 0.84

Sprague et al 2007 [340]

To investigate the relationship between LTPA and strenuous OPA and BC risk.

• n = 15,710 (1,689 cases in situ; 6,391 invasive and 7,630 controls)

PA assessment: Questionnaire for lifetime TPA (hr/wk and MET hr/wk), divided into groups

Adjusted OR (95% CI) for in situ BC by lifetime TPA (hr/wk)

The results provide further evidence that for most women, PA may reduce the risk of invasive BC.

USA

• Sex: Women

Lifetime total PA (hr/wk)

   • G1 = 1.00 (referent)

Case control

• Age: 20-69

G1 = 0

   • G2 = 0.92 (0.72-1.19)

D & B score = 13

• The Collaborative Breast Cancer Study

G2 = 0.1-15.0

   • G3 = 0.83 (0.62-1.13)

G3 = 15.1-30.0

   • G4 = 0.86 (0.59- 1.24)

G4 = > 30.0

Trend p = 0.22

MET hr/wk

Adjusted OR (95% CI) for in situ BC by lifetime TPA (MET hr/wk)

G1 = 0.0

   • G1 = 1.00 (referent)

G2 = 0.1-62.5

   • G2 = 0.93 (0.72- 1.20)

G3 = 62.6-125.0

   • G3 = 0.82 (0.61- 1.10)

G4 = >125.0

   • G4 = 0.82 (0.57- 1.17)

Trend p = 0.10

Adjusted OR (95% CI) for invasive BC by lifetime TPA (hr/wk)

   • G1 = 1.00 (referent)

   • G2 = 0.88 (0.76- 1.03)

   • G3 = 0.87 (0.73- 1.05)

   • G4 = 0.85 (0.67- 1.07)

Trend p = 0.22

Adjusted OR (95% CI) for invasive BC by lifetime TPA (MET hr/wk)

   • G1 = 1.00 (referent)

   •G2 = 0.89 (0.76- 1.04)

   • G3 = 0.82 (0.68- 0.99)

   • G4 = 0.88 (0.71- 1.09)

Trend p = 0.12

Steindorf et al 2003 [341]

To clarify the relationship between PA and BC risk.

• n = 1,246 (360 cases; 886 controls)

PA assessment: Computer assisted telephone interview for TPA (MET hr/wk) at various ages

360 cases of BC

The data do not suggest an inverse association between PA and BC risk in pre-menopausal women.

Germany

• Sex: Women

TPA at age 12-19 yr

Multivariate adjusted OR (95% CI) by TPA at age 12-19 yr

Case control

• Age: Mean. cases 41.9 yr; controls 42.5 yr

G1 = 13.0-55.7

   • G1 = 1.00 (referent)

D & B score = 13

G2 = 55.8-88.7

   • G2 = 1.07 (0.75- 1.52)

G3 = 88.8-134.0

   • G3 = 1.00 (0.70- 1.42)

G4 = 134.1-695.9

   • G4 = 0.73 (0.50- 1.07)

TPA at age 20-30 yr

Trend p = 0.44

G1 = 6.4-69.0

Multivariate adjusted OR (95% CI) by TPA at age 20-30 yr

G2 = 69.1-109.0

   • G1 = 1.00 (referent)

G3 = 109.1-160.4

   • G2 = 0.95 (0.67- 1.37)

G4 = 160.5-728.8

   • G3 = 0.85 (0.59- 1.23)

TPA at age 12-30 yr (both)

   • G4 = 0.96 (0.67- 1.39)

G1 = 17.2-70.4

Trend p = 0.32

G2 = 70.5-104.0

Multivariate adjusted OR (95% CI) by TPA at age 12-30 yr

G3 = 104.1-145.5

   • G1 = 1.00 (referent)

G4 = 145.6-564.4

   • G2 = 0.97 (0.68- 1.38)

Logistic regression

   • G3 = 0.68 (0.46- 0.99)

G4 = 0.94 (0.65- 1.35)

Trend p = 0.29

Tehard et al 2006 [342]

To investigate the type, duration, frequency and intensity of PA required to reduce the risk of BC.

• n = 90,509

Baseline and follow-up every 2 years for 12 years

3,424 cases of BC

BC risk was reduced, especially with VPA.

France

• Sex: Women

PA assessment: Questionnaire for various PA variables, all divided into groups

Multivariate adjusted RR (95% CI) by TPA

Prospective cohort

• Age: 40-65

TPA (MET hr/wk)

   • G1 = 1.00 (referent)

D & B score = 13

• Characteristics: French women insured with Mutuelle Generale de l'Education Nationale

G1 = <28.3

   • G2 = 1.05 (0.93- 1.17)

• E3N Cohort Study

G2 = 28.3-41.8

   • G3 = 0.94 (0.83- 1.05)

G3 = 41.8-57.8

   • G4 = 0.90 (0.80- 1.02)

Trend p < 0.05

G4 = ≥ 57.8

Multivariate adjustedRR (95% CI) by total recreational PA

Total recreational PA (MET hr/wk)

   • G1 = 1.00 (referent)

G1 = Inactive

   • G2 = 0.82 (0.71- 0.93)

G2 = <16.0

   • G3 = 0.94 (0.84- 1.06)

G3 = 16.0-22.3

   • G4 = 0.88 (0.79- 0.98)

G4 = 22.3-33.8

   • G5 = 0.81 (0.72- 0.92)

G5 = ≥33.8 = 0.81

Trend p < 0.01

Walking (min/d)

Multivariate adjusted RR (95% CI) by walking duration

G1 = <500

   • G1 = 1.00 (referent)

G2 = 500-2000

   • G2 = 1.03 (0.95- 1.11)

G3 = >2000

   • G3 = 0.91 (0.81- 1.02)

MPA (hr/wk)

Trend p = 0.45

G1 = Inactive

Multivariate adjusted RR

G2 = 0

(95% CI) by MPA

G3 = 1-4

   • G1 = 1.00 (referent)

G4 = 5-13

   • G2 = 0.80 (0.60- 1.05)

G5 = 14

   • G3 = 0.87 (0.79- 0.94)

VPA (hr/wk)

   • G4 = 0.86 (0.74- 0.99)

G1 = Inactive

   • G5 = 0.89 (0.65- 1.24)

G2 = 0

Trend p<0.01

G3 = 1-2

Multivariate adjusted RR (95% CI) by VPA

G4 = 3-4

   • G1 = 1.00 (referent)

G5 = 5

   • G2 = 0.90 (0.81- 0.99)

Cox proportional hazard ratio

   • G3 = 0.88 (0.79- 0.97)

   • G4 = 0.82 (0.71- 0.95)

   • G5 = 0.62 (0.49- 0.78)

Trend p < 0.0001

Thune et al 1997 [343]

To investigate whether everyday exercise is related to the risk of BC.

• n = 25,624

Baseline and mean follow- up of 14 years

351 cases of BC (110 pre-menopausal and 251 post-menopausal women)

LTPA and OPA are associated with a reduced risk of BC.

Norway

• Sex: Women

PA assessment: Self-reported LTPA and OPA, divided into groups

   • G1 = 1.00 (referent)

Adjusted RR (95% CI) by LTPA

Prospective cohort

• Age: 20-54

LTPA

   • G2 = 0.93 (0.71- 1.22)

D & B score = 14

G1 = Sedentary

   • G3 = 0.63 (0.42- 0.95)

G2 = Moderate

Trend p = 0.04

G3 = Regular exercise OPA

Adjusted RR (95% CI) by LTPA, pre- menopausal

G1 = Sedentary

   • G1 = 1.00 (referent)

G2 = Walking

   • G2 = 0.77 (0.46- 1.27)

G3 = Lifting

   • G3 = 0.53 (0.25- 1.14)

G4 = Heavy manual labor

Trend p = 0.10

During work

Adjusted RR (95% CI) by LTPA, post- menopausal

Pre-menopausal

   • G1 = 1.00 (referent)

G1 = Sedentary

   • G2 = 1.00 (0.72- 1.39)

G2 = Walking

   • G3 = 0.67 (0.41- 1.10)

G3 = Lifting or heavy manual labor

Trend p = 0.15

Adjusted RR (95% CI) by OPA

   • G1 = 1.00 (referent)

   • G2 = 0.84 (0.63- 1.12)

   • G3 = 0.74 (0.52- 1.06)

   • G4 = 0.48 (0.25- 0.92)

Trend p = 0.02

Adjusted RR (95% CI) by OPA, pre- menopausal

   • G1 = 1.00 (referent)

   • G2 = 0.82 (0.50- 1.34)

   • G3 = 0.48 (0.24- 0.95)

Trend p = 0.03

Adjusted RR (95% CI) by OPA, post- menopausal

   • G1 = 1.00 (referent)

   • G2 = 0.87 (0.61- 1.24)

   • G3 = 0.78 (0.52- 1.18)

Trend p = 0.24

Zheng et al 1993 [344]

To assess the role of OPA in the risk of BC.

• n= 3,783 (BC = 2,736)

PA assessment: Interview for OPA, divided into groups

2,736 cases of BC

Women with low OPA had an increased risk of BC; the incidence of BC was reduced in women with high-activity jobs.

China

• Sex: Women

G1 = Low

Standardized incidence ratios

D & B score = 9

• Age: 30

G2 = Moderate

   • G1= 131

G3 = High

   • G2 = 95

   • G3 = 79