-
The CHNS is an ongoing longitudinal and household-based survey that has completed ten waves (1989–2015). The survey originally covered eight provinces (including Jiangsu, Hubei, Hunan, Guangxi, Guizhou, Liaoning, Shandong, and Henan). In 2011, the CHNS included additional provinces of Beijing, Shanghai, and Chongqing and in 2015 it included Shanxi, Yunnan, and Zhejiang. Although the CHNS is not considered as a national representative survey, the data gathered from these provinces provided significant statistical variability in terms of demographic distribution, geographic distribution, economic strata, and public resources which could be considered as a representative dataset for all the provinces in China. It was approved by the institutional review committees of the University of North Carolina at Chapel Hill (UNC-CH) and the National Institute for Nutrition and Health, Chinese Center for Disease Control and Prevention. The detailed survey protocol was described previously elsewhere[25].
Fasting blood samples were collected from individuals in 2009 and 2015. The present study used samples from 5,132 eligible participants (6,378 participated in both the two waves; then excluded those age over 75 in 2009) who aged 18–75 in 2009 and participated in these two waves. Pregnant women or lactating mothers were excluded. Participants who were previously diagnosed with diabetes, hypertension, cancer, and MetS in the 2009 wave were excluded because these diseases affected their dietary habits. Participants lacking complete data of dietary measures, blood pressure, waist circumference, fasting blood, and serum ferritin determinations in 2009 and 2015, and those with missing sociodemographic variables in 2009 were excluded as well. Finally, a total of 2,797 participants were included in the analysis.
-
Dietary assessment used foods usual intake which was estimated by 3 consecutive days 24-hour recalls and FFQ in 2009[26-28]. The three consecutive days 24-hour recalls mainly recorded the type, amount, type of meal, and place for all the food items consumed at an individual level and were entered into a household level inventory for the same three days. These 3 consecutive days were randomly allocated from Monday to Sunday and included at least one weekend. Changes in the household inventory as well as the wastage were used to calculate total household consumption. The percentage of the oil and condiments were determined from the household inventory for each member as the ratio of their individual energy intake and to the energy intake of all members combined. Dietary intake data was linked to the China Food Composition Table[29], and we calculated the average of total energy intake (TEI) and nutrients intake (dietary fiber, carbohydrate, and dietary fat) from all the food items consumed over the period of 3 days. The FFQ included 74 food items and 9 food categories (including rice, wheat, and wheat products, other staple cereals and tubers, legumes, vegetables, fruits, dairy products, meats including pork, beef, mutton, poultry, fish, and shrimp, eggs, snacks, alcohol, and soft drinks). For each of the food items listed above, participants reported the frequency of habitual consumption [daily, weekly, monthly, annually, or never (the reference category)] and the amount consumed each time for a period of 12 months.
For this study, we have defined total red meat as fresh red meat and processed red meat. Fresh red meat included muscle and organ meat from pork, beef, and mutton that had not been treated, whereas processed red meat included all the products of all types of red meat that were undergo treated, such as sausages, salami, ham, and luncheon meats[13]. Red meat is episodically consumed food in China, so we performed the National Cancer Institute (NCI) method to estimate the usual intake of foods. The NCI method was developed specifically to deal with episodically consumed or no daily consumed foods, which consists of a two-part model with correlated person-specific effects[30-32]. The first part consists of estimating the probability of consuming a specific food using logistic regression analysis with a person-specific random effect (mixed model). Part I was represented as:
$$ \begin{aligned} & \text{Logit} \left( {24\text{HR}\;\text{Probability}} \right) = {\text{Intercept}}_{\text{I}} + {\text{Slope}}_{\text{I}}\\ & \quad \times \text{Coveriate} + \text{Person}_- \text{Specific} \; \text{Effect}_{\text{I}} \end{aligned}$$ (1) where, for probability P, Logit (P) = log (P/1 − P). The intercept, slope, and variance was person-specific, and I (shown in subscript) indicated their association with Part I. Covariate represented the effect of personal characteristics, such as age, sex, or body mass index. The model allowed for multiple covariates[31]. The second part specified the consumption-day amount and may be represented as:
$$ \begin{aligned} & \text{Transformed} \left( \text{24HR Amount} \right) \!=\! {\text{Intercept}}_{\text{II}} \!+\! {\text{Slope}}_{\text{II}} \\ &\times {\text{Coveriate}} + {\text{Person}}_-{\text{Specific Effect}}_{\text{II}} \\ &+ {\text{Within}}_- {\text{person Variability}}_{\text{II}} \end{aligned}$$ (2) where II (shown in subscript) indicated that these parameters were associated with Part II, and differed from those in Part I. Two or more 24 HR on a number of individuals with reports of the food of interest were required to distinguish between- and within-person variation in the model of part II[31]. The model was specified on the transformed scale where the person-specific effect and within-person random variability were normally distributed. Moreover, additional information from FFQ was included in the model as a covariate[31, 33]. The details of the NCI method were described previously elsewhere [30-32].
-
Height, weight, and waist circumference were measured by health workers who were trained with standardized procedures. Height was measured to the nearest 0.1 cm using a Stadiometer (model 206, SECA) without wearing shoes. Weight was measured to the nearest 0.1 kg with SECA 880, with participants standing and wearing a single layer of clothing. We calculated the body mass index (BMI) using weight in kilograms divided by measured height in squared meters, and divided it into three categories underweight (BMI < 18.5 kg/m2), normal (18.5 kg/m2 ≤ BMI < 24.0 kg/m2) and overweight/obesity (BMI ≥ 24.0 kg/m2)[34]. Waist circumference was measured by a SECA tape measure. At least a 10-min rest period between the systolic blood pressure (SBP) and diastolic blood pressure (DBP) was set between the BP measurements. All the measurements were done on the participant’s left arm in a seated position and averaged to the nearest mmHg according to standard procedures. We used the mean of three satisfactory measurements for all the analyses. The requirements for the measurement process were described previously [35].
Blood samples were collected by health workers by venipuncture after an overnight fast. Glucose, TG, HDL-C, and ferritin of all the samples were measured according to standard procedures in a national lab in Beijing[36]. FPG was measured by the GOD-PAP (Randox Laboratories Ltd., London, UK). The concentration of serum HDL-C and TG were measured by the enzymatic method and CHOD-PAP (Kyowa Medex Co., Ltd, Tokyo, Japan), respectively. Serum ferritin levels were determined by radioimmunoassay method.
-
MetS was diagnosed as per the guidelines of the International Diabetes Federation (IDF) and AHA/NHLBI criteria with a modified waist circumference cutoff for Chinese adults[37]. A person diagnosed with MetS had to meet any three or more of the criteria listed: (1) Elevated WC (using the Chinese values: ≥ 90 cm in man, ≥ 80 cm in female); (2) raised TG (TG ≥ 150 mg/dL (1.7 mmol/L) or drug treatment for elevated triglycerides); (3) reduced HDL-C [< 40 mg/dL (1.0 mmol/L) in males, < 50 mg/dL (1.3 mmol/L) in females or drug treatment for reduced HDL-C]; (4) raised BP [systolic BP (SBP) ≥ 130 mmHg or diastolic BP (DBP) ≥ 85 mmHg or antihypertensive drug treatment in a patient with a history of hypertension]; (5) raised FPG (≥ 100 mg/dL or drug treatment of elevated glucose).
-
Standard questionnaires were used to collect the sociodemographic characteristics data such as age, gender, educational level, lifestyle factors (e.g. smoking, drinking alcohol, physical activity), household income, and community information. Participants were grouped into three categories based on their age (18–44 years, 45–59 years, and 60–75 years) and classified smoking status as current (Yes) and ever/never (No). Alcoholic status was categorized as drinking (Yes) and non-/ever drinking (No). The household incomes were calculated according to the household size and grouped as low, middle, and high-income groups. The urbanization index was calculated based on 12 multidimensional components reflecting the heterogeneity in economic, social, demographic, and infrastructural characteristics at the community level[38], and categorized as low, middle, and high. Physical activity classification included four domains: occupational, household chores, leisure time activities, and commute related activities. Participants reported all activities as an average of all the hours spent in activities per week, and we converted the time spent in each activity into a metabolic equivalent of task (MET) hours per week based on the Compendium of Physical Activities. We grouped the total MET-hours per week as low, middle, and high groups. Based on the National Bureau of Statistics, we divided all provinces covered in the present study into east (include Jilin, Jiangsu, and Shandong), central (include Hubei, Heilongjiang, Hunan, and Henan) and west (include Guangxi and Guizhou).
-
Total red meat usual intake was grouped into quartiles for all the subsequent analyses. For the baseline characteristics of the participants, categorical variables were presented as percentages and used chi-square analysis to test the differences across quartiles of total red meat usual intake. All continuous variables were tested for normality and expressed with skewed distribution variables as medians (interquartile ranges) and were subjected to non-parametric statistical hypothesis tests of the Kruskal-Wallis test. A series of multivariable logistic regression models were constructed to assess the association between the risk of the usual intake of total red meat and specific red meats and MetS, adjusted for potential confounding factors including demographic, socioeconomic, lifestyle factors, and dietary factors in the models. Because the small amount consumed, processed red meat were consumed for the first three quartiles, total red meat was used in the multivariable logistic regression. Linear trends were tested by assigning median values to quartiles of consumption of total red meat or specific red meats and this variable was modeled as a continuous term. In addition, we estimated the association of the total red meat or specific red meats usual intake and serum ferritin levels by quantile regression model by adjusting the confounding factors. All the statistical analyses were conducted using the SAS 9.4 (SAS Institute, Inc., Cary, NC, USA). P < 0.05 was considered significant.
-
The baseline characteristics of all the participants across the quartile of total red meat usual intake are listed in Table 1. Baseline characteristics data revealed that participants who had a higher total red meat usual intake were overweight young men who smoked and had higher urbanicity index, higher household income, lower physical activities, high alcohol consumption rates. Furthermore, the top quartile of total red meat usual intake appeared to be those with relatively higher TG and FBG levels, and lower HDL-C and blood pressure. Participants showed significant differences in the dietary intake among the quartile of total red meat usual intake. To summarize, participants with higher total energy, dietary fat intake, and lower dietary fiber intake had higher total red meat intake. It was interesting to observe that adults with the lowest and highest quartile of total red meat usual intake had higher vegetables and fruits intakes.
Factors Q1 Q2 Q3 Q4 P-value Participants 699 699 700 699 Total red meat (g/d) 15.26
(11.44, 20.07)40.46
(33.76, 45.75)66.99
(59.89, 73.78)104.18
(91.17, 123.41)Age (years), n (%) < 0.001 18−44 204 (29.18) 228 (32.62) 265 (37.86) 276 (39.48) 45−59 193 (27.61) 207 (29.61) 204 (29.14) 210 (30.04) 60−75 302 (43.2) 264 (37.77) 231 (33.00) 213 (30.47) Gender, n (%) < 0.001 Men 297 (42.49) 308 (44.06) 337 (48.14) < 0.001 Women 402 (57.51) 391 (55.94) 363 (51.86) < 0.001 Regions, n (%) < 0.001 East 231 (33.05) 285 (40.77) 230 (32.86) < 0.001 Central 391 (55.94) 273 (39.06) 271 (38.71) < 0.001 West 77 (11.02) 141 (20.17) 199 (28.43) < 0.001 Urbanization index, n (%) < 0.001 Low 381 (54.51) 234 (33.48) 187 (26.71) < 0.001 Middle 218 (31.19) 254 (36.34) 227 (32.43) < 0.001 High 100 (14.31) 211 (30.19) 286 (40.86) < 0.001 Household income, n (%) < 0.001 Low 304 (43.80) 226 (32.47) 193 (27.81) < 0.001 Middle 225 (32.42) 222 (31.90) 243 (35.01) < 0.001 High 165 (23.78) 248 (35.63) 258 (37.18) < 0.001 Physical activity, n (%) < 0.001 Low 244 (34.91) 205 (29.33) 231 (33.00) 252 (36.05) Middle 174 (24.89) 233 (33.33) 266 (38.00) 260 (37.20) High 281 (40.20) 261 (37.34) 203 (29.00) 187 (26.75) Smoking, n (%) 0.316 No 479 (68.53) 486 (69.53) 471 (67.29) 462 (66.09) Yes 220 (31.47) 212 (30.33) 229 (32.71) 237 (33.91) Alcohol, n (%) < 0.001 No 494 (70.67) 469 (67.10) 445 (63.57) 427 (61.09) Yes 205 (29.33) 229 (32.76) 255 (36.43) 272 (38.91) BMI, n (%) 0.305 Underweight 47 (6.72) 56 (8.01) 63 (9.00) 50 (7.15) Normal 427 (61.09) 413 (59.08) 431 (61.57) 434 (62.09) Overweight/obesity 225 (32.19) 230 (32.90) 206 (29.43) 215 (30.76) WC (cm) 80 (75, 87) 80 (74, 87) 79 (73, 86) 79 (73, 86) 0.006 SBP (mmHg) 120 (111, 131) 120 (110, 130) 120 (110, 129) 119 (109, 125) < 0.001 DBP (mmHg) 80 (71, 87) 79 (71, 84) 79 (71, 83) 77 (70, 81) < 0.001 HDL-C (mg/dL) 57 (50, 66) 58 (50, 68) 57 (50, 67) 56 (49, 65) 0.004 TG (mg/dL) 92 (66, 124) 92 (66, 124) 93 (65, 125) 97 (70, 132) 0.024 FBG (mg/dL) 89 (83, 96) 89 (83, 96) 90 (84, 96) 90 (84, 98) < 0.001 Serum ferritin (ng/mL) 65.46
(31.57, 119.56)66.08
(30.14, 121.66)70.54
(32.18, 131.96)84.88
(42.34, 155.56)< 0.001 TEI (kcal/day) 2047.89
(1633.48, 2491.71)2105.54
(1748.4, 2539.48)2157.35
(1821.03, 2680.92)2417.6
(2034.1, 2862.35)< 0.001 Dietary fiber (g/d) 11.33
(8.48, 15.12)10.54
(8.09, 15.11)10.27
(7.51, 14.18)9.79
(7.28, 13.53)< 0.001 Carbohydrate (g/d) 310.5
(241.2, 396.7)284.91
(226.86, 361.85)281.52
(226.69, 358.41)278.44
(222.87, 351.41)< 0.001 Dietary fat (g/d) 54.68
(37.28, 74.39)68.4
(50.6, 89.84)78.88
(59.87, 102.76)99.62
(75.21, 122.05)< 0.001 Vegetables (g/d) 296.71
(212.55, 392.42)299.36
(215.55, 390.75)303.85
(231.07, 386.47)332.79
(266.25, 400.73)< 0.001 Fruit (g/d) 0.77
(0.37, 4.07)1.19
(0.52, 53.94)1.57
(0.53, 56.03)1.89
(0.67, 63.95)< 0.001 Fresh red meat (g/d) 15.11
(11.38, 19.87)39.63
(32.94, 45.1)65.52
(58.68, 72.32)102.84
(89.7, 121.62)< 0.001 Processed red meat (g/d) 0.00 (0.00, 0.51) 0.00 (0.00, 0.73) 0.00 (0.00, 0.85) 0.38 (0.00, 0.94) < 0.001 Note. Q = quartile. Data of categorical variables are expressed as a percentage (%); Medians (interquartile ranges) were calculated for skewed parameters. Table 1. Baseline characteristics of all the participants according to the quartile of red meat usual intake
-
The risk of metabolic syndrome was analyzed by red meat and specific red meat usual intake, which was divided into quartiles as listed in Table 2. The risk of metabolic syndrome increased linearly with an increase in red meat usual intake, after adjusting for all the potential confounders (age, gender, regions, urbanicity index, household income level, physical activity, smoking, alcohol, BMI, TEI, dietary fiber, carbohydrate, dietary fat, of vegetables and fruits usual intake). The risk of metabolic syndrome was highest in the top quartile of red meat usual intake (RR: 1.41; 95% CI: 1.05–1.90). In addition, for specific red meat, the risk of metabolic syndrome in relation to the amount of intake showed a significant difference only in fresh red meat usual intake. The top quartile of fresh red meat usual intake (RR: 1.37; 95% CI: 1.02–1.85) exhibited the highest risk of metabolic syndrome. As the intake increased, the risk of metabolic syndrome showed an increasing trend (P < 0.05). However, there was no association between the risk of metabolic syndrome and the processed red meat usual intake.
Groups Q1 Q2 Q3 Q4 P-trend Total red meat Crude Ref 0.99 (0.78, 1.26) 1.02 (0.80, 1.29) 1.16 (0.91, 1.47) 0.197 Model 1 Ref 1.01 (0.80, 1.29) 1.08 (0.84, 1.38) 1.31 (1.02, 1.67)* 0.025 Model 2 Ref 1.00 (0.78, 1.30) 1.11 (0.86, 1.45) 1.37 (1.05, 1.80)* 0.013 Model 3 Ref 1.03 (0.79, 1.34) 1.14 (0.87, 1.49) 1.41 (1.05, 1.90)* 0.016 Fresh red meat Crude Ref 1.00 (0.79, 1.27) 1.12 (0.89, 1.43) 1.14 (0.90, 1.46) 0.184 Model 1 Ref 1.02 (0.80, 1.30) 1.19 (0.93, 1.51) 1.30 (1.01, 1.66)* 0.021 Model 2 Ref 1.01 (0.78, 1.31) 1.22 (0.94, 1.58) 1.35 (1.03, 1.77)* 0.014 Model 3# Ref 1.03 (0.79, 1.34) 1.24 (0.95, 1.63) 1.37 (1.02, 1.85)* 0.019 Processed red meat Crude Ref 1.27 (1.02, 1.59)* 1.22 (1.00, 1.50)* 0.024 Model 1 Ref 1.26 (1.01, 1.57)* 1.21 (0.98, 1.49) 0.037 Model 2 Ref 1.15 (0.91, 1.45) 1.13 (0.90, 1.41) 0.212 Model 3# Ref 1.14 (0.90, 1.45) 1.13 (0.90, 1.42) 0.164 Note. Q = quartile; Ref = reference group. Data are expressed as RR (95% CI). Crude: adjusted age and gender; Model 1: based on crude, adjusted regions and household income level; Model 2: based on model 1, adjusted baseline of body mass index, urbanicity index, smoking, drinking alcohol, physical activity, and TEI; Model 3: based on model 2, adjusted dietary fiber, fat, carbohydrate, usual intake of vegetables and fruits. #Model 3: based on model 2, adjusted dietary fiber, fat, carbohydrate, usual intake of vegetable, fruit and other types of red meat; *P < 0.05. Table 2. The association between total red meat or its subtype usual intakes and the risk of MetS
-
The association between metabolic syndrome risk factors and serum ferritin levels is shown in Figure 1. A number of risk factors responsible for metabolic syndrome and serum ferritin levels were positively correlated. The mean value of serum ferritin increased significantly with the number of risk factors related to metabolic syndrome (P < 0.05). In brief, a higher number of metabolic syndrome components were associated with greater serum ferritin concentrations.
-
Table 3 shows the serum ferritin quantile regression analysis among adults. The red meat usual intake had a significant positive association with serum ferritin levels across the entire conditional serum ferritin distribution (P < 0.05). The lowest serum ferritin association was seen at the 10th quantile and the highest at the 90th quantile (0.09, 0.20, 0.39, 0.61, and 0.89 at the 10th, 25th, 50th, 75th, and 90th quantiles, respectively). In addition, the fresh red meat usual intake was also positively associated with serum ferritin and the coefficients were statistically significant across the entire conditional distribution (0.22, 0.42, 0.63, and 0.88 at the 25th, 50th, 75th, and 90th quantiles, respectively). However, there was no association between processed red meat and serum ferritin levels.
Groups Quantiles 10th 25th 50th 75th 90th Total red meat 0.09 (0.00, 0.17)* 0.20 (0.10, 0.31)* 0.39 (0.22, 0.55)* 0.61 (0.40, 0.82)* 0.89 (0.55, 1.23)* Fresh red meat1 0.08 (−0.01, 0.17) 0.22 (0.11, 0.33)* 0.42 (0.26, 0.57)* 0.63 (0.42, 0.85)* 0.88 (0.56, 1.20)* Processed red meat1 −0.15 (−0.86, 0.56) −0.23 (−0.94, 0.47) −0.66 (−1.61, 0.30) 0.23 (−0.89, 1.35) 0.14 (−1.76, 2.04) Note. Data are expressed as a coefficient (95% CI) and adjusted for gender, age, regions, urbanization index, household income, physical activity, smoking, drinking alcohol, baseline BMI, baseline serum ferritin, TEI, dietary fiber, carbohydrate, dietary fat, baseline of vegetables and fruits usual intake. 1: Adjusted for gender, age, regions, urbanicity index, household income, physical activity, smoking, alcohol drinking, baseline BMI, serum ferritin baseline, TEI, dietary fiber, carbohydrate, dietary fat, usual intake of vegetable, fruit and other types of red meat; *P < 0.01. Table 3. Estimation of coefficients from quantile regression analysis of serum ferritin by total red meatand its subtype usual intake
Association of Red Meat Usual Intake with Serum Ferritin and the Risk of Metabolic Syndrome in Chinese Adults: A Longitudinal Study from the China Health and Nutrition Survey
doi: 10.3967/bes2020.003
- Received Date: 2019-06-21
- Accepted Date: 2019-09-16
-
Key words:
- Usual intake /
- Red meat /
- Metabolic syndrome /
- Serum ferritin
Abstract:
Citation: | HUANG Li Na, WANG Hui Jun, WANG Zhi Hong, ZHANG Ji Guo, JIA Xiao Fang, ZHANG Bing, DING Gang Qiang. Association of Red Meat Usual Intake with Serum Ferritin and the Risk of Metabolic Syndrome in Chinese Adults: A Longitudinal Study from the China Health and Nutrition Survey[J]. Biomedical and Environmental Sciences, 2020, 33(1): 19-29. doi: 10.3967/bes2020.003 |