Vegans and ovo-lacto vegetarians are the two common varieties of vegetarians. Vegetarians consume smaller amounts of total fat and saturated fat and larger amounts of unsaturated fats and fibre than omnivores( Reference Craig 1 ). The consumption of a vegetarian diet is believed to reduce the risk of CVD( Reference Key, Fraser and Thorogood 2 – Reference Pettersen, Anousheh and Fan 5 ). HDL-cholesterol (HDL-C), LDL-cholesterol (LDL-C), TAG and the ratio of HDL-C to total cholesterol (TC) are often used to predict the risk of having CVD( Reference Ridker, Rifai and Cook 6 , 7 ). Better plasma lipid and lipoprotein profiles are proven strategies to reduce risk for CVD( 7 ).
Vegetarians tend to have lower LDL-C and TC than omnivorous individuals( Reference Chen, Lin and Lin 8 ). Despite having lower HDL-C and LDL-C levels, the ratio TC:HDL-C ratio was lower in female vegetarians( Reference Chiang, Lin and Chen 9 ). Healthy postmenopausal vegetarians had lower TC, LDL-C and TAG( Reference Fu, Yang and Lin 10 ). Fu et al. also reported that postmenopausal vegetarians had lower TC, HDL-C and LDL-C than age-matched omnivores. In a meta-analysis, vegetarian diets were effective in lowering plasma TAG concentrations, especially in developing countries( Reference Zhang, Ma and Chen 11 ). HDL-C is an independent risk factor of CVD( Reference Ashen and Blumenthal 12 ). In our previous work, coronary artery disease risk was higher in patients with low levels of HDL-C than in those with high levels of LDL-C( Reference Huang, Ho and Lin 13 ). Results from previous studies on the effects of vegetarian diets on lipids, especially HDL-C, have been inconsistent. Besides, CVD is relatively more common in postmenopausal women compared with age-matched men or premenopausal women( Reference Villablanca, Jayachandran and Banka 14 ). In this regard, the present study focused on the effects of vegetarian diets on HDL-C, LDL-C, TAG and HDL-C:TC by gender.
Materials and methods
The data were obtained from the 2002 Taiwanese Survey on the Prevalence of Hyperglycemia, Hyperlipidemia and Hypertension (TwSHHH) for individuals aged 15 years and above. The survey was approved, conducted and provided by the National Health Research Institutes and Bureau of Health Promotion. Information was collected on age, gender, exercise, dietary habits and organic food. Measurements included waist:hip ratio (WHR), blood pressure and serum biochemistry. Individuals who had received lipid-lowering drugs were excluded. Participants were instructed to fast for more than 12 h prior to blood sampling. Blood samples were collected to determine blood glucose, HDL-C, LDL-C, apo-A1, apo-B, TAG, TC, uric acid and creatinine.
Vegetarians were identified from the responses to the interview questions. The questions were related to dietary type (vegan and ovo-lacto vegetarian). The questionnaires did not contain the exact amount of diet and the recommended daily intake. Vegan diet was defined as a diet that excluded eggs, milk, meat, poultry, seafood and by-products of animals, while ovo-lacto vegetarian diet was one that excluded all meat and fish but included eggs and dairy products, for a period of 1 year.
Confounders included gender, age, exercise, organic food, WHR, blood pressure, fasting plasma glucose, uric acid and creatinine. Two seated blood pressure readings were taken 10 s apart in the left arm after a 5–10 min rest. Another blood pressure measurement was taken if the first two readings varied by more than 10 mmHg. The mean of the two closest readings was calculated. Hip circumference was measured at the level of the greater trochanter. Waist circumference was measured midway between the iliac crest and the last rib. WHR was calculated as waist circumference (centimetres) divided by hip circumference (centimetres) as the index of central obesity. Because body weights were not measured during the survey, obesity was defined as WHR ≥0·90 for men and ≥0·85 for women( 15 ). Exercise was defined by individuals who had regular exercise at least three times per week (30 min each time) for 3 months or more. An organic food diet was defined as a diet in which half of the food was from an organic grocery store and for consumers who had followed such a diet for at least 1 year prior to the study.
Statistical analysis
All analyses were conducted by using the SAS statistical software package version 9·2. Continuous data were expressed as mean and standard deviation. Comparisons among vegans, ovo-lacto vegetarians and omnivores were made using ANOVA. Multiple linear regression analysis was used to estimate the association between serum lipid levels, associated factors and the three diet groups. A P value <0·05 was considered to be statistically significant.
Results
The study included 3257 men (forty-five vegans, sixty-eight ovo-lacto vegetarians and 3144 omnivores) and 3551 women (ninety-nine vegans, 127 lacto-ovo vegetarians and 3325 omnivores). The demographic, laboratory and anthropometric characteristics of males are shown in Table 1. There were no significant differences in waist circumference, WHR, blood pressure, fasting plasma glucose, apo-A1, apo-B, TAG and HDL-C:TC among the three diet groups. Compared with omnivores, vegans were significantly older and had lower levels of HDL-C, LDL-C and TC.
Table 1 Demographic, laboratory and anthropometric characteristics of vegan, ovo-lacto vegetarian and omnivorous men, Taiwanese Survey on the Prevalence of Hyperglycemia, Hyperlipidemia and Hypertension (TwSHHH), 2002
WC, waist circumference; HC, hip circumference; WHR, waist:hip ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure; FPG, fasting plasma glucose; UA, uric acid; HDL-C, HDL-cholesterol; LDL-C, LDL-cholesterol; TC; total cholesterol.
* Mean values were significantly different from those of omnivores (P<0·05).
Table 2 shows the demographic, laboratory and anthropometric characteristics of female vegans, ova-lacto vegetarians and omnivores. There were no significant differences between diet groups for waist circumference, hip circumference, fasting plasma glucose, LDL-C and TC. Vegans registered lowest HDL-C:TC and highest age, WHR and TAG levels, which were significantly different compared with omnivores. HDL-C was significantly lower in vegans and ovo-lacto vegetarians than in omnivores.
Table 2 Demographic, laboratory and anthropometric characteristics of vegan, ovo-lacto vegetarian and omnivorous women, Taiwanese Survey on the Prevalence of Hyperglycemia, Hyperlipidemia and Hypertension (TwSHHH), 2002
WC, waist circumference; HC, hip circumference; WHR, waist:hip ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure; FPG, fasting plasma glucose; UA, uric acid; HDL-C, HDL-cholesterol; LDL-C, LDL-cholesterol; TC; total cholesterol.
* Mean values were significantly different from those of omnivores (P<0·05).
† Mean values were significantly different from those of ovo-lacto vegetarians (P<0·05).
The associations between lipid profiles and dietary groups for males assessed by multiple linear regression analyses are shown in Table 3. After adjusting for potential confounders, vegan and ovo-lacto vegetarian diets had significant effects on LDL-C (β=−10·98, P=0·005 and β=−7·12, P=0·025, respectively). Significant associations were observed between vegan diet and HDL-C (β=−6·53, P=0·004) compared with omnivorous diet. Exercise was associated with higher HDL-C (β=1·16, P=0·045) and HDL-C:TC (β=0·01, P=0·001) and lower TAG levels (β=−7·91, P=0·026). There was no significant association between organic foods and lipid profiles.
Table 3 Regression analysis between diet types and serum lipid levels in men, Taiwanese Survey on the Prevalence of Hyperglycemia, Hyperlipidemia and Hypertension (TwSHHH), 2002
HDL-C, HDL-cholesterol; LDL-C, LDL-cholesterol TC, total cholesterol; SBP, systolic blood pressure; DBP, diastolic blood pressure; FPG, fasting blood glucose; UA, uric acid.
* Obesity is defined as waist:hip ratio ≥0·90.
The associations between serum lipid levels and dietary groups for females generated by multiple linear regression analyses are shown in Table 4. Vegan and ovo-lacto vegetarian diets were significantly associated with lower HDL-C (β=−5·72, P<0·0001 and β=−4·86, P=0·002, respectively) and HDL-C:TC (β=−0·02, P<0·012 and β=−0·01, P=0·026, respectively), and higher TAG (β=16·51, P=0·011 and β=15·09, P=0·008, respectively), compared with omnivorous diet. The difference associated with menopausal status was 2·60 (P=0·003) for HDL-C and 8·40 (P<0·0001) for LDL-C. Exercise was associated with a decrease in TAG level (β=−7·50, P=0·004) and an increase in HDL-C:TC (β=0·01, P=0·013). No significant association was found between organic foods and lipid profiles.
Table 4 Regression analysis between diet types and serum lipid levels in women, Taiwanese Survey on the Prevalence of Hyperglycemia, Hyperlipidemia and Hypertension (TwSHHH), 2002
HDL-C, HDL-cholesterol; LDL-C, LDL-cholesterol TC, total cholesterol; SBP, systolic blood pressure; DBP, diastolic blood pressure; FPG, fasting blood glucose; UA, uric acid.
* Obesity is defined as waist:hip ratio≥0·85.
Discussion
Lower levels of TAG, TC and LDL-C have been reported in vegans( Reference De Biase, Fernandes and Gianini 16 ). However, to our knowledge, no nationwide study has assessed the health effects of different diet groups on lipid profiles by gender in the Taiwanese population. In our study, vegetarian diet lowered LDL-C and HDL-C levels in men. HDL-C and HDL-C:TC were equally lowered in women, whereas TAG level was increased. Vegetarian diets were associated with higher CVD risk in women than men.
From the survey, the percentage of male vegans was 1·3 % while female vegans constituted 2·8 %. The traditional Taiwanese diet contains plenty of plant-based foods, fish in low to moderate amounts, poultry and eggs used in low amounts, red meat used sparingly and the absence of dairy products( Reference Lin, Yen and Chen 17 ). The tofu and other soyabean products, wheat protein, beans, nuts, pickles and plain rice in Taiwanese folk vegetarian meals are normally considered as a ‘light’ type of diet with less amounts of fat( Reference Chen, Chiang and Lin 18 ). From a national survey, elderly vegetarians in Taiwan were noted to have significantly lower daily total energy intake, lower TC, higher percentage of PUFA, higher Ca and higher crude fibre intakes compared with omnivores( Reference Huang, Fan and Liu 19 ).
However, a study conducted in a private membership chain clinic in Taiwan showed that veganism increased the risk of metabolic syndrome, high TAG and low HDL-C( Reference Shang, Shu and Wang 20 ). From a hospital-based study, healthy male vegetarians were noted with low TC and LDL-C while healthy female vegetarians had low TC levels( Reference Chen, Lin and Lin 8 ). In addition to lower TC, LDL-C and HDL-C, Taiwanese female vegetarians had higher levels of high-sensitivity C-reactive protein, homocysteine and TAG( Reference Chen, Lin and Lin 21 ). Besides, female Buddhists had significantly lower TC:HDL-C and LDL-C:HDL-C ratios( Reference Chiang, Lin and Chen 9 ). Postmenopausal vegetarians had lower TC, LDL-C and HDL-C compared with age-matched omnivores( Reference Su, Torng and Jeng 22 ). In the present study, female vegetarians had lower HDL-C and higher levels of TAG compared with omnivores, whereas Rizzo et al. reported higher HDL-C and lower TAG concentrations in vegetarians( Reference Rizzo, Sabate and Jaceldo-Siegl 23 ). Reductions in the intake of dietary fats are associated with lower plasma LDL-C and HDL-C levels( Reference Ashen and Blumenthal 12 ). Healthy untrained individuals consuming a low-fat diet were found to have lower HDL-C( Reference Meksawan, Pendergast and Leddy 24 ). Vegetarians in Taiwan had lower fat intake with low serum TC and TAG levels compared with omnivores( Reference Yen, Yen and Huang 25 ). In our study, vegan diet contributed to low HDL-C levels in both genders. There were sex disparities in the LDL-C, TAG and HDL-C:TC levels among diet patterns. However, the cross-sectional design was inadequate to explain such disparities.
Vegetarian diets were associated with lower HDL-C:TC in women after adjusting for confounders. The ratio TC:HDL-C (the reverse of HDL-C:TC) has been reported to be a better index for CVD than TC, HDL-C or LDL-C( Reference Ridker, Rifai and Cook 6 , Reference Jiang, Schulze and Li 26 ), or even LDL-C:HDL-C( Reference Ballantyne and Hoogeveen 27 ). Lack of all animal products in a diet increases the risk of certain nutritional deficiencies, including those of vitamins B12 and D, Ca and long-chain n-3 fatty acids( Reference Craig 1 ). Because of a substantial portion of essential nutrients in dairy products, especially Ca, K and Mg, the ovo-lacto vegetarian diet is beneficial for reducing the risk of metabolic syndrome, stroke and some cancers( Reference Weaver 28 ). Our results indicated that the effects of vegetarian and ovo-lacto diets on lipid profiles were similar among women, unlike men.
Central obesity, low HDL-C and high TAG have been considered as components of the metabolic syndrome( Reference Alberti, Zimmet and Shaw 29 ) and independent risk factors for CVD( Reference Toth 30 , Reference Schneider, Friedrich and Klotsche 31 ). Measures of abdominal obesity, principally waist circumference and WHR, are used as surrogates of body fat centralization and have been used to evaluate CVD risk( Reference Schneider, Friedrich and Klotsche 31 , Reference Schneider, Glaesmer and Klotsche 32 ). Obesity was more strongly related to TAG than other variables in both sexes. Some important confounding factors such as exercise can contribute in increasing HDL-C:TC( Reference Tseng, Ho and Chen 33 ).
Consumers purchase organic foods for different reasons some of which include concerns about pesticide and antibiotic residues, human health and delicacy. In a systematic review, biomarkers and nutrient levels in serum, urine, breast milk and semen in adults did not show any clinically meaningful differences between conventional and organic foods. However, residues of pesticides and antibiotic-resistant bacteria were present in small amounts( Reference Smith-Spangler, Brandeau and Hunter 34 ). In the present study organic food did not have any significant effect on lipid profiles.
There were several limitations in the present study. First, TwSHHH was a cross-sectional study that was susceptible to unmeasured confounders and reverse causation. Second, a potential self-selection effect may have existed among persons who adopted vegetarian diets because they may have been more aware of health concerns. We stratified the participants by sex, age, organic food consumption and exercise. Participants who received lipid-lowering therapy were excluded. Third, the questionnaires did not contain detailed information on the dietary contents to clarify whether it was a diet with mostly polyunsaturated rather than monounsaturated fat. This made it harder to clarify the relationship between cardiometabolic profiles. Finally, the small sample size of the vegans and ovo-lacto vegetarians may not have been sufficient to detect the difference between vegetarians and omnivores. Multivariate regression was used to evaluate the relationship between lipid profiles and explanatory variables, including age, WHR, menopause and other variables. We did not analyse the data by the matching method. Even with careful matching of the pairs, there would still be many other factors influencing lipid profiles which may be impossible to match.
Conclusions
Vegetarian diet resulted in a significant decrease in HDL-C:TC and a significant increase in TAG levels in females, as well as a significant decrease in HDL-C concentrations in both males and females. Ovo-lacto vegetarian diet may be more appropriate for men. The present results cannot directly address biological and nutritional mechanisms underlying the findings; further studies are needed to investigate such mechanisms.
Acknowledgements
Acknowledgments: This study is based on the data obtained from the 2002 Taiwanese Survey on Hypertension, Hyperglycemia, and Hyperlipidemia (TwSHHH), provided by the Bureau of Health Promotion, Department of Health, Taiwan. The descriptions or conclusions herein do not represent the viewpoints of the Bureau. Financial support: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Conflict of interest: None. Authorship: Y.-P.L. conceived and designed the study. C.-C.L., P.-C.K. and C.-C.H. analysed and interpreted the data. Z.-H.J., H.-C.C., Y.-C.C., Y.-C. Liaw and Y.-C. Liang drafted the manuscript. Z.-H.J. and O.N.N. did a critical revision of the manuscript. All authors contributed to, read and approved the final version of the manuscript. Ethics of human subject participation: Ethical approval was not required.
