Abstract

Objectives

Previous studies had reported the association between famine exposure in early life and subsequent non-communicable diseases risk. In current study, we aimed to evaluate the associations between famine exposure on multimorbidity prevalence and incidence in middle-aged and older Chinese population.

Methods

A total of 13,254 participants from the China Health and Retirement Longitudinal Study 2011 were included in cross-sectional analyses. 5,780 participants were including in longitudinal analyses and were followed up in 2020. Based on the questionnaire, participants were divided into non-exposed, mild famine, moderate famine and severe famine subgroups.

Results

In cross-sectional analyses, both mild, moderate and severe famine groups were significantly associated with higher multimorbidity prevalence. During the 9 years of follow-up, a total of 2,643(45.73%) participants with multimorbidity incident were identified. After adjusting for all covariates, individuals who experienced mild famine, moderate famine and severe famine were significantly associated with increased risk of multimorbidity incident, with the corresponding ORs (95%CI) being 1.32(1.14鈥�1.52), 1.54(1.21鈥�1.96) and 1.62(1.32鈥�1.99), respectively.

Discussion

Our findings indicate that famine exposure in mild, moderate and severe famine groups might be associated with an increased multimorbidity risk.

Introduction

Multimorbidity, the presence of two or more long-term conditions, is a growing global challenge, which brings heavy burden on individuals, caregivers and society [1]. It is reported that the estimated prevalence of multimorbidity increased rapidly with age, which ranges from 3鈥�68% [2,3,4]. Compared to individuals with a single chronic condition, those with multimorbidity are also associated with poorer function and health-related quality of life, mental illness and intake of multiple drugs as well as greater socioeconomic costs [5,6,7,8]. Furthermore, people with multimorbidity are more likely to die prematurely, be admitted to hospital and have an increased length of stay [9, 10].

In recent years, the 鈥淒evelopmental origins of health and disease鈥� hypothesis suggested that early life exposure to malnutrition would affect the health in adulthood [11, 12]. Due to the ethical limitations, famine, which may cause severe malnutrition, was considered as a natural experiment to assess the impact of early life undernutrition on adverse health outcomes in adulthood. The Dutch famine of 1944鈥�1945 lasted for a shorter period of time, had an abrupt beginning and end in an otherwise affluent and well-nourished population, which offers a unique opportunity to consider whether timing of famine exposure during different pregnancy trimesters may lead to differences in adult diseases [13]. Comparted the Dutch Famine, the Great Chinese Famine from late 1950s to early 1960s lasted for a longer period and was more severe in terms of extent. Almost the entire mainland China was stuck by extreme food scarcity, leading to about 30听million premature deaths during this period [14]. A growing evidence based on the Dutch Famine and the Great Chinese Famine suggested that both prenatal exposure, postpartum exposure, childhood famine exposure and adolescence exposure to the famine might increase the risk of type 2 diabetes [15], obesity [16], coronary heart disease [17], atherogenic lipid profile [18], hypertension [19], microalbuminuria [20], depression [21], stroke [22] and other noncommunicable diseases in adulthood [23]. Evidence from the China Kadoorie Biobank, Guangzhou Nutrition and Health Study, Guangdong Gut Microbiome Project, and China Health and Nutrition Survey (CHNS) also found prenatal exposure to the Chinese famine might be associated with an increased diseases risk [23, 24]. In addition, two large cross-sectional KORA (Cooperative Health Research in the Region of Augsburg) -Age studies in 2008/9 and 2016 found adverse early-life exposure (born during the late war phase) may increase the risk of multimorbidity in adults aged 65鈥�71 years [25]. However, most of the studies were cross-sectional, and no previous study had examined the association between early life undernutrition and multimorbidity prevalence and incidence among Chinese.

The Great Famine in China (1959鈥�1961) was one of the largest famines in human history, resulting in insufficient nutritional supply for a large number of people exposed to the famine environment. Therefore, in current study, we aimed to examine the association between exposure to the Great Chinese Famine and the prevalence and incidence of multimorbidity in adults, using data from the China Health and Retirement Longitudinal Study (CHARLS).

Methods

Study population

The CHARLS is an ongoing nationally representative and population-based study, that uses a multistage clustering sample method to select participants and conducted to collect a series of data regarding demographics, economic status, social networks, physical and psychological health in China [26]. The first visit was accomplished in 2011鈥�2012 (Wave 1) of 17,708 patients, subsequently four follow-up visits carried out after that, each nearly two years apart among survivors (2013鈥�2014: Wave 2, 2015鈥�2016: Wave 3, 2017鈥�2018: Wave 4, 2019鈥�2020: Wave 5). The ethics application for collecting data on human subjects in CHARLS was approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052-11015), and all CHARLS participants provided written informed consent. The details of the CHARLS data are available at its website (). The subjects were treated according to the declaration of Helsinki.

In current study, we conducted a cross-sectional and a longitudinal analysis using data from the four waves of CHARLS (from 2011 to 2020). In the cross-sectional analysis, we included participants according to the following criteria: [1] individuals鈥夆墺鈥�45 years old [2], individuals with complete information about date of famine information and multimorbidity in 2011. A total of 13,254 participants were included (Fig.听1). In the longitudinal analysis, we further included participants according to the following criteria: [1] individuals without multimorbidity in baseline [2], individuals with complete information about multimorbidity in follow-up [3], individuals who were successfully followed-up. Finally, a total of 5,780 individuals were eligible for longitudinal analysis (Fig.听1).

Fig. 1

Flow chart of sample selection and the exclusion criteria

Famine exposure and grouping

In current study, the famine exposure was defined according to the 2014 life course survey with three questions according to previous reported studies [27]: [1] Between 1958 and 1962 did you and your family (including your grandparents, parents, siblings, children and so on) experience starvation? If the respondents answered 鈥淵es鈥�, they would be defined as mild famine exposure [2]. During those days, did you and your family (including your grandparents, parents, siblings, children and so on) move away from the famine-stricken area? If the respondents answered 鈥淵es鈥�, they would be defined as moderate famine exposure [3]. During those days, had any of your family (including your grandparents, parents, siblings, children and so on) starved to death? If the respondents answered 鈥淵es鈥�, they would be defined as severe famine exposure. Furthermore, similar to other studies [22, 28], we used birth date as a surrogate variable for famine exposure in sensitivity analysis, converting the lunar calendar to the solar calendar date, and participants in this study were divided into four groups based on birth date: Non-exposed group (born between October 1, 1962 and September 30, 1964); Fetal exposure group (born between October 1, 1959 and September 30, 1961); Preschool-age exposed group (born between October 1, 1954 and September 30, 1956); School-age exposed group (born between October 1, 1951 and September 30, 1953).

Definition of chronic diseases and multimorbidity

Multimorbidity was defined by the coexistence of two or more of these 14 chronic conditions in the same individual [29, 30]. Participants without any chronic disease or with only one chronic disease were classified into non-multimorbidity group. Heart disease, stroke, chronic lung disease, asthma, kidney disease, liver disease, digestive disease, cancer, psychiatric disease, memory-related disease (including dementia, Parkinson鈥檚 disease, and cerebral atrophy), and arthritis were defined by self-reported physician diagnosis. Blood pressure was measured three times and hypertension was assessed as mean systolic blood pressure (SBP)鈥夆墺鈥�140 mmHg, and/or mean diastolic blood pressure (DBP)鈥夆墺鈥�90 mmHg, and/or using anti-hypertensive drugs. Venous blood samples were also collected from each participant to obtain levels of blood glucose and glycated hemoglobin. Diabetes was diagnosed as meeting any of following criteria: fasting plasma glucose鈥夆墺鈥�7.0 mmol/L, random plasma glucose鈥夆墺鈥�11.1 mmol/L, or using glucose-lower drugs/insulin treatment. Dyslipidemia was defined as triglycerides鈥夆墺鈥�150听mg/dl, total cholesterol鈥夆墺鈥�240听mg/dl, high density lipoprotein cholesterol鈥�<鈥�40听mg/dl, low density lipoprotein鈥夆墺鈥�160听mg/dl, current use of the lipid-lowering medications, or self-reported history of dyslipidemia.

Covariates assessments

The covariates were collected at baseline including age, sex, place of residence (rural vs. urban), smoking status (ever smoking vs. never smoking), educational level (illiteracy; primary school; middle school; high school or above), drinking status (ever drinking vs. never drinking), marriage status and family鈥檚 economic status (鈮モ�塧verage level or <鈥塧verage level, which was assessed by the individuals鈥� standard of living).

Statistical analyses

Participants鈥� baseline characteristics are presented as percentages for categorical variables, as the means with standard deviation for normally distributed continuous variables and as medians with interquartile range for non-normally distributed variables. Demographic and clinical characteristics were compared by Kruskal-Wallis test for continuous variables and 蠂² test for categorical variables among participants with different famine exposure.

In cross-sectional and longitudinal analyses, multivariable logistic regression models were applied to calculate the odds ratio (OR) and 95% confidence interval (95% CI) between famine exposure with the risk of multimorbidity. A Poisson regression model with quasi- likelihood estimation was applied to examine whether famine exposure status could increase the number of morbidities. Furthermore, we explored the associations between famine exposure status with risk of 14 chronic diseases separately. In all models, age, sex, BMI, educational level, marriage status, family鈥檚 economic status, cigarette smoking, alcohol consumption, and residential locations were commonly adjusted. Subgroup analyses were performed to evaluate the association between famine exposure status and the risk of multimorbidity according to sex, age, place of residence, smoking, drinking and family鈥檚 economic status. Two tailed P鈥�<鈥�0.05 was considered statistical significance. All statistical analyses were conducted using SAS statistical software (version 9.4, Cary, NC).

Results

Characteristics of participants in the cross鈥憇ectional study

In the cross鈥憇ectional study, a total of 13,254 participants (6,414 men and 6,840 women) were included, and the average age was 59.27鈥壜扁��9.35 years. Among them, there were 2,217 (16.73%) in non-exposed famine group, 8,675 (66.45%) in the mild famine group, 830 (6.26%) in the moderate famine group, and 1,532 (11.56%) in the severe famine group, respectively. There were significant differences among the four subgroups in terms of age, sex, living place, smoking status, educational level, smoking and drinking (Table听1).

Table 1 Baseline characteristics of the study participants according to exposure to the Chinese famine (N鈥�=鈥�13,254)

Cross-sectional associations of famine exposure status with multimorbidity prevalence and chronic conditions

In the cross-sectional study, the overall prevalence of chronic multimorbidity was 40.12%. Table听2 shows the associations of famine exposure status with multimorbidity prevalence. Compared with the non-exposed group, the prevalence of multimorbidity was higher in mild, moderate and severe famine groups (35.50% vs. 40.09%, 40.72%, 46.67%, P鈥�<鈥�0.001). After controlling for the potential confounders, those in mild, moderate and severe famine groups had higher risk of multimorbidity with the corresponding ORs (95%CI) being 1.25(1.13鈥�1.38), 1.25(1.06鈥�1.48) and 1.62(1.41鈥�1.85), respectively, compared to the non-exposed group (Table听2). Similarly, the Poisson regression results showed that individuals in mild, moderate and severe famine groups were positively related to an increased number of morbidities in cross-sectional analysis (Table听3). In the subgroup analysis, the significant associations between severe famine with higher risk of multimorbidity were observed in almost all subgroups, expect for male, age鈥夆墺鈥�60 and living in urban subgroups (Table S1). Table S2 shows the association of famine exposure status with the risk of 14 chronic conditions. After controlling for confounders, we found that individuals in severe famine group were significantly associated with increased risks of hypertension, dyslipidemia, diabetes, heart disease, chronic lung disease, asthma, liver disease, cancer, digestive disease, arthritis and psychiatric disease. When birth date was used as a surrogate variable for famine exposure in sensitivity analysis, those in fetal, preschool and school-aged exposed groups had higher risk of multimorbidity with the corresponding ORs (95%CI) being 1.22(1.06鈥�1.41), 1.68(1.47鈥�1.92) and 2.11(1.84鈥�2.41), respectively, compared to the non-exposed group (Table听4).

Table 2 Association of famine exposure with multimorbidity prevalence in the cross-sectional study (N鈥�=鈥�13,254)

Table 3 Associations of famine exposure with the number of morbidities in cross-sectional and longitudinal study

Table 4 Sensitivity analysis the association of famine exposure with multimorbidity prevalence and incidence

Longitudinal association between famine exposure status with multimorbidity incident and chronic conditions

From 2011 to 2020, a total of 1,704(43.19%) participants with multimorbidity incident were identified. After adjusting for all covariates, individuals who mild, moderate and severe famine were significantly associated with increased risk of multimorbidity incident, with the corresponding ORs (95%CI) being 1.25(1.13鈥�1.38), 1.25(1.06鈥�1.48) and 1.62(1.41鈥�1.85), respectively (Table听5). The Poisson regression results also showed that individuals with severe famine was positively related to an increased number of morbidities in longitudinal study (Table听3). In the subgroup analysis, the significant associations between severe famine with risk of multimorbidity incident were observed in almost all subgroups (Table S3). As shown in Table S4, individuals with severe famine were more likely to have a higher risk of hypertension, diabetes, heart disease, stroke, chronic lung disease, digestive disease, kidney disease, arthritis and psychiatric disease. Similarly, in the sensitivity analysis, individuals who experienced preschool and school-aged exposed famine were significantly associated with increased risk of multimorbidity incident, with the corresponding OR (95%CI) was 1.32(1.10鈥�1.60) and 1.32(1.09鈥�1.61), respectively (Table听4).

Table 5 Association of famine exposure with multimorbidity incident in longitudinal study (from 2011 to 2020, N鈥�=鈥�5,780)

Discussion

In this nationwide longitudinal prospective cohort study of Chinese adults, we first demonstrated that populations who were exposed to Chinese famine in early life stage had higher risk of multimorbidity. Exposure to the Great Chinese Famine was positively related to multimorbidity prevalence and multimorbidity incident in adult. In addition, the main results were consistent, when birth date was used as a surrogate variable for famine exposure. All those evidences provided a more valid association between famine exposure and multimorbidity risk. The results might help to elucidate the pathogenesis of multimorbidity and emphasize the importance of adequate nutrition during the early life stage.

Actually, studies on associations between famine exposure and individuals鈥� chronic disease started from the Dutch famine studies in 1970s and afterwards [31]. As of December 2023, more than 300 original research articles had been published relating nineteenth-century crop failures in Sweden and Finland, the Siege of Leningrad of 1941鈥�1944, the Dutch Hunger Winter of 1944鈥�1945, seasonal famines in the Gambia between 1949 and 1994, the Chinese Great Leap Forward famine of 1959鈥�1961, and recent seasonal famines in Bangladesh, which reported that the famine related to a variety of adult diseases, including metabolic and cardiovascular conditions, reproductive health, psychological disorders, and many others [31,32,33]. Findings from the cross-sectional study of CHARLS 2011 also found that females exposed to famine in China during infancy were more likely to report poor self-rated health in their adulthood [34]. Therefore, exposure to famine may have harmful effects on the multimorbidity risk. Arshadipour et al. using two large cross-sectional KORA (Cooperative Health Research in the Region of Augsburg) -Age studies in 2008/9 and 2016 found that individuals born during the late war phase had the highest prevalence of multimorbidity and single chronic diseases compared to participants born during the other phases [25]. In consistent with previous reported studies, our results indicated exposed to Chinese famine in early life stage was associated with multimorbidity prevalence and number of morbidities in cross-sectional study. Moreover, the current study conducted longitudinal analyses by using five-wave data from the CHARLS, and the results also indicated that early famine exposure was associated with multimorbidity incidence and number of morbidities in longitudinal prospective cohort study. As a result, our study corroborates prior findings and provides new evidence on the possible effect of early famine exposure on the later progression of multimorbidity.

The second important finding of the present study is that exposure to famine was positively related to multimorbidity prevalence, and exposure to famine in early postnatal life was associated with increased risk of multimorbidity incident in adult. These findings suggested that both utero and early postnatal life could be a critical time window for multimorbidity. A study from a cohort of 2414 people, aged 50 years, born as term singletons around the time of the 1944鈥�1945 Dutch famine indicated that exposure to famine in early gestation was associated with higher risk of coronary heart disease, raised lipids, altered clotting and more obesity, and exposure in mid gestation was associated with obstructive airways disease and microalbuminuria [35]. While, exposed to famine in late gestation was associated with decreased glucose tolerance in adult people [35]. Furthermore, Chen et al. found that individuals exposed to famine before age 18 had a higher risk of noninfectious chronic diseases in later life, and the risk was particularly high for those exposed to famine in-utero and in the 鈥渇irst 1,000 days鈥� of life, namely, between 0 and 2 years (particularly 0鈥�6 months) than individuals unexposed to famine over their life course [33]. All of those findings suggested that exposure to famine was associated with a higher risk of non-communicable diseases and multimorbidity.

In consistent with some previous studies that reported a more pronounced detrimental impact of famine exposure among females than males [36, 37], we found that the association between famine exposure and multimorbidity incident in later life was significant in females other than males. Several biological and environmental interactions may partly explain such gender differences such as the healthy survivor effect among males and the food allocation priority given to sons in the family [38]. The association between famine exposure in early life and multimorbidity in adulthood appeared to be different in rural and urban areas. We only found significant association in rural subgroups. It has been documented that during the Great Chinese Famine, rural areas suffered from more severe food shortages than did urban areas due to the central purchase and supply system [39]. Future studies are warranted to investigate the possible sex- and area-specific mechanisms underlying the famine exposure and later-life multimorbidity.

Whether famine exposure in utero and early postnatal life, malnutrition would result in caloric restriction, deficiency in macro and micronutrients, which may have effects on structure and function of the reproductive system, including pathological organ development and abnormal endocrine status [39, 40]. Exposure to famine increases the preference for high-fat foods and a high prevalence of dyslipidemia, which in turn increases the risk of noninfectious chronic diseases [18]. Furthermore, several studies have shown that there were was an interaction between hypertension, obesity, hyperglycemia and famine on increase the risk of noninfectious chronic diseases [16, 41]. Furthermore, both general and abdominal obesity could modify and mediate the famine-diseases association, which may increase the risk [42].In addition, exposure to famine may cause life-long changes in DNA demethylation and metabolic reprogramming, and subsequently increase disease risk [43,44,45]. Further studies are warranted to explore the exact mechanism.

The current study provided some evidence to prevent multimorbidity. It suggested that we should be sensitive to the nutritional status of individuals and provide them with adequate and optimal nutrition. And we also should popularize the nutrition and health knowledge. Some strengths of our study include, first, the current study was based on the data from the CHARLS study, which is a large nationally representative cohort study with a high response rate, and potential confounders were collected and controlled in the multivariable models. Second, the innovative grouping and stratifying methods make it clarified to distinguish the exact stage in which famine exerts influences on the multimorbidity. Several potential limitations of present study need to be mentioned. First, unlike the Dutch famine which lasted only 5鈥�6 months and had defined starting and ending points, the long and imprecise duration of the Chinese Famine in 1959鈥�1961 increased difficulties in grouping. Second, the retrospective measures of famine exposure and severity of famine exposure were self-reported and might be subject to recall bias and misclassification. Third, the information about birth weight or gestational age were not collected in CHARLS, which might be important confounders. Fourth, survivor bias might be part of the story. Finally, the results were only conducted from the CHARLS, data from a large-sample study like China Kadoorie Biobank may have higher representativeness. Finally, although we had adjusted the covariates in the model, the socioeconomic status, lifestyle, and genetic predispositions may also influence the prevalence and incidence of multimorbidity. Future studies are needed to evaluate the effects of survivor bias by investigating the possible selection mechanisms and adjust the effect estimates of famine exposure.

Conclusion

Based on CHARLS, a large-scale, nationally representative, longitudinal study, which provides both retrospective and prospective data, our findings indicted famine exposure has long-term detrimental consequences for later-life multimorbidity and non-communicable diseases. Our findings provided the large body of evidence underlining the importance of adequate nutrition of pregnant women and children for the health of them in the adult.