In 2019, the total number of deaths from CVD in China was 5·09 million, and the age-standardised death rate reached 2·760 per 100 000(Reference Liu, Qi and Yin1). CVD remains the top cause of death in China, which poses a strong economic burden(Reference Liu, Li and Zeng2). Dyslipidaemia, hypertension and diabetes have long been recognised as risk factors for CVD(Reference Rader3–Reference Mangiapane5). Diet, as one of the ways to improve CVD risk factors, has been the focus of researchers(Reference Badimon, Chagas and Chiva-Blanch6). In recent years, a growing number of studies have found that limiting carbohydrate intake is associated with lower body weight(Reference Churuangsuk, Kherouf and Combet7), improved diabetes(Reference Alzahrani, Skytte and Samkani8–Reference Samkani, Skytte and Kandel10) and reduced risk of CVD(Reference Antonio de Luis, Aller and Izaola11–Reference Dong, Guo and Zhang13). Therefore, some experts believed that low carbohydrate diet (LCD) could be incorporated into dietary guidelines as a healthy way of diets in the future(Reference Volek, Phinney and Krauss14,Reference Hu and Bazzano15) .
It is worth noting that the role of the LCD in the treatment of CVD is currently controversial(Reference Kirkpatrick, Bolick and Kris-Etherton16–Reference Diamond, O’Neill and Volek19). Several studies have suggested that the consumption of unrestricting saturated fat may increase LDL-cholesterol levels, which could increase the risk of cardiovascular mortality(Reference Kirkpatrick, Bolick and Kris-Etherton16–Reference Diamond, O’Neill and Volek19). However, a review pointed out that the current review focuses only on LDL-cholesterol, a poor indicator of CVD risk, rather than a more reliable CVD risk factor (i.e. diabetes, hypertension, TAG, HDL)(Reference Diamond, O’Neill and Volek19).
The LCD score was designed by Halton et al. to assess dietary carbohydrate intake(Reference Halton, Willett and Liu20). Halton et al. also created animal-based and plant-based LCD scores. Few studies have examined the relationship between the LCD score and cardiometabolic risk factors in adults, including blood lipid(Reference Mirmiran, Asghari and Farhadnejad21–Reference Sangsefidi, Lorzadeh and Nadjarzadeh28), fasting blood glucose (FBG)(Reference Mirmiran, Asghari and Farhadnejad21–Reference Jafari-Maram, Daneshzad and Brett25), blood pressure(Reference Mirmiran, Asghari and Farhadnejad21,Reference Ha, Joung and Song22,Reference Shirani, Esmaillzadeh and Keshteli24,Reference Jafari-Maram, Daneshzad and Brett25,Reference Sangsefidi, Lorzadeh and Nadjarzadeh28) and uric acid(Reference Nakamura, Ueshima and Okuda23). There are several problems with these studies. First, the positive association between the LCD score and HDL-cholesterol has been demonstrated in most studies(Reference Ha, Joung and Song22,Reference Nakamura, Ueshima and Okuda23,Reference Kim, Lim and Shin26,Reference Sangsefidi, Lorzadeh and Nadjarzadeh28) . However, inconsistent results have been found between other cardiovascular risk factors and the LCD score. Only a few studies have found the negative relationship between the LCD score and FBG(Reference Mirmiran, Asghari and Farhadnejad21,Reference Shirani, Esmaillzadeh and Keshteli24) or blood pressure(Reference Mirmiran, Asghari and Farhadnejad21). Instead, the two other studies found that the LCD score could increase the risk of hypercholesterolaemia(Reference Kim, Lim and Shin26,Reference Tan, Kim and Shin27) or hypertriacylglycerolaemia(Reference Tan, Kim and Shin27). Inconsistent results may be due to differences in study participants. However, only one study has been conducted in the Chinese population(Reference Tan, Kim and Shin27). Therefore, more studies are needed to explore the effects of the LCD on cardiovascular risk factors in the Chinese population. Second, sex and dietary source differences may exist in the association of the LCD score with CVD risk factors(Reference Ha, Joung and Song22,Reference Nakamura, Ueshima and Okuda23,Reference Kim, Lim and Shin26–Reference Sangsefidi, Lorzadeh and Nadjarzadeh28) . Two studies found that animal-based and plant-based LCD scores were significantly positively related to HDL-cholesterol in Japanese populations(Reference Nakamura, Ueshima and Okuda23) or Korean females(Reference Ha, Joung and Song22). However, other studies showed more adherence to LCD(Reference Sangsefidi, Lorzadeh and Nadjarzadeh28) or animal-based LCD(Reference Ha, Joung and Song22) might be associated with the lower risk of low HDL-cholesterol, especially in males. The relationship between animal-based LCD score and hypertriacylglycerolaemia was opposite in Korean(Reference Kim, Lim and Shin26) and Chinese males(Reference Tan, Kim and Shin27). Based on the inconsistency of the above findings, more studies were needed to investigate the effect of sex and different dietary source on the relationship between LCD and cardiovascular risk factors. Finally, no studies have comprehensively explored the relationship between LCD and blood lipid, blood glucose, blood pressure and uric acid.
At the same time, carbohydrate is the main source of energy for Chinese; it is of great public health significance to explore the effects of carbohydrate diet on cardiovascular health of Chinese. The primary aim of the present study was to assess the associations between three LCD scores and cardiometabolic risk factors (blood lipid, FBG, blood pressure and uric acid) among Chinese adults. Second, the association between three LCD scores and cardiometabolic risk factors was assessed after stratification for sex.
Methods
Study population
Detailed methods of the Community-based Cohort Study on Nervous System Diseases have been described in another study(Reference Huang, Jia and Zhang29). In this study, Hebei, Zhejiang, Shanxi and Hunan provinces were selected as survey sites. Each province randomly selected two cities and two counties. Urban and suburban communities, as well as towns and villages within the county, were randomly selected. This study was reviewed and approved by the Institutional Review Board of the National Institute of Nutrition and Health (No. 2017020, 6 November 2017). Each participant signed an informed consent form before the study.
Participants in this survey were selected only from Hebei province. The baseline survey was conducted in 2018 and 6720 people participated in the survey. There were 5920 people aged 18 and older. For this study, participants with incomplete data on blood sample data (n 906) and incomplete answers to dietary data (n 215) were excluded. Additionally, 190 participants with energy intake < 2092 kJ/d (500 kcal/d) or ≥ 20920 kJ/d (5000 kcal/d) were excluded. Finally, the remaining 4609 participants were included in the study.
Dietary assessment and calculation of the low carbohydrate diet score
Participants’ diets were assessed using a validated semi-quantitative FFQ including eighty-one food items. Participants were asked if they had eaten the food item in the previous 12 months, and if so, how often (choosing one of each day, week, month or year) and how much they ate each time. If they had not, zero was recorded. The total amount of one food item equal to the frequency of food intake multiplied by each intake. Finally, the total amount of food item consumed was translated into daily grams. The remaining sixty-five foods were analysed after excluding sixteen nutritional supplements. The validity and reproducibility of the questionnaire were documented in other studies(Reference Zhang and Ho30,Reference Villegas, Yang and Liu31) . The correlations of nutrient intake between the FFQ and the second FFQ were 0·46–0·71(Reference Zhang and Ho30) and 0·38–0·52(Reference Villegas, Yang and Liu31), respectively. And the correlations of nutrient intake between the FFQ and the 24-h dietary recalls were 0·25–0·65 and 0·33–0·64(Reference Villegas, Yang and Liu31), respectively. In particular, the correlations of carbohydrate intake between the FFQ and the 24-h dietary recalls in Shanghai men were as high as 0·64(Reference Villegas, Yang and Liu31). Food nutrient composition based on China Food Composition Book (2009 edition) compiled by National Institute of Nutrition and Food Safety, China CDC. Condiment intake was collected by asking participants how much each condiment was consumed in their households for a month. There are seven condiments including vegetable oil, lard oil, salt, soya sauce, monosodium glutamate, fermented soya paste and sugar. Every condiment intake was computed by the total amount of each condiment by number of family members.
The calculation method of LCD was proposed by Halton et al.(Reference Halton, Willett and Liu20). Fat, protein and carbohydrate intakes, expressed as percentage of energy, were divided into eleven strata. For protein and fat, the higher the stratum, the higher the score (0–10). For carbohydrates, the opposite is true (10–0). The final three macronutrient scores were added together for a total score of 0–30, with higher scores representing participants intaking more protein and fat and less carbohydrates. In the study, two other LCD scores (animal-based LCD score and plant-based LCD score) were also calculated. Animal-based LCD score was calculated from the percentage of energy of carbohydrate, animal fat and animal protein. Plant-based LCD score was calculated from the percentage of energy of carbohydrate, plant fat and plant protein, please refer to Table 1 for details.
Table 1. Energy percentage of macronutrients used in calculating the low carbohydrate diet (LCD) scores, animal-based LCD scores and plant-based LCD scores of Chinese adults, Community-based Cohort Study on Nervous System Diseasesa
a Energy from diet carbohydrate, total protein, total fat, animal protein, animal fat, plant protein and plant fat is shown according the score assigned to the eleven groups after ranking the participants’ macronutrient intake, respectively.
Biochemical measurements
The participants’ blood was collected without breakfast. Blood samples were immediately sent to the First People’s Hospital of Hebei Province for testing. HDL-cholesterol, LDL-cholesterol, total cholesterol (TC), TAG, FBG and uric acid were measured on an AU5800 instrument (Beckman Coulter, Inc.). According to the Chinese guideline for the management of dyslipidaemia(32), high LDL-cholesterol was defined as LDL-cholesterol ≥ 160 mg/dl; low HDL-cholesterol was defined as HDL-cholesterol < 40 mg/dl; hypercholesterolaemia was defined as TC ≥ 240 mg/dl and hypertriacylglycerolaemia was defined as TAG ≥ 200 mg/dl. Impaired fasting glucose (IFG) was defined as FBG ≥ 5·6 mmol/l according to the American Diabetes Association criteria(33). According to the guidelines(Reference Zhang, Doherty and Bardin34), hyperuricaemia was defined as uric acid ≥ 7 mg/dl in men and ≥ 6 mg/dl in women.
Blood pressure was measured three times using an automated electronic sphygmomanometer (HBP-1300; Omron Corporation). The average of the three measurements was used as the final analysis. Of the 547 participants, only one blood pressure measurement was included in the analysis. High blood pressure was defined as systolic blood pressure ≥ 130 mm Hg and/or diastolic blood pressure ≥ 85 mm Hg(Reference Alberti, Eckel and Grundy35).
Assessment of other variables
Height was measured during the baseline period using a stable stadiometer (Seca) with a 0·1 cm precision. Weight was measured using an electronic scale. BMI was calculated by dividing body weight in kilograms by height in meters. Weight status was divided by four groups based on BMI: underweight (<18·5 kg/m2), normal weight (≥ 18·5 and <24 kg/m2), overweight (≥ 24 and <28 kg/m2) and obesity (≥ 28 kg/m2) according to the Guidelines for the Prevention and Control of Overweight and Obesity in Chinese Adults (2003).
Other covariates were obtained by questionnaire, including sex, age, the area of residence, monthly income per family, education (primary school or below, junior high, and senior high school and above), smoking (current/previous smoking and non-smoking), alcohol consumption (yes and no) in the past year (2017), physical activity, hypertension and diabetes. The area of residence was classified as rural or urban based on where they currently live. Monthly income per family was collected by asking each family about their per capita monthly income. Participants could choose from three levels (<1000, 1000–3999, ≥ 4000). Anyone who had consumed alcohol in the past year is considered ‘yes’. Physical activity was expressed in metabolic equivalent hours per day (Met-h/day). According to the Compendium of Physical Activities(Reference Ainsworth, Haskell and Whitt36), MET per hour for every sport were: 10·0 for martial arts, 7·0 for running or swimming, 4·5 for gymnastics, dancing or acrobatics, 3·5 for walking, 7·0 for playing football, basketball or tennis, 3·75 for playing badminton or volleyball and 4·0 for playing table tennis or tai chi. The history of the disease was determined by asking participants whether their doctor had given them a diagnosis of the disease.
Statistical analyses
The Mantel–Haenszel χ 2 statistical test for nominal variables and the ‘contrast’ option for linear regression analysis were used to assess whether there were significant differences in variables across quintiles of three LCD scores. Trend P values were obtained. All results for the continuous variables are presented as the mean values with their standard error, and the results for the categorical variables are presented as n (%). Multivariable logistic regression analyses were used to estimate OR with 95 % CI for the association between quintiles of three LCD scores and CVD indicators (including low HDL-cholesterol, high LDL-cholesterol, hypercholesterolaemia, hypertriacylglycerolaemia, IFG, high blood pressure and hyperuricaemia). Based on previous studies(Reference Jafari-Maram, Daneshzad and Brett25–Reference Tan, Kim and Shin27), model 1 adjusted for age, sex, area of residence, monthly income per family, weight status, smoking, alcohol, education level, physical activity, history of diabetes and hypertension. In addition, model 2 adjusted for model 1 covariates + salt, soya sauce, monosodium glutamate and sugar, as condiment intakes were associated with an increased risk of CVD(Reference He and MacGregor37–Reference Pan and Kong39). Tests for linear trend for OR were performed using the median value for each quintile as a continuous variable. All statistical analyses were performed using R software (Version 4.0.5). All P values were two tailed. P < 0·05 was considered significant.
Results
The general characteristics of the study population according to the LCD score quintiles are shown in Table 2. In all LCD scores, participants with a higher LCD score tended to live in urban, have a higher household income level, consume alcohol, history of hypertension and lower mean diastolic blood pressure than those with a lower score (all P < 0·05). In animal-based LCD score, participants with a higher LCD score tended to be female, have a higher physical level and a higher education level, have higher LDL-cholesterol and TC and have a lower systolic blood pressure than those with a lower score (all P < 0·05). Moreover, participants with a higher plant-based LCD score tended to be older, obese, smoking, history of diabetes and have higher systolic blood pressure and uric acid than those with a lower score (all P < 0·05).
Table 2. General characteristics of Chinese adults according to the quintiles of the low carbohydrate diet (LCD) scores, Community-based Cohort Study on Nervous System Diseases
(Numbers and percentages; mean values with their standard errors)a, b
a TC, total cholesterol; FBG, fasting blood glucose; SBP, systolic blood pressure; DBP, diastolic blood pressure; Q, quintile.
b The numbers of missing values were 7, 22, 32, 53, 7, 68 and 72 for weight status, smoke, alcohol consumption, smoking status, physical activity, education level, diabetes and hypertension, respectively. P values were calculated by the Mantel–Haenszel χ 2 statistical test for nominal variables and the ‘contrast’ option for linear regression analysis.
Macronutrient intake and seasoner intakes according to the three LCD scores are shown in Table 3. In the three LCD scores, participants with a higher LCD score had higher protein and fat intakes instead of lower carbohydrate intake compared with those with a lower score (all P < 0·0001).
Table 3. Macronutrient and condiment intake according to the low carbohydrate diet (LCD) scores, Community-based Cohort Study on Nervous System Diseases
(Mean values with their standard errors) a
a P values were calculated by the ‘contrast’ option for linear regression analysis.
In animal-based LCD score, participants with a higher LCD score had higher lard oil, monosodium glutamate and fermented soya paste intake, and lower vegetable oil and salt than those with a lower score (all P < 0·05). In plant-based LCD score, participants with a higher LCD score had higher fermented soya paste and sugar, and had lower vegetable oil intake than those with a lower score (all P < 0·0001). Refer to online Supplementary Table S1 for additional food intake and energy percentage according to the LCD score quintiles.
Carbohydrate and fat intake levels according to the quintiles of the three LCD scores are shown in Fig. 1. According to the acceptable macronutrient distribution range (AMDR) carbohydrate and fat recommendations, 12·8 % of participants intake was below the recommended levels of carbohydrates and 11·5 % of participants intake was above the recommended levels of fat. As quantiles of the three LCD scores increased, more participants had lower carbohydrate intake and higher fat intake (all P < 0·0001).
Fig. 1. Carbohydrate and fat intakes levels among Chinese adults according to the quintiles (Q) of the low-carbohydrate diet (LCD) scores. Values are presented as n (%). (a) Classification of the dietary carbohydrate level based on acceptable macronutrient distribution range (AMDR) Chinese Dietary Reference Intakes (CDRI) Handbook (2013). (b) Classification of the dietary fat level based on AMDR CDRI Handbook (2013). (a) 
, low (< 50%); 
, moderate; 
, high (> 65%). (b) , low (< 20%); , moderate; , high (> 30%).
Multivariate-adjusted OR for CVD indicators across quintiles of three LCD scores are presented in Table 4. Adjusted OR of low HDL-cholesterol for comparisons of Q5 with Q1 were 0·65 (95 % CI 0·50, 0·83) for the total LCD score (P for trend = 0·0001), 0·72 (95 % CI 0·56, 0·91) for animal-based LCD score (P for trend = 0·001) and 0·73 (95 % CI 0·57, 0·93) for plant-based LCD score (P for trend = 0·01). Adjusted OR of IFG for comparisons of Q5 with Q1 were 0·65 (95 % CI 0·51, 0·81) for the total LCD score (P for trend = 0·001) and 0·74 (95 % CI 0·59, 0·92) for plant-based LCD score (P for trend = 0·005). No significance was found between IFG and animal-based LCD score for comparisons of Q5 with Q1 (P for trend = 0·070). Three LCD scores were not related to high LDL-cholesterol, hypercholesterolaemia, hypertriacylglycerolaemia, high blood pressure and hyperuricaemia (P for trend > 0·05).
Table 4. Risk of CVD indicators according to the quintiles of the low carbohydrate diet (LCD) scores, Community-based Cohort Study on Nervous System Diseases
a IFG, impaired fasting glucose; Q, quintile.
b Values are presented as OR (95 % CI). Tests for linear trend for ORs were performed using the median value for each quintile as a continuous variable. P < 0.05 was considered significant. Model 1 adjusted for age, sex, area of residence, monthly income per family, weight status, smoking, alcohol, education level, physical activity, history of diabetes and hypertension. Model 2 adjusted for model 1 + salt, soya sauce, monosodium glutamate and sugar.
After stratification for sex, males in the highest quintile of the animal-based or plant-based LCD scores showed a decreased risk of low HDL-cholesterol (animal-based LCD score: OR: 0·60 (95 % CI 0·42, 0·87), P = 0·002; plant-based LCD score: OR: 0·58 (95 % CI 0·40, 0·83), P = 0·001), and females in the highest quintile of the animal-based or plant-based LCD scores showed a decreased risk of IFG than those in the lowest quintile of the LCD score (animal-based LCD score: OR: 0·69 (95 % CI 0·51, 0·94), P = 0·021; plant-based LCD score: OR: 0·71 (95 % CI 0·53, 0·96), P = 0·012) (Table 5). Refer to online Supplementary Table S2 for the risk of CVD indicators according to the quintiles of the total LCD score after stratification for sex.
Table 5. Risk of CVD indicators according to the quintiles of the low carbohydrate diet (LCD) scores after stratification for sex
a IFG, impaired fasting glucose; Q, quintile.
b Values are presented as OR (95 % CI). Tests for linear trend for ORs were performed using the median value for each quintile as a continuous variable. P < 0.05 was considered significant. Model adjusted for age, area of residence, monthly income per family, weight status, smoking, alcohol, education level, physical activity, history of diabetes and hypertension, salt, soya sauce, monosodium glutamate and sugar.
Discussion
This study found that Chinese adults who adhered to LCD obtained 47·6 % of energy from carbohydrate and 29·8 % of energy from fat. This was similar to the results of another study among Chinese adults(Reference Tan, Kim and Shin27). Tan et al. found that Chinese adults in the highest quartile of the LCD score obtained 53·3–53·8 % of energy from carbohydrate and 28·7–29·3 % of energy from fat(Reference Tan, Kim and Shin27). However, this is close to the energy intake from carbohydrate (54·7 %) and fat (28·3 %) in the lowest decile of the LCD score in the USA(Reference Halton, Liu and Manson40). American adults who adhered to LCD obtained 29·6 % of energy from carbohydrate and 46·1 % of energy from fat(Reference Halton, Liu and Manson40). Similarly, normal carbohydrate intake (≥ 45 %) is 52·6 % of total energy in the UK(Reference Shafique, Russell and Murdoch41). Current studies restrict carbohydrate energy to less than 45 % of total energy as the LCD(Reference Hu, Mills and Yao42,Reference Jovanovski, Zurbau and Vuksan43) . However, only a small part of the participants were able to meet this standard without intervention due to Chinese eating habits.
The study showed all LCD scores were positively associated with HDL-cholesterol after multivariable logistic regression analyses. Similar results have been found in other two studies(Reference Ha, Joung and Song22,Reference Nakamura, Ueshima and Okuda23) . Ha et al. found animal-based and plant-based LCD scores significantly decreased the risk of reduced HDL-cholesterol in females(Reference Ha, Joung and Song22). The INTERMAP study also found that all three LCD scores were significantly positively related to HDL-cholesterol (all P < 0·001) in a Japanese population(Reference Nakamura, Ueshima and Okuda23). The beneficial effects of LCD on HDL-cholesterol have been demonstrated in several systematic reviews(Reference Mansoor, Vinknes and Veierod44–Reference Huntriss, Campbell and Bedwell47). The benefit of LCD on HDL-cholesterol may be due to an increase in fatty acids intake(Reference Mensink, Zock and Kester48). And this study found that the intake of fatty acids increased HDL-cholesterol levels independent of the type of fatty acids. This speculation was also reflected in other studies(Reference Kim, Lim and Shin26,Reference Mensink, Zock and Kester48) .
Surprisingly, this study found that LCD increased HDL-cholesterol levels only in males. This finding was similar to another study in an Iranian population(Reference Sangsefidi, Lorzadeh and Nadjarzadeh28). A randomised controlled study also demonstrated that HDL-cholesterol levels increased significantly with carbohydrate restriction in men but not in women(Reference Can, Uysal and Palaoglu49). This result contradicts a current view. The view suggests that hormone-dependent differences between men and women cause women to have higher HDL-cholesterol levels than men(Reference Knopp, Paramsothy and Retzlaff50). In addition, high-fat diets also lead to higher levels of HDL-cholesterol in women than in men(Reference Knopp, Paramsothy and Retzlaff50). Therefore, LCD should be more able to elevate HDL-cholesterol levels in women than in men. However, this study found that it would be inappropriate to use only hormone-dependent differences to explain sex differences in the relationship between LCD and HDL-cholesterol levels. The specific mechanism needs to be explored in the future.
The study showed that substituting animal protein and fat for carbohydrates could not lead to an increase in LDL-cholesterol, hypercholesterolaemia and hypertriacylglycerolaemia. This result has been confirmed in most studies(Reference Ha, Joung and Song22,Reference Shirani, Esmaillzadeh and Keshteli24,Reference Jafari-Maram, Daneshzad and Brett25) . The three studies did not find a significant association between the LCD scores and low LDL-cholesterol, hypercholesterolaemia or hypertriacylglycerolaemia. Only one study found that a higher animal-based LCD score was significantly associated with higher odds of hypercholesterolaemia and hypertriacylglycerolaemia in males(Reference Tan, Kim and Shin27). This study explains that higher consumption of an animal-based diet leads to higher TC levels. A prospective study also shown that animal protein substitution of carbohydrates was positively associated with LDL-cholesterol or TC(Reference Meng, Cui and Li51). However, the notion that animal-based LCD may have a deleterious effect on blood lipids remains speculative. More studies are needed in the future to carefully explore whether replacing LCD with more animal protein and fat increases the risk of dyslipidaemia.
Plant-based LCD score but not animal-based LCD score was negatively associated with IFG after the multivariate analysis in this study. The likely reason was that participants with high plant-based LCD scores consumed more legumes and nuts. A prospective study shown that replacing similar bread or rice with half a daily serving of beans may reduce the incidence of diabetes(Reference Becerra-Tomas, Diaz-Lopez and Rosique-Esteban52). A systematic review showed that nuts (walnuts, almonds and hazelnuts) reduced FBG and glycated Hb levels by varying degrees(Reference Muley, Fernandez and Ellwood53). After sex stratification, the association between LCD scores and IFG was found only in women. Shirani et al. also found that LCD score was associated with low FBG in Iranian women but did not study in men(Reference Shirani, Esmaillzadeh and Keshteli24). Ha et al. showed that both males and females who adhered to the LCD had no association with FBG(Reference Ha, Joung and Song22). There is no study to explore the sex difference of LCD on IFG, and the specific mechanism needs to be solved in future studies.
The study found all LCD scores were not significantly associated with blood pressure, similar to previous findings(Reference Ha, Joung and Song22,Reference Shirani, Esmaillzadeh and Keshteli24,Reference Jafari-Maram, Daneshzad and Brett25) . There were no association between total LCD score with high blood pressure. At the same time, meta-analysis did not find a significant difference between LCD and isoenergetic balanced or higher carbohydrate diets for either systolic blood pressure or diastolic blood pressure(Reference Naude, Schoonees and Senekal54,Reference Korsmo-Haugen, Brurberg and Mann55) . Even if LCD showed a short-term advantage in lowering systolic or diastolic blood pressure compared with the high carbohydrate diet, the effect disappeared after a year(Reference Klemsdal, Holme and Nerland56,Reference Wycherley, Brinkworth and Clifton57) . Only a cohort study shown that total LCD score was a faint association with blood pressure in Tehranian adults (P = 0·048)(Reference Mirmiran, Asghari and Farhadnejad21). The study also found all LCD scores were not significantly associated with hyperuricaemia. Similar results were also found in the study by Nakamura et al.(Reference Nakamura, Ueshima and Okuda23). However, there was a randomised controlled trial showing that a 24-month non-energy-restricted LCD improved uric acid levels(Reference Yokose, McCormick and Rai58). The study restricted carbohydrates to 20–120 g/d, compared with 165 g in the highest quintile of the LCD score in this study. It is very difficult for Chinese people to meet this requirement in real life without any intervention.
This study has several strengths. First, this was the study to use the LCD scores to study multiple cardiometabolic risk factors among Chinese adults. Second, few studies had considered the effect of condiment intake on the results. Condiment was adjusted as confounding factors in this study.
This study has several limitations. First, this was a cross-sectional study, and the causal relationship between three LCD scores and cardiometabolic risk could not be established. Further large prospective studies are required to examine the effect of three LCD scores on cardiometabolic risk factors in the Chinese population. Second, this study used a semi-quantitative FFQ which may have a large recall bias. This study found that FFQ underestimated energy intake (1306–1531 kcal/d). Part of the reason may be that the total energy intake of the elderly in China is low, and 44·5 % of the participants in this study were elderly. A study showed that mean total energy intake was 1463 kcal/d among older Chinese adults in 2009(Reference Pan, Smith and Batis59). A 3-d 24-h dietary recalls should be applied to assess dietary intake in future studies. Third, the questionnaire on condiments was based on the household consumption divided by the number of family members. This calculation method cannot accurately reflect the actual situation of personal condiment intake. Fourth, the participants in this study only included the population in northern China. However, there is a big difference in eating habits in the north and south of China. A study has shown that people in southern China consume more grains, beans, milk and eggs, and less fish and seafood than people in the north(Reference Shi, Wang and Cao60). Therefore, it is difficult to extend the results of this study to the general Chinese population. Finally, blood lipids, blood pressure and uric acid are good predictors of CVD in this study. However, other risk factors (i.e. small, dense LDL, lipoprotein-a and inflammatory biomarkers) are more closely linked to CVD outcomes(Reference Diamond, O’Neill and Volek19), which should be used to explore the relationship with three LCD scores.
Conclusions
This study found that the LCD score was negatively associated with low HDL-cholesterol and IFG. Males in the highest quintile of the animal-based or plant-based LCD scores showed a decreased risk of low HDL-cholesterol, and females in the highest quintile of the animal-based or plant-based LCD scores showed a decreased risk of IFG than those in the lowest quintile of the LCD score. These results suggest that sex differences should be considered when using LCD to treat dyslipidaemia and reduce FBG. Further studies were needed to explore the specific mechanisms of the sex difference.
Acknowledgements
The authors are very grateful to all the participants in this study.
This research was supported by Community Cohort Study on Specialized Nervous System Diseases (No.2017YFC0907701). All funders had no role in the design, analysis or writing of this article.
All authors carried out the study. J. W. analysed the data, interpreted the findings and wrote the first draft. S. L. and Y. Z. designed the study. Y. M. made the funding acquisition and project administration. Y. S., H. Z., G. Y. and Y. W. provided critical comments and approved the final manuscript.
All authors declare that they have no conflict of interest.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114522001076
