Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention

Transcription

1 Cancer Epidemiology 33 (2009) Contents lists available at ScienceDirect Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention journal homepage: Education and cancer incidence in a rural population in south India Rajaraman Swaminathan a, Ramanujam Selvakumaran a, Jissa Vinodha b, Jaques Ferlay c, Catherine Sauvaget c, Pulikattil Okkuru Esmy b, Viswanathan Shanta a, Rengaswamy Sankaranarayanan c, * a Cancer Institute (WIA), Chennai , India b Christian Fellowship Community Health Center, Ambilikkai , India c International Agency for Research on Cancer, Lyon 69008, France ARTICLE INFO ABSTRACT Article history: Accepted 29 June 2009 Keywords: Cancer Population Registry Epidemiology Education Incidence Risk Background: Population-based studies describing the association between education and cancer incidence has not yet been reported from India. Methods: Information on the educational attainment of 4417 cancer cases aged 14 years and above, diagnosed during in Dindigul district, Tamil Nadu, India, was obtained from the Dindigul Ambilikkai Cancer Registry, which registers invasive cancer cases by active methods from 102 data sources. Population distribution by 5-year age groups and for four educational levels namely no education, education 5 years, 6 12 years and >12 years, was obtained from census data. Standardized rate ratios based on age-standardized rates were calculated to study cancer risks for different educational levels. Results: Men and women with no education had higher overall cancer incidence rates compared to the educated population. The risk of cervix, mouth, esophagus, stomach and lung cancers were inversely associated with higher levels of education whereas a high incidence of breast cancer was observed with increasing educational levels. The standardized rate ratio of cervical cancer 0.32 (95% CI: ) and of breast cancer was 6.08 (95% CI: ) for women with more than 12 years of education compared to those with no education. There was paucity of cases in the highest education level for most cancers. Conclusion: With more and more women in rural India becoming educated, one could foresee breast cancer becoming more frequent even in rural areas of India in future. ß 2009 Elsevier Ltd. All rights reserved. 1. Introduction Educational attainment is an important indicator of socioeconomic status through its association with occupation, disposable income, residential and life-style factors. A number of studies in developed countries have shown that cancer incidence varies between people with different levels of education [1 4]. A high incidence of breast, large bowel, melanoma and prostate cancers has been found among those with high levels of education whereas an inverse association has been found for the incidence of cancers of the upper aero digestive tract, esophagus, stomach, lung and uterine cervix [1 4]. Such differences in cancer risks associated with educational attainment may reflect the differences in life-style factors and exposure to both environmental and work related carcinogens. While consistent patterns of cancer risk and educational correlates in general populations have been amply documented * Corresponding author at: Section of Early Detection & Prevention, International Agency for Research on Cancer, 150 cours Albert-Thomas, Lyon Cedex 08, France. Tel.: ; fax: addresses: sankar@iarc.fr, bayle@iarc.fr (R. Sankaranarayanan). in developed countries, such information from less developed is not widely available [5,6]. Population-based studies from less developed countries had so far reported only on the proportions of specific cancers to total cancers for different categories of literacy levels [7]. The aim of this study is to describe the association between cancer incidence and educational levels in a rural population in South India, using cancer incidence data from a population-based cancer registry and census data on education levels. 2. Materials and methods All incident cancer cases, occurring in the resident population of predominantly rural and agrarian Dindigul district, Tamil Nadu state, South India, have been actively registered by the Dindigul Ambilikkai population-based cancer registry (DACR) since January 1, This district has a population of 2 million with a sex ratio of 990 women to 1000 men living in 6267 km 2 of surface area. The DACR employs active case finding methods, as advocated by the International Association of Cancer Registries (IACR), to identify and register incident cancer cases [8]. The registry staff visit 102 data sources (70 sources outside the district) including public and private hospitals, clinics, hospices, imaging centers and pathology /$ see front matter ß 2009 Elsevier Ltd. All rights reserved. doi: /j.canep

2 90 R. Swaminathan et al. / Cancer Epidemiology 33 (2009) laboratories where cancer cases are likely to be diagnosed and treated; they also visit the death registry offices in the district to collect information on cancer deaths. A standardized proforma is used to collect information on personal identifiers, education, disease, treatment and outcome. Data are collected by abstracting the information available in the case records in the data sources and also by direct interview of patient and/or household members of the patient wherever possible. The primary site of cancer and morphology were coded using International Classification of Diseases for Oncology, third edition (ICD-O III) [9]. The CONVERT [10] software developed by International Agency for Research on Cancer was then used to get ICD-10 disease codes [11]. Incident cancer cases were then pooled after elimination of duplicate registrations at the end of every year and house visits were undertaken for completion of any missing details of the registered cancer cases. A total of 4516 incident cancers of all sites for all ages were registered by the DACR during Incident cancer cases and population in the age group of 0 14 years were excluded from this study, and 1938 men and 2479 women with cancers comprising uterine cervix (916), female breast (446), stomach (319), mouth (269), esophagus (140), ovary (133), lung (123) and other sites were included in this study. The estimation of Dindigul district population by 5-year age group and sex, annually between 2003 and 2006, was done by assuming exponential growth rate based on the census populations of years 1991 and 2001 [12,13]. Information on education levels of the population was obtained from the census data [13]. Educational attainment of cancer patients and the population of the district were classified as no education, 1 5 years of schooling, 6 12 years of schooling and more than 12 years of education. The populations estimated annually during were added to estimate the person years for the calculation of incidence rates per 100,000 person years. Age-standardized incidence rates for each education level were calculated by direct standardization using the world standard population [14]. Rate ratios based on agestandardized rates were calculated to obtain the breast and cervical cancer risks by education levels. 3. Results The distribution of the mid-period population by different levels of education and sex are given in Table 1. Almost half of the Table 1 Distribution of mid-period population by educational level and sex in Dindigul district among subjects aged 15 years and above (as on January 1, 2005). Educational level Men Women Number % Number % Nil 169, , , , , , >12 30, , Total 739, , women (47.3%) and a quarter of men (23%) had no education and less than 5% had more than 12 years of schooling. Education specific crude incidence rates for all cancer sites combined and for selected cancer sites in men and in women are given in Tables 2 and 3 respectively. The number of cases for the highest education category was scanty. The all cancer incidence rate for men with no education (148.1) was more than two- to three-fold higher than men in the different education categories. A decreasing incidence of stomach, mouth, esophagus and lung cancers was observed with increasing levels of education; men with no education had an eight-fold increased incidence rate of stomach cancer (20.1) and fourteen-fold increased incidence rate of esophageal cancer (11.2) compared to those with more than 12 years of schooling (Table 2). The all cancer incidence rate among women with no education (111.8) was almost two-fold higher than those with 1 12 years of schooling; the incidence rates of cervical cancer strikingly declined, whereas those of breast cancer increased with increasing levels of education (Table 3). These two cancers together constituted almost half of all cancers in all educational levels. The education specific age-standardized incidence rates and standardized rate ratios (SRR) with 95% confidence intervals for breast and cervix cancers are given in Table 4. There was a significant inverse association between the risk of cervical cancer and increasing educational attainment. Cervical cancer risk was significantly reduced by 68% among women with the highest educational attainment and there was a significant 39 43% reduction in the risks of cervical cancer among those with 1 12 years of schooling compared to women with no education. The risk Table 2 Education specific crude incidence rates of major cancers, Dindigul Ambilikkai Cancer Registry, , men. Educational level Education specific crude incidence rate per 100,000 person years (number of incident cancer cases) All sites (1938) Stomach (218) Mouth (161) Esophagus (140) Lung (123) Nil (1019) 20.1 (138) 11.3 (78) 11.2 (77) 9.4 (65) (443) 3.4 (38) 4.6 (51) 3.3 (37) 2.2 (25) (393) 3.8 (39) 2.6 (27) 2.4 (25) 2.6 (27) > (83) 2.5 (3) 4.2 (5) 0.8 (1) 5.0 (6) Note: Italicized values are based on scanty numbers. Table 3 Education specific crude incidence rates of major cancers, Dindigul Ambilikkai Cancer Registry, , women. Education level Education specific crude incidence rate per 100,000 person years (number of incident cancer cases) All sites Cervix (916) Breast (446) Ovary (133) Mouth (108) Stomach (101) Nil (1590) 48.1 (684) 10.8 (153) 10.2 (215) 5.6 (80) 5.4 (77) (458) 17.8 (142) 17.8 (142) 2.8 (53) 2.1 (17) 1.9 (15) (359) 11.9 (84) 17.3 (122) 2.6 (45) 1.4 (10) 0.9 (6) > (72) 8.6 (6) 41.5 (29) 3.2 (6) 1.4 (1) 4.3 (3) Note: Italicized values are based on scanty numbers.

3 R. Swaminathan et al. / Cancer Epidemiology 33 (2009) Table 4 Education specific age-standardized rate (ASR) and standardized rate ratio (SRR) with 95% CI of cervical and breast cancers in Dindigul Ambilikkai Cancer Registry, Education level (in years) Cancer cervix Cancer breast ASR SRR a 95% CI ASR SRR a 95% CI No education b b 5 years , , years , , 7.76 >12 years , , All education levels ASR: Age-standardized rate; SRR: standardized rate ratio; CI: confidence interval. a p-value for trend < b Reference category. for breast cancer increased with increasing levels of education with a two- to six-fold elevated risk among different levels of educated women. The age specific incidence rates for women with no education and for educated women (by combining the three categories of education) for cervical and female breast cancers are shown in Fig. 1. The rates among uneducated compared with educated women were consistently higher for cervical and lower for breast cancers. 4. Discussion The overall cancer incidence in men and women followed a U shaped pattern in our study, with higher incidence among men and women with no education, low rates for those with 1 12 years of schooling and a high incidence for those with more than 12 years of schooling. However, the overall cancer incidence rate for men with no education was two-fold higher than that of those with highest educational attainment. A high risk of stomach, mouth, esophagus and lung cancers seems to contribute to this high overall burden in men with no education. Thehighburdenofcanceramongwomenwithnoeducationis predominantly contributed by the high risk of cervical cancer, whereas the high risk of breast cancer accounted for the high burden among women with high educational attainment. The identification of life-style and behavioral factors contributing to the low risk of cancer in the subset of subjects with 1 12 years of schooling is likely to provide valuable leads to cancer control. Our results on the association between education and incidence of cancers in individual cancer sites are consistent with observations in developed countries and largely in directions that can be predicted from literature on socioeconomic gradients of cancer incidence [1 4,15]. We found an inverse association between increasing educational levels and incidence of mouth, esophagus, lung and cervical cancers, whereas a positive association was found for breast cancer incidence. The paucity of cases in the highest education group for most cancers has to be borne in mind and results have to be interpreted with caution. The findings of a negative association between educational level and the risk of cervical cancer are in accordance with other studies [1 4,16]. The main explanations for the educational gradient in cervical cancer risk are related to reproductive factors and sexual behavior. A prevalence survey from the Dindigul district revealed that the high-risk human papillomavirus (HPV) infection, which is necessary for the Fig. 1. Age-specific incidence by education status, Dindigul Ambilikkai Cancer Registry, Description: The figure gives the distribution of age (x-axis) specific incidence rate (y-axis) observed among uneducated and educated (legend) for cervical and female breast cancers separately.

4 92 R. Swaminathan et al. / Cancer Epidemiology 33 (2009) development of cervical cancer, was inversely associated with education [17]. The positive relationship between increasing educational level and breast cancer incidence is well known [1 4]. Among the factors associated with an increased risk of breast cancer are nulliparity or low parity, higher age at first childbirth, lower age at menarche and higher age at menopause whereas breast feeding and physical activity are associated with a decreased risk. Higher education is usually associated with postponement and reduction of childbirths, which in turn increase the risk of breast cancer. In a Nordic study, the educational gradient in breast cancer risk could be explained by established breast cancer risk factors [4]. Education is a good surrogate for many aspects of the socioeconomic status and social classes. Whereas an incident cancer may lead to downward occupational mobility and reduced income, educational status achieved in early adulthood is not affected by it, thus avoiding reverse causation bias. A number of demographic, behavioral and biologic factors, including smoking, nutrition, energy balance, cancer detection and reproductive factors are likely to act in the causal pathway between education and cancer risk. It has been shown that inflammation biomarkers such as high levels of interleukin-6, C-reactive protein and tumor necrosis factor-alpha are potentially causal with respect to cancer and overall mortality are inversely associated with education [18,19]. Behavioral factors such as smoking, drinking, obesity explained a substantial part of this inverse association [18]. Unhealthy behaviors have been shown to account for the association between education and cancer incidence in a Nordic study [4]. Although these factors may potentially explain the education-cancer connection, they do not in anyway underemphasize the public health importance of educational attainment as a predictor of cancer patterns. The information on socio-economic status in the whole population, especially the family income, was not routinely available in the census data in India. But data on educational level were collected from the entire population during census operations. Also, using educational level as a surrogate for socioeconomic status substantially reduced the problem of misclassification at the population level than any other indirect method based on residential area. In DACR, educational level was one of the socio-cultural factors on which data were collected by trained cancer registrars during house visits through direct interview of every cancer case/relative as part of registration procedure. This ensured unambiguous data collection and enhanced data quality in DACR. The exclusion of childhood population of 0 14 years of age from this analysis was justified by our perception that education status would be apt to serve as a surrogate for socio-economic status only among an adult population. Cervical and breast cancers together constituted about 55% of all cancers in women and the incidence rates of both in the years age group were higher than in the 65+ years in all the rural areas in India [20]. Our study had now shown a 68% reduced risk of cervical cancer and six-fold higher risk of breast cancer among women with the highest level of education compared to uneducated women. The role of education in explaining the socio-economic status differences in breast cancer occurrence with the highest risk observed among those with more than 12 years of education had been observed earlier in a developed environment [21]. Furthermore, the persistence of differentials in reproductive behavior by social class across countries with low and high levels of modernization and fertility had been documented [22]. We believe that education status clearly reflected the differences in the profile of reproductive factors such as age at first childbirth and number of children, between women belonging to different education levels possibly representing different socio-economic status in Dindigul district. With the demographic transition, time trends in social distribution and the continuing process of more and more women in rural India becoming educated, one could foresee breast cancer becoming more frequent even in rural areas of India in future, as already observed in urban registries in India [20]. The implications of the present study are clear that cancer control efforts in India should take into account the changing profile of cancers in relation to education and socio-economic progress and incorporate life-style intervention measures accordingly, including cessation of tobacco use in any form, dietary interventions and promotion of physical activity; cancer early detection programs must offer interventions covering both cervical and breast cancers. Conflict of interest None. Acknowledgements Thanks are due to the heads, officers and other staff at all the sources of data collection for DACR and officials in the ministry of Health and Family Welfare, government of Tamil Nadu and its affiliated departments, for their assistance, cooperation and support. We are grateful to the Bill & Melinda Gates Foundation for their generous support through the Alliance for Cervical Cancer Prevention. R. Swaminathan carried out the analysis of data during his post-doctoral fellowship at the International Agency for Research on Cancer, Lyon, France. References [1] Hemminki K, Li X. Level of education and the risk of cancer in Sweden. Cancer Epidemiol Biomarkers Prev 2003;12: [2] Vidarsdottir H, Gunnarsdottir HK, Olafsdottir EJ, Olafsdottir GH, Pukkala E, Tryggvadottir L. Cancer risk by education in Iceland; a census based cohort study. Acta Oncol 2008;47(3): [3] Mouw T, Koster A, Wright ME, Blank MM, Moore SC, Hollenbeck A, et al. Education and risk of cancer in a large cohort of men and women in the United States. PLoS ONE 2008;3(11):e3639. doi: /journal.pone [4] Braaten T, Weiderpass E, Kumle M, Lund E. Explaining the socioeconomic variation in cancer risk in the Norwegian Women and Cancer Study. Cancer Epidemiol Biomarkers Prev 2005;14(11 Pt 1): [5] Bhattacharyya SK, Basu S, Banerjee S, Dastidar AG, Bagchi SR. An epidemiological survey of carcinoma cervix in north Bengal zone. J Indian Med Assoc 2000;98(2). pp. 60-1, 66. [6] Capalash N, Sobti RC. Epidemiology of cervical cancer a case control study on north Indian population. Indian J Cancer 1999;36(2 4): [7] Kurkure AP, Yeole BB. Social inequalities in cancer with special reference to south Asian countries. Asian Pac J Cancer Prev 2006;7: [8] Jensen OM, Parkin DM, Maclennan R, Muir CS, Skeet RG. Cancer registration principles and methods. IARC scientific publications no. 95. Lyon: International Agency for Research on Cancer, [9] World Health Organization. International classification of diseases for oncology ICD-O, 3rd ed.. Geneva: World Health Organization, [10] Ferlay J, Burkhard C, Whelan S, Parkin DM. Check and conversion programs forcancerregistries(iarc/iacrtoolsforcancerregistries).iarctechnical report no. 42. Lyon: International Agency for Research on Cancer; [11] World Health Organization. International statistical classification of diseases and related health problems, tenth revision, vol. 1. Geneva: World Health Organization, [12] Census of India Social and cultural tables, series-23, Tamil Nadu: Part IV- A, Table C-6. Chennai: Directorate of Census Operations; [13] Census of India Social and cultural tables, series-33, Tamil Nadu: Table C-8. Chennai: Directorate of Census Operations; [14] Curado MP, Edwards B, Shin HR, Storm H, Ferlay J, Heanue M, et al. Cancer incidence in five continents, vol. IX. IARC scientific publications no Lyon: International Agency for Research on Cancer, [15] Louwman WJ, van Lenthe FJ, Coebergh JW, Mackenbach JP. Behavior partly explains educational differences in cancer incidence in the southeastern Netherlands: the longitudinal GLOBE study. Eur J Cancer Prev 2004;13(2): [16] Parikh S, Brennan P, Boffetta P. Meta-analysis of social inequality and the risk of cervical cancer. Int J Cancer 2003;105(5):

5 R. Swaminathan et al. / Cancer Epidemiology 33 (2009) [17] Franceschi S, Rajkumar R, Snijders PJ, Arslan A, Mahé C, Plummer M, et al. Papillomavirus infection in rural women in southern India. Br J Cancer 2005;92(3): [18] Koster A, Bosma H, Penninx BW, Newman AB, Harris TB, van Eijk JT, et al. Association of inflammatory markers with socioeconomic status. J Gerontol A Biol Sci Med Sci 2006;61: [19] Coussens LM, Werb Z. Inflammation and cancer. Nature 2002;420: [20] National Cancer Registry Program. Consolidated report of population based cancer registries, New Delhi: Indian Council of Medical Research; [21] Dano H, Hansen KD, Jensen P, Petersen JH, Jacobsen R, Ewertz M, et al. Fertility pattern does not explain social gradient in breast cancer in Denmark. Int J Cancer 2004;111(3): [22] Santos Silva ID, Beral V. Socioeconomic differences in reproductive behavior. IARC Sci Publ 1997;138: