Explaining the Relation Between Education and Postmenopausal Breast Cancer

American Journal of Epidemiology Copyright O 1997 by The Johns Hopkins University School of Hygiene and Public Health All rtght3 reserved Vol. 145, N...
Author: Marjorie Hardy
7 downloads 2 Views 631KB Size
American Journal of Epidemiology Copyright O 1997 by The Johns Hopkins University School of Hygiene and Public Health All rtght3 reserved

Vol. 145, No. 4 Printed in U.S.A

Explaining the Relation Between Education and Postmenopausal Breast Cancer

Katherine E. Heck and Elsie R. Pamuk The authors examined the relation between socioeconomic status, as defined by education level, and postmenopausal breast cancer incidence using data from the National Health and Nutrition Examination Survey I Epidemiologic Followup Study. Female participants in the study were followed from 1971-1974 to 1992-1993. Cox proportional hazards modeling was used to determine the relation between breast cancer incidence and education level. There was a direct dose-response association between education level and postmenopausal breast cancer risk. Several breast cancer risk factors, including height and reproductiverelated risks such as nulliparity, were found to mediate this relation. Adjustment for these factors reduced, but did not eliminate, the positive relation between education level and risk of postmenopausal breast cancer; however, the association was no longer statistically significant. The association between higher education and increased risk of breast cancer appears to be largely explained by differences in the known risk factors for breast cancer. Am J Epidemiol 1997; 145:366-72. breast neoplasms; education; socioeconomic factors

Unlike most other illnesses, breast cancer has been associated with higher socioeconomic status. This correlation has appeared at both the individual and the community level. Case-control studies in the United States, Canada, and Israel (1, 2) indicate that women with more than 12 years of education have a significantly increased risk of breast cancer. In a Canadian study (3), education was not related to breast cancer risk, but cases tended to have greater family incomes than controls. A number of ecologic analyses of US data (4-9) have found higher rates of breast cancer incidence and mortality among women living in more affluent communities. Smaller, individual-level studies (10-15) have suggested the same disparity. Research in countries as diverse as Brazil (16), Finland (17-19), Italy (20), Denmark (21), and Australia (22) have found positive relations between breast cancer incidence and socioeconomic status. Although most research suggests this link between higher socioeconomic status and breast cancer, few US studies have been able to examine nationally representative individual-level data or to control for mediating variables. The first National Health and Nutri-

tion Examination Survey (NHANES I) Epidemiologic Followup Survey (NHEFS) has been used to examine multiple risk factors for breast cancer because it provides an opportunity to use individual-level data to examine the incidence of breast cancer among a nationally representative cohort of several thousand women, followed for up to 22 years, and to examine the effects of known breast cancer risk factors as well as socioeconomic variables. In one study using NHEFS, Carter et al. (23) found that education past high school was associated with a greater risk of breast cancer. Another socioeconomic variable, income, also has been positively associated with breast cancer risk in NHEFS (24). Additional examination of the link between breast cancer and socioeconomic status may provide more information about the etiology of the disease and why breast cancer risk appears to increase with higher socioeconomic status. Determining the extent to which the unusual association between breast cancer and socioeconomic status is explained by known risk factors should provide more specific information about which women are at high risk of acquiring the disease.

Received for publication April 12, 1996, and accepted for publication October 9, 1996. Abbreviations: NHANES I, first National Health and Nutrition Examination Survey; NHEFS, NHANES I Epidemiologic Followup Study. From the National Center for Health Statistics, Hyattsville, MD. Reprint requests to Katherine E. Heck, National Center for Health Statistics, 6525 Belcrest Road, Room 730, Hyattsville, MD 20782.

MATERIALS AND METHODS

The research question to be addressed by this analysis was whether socioeconomic status is related to the incidence of breast cancer and whether such a relation could be explained by variation in reproductive and 366

Education and Postmenopausal Breast Cancer

other breast cancer risk factors, which may vary between women of different socioeconomic standing. The NHEFS data include several indicators of socioeconomic status, such as education, income, poverty level of the census tract, and urban/rural status. However, many women in the sample were older than retirement age, when income and other social circumstances can change rapidly. Education was chosen to represent socioeconomic status because it is a more constant measure of lifelong social status than variables such as income or residence. The sample available for this study consisted of all women (n = 8,596) who took part in the NHANES I Epidemiologic Followup Study (25). NHANES I, a survey and physical examination focusing on nutrition status and other health issues, was conducted during 1971-1975 on a national probability sample of the noninstitutionalized population of the United States. The NHEFS sample consisted of all NHANES I participants who were 25-74 years of age at the original survey. NHEFS participants were tracked and subsequently surveyed in three periods, 1982-1984, 1987, and 1992-1993. Participants older than 55 years at baseline were also surveyed in 1986. The NHEFS collected health and demographic information through an interview of the subject or a proxy, obtained death certificates for participants who had died, and requested medical records from hospitals in which the subject or proxy reported an overnight stay during the follow-up period. The response rate to the full NHANES I (survey and medical examination) was 69.5 percent. Of the original sample, 93 percent were traced by the 1982-1984 wave, and 87 percent were either interviewed or a proxy was interviewed, with a loss to follow-up of 3.5 percent between 1982-1984 and the 1992-1993 follow-up survey. The study population excluded female participants who were not traced (n = 376), women of a race other than black or white (n = 87; race in NHANES I was collected by interviewer observation, and no ethnicity information was collected), women with missing information on education level (n = 51), and women identified as having breast cancer prevalent at baseline (n = 56). Staff of the National Cancer Institute completed a medical record review of breast cancer cases in NHEFS in August 1996. This study used the National Cancer Institute designations of case status and date of diagnosis. Case ascertainment and determination of diagnosis date was made from a thorough review of available medical records, self-reports, and death certificates. For the eight cases for whom ascertainment was made only through the death certificate, a date of diagnosis was imputed based on age-specific survival Am J Epidemiol

Vol. 145, No. 4, 1997

367

rates from the Surveillance, Epidemiology, and End Results (SEER) cancer surveillance program. There were too few premenopausal cases to examine separately, so this analysis was limited to postmenopausal breast cancer. Women who never reached menopause during the follow-up period or who had an incidence of premenopausal breast cancer were excluded (n = 1,669), as were 96 additional women for whom data were missing for the variables of interest. The final sample consisted of 229 cases and 6,032 noncases. The relative risk of breast cancer incidence at four levels of education by years of completion (