Abstract
Objective
To investigate the association of early-life exposure to famine with the risk of mortality in later life, as well as to assess the mediating effect of BMI in adulthood on the relationship between early-life famine exposure and long-term mortality risk.
Methods
The China Health and Nutrition Survey (CHNS) served as the basis for the study’s data analysis. A total of 8,085 individuals participated in this study. All subjects were categorized into five groups: no exposure group, fetal exposure group, early childhood exposure group, mid-childhood exposure group, and late childhood exposure group. The association of famine exposure with mortality was analyzed using Cox proportional hazards regression models.
Results
Exposure to famine in mid-childhood and late childhood exposure was significantly associated with an increased mortality risk (HR: 2.453 and 3.016, 95%CI: 1.532–3.927 and 1.830–4.971, P < 0.001 and < 0.001, respectively). BMI mediated the association of famine exposure with mortality in mid-childhood (P = 0.004) and late childhood exposure (P < 0.001), the mediated proportion were 15.01% for exposure in mid-childhood and 10.22% for exposure in late childhood. However, when the sample was restricted to individuals with a narrower age range for enhanced comparability, the effect of fetal exposure became statistically significant (HR: 2.483, 95% CI: 1.346–4.579, P = 0.004).
Conclusion
Exposure to China’s Great Famine during mid- and late childhood resulted in a higher long-term mortality rate. Additionally, BMI acted as a mediator between famine exposure and the mortality. Moreover, the age-limited analysis highlighted the significant effect of fetal exposure on mortality.
Keywords: Chinese famine, Mortality, BMI, Mediating effect, Cohort study
Introduction
The fetal origins hypothesis suggests that malnutrition during fetal development significantly influences the programming of susceptibility to various adult diseases, including hypertension, cardiovascular disease (CVD), and diabetes mellitus [1–3]. However, the impact of nutritional deprivation extends beyond fetal life; periods of food scarcity or malnutrition at any stage of early life increase the risk of chronic diseases in adulthood [4]. Pregnant women experiencing malnutrition may give birth to infants with growth restrictions and a higher risk of neonatal death. Moreover, malnutrition during the first two years of life can lead to stunted growth and increase the likelihood of developing chronic diseases later in life [5], ultimately affecting life expectancy [6]. Despite these known outcomes, little research conducted on the association between malnutrition in early life and mortality risk later in life.
The China’s Great Famine (1959–1961), which affected approximately 600 million people and caused an estimated 30 million premature deaths [7]. This tragic event provides a unique opportunity to analyze the association between malnutrition in early life and mortality in later life. It is widely acknowledged that early exposure to unfavorable environments, particularly during the critical period of fetal development, has a long-lasting influence on the emergence of chronic illnesses in later life, such as overweight/obesity [8, 9]. Therefore, it is hypothesized that early-life famine exposure is correlated with mortality in later life. Moreover, the body mass index (BMI) in adulthood might have a mediating effect on the association of the famine exposure with the risk of mortality.
While the link between early-life famine and adult morbidity (e.g., diabetes, CVD) is well-documented, the impact on long-term all-cause mortality remains less explored. This study aims to fill this gap by examining how nutritional shocks in early life influence lifespan reduction over several decades. For a robust analysis, we utilized data from the China Health and Nutrition Survey (CHNS), which provides several advantages, including broad geographical coverage across multiple provinces and extensive longitudinal follow-up spanning 26 years (1989–2015). The long-term tracking of a large-scale national sample enables a more comprehensive assessment of mortality trajectories compared to smaller, localized studies.
The primary goal of this study is to explore the association between early-life famine exposure and mortality risk, using data from the CHNS. Additionally, we aim to assess the potential mediating effect of BMI in adulthood on this relationship. The findings of this study could inform early-life nutrition interventions, ultimately contributing to improved health outcomes and increased life expectancy.
Materials and methods
Study design and setting
This study was conducted upon an ongoing, prospective cohort study of the China Health and Nutrition Survey (CHNS), which is an international collaboration between the University of North Carolina and the Nutrition and Health Center of the Chinese Center for Disease Control and Prevention [10]. Objective to reflecting the nutritional and physical condition of the Chinese population, a sample selected through a method of multistage, random cluster sampling was applied to establish the CHNS [11]. The program was initiated in 1989 and subsequently included nine waves of measurements in nine provinces and two autonomous regions in 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011, and 2015.
Participants
As much of the Chinese famine burden from 1959 to 1961, individuals born between 1949 and 1965 among the data were included in the analyzed population. Individual demographic information with complete values on age, sex, BMI, ethnicity, area of residence, gross family income, current smoking, alcohol consumption, physical activity, prevalence of hypertension, CVD, and diabetes mellitus, year of birth, famine exposure, severity of famine exposure, survival status, and survival time was included. Participants with implausible BMI values (< 10 or > 60 kg/m²) were excluded to remove potential data-entry or measurement errors. After excluding missing values and outliers, the study obtained a total of 8,085 data points [Figure 1]. This study was granted approval by the Institutional Review Board (IRB) of the National Institute for Nutrition and Food Safety, China Center for Disease Control and Prevention, as well as University of North Carolina at Chapel Hill. It is important to note that written informed consent was obtained from all participants.
Fig. 1.
Flow diagram of the screening and enrollment of study participants
Famine exposure
The famine in China lasted for three years from 1959 to 1961, and people born in different years may experience the Chinese Great Famine at different age in their childhood. Exposure to famine was defined by the year of birth, and participants were categorized into five groups: (1) No exposure group included individuals born between 1962 and 1965 served as a reference group because they were born after the period of the famine; (2) Fetal exposure group included individuals born between 1959 and 1961 who were exposed to famine while in the womb; (3) Early childhood exposure group included individuals born between 1956 and 1958 who were exposed to famine as toddlers; (4) Mid-childhood exposure group included individuals born between 1953 and 1955 who were exposed to famine during preschool ages; (5) Late childhood exposure group included individuals born between 1949 and 1952 who were exposed to famine during school ages [12].
Ascertainment of deaths
Data recorded prior to the last 2015 survey were used to determine the survival status and time of death for subjects in this study. At each survey, investigators recorded the survival status (alive or dead) and the death time of participants reported by family members. The first reported outcome was chosen when reporting was repeated in different rounds of the survey. Due to the specific cause of death was not reported during the survey, all-cause mortality was used as the end-event in this study.
Covariates
Potential confounders included age, sex, BMI, ethnicity (Han vs. others), area of residence (rural vs. urban), famine severity (less severe vs. severe), gross family income (≤ 20000 RMB vs. > 20000 RMB), current smoking (yes vs. no), current alcohol consumption (yes vs. no), physical activity (adequate vs. inadequate), history of hypertension (yes vs. no), history of diabetes (yes vs. no) and history of cardiovascular disease (CVD) (yes vs. no).
Since the severity of famine varies between regions and years, excess mortality rate (EDR) was considered as an indicator for famine severity grading to evaluate the time, region and scale of famine exposure. EDR was calculated by increasing the average level in 1956–1958 to the maximum level in 1959–1961. Excess deaths in 100% were used as the threshold for famine exposure classification. Greater than 100% were classified as severely affected areas of famine exposure and less than 100% were classified as less affected areas of famine [13].
Besides, the gross family income was normalized using 2011 as the standard and split at 20,000 RMB into two levels. Previous research based on CHNS perceives at least 150 min of moderate or vigorous activity per week as “Adequate physical activity”, and others as “Inadequate physical activity“ [14].
Statistics analysis
Baseline characteristics of the population were categorized into five groups according to various famine exposure periods: no exposure, fetal childhood exposure, early childhood exposure, mid-childhood exposure, and late childhood exposure. Continuous variables, i.e. age and BMI, were compared using an Analysis of Variance (ANOVA) test, while categorical variables were compared using chi-square analysis. The trend test was utilized to investigate the trend of each variable throughout five periods of famine exposure. In parallel, the association of famine exposure with mortality was analyzed using Cox proportional hazards regression models. The crude model did not adjust for any covariate, while the adjusted model incorporated all covariates, including age, BMI, sex, ethnicity, gross family income, area of residence, severity of famine exposure, current smoking, current alcohol consumption, physical activity, history of hypertension, CVD, and diabetes mellitus. The impacts of all variables on survival risk satisfied proportion hazards assumption. Sex differences in the long-term mortality risk of early-life exposure to famine was explored in the stratified analysis by sex. Famine severity was perceived as an effect modifier and used in the subgroup analysis of the association of famine exposure with mortality. Additionally, bootstrap analysis was used to assess the mediating effects of BMI on the association between famine exposure and mortality.
We employed several robust statistical approaches to minimize potential confounding and ensure the validity of our findings. First, a sensitivity analysis was performed to demonstrate the stability of the model by excluding individuals born in 1962 and 1958. Second, inverse probability weighting was applied to balance the distribution of baseline covariates, with a particular focus on the age structure, across different exposure groups. By assigning weights based on propensity scores, we constructed a pseudo-population in which famine exposure was independent of measured confounders, thereby reducing selection bias. Third, to address the inherent age gap between cohorts, an age-limited sensitivity analysis was conducted. This analysis was restricted to individuals whose age did not exceed the mean age of the unexposed group, thereby confining the sample to a narrower overlapping age range and enhancing the comparability between the exposed and unexposed groups. Finally, martingale residual analysis was utilized to further isolate the independent effect of famine from chronological age. We first fitted a Cox proportional hazards model using age as the sole covariate to account for its contribution to mortality. The resulting martingale residuals, which represent the unexplained variance in mortality risk after fully adjusting for age, were then analyzed via linear regression to evaluate their specific association with famine exposure.
All analysis were carried out using R 4.2.2, and a statistically significant result was defined as a two-tailed P value less than or equal to 0.05.
Results
The study involved a total of 8,085 participants. Among them, 228 individuals were reported to have deceased, resulting in an observed mortality rate of 2.82%. Total population average age was 53.21 ± 8.34 years, and the mean follow-up was 53.92 ± 8.08 years. All subjects were divided into no exposure (n = 2253, 27.87%), fetal exposure (n = 1031, 12.75%), early childhood exposure (n = 1449, 17.92%), mid-childhood exposure (n = 1594, 19.72%), and late childhood exposure (n = 1758, 21.74%) at baseline, respectively [Table 1]. There were no significant differences or discernible trends observed pertaining to the sex, BMI, or current smoking status of the groups. The gross family income among the groups didn’t exist linear trend, nonetheless, there were significant differences in family income between the groups. Additionally, there were significant differences and linear trends in other characteristics among the groups.
Table 1.
The characteristics of all participants at baseline
| Characteristics | All subjects (N = 8085) |
No exposure (n = 2253) |
Fetal exposure (n = 1031) |
Early childhood exposure (n = 1449) |
Mid-childhood exposure (n = 1594) |
Late childhood exposure (n = 1758) |
P | Ptrend b |
|---|---|---|---|---|---|---|---|---|
| Age (mean ± SD, years) a | 53.21 ± 8.34 | 46.38 ± 6.93 | 50.38 ± 6.65 | 53.54 ± 6.51 | 56.67 ± 6.29 | 60.21 ± 6.09 | < 0.001 | < 0.001 |
| BMI (mean ± SD, kg/m2) a | 24.26 ± 3.64 | 24.26 ± 3.68 | 24.48 ± 3.91 | 24.14 ± 3.66 | 24.11 ± 3.41 | 24.36 ± 3.61 | 0.052 | 0.818 |
| Sex (n (%)) | 0.958 | 0.635 | ||||||
| Male | 3889 (48.10) | 1080 (47.94) | 493 (47.82) | 696 (48.03) | 760 (47.68) | 860 (48.92) | ||
| Female | 4196 (51.90) | 1173 (52.06) | 538 (52.18) | 752 (51.97) | 834 (52.32) | 898 (51.08) | ||
| Ethnicity (n (%)) | 0.021 | 0.029 | ||||||
| Han | 7288 (90.14) | 2036 (90.37) | 941 (91.27) | 1329 (91.72) | 1425 (89.40) | 1557 (88.57) | ||
| Other | 797 (9.86) | 217 (9.63) | 90 (8.73) | 120 (8.21) | 169 (10.60) | 201 (11.43) | ||
| Area of residence (n (%)) | 0.002 | 0.001 | ||||||
| Urban | 3593 (44.44) | 923 (40.97) | 482 (46.75) | 647 (44.65) | 723 (45.36) | 818 (46.53) | ||
| Rural | 4492 (55.56) | 1330 (59.03) | 549 (53.25) | 802 (55.35) | 871 (54.64) | 940 (53.47) | ||
| Famine severity (n (%)) | < 0.001 | 0.009 | ||||||
| Less severe | 3321 (41.08) | 839 (37.24) | 472 (45.78) | 590 (40.72) | 690 (43.29) | 730 (41.52) | ||
| Severe | 4764 (58.92) | 1414 (62.76) | 559 (54.22) | 859 (59.28) | 904 (56.71) | 1028 (58.48) | ||
| Gross family income (n (%)) | 0.039 | 0.468 | ||||||
| ≤ 20,000 RMB | 2047 (25.32) | 573 (25.43) | 223 (21.63) | 392 (27.05) | 414 (25.97) | 445 (25.31) | ||
| > 20,000 RMB | 6038 (74.68) | 1680 (74.57) | 808 (78.37) | 1057 (72.95) | 1180 (74.03) | 1313 (74.69) | ||
| Current smoking (n (%)) | 0.698 | 0.997 | ||||||
| No | 5666 (70.08) | 1589 (70.53) | 705 (68.38) | 1015 (70.05) | 1130 (70.89) | 1227 (69.80) | ||
| Yes | 2419 (29.92) | 644 (29.47) | 326 (31.62) | 434 (29.95) | 464 (29.11) | 531 (30.20) | ||
| Current alcohol consumption (n (%)) | 0.002 | < 0.001 | ||||||
| No | 5476 (67.73) | 1467 (65.11) | 683 (66.25) | 978 (67.49) | 1102 (69.13) | 1246 (70.88) | ||
| Yes | 2609 (32.27) | 786 (34.89) | 348 (33.75) | 471 (32.51) | 492 (30.87) | 512 (29.12) | ||
| Physical activity (n (%)) | 0.013 | 0.003 | ||||||
| Adequate | 898 (11.11) | 221 (9.81) | 119 (11.54) | 142 (9.80) | 193 (12.11) | 223 (12.68) | ||
| Inadequate | 7187 (88.89) | 2032 (90.19) | 912 (88.46) | 1307 (90.20) | 1401 (87.89) | 1535 (87.32) | ||
| History of hypertension (n (%)) | < 0.001 | < 0.001 | ||||||
| No | 5246 (64.89) | 1623 (72.04) | 719 (69.74) | 948 (65.42) | 961 (60.29) | 995 (56.60) | ||
| Yes | 2839 (35.11) | 630 (27.96) | 312 (30.26) | 501 (34.58) | 633 (39.71) | 763 (43.40) | ||
| History of CVD (n (%)) | < 0.001 | < 0.001 | ||||||
| No | 7917 (97.92) | 2238 (99.33) | 1019 (98.84) | 1426 (98.41) | 1555 (97.55) | 1679 (95.51) | ||
| Yes | 168 (2.08) | 15 (0.67) | 12 (1.16) | 23 (1.59) | 39 (2.45) | 79 (4.49) | ||
| History of diabetes (n (%)) | < 0.001 | < 0.001 | ||||||
| No | 7695 (95.18) | 2198 (97.56) | 985 (95.54) | 1380 (95.58) | 1504 (94.34) | 1628 (92.61) | ||
| Yes | 390 (4.82) | 55 (2.44) | 46 (4.46) | 69 (4.76) | 83 (5.65) | 130 (7.39) | ||
| Survival status (n (%)) | < 0.001 | < 0.001 | ||||||
| Living | 7857 (97.18) | 2213 (98.22) | 1006 (97.58) | 1415 (97.65) | 1544 (96.86) | 1679 (95.51) | ||
| Death | 228 (2.82) | 40 (1.78) | 25 (2.42) | 34 (2.35) | 50 (3.14) | 79 (4.49) |
BMI: body mass index, CVD: cardiovascular disease
a These variables were analyzed by ANOVA-test. The other variables were compared by chi-square test
b P value for trend of all variables among the 5 groups stratified by famine exposure was obtained by Cochran-Armitage trend test
After adjusting for age only [Model 1, Table 2], the mortality risks for mid-childhood (P < 0.001) and late childhood exposure (P < 0.001) were significantly higher compared to the non-exposed group. Mid-childhood exposure was associated with a 190.9% (HR: 2.909; 95% CI: 1.823, 4.643) higher mortality risk, and late childhood exposure had a 289.5% (HR: 3.895; 95% CI: 2.401, 6.318) higher mortality risk. Fetal and early childhood exposures were not significantly associated with mortality risk compared to the non-exposed group (P = 0.066 and 0.051, respectively). After full adjustment for covariates [Model 2, Table 2], the mortality risks of mid-childhood (P < 0.001) and late childhood exposure (P < 0.001) were significantly different from that of non-exposed group. Compared with non-exposed group, mid-childhood famine exposure was related to a 145.3% (HR: 2.453; 95% CI: 1.532, 3.927) higher mortality risk, and late childhood exposure had a 201.6% (HR: 3.016; 95% CI: 1.830, 4.971) higher mortality risk. However, compare to non-exposed group, fetal and early childhood exposure were insignificantly associated with the risk of mortality (P = 0.077 and 0.212, respectively).
Table 2.
HRs (95% CI) for the association of famine exposure with mortality from Cox regression a
| Famine exposure | Model 1 | Model 2 | ||
|---|---|---|---|---|
| HR (95% CI) | P | HR (95% CI) | P | |
| No exposure | Ref | |||
| Fetal exposure | 1.616(0.969,2.697) | 0.066 | 1.593(0.951–2.671) | 0.077 |
| Early childhood exposure | 1.621(0.998,2.633) | 0.051 | 1.367(0.837–2.232) | 0.212 |
| Mid-childhood exposure | 2.909(1.823,4.643) | < 0.001 | 2.453(1.532–3.927) | < 0.001 |
| Late childhood exposure | 3.895(2.401,6.318) | < 0.001 | 3.016(1.830–4.971) | < 0.001 |
a Model 1: Adjustment for age
Model 2: Adjustment for age, sex, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
In males, the mortality risks of mid-childhood (P = 0.015) and late childhood exposure (P = 0.003) were significantly different from that of non-exposed group, the hazards ratios when mid-childhood and late childhood exposure were compared with non-exposed group were 2.138 (95% CI: 1.160–3.943) and 2.732 (95% CI: 1.418–5.265) for mortality risk. However, in female, the mortality risks of fetal exposure (P = 0.020), mid-childhood exposure (P = 0.003) and late childhood exposure (P = 0.001) were significantly different from that of non-exposed group, the hazards ratios when fetal exposure, mid-childhood exposure and late childhood exposure were compared with non-exposed group were 2.520 (95% CI: 1.161–5.473), 3.135 (95% CI: 1.471–6.679), and 4.029 (95% CI: 1.805–8.995) for mortality risk [Table 3].
Table 3.
The association of famine exposure with mortality stratified by sex a
| Subgroups | HR | 95% CI | P | P for interaction b |
|---|---|---|---|---|
| Sex | 0.446 | |||
| Male (n = 3889) | ||||
| No exposure | Ref | |||
| Fetal exposure | 1.225 | 0.607–2.470 | 0.572 | |
| Early childhood exposure | 1.402 | 0.750–2.620 | 0.290 | |
| Mid-childhood exposure | 2.138 | 1.160–3.943 | 0.015 | |
| Late childhood exposure | 2.732 | 1.418–5.265 | 0.003 | |
| Female (n = 4196) | ||||
| No exposure | Ref | |||
| Fetal exposure | 2.520 | 1.161–5.473 | 0.020 | |
| Early childhood exposure | 1.500 | 0.656–3.430 | 0.337 | |
| Mid-childhood exposure | 3.135 | 1.471–6.679 | 0.003 | |
| Late childhood exposure | 4.029 | 1.805–8.995 | 0.001 |
aAdjustment for age, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
b P for interaction indicates the multiplicative interaction between famine exposure and sex on the mortality
As stratified by the severity of famine, the mortality risks of mid-childhood (P = 0.001) and late childhood exposure (P < 0.001) were significantly different from that of non-exposed group among people exposed to less severe famine, the hazards ratios when mid-childhood and late childhood exposure were compared with non-exposed group were 4.770 (95% CI: 1.827–12.451) and 6.936 (95% CI: 2.503–19.219) for mortality risk. In people exposed to severe famine, the mortality risks of mid-childhood (P = 0.023) and late childhood exposure (P = 0.007) were significantly different from that of non-exposed group, the hazards ratios when mid-childhood and late childhood exposure were compared with non-exposed group were 1.920 (95% CI: 1.096–3.363) and 2.258 (95% CI: 1.251–4.077) for mortality risk [Table 4].
Table 4.
The association of famine exposure with mortality stratified by the severity of famine exposure a
| Subgroups | HR | 95% CI | P | P for interaction b |
|---|---|---|---|---|
| Area | 0.115 | |||
| Less severe (n = 3321) | ||||
| No exposure | Ref | |||
| Fetal exposure | 2.165 | 0.725–6.471 | 0.167 | |
| Early childhood exposure | 0.862 | 0.241–3.088 | 0.820 | |
| Mid-childhood exposure | 4.770 | 1.827–12.451 | 0.001 | |
| Late childhood exposure | 6.936 | 2.503–19.219 | < 0.001 | |
| Severe (n = 4764) | ||||
| No exposure | Ref | |||
| Fetal exposure | 1.568 | 0.868–2.832 | 0.136 | |
| Early childhood exposure | 1.514 | 0.879–2.609 | 0.135 | |
| Mid-childhood exposure | 1.920 | 1.096–3.363 | 0.023 | |
| Late childhood exposure | 2.258 | 1.251–4.077 | 0.007 |
a Adjustment for age, sex, BMI, ethnicity, area of residence, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
b P for interaction indicates the multiplicative interaction between famine exposure and the severity of famine exposure on the mortality
With BMI as a mediator, after adjustment for covariates [Table 5], BMI mediated the association of famine exposure with mortality in mid-childhood (P = 0.004) and late childhood exposure (P < 0.001). BMI partly mediates this association, with the proportion being 15.01% for exposure in mid-childhood and 10.22% for exposure in late childhood.
Table 5.
The mediated effect of BMI on the association of famine exposure with mortality a
| Famine exposure | Total effect (95%CI) | P | Direct effect (95%CI) | P | Indirect effect (95%CI) | P | Proportion mediated (%) | P |
|---|---|---|---|---|---|---|---|---|
| No exposure | Ref | Ref | Ref | Ref | ||||
| Fetal exposure | 9.701(-2.077,20.601) | 0.116 | 9.318(-2.423,20.246) | 0.122 | 0.383(-0.212,1.048) | 0.241 | 3.09 | 0.348 |
| Early childhood exposure | 4.494(-9.065,16.741) | 0.492 | 2.629(-10.510,14.848) | 0.681 | 1.864(0.853,3.140) | < 0.001 | 20.49 | 0.492 |
| Mid-childhood exposure | 16.724(7.013,26.476) | 0.004 | 14.210(4.334,23.710) | 0.008 | 2.515(1.236,4.084) | < 0.001 | 15.01 | 0.004 |
| Late childhood exposure | 23.816(14.541,31.590) | < 0.001 | 21.423(11.989.29.375) | < 0.001 | 2.393(1.112,3.849) | < 0.001 | 10.22 | < 0.001 |
a Adjusted for age, sex, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
Sensitivity analysis
To further validate the robustness of our findings and address potential confounding by age, four additional sensitivity analyses were performed.
As accurate dates for the start and end of the Chinese Famine were not available, we used previous literature to infer the most widely accepted definition of the famine period as spanning from January 1, 1959, to December 31, 1961. To avoid misclassification bias and confirm the stability of the results, individuals born in 1958 and 1962 were excluded, and the associations were reanalyzed. Compared with the unexposed group, fetal exposure was associated with a 73.4% (HR: 1.734; 95% CI: 1.002–3.003) higher risk of mortality. While early childhood exposure showed an insignificant association (P = 0.133), mid-childhood and late childhood exposure were associated with a 170.1% (HR: 2.701; 95% CI: 1.612–4.527) and 234.2% (HR: 3.342; 95% CI: 1.937–5.767) higher risk of mortality, respectively [Table 6].
Table 6.
HRs (95% CI) for the association of famine exposure with mortality from Cox regression with removing participants born in 1958 or 1962 a
| Famine(n = 7086) | Total sample (n (%)) | Event (n (%)) | HR (95%CI) | P |
|---|---|---|---|---|
| No exposure | 1700 (23.99) | 30 (1.76) | Ref | |
| Fetal exposure | 1031 (14.55) | 25 (2.42) | 1.743(1.002–3.033) | 0.049 |
| Early childhood exposure | 1003 (14.15) | 24 (2.39) | 1.563(0.873–2.798) | 0.133 |
| Mid-childhood exposure | 1594 (22.50) | 50 (3.14) | 2.701(1.612–4.527) | < 0.001 |
| Late childhood exposure | 1758 (24.81) | 79 (4.49) | 3.342(1.937–5.767) | < 0.001 |
a Adjustment for age, sex, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
We employed inverse probability weighting to balance the distribution of baseline covariates across exposure groups. Compared with the unexposed group, fetal exposure was associated with an 87.4% (HR: 1.874; 95% CI: 1.055–3.330) higher risk of mortality, and early childhood exposure was associated with a 123.8% (HR: 2.238; 95% CI: 1.228, 4.082) higher risk, and mid-childhood exposure was associated with a 357.6% (HR: 4.576; 95% CI: 2.382–8.788) higher risk, and late childhood exposure was associated with a 414.7% (HR: 5.147; 95% CI: 2.522–10.507) higher risk [Table 7].
Table 7.
Association between famine exposure and mortality using inverse probability weighting to balance baseline covariates a
| Famine exposure (N = 8085) | HR | 95% CI | P |
|---|---|---|---|
| No exposure | Ref | ||
| Fetal exposure | 1.874 | 1.055–3.330 | 0.032 |
| Early childhood exposure | 2.238 | 1.228–4.082 | 0.009 |
| Mid-childhood exposure | 4.576 | 2.382–8.788 | < 0.001 |
| Late childhood exposure | 5.147 | 2.522–10.507 | < 0.001 |
a Adjustment for age, sex, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
To minimize the inherent age gap between cohorts, we conducted an age-limited analysis restricted to individuals whose age did not exceed the mean age of the “No exposure” group. In this restricted sample, fetal exposure was associated with a 148.3% (HR: 2.483; 95% CI: 1.346–4.579) higher risk of mortality. Similarly, mid-childhood and late childhood exposure were associated with a 235.2% (HR: 3.352; 95% CI: 1.907–5.890) and 221.3% (HR: 3.213; 95% CI: 1.580–6.533) higher risk of mortality, respectively. Early childhood exposure showed a similar direction but was not statistically significant (HR: 1.786; 95% CI: 0.948–3.332; P = 0.068) [Table 8].
Table 8.
Association between famine exposure and mortality, restricted to individuals whose age does not exceed the mean age of the unexposed group
| Famine exposure (n = 2971) | HR | 95% CI | P |
|---|---|---|---|
| No exposure | Ref | ||
| Fetal exposure | 2.483 | 1.346–4.579 | 0.004 |
| Early childhood exposure | 1.786 | 0.948–3.332 | 0.068 |
| Mid-childhood exposure | 3.352 | 1.907–5.890 | < 0.001 |
| Late childhood exposure | 3.213 | 1.580–6.533 | 0.001 |
a Adjustment for age, sex, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
Finally, we performed a martingale residual analysis to evaluate the mortality risk remaining after fully accounting for chronological age. Linear regression revealed that mid-childhood exposure was associated with significantly higher martingale residuals (β: 0.015; 95% CI: 0.003–0.028), as was late childhood exposure (β: 0.018; 95% CI: 0.006–0.031). Although the coefficients for fetal exposure (β: 0.009; 95% CI: -0.006, 0.023; P = 0.231) and early childhood exposure (β: 0.004; 95% CI: -0.009, 0.017; P = 0.519) were positive, they did not reach statistical significance in this model [Table 9].
Table 9.
Linear regression analysis of martingale residuals from an age-adjusted Cox model by famine exposure group
| Famine exposure (N = 8085) | β | 95% CI | P |
|---|---|---|---|
| No exposure | Ref | ||
| Fetal exposure | 0.009 | -0.006-0.023 | 0.231 |
| Early childhood exposure | 0.004 | -0.009-0.017 | 0.519 |
| Mid-childhood exposure | 0.015 | 0.003–0.028 | 0.019 |
| Late childhood exposure | 0.018 | 0.006–0.031 | 0.003 |
a Adjustment for sex, BMI, ethnicity, area of residence, the severity of famine exposure, gross family income, and current smoking, alcohol consumption, physical activity, and prevalent hypertension, CVD and diabetes mellitus
Discussion
To sum up, in the main results, people exposed to famine in fetal and early childhood had insignificant risk of death later in life, whereas, exposure to famine in mid-childhood and late childhood were significantly related to elevated mortality. In the subgroup analyses, the relationship between famine and mortality is independent of the severity of the famine. As classified according to sex, in females, fetal exposure, exposure to famine in mid-childhood and exposure to famine in late childhood were significantly related to elevated mortality. BMI acted as a mediator between famine exposure and the mortality. Moreover, the age-limited analysis highlighted the significant effect of fetal exposure on mortality.
Our findings, showing that famine exposure during mid- and late childhood significantly raises mortality risk, align with other studies based on the CHNS. These studies suggest that nutritional deprivation during critical growth periods leads to the development of metabolic syndromes such as obesity and insulin resistance [12, 13]. Additionally, our age-limited sensitivity analysis aligns with the Dutch Hunger Winter studies [15], suggesting that prenatal exposure’s effects may be masked by survivor bias and chronological age differences in broader samples [16].
Exposure to the Great Chinese Famine during mid-childhood and late childhood were significantly associated with higher mortality. It indicated that the later the exposure in childhood, the higher the risk of death in later life. The phenomenon could be explained in three aspects. Firstly, health-related fitness, including cardiorespiratory fitness, muscular strength, and muscular endurance, undergoes rapid growth during mid- to late childhood, and inadequate nutrition can negatively impact the normal growth and development of the body and physiological functions [17]. Secondly, since their brains and nerve system are profoundly sensitive to nourishment due to experiencing a high-speed development [18], under-nutrition in this periods may inhibit brain and nerve system development and cause harm to the physical and mental health. Thirdly, the rate of infectious and immunological diseases increased among people in mid- and late childhood because of secondary immunodeficiency, and it may impact their long-term immunity.
The absence of excess mortality in the fetal and early childhood famine-exposed cohorts may seem inconsistent with the fetal programming hypothesis. However, this pattern is likely due to survivor bias, background mortality differences, and age structure effects, rather than the absence of developmental influences. These results highlight the complexity of famine exposure effects, shaped by both early-life conditions and later-life factors. Therefore, these findings should be interpreted with caution within the context of the developmental origins framework.
The impact of famine exposure in early life has been observed to have differing effects on males and females, and it showed a significant difference on exposure to famine in fetus period. The result suggested that sex difference might have originated in uterus. Male had been defined as the independent risk factor for adverse pregnancy outcomes and complications such placenta insufficiency, premature delivery and spontaneous abortion [19]. The male fetuses exposed to famine would have a higher risk of death in early life than female fetus, whereas survival male fetus had a physical advantage compared with female fetuses. Besides, the fetal-origin hypothesis contending that nourishment during the individual’s fetus and infancy programs leads to the development of risk factors for several important diseases later in life [20]. Therefore, these reasons may lead to female, the mortality risks of fetal exposure were significantly different from that of the non-exposed group. Furthermore, the research pointed that sex difference in under-nutrition had been reduced over time [21]. Hence, males and females exposed to famine in same periods had a similar effect on their mortality later in life. In stratified analyses, mortality HRs were numerically higher in regions with less-severe famine. However, the interaction p-value (0.115) suggests no statistically significant difference across severity levels. This trend may reflect survivor bias: in regions with severe famine, the most vulnerable individuals may have died during the famine, leaving a more resilient survivor cohort, which could attenuate the observed HRs.
Our analysis demonstrated that BMI in adulthood mediated the relationship between early-life famine exposure and mortality. Previous studies have shown that famine exposure is consistently associated with an increased risk of obesity, particularly central obesity [22, 23]. This may be due to early malnutrition altering neuroendocrine function, including activation of the hypothalamic-pituitary-adrenal axis, leading to excessive secretion of glucocorticoids and fat accumulation in later life [24, 25]. Epigenetic studies have indicated that early malnutrition may result in aberrant DNA methylation of genes, which has been linked to the development of obesity and insulin resistance in adulthood [26]. Obesity is known to be a main contributor of metabolic disease-related deaths [27]. This association between childhood famine exposure and elevated risk of mortality further reinforces the significance of this finding. Adequate nutrition early in life would decrease the risks of deaths and prolong life expectancy. Therefore, it is crucial that enhancing and concentrating on nutrition early in life.
Due to differences in age between the exposure groups, we conducted several sensitivity analyses to minimize potential confounding by age. The results of these analyses were consistent with the main findings. After adjusting for age-related confounders using inverse probability weighting, restricting the analysis to individuals with similar ages, and evaluating martingale residuals from an age-adjusted Cox model, the associations between famine exposure and mortality remained robust. In particular, individuals exposed to famine during mid- and late childhood consistently showed significantly higher mortality risks. These findings suggest that the observed associations are likely genuine long-term effects of early-life nutritional deprivation, rather than being driven by age structure or methodological artifacts.
A key finding emerged when comparing the main analysis with the age-limited sensitivity analysis. Main analysis fetal exposure showed a non-significant association. However, when the sample was restricted to individuals with a narrower age range to enhance comparability, the effect of fetal exposure became statistically significant. This indicates that the initial analysis likely underestimated the vulnerability of the fetal period due to the inherent age gap between cohorts, thus providing stronger support for the fetal origins hypothesis.
The long-term health consequences of famine exposure are shaped by various interacting factors, such as the intensity and duration of the famine, the speed of recovery to a normal or richer diet, and the presence of later-life risk factors for non-communicable diseases (NCDs) or type 2 diabetes (T2D), in line with the capacity–load hypothesis. Survivor bias should also be considered—severe famines with high short-term mortality may leave a particularly resilient survivor population, while milder famines may produce different long-term outcomes. Therefore, the effects of famine exposure may vary across regions and countries, with different associations depending on local famine characteristics and post-famine environments.
While famine studies serve as unique ‘natural experiments’ to examine long-term health outcomes, interpretations of causal links must be approached with caution due to inherent methodological limitations in the existing literature. As highlighted by recent rigorous reviews, a major threat to validity in Chinese famine studies is the uncontrolled age difference between comparison groups [15, 28]. A recent systematic review and meta-analysis suggests that many reported positive associations may be driven by the aging process rather than famine exposure itself. Furthermore, studies often suffer from poorly assessed famine intensity and unsuitable study settings [16]. Therefore, to advance the field, future research must move beyond simple cohort comparisons. Studies with more rigorous methods, including age-balanced controls and robust famine intensity measures, such as age–period–cohort (APC) models, difference-in-differences (DID) analyses, and residual analyses to distinguish genuine famine effects from secular time trends and other confounding influences [15, 16]. We used excess mortality rates to stratify famine severity; however, we acknowledge the growing body of literature proposing more systematic and refined approaches for measuring famine intensity [29–32]. Adopting these refined metrics in future research will be pivotal for verifying dose-response effects and minimizing potential classification bias.
Strengths and Limitations
The robustness lied in the compelling evidence linking famine exposure to the risk of mortality, and its investigation of BMI as a possible mediator of this association. The primary strength of this study is the use of the CHNS, a large-scale national longitudinal survey with broad geographical coverage across multiple provinces. However, the limitations of this study also were stated. Given that famine could generate the short-term excess mortality, there was a survivorship bias caused by different excess mortality during Chinese famine. The part of fetuses, infants, and children exposed to famine died during three-years period, and survivors may be healthier and expose to less severe famine. In addition, the fetus is more fragile and has a higher risk of death compared with infants and children, which could exist selection bias. The survey was based on questionnaires that were answered by respondents and their families. There were recalling bias. Besides, we are using data up to 2015. The data from subsequent surveys after 2015 has not yet been made public, but up to 2015, the target cohorts (born 1949–1965) had reached approximately 60 years of age, which is enough to observe long-term survival outcomes. Furthermore, given the universal biological mechanisms underlying early-life nutritional shocks, our findings offer significant generalizability to other geographical regions facing similar challenges.
Conclusions
In conclusion, experiencing famine during childhood has been proven to significantly increase the risk of mortality in later life, with a stronger correlation found among those who were exposed to famine at a comparative later age in childhood. Results confirmed the effect of experiencing famine on the risk of death was partially mediated by BMI. Accordingly, we should lay more emphasis on nutrition for children to extend the life expectancy.
Acknowledgements
Not applicable.
Abbreviations
- CHNS
China Health and Nutrition Survey
- CVD
cardiovascular disease
- BMI
body mass index
- EDR
excess mortality rate
Author contributions
SZ contributed to data curation, formal analysis and writing the original draft. ZY and SW contributed to analysis and interpretation of data and have drafted the work; YC contributed to study design, review and editing the draft. All authors read and approved the final manuscript.
Funding
Role of the funding source: National Natural Science Foundation of China (81903416).
Data availability
The datasets analysed during the current study are available in the China Health and Nutrition Survey, [https://www.cpc.unc.edu/projects/china](https:/www.cpc.unc.edu/projects/china).
Declarations
Ethics approval and consent to participate
This CHNS was granted approval by the Institutional Review Board (IRB) of the National Institute for Nutrition and Food Safety, China Center for Disease Control and Prevention, as well as University of North Carolina at Chapel Hill. It is important to note that written informed consent was obtained from all participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Shichao Zhao, Zhiying Yin and Shuangqing Wang contributed equally to this work.
References
- 1.Harding JE. The nutritional basis of the fetal origins of adult disease. Int J Epidemiol. 2001;30:15–23. [DOI] [PubMed] [Google Scholar]
- 2.Chen C, Nie Z, Wang J, Ou Y, Cai A, Huang Y, et al. Prenatal exposure to the Chinese famine of 1959-62 and risk of cardiovascular diseases in adulthood: findings from the China PEACE million persons project. Eur J Prev Cardiol. 2022;29:2111–9. [DOI] [PubMed] [Google Scholar]
- 3.Liu H, Chen X, Shi T, Qu G, Zhao T, Xuan K, et al. Association of famine exposure with the risk of type 2 diabetes: A meta-analysis. Clin Nutr. 2020;39:1717–23. [DOI] [PubMed] [Google Scholar]
- 4.Campisano S, La Colla A, Echarte SM, Chisari AN. Interplay between early-life malnutrition, epigenetic modulation of the immune function and liver diseases. Nutr Res Rev. 2019;32:128–45. [DOI] [PubMed] [Google Scholar]
- 5.Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, de Onis M, et al. Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet. 2013;382:427–51. [DOI] [PubMed] [Google Scholar]
- 6.Foreman KJ, Marquez N, Dolgert A, Fukutaki K, Fullman N, McGaughey M, et al. Forecasting life expectancy, years of life lost, and all-cause and cause-specific mortality for 250 causes of death: reference and alternative scenarios for 2016-40 for 195 countries and territories. Lancet. 2018;392:2052–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Smil V. China’s great famine: 40 years later. BMJ. 1999;319:1619–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hanson MA, Gluckman PD. Early developmental conditioning of later health and disease: physiology or pathophysiology? Physiol Rev. 2014;94:1027–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jiang H, Yu Y, Li L, Xu W. Exposure to the Great Famine in Early Life and the Risk of Obesity in Adulthood: A Report Based on the China Health and Nutrition Survey. Nutrients. 2021;13:1285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zhang B, Zhai FY, Du SF, Popkin BM. The China Health and Nutrition Survey, 1989–2011. Obes Rev. 2014;15(Suppl 1 0 1):2–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Zhang Y-B, Li Y, Geng T-T, Pan X-F, Zhou Y-F, Liu G, et al. Overall lifestyles and socioeconomic inequity in mortality and life expectancy in China: the China health and nutrition survey. Age Ageing. 2022;51:afac167. [DOI] [PubMed] [Google Scholar]
- 12.Yu C, Wang J, Wang F, Han X, Hu H, Yuan J, et al. Victims of Chinese famine in early life have increased risk of metabolic syndrome in adulthood. Nutrition. 2018;53:20–5. [DOI] [PubMed] [Google Scholar]
- 13.Li Y, Jaddoe VW, Qi L, He Y, Lai J, Wang J, et al. Exposure to the Chinese famine in early life and the risk of hypertension in adulthood. J Hypertens. 2011;29:1085–92. [DOI] [PubMed] [Google Scholar]
- 14.Xi L, Yang X, Wang R, Ku C, Wu B, Dai M, et al. Waist Circumference-Years Construct Analysis and the Incidence of Type 2 Diabetes: China Health and Nutrition Survey, 1997–2015. Nutrients. 2022;14:4654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lumey LH, Stein AD, Susser E. Prenatal famine and adult health. Annu Rev Public Health. 2011;32:237–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Li C, Lumey LH. Early-Life Exposure to the Chinese Famine of 1959–1961 and Type 2 Diabetes in Adulthood: A Systematic Review and Meta-Analysis. Nutrients. 2022;14:2855. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Black RE, Allen LH, Bhutta ZA, Caulfield LE, de Onis M, Ezzati M, et al. Maternal and child undernutrition: global and regional exposures and health consequences. Lancet. 2008;371:243–60. [DOI] [PubMed] [Google Scholar]
- 18.Prado EL, Dewey KG. Nutrition and brain development in early life. Nutr Rev. 2014;72:267–84. [DOI] [PubMed] [Google Scholar]
- 19.Broere-Brown ZA, Adank MC, Benschop L, Tielemans M, Muka T, Gonçalves R, et al. Fetal sex and maternal pregnancy outcomes: a systematic review and meta-analysis. Biol Sex Differ. 2020;11:26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Barker DJ. Fetal origins of coronary heart disease. BMJ. 1995;311:171–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Garenne M, Thurstans S, Briend A, Dolan C, Khara T, Myatt M, et al. Changing sex differences in undernutrition of African children: findings from Demographic and Health Surveys. J Biosoc Sci. 2022;54:847–57. [DOI] [PubMed] [Google Scholar]
- 22.Jornayvaz FR, Vollenweider P, Bochud M, Mooser V, Waeber G, Marques-Vidal P. Low birth weight leads to obesity, diabetes and increased leptin levels in adults: the CoLaus study. Cardiovasc Diabetol. 2016;15:73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Liu D, Yu D, Zhao L, Fang H, Zhang J, Wang J, et al. Exposure to Famine During Early Life and Abdominal Obesity in Adulthood: Findings from the Great Chinese Famine During 1959–1961. Nutrients. 2019;11:903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bale TL, Baram TZ, Brown AS, Goldstein JM, Insel TR, McCarthy MM, et al. Early life programming and neurodevelopmental disorders. Biol Psychiatry. 2010;68:314–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Barker DJP, Lampl M, Roseboom T, Winder N. Resource allocation in utero and health in later life. Placenta. 2012;33(Suppl 2):e30–34. [DOI] [PubMed] [Google Scholar]
- 26.Tobi EW, van den Heuvel J, Zwaan BJ, Lumey LH, Heijmans BT, Uller T. Selective Survival of Embryos Can Explain DNA Methylation Signatures of Adverse Prenatal Environments. Cell Rep. 2018;25:2660–e26674. [DOI] [PubMed] [Google Scholar]
- 27.Chong B, Kong G, Shankar K, Chew HSJ, Lin C, Goh R, et al. The global syndemic of metabolic diseases in the young adult population: A consortium of trends and projections from the Global Burden of Disease 2000–2019. Metabolism. 2023;141:155402. [DOI] [PubMed] [Google Scholar]
- 28.Li C, Tobi EW, Heijmans BT, Lumey LH. The effect of the Chinese Famine on type 2 diabetes mellitus epidemics. Nat Rev Endocrinol. 2019;15:313–4. [DOI] [PubMed] [Google Scholar]
- 29.Lumey LH, Khalangot MD, Vaiserman AM. Association between type 2 diabetes and prenatal exposure to the Ukraine famine of 1932-33: a retrospective cohort study. Lancet Diabetes Endocrinol. 2015;3:787–94. [DOI] [PubMed] [Google Scholar]
- 30.Lumey LH, Li C, Khalangot M, Levchuk N, Wolowyna O. Fetal exposure to the Ukraine famine of 1932–1933 and adult type 2 diabetes mellitus. Science. 2024;385:667–71. [DOI] [PubMed] [Google Scholar]
- 31.Xu H, Zhang Z, Li L, Liu J. Early life exposure to China’s 1959-61 famine and midlife cognition. Int J Epidemiol. 2018;47:109–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Li C, Zhou Z, Lumey LH. Early-life exposure to the Chinese famine and tuberculosis risk: Unrecognized biases from different measures of famine intensity. Proc Natl Acad Sci U S A. 2021;118:e2102809118. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets analysed during the current study are available in the China Health and Nutrition Survey, [https://www.cpc.unc.edu/projects/china](https:/www.cpc.unc.edu/projects/china).