Journal of Obesity / 2016 / Article / Tab 3

Research Article

Weight Fluctuation and Postmenopausal Breast Cancer in the National Health and Nutrition Examination Survey I Epidemiologic Follow-Up Study

Table 3

Studies with weight cycling as risk factor for postmenopausal breast cancer.

Author(s)/year/[ref]Study design 
population/database
Cancer 
Outcome
Exposure: weight cycling/analysis and covariatesResults/ConclusionComments

French et al., 
1997 [29]
Cohort
1986–1992 
Iowa 
Women’s 
Health Study
Postmeno
older 
women 
55–69 years 
Case = 660 
self- 
reported 
WT change 
18–62 years
Stable WT (ref) (<5% 
change between ages 18 and 62 years) and small gain (at least one gain between 5 and 10%) 
Large cycle (at least one gain 10% and at least one loss 10%) 
Small cycle  (at least one gain ≥5% but <10% and at least one loss ≥ 5% but <10%) 
Large gain (at least one gain ≥10%) 
Weight loss maintenance
(a > 10% weight loss with 5% maintenance of reduced value) 
Others (all other unspecified weight change patterns)  
Cox proportional hazard regression with multivariate adjustment for baseline values of age, waist/hip ratio, BMI, BMI2, smoking status (never, former, and current), pack years of cigarettes, education (<high school, high school, and >high school), physical activity (low, medium, and high), alcohol, marital status (yes/no), and hormone replacement (never, former, and current)
Stable W T(ref)
versus
large cycle
0.83 (0.60–1.14), 
small cycle
0.86 (0.58–1.27), 
large gain
1.29 (1.02–1.63),
weight loss
maintenance
0.81 (0.50–1.30), and
others
1.01 (0.80–1.28) 
Large gainers
were at significant risk of BC (versus stable weight)
First study 
Large sample size 
Weight variability 
was defined as () 
the root-mean-square-error (RMSE)
around the slope 
of weight on age and 
() categorical 
measures of weight change

Trentham-Dietz et al., 2000 
[30]
Case-Control
50–79 years old 
Massachusetts, 
New 
Hampshire, 
Wisconsin
Postmeno
invasive BC 
Case = 5031 
Cont. = 5255
Weight cycling: at least 1 cycle loss of at least 20 pounds and regain back of at least half during a year
Conditional logistic regression 
on age and state, adjusted for parity, age at first full-term pregnancy, family history of breast cancer, recent alcohol consumption, education, and age at menopause
≥1 cycle versus 
no cycles 
OR = 1.0 (0.9–1.1)
No association
~18.6% cases and ~17% controls 
were weight 
cyclers

Eng et al., 2005 
[31]
Case-Control
20–98 years old 
Long Island 
Breast CSP: 
1996-1997
Postmeno diganosed: first primary in situ or invasive Cases = 990 
Cont. = 1,006
Weight cycling as switching from “low” to “high” status and back at 4 time-points: 20, 30, 50, and 1 y before ref/diagnosis: “high” and “low” versus control medium weight (kg)
Unconditional logistic regression adjusted for age at reference day (diagnosis), number of pregnancies, lactation status, nulliparity, history of breast cancer in a mother, sister, or daughter, and history of benign breast disease
OR = 2.11 (1.00–4.44)Twofold increase in risk versus mainlining the “low” status