Fetal and infant exposure to severe Chinese famine increases the risk of adult dyslipidemia: Results from the China health and retirement longitudinal study

The CHARLS is a longitudinal large-scale national survey of the middle-aged and elderly Chinese population and was followed up every two years. The baseline data collection of the survey was performed from June 2011 to March 2012. The protocol was the same as before [24]. Briefly, 17,708 participants aged more than 45 years were selected from 10,257 households through four-staged, stratified, cluster sampling among 28 provinces, municipalities, and autonomous regions in Mainland China. In the current study, 2883 participants were enrolled into four cohorts as defined by birthdates. After excluding 131 participants with missing values in blood lipid profile, 2752 subjects participated in the current study (Fig. 1).

We categorized all the subjects by birthdate into four birth cohorts based on the life stage when the famine happened: non-exposed cohort, fetal stage-exposed, infant stage-exposed, and preschool stage-exposed cohorts. Because the Chinese famine lasted for three years from 1959 to 1961, but the exact dates of famine starting and ending were not clear. In order to minimize the misclassification bias, participants born between January 1st, 1959 and September 30th, 1959, and between October 1st, 1961 and September 30th, 1962 were removed. The participants’ birthdate ranges of four cohorts were the same with our previous paper [23] and were listed briefly below: the non-exposed cohort (10/01/1962–09/30/1964), the fetal stage-exposed cohort (10/01/1959–09/30/1961), the infant stage-exposed cohort (01/01/1958–12/31/1958), the preschool stage-exposed cohort (01/01/1956–12/31/1957). The participant’s mean age (standard deviation) of the four cohorts was 47.4 (0.5), 50.9 (0.7), 53.2 (0.4), and 55.1 (0.4) years, respectively.

As described [23], although the Chinese famine affected all the provinces in China, the famine severity was different in different provinces due to variation of climate, population density, and policies regarding shortage of foods [25]. Referring to the method in previous studies [23, 25], the current study also used excess mortality to partly reflect the severity of the famine exposure. The computing method of excess death rate was described in our previous study [23]. An excess mortality of 50% was used to differentiate the severe affected areas from less affected areas in the present study.

Frozen plasma total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglycerides (TG) from venous blood were examined by the standard enzymatic colorimetric method [26‐28]. The classifications from the Guidelines for The China Adult Dyslipidemia Prevention, Evaluation and Treatment [29] were used to define elevated TC (TC ≥6.216 mmol/L; 240 mg/dL), low HDL-c (HDL <1.036 mmol/L; 40 mg/dL), elevated LDL-c (LDL-c ≥ 4.138 mmol/L; 160 mg/dL), and elevated TG (TG <2.250 mmol/L; 200 mg/dL). We defined dyslipidemia as TC/HDL-c > 5.0 [30] or self-report dyslipidemia to avoid the influence of lipid-lowering drugs. The participants were recognized as suffering from self-reported dyslipidemia if they answered ‘yes’ to the question below: Have you been diagnosed with the dyslipidemia (elevated LDL-c, TG, and TC, or a low HDL-c) by a registered physician?

The mean family income (12,000 Chinese Yuan per person per year) of the current samples was selected as a threshold to divide all participants into two groups (low economic status group and high economic status group) [31]. The body mass index (BMI) was calculated using weight in kilometer divided by square of height in meter, and 24.0 kg/m² of BMI was used as a threshold (recommended for Chinese adults) to divide all participants into normal weight and overweight/obese groups. Actual age of the participants was calculated when detect time subtracted the birthdates.

Data were collected by computer-aided personal survey (CAPI) system. SPSS version 20.0 (SPSS Inc. Chicago, IL) was used for the statistical analysis.

ANOVA and ANCOVA with TC, TG, HDL, and LDL, respectively as dependent variables and age as a covariant were conducted to test the difference of TC, TG, HDL, and LDL among the four birth cohorts. The Dunnett-t test was used to perform multiple comparisons between the three exposed cohorts and the non-exposed cohort, respectively. Logarithmic transformation was used to transfer the asymmetrical distribution data to a normal distribution or near-normal distribution.

The method of maximum likelihood by the binary logistics regression model, control for gender, and current family economic status, was used to examine the risks of dyslipidemia among fetal, infant, and preschool stage-exposed cohorts, where the non-exposed cohort was the reference group. The interaction between the famine exposure cohorts and famine effect areas, BMI, and gender was examined by providing a multiplicative factor in the logistics regression model. The analyses were controlled for sex and the current family economic status.

To explore whether the associations of early-life the Chinese famine exposure with the dyslipidemia were sex-specific, the stratified analysis by sex was performed in adulthood.

This study has several limitations. Firstly, the selection bias caused by excess mortality in early life could be a main limitation. Subjects with abnormal metabolism and structure may succumb to the famine, leaving only the stronger and healthier participants. The bias may lower the real effect of famine exposure on dyslipidemia, and thus it could not overestimate the link of the famine exposure with dyslipidemia. Secondly, we cannot completely divide the fetal exposed cohort from infant exposed cohort because the Chinese famine lasted for three years (1959–1961). A lack of an adequate method accurately distinguishes whether they were fetal or infant exposed to the Chinese famine. However, in this study, subjects who born from January 1, 1958 to December 31, 1958 was defined as the infant-exposed cohort to ensure that the vast majority of the subjects in this cohort were exposed to Chinese famine in an infant stage. Furthermore, we also found that both the fetal and infant stage famine exposure increased the risk of dyslipidemia in later life. Thirdly, the excess death rate was used to evaluate the severity of suffering from the Chinese famine in this study. With this indicator, the confounding factors, such as adverse climate conditions or infections that may also contribute to the deaths are not taken into consideration. Also, the information of personal energy intake during famine period was not available in the current study. Thus, we cannot accurately attribute the effect of famine exposure in early life on elevated risks of dyslipidemia. Additional, similar with the other studies focused on the Chinese famine [25], the current study also lacked the objective indexes, such as birth weight and length to reflect the effect of famine exposure on physical health outcomes. Although the current study was open to residual confounding or selection bias, it used a national data with broad representativeness, combining self-report dyslipidemia with serum lipid profile examination to evaluate the association between early-life the Chinese famine exposure and dyslipidemia in later life and found that early life exposed to severe malnutrition could be associated with the higher risk of adult non-communicable diseases. Despite China has been food secure now; many people are still exposed to food shortages in low-income and middle-income countries [47, 48]. Hence many children are still at short term risk of malnutrition associated mortality and morbidity [47], but also at long term risk of adult non-communicable diseases. The heavily short and long term adverse effects from severe food shortage ask us to implement the urgent interventions to lower their occurrence or improve their consequences.