This study showed that the temporal trends in breast cancer incidence are different among the Chinese, Malays and Indians living in Singapore. It is unlikely that this could be explained by ethnic differences in access to health care as the Ministry of Health in Singapore ensures that universal health coverage is provided to all Singaporeans regardless of their ability to pay [25]. Thus the three ethnic groups have similar access to healthcare but differ with regard to lifestyles, especially in the past. Furthermore, the proportions of breast cancer that went unnoticed until the time of death in the three ethnic groups were very low.
Our data suggest that the temporal trends in Chinese and Malays are most likely attributed to risk factors associated with birth cohort. Specifically, breast cancer incidence in Chinese and Malay women showed a steady increase over birth cohorts from the 1930s, and the age-cohort model provided a good fit for both pre- and post-menopausal Chinese, and pre-menopausal Malays. Results for the Indians indicate that the age-drift models best fit both the pre- and postmenopausal Indians.
The age-specific breast cancer rates between the Indians and the other two ethnic groups also differ. Yansui and Potter suggested that different patterns of age-specific rates may be due to differences in the proportions of ER+ and PR+ breast cancers [26]. Preliminary results on available ER+ and PR+ status of breast cancer tissues did not show any ethnic differences. Based on a series from the largest hospital in Singapore between 1995 and 2005, the proportions of ER+ cases were 65% in Chinese, 62% in Malay and 61% in Indians while those of PR+ cases were 50%, 47% and 50% respectively (unpublished data).
These observed ethnic differences in age-specific rates and results from APC modelling suggest that there may be i) ethnic differences in exposure to certain risk factors, or ii) ethnic differences in the response to similar changes in exposure, or iii) both differences in ethnic exposure and responses.
The changes in breast cancer incidence may be related to Singapore's transition from an industrialized country to a developed country, bringing about changes in lifestyle in the population, including late marriages, delay in childbearing and smaller family sizes. This economic transition led to demographic changes such as low population growth rates, accompanied by changes in socioeconomic status such as increase in incomes and women's participation in paid employment; and higher education levels [27]. Furthermore, in 1966, the Singapore Family Planning and Population Board (SFPPB) introduced family planning campaigns to curb increasing population growth, resulting in a dramatic decline in the total fertility in Singaporean women (Figure 4). However, in the 1980s, pro-natalist family planning programs were introduced in view of the low birth rates and population growth.
From recent data on parity, there was a general decline in number of children born per ever-married female. In 1980, Malay women had the highest mean number of children (3.9) compared to the other two ethnic groups (3.4 for Chinese and for Indians). By year 2000, the numbers declined to 3.1 for the Malays, 2.5 for the Chinese and 2.4 for the Indians [28, 29]. The median age at first birth increased in all three ethnic groups from 1970 to 2000 (Table 4). Malay women have consistently had the youngest age at first birth and the highest parity among the three ethnic groups [28–30].
Delayed age at first birth and decreasing parity are possibly important factors governing the changes in breast cancer rates of the Chinese and Malays, and may contribute to the increasing IRRs seen in the birth cohorts after 1930, particularly after 1945. Implementation of the anti-natalist family planning campaign in the early 70s would have had more influence on women in their late teens and twenties, i.e. those who were likely to be born after 1940. While among those who had already given birth we may expect to see lower parity, we would expect to see both delayed first birth and lower parity for the remaining women. This is consistent with the increased rate of change of risk for cohort 1946 – 1950 (Figure 3c) and the marked increase in cohort IRR (Table 3), particularly for the Malay women. Following the changes in family planning after 1980s, the rate of change of risk slowed down in subsequent cohorts from 1951 (Figure 3c).
The impact of delaying first birth and decreasing parity brought by the changes in lifestyle patterns and family planning campaign was probably not as great for the Chinese as it was for the Malays since the Chinese already tended to delay age at first birth and had smaller family sizes relative to the Malays (and Indians) in the 1970s. This is consistent with Figure 3c where the Chinese had an increased rate of change in risk at an earlier cohort (1936–1940) than the Malays (1946–1950). The introduction of family planning campaign may have prolonged the increased rate of change in the Chinese via changes in fertility patterns. These differential changes in age at first birth and parity between the Chinese and Malays may also explain why the increased risk of developing breast cancer observed in Malay women relative to their counterparts in the earlier cohorts is higher than the increased risk of developing breast cancer observed in the Chinese (Table 3).
Although all three ethnic groups experienced delayed age at first birth and decreasing parity, there appears to be ethnic differences in its influence on breast cancer incidence. Decreasing total fertility is found to be associated with increasing cumulative incidence rates in later birth cohorts for Chinese and Malay women aged 25 to 49 (Figure 5). For the Indians, the association was weaker compared to the other two ethnic groups, and the age-drift model fitted them well. The difference may be due to a less pronounced protective effect of reproductive factors in the Indians.
A significant period effect was seen in the Chinese population. While the period curvature effects were minimal compared to the cohort curvature effects, increasing self-awareness of breast cancer in the Chinese could be a contributing factor. The absence of cohort effects in the post-menopausal Malays despite the good age-cohort model fit for the entire Malay population may be due to a cohort influence that is only beginning to emerge in the younger Malay women, particularly those born after 1941 who had just begun to move into their post-menopausal years (≥50).
Our preliminary analyses showed ethnic differences in the association between fertility and pre-menopausal breast cancer. As this cohort matures, the ethnic differences in post-menopausal breast cancer can be further examined. However, total fertility rates may not be an adequate index of the changing fertility patterns as it does not capture the changes in age at first birth of a population, which is known to be a strong protective factor for breast cancer. A record-linkage study between birth registry and cancer registry may confirm the ethnic differences associated with fertility.
Besides reproductive factors, height, dietary habits, obesity, duration of breastfeeding, childhood growth, age at menarche and menopause are some other possible environmental risk factors whose distribution may vary across the ethnic groups, and could also influence breast cancer risk [10, 11, 31]. Obesity (body mass index ≥ 30 kg/m2) has generally been more prevalent among the Malay and Indian women compared to the Chinese women [32–34]. With increasing affluence, there has been an increase in the consumption in all food groups, including dietary fats. From the National Nutrition Survey 2004, the mean fat intake was highest among the Malays (85.2 g), followed closely by the Indians (83.3 g) and lastly the Chinese (76.2 g) [35]. However, it is difficult to establish a direct link between each type of dietary fats and the development of breast cancer since it is difficult to compare trends in consumption of specific food items at ethnic or even population level. Changes in dietary habits, e.g. increased intake of dietary fats, is likely to be confounded with low parity and higher age at first birth which are more common in affluent countries [13]. Thus with possibly limited effect of fertility, the obesity factor may have contributed to an increasing trend in age-specific breast cancer incidence for the Indians. Further, as women progress into sedentary jobs, decreased physical activity could contribute to the cohort effect in the older Chinese women. Breastfeeding has been more common among the Malays [36, 37]. As mothers with more children could potentially breastfeed longer, fertility and duration of breastfeeding could be markers of each other. Though there has been an increase in breastfeeding among the better-educated mothers, there is no readily available detailed information about the changes in breastfeeding practices over the study period. The use of menopausal hormones has never been prevalent in any ethnic group in Singapore and are unlikely to explain the observed differences [38].