Dietary Fat Intake and Risk of Uterine Leiomyomata

A Prospective Ultrasound Study

Theodore M. Brasky; Traci N. Bethea; Amelia K. Wesselink; Ganesa R. Wegienka; Donna D. Baird; Lauren A. Wise


Am J Epidemiol. 2020;189(12):1538-1546. 

In This Article


The Study of Environment, Lifestyle and Fibroids

SELF is a prospective cohort of Black women living in the Detroit, Michigan, metropolitan area.[14] Recruitment of SELF participants occurred from 2010 through 2012, with community outreach through radio, television, newspapers, event booths, and informational letters to women in the Henry Ford Health System, the clinical institution collaborating on SELF.[14] Enrolled women (n = 1,693) were self-identified as Black/African-American, were 23–35 years of age without self-reported history of hysterectomy (partial or total, verified at baseline ultrasound), and reported no prior diagnoses of UL, cancer, or autoimmune diseases requiring regular medication. Women who were pregnant at the time of recruitment delayed their enrollment into the study until after delivery to assure optimum ultrasound imaging.

At baseline and every 20 months during a 5-year follow-up period, participants completed computer-assisted telephone and Web-based questionnaires, self-administered paper questionnaires, and attended in-person clinic visits. Participants were queried on their personal medical history, physical activity, lifestyle and behaviors (e.g., smoking, alcohol intake), reproductive history, and use of contraceptives. Transvaginal ultrasound was performed during in-person clinic visits at enrollment (baseline), and at 20, 40, and 60 months of follow-up. Pregnant women were asked to return to the clinic at 4 months postpartum. For women whose pregnancy was identified at follow-up ultrasound, data were recorded if the participant was ≤12 weeks pregnant based upon fetal measures, otherwise they were asked to return to the clinic at 4 months postpartum. Study sonographers were registered diagnostic medical sonographers with ≥3 years of experience in gynecological sonography.[15] Sonographers received additional training for the study to assure consistency in completing research documentation about the ultrasound and received regular refresher trainings during the course of the study.[14] At each clinic visit trained study staff measured height and weight, from which body mass index was calculated (as weight (kg)/height (m)2). All participants gave written informed consent, and the study was approved by the institutional review boards of the Henry Ford Health System, the National Institute of Environmental Health Sciences, and Boston University Medical Campus.

Dietary Measurement

At baseline, women completed a validated Web-based semiquantitative Block food frequency questionnaire.[16–18] Participants reported their usual frequencies and serving sizes (in cups: 0.25, 0.50, 1, or 2) of 110 foods and beverages in the past 12 months. Average daily intakes of dietary fat and individual fatty acids were calculated by multiplying the serving-size and season-adjusted frequency of each food item by its fat content as determined by the US Department of Agriculture Food and Nutrient Database for Dietary Studies.[19] Data for total dietary fat and fat groups (SFA, MUFA, PUFA, and trans-fatty acids) are given in grams/day, whereas data for individual fatty acids were given in milligrams/day. Summary measures were calculated for total dietary marine ω-3 PUFA, as the sum of eicosapentaenoic (20:5n3), docosapentaenoic (22:5n3), and docosahexaenoic acids (22:6n3), and total ω-6 PUFA as the sum of linoleic (18:2n6) and arachidonic acids (20:4n6). Total fin-fish and shellfish intake was calculated as the sum of serving-size adjusted grams/day of tuna (including tuna salad and tuna casserole), fried fish or fish sandwich, other fish (not fried), shellfish (including shrimp, scallops, and crabs), and oysters.

For the present analysis, we excluded 384 women with prevalent UL identified by ultrasound at enrollment. We excluded an additional 78 women with total energy intakes of <400 or ≥5,000 kcal/day and 60 women who were missing follow-up data, leaving 1,171 women followed for a median of 5 years for incident UL. Among them, we identified 277 incident UL cases (primary outcome) first detected at their 20- (n = 110), 40- (n = 88), or 60-month follow-up visit (n = 79). The remaining women were right-censored at hysterectomy (n = 8), withdrawal from the study after completing at least 1 follow-up visit (n = 66), or their 60-month follow-up visit (n = 820), whichever came first. There were no apparent differences in baseline characteristics among women with no follow-up data (n = 60; excluded), women who withdrew from the study after completing at least 1 follow-up visit (n = 66; censored), and the remaining analytical cohort (n = 1,105) (Web Table 1, available at

Statistical Analysis

Dietary fats were energy-adjusted using the nutrient residual method.[20] We categorized dietary variables into quartiles to avoid the assumption of linearity between dietary fats and UL risk and to make our findings comparable with those of the BWHS.[13] We used Cox proportional hazards models, with follow-up time as the time metric, to estimate hazard ratios and 95% confidence intervals for the association between dietary fat and UL incidence. In addition to accounting for follow-up time, all models adjusted for age (stratification variable) and total energy intake. A priori we determined additional baseline factors for inclusion in multivariable models; they included education (up to high school, some college, college or advanced degree), body mass index (continuous), age at menarche (in years: <11, 11, 12, 13, ≥14), parity (nulliparous and tertiles of births: 1, 2, and ≥3 births), age at first birth (nulliparous and tertiles of age in years at first birth: <19, 19–22, ≥23), years since last birth (nulliparous and tertiles of years since last birth: 0–2.6, 2.7–5.7, ≥5.8), ever used oral contraceptives, ever used progestin-only injectable contraceptives, alcohol intake (quartiles of drinks/week: <0.33, 0.33–1.27, 1.28–3.98, ≥3.99), smoking status (none, former, current for <10 years, current for ≥10 years), and history of hypertension (yes/no). We calculated P values from tests for trend (P for trend) by assigning the median value to each category of ordinal dietary variables and including them as continuous variables in regression models.

Because progestin-only injectable contraceptives strongly suppress ovarian hormones (estradiol and progesterone), we performed a sensitivity analysis in which we additionally excluded 76 women (including 10 with incident UL cases) who were using these contraceptives at baseline. Statistical analyses were performed using SAS, version 9.4 (SAS Institute, Inc., Cary, North Carolina). All reported P values are 2-sided, and P < 0.05 was considered statistically significant.