ORIGINAL ARTICLE BMI and Breast Cancer in Korean Women: A Meta-Analysis Dukyoo Jung 1, Sun-Mi Lee 2 * 1 Division of Nursing Science, College of Health Sciences, Ewha Womans University, Seoul, Korea 2 College of Nursing, The Catholic University of Korea, Seoul, Korea Introduction The number of breast cancer women has increased dramatically in Korea. The cause is perceived to stem from adaptation to a westernized life style which increases body mass index (BMI). However, there are no meta-analysis data available that could help in understanding the relationship between Korean females BMI and breast cancer occurrence. Method All the published articles that investigated the relationship of Korean women s BMI with breast cancer prevalence between 1950 and 2007 were included in this study, based on a screen of the computerized databases that search for these articles (MEDLINE, RISS4U and KMBase). The commercial software Comprehensive Meta Analysis was used for the analysis. Results The high BMI score group presented a higher prevalence of breast cancer on both a fixed-effects model [odds ratio (OR) = 1.282; 95% confidence interval (CI) = 1.209, 1.361] and a random-effects model (OR = 1.388; 95% CI = 1.129, 1.706). In addition, a high BMI score on pre- and postmenopausal groups was found to have a significantly higher prevalence of breast cancer on both a fixed-effects model (OR = 1.467; 95% CI = 1.268, 1.698, OR = 1.614; 95% CI = 1.360, 1.917, pre- and postmenopausal, respectively) and a random-effects model (OR = 1.387; 95% CI = 1.134, 1.696, OR = 1.681; 95% CI = 1.149, 2.461, pre- and postmenopausal, respectively). Conclusion This meta-analysis of Korean women showed that a high BMI was related to a higher incidence rate of breast cancer. This study used a subgroup analysis of pre- and postmenopausal groups; the high BMI subset in both the pre- and postmenopausal groups was shown to have a higher incidence rate of breast cancer. [Asian Nursing Research 2009;3(1):31 40] Key Words body mass index, breast cancer, Korean women, meta-analysis, systematic review INTRODUCTION Breast cancer is one of the most common types of cancer and affects millions of women around the world, with a noticeable fatality rate. The prevalence rate of breast cancer, particularly in Korean females, has shown a spike in recent years. In 2002, breast cancer was reported as the most frequent form of cancer being treated in women (National Cancer Center, 2002), and in 2003 the death rate associated with the cancer ranked second among various types of cancers affecting female patients (Korean National *Correspondence to: Sun-Mi Lee, College of Nursing, The Catholic University of Korea, Banpo-dong, Seochu-gu, Seoul, Korea 137-701. E-mail: leesunmi@catholic.ac.kr Received: January 21, 2009 Revised: January 30, 2009 Accepted: March 3, 2009 Asian Nursing Research March 2009 Vol 3 No 1 31
D. Jung, S.M. Lee Statistical Office, 2003). Numerous factors have been identified as major contributing factors of breast cancer in women. These include race (Axelrod et al., 2008), age (Gloecker-Reis, Pollack, & Young, 1983; Muller, Ames, & Anderson, 1978; Sweeney, Blair, Anderson, Lazovich, & Folsom, 2004), family history (Axelrod et al., 2006; Sweeney et al.), older age of menopause (Muller et al.), estrogen receptors (Axelrod et al., 2006; Clark, McGuire, Hubay, Peterson, & Marshall, 1983; Manni, 1983), weight (Song, Sung, & Ha, 2008; Yoo, 2003) and decreased birth rates and breast-feeding (National Cancer Center, 2002;Yoo, Yoo et al., 2002). Obesity is another important factor (Hede, 2008; Hjartaker, Langseth, & Weiderpass, 2008) to consider. When viewed in the light of recent changes in the diet of Korean women, i.e., westernization, and subsequent increases in fast food consumption and greasy, fatty food intake, obesity is a significant element contributing to the female population s rising body mass index (BMI) and potentially to an increased risk of breast cancer diagnosis (Bissonauth, Shatenstein, & Ghadirian, 2008; Hjartaker et al.). Another major factor associated with obesity is a lack of exercise due largely to a sedentary lifestyle (Carpenter, Ross, Paganini-Hil, & Bernstein, 2003; Jakicic & Otto, 2006; Maruti, Willett, Feskanich, Rosner, & Colditz, 2008). Given the considerably high level of prevalence and fatalities observed in Korean women, a more accurate causal analysis is urgently needed. However, little or no metaanalysis data are available, where these could help to better understand the relationship between BMI and breast cancer occurrence in Korean females. Therefore, the purpose of this study was to investigate the relationships among Korean women s BMI and breast cancer occurrence, and to determine through meta-analysis if obesity during the pre- or postmenopausal stage was related to breast cancer occurrence. METHODS In this study, the relationship between BMI and breast cancer among Korean women was investigated, as examined through meta-analysis while executing the guideline provided in the MOOSE (Meta-analysis of Observational Studies in Epidemiology Group; Stroup et al., 2000). Research strategies All the published articles that investigated the relationship of Korean women s BMI with breast cancer prevalence between 1950 and September 2006 from the authors previous work (Lee, Park, & Park, 2008) and between October, 2006 and November, 2007 from this study were included. A screen of the computerized databases was conducted to search for these articles. Those databases were Medline, RISS4U, which is the typical database supplied by KERIS (Korean Education & Research Information Service, http://www.riss4u.net) in Korean Healthcare Academia and the Korean Medical Database (KMBase; http://kmbase.medric.or.kr). The keywords used for all areas were breast cancer or breast neoplasm and Korea or Korean in the Medline database, and the keywords of breast cancer or breast neoplasm were used for title screening in the local database. The articles were selected through title and abstract screening from breast cancer-related articles among domestically and internationally published studies with Koreans as subjects. The secondary abstract review was performed on the first screened articles with review of the full text. The selection procedure was managed by two investigators with consensus required, when they integrated the different views on particular articles. Inclusion criteria and study selection The eligible articles for the meta-analysis included Korean women living in Korea, and the studies examined the relationship between breast cancer and obesity (BMI). The selected study design included was the observational study (cross-sectional, casecontrol and cohort study). The research that included males or patients living in a foreign country, or required an invasive test to verify a risk factor such as genetic factor, were excluded from this study. In addition, non-observational studies and dual-published research were not included. 32 Asian Nursing Research March 2009 Vol 3 No 1
BMI and Breast Cancer in Korea: A Meta-Analysis Selected observational studies Figure 1 shows the selection process of how the final 26 articles were selected as subjects for the metaanalysis. The 25 papers, which investigated BMI as a breast cancer risk factor, were selected from 34 papers in the authors previous work (Lee et al., 2008). We originally retrieved those articles for systematic review of the breast cancer risk factors in Korean women between January, 1950 and September, 2006. For the meta-analysis, one additional paper published from September, 1996 to November, 2007 was selected by following the exactly same search strategies as the authors previous work. As shown in Figure 1, a total of additional 213 were initially screened: 175 articles on RISS4U & KMBase and 38 articles on MEDLINE. Nine articles were selected based on a screening through the title and abstract review. Then, one article was selected, after eight articles out of the initial nine articles were excluded based on full text review. Data extraction The selected articles were analyzed with checklists. The contents of the checklists were as follows: year of publication, type of publication, research question or purpose, study design, data collection method including response format (e.g., questionnaire, etc.), study sites (or region), source and number of study subjects, inclusion and exclusion criteria of subjects, demographics of the subjects, period of study, response rate, sampling methods, possible sources of bias, confounding variables, analytic methods, number of subgroups according to BMI cut-off, effect size and p values, 95% CI and adjustment factors. In particular, BMI categories used in the 26 studies were generally binary, tertiary, quartile or quintiles. Therefore, a BMI score of 25, which was generally accepted as normal, was decided as the cut-off point, and each effect size was recalculated after transformation to a binary category (WHO, 2008). All data from the studies and quality evaluation were independently rated by the two reviewers. The agreement rate to determine the reliability of the ratings was 92.4% and disagreements were adjusted through discussion. Statistical method Comprehensive Meta Analysis, a commercial software program, was used for the analysis. Heterogeneity was Period: 1996 to September, 2006 (October, 2006 to November, 2007) Papers retrieved from the databases = 2,199 (213) Papers from RISS4U & KMBase = 2,090 (175) Papers from MEDLINE = 109 (38) Paper excluded on the basis of titles and abstracts = 2,136 (204) Full text retrieved for more detailed evaluation = 63 (9) Papers excluded on exclusion criteria = 29 (6) Observational studies selected = 34 (3) Non BMI measure papers excluded = 9 (2) The papers finally selected for BMI meta-analysis = 25 (1) Figures in brackets = additional search from October, 2006 to November, 2007 Figure 1. Flow diagram of article selection. Asian Nursing Research March 2009 Vol 3 No 1 33
D. Jung, S.M. Lee analyzed with Q statistics based on the calculated values at a.1 level of significance, and the I 2, which represent the percentage of degree of variation among selected studies. When heterogeneity did not exist, the I 2 score was 0, with a larger I 2 score presenting a higher probability of heterogeneity. Publication bias was tested for by Funnel Plot and Begg s & Egger s tests. Because obesity (BMI > 25) is more likely to occur during the postmenopausal stage where there is an increased risk for breast cancer, pre- and postmenopausal subgroup analysis was also performed (Song et al., 2008). RESULTS General description The articles used for the meta-analysis are shown in Table 1. A total of 26 articles were included for the meta-analysis, and those consisted of 20 periodic articles (76.9%), four PhD dissertations (15.4%) and two Master theses (7.7%). A total of 8,559 cases and 63,605 controls were included for the analysis. In terms of study design, there were 25 case control studies and one cohort study out of the 26 studies. Eight of those studies were conducted in the 1990s, and the remaining 18 studies were performed in the 2000s. Fifteen studies out of those 26 articles focused on BMI and prevalence of breast cancer between the pre- and postmenopausal groups, but only 11 studies drew the results by classifying the pre- and postmenopausal group into subgroups. Therefore, the results from those 11 studies were applied to the meta-analysis. Meta analysis A high BMI score group presented a higher prevalence of breast cancer in both a fixed-effects model odds ratio (OR) = 1.282; 95% confidence interval (CI), 1.209 to 1.361; p =.000) and a random-effects model (OR = 1.388; 95% CI, 1.129 to 1.706; p = ), as shown in Table 2 and Figure 2. Also, a high BMI score for pre- and postmenopausal groups was found to have a significantly higher prevalence of breast cancer in both a fixed-effects model (OR = 1.467; 95% CI, 1.268 1.698, OR = 1.614; 95% CI, 1.360 1.917, pre- and postmenopausal, respectively) and a randomeffects model (OR = 1.387; 95% CI, 1.134 1.696, OR = 1.681; 95% CI, 1.149 2.461, pre- and postmenopausal, respectively). The homogeneity was analyzed with Q statistics and I 2. None of the 26 articles homogeneity was found to be statistically significant, as shown in Table 3 (Q = 255.350; p =.000; I 2 = 90%). The premenopausal group analyzed from the 11 articles was found to have relative homogeneity (Q = 15.158; p =.126; I 2 = 34%), but the findings of the postmenopausal group was revealed to have the opposite result (Q = 43.337; p =.000; I 2 = 77%). The results of the Funnel Plot and Random Effect assumption to check publication bias are shown in Figure 3 because the homogeneity was not satisfied. The Funnel Plot was shown with relative symmetry, and there were large studies among the 11 studies. Begg s & Egger s tests failed to identify publication bias. For the 26 articles, the p value of Begg s test was.537 and the p value of Egger s test was.513. No bias was found for either premenopausal status (Begg s test, p =.350; Egger s tests, p =.183) or postmenopausal status (Begg s test, p = 1.00; Egger s tests, p =.706), as shown in Table 4. To measure the stability of the result, sensitivity analysis was used. As shown in Figure 4, each p value of the 25 articles (except for the subjected article) was calculated for overall effect size and statistical significance, and the stability of this study was confirmed accordingly. DISCUSSION A meta-analysis of western women previously reported that obesity might increase the incidence rate of breast cancer in general (Connolly et al., 2002; Harvie, Hooper, & Howell, 2003). Similarly, this meta-analysis of Korean women showed that a high BMI was more likely to be related to a higher incident rate of breast cancer. This study used a subgroup analysis of pre- and postmenopausal groups; the high BMI group for both 34 Asian Nursing Research March 2009 Vol 3 No 1
BMI and Breast Cancer in Korea: A Meta-Analysis Table 1 Observational Studies Included in the Meta-analysis First author Case Control Subgroup analysis- BMI cut off 95% Study design Odds ratio confidence (year) number (n) number (n) menopausal status Reference Comparison interval Ahn (1999) 683 501 Case-control Pre-post < 25.6 25.6 1.717 1.234 2.389 Choi (2004) 22 22 Case-control < 23 23 1.204 0.365 3.974 Chun (2004) 97 97 Case-control Pre-post < 25 25 1.205 0.774 1.877 Do (2000) 108 121 Case-control Pre-post < 25 25 1.156 0.686 1948 Do (2003) 108 121 Case-control Pre-post < 25 25 1.082 0.637 1.839 Jeon (1991) 106 110 Case-control < 25 25 1.045 0.502 2.174 Joo (1994) 162 150 Case-control Pre-post < 25 25 5.693 3.857 8.403 Kang (1998) 146 146 Case-control < 26 26 1.299 0.734 2.299 Lee (2005) 103 159 Case-control Pre-post < 25 25 1.996 1.272 3.131 Lee (2003) 891 713 Case-control < 25 25 0.426 0.349 0.521 Lee (2001) 108 131 Case-control < 24 24 1.850 1.082 3.163 Lim (2005) 64,149 Cohort < 23.9 23.9 1.020 0.855 1.215 Moon (1997) 128 251 Case-control Pre-post < 25.6 25.6 1.667 1.201 2.313 Park (2003) 1,136 1,136 Case-control Pre-post 22 > 22 1.715 1.517 1.938 Shin (1995) 206 214 Case-control < 25 25 2.859 1.723 4.746 Shin (2000) 210 249 Case-control < 25 25 2.967 1.833 4.802 Suh (1995) 190 190 Case-control < 25 25 0.801 0.544 1.180 Yoo (1993) 162 150 Case-control < 25 25 1.640 1.025 2.623 Yoo (1998) 280 930 Case-control Pre-post < 25 25 1.325 1.001 1.756 Yoon (2000) 72 86 Case-control < 25.6 25.6 0.330 0.087 1.250 Choi (2001) 1,011 1,011 Case-control < 25 25 1.143 0.932 1.402 Choi (2003) 346 377 Case-control Pre-post < 26 26 0.967 0.746 1.253 Do (2003) 224 250 Case-control < 25 25 1.122 0.777 1.619 Lee (2005) 574 502 Case-control < 25 25 1.500 1.070 2.103 Park (2000) 189 189 Case-control Pre-post 26 26 2.118 1.251 3.583 Kim (2007) 753 753 Case-control < 25 25 1.225 0.963 1.557 Asian Nursing Research March 2009 Vol 3 No 1 35
D. Jung, S.M. Lee Sub-groups No. of studies Table 2 Meta-analysis Results Fixed-effects model Random-effects model OR 95% CI p OR 95% CI p All studies 26 1.282 1.209, 1.361.000 1.388 1.129, 1.706 Premenopause only 11 1.467 1.268, 1.698.000 1.387 1.134, 1.696.001 Postmenopause only 11 1.614 1.360, 1.917.000 1.681 1.149, 2.461.008 OR = odds ratio; CI = confidence interval. Study Sub-scale Odds ratio and 95% CI Ahn (1998) Choi (2004) Chun (2004) Do (2000) Do (2003a) Jeon (1991) Joo (2004) Kang (1998) Lee (2005) Lee (2003) Lee (2001) Lim (2005) Moon (1997) Park (2003) Shin (1995) Shin (2000) Suh (1995) Yoo (1993) Yoo (1998) Yoon (2000) Choi (2001) Choi (2003) Do (2003b) Lee (2005) Park (2000) Kim (2007) 0.1 0.2 0.5 1 2 5 10 Figure 2. Odds ratio and 95% confidence interval from the individual studies for BMI and Forest Plot (Random Effects Model). pre- and postmenopausal groups was shown to have a higher incidence rate of breast cancer; additionally, the incidence of breast cancer in women in the postmenopausal stage was highly correlated with BMI, as compared to the incidence rate for women in the premenopausal stage. However, this result was not consistent with a previous meta-analysis that used 23 observational studies of the pre-menopausal stage. Ursine, Longnecker, Haile, & Greenland (1995) reported a moderate negative correlation between BMI and breast cancer. This inconsistency was based on the heterogeneity of the studies used in the metaanalysis by Ursin et al., and an inverse association was revealed for 19 control studies and community control studies. However, a positive association was shown after adjustment for confounding factors like age. Another meta-analysis of eight prospective studies of women with breast cancer reported that increased BMI might cause a significant elevation of breast cancer rate (Endogenous Hormones and Breast Cancer 36 Asian Nursing Research March 2009 Vol 3 No 1
BMI and Breast Cancer in Korea: A Meta-Analysis Table 3 Tests of Homogeneity Sub-groups No. of studies Test of homogeneity Q p I 2 (%) All studies 26 255.350.000 90.210 Premenopausal only 11 15.158.126 34.028 Postmenopausal only 11 43.337.000 76.925 Standard error 0.0 0.2 0.4 0.6 0.8 2.0 1.5 1.0 0.5 0.0 0.5 1.0 1.5 2.0 Log odds ratio Precision (1/Standard error) 20 15 10 5 0 2.0 1.5 1.0 0.5 0.0 0.5 1.0 1.5 2.0 Log odds ratio Figure 3. Funnel Plot of standard error and precision by log odds ratio. Table 4 Tests of Publication Bias Sub-groups Begg s test Egger s test Kendall s tau p SE p All studies.089.537 1.307.513 Premenopausal only.236.350.835.183 Postmenopausal only.018 1.000 1.955.706 Collaborative Group, 2003); this conclusion is consistent with the findings of the current study. Nonetheless, this result showed a decreased relative risk (RR) after adjusting for the effect of the serum estrogen, and such a fact indicates that the effect of serum estrogen seems to be critical to the relationship between BMI and the rate of breast cancer for the women in postmenopausal stage. A meta-analysis that investigated the relationship between body fat distribution and breast cancer incidence rate through waist-to-hip ratio (WHR) reported that the breast cancer rate increased when WHR was elevated to 1.62 (95% CI = 1.28 2.04) as a risk estimate (Connolly et al., 2002). Connolly et al. adopted 14 case control studies and five cohort studies for their meta-analysis. They found that the risk estimate for premenopausal women was 1.79 (95% CI = 1.22 2.62) and the risk estimate for postmenopausal women was 1.50 (95% CI = 1.10 2.04). As a result of separate analysis of pre- and postmenopausal Asian Nursing Research March 2009 Vol 3 No 1 37
D. Jung, S.M. Lee Study Statistics with study removed Odds Ratio (95% CI) with study removed Point Lower limit Upper limit z-value p-value Ahn (1998) Choi (2004) Chun (2004) Do (2000) Do (2003) Jeon (1991) Joo (2004) Kang (1998) Lee (2005) Lee (2003) Lee (2001) Lim (2005) Moon (1997) Park (2003) Shin (1995) Shin (2000) Suh (1995) Yoo (1993) Yoo (1998) Yoon (2000) Choi (2001) Choi (2003) Do (2003) Lee (2005) Park (2000) Kim (2007) 1.375 1.392 1.396 1.397 1.401 1.400 1.307 1.391 1.368 1.473 1.373 1.408 1.377 1.373 1.350 1.347 1.421 1.379 1.391 1.420 1.400 1.411 1.401 1.383 1.366 1.396 1.388 1.110 1.129 1.128 1.130 1.133 1.134 1.081 1.126 1.107 1.254 1.111 1.129 1.111 1.100 1.096 1.094 1.150 1.115 1.120 1.154 1.123 1.138 1.131 1.116 1.106 1.122 1.129 1.703 1.715 1.727 1.728 1.732 1.728 1.581 1.719 1.690 1.730 1.696 1.756 1.705 1.715 1.663 1.659 1.755 1.705 1.727 1.748 1.745 1.750 1.735 1.714 1.687 1.736 1.706 2.919 3.096 3.068 3.091 3.116 3.130 2.763 3.054 2.901 4.722 2.939 3.034 2.926 2.801 2.820 2.809 3.252 2.964 2.986 3.318 2.994 3.132 3.085 2.963 2.899 2.994 3.110.004.006.004.000.005.005.005.001.001.004 0.1 0.2 0.5 1 2 5 Figure 4. Overall effect size without a study at a time (Random Effect Assumption). CI = confidence interval. stage and body fat distribution, the breast cancer incidence rate was higher in premenopausal women. When this study was compared with previous metaanalyses, the relationship between BMI and breast cancer showed that the breast cancer incidence rate was not only higher for women in the postmenopausal stage, but also for women in the premenopausal stage. Because this meta-analysis adopted a total of 26 observational studies consisting of 25 case control studies and a cohort study, comparison of the difference between BMI and breast cancer incidence rate in the case control study and cohort study was not performed. However, the risk estimate was higher in the case control study than the cohort study according to previous meta-analysis between obesity and breast cancer rate (Connolly et al., 2002; Ursin et al., 1995). This meant that the large size cohort study in Korean women was insufficient. To obtain a significant result in the meta-analysis studies, there should be a qualitative study and the level of study integration should be improved by performing a high quality cohort study of female breast cancer. 38 Asian Nursing Research March 2009 Vol 3 No 1
BMI and Breast Cancer in Korea: A Meta-Analysis The Q statistic of meta-analysis on homogeneity among 26 articles was not found to be statistically homogeneous. The possible reasons for this are the variety of study designs for articles used for metaanalysis, variety of study population (i.e. age, menopausal status), uncontrolled confounding factors (i.e. comorbidity, cancer stage), and unmatched BMI cutoff for the comparison of true effects. When the variety of true effects of integrated articles was considered, interpretation of the meta-analysis result using a random effect model was required (DerSimonian & Laird, 1986; National Cancer Center, 2002). Because the analyzed studies consisted of observational studies, the limitation of the meta-analysis was that the confounding effect of each study needed to be considered when the relationship between BMI and breast cancer rate from the result was generalized. The meta-analysis using observational studies cannot control for the inherent biases compared to the randomized controlled study, and the study design and subject selection for each analysis are not coherent; thus, it is somewhat difficult to generalize the results (Stroup et al., 2000). Moreover, a cross-sectional study is more likely performed in a particular area with a high incidence, and is less likely in an area with a rare incidence; therefore, those results can have bias and are a study limitation (Stroup et al., 2000). However, in many circumstances a randomized controlled study may not be possible in reality. The meta-analysis using observational studies may be required because of the necessity of integrating real situations rather than controlled situations. In addition, observational studies can be useful because research on breast cancer rate and increased BMI cannot be performed with randomized clinical trials using an experimental study design. Although there is much debate on metaanalysis using observational studies, this method may be required because of the characteristics of breast cancer research. Also, this study s meta-analysis adopted the investigator s cut-off point to pull in raw data because all of the 26 studies utilized different BMI references. In addition, all the adjusted variables including age used in the actual studies were ignored, which may be a possible limitation. Although these study limitations exist, this metaanalysis is still meaningful as an initial study because no other studies analyzing the relationship between BMI and breast cancer incidence rate among Korean women have been previously conducted. REFERENCES Axelrod, D., Smith, J., Kornreich, D., Grinstead, E., Singh, B., & Cangiarella, J. (2008). Breast cancer in young women. Journal of American College Surgeons, 206, 1193 1203. Bissonauth, V., Shatenstein, B., & Ghadirian, P. (2008). Nutrition and breast cancer among sporadic cases and gene mutation carriers: An overview. Cancer Detection and Prevention, 32,52 64. Carpenter, C. L., Ross, R. K., Paganini-Hill, A., & Bernstein, L. (2003). Effect of family history, obesity and exercise on breast cancer risk among postmenopausal women. International Journal of Cancer, 106,96 102. Clark, G. M., McGuire,W. L., Hubay, C.A., Peterson, O. H., & Marshall, J. S. (1983). Progesterone receptors as a prognostic factor in stage II breast Cancer. The New England Journal of Medicine, 309, 1343 1347. Connolly, B. S., Barnett, C., Bogt, K. N., Li, T., Sone, J., & Boyd, N. F. (2002). A meta-analysis of published literature on waist-to-hip ratio and risk of breast cancer. Nutritional and Cancer, 44, 127 138. DerSimonian, R., & Laird, N. (1986). Meta-analysis in clinical trials. Controlled Clinical Trials, 7, 177 188. Endogenous Hormones and Breast Cancer Collaborative Group (2003). Body mass index, serum sex hormones, and breast cancer risk in post menopausal women. Journal of the National Cancer Institute, 95, 1218 1226. Gloecker-Reis, L., Pollack, E. S., & Young, J. L. (1983). Cancer patient survival. Journal of the National Cancer Institute, 70,693 707. Harvie, M., Hooper, L., & Howell, A. H. (2003). Central obesity and breast cancer risk: A systematic review. Obesity Reviews, 4, 157 173. Hede, K. (2008). Fat may fuel breast cancer growth. Journal of the National Cancer Institute, 100,298 299. Hjartaker, A., Langseth, H., & Weiderpass, E. (2008). Obesity and diabetes epidemics: Cancer repercussions. Advances in Experimental Medicine and Biology, 630, 72 93. Asian Nursing Research March 2009 Vol 3 No 1 39
D. Jung, S.M. Lee Jakicic, J. M., & Otto, A. D. (2006). Treatment and prevention of obesity: What is the role of exercise? Nutrition Reviews, 64,57 61. Korean National Statistical Office (2003). The cause of death statistics, 2002, Seoul, Korea: Author. Lee, S. M., Park, J. H., & Park, H. J. (2008). Breast cancer risk factors in Korean women: A literature review. International Nursing Review, 55,355 359. Lee, S. R. (2001). A case-control study on related factors of breast cancer in Korean women. Journal of Public Health and Social Science, 10,97 115. Lim, S. M. (2005). BMI and breast cancer risk Korean medical insurance corporation study. Unpublished master s thesis, Yonsei Univesity, Seoul, Korea. Manni, A. (1983). Hormone receptors and breast cancer. The New England Journal of Medicine, 309, 1383 1385. Maruti, S. S., Willett, W. C., Feskanich, D., Rosner, B., & Colditz, G. A. (2008). A prospective study of agespecific physical activity and premenopausal breast cancer. Journal of the National Cancer Institute, 100, 728 737. Muller, C. B., Ames, F., & Anderson, G. D. (1978). Breast cancer in 3558 women: Age as a significant determinant in the rate of dying and causes of death. Surgery, 83, 123 132. National Cancer Center (2002). Annual report of the central cancer registry in Korea, Seoul, Korea: Author. Song, Y. M., Sung, J., & Ha, M. (2008). Obesity and risk of cancer in postmenopausal Korean women. Journal of Clinical Oncology, 26, 3395 3402. Stroup, D. F., Berlin, J. A., Morton, S. C., Olkin, I., Williamson, G. D., & Rennie, D. (2000). Meta-analysis of observational studies in epidemiology. Journal of the American Medical Association, 283, 2008 2012. Sweeney, C., Blair, C. K., Anderson, K. E., Lazovich, D., & Folsom, A. R. (2004). Risk factors for breast cancer in elderly women. American Journal of Epidemiology, 160, 868 875. Ursin, G., Longnecker, M. P., Haile, R. W., & Greenland, S. (1995). A meta-analysis of body mass index and risk of premenopausal breast cancer. Epidemiology, 6, 137 141. WHO. BMI classification. Retrieved October 8, 2008., from http://www.who.int/bmi/index.jsp?intropage = intro_ 3.html Yoo, K. Y. (2003). Epidemiology and risk factors of breast cancer. Journal of Korean Medical Association, 46, 482 489. Yoo, K. Y., Kang, D., Park, S. K., Kim, S. U., Shin, A., & Yoon, H. (2002). Epidemiology of breast cancer in Korea: Occurrence, high-risk groups, and prevention. Journal of Korean Medical Science, 17,1 6. 40 Asian Nursing Research March 2009 Vol 3 No 1