The Mediterranean score of dietary habits in Chinese populations in four different geographical areas

(2001) 55, 215±220 ß 2001 Nature Publishing Group All rights reserved 0954±3007/01 $15.00 www.nature.com/ejcn The Mediterranean score of dietary habits in Chinese populations in four different geographical areas J Woo 1 *, KS Woo 1, SSF Leung 1, P Chook 1, B Liu 1,RIp 1,SCHo 2, SW Chan 3, JZ Feng 4 and DS Celermajer 5 1 Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong, People's Republic of China; 2 Department of Community and Family Medicine, The Chinese University of Hong Kong, Hong Kong, People's Republic of China; 3 Chinese Hospital, San Francisco, California, USA; 4 Guangdong Provincial Cardiovascular Institute, Guangzhou, People's Republic of China; and 5 Department of Cardiology, The Royal Prince Alfred Hospital, Sydney, Australia Objective: To compare the dietary intake of Chinese people living in Pan Yu, Hong Kong, San Francisco and Sydney with respect to cardiovascular health, using the Mediterranean diet score, examining the effects of age, gender, urbanization and acculturation on the diet score. Subjects: A total of 500 men and 510 women in Hong Kong were recruited as a territory-wide strati ed random sample. Subjects were recruited in response to local advertisements for the other three sites: Pan Yu, 58 men, 95 women; San Francisco, 166 men, 192 women; Sydney, 95 men, 73 women. Method: Food-frequency questionnaire over a 7 week period. A high=healthy score was taken as 4 for men and 3 for women, representing a dietary pattern bene cial for cardiovascular health. Results: In Hong Kong, more women in the middle age group (35 ± 54) had a high score than other age groups, and overall more women had high scores than men. In comparing the four geographical regions, Pan Yu had the highest number of subjects with high score, and Hong Kong had the lowest. With the exception of the younger population and men in Hong Kong, the percentage of the population with a high score in all sites is greater than among elderly Greeks consuming a more traditional heart-healthy Mediterranean diet. Conclusion: Considerable variations in Chinese dietary patterns exist with respect to age, gender and geographic location. Overall, the Chinese diet is comparable to the Mediterranean diet and may be expected to have similar health bene ts that have been documented for the traditional Mediterranean diet. Descriptors: Chinese diet; cardiovascular disease; acculturation; urbanization; Mediterranean diet (2001) 55, 215±220 Introduction The dietary pattern of inhabitants around the shores of the Mediterranean has long been noted to have health bene ts, particularly with respect to coronary heart disease (Keys, 1995; Nestle, 1995). The main features of the diet are: high consumption of fruits, vegetables, legumes and grains; foods with high monosaturated to saturated fats ratio; moderate consumption of dairy products and ethanol (mainly wine); and low consumption of meat and meat products (Trichopoulou & Lagiou, 1997). Longitudinal studies and randomized controlled trials showed that a Mediterranean dietary pattern is associated with improved survival among elderly Europeans (de Groot et al, 1996; Osler & Schroll 1997), possibly reduced cancer rate (de *Correspondence: J Woo, Department of Medicine and Therapeutics, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR, People's Republic of China. E-mail: jeanwoowong@cuhk.edu.hk Received 6 September 2000; revised 3 November 2000; accepted 6 November 2000 Lorgeril et al, 1998), and reduced cardiovascular complications after myocardial infarction (de Lorgeril et al, 1999). The diet emphasizes food groups rather than speci c nutrients, and this traditional plant-based diet may provide bene t through the intake of antioxidants, bre, or indirectly by lowering blood pressure (de Lorgeril, 1998). While individual items of the Mediterranean diet may not be predictive of outcome, the extent to which dietary patterns conform to the Mediterranean diet, quanti ed as a score, is predictive of outcome (de Groot et al, 1996; Osler & Schroll 1997). Among Chinese populations, the incidence of coronary heart disease is lower than in Caucasians (Woo & Donnan, 1989), and a difference in dietary habits may be a contributory factor. The Chinese diet has many similar features with the Mediterranean diet, in that vegetable and fruit consumption is high, and fat and meat consumption is low. However, there is no longitudinal observation study or randomized controlled trial to determine if a traditional Chinese diet is related to improved survival or better health outcomes. Since such evidence is available for the Mediterranean diet, it would be of interest to determine

216 whether the dietary habit of Chinese populations is similar, by quantifying the features using the Mediterranean diet score, developed by other researchers (de Groot et al, 1996). The proportion of the population with a high score representing a dietary pattern bene cial to cardiovascular health may be compared with published studies for Caucasian populations, as a useful indicator of the dietary pattern of the community with respect to chronic diseases. In this study, the Mediterranean score is calculated for Chinese populations in four geographic regions throughout the world, and the effect of age, gender, urbanization, acculturation and, indirectly, the effect of public health education in different countries on the Mediterranean score is also examined. Subjects and method Four cohorts of ethnic Chinese subjects living in Hong Kong, a rural village on the outskirts of Pan Yu in Guangdong province in Southern China, in Sydney and in San Francisco were studied. (Hong Kong), 500 M, 510 F; Pan Yu, 58 M, 95 F; San Francisco, 166 M, 192 F; Sydney, 95 M, 73 F). Subjects in the Hong Kong cohort were recruited as part of a territory-wide cardiovascular risk factor study from October 1995 to May 1996, a strati ed random survey where the sample was representative of the Hong Kong population. Details of the recruitment have been given elsewhere (Woo et al, 1998). The other three cohorts were community volunteers. The total population of the rural village, Pan Yu district, China, is approximately 3000, and has one of the lowest annual rates of acute myocardial infarction in this region, being approximately 40 per 100 000 among those age 40 y and over, compared with 90.8 ± 117.3 per 100 000 in Hong Kong (Woo & Donnan, 1989). Posters advertising a cardiovascular risk factor survey were placed in the community in the vicinity of the hospital or health centre. All advertising material was written only in the local language in each centre. For San Francisco and Sydney, subjects were included if they were originally of southern Chinese origin (for the past three generations), and had lived in these cities for at least 10 y. Dietary intake was assessed in all cohorts using the same food frequency method. The quantity and frequency of consumption per day and per week for each item were recorded with the aid of photographs to illustrate portion sizes, and nutrient quantity was calculated using food tables from multiple sources Ð UK and two mainland Chinese institutions. Details of the method have been described elsewhere (Woo et al, 1997). The Mediterranean score was calculated according to the consumption of eight categories of food, and a score of one was given if the criteria for each category was full lled (de Groot et al, 1996): 1. High monounsaturated: saturated fat ratio (> 1.6). 2. Moderate ethanol consumption (men: < 10 g=day). 3. High consumption of legumes (men > 60 g=day, women > 49 g=day). 4. High consumption of cereals (mean > 291 g=day, women > 248 g=day). 5. High consumption of fruits (men > 249 g=day, women > 216 g=day). 6. High consumption of vegetables (men > 303 g=day, women > 248 g=day). 7. Low consumption of meat and meat products (men < 109 g=day, women < 91 g=day). 8. Low consumption of milk and dairy products (men < 201 g=day, women < 194 g=day). These values were adjusted to daily intakes of 2500 kcal for men and 2000 kcal for women. For our population, since very few women drink, item 2 was deleted, so that a total score of 7 for women was used instead of 8. For this study, a high score for men is de ned as 4 for men, and 3 for women, and represents a dietary pattern that is bene cial for cardiovascular health. A score of 3 or 4 has also been used to relate Mediterranean dietary pattern to better health outcomes (Osler & Schroll, 1997). Mean nutrient intakes between the four cohorts were compared using Tukey's method of pairwise multiple comparisons. Chi-square test was used to compare the number of subjects with high score between different age and sex groups, and between different geographical regions. Results The mean nutrient intake of Chinese people in the four regions is shown by age and sex in Table 1. For the younger age group (34 y), there was no difference in energy intake among men, but women in San Francisco had the lowest intake. Men and women in Pan Yu had the highest carbohydrate and lowest protein percentage of total energy compared to the other regions. Men in Pan Yu also consumed the lowest percentage fat. The MUFA:SFA ratio was also the highest for both men and women in Pan Yu, while the PUFA:SFA ratio was also the highest among men in Pan Yu. Both men and women in Hong Kong had the highest cholesterol intake per 1000 kcal of energy compared with other regions, while women in Hong Kong also had the highest percentage fat consumption. Among middle aged Chinese (35 ± 54 y), men in Sydney and Hong Kong had higher energy intakes. The lowest percentage protein was still observed for the Pan Yu population, while it was highest for San Francisco. Interestingly, the percentage fat was highest for Pan Yu; at the same time the MUFA:SFA and PUFA:SFA ratio were also high. The cholesterol intake was also the highest in Hong Kong, as for the younger age group. For the older age group (55 y), fewer differences between regions were observed. The pattern of lowest protein percentage, higher fat percentage and MUFA:SFA ratios in the Pan Yu population was also observed.

Table 1 Mean (s.d.) nutrient intake in four regions by age and sex 217 Male Female Macronutrient composition Sydney Pan Yu San Francisco Hong Kong Sydney Pan Yu San Francisco Hong Kong (A) 34-y-old or below n 11 21 28 103 11 45 28 106 Energy (kcal) 2622 2596 2173 2557 1931 1982 b 1502 b ** c 1891 c * (736) (879) (749) (723) (766) (649) (592) (583) Carbohydrate (%) 51 a 60 a * b 49 b *** 52 b ** 55 59 b 53 50 b *** (10) (12) (9) (7) (12) (10) (12) (7) Protein (%) 18 15 b 21 b *** 19 b *** 16 a 15 b 20 b *** 20 a * b *** (4) (4) (7) (3) (6) (4) (6) (3) Fat (%) 31 25 b 30 30 b * 25 a 26 b 27 c 32 a ** b *** c * (8) (12) (8) (6) (8) (9) (9) (5) MUFA:SFA 1.26 a 1.62 a *** b 1.26 b *** 1.27 b *** 1.15 a 1.48 a *** b 1.27 b *** 1.27 b *** (0.21) (0.38) (0.28) (0.01) (0.18) (0.30) (0.21) (0.13) PUFA:SFA 0.87 1.01 b 0.73 b ** 0.77 b ** 0.86 0.85 0.79 0.84 (0.23) (0.48) (0.35) (0.21) (0.44) (0.45) (0.42) (0.24) Cholesterol (mg=1000 kcal) 137.80 117.00 b 145.6 170.2 b *** 88.60 a 134.3 b 150.1 a * 171.5 a *** b *** (57.6) (68.2) (53.8) (45.6) (39.4) (61.5) (78.7) (47.8) (B) 35 ± 54-y-old n 67 19 91 267 48 29 110 276 Energy (kcal) 2556 a 2147 2200 a * c 2470 c *** 2061 2113 1759 1857 (1051) (499) (810) (650) (1160) (668) (676) (598) Carbohydrate (%) 55 54 52 53 53 55 52 53 (8) (12) (9) (8) (8) (10) (8) (7) Protein (%) 19 a 16 a * b 23 a *** b *** c 19 c *** 19 a 14 a *** b 21 a *** b *** c 19 b *** c *** (5) (4) (6) (3) (4) (5) (6) (3) Fat (%) 26 a 30 a * b 25 b * c 29 a *** c *** 30 32 b 28 b * 29 (6) (11) (8) (6) (8) (10) (7) (5) MUFA:SFA 1.35 a 1.56 a *** b 1.41 b * c 1.29 b *** c *** 1.41 a 1.70 a *** b 1.38 b *** c 1.29 a * b *** c ** (0.20) (0.30) (0.35) (0.12) (0.34) (0.30) (0.36) (0.16) PUFA:SFA 1.01 a 1.11 b 0.89 b * 0.81 a *** b *** 1.10 a 1.22 b 0.95 b ** 0.94 a * b *** (0.40) (0.65) (0.34) (0.24) (0.39) (0.62) (0.40) (0.31) Cholesterol (mg=1000 kcal) 109.80 a 162.6 a *** b 124.8 b * c 171.0 a *** c *** 110.00 a 125.8 b 132.4 c 154.9 a *** b * c ** (47.1) (71.6) (56.6) (53.90) (48.2) (70.1) (54.60) (54.1) (C) 55-y-old or above n 17 18 47 130 14 21 54 128 Energy (kcal) 2336 2379 1966 2211 1948 1827 1657 1654 (749) (686) (817) (612) (834) (589) (928) (454) Carbohydrate (%) 53 52 55 57 57 58 54 57 (6) (10) (9) (8) (10) (9) (9) (7) Protein (%) 20 a 15 a * b 21 b *** c 18 c *** 19 15 b 22 b *** c 18 b * c *** (4) (5) (7) (3) (4) (3) (7) (3) Fat (%) 28 29 b 24 b ** c 27 c * 25 28 25 27 (7) (8) (6) (7) (7) (9) (9) (5) MUFA:SFA 1.36 a 1.67 a *** b 1.42 b *** c 1.31 b *** c ** 1.42 1.61 b 1.41 b * 1.31 b *** (0.23) (0.22) (0.29) (0.14) (0.17) (0.27) (0.44) (0.17) PUFA:SFA 0.99 0.94 1.13 c 0.91 c *** 1.01 1.04 0.98 0.96 (0.32) (0.29) (0.45) (0.30) (0.41) (0.41) (0.46) (0.35) Cholesterol (mg=1000 kcal) 118.9 141.8 122.8 137.8 118.9 123.6 132 136.7 (38.3) (71.0) (63.5) (55.7) (53.2) (47.0) (60.1) (48.4) a Differ signi cantly from Sydney. b Differ signi cantly from Pan Yu. c Differ signi cantly from San Francisco. *P < 0.05; **P < 0.01; ***P < 0.001. Post hoc pairwise multiple comparisons (Tukey method). The Mediterranean score was calculated for men and women, and the number of subjects divided into those with low and high scores. The score differed by age and gender among the Hong Kong population only (Table 2). More women in the middle age group (35 ± 54) had a higher score compared with other age groups, while more men had lower scores than women. The number of men and women with high scores was the highest in Pan Yu and

218 Table 2 gender Mediterranean score of the Hong Kong population by age and Mediterranean score high score a Sex and age group n Count % Female age 34 or below 106 66 62 Female age 35 ± 54 276 209 76 Female age 55 or above 128 85 66 Male age 34 or below 103 42 41 Male age 35 ± 54 267 129 48 Male age 55 or above 130 81 62 a Score 4 for male, score 3 for female. Chi-square: 62.359, P ˆ < 0.001. lowest in Hong Kong (Table 3). When the dietary pattern was examined by age group, for all age groups, Pan Yu had the highest percentage of subjects with a high score (Table 4). For all age groups, the dietary pattern was worst in Hong Kong. In summary, the results show considerable variations in Table 3 Mediterranean score in different regions by sex Mediterranean score high score a Sex Place n Count % Female b Hong Kong 510 360 71 Pan Yu 95 81 85 San Francisco 192 147 77 Sydney 73 58 79 Male c Hong Kong 500 252 50 Pan Yu 58 56 97 San Francisco 166 117 70 Sydney 95 61 64 a Score 4 for male; score 3 for female. b Pearson chi-square ˆ 11.174; P ˆ 0.011. c Pearson chi-square ˆ 59.325; P ˆ < 0.001. Table 4 Mediterranean score in different regions by age group Mediterranean score high score a Age group Place n Count % 34 or below b Hong Kong 209 108 51 Pan Yu 66 55 83 San Francisco 56 35 63 Sydney 22 13 59 35 ± 54 c Hong Kong 543 338 62 Pan Yu 48 46 96 San Francisco 201 154 77 Sydney 115 80 70 55 or above d Hong Kong 258 166 64 Pan Yu 39 36 92 San Francisco 101 75 74 Sydney 31 26 84 a Score 4 for male; score 3 for female. b Pearson chi-square ˆ 21.115, P ˆ < 0.001. c Pearson chi-square ˆ 34.402; P ˆ < 0.001. d Pearson chi-square ˆ 17.014; P ˆ < 0.001. Chinese dietary patterns with age, gender, and geographical locations. Discussion To date, the only studies relating dietary patterns to health outcomes are those relating to the Mediterranean diet (Keys, 1995; Nestle, 1995; Trichopoulou & Lagiou, 1997; de Groot et al, 1996; Osler & Schroll, 1997; de Lorgeril et al, 1998; de Lorgeril, 1998). These studies suggest that dietary habits may be considered as a risk factor for cardiovascular diseases or survival in older populations. It is likely that the lower incidence of cardiovascular diseases in Chinese populations may be partly attributed to the Chinese diet. Although there are many variants of the Mediterranean and Chinese diets, the development of a Mediterranean diet score in some previous studies (de Groot et al, 1996; Osler & Schroll, 1997) enables a quantitative comparison of the Chinese and Mediterranean dietary habits. There are limitations in using a dietary score devised by another author, in that the characteristics of the population to which the score applies will be different to Chinese populations. Furthermore, there may be other characteristics in the Chinese diet with adverse or bene cial effects to health that have not been taken into account in comparison with different populations. The percentage of the population with a high score ranges from 51% to 96% in four geographic regions with different culture, educational background and socioeconomic characteristics. With the exception of the younger population and men in Hong Kong, the percentage with a high score was greater than the percentage among elderly Greeks (57%) (Trichopoulou et al, 1995). Therefore Chinese populations as a whole have a dietary pattern similar to the Mediterranean dietary pattern, the majority achieving a high score irrespective of acculturation, age or socioeconomic factors. It is possible that the current dietary patterns of Chinese populations may be `healthier' with respect to cardiovascular disease compared with the `Mediterranean' population, since the Mediterranean diet is a traditional diet and studies suggest that there has been a deviation from traditional dietary patterns with time (Ferro- Luzzi & Branca, 1995; Alberti-Fidanza et al, 1999). Among the Chinese population, the dietary pattern shows interesting variations with age, gender and geographical region, re ecting in uences of local culture, health messages promulgated by media and the government, differences in education, economic status and degree of urbanization. In Pan Yu, being a rural area, the population will tend to preserve a more traditional dietary pattern, and their relatively lower income may preclude a diet high in protein from meat or seafood. However, they will be less exposed to nutrition education compared with societies in San Francisco or Sydney, and this may be re ected in the observation that the percentage fat comsumption increased with age in the Pan Yu residents, perhaps re ecting an

increase in income. Elsewhere it has been documented that the consumption of fat and meat in China increases with improvement in socioeconomic status (Ge et al, 1995). In contrast, Hong Kong is entirely urban, and exposed to Western in uences from other countries. Yet the dietary habit of the population appears worse than that of Chinese populations in Sydney or San Francisco. A difference in socioeconomic status or a lower level of public health knowledge regarding nutrition and disease in Hong Kong may account for the difference. It is also possible that the lifestyle in Hong Kong, particularly for men and the young and middle-aged groups, consists of eating away from home much of the time, the fast food outlets being a regular source of meals for a large percentage of the population. In general, there is less geographical variation in dietary habits among the older population, perhaps re ecting a habit of taking more traditional Chinese meal patterns rather than adopting `Westernized' foods. The gender difference in dietary habits in Hong Kong is of interest. Young and middle-aged women had a better pattern compared with men. This may be a difference in lifestyle, in that more women may stay at home to look after children, or that women have better nutritional knowledge than men. Since meals are largely provided by women, one would not have expected a marked gender difference if men ate most of the meals at home (if they were not single). Just as there are variations in Mediterranean dietary patterns in populations in countries around the Mediterranean (Ferro-Luzzi & Branca, 1995), variations exist for the Chinese diet. The public health implications are that Chinese populations living in Western countries appear to have healthier dietary habits than urban populations in China. Therefore it is probably not necessary to particularly promote the traditional Chinese diet for health bene ts in Chinese populations in Western countries. Indirectly this observation re ects the impact of public health programmes in these countries. However there is a need to raise the level of nutritional knowledge for Chinese populations in urban and socioeconomically advanced cities in China, such as Hong Kong, where the worst patterns occur in the younger and middle age groups. Since there are adverse features of the Chinese diet with respect to blood pressure and osteoporosis (high salt and low calcium content) (Woo et al, 1999), the standard dietary advice in the form of food pyramids (US Department of Agriculture, 1992; Willet et al, 1995) that is widely used in Western countries would be equally appropriate for the Chinese population in China. There are limitations for this study. The number of subjects in Pan Yu, Sydney and San Francisco may be too small in each age and sex group to detect any age and gender differences in dietary habits within each city. Also, the sampling method is different for the four geographical areas. The sample in Hong Kong is representative of the Hong Kong population (Woo et al, 1998). However, it is uncertain and dif cult to determine `representativeness' of Chinese populations living overseas. Similarly we were not able to comment on whether the Pan Yu subjects are representative of the whole of the Pan Yu district. Nevertheless, the strength of the study lies in that the same dietary assessment instrument had been used in all four regions, and was administered by the same team of interviewers. In spite of the limitations, one can conclude that currently the large majority of the Chinese population consume a diet comparable to the traditional Mediterranean diet, whether in China or in Western countries, and by extrapolating from the data for Mediterranean diet, are likely to have health bene ts in terms of coronary heart disease and survival. However, the adverse features of the Chinese diet have not been taken into account. There is need for improved nutritional knowledge in urban Chinese populations in China, in order to prevent a worsening trend in dietary habits. Acknowledgements ÐThis study is partially supported by the Bristol Myers Squibbs Foundation Unrestricted Nutrition Grant, and the Hong Kong Heart Foundation. References Alberti-Fidanza A, Fidanza F, Chiuchiu MP, Verducci G & Fruttini D (1999): Dietary studies on two rural Italian population groups of the seven countries study. 3. Trend of food and nutrient intake from 1960 to 1991. Eur. J. Clin. Nutr. 53, 854 ± 860. de Groot LCPGM, Van Staveren WA & Burema J (1996): Survival beyond age 70 in relation to diet. Nutr. Rev. 54, 211 ± 219. de Lorgeril M (1998): Mediterranean diet in the prevention of coronary heart disease. Nutrition 14, 55 ± 57. de Lorgeril M, Salen P, Martin JL, Monjaud I, Boucher P & Mamelle N (1998): Mediterranean dietary pattern in a randomized trial: prolonged survival and possible reduced cancer rate. Arch. Intern. Med. 158, 1181 ± 1187. de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J & Mamelle N (1999): Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: nal report of the Lyon Diet Heart Study. Circulation 99, 779 ± 785. Ferro-Luzzi A & Branca F (1995): Mediterranean diet, Italian-style: prototype of a healthy diet. Am. J. Clin. Nutr. 61(Suppl), 1338S ± 1345S. Ge K, Cheng L, Guo X, Wang Q, Hu J, Jia F, Liu J & Li J (1995): The dietary and nutritional status of the Chinese population (1992 National Nutrition Survey), Vol 1. Beijing China: People's Medical Publishing House Vol 1. Keys A (1995): Mediterranean diet and public health: personal re ection. Am. J. Clin. Nutr. 61(Suppl), 1321S ± 1323S. Nestle M (1995): Mediterranean diets: historical and research overview. Am. J. Clin. Nutr. 61(Suppl), 1313S ± 1320S. Trichopoulou A & Lagiou P (1997): Healthy traditional Mediterranean diet: an expression of culture, history and lifestyle. Nutr. Rev. 55, 383 ± 389. Osler M & Schroll M (1997): Diet and mortality in a cohort of elderly people in a North European community. Int. J. Epidemiol. 26, 155 ± 159. Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, Gnardellis C, Lagiou P, Polychronopoulos E, Vassilakou T, Lipworth L & Trichopoulos D (1995): Diet and overall survival in elderly people. Br. Med. J. 311, 1457 ± 1460. US Department of Agriculture (1992): The food guide pyramid. (Publication HG252) Hyattsville, MD: Human Nutrition Information Service. Willet WC, Sacks F, Trichopoulou A, Drescher G, Ferro-Luzzi A, Helsing E & Trichopoulous D (1995): Mediterranean diet pyramid: a culture model for healthy eating. Am. J. Clin. Nutr. 61(Suppl), 1402S ± 146S. 219

220 Woo KS & Donnan SPB (1989): Epidemiology and coronary arterial disease in the Chinese. Int. J. Cardiol. 24, 83 ± 93. Woo J, Leung SSF, Ho SC, Lam TH & Janus ED (1997): A food frequency questionnaires for use in the Chinese population in Hong Kong: description and examination of validity. Nutr. Res. 17, 1633 ± 1641. Woo J, Leung SSF, Ho SC, Lam TH & Janus ED (1998): Dietary intake and practices in the Hong Kong Chinese population. J. Epidemiol. Community Health 52, 631 ± 637. Woo J, Leung SSF, Ho SC & Chan SM (1999): Is there a typical Chinese diet and what are the health implications? Ecol. Food Nutr. 38, 491 ± 503.