SEARCH

SEARCH BY CITATION

Keywords:

  • education;
  • socio-economic status;
  • breast cancer;
  • cohort;
  • Sweden;
  • Norway;
  • epidemiology

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

A positive relationship between level of education and female breast cancer risk is well supported by scientific evidence, but few previous studies could adjust for all relevant potential confounding factors. The authors' purpose was to examine how risk for breast cancer varies with level of education and to identify factors that explain this variation, using data from a prospective cohort study including 102,860 women from Norway and Sweden who responded to an extensive questionnaire in 1991/1992; 1,090 incident primary invasive breast cancer cases were revealed during follow-up, which ended in December 1999. The Cox Proportional Hazards Model was used to calculate relative risks (RR) with 95% confidence intervals (CI). Women with more than 16 years of education had a 36% increased risk compared to the lowest educated (7–9 years) (Age adjusted RR=1.36, 95% CI: 1.10, 1.68). This relationship was slightly stronger among postmenopausal (RR 1.51) than among premenopausal (RR 1.25) women. In both groups, however, the relative risk estimates turned close to unity by adjustment for parity, age at first birth, body mass index (BMI), height, age at menarche, menopausal status, use of oral contraceptives and consumption of alcohol. The overall multivariate relative risk among the highest educated women was 1.04 (95% CI 0.82–1.32). The results of our study suggest a clear positive gradient in risk for breast cancer by level of education, which can be fully explained by established breast cancer risk factors. © 2004 Wiley-Liss, Inc.

Socioeconomic differentials concerning a wide range of diseases, including cancer, have been frequently reported during the last decades. The direction of the socioeconomic gradient in risk differs, however, between cancer sites.1 Among women, it tends to be negative for lung, stomach, oesophagus and cervical cancer, while a positive association has been observed for malignant melanoma and cancers of the colon, breast and ovaries. The excess risk of breast cancer among women with high socioeconomic status (SES) is confirmed by a number of epidemiological studies. Different measures of SES have been applied, but the link exists both with income,2, 3, 4, 5, 6, 7, 8 occupation or socioeconomic group,9, 10, 11, 12, 13, 14, 15, 16 and level of education.2–5, 9–12, 17–21 Although level of education obviously acts only as an indicator of aetiologically relevant factors, no study has fully explained the relation by multivariate adjustment for possible confounding factors. Among the few prospective studies, one found no association,22 while two did,2, 17 one of them being restricted to postmenopausal women. However, the positive association between SES and breast cancer risk observed in these studies was explained only partially by known confounding factors. Thus, further investigation is required to increase our understanding of the correlates of education that affect risk for breast cancer.

We present here results from a large, prospective cohort study carried out in Norway and Sweden, with comprehensive information on the characteristics of a woman's life and behaviour that might affect the risk of developing breast cancer. The aim of our study was to assess how risk for breast cancer varies with level of education and to identify the underlying causal factors leading to this variation.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The cohort

The cohort was enrolled during 1991 and 1992. In Norway, a sample of 100,000 women born between 1943 and 1957 (34–49 years of age) was randomly selected from the Central Population Register. This register records the addresses of all persons alive and residing in the country, and the dates of death or migration to or from Norway since 1960. In this register each person is identified by an individually unique national registration number; the first 6 digits encode information on the date of birth, and the last 5 digits are based on an algorithm that ensures a unique number, including information on gender. In Sweden, a sample of 96,000 women born between 1942 and 1962 (30–50 years of age), residing in the Uppsala Health Care Region (comprising about 1/6 of the Swedish population) was randomly selected from the Swedish Central Population Register at Statistics Sweden. In this register, each individual is identified by a unique 10-digit national registration number, which encodes information on date of birth and gender.

A letter of invitation to participate in the study and a health-survey questionnaire were sent to all women. In Norway, the questionnaire was mailed to 10 subgroups at regular intervals. In Sweden, 2 mailings were done: 1 in 1991 and 1 in 1992. Of the 100,000 invited women in Norway, 57,582 (57.6%) returned a completed questionnaire, as did 49,259 of the 96,000 invited women (51.3%) in Sweden. Thus, the overall crude participation rate was 54.5% (106,841 out of 196,000). The questions relevant to this analysis were identical in the 2 countries. This common set of questions included a detailed assessment of reproductive history, height and weight, contraceptive use, prevalent diseases, history of breast cancer in mother and sister(s), lifestyle habits and total number of years of education.

Follow-up

Follow-up was achieved through linkages between the cohort data set and various population-based registries. These linkages were possible through the use of the individually unique national registration numbers present in all national registries in Norway and Sweden.23 We obtained information on dates of death for deceased persons from the death registers and on dates of emigration from the registers of population migration. The national cancer registries, established in the 1950s in both countries, provided data on prevalent cancer cases at cohort enrolment and incident cancers diagnosed in the cohort during the follow-up. These registers are considered to be almost complete. We excluded from the cohort 15 women who were dead or had emigrated before the start of follow-up. Another 1,663 women with a prevalent cancer diagnosis at study enrolment were also excluded, as were 2,303 women who did not state educational length in the questionnaire. Hence, the study population includes 102,860 subjects. The follow-up ended on 31 December 1999, or at emigration, death, or primary cancer diagnosis, whichever occurred first.

Classification of education

In the questionnaire, women were asked to give the total number of years they attended school. The choice of classification is yet related to levels in the educational system in Norway and Sweden, and hence the term educational level will be used in the following. In Sweden, compulsory school attendance increased from 7 to 9 years in 1959. In Norway, this happened about 7 years later. Thus 7–9 years of education means primary school with at most 2 years of additional professional education. Women with 10–12 years of education may have completed secondary school, or up to 5 years of professional training. Education lasting 13–15 years corresponds to a university bachelor degree, or, in some instances, several professional training sessions at a lower level. The highest category comprises women with more than 16 years of education, which mainly corresponds to a university master level.

Statistical analysis

The Cox Proportional Hazards Model was applied to perform the statistical analyses, using the SAS Software Package (version 8.2) to calculate hazard ratios with corresponding 95% confidence intervals. The hazard ratios are interpreted as estimates of relative risks (RR).

The relationship between years of education and breast cancer incidence was first examined in age-adjusted analyses. Subsequently, other explanatory variables were added stepwise to the model whenever they tended to confound the association of interest, which was defined as a change in the RR of at least 1%. Age at first birth (<21, 22–24, 25 years or more) and parity (0, 1, 2, 3 or more children) were considered as a set of combined indicator variables, while age at start of follow-up, BMI (weight in kilos divided by height squared), height, age at menarche and alcohol consumption were treated as continuous variables. We tested BMI as a categorical variable in the statistical models, which gave a poorer model fit than treating it as continuous variable. Information on menopausal status was obtained from the questionnaire. Only women who reported natural menopause or a bilateral oophorectomy at cohort enrolment were considered postmenopausal, regardless of hysterectomy, or use of hormonal replacement therapy (HRT). Unknown age at menopause was set to 50 in the separate analyses. Family history of breast cancer was not related to level of education in our data and hence not included in the model. Tests for linear trend were carried out by introduction of an ordinal variable obtained by assigning consecutive integers to the categories of education.

The responsible Data Inspection Boards and Ethical Committees in both countries approved the study design, and all women gave informed consent to participate in the study.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Characteristics of the study population by country of residence are given in Table I. A total of 1,090 incident breast cancers were diagnosed during the follow-up. The slight difference in mean age at entry among Norwegian and Swedish women is attributable to a small discrepancy in range of age. Table II shows the distribution by education of the covariates included in the analysis. Well-educated women were on average younger, had fewer children and were older at their first birth. They also had a lower BMI and were taller than the less educated. Mean alcohol consumption increased with education, as did use of hormonal contraceptives. Age at menarche was on average slightly higher for the lowest educated women in our study population.

Table I. Characteristics of the Study Population and the Incident Cases of Breast Cancer According to Country of Residence: the Norwegian-Swedish Women's Lifestyle and Health Cohort Study 1991–1999
CharacteristicsNorwaySwedenTotal
  • 1

    Reported postmenopausal at cohort enrolment or passed age 50 at time of diagnosis.

Number of women55,60347,257102,860
Age at entry, mean (range)41.1 (34–49)39.5 (30–50)40.4
Person-years of follow-up451,382380,510831,892
Number of invasive breast cancer cases6224681,090
Age at diagnosis of premenopausal breast cancer, mean (range)44.8 (36–50)44.4 (30–50)44.6
Age at diagnosis of postmenopausal breast cancer, mean (range)152.0 (44–56)52.5 (38–57)52.2
Table II. Characteristics By Education: the Norwegian-Swedish Women's Lifestyle and Health Cohort Study 1991–1999
 NYears of education
7–9 %10–12 %13–16 %≥17 %
  • 1

    Weight (kg)/height squared (m2).

Total102,86021.737.830.210.3
Breast cancer cases1,09022.335.330.511.9
Characteristics     
Age at entry     
30–34 years14,2647.416.416.510.4
35–39 years31,78820.932.235.035.4
40–44 years29,90730.728.228.530.7
45–49 years26,90141.023.220.023.5
Mean age (± SD) 42.3 years (± 4.8)39.9 years (± 5.1)39.6 years (± 4.9)40.4 years (± 4.8)
Age at first birth     
Less than 20 years12,98227.716.05.92.3
20–24 years40,62151.451.238.023.6
25–29 years26,86415.925.041.045.0
30 years or more10,3165.07.815.129.1
Mean age at first birth (± SD) 22.0 years (± 3.9)23.3 years (± 4.0)25.4 years (± 4.1)27.4 years (± 4.3)
Parity at entry     
Nulliparous12,0729.010.113.119.5
One child14,50212.314.014.616.7
Two children44,89340.745.744.839.1
Three children or more31,39338.030.227.524.7
Mean number of children (± SD) 2.2 (± 1.2)2.0 (± 1.1)1.9 (± 1.1)1.8 (± 1.2)
BMI1     
Less than 18.5 kg/m22,1482.12.02.32.4
18.5–24 kg/m272,47964.071.976.879.3
25–29 kg/m220,24726.420.817.015.3
30 kg/m2 or more5,1127.55.33.93.0
Mean BMI (± SD) 24.0 (± 3.9)23.3 (± 3.6)22.8 (± 3.4)22.5 (± 3.3)
Mean height (± SD) 165.4 cm (± 5.7)166.1 cm (± 5.6)166.8 cm (± 5.7)167.3 cm (± 5.7)
Mean age at menarche (± SD) 13.2 (1.4)13.1 (1.4)13.1 (1.4)13.1 (1.4)
Use of hormonal contraceptives     
Ever used74,35036.125.423.325.1
Never used27,52863.974.676.774.9
Mean alcohol consumption (± SD) 2.3 (± 5.5)2.7 (± 5.0)3.0 (± 4.8)3.7 (± 5.4)

The relative risks for the total cohort comprising both pre- and postmenopausal breast cancer cases are given in Table III. We observed a steadily increasing positive association between educational level and breast cancer risk (p for linear trend = 0.001). When we added age at first birth and number of children to the model the magnitude of the association decreased considerably. Low BMI accounted for a modest increase in risk. The slight variation in risk still left was almost completely explained by the use of hormonal contraceptives, height, age at menarche, alcohol consumption and menopausal status at cohort entry. Hence, in the ultimate multivariate model no association between education and breast cancer risk persisted (p for trend=0.66).

Table III. Relative Risks (RR) with 95% Confidence Intervals of Developing Breast Cancer in Relation to Years of Education: the Norwegian-Swedish Women's Lifestyle and Health Cohort Study 1991–1999
 Years of education
7–910–1213–16≥17
Adjustment
Age1.00 (ref.)1.12 (0.95–1.32)1.26 (1.06–1.49)1.36 (1.10–1.68)
Age0.001
p for linear trend1.00 (ref.)1.08 (0.91–1.27)1.13 (0.95–1.35)1.16 (0.92–1.45)
Age, parity, age at first birth1.00 (ref.)1.06 (0.90–1.25)1.11 (0.93–1.32)1.11 (0.89–1.40)
Age, parity, age at first birth, BMI1.00 (ref.)1.03 (0.86–1.23)1.05 (0.87–1.27)1.04 (0.82–1.32)
Age, parity, age at first birth, BMI, height, age at menarche, menopausal status at entry, ever use of hormonal contraceptives, consumption of alcohol    
p for linear trend0.66

In Table IV, the cohort is separated by estimated menopausal status at follow-up. Among premenopausal women none of the categories of educational level showed a significantly elevated risk of breast cancer compared to the reference group, although there was a significant trend across educational groups (p=0.03). This trend levelled off by subsequent multivariate adjustment, as described above. The analysis of postmenopausal women revealed a steeper increase in risk by level of education. However, as for the total cohort, the RRs were reduced after controlling for parity in the model and turned close to unity in the multivariate analysis when other risk factors were adjusted for.

Table IV. Relative Risks (RR) with 95% Confidence Intervals of Developing Breast Cancer in Relation to Years of Education, According to Menopausal Status: the Norwegian-Swedish Women'S Lifestyle and Health Cohort Study 1991–1999
Years of educationRR (95% CI)
PremenopausalPostmenopausal
Age adjustedMultivariate1Age adjustedMultivariate1
  • 1

    Adjusted for age, parity, age at first birth, BMI, height, age at menarche, ever use of hormonal contraceptives, and consumption of alcohol

7–91.00 (ref.)1.00 (ref.)1.00 (ref.)1.00 (ref.)
10–121.02 (0.82–1.25)0.96 (0.77–1.21)1.30 (1.00–1.68)1.12 (0.85–1.48)
13–161.19 (0.96–1.47)1.03 (0.81–1.31)1.32 (0.99–1.74)1.03 (0.75–1.40)
≥171.25 (0.96–1.64)0.99 (0.74–1.34)1.51 (1.05–2.16)1.09 (0.74–1.61)
p for linear trend0.030.810.020.80

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Our finding of a positive association between level of education and risk of breast cancer is consistent with most2–5, 9–12, 17–21 but not all22 previous studies. Moreover, our hypothesis that this association could be explained by known risk factors was supported. Differences in parity and age at first birth accounted for more than 50% of the difference in risk between the lower and the higher educated group of women. The remaining variation in risk was attributable to lower BMI, increased height, lower age at menarche, later age at menopause and more frequent use of both alcohol and hormonal contraceptives among the higher educated group. The association of parity and age at first birth with breast cancer risk is well established,24 while high BMI is found to be a protective factor before but not after menopause.25 We also observed a persisting negative linear relationship between BMI and breast cancer risk after menopause, although it weakened with increasing age. The lack of turn in effect may be due to a possible underestimation of age at menopause in our cohort, as explained below. The minor contribution to breast cancer risk by other factors included in the multivariate model is supported by previous studies,26, 27, 28, 29 as is the distribution of these reproductive, anthropometrical and lifestyle characteristics by level of education.30, 31, 32, 33

A positive gradient in risk by level of education has been documented in one previous prospective study comprising both pre- and postmenopausal women.2 However, even after controlling for parity, age at first birth, status of menopause, weight and height, and consumption of alcohol, a borderline significant excess risk remained among highly educated women. The lack of agreement with our study could relate to the great difference in cohort size.

Although age at menopause was unknown for most of the cohort members, we performed analyses separated by menopausal status, using age 50 as an estimate when menstrual history was unavailable.34 This entails a possible misclassification that might have attenuated any true difference between pre- and postmenopausal women.35

We found indications of a slightly steeper increase in risk associated with educational level after menopause rather than before. The lack of previous published studies considering menopausal status hampers any comparison, while the few prospective studies examining only postmenopausal women show inconsistent results.17, 22

One possible explanation for the observed lack of consistency in relative risks in pre- and postmenopausal women is that the meaning of education length varies by birth cohort. Certain occupational groups correspond to different levels of education, according to age. Compulsory school expanded during adolescence of the study population, and several professions at a middle or lower level (such as nursing and teaching) have required more years of total education during the last decades than in earlier ones. Thus, in our cohort, the younger women of a given education group may be comparable to the older women within a lower group.

Another possible explanation for the more pronounced association between educational level and breast cancer risk after menopause observed in our study may also be a birth cohort effect: the distribution of reproductive and lifestyle behaviour has changed over time according to the educational level achieved. Because the younger women in our cohort were at reproductive ages at time of cohort entry, we cannot compare their reproductive pattern according to education in all age groups, but figures from the Norwegian Population Register show a narrowing gap by birth cohort between education levels according to both average number of children and childlessness.30 On the other hand, the disparity in age at first birth has widened between education groups during the last decades, as average age at first birth has increased in all groups.36 However, different age at first birth seems to give smaller differentials in risk than differences in parity.37, 38 Alcohol consumption, age at menarche, menopausal status at start of follow-up and proportion of women using hormonal contraceptives increased with increasing age in our study at all levels of education.

Risk pattern for breast cancer most probably also differs according to menopausal status. Family history of breast cancer, particularly breast cancer in young first-degree relatives, is a stronger determinant of premenopausal breast cancer risk.39 Hence, other behavioural and reproductive risk factors will be more prominent postmenopausally. Therefore, an additional reason why we were able to explain the positive relationship between education and breast cancer risk, moreso than previous studies, may be that our cohort is younger and was collected at a later time.

The strengths of our study include its prospective design, large size and complete follow-up. Our data offer sufficient variability in years of education as well as in related exposures to exhibit any differential in risk.

The use of self-reported information on education may represent a weakness of the study. Self-reported education often exceeds the number of years recorded in official statistics because the participants are likely to state both incomplete and informal training sessions. Moreover, as frequently observed in studies with volunteers, an over-representation of highly educated women as compared to the source population is present. The selection bias by education has been assessed in a part of the cohort by comparing the distribution of education among those who responded with the total invited sample using information from the Norwegian national register of education. Of the 11,600 women who responded, 26% had completed 13 or more years of education, compared to 22% in the invited sample of 18,900 women (our own unpublished data). However, since all comparisons we did in our analysis are within cohort members, we do not believe that selection bias affected any results.

Almost all studies on SES and breast cancer risk have reported a positive association irrespective of how SES was operationalised. Some of them combined education and income2, 3, 4, 5 or education and occupational or socioeconomic group.10–12, 40 Compared to income (measured as gross household income or poverty index ratio), years of education tends to be more strongly associated with risk.2–4, 17 Occupational class measures, however, generally provide a reinforced effect among the higher (professional) group.

There are several advantages of using years of education as a social class indicator. It applies to every adult individual, is more stable over one's lifetime than either occupation or income41 and is easy obtainable and recordable.40

When the objective of a study is to estimate risks in various social strata and further explain an observed social class gradient in risk, the benefit of an indicator also depends on its ability to discriminate across strata according to the present outcome, which is conditioned by its strength of association with underlying causal factors. Education may be the most relevant measure in the analysis of social class and breast cancer, owing to its close relationship with reproductive pattern.30, 36

The identification of the underlying factors that explain variations in risk by level of education also raises the question of whether it is still necessary to adjust for years of education in the analysis of breast cancer. We suggest that when information on reproductive factors and anthropometry is collected, it is superfluous to keep education as a covariate in the model, at least for young adults and middle-aged women. Since aetiological risk factors for breast cancer are probably similar in most populations, we believe that this statement can be applied in general.

We found a straight-line positive relationship between years of education and risk for breast cancer in a cohort of Norwegian and Swedish women at most 50 years old at enrolment, which can be fully explained by known risk factors. Dividing the analysis by pre- and postmenopausal follow-up time revealed a more pronounced relationship postmenopausally, but we were still able to identify the underlying differentials in exposure.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors certify that they have not entered into any agreement that could interfere with their access to the data on the research, nor upon their ability to analyse the data independently, to prepare articles and to publish them. We are grateful to M. Ustad and J. Mathiassen for their contribution in earlier phases of the study.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Kogevinas M, Pearce N, Susser M, Boffetta P. Social inequalities and cancer. IARC Scientific Publications No. 138; ed. Lyon: International Agency for Research on Cancer, 1997.
  • 2
    Carter CL, Jones DY, Schatzkin A, Brinton LA. A prospective study of reproductive, familial and socioeconomic risk factors for breast cancer using NHANES I data. Public Health Rep 1989; 104: 4550.
  • 3
    Baquet CR, Horm JW, Gibbs T, Greenwald P. Socioeconomic factors and cancer incidence among blacks and whites. J Natl Cancer Inst 1991; 83: 5517.
  • 4
    Devesa SS, Diamond EL. Association of breast cancer and cervical cancer incidence with income and education among whites and blacks. J Natl Cancer Inst 1980; 65: 51528.
  • 5
    Williams RR, Horm JW. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey. J Natl Cancer Inst 1977; 58: 52547.
  • 6
    Gorey KM, Holowaty EJ, Laukkanen E, Fehringer G, Richter NL. Association between socioeconomic status and cancer incidence in Toronto, Ontario: possible confounding of cancer mortality by incidence and survival. Cancer Prev Control 1998; 2: 23641.
  • 7
    Gomes AL, Guimaraes MD, Gomes CC, Chaves IG, Gobbi H, Camargos AF. A case-control study of risk factors for breast cancer in Brazil, 1978–1987. Int J Epidemiol 1995; 24: 2929.
  • 8
    Mackillop WJ, Zhang-Salomons J, Boyd CJ, Groome PA. Associations between community income and cancer incidence in Canada and the United States. Cancer 2000; 89: 90112.
  • 9
    Faggiano F, Zanetti R, Costa G. Cancer risk and social inequalities in Italy. J Epidemiol Community Health 1994; 48: 44752.
  • 10
    Barbone F, Filiberti R, Franceschi S, Talamini R, Conti E, Montella M et al. Socioeconomic status, migration and the risk of breast cancer in Italy. Int J Epidemiol 1996; 25: 47987.
  • 11
    Talamini R, La Vecchia C, Decarli A, Franceschi S, Grattoni E, Grigoletto E, Liberati A, Tognoni G. Social factors, diet and breast cancer in a northern Italian population. Br J Cancer 1984; 49: 7239.
  • 12
    Ewertz M. Risk of breast cancer in relation to social factors in Denmark. Acta Oncol 1988; 27: 78792.
  • 13
    Yost K, Perkins C, Cohen R, Morris C, Wright W. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control 2001; 12: 70311.
  • 14
    Krieger N, Quesenberry C, Jr, Peng T, Horn-Ross P, Stewart S, Brown S, Swallen K, Guillermo T, Suh D, Alvarez-Martinez L, Ward F. Social class, race/ethnicity, and incidence of breast, cervix, colon, lung, and prostate cancer among Asian, Black, Hispanic, and White residents of the San Francisco Bay Area, 1988–92 (United States). Cancer Causes Control 1999; 10: 52537.
  • 15
    Williams RR, Stegens NL, Goldsmith JR. Associations of cancer site and type with occupation and industry from the Third National Cancer Survey Interview. J Natl Cancer Inst 1977; 59: 114785.
  • 16
    Lynge E, Thygesen L. Occupational cancer in Denmark. Cancer incidence in the 1970 census population. Scand J Work Environ Health 1990; 16: 335.
  • 17
    Heck KE, Pamuk ER. Explaining the relation between education and postmenopausal breast cancer. Am J Epidemiol 1997; 145: 36672.
  • 18
    La Vecchia C, Negri E, Franceschi S. Education and cancer risk. Cancer 1992; 70: 293541.
  • 19
    Paffenbarger RS, Jr., Kampert JB, Chang HG. Characteristics that predict risk of breast cancer before and after the menopause. Am J Epidemiol 1980; 112: 25868.
  • 20
    Helmrich SP, Shapiro S, Rosenberg L, Kaufman DW, Slone D, Bain C, Miettinen OS, Stolley PD, Rosenshein NB, Knapp RC, Leavitt T Jr, Schottenfeld D, et al. Risk factors for breast cancer. Am J Epidemiol 1983; 117: 3545.
  • 21
    Tavani A, Braga C, La Vecchia C, Negri E, Russo A, Franceschi S. Attributable risks for breast cancer in Italy: education, family history and reproductive and hormonal factors. Int J Cancer 1997; 70: 15963.
  • 22
    Van Loon AJ, Goldbohm RA, Van den Brandt PA. Socioeconomic status and breast cancer incidence: a prospective cohort study. Int J Epidemiol 1994; 23: 899905.
  • 23
    Lunde AS, Lundeborg S, Lettenstrom GS, Thygesen L, Huebner J. The person-number systems of Sweden, Norway, Denmark and Israel. Vital Health Stat 2 1980; 2: 159.
  • 24
    Adami HO, Hunter D, Trichopoulos D. Textbook of cancer epidemiology. New York: Oxford University Press, Inc., 2002.
  • 25
    Van den Brandt PA, Spiegelman D, Yaun SS, Adami HO, Beeson L, Folsom AR, Fraser G, Goldbohm RA, Graham S, Kushi L, Marshau JR, Miller AB, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol 2000; 152: 51427.
  • 26
    Monninkhof EM, van der Schouw YT, Peeters PH. Early age at menopause and breast cancer: are leaner women more protected? A prospective analysis of the Dutch DOM cohort. Breast Cancer Res Treat 1999; 55: 28591.
  • 27
    Titus-Ernstoff L, Longnecker MP, Newcomb PA, Dain B, Greenberg ER, Mittendorf R et al. Menstrual factors in relation to breast cancer risk. Cancer Epidemiol Biomarkers Prev 1998; 7: 7839.
  • 28
    Kumle M, Weiderpass E, Braaten T, Persson I, Adami HO, Lund E. Use of oral contraceptives and breast cancer risk: the Norwegian-Swedish women's lifestyle and health cohort study. Cancer Epidemiol Biomarkers Prev 2002; 11: 137581.
  • 29
    Smith-Warner SA, Spiegelman D, Yaun SS, Van den Brandt PA, Folsom AR, Goldbohm RA, Graham S, Holmberg L, Howe GR, Marshall JR, Miller AB, Potter JD, et al. Alcohol and breast cancer in women: a pooled analysis of cohort studies. JAMA 1998; 279: 53540.
  • 30
    Kravdal O. The high fertility of college educated women in Norway. Demographic Res 2001; 5: 187215.
  • 31
    Abramson JH, Gofin R, Habib J, Pridan H, Gofin J. Indicators of social class: a comparative appraisal of measures for use in epidemiological studies. Soc Sci Med 1982; 16: 173946.
  • 32
    Jacobsen BK, Lund E. Level of education, use of oral contraceptives and reproductive factors: the Tromso Study. Int J Epidemiol 1990; 19: 96770.
  • 33
    Jacobsen B, Nilsen H. High education is associated with low fat and high fibre, beta-carotene and vitamin C. Norsk Epidemiologi 2000; 10: 5762.
  • 34
    Morabia A, Flandre P. Misclassification bias related to definition of menopausal status in case-control studies of breast cancer. Int J Epidemiol 1992; 21: 2228.
  • 35
    Rothman KJ. Epidemiology: an introduction. New York: Oxford University Press, 2002.
  • 36
    Lappegård T. New fertility trends in Norway. Demographic Res 2000; 2: 3.
  • 37
    Albrektsen G, Heuch I, Tretli S, Kvale G. Breast cancer incidence before age 55 in relation to parity and age at first and last births: a prospective study of one million Norwegian women. Epidemiology 1994; 5: 60411.
  • 38
    Kvale G, Heuch I. A prospective study of reproductive factors and breast cancer. II. Age at first and last birth. Am J Epidemiol 1987; 126: 84250.
  • 39
    Collaborative Group on Hormonal Factors in Breast Cancer. Familial breast cancer: collaborative reanalysis of individual data from 52 epidemiological studies including 58,209 women with breast cancer and 101,986 women without the disease. Lancet 2001; 358: 138999.
  • 40
    Van Loon AJ, Brug J, Goldbohm RA, Van den Brandt PA, Burg J. Differences in cancer incidence and mortality among socio-economic groups. Scand J Soc Med 1995; 23: 11020.
  • 41
    Susser MW, Watson W, Hopper K. Sociology in medicine, 3 ed. New York: Oxford University Press., 1985.