It is now well recognized that preventive health care requires going beyond the immediate causes of diseases and understanding their fundamental socio-economic determinants. Cancer is one of these diseases that has emerged as the top killer, both physically and economically, surpassing heart disease, in many high-income countries. Does stress contribute to these increasing trends in cancer incidence and deaths? This is a question that does not have an easy answer. Opinion surveys tend to over-estimate or under-estimate the role of stress in causing cancer. Although it is not possible to conduct human experiments similar to those on animals, stressful events like the Chinese Cultural Revolution are a natural choice to study this link. Increase in cancer incidence among the cohorts born in the 1950s in two of China’s largest cities, Shanghai and Tianjin, show similar patterns. A regression analysis of the data from Shanghai shows a robust link between the Cultural Revolution and cancer; young adults who faced the Cultural Revolution had become more susceptible to cancer as they aged. These results bring out an unexpected link, a link to the so called ‘lost generation’ of China, those born in the 1950s. It shows that the plight of the lost generation has gone beyond the already documented socio-economic circumstances to other manifestations such as cancer.
Cancer is emerging as the top killer in many developed countries. Although cancer is not yet the top killer in some high income countries, in terms of economic cost the American Cancer Society has ranked cancer as the top economic killer . Estimates by the World Health Organization (WHO) showed that in 2008 the risk of getting cancer before the age of 75 for US men and women was 33.5% and 26.7%, respectively. In contrast, in India, the numbers for men and women were substantially lower at 10.2% and 10.8% . In fact, there is a close positive linear relationship between per capita income of a country and the risk of getting cancer before age 75 (Figure 1).
Figure 1. Linear relation between per capita income and risk of getting cancer before age 75 (Correlation = 0.78). Per capita income in international dollar is from Penn World Table 7.0, series RGDPCH (online) and cancer risk is from WHO. Systematically sampled 55 countries from a list of 190 ranked by per capita income (excludes oil rich outliers)
As countries move up the income per capita ladder the cancer risk also tends to increase. Cancer appeared to be rare in antiquity . The rapid increase in cancer incidence in modern societies is largely attributed to increasing exposure to carcinogens and behavioral factors such as improper diet, smoking and a lack of physical activity . In Japan, however, the cancer incidence rates have trended upward despite falling trends in cigarette smoking and heart disease rates . Does stress contribute to these increasing trends in cancer incidence and deaths? This is a question that does not have an easy answer because of the difficulties involved in assessing if there is a link. Nevertheless, examining the relationship between stressful events and cancer may shed light on this important health issue.It is now well recognized that many present-day chronic diseases and health problems have socio-economic origins . The basic thrust here is to go beyond immediate causes of diseases and examine the fundamental socio-economic causes in order to formulate preventive-care measures. The conceptual framework that the WHO has presented circumscribes how circumstances of daily life interacting with socio-economic structures work through material, social, psychological, behavioral and biological factors to produce different health outcomes in different societies and different layers of the same society [2, 6]. Wilskinson and Pickett  present an extensive analysis of the link between income inequality and health and other social malaise. A collection of articles published in Social Science & Medicine (2012, Vol. 74, Issue 5) reviews a sizable amount of literature on how economic crises affect health outcomes through increased unemployment, reduced income and wealth and altered educational opportunities, health care provisions and housing conditions.
An important mechanism through which socio-economic conditions affect health outcomes is psychological well-being . Some research demonstrates that the nature of the socio-economic climate experienced as a child may become an important determinant of adult mental health [8, 9]. Physiological distress is likely to affect health adversely both indirectly through lifestyle changes and directly through the effects on stress hormones [10, 11]. However, data constraints have been a major deterrant to assessing the impact of psychological distress on health.
Research on the link between psychological factors and cancer also remains limited and existing scientific research has not established a definitive link between the two [12, 13]. Nevertheless, recent randomized controlled experiments on fruit flies  seem to provide new evidence on a direct link between stress and cancer. How these results extend to humans remains to be seen. Scientific experimentation on healthy humans similar to those on animals is not ethically possible. As a result, the assessment of a possible link between stress and cancer is mostly confined to the opinions of cancer patients and as to whether they were under prolonged stress before the onset of their cancers. Such opinion surveys unfortunately may over- or under-emphasize the role of stress and produce tenuous conclusions .
While it is difficult to conduct scientific experiments on humans similar to those on animals, human history is replete with ‘experiments’ that have caused enormous stress on some human populations. Had there been data available from such historical events they would provide a rich source of information to examine the impact of stressful events on cancer incidence.
The objective of our exercise is to focus on one such ‘human experiment’, the Cultural Revolution in China, to examine how it relates to cancer incidence. It is well documented that the Cultural Revolution began in October, 1966, with the objective of further propagating the socialist ideology and transforming the bourgeoisie. This period lasted more than a decade, took violent forms, caused enormous social, political and economic upheaval and subjected large segments of the population to unbearable mental and physical stress [16, 17]. We were motivated to study this by a Singaporean cancer specialist who talked about a possible link between stress and cancer and drew attention to an increase in cancer incidence during the Chinese Cultural Revolution . Furthermore, in an intriguing account of traumatic experiences of some Chinese who have lived through the Cultural Revolution, Thurston  also draws attention to the possibility of a link between stress and cancer. After searching through the literature we came across only one other study that examined the impact of another major stressful event on cancer, namely, the Nazi Holocaust . The conclusion of that study was that the Israeli Jews who were exposed to the Holocaust experienced higher incidence of all cancers later in their lives.
In our review of the data we wanted to focus on a cohort analysis to assess the impact of the Cultural Revolution on cancer incidence in different birth cohorts of populations across China. City populations are of particular interest because of the ‘send down policies’ that made urban people work in rural areas. However, the paucity of data has been a challenge for the study; complete data records pertaining to our study are hardly available over the Cultural Revolution period. After some considerable effort we were able to compile a good data set for Shanghai and a limited data set for Tianjin. We also examined the data from Osaka, Japan, for comparison with a non-Chinese Asian city population. The main statistical technique we used is a variation of the age-period-cohort analysis (The full paper with methodology can be obtained from the authors).
Growth of cancer incidence by birth cohort
The Shanghai overall cancer incidence rate (per 100,000), by year, shows a spike in 1978 and then again in 1989 (Figure 2).
Figure 2. Shanghai cancer incidence rate per 100,000
From this type of data plot alone it is difficult to ascertain the impact of the Cultural Revolution on cancer incidence. Table 1 provides a clearer picture displaying how the rate of change of cancer incidence from one birth cohort to the next progressed as people grew older.
Growth rate (%) of cancer incidence at the same age in Shanghai and Tianji
To keep the table size small, any age groups >70 were omitted. Although the Tianjin data set is not complete, the cancer pattern in the two cities is very similar. Both cities show three sets of positive numbers (from left to right in the table), at the beginning, in the middle and at the end. In the table we have presented the diagonal averages with corresponding birth cohorts in parentheses and highlighted the middle set. Positive numbers observed at the end of the table are not surprising because cancer is generally an old-age disease. But what is interesting to note is that there had been a substantial drop in cancer incidence in the age group 55-69 since the early 1990s and old-age cancer incidence had shifted to above age 70. The positive numbers at the beginning are unexpected and we will discuss this later. The positive numbers in the middle of the table show that the birth cohorts indicated by years 1948, 1953, 1958 experienced, on average, a 58%, 63% and 28% increase in cancers in Shanghai and 38%, 57%, and 16% increase in Tianjin, respectively. These people were young adults during the Cultural Revolution. The most affected appear to be those who were born in the early 1950s, i.e. those who were in their late teens when the Cultural Revolution began in 1966. Since the numbers in Table 1 do not control for some important confounding effects we carried out a regression analysis on the Shanghai data. Similar analysis was carried out on the Osaka data for comparison (Tables are not presented for brevity). After adjusting for population size, age, income, a pollution indicator and age-pollution interaction variables, Figure 3 shows the growth rate of cancer incidence in successive birth cohorts in the Shanghai data.
Figure 3. Shanghai cancer incidence: Change in cohort effects by birth cohort and sex. Cohort effects are obtained after removing the effect of some confounding factors
The Shanghai cohort effects show two distinct sharp rises in the curves for both males and females. The estimates pertaining to these increases are statistically significant and not sensitive to different specifications of the regression model. Regardless of whether we use age dummies or an age polynomial, period dummies or period variables and with or without period controls, these hump estimates remain roughly the same. Figure 4 plots the level of cohort effects (in index format) to highlight the relative position and trends of the curves for males and females.
Figure 4. Shanghai cancer incidence: Level (index) of cohort effects by birth cohort and sex. Cohort effects are obtained after removing the effect of some confounding factors
A similar plot of the Osaka cohort effects does not show a hump corresponding to that of the Shanghai data. The first sharp increase in cancer in Figure 3 pertains to cohorts born between 1944 and 1958. These cohorts were in their young working ages during the Cultural Revolution. Having controlled for a number of confounding effects and also observing the absence of a corresponding rise in the Osaka data, we have to attribute the sharp increases in cancer incidence to the Cultural Revolution. As shown in Table 1, these cohorts experienced higher incidence of cancer in the late 1970s not only when they were still young (below age 35), but also as they grew older. The cohorts who were the most harshly affected appear to be those born in the 1950s; most of them must have been in their late teens at the peak (around 1970) of the Cultural Revolution.
What is further interesting to note in Figures 3 and 4 is the second large rise that peaks for the cohort born between 1983 and 1987. The data show that this peak is due to a sharp increase in leukemia cases in the youngest (0-4) age group. The positive numbers in Table 1 at young ages correspond to this jump. This appears to be a ripple effect of the Cultural Revolution itself; parental exposure to stresses of the Cultural Revolution and the incidence of leukemia in babies is likely. Whether such an association is mediated through behavioral factors like maternal alcohol consumption and smoking needs to be studied .
Cancer by type
As stated in the introduction, a depressive socio-economic environment may cause psychological distress which may result in poor health either through changes in behavioral factors or through the effects on stress hormones or both. The difficultly, however, is how to separate these effects and relate them to the jump in cancer incidence that resulted from the Cultural Revolution. The link between some behavioral factors such as smoking, physical inactivity, excess weight, improper diet and cancer is well established . In the absence of good data on these behavioral factors we cannot isolate the direct stress effect by controlling for other factors. However, we can shed some light on this issue by applying the cohort analysis to the Shanghai data by type of cancer. Because of data limitations, we limit our analysis to only some illustrative cancer types. The estimated cohort effects for these cancer types are presented in Figure 5.
Figure 5. Shanghai cancer incidence by type: Change in cohort effects by birth cohort and sex. Cohort effects are obtained after removing the effect of some confounding factors
Figure 5 shows that the first hump observed in Figure 3 is present in each case though with some variation between males and females. In Figure 5, notable differences in the peak shape between males and females occur in liver cancer and to some extent in lung cancer. Since these cancers are more prevalent among men it is possible that the interaction between stress and men, who tend to indulge more than women in substances that harm the body, i.e. alcohol, cigarette smoking, and/or inhaled other intoxicants such as opium, may have caused a higher incidence of cancers among men. It should be noted, however, that, strict rules under the new Communist Government brought opium abuse that plagued China for centuries under control in the early 1950s . Data also show that cigarette smoking grew only moderately over the Cultural Revolution period, 1-1.3 cigarettes/person/day. Cigarette smoking increased substantially only after 1977 and peaked around 1990, about 4 cigarettes/person/day, before declining mildly in later years . Among women, tobacco use was substantially lower, about 4% of women smoked in the 1990s as compared with 67% in men [23, 24]. Furthermore, studies on rural counties of China show that a higher vegetarian based diet had kept chronic diseases like cancer and heart diseases low in rural China at the time of the Cultural Revolution . Therefore, city populations that were deployed in rural areas were less likely to have faced higher incidence of cancer as a result of improper diet. These observations preclude us from attributing the entire increase in cancer incidence to changes in behavioral factors mentioned above.
The picture becomes even clearer in charts for leukemia, prostate, testis, penis and breast cancers. Since leukemia incidence had been somewhat higher in women, similarly jumps for leukemia cannot be attributed purely to the changes in behavioral factors. Likewise, the jumps in prostate, testis and penis cancers in the case of males and breast cancer in the case of females cannot be attributed purely to changes in behavioral factors. Therefore, we cannot rule out the possibility that there was a direct link between stress and cancer incidence resulting from the Cultural Revolution.
The lost generation of China
Our results suggest a real possibility of an unexpected link, a link to the so called ‘lost generation’ of China, those born in the 1950s. Hung and Chiu  and many others have documented that the city folks of this generation not only faced the wrath of the Cultural Revolution through deprived education and hard labor in the country side, they, lacking the necessary education and skills, also became victims of the subsequent transition of the country to the capitalist system. Our results show that the plight of the lost generation has gone beyond socio-economic circumstances to other manifestations like cancers. Exposure to traumatic events and physical and psychological stresses of the Cultural Revolution may have affected the young adults more severely such that they eventually became victims of cancer. As Thurston  observes: “Those who were middle-aged at the start of the Cultural Revolution had already experienced, at a minimum, the Japanese invasion, the civil war, the anti-rightist campaign, and the natural disasters of the ‘three bad years’ from 1959-61”. As for the younger people, Thurston comments that there were both direct and indirect victims; the latter refers to those who began as active supporters of the Cultural Revolution but subsequently became its victims.
Whether stress is going to cause cancer in an individual depends on each individual’s mechanism for coping with stress. We can observe three types of responses by individuals to stress: some resort to external stimuli like consuming alcohol and/or cigarette smoking; others simply suffer mentally, while still others manage stress by cultivating mental strength. Therefore, data at an individual level show large variations and make it difficult to measure a link between mental stress and cancer, an aspect scientists are more interested in. However, data from stressful events like the Cultural Revolution that lasted over a prolonged period are likely to show both the direct and indirect link between stress and cancer.
Our analysis of the Shanghai (and Tianjin) cancer data shows a very clear robust link between the Cultural Revolution and the increase in cancer incidence. After controlling for population, age, time period and age-pollution interaction effects, the regression estimates show a statistically significant sharp jump in cohort coefficients pertaining to those who were exposed to the Cultural Revolution during their young working ages. Rather unexpectedly, these birth cohorts also belong to the so called ‘lost generation’ of China, an outcome of the Cultural Revolution. Some of the increase in cancers in this generation may be attributed to changes in behavioral factors such as increased alcohol consumption and/or smoking. However, the similarity in the jump in the incidence of different types of cancers between men and women is suggestive of a possible direct link between stress and cancer. As the effects of the Cultural Revolution withered away, growth of cancer incidence (adjusted for other effects) in subsequent cohorts also came down.
Based on death and disability statistics related to lost years of life, John & Ross  ranked cancer as the top economic killer globally. Their estimates indicate that this indirect cost of cancer consumes 1.5% of world gross domestic product that is nearly 19 percent higher than heart disease. Estimates by Zhao et al.  indicate that cancer has become the top economic killer in China as well. Their estimates show that the indirect cost of cancer was twice the direct cost in 2003. Although the authors do not discuss in detail, it is likely that the indirect cost estimates do not include lost productivity of family caregivers. Total economic cost of cancer is likely to be much higher than estimated. Therefore, the importance of cancer prevention, both from individual and national perspectives, need not be over emphasized. In this regard, the role of stress, especially considering the long latency of cancers, may require more emphasis than it currently receives in cancer prevention programs.
The authors would like to thank the participants of seminars and conferences at Fudan University, Monash University (Sunway campus), National University of Singapore, Institute of Policy Studies, Sri Lanka, Daniel Bernhardt, Roger Koenker, Edward Norton and Xueyan Zhao for their valuable comments on earlier drafts of the paper.
Tilak Abeysinghe – Department of Economics and FASS Health Cluster, National University of Singapore, 117570 Singapore
Jiaying Gu – Department of Economics, University of Illinois at Urbana-Champaign, Urbana IL 6180, USA
Quesnel-Vallée, A. and M. Taylor, Socioeconomic pathways to depressive symptoms in adulthood: Evidence from the National Longitudinal Survey of Youth 1979. Social Science & Medicine, 2012. 74: p. 734-743.
Gelkopf, M., et al., Protective factors and predictors of vulnerability to chronic stress: A comparative study of 4 communities after 7 years of continuous rocket fire. Social Science & Medicine, 2012. 74: p. 757-766.
Kelly, A.B., et al., Longitudinal associations between telomere length, chronic stress, and immune stance in cervical cancer survivors, in 102nd American Association for Cancer Research Annual Meeting, April 4, 2011.
Kristenson, M., Impact of socioeconomic determinants on psychosocial factors and lifestyle – implications for health services. The Swedish experience. Social Science & Medicine, 2012. 74: p. 661–664.
Dalton, S., et al., Mind and cancer: do psychological factors cause cancer? European Journal of Cancer, 2002. 38(10): p. 1313-1323.
Garssen, B., Psychological factors and cancer development: evidence after 30 years of research. Clinical psychology review, 2004. 24(3): p. 315-338.
Wu, M., J.C. Pastor-Pareja, and T. Xu, Interaction between RasV12 and scribbled clones induces tumour growth and invasion. Nature, 2010. 463(7280): p. 545-548.
Catharine, W., et al., Beliefs about the causes of breast and colorectal cancer among women in the general population. . Cancer Causes and Control 2010. 21: p. 99–107.
Thurston, A.F., Victims of China’s Cultural Revolution: The Invisible Wounds: Part I. Pacific Affairs, 1984. 57(4): p. 599-620.
Thurston, A.F., Victims of China’s Cultural Revolution: The Invisible Wounds: Part II. Pacific Affairs, 1985. 58(1): p. 5-27.
Ang, P.T., The link between stress and cancer. Mind Your Body, The Straits Times. Singapore, December 22, 2008.
Keinan-Boker, L., et al., Cancer incidence in Israeli Jewish survivors of World War II. Journal of the National Cancer Institute, 2009. 101(21): p. 1489-1500.
MacArthur, A.C., et al., Risk of childhood leukemia associated with parental smoking and alcohol consumption prior to conception and during pregnancy: the cross-Canada childhood leukemia study. Cancer Causes and Control, 2008. 19(3): p. 283-295.
Lin, L., F. Yuxia, and W. Xi, Drug Abuse in China: Past, Present and Future. Cell Molecular Neurobiology, 2008. 28: p. 479–490.
USDA, (US Department of Agriculture), Production, Supply & Distribution, electronic database, updated 30 January, , 2004.
Yang, G., et al., Smoking in China. JAMA: the journal of the American Medical Association, 1999. 282(13): p. 1247-1253.
Peto, R., Z. Chen, and J. Boreham, Tobacco—the growing epidemic in China. JAMA: the journal of the American Medical Association, 1996. 275(21): p. 1683-1684.
Chen, J., et al., Life-style and Mortality in China1990: Oxford University Press.
Hung, E.P.W. and S.W.K. Chiu, The Lost Generation. Modern China, 2003. 29(2): p. 204-236.
Zhao, P., et al., Cancer trends in China. Japanese journal of clinical oncology, 2010. 40(4): p. 281-285.