行政院國家科學委員會專題研究計畫 期中進度報告
子計畫一:建立癌症研究的資料及統計中心(1/3)
計畫類別： 整合型計畫 計畫編號： NSC93-2745-B-039-001-URD 執行期間： 93 年 08 月 01 日至 94 年 07 月 31 日 執行單位： 中國醫藥大學環境醫學研究所 計畫主持人： 郭憲文 共同主持人： 鍾景光，梁文敏 計畫參與人員： 蔡清讚教授、宋鴻樟教授、王昶弼、陳宏偉、吳泰進、許哲瑋、 林孟宏、鄭晏宗 報告類型： 精簡報告 處理方式： 本計畫可公開查詢中 華 民 國 94 年 6 月 16 日
Abstract
The objective of this study was to investigate the roles of spatio-temporal factors
in the incidence and mortality of breast cancer among Taiwanese women from 1979 to
1998. Breast cancer incidence and mortality data were provided by Taiwan’s
Department of Health (DOH). The spatial factors included urbanization, residential
areas, population density and allocation of medical resources. The rate of increase of
breast cancer incidence over the 20-year period was much higher than that of breast
cancer mortality. Overall, the peak of breast cancer incidence was in the age range of
45 to 49 years. Breast cancer incidence in the 336 regions varied considerably and
correlated strongly with mortality (r=0.37). Breast cancer incidence and mortality
were strongly correlated with each of the four spatial factors. In conclusion, the
significant differences in breast cancer incidence and mortality based on
spatio-temporal factors indicate that environmental factors play an important role in
the etiology of this disease. The results of this study may be used as a reference in
strategies aimed at preventing breast cancer.
摘要
此研究的目標在於調查空間時間因子在台灣女性乳癌的發生率
和死亡率中所扮演的角色。研究期間為 1979 年至 1998 年。而乳癌的
發生率和死亡率的資料來源為衛生署。空間因子包括都市化程度、地
形、人口密度和醫療資源的分布。結果顯示女性年齡大於 20 歲其乳
癌發生率的增加比死亡率來的高，而乳癌的發生率在 45 歲至 50 歲的
期間達到巔峰，但是最近幾年發現，乳癌的發生率有逐漸年輕化的趨
勢。此外，乳癌的發生率在 336 個鄉鎮市區中和死亡率有顯著的相關
性(r=0.37)，且研究發現乳癌的死亡率及發生率和四個空間因子有顯
著的相關。整體而言，乳癌的死亡率和發生率在環境中的時空因子對
乳癌的發生率和死亡率扮演著重要的角色，本研究的結果可提供有關
機構作為預防乳癌政策上的參考及預防乳癌的目標。
關鍵字：時間空間因子、乳癌發生率、死亡率
Introduction
Breast cancer is a major cause of death among Taiwanese women. In 2000, breast
cancer mortality rate among women was 10.61 per 100,000. There has been a steady
rise in breast cancer incidence over the last twenty years which may reflect, in part,
changes in lifestyle, such as diet. In 1990, Taiwan’s DOH implemented a nationwide
breast cancer screening program in the community. Between 1999 and 2001, one
million women aged over 35 years were given breast palpitation. The screening
program included education about self-examining and the importance of early
detection. Suspected positive cases were referred to local hospitals for diagnosis and
treatment. Tabar (2003) investigated mortality in breast cancer patients and found that
mammography service screening substantially reduced mortality rates. In Taiwan,
such screening programs have not been comprehensively evaluated. Recent reports in
Taiwan indicate that the age of onset of malignant breast cancer has become lower
over the past few decades. Family history of breast cancer is a well-known risk factor
in early onset of this disease. Claus (1990) conducted a population-based case-control
study. A total of 4730 breast cancer patients were age- and residence-matched with
4688 controls. Family history of breast cancer, especially in a first-degree relative,
was found to be a significant factor, after adjustment for confounders. Sattin (1985)
family history of breast cancer compared to those without family history of the
disease. The Collaborative Group on Hormonal Factors in Breast Cancer (2001)
concluded that the lifetime excess incidence of breast cancer is 5.5% with one
affected first-degree relative and 13.3% for women with two. Although family history
most likely affects incidence of breast cancer through genetic predisposition, various
environmental factors may also play important roles. Currently, there are no available
data in Taiwan on relationships between the incidence and mortality of breast cancer.
Moreover, there are no data on the spatio-temporal factors that may affect incidence
and mortality of this disease in Taiwan. The findings of this study will be used to
establish baseline data on breast cancer incidence and mortality in order to develop
nationwide breast cancer prevention programs.
Materials and Methods
Data on the general population from 1979 to 1998 in Taiwan were provided by the
Ministry of the Interior, while data on breast cancer cases and breast cancer death
cases were provided by the Department of Health. Data from 1996 were used as the
standard population to calculate the standardized incidence and mortality of breast
cancer in each year. Standardized incidence and mortality in each district from 1992
labeled as high risk, moderate high risk, moderate, moderate low risk and low risk.
Urbanization data, residential areas, population density and medical resources (no. of
medical staff) and population density were provided by the government office of
statistics (Executive Yuan). The urbanization level was classified into seven categories
according to population density, social index, and economic status. In the current
study three main categories of urbanization were used: metropolitan (high), suburban
(moderate) and rural areas (low). Residential areas were classified into three
categories: plains, mountains and outlying islands. The proportions of the population
and medical staff were calculated by dividing numbers in each district by numbers in
the total population. These proportions were classified into high (over 33.3%), low
(under 66.6%) and medium (between 33.3% and 66.6%). Population density in each
district was calculated in the same way. Age-adjusted incidence and mortality were
graded into five levels. Breast cancer incidence was graded from 0-11.95, 11.95-15.95,
15.95-20.75, 20.75-27.24, 27.24-67.46 per 100,000 population. Breast cancer
mortality was graded from 0-3.93, 3.93-6.61, 6.61-8.41, 8.41-10.45, 10.45-26.25 per
100,000 population.
The statistical model used was proportional odds model
k k j x j x x x ODD ( ))=α +β1 1+β2 2+Λ +β log( ,
2 , 3 , 4 , 5
, j= . Since the assumption of proportional odds hold (which was shown by the goodness-of-fit test for model fit), for any fixed j, the estimated odds that a spatial
situation (denoted by A) is in the high risk direction rather than the low risk direction
equal exp(β ) times the estimated odds for the other spatial situation (denoted by B). For example, for urbanization, let x=A meant in the moderate urbanization area, x=B
meant in the low urbanization area, the odds ratio (OR) was calculated from these
results as follows: . 1 ) 1 1 ) 1 1 ) 1 1 ) 1 ) ( 2 ) ( 2 ( 2 ) ( 2 ) ( 3 ) ( 3 ( 3 ) ( 3 ) ( 4 ) ( 4 ( 4 ) ( 4 ) ( 5 ) ( 5 ( 5 ) ( 5 B x B x A x A x B x B x A x A x B x B x A x A x B x B x A x A x p p p p p p p p p p p p p p p p Ratio Odds = = = = = = = = = = = = = = = = − − = − − = − − = − − =
When OR was over 1, this meant that the population in the moderate urbanization
area had a higher risk of developing breast cancer than the population in areas with
low level urbanization. The same calculation method was used for the other spatial
factors. Age standardized incidence and mortality were plotted for each year (1979 to
1998), for each age group, and for each birth cohort. Age standardized incidence and
mortality of breast cancer in the 336 districts were calculated and classified into five
categories and plotted on the map of Taiwan. The Pearson’s correlation coefficient
Results
Age standardized incidence and mortality of breast cancer in the 336 districts in
Taiwan are shown in Figure 1. Breast cancer incidence and mortality were graded into
five levels with the darkest areas indicating highest incidence and mortality. The
darkest areas tend to coincide with the metropolitan areas, such as in the north.
Pearson’s correlation coefficient between incidence and mortality of breast cancer was
0.37. Incidence was lowest in the east of Taiwan, although some areas with low breast
cancer incidence had high mortality.
Figure 2 shows the trends in age-specific breast cancer incidence and mortality
rates from 1979 to 1998. Incidence of breast cancer increases with time, with a sharp
increase at 1990. However, the mortality rate increases only slightly (range from
5/100,000 population to 10/100,000 population). The peak in mortality rate occurred
in 1997 and decreased gradually afterwards.
Figure 3 shows the trends in age-specific incidence and mortality rates of breast
cancer by birth cohort (1902-1911, 1907-1916, …, 1942-1951). Incidence of breast
cancer occurred at an earlier age in the younger generations. Also, the rate of increase
in incidence of breast cancer was higher in the younger generations.
The trends in age-specific incidence and mortality rates of breast cancer in three
There was a similar trend to that in Fig.3 with an increase of breast cancer mortality
with age. The increase in incidence was highest in the most recent five-year time
period (1992-1996), with a peak at 45-49 years for each five-year age interval. In
principle, mortality rates increased in each of the three five-year time periods.
Mortality rate was highest at ages over 80 years. Except for ages over 80 years, the
mortality in the most recent five-year time period was consistently higher than in the
other time periods. For all ages below 65, incidence of breast cancer was higher than
breast cancer mortality.
Table 1 shows the correlations of the five levels of breast cancer incidence and
mortality with the spatial factors. Overall, there were significant associations of both
incidence and mortality of breast cancer with each of the spatial factors. There was a
significant positive correlation of high urbanization with breast cancer incidence and
mortality. There was an even spread of levels of breast cancer incidence and mortality
in the plain areas, but for the mountainous areas and outlying islands, there were
negative correlations with breast cancer incidence and mortality. Population density
correlated negatively with breast cancer incidence and mortality. Low population
density correlated significantly with high incidence of breast cancer, but breast cancer
mortality was similar in each level of mortality at low population density. There was a
The odds ratios of breast cancer incidence and mortality within levels of each
spatial factor can be seen in Table 2. There were dose-dependently significant
correlations of urbanization with breast cancer incidence and mortality. There was a
significant difference between plains and mountains in breast cancer incidence and
mortality, but there were no significant associations among other topographical areas.
Significant negative correlations between each of the levels of population density
were found in breast cancer incidence, but for breast cancer mortality only high and
low population density correlated significantly. High level of medical resources was
higher than medium and low levels of medical resources in both breast cancer
incidence and mortality. Incidence and mortality of breast cancer were similar
between medium and low level of medical resources.
Discussion
Previous studies of breast cancer in Taiwan have shown that breast cancer mortality is
highest in urban areas, while it is highly variable around the island as a whole, which
may be due to the small number of breast cancer cases. Long-term monitoring of
breast cancer mortality is required to identify trends in different parts of the country.
In addition, cancer mapping, such as the “Geographic Information System” (GIS) can
provide a dynamic exhibition of patterns and trends which yields information about
their intercorrelations. The current study found that breast cancer mortality and
incidence were highly correlated and has increased gradually over the past twenty
years. This may be attributed to the introduction of the national cancer screening
program in 1995 in Taiwan. Furthermore, dietary and lifestyle changes may have
contributed to the increase in breast cancer mortality and incidence. Another
important trend is the considerably earlier onset of breast cancer incidence over the
past twenty years. As such, the Department of Health may lower the age of breast
cancer screening. Ongoing health awareness programs have also helped to catch
breast cancer in many women in Taiwan. Levi (1994) reported that overall incidence
of various cancers increased from 10% to 30% but the cancer mortality remained
approximately the same. This may be due to improved diagnosis and therapeutic
advancements. Menegoz (1997) investigated the cancer incidence and mortality in
France in 1975-95 and found that mortality remained relatively even while incidence
increased.
Cancer mapping studies have shown that the south-west of Taiwan has a
clustering effect for certain cancers, such as cancer of the bladder, skin, liver and lung
which have been attributed to arsenic in the groundwater. However, breast cancer
incidence in this part of the country was not significantly from that in other areas.
pesticides, such as DDT which was used 50 years ago to eradicate malaria in Taiwan,
and air pollutants from incinerators, such as dioxin. The Taiwan EPA has recently
reduced the maximum permissible limit for dioxin. Breast cancer incidence is
generally much higher in highly urbanized areas. This finding is consistent with those
of many other studies in other countries, such as Japan and Spain. Previous studies
have identified other factors associated with increased incidence of breast cancer,
such as decreasing age of menarche, increasing age at first marriage, age at first birth,
decreasing fertility rate and increasing adult height. Minami (1996) reported
significant period effects on breast cancer incidence, although the cohort effect was
marginal. The relative risks by birth cohort suggested a declining trend in younger
birth cohorts. Mackillop (2000) evaluated associations between community incomes
and cancer incidence in North America and found that breast cancer incidence was
highest among highest socioeconomic status, although the underlying mechanisms
remain unclear. Over the years the cancer registry has developed comprehensive data
on breast cancer. In addition, breast cancer diagnostic techniques have improved
considerably, as proven by follow-up biopsy in 80% of diagnoses. This database can
be used to correlate with environmental and ecological factors as well as personal risk
factors, such as predisposition, alcohol consumption and smoking habit. Analyses of
programs. The database can also be used to evaluate the effectiveness of such
programs.
ACKNOWLEDGEMENTS
This study was supported by a specific grant (NSC93－2745－B－039－001－URD)
REFERENCES
1. Tabar L ,Yen MF ,Vitak B ,Tony Chen HH. Mammography service screening and mortality in breast cancer patients: 20-year follow-up before and after introduction of screening. The Lancet 2003; 361:9367.
2. Claus EB ,Risch NJ ,Thompson WD. Using age of onset to distinguish between subforms of breast cancer. Ann Hum Genet 1990; 54(2):169-77.
3. Claus EB ,Stowe M ,Carter D ,Holford T. The risk of a contralateral breast cancer among women diagnosed with ductal andlobular breast carcinoma in situ: data from the Connecticut Tumor Registry. The Breast 2003;12: 451-56.
4. 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(9291):1389-99.
5. Sattin RW ,Rubin GL ,Webster LA ,Huezo CM ,Wingo PA ,Ory HW ,Layde PM. Family history and the risk of breast cancer. JAMA 1985;253(13):1908-13.
6. Levi F ,Lucchini F ,La Vecchia C. Worldwide patterns of cancer mortality,1985-89. Eur J Cancer Prev 1994;3(2):109-43.
7. Hannoun-Levi J M, Courdi A, Marsiglia H , Namer H, Gerard J P. Breast cancer in elderly women: is partial breast irradiation a good alternative? Breast Cancer Research and Treatment 2003;81(3): 243.
8. Menegoz F ,Black RJ ,Arveux P ,Magne V ,Ferlay J ,Buemi A ,Carli PM ,Chapelain G ,Faivre J ,Gignoux M ,Grosclaude P ,Mace-Lesec'h J ,Raverdy N ,Schaffer P.
Cancer incidence and mortality in France in 1975-95. Eur J Cancer Prev 1997 ;6(5):442-66.
9. Minami Y ,Takano A ,Okuno Y ,Fukao A ,Kurihara M ,Hisamichi S. Trends in the incidence of female breast and cervical cancers in Miyagi Prefecture, Japan, 1959-1987. Jpn J Cancer Res 1996;87(1):10-7.
10. Colonna M ,Grosclaude P ,Faivre J ,Revzani A. Cancer registry data based estimation of regional cancer incidence. Journal of Epidemiology and Community Health 1999; 53(9):558
11. Minami Y ,Ohuchi N ,Fukao A ,Hisamichi S ,Determinants of Infant Feeding Method in Relation to Risk Factors for Breast Cancer. Preventive Medicine
2000;30:363-370 ).
12. Mackillop WJ ,Zhang-Salomons J ,Boyd CJ ,Groome PA. Associations between community income and cancer incidence in Canada and the United. StatesCancer 2000;89(4): 901-912.
13. Chen CJ, You SL, Pwu RF, Wang LY, Lin YP, Hsi GC, Hsi MS, Ho HC, Lee CT, Lin CG, et al. Community-based cervical cancer screening in seven townships in Taiwan. J Formos Med Assoc.1995;94(2):103-11.
Table1.Correlations of the five levels of breast cancer incidence and mortality with the
spatial factors(1992 to 1996)
Rate of Incidence rate （％） # P Rate of Mortality （％） # P Spatial factors N Low Mod Low Mod Mod
high High r* Low Mod Low Mod Mod high High r* Urbanization <.0001 <.0001 Low 168 35.7 24.4 19.0 14.9 6.0 0.476 32.7 21.4 17.3 12.5 16.1 0.327 Moderate 143 7.7 20.3 23.8 27.3 21.0 10.5 23.1 24.5 25.9 16.1 High 48 4.2 12.5 16.7 66.7 0.0 2.1 6.3 16.7 29.2 45.8 Region <.0001 <.0001 Outlying islands 9 22.2 44.4 11.1 11.1 11.1 0.273 44.4 11.1 22.2 0.0 22.2 0.204 Mountainous areas 29 65.5 13.8 10.3 6.9 3.4 58.6 10.3 10.3 6.9 13.8 Plain areas 321 15.6 19.9 21.2 21.5 21.8 15.6 21.2 20.9 21.8 20.6 Pop. Density Low 119 13.4 10.9 16.8 23.5 35.3 <.0001 12.6 18.5 21.0 21.8 26.1 <.0001 Moderate 120 14.2 23.3 23.3 22.5 16.7 -0.298 16.7 20.8 19.2 25.8 17.5 -0.194 High 120 31.7 25.8 20.0 14.2 8.3 30.0 20.8 20.0 12.5 16.7 Dist. of Medical Resources Low 119 23.5 26.9 22.7 19.3 7.6 <.0001 23.5 31.9 18.5 9.2 16.8 <.0001 Moderate 120 25.8 20.8 18.3 17.5 17.5 0.271 21.7 19.2 19.2 22.5 17.5 0.218 High 120 10.0 12.5 19.2 23.3 35.0 14.2 9.2 22.5 28.3 25.8
#P：P value is according to chi-square tendency test
Table2. Odds ratios of breast cancer incidence and mortality within levels of each spatial
Factor (1992 to 1996)
P：P value is according to chi-square test
Incidence rate Mortality rate Spatial factor
Odds ratios (95%CI) P Odds ratios (95%CI) P
Urbanization Low 1 - 1 - Moderate 2.9(1.9-4.5) <.0001 2.0(1.3-3.0) <.0001 High 22.4(11.2-45.0) <.0001 6.1(3.4-11.2) <.0001 Region Outlying islands 1 - - Mountainous areas 1.3(0.3-5.1) 0.002 0.4(0.1-1.9) <.0001 Plain areas 3.4(1.0-11.8) 0.054 2.0(0.5-7.9) 0.318 Pop. Density Low 1 - 1 - Moderate 0.4(0.3-0.7) 0.010 0.7(0.4-1.1) 0.105 High 0.2(0.1-0.4) <.0001 0.5(0.3-0.8) 0.002 Distribution of Medical Resources Low 1 - 1 - Moderate 1.4(0.9-2.2) <.0001 1.3(0.8-2.0) 0.014 High 3.8(2.4-6.0) <.0001 2.3(1.4-3.6) 0.001
Figure 1. Age standardized incidence and mortality of breast cancer in the 336 districts in Taiwan.
*r：Pearson’s correlation coefficient between incidence and mortality
Incidence Mortality
Figure 2. Trends in age-specific breast cancer incidence and mortality rates from 1980 to 1998.
Figure 3. Trends in age-specific incidence and mortality rates of breast cancer by year of birth
Figure 4. Trends in age-specific incidence and mortality rates of breast cancer in three five-year time periods.
