Breast cancer mortality is higher in black than in white women. The incidence of pre-menopausal breast cancer is higher in black than in white women and, in some age groups, is increasing at a faster rate than in white women. These and other aspects of breast cancer epidemiology suggest the need for within-race studies of the causes and prevention of breast cancer in black women. Dietary factors, as universal and potentially-modifiable risk factors are of substantial interest in this respect. Yet, the contribution of diet to breast cancer etiology has been difficult to specify, partly due to methodological problems with dietary assessment. The proposed studies are motivated by evidence suggesting that effective implementation of the needed studies of diet in breast cancer etiology in black women is especially impeded by methodological problems: food frequency or dietary history methods that have acceptable validity in white populations appear to have less validity in black populations. These problems can be overcome, in part, by the application of certain statistical approaches to correct for measurement error-- approaches requiring repeated-measures data collected from the same population with a more accurate method. Studies are proposed to 1) assess the validity of food frequency questionnaire data collected from a sample of black women, and 2) quantify the impact of the associated error on epidemiologic analyses of diet and breast cancer risk. We will use telephone and mail contacts to collect 3 unannounced 24-hour dietary recalls and one 3-day food record from a geographically dispersed sample of 400 black women. The women will be sampled from a large cohort of black women who completed a food frequency questionnaire as part of their enrollment in a large cohort study. The recall and record data will constitute reference data approximating """"""""true"""""""" intake and will support comprehensive statistical modeling to quantify random variation (i.e., intraperson differences in nutrient intakes from day-to-day) and bias (systematic departures of nutrient intakes estimated from the food frequency questionnaire). Separate analyses will be done for each of several key nutrients of interest in relation to breast cancer etiology, including energy, total fat, saturated fat, dietary fiber, beta carotene, vitamin C, vitamin E, and alcohol. Deattenuation and correction factors derived from the dietary assessment studies will be used in exploratory analyses of dietary risk factors for incident breast cancer using data from the prospective study data base. Thus, this research will yield statistical methods for enhancing the ability to assess diet-related breast cancer risks in black women as well as relevant pilot data to support future studies.
| Kumanyika, Shiriki K; Mauger, David; Mitchell, Diane C et al. (2003) Relative validity of food frequency questionnaire nutrient estimates in the Black Women's Health Study. Ann Epidemiol 13:111-8 |
| Lancaster, K J; Smiciklas-Wright, H; Kumanyika, S K et al. (2000) Food diary response rates in a sample of African-American women: respondent characteristics and nutrient intake. J Am Diet Assoc 100:1532-5 |
