Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
European Journal of Clinical Nutrition (2000) 54, 530±539 ß 2000 Macmillan Publishers Ltd All rights reserved 0954±3007/00 $15.00 www.nature.com/ejcn A comparison of dietary behaviour in Central England and a French Mediterranean region M Holdsworth1*, M Gerber2, C Haslam3, J Scali2, A Beardsworth4, MH Avallone2 and E Sherratt5 1 School of Biological Sciences, University of Nottingham, Nottingham, UK; 2Groupe d'EpideÂmiologie MeÂtabolique, INSERM-CRLC, Montpellier, France; 3Department of Human Sciences, Loughborough University, Loughborough, UK; 4Department of Social Sciences, Loughborough University, Loughborough, UK; and 5Department of Public Health and Epidemiology, The Medical School, University of Birmingham, Birmingham, UK Objective: The investigation involves comparison of dietary behaviour between UK and Mediterranean France by characterizing the pattern of the current French Mediterranean diet compared with the current British diet. Design: The ®ndings of two dietary surveys, one in the UK and one in France, are compared. An intervieweradministered questionnaire was used in both countries. Questions on food frequency were used to assess dietary behaviour, which were regrouped in the French survey to correspond with UK groupings. Dietary indices were constructed to describe dietary behaviour in relation to cancer recommendations for intake of fat, ®bre, meat, fruit and vegetables. Setting: The UK study was conducted in Leicestershire, central England and the French study was carried out in HeÂrault, southern France. Subjects: UK: n 418 subjects (57.9% female and 42.1% male; mean age 45.0 y); France: n 635 subjects (50.1% female and 40.9% male; mean age 49.8 y). Age range of both samples: 20 ± 74 y. Results: There were positive and negative trends in food consumption in each country. UK respondents reported eating more beans and pulses (P 0.000), less cheese (P 0.000), red meat (P 0.001), and processed meats (P 0.000) than French respondents. However, on the negative side, they ate less fruit and vegetables (P 0.000), ®sh and poultry (P 0.000), cereals (P 0.000), and more sweets and chocolates (P 0.000), and cakes, pastries, biscuits and puddings (P 0.000). Women had healthier diets in both countries. Conclusions: Overall the southern French diet was healthier as French respondents scored signi®cantly better for indices for fat, dietary ®bre, fruit and vegetables (P 0.000 in all cases). However, the French sample scored poorer for the meat index (P 0.000). Sponsorship: This study was supported by a grant from l'Association de la Recherche contre le Cancer (ARC) awarded to M Holdsworth. Descriptors: nutrition survey; dietary habits; food habits; Mediterranean France; UK European Journal of Clinical Nutrition (2000) 54, 530±539 Introduction An estimated 30 ± 40% of cancer cases throughout the world are preventable by dietary means (World Cancer Research Fund, 1997). Cancers of the stomach, colon and rectum are mostly preventable by appropriate diets and related factors. Mediterranean populations have lower rates of cancer (Cummings & Bingham,1998) and coronary heart disease (Corpet & Gerber,1997) than Northern European countries, suggesting that incorporating a Mediterranean style diet in the UK could reduce the incidence of these diseases. One of the factors contributing to the increasing *Correspondence: M Holdsworth, Division of Nutritional Biochemistry, School of Biological Sciences, University of Nottingham, Sutton Bonington Campus, Loughborough, Leicestershire LE12 5RD, UK. Guarantor: M Holdsworth. Contributors: MH re-organized both data sets for comparison and carried out the statistical analyses for the comparative study. MH prepared the paper, which all investigators edited. MG supervised both the comparative study and the French study, initiating and designing the French study. CH and AB designed the UK study protocol and supervised data collection and analysis. MHA collected the data for the French study and JS compiled the database. ES collected the data for the UK study and compiled the database. Received 24 August 1999; revised 2 February 2000; accepted 17 February 2000 burden of cancer and coronary heart disease, particularly in the UK, is the consumption of a diet low in fruit and vegetables and high in saturated fat. The conclusion that most cancers and heart disease are preventable has enormous implications for public health policy in the UK and France. The need for effective nutrition education strategies that acknowledge the complex in¯uences of food choice and eating behaviour has been identi®ed on a local (Holdsworth & Spalding, 1997) and national level (WHO, 1996). One of the dif®culties encountered by nutrition educators in promoting a `healthier diet' is that many consumers perceive such a diet as tasteless and unappetising (Wardle & Solomons, 1994; Holdsworth et al, 1997). A Mediterranean-style diet may be a more attractive and successful way of marketing healthier eating to consumers, as it is regarded as being palatable as well as bene®cial to health. Cost is also a major in¯uence on dietary change (eg LennernaÈs et al, 1997; Glanz et al, 1998) and may indeed be the key obstacle to increasing fruit and vegetable intake in the UK (Cox et al, 1996). The `Mediterranean diet' is characterized by a large intake of cereals, and of diverse fresh vegetables and fruit; a low intake of red meat, a large intake of ®sh and seafood; almost no milk or butter, but cheese and=or yoghurts; A comparison of dietary behaviour M Holdsworth et al visible fat as olive oil, and a moderate amount of red wine during meals (Corpet & Gerber, 1997). The bene®ts to health of such a diet have been recognized since the 1960's, with lower incidences of coronary heart disease and cancer, and a longer life expectancy compared with the UK (James, 1995; Gerber & Corpet, 1997, 1998). In the UK, eating patterns have changed more than in France, incorporating `foreign cuisine' including `Mediterranean' style cooking (Mennell, 1996). The change in eating habits appears to be less evident in France, where national traditions appear to have persisted; even so, a more North American type diet is becoming popular in France, with convenience meals becoming more widespread (Askegaard, 1993). This shift away from traditional eating habits is also evident in southern France, particularly amongst the young (Gerber, 1999). In a European-wide survey of dietary attitudes and beliefs (Gibney, 1997), French respondents reported that time and the cost of food were less of a barrier to healthy eating than for UK respondents. The French were less likely than UK respondents to perceive a need to change eating habits (Kearney et al, 1997) and were less likely to report that they were trying to eat healthily (LennernaÈs et al, 1997). The investigation reported here involves comparison of dietary behaviour between the UK and Mediterranean France by comparing the current French Mediterranean diet with the current British diet. Other aspects of dietary behaviour are also contrasted, for example use of vitamin and mineral supplements and alcohol consumption. Recommendations are made on how the positive aspects of eating patterns can be promoted in France and the UK for the prevention of cancer and other diet-related diseases. Methods UK study The Dietary Dilemmas study was conducted in Leicestershire, UK and is a cross-sectional study investigating dietary behaviour and attitudes to diet on a range of broad dimensions. These relate to frequency and patterns of food consumption, dietary change, self perceptions of health and health-related behaviour (smoking, alcohol and exercise), attitudes towards diet and dietary issues, nutritional knowledge and awareness, dietary concerns, and food purchase and preparation. The sample comprised 421 individuals drawn from the Family Health Services register of Leicestershire Health. The sample drawn was comparable with age pro®le statistics from the General Household Survey in Britain in 1994 (Of®ce of Population Censuses and Surveys, 1996a,b). Three individuals were excluded from the analysis as they were aged less than 20 y, to allow comparability with the sample in the MEDHEA study, reducing the sample size to 418. The ®nal response rate of the study was 35%. Subjects were interviewed in their own homes using an interviewer-administered, structured questionnaire (available from the ®rst author). The design of the study has been reported before (Goode et al, 1995, 1996). This paper reports on the data collected on food consumption. Questions on food frequency were used to generate data on the frequency of consumption of key food items to allow an assessment of dietary behaviour. Foods were listed which were the major sources of fat, dietary ®bre and sugar using data from the National Food Survey (Ministry of Agriculture, Fisheries and Food, 1992). Food types= groupings were listed and respondents were asked to record how often they ate the foods. Foods with comparable nutrient content were grouped together and groupings were avoided that combined foods eaten in different circumstances. These considerations have been identi®ed as crucial (Kemm & Booth, 1992). Portion size was not recorded, because of the decision not to measure nutrient intake. A full list of the 12 food groupings used is illustrated in Table 2. Questions were also included to measure vitamin and mineral supplement intake, breakfast cereal consumption, type of spreading fat used and alcohol intake. 531 French study The population under study is a sub-sample of a larger sample randomly recruited in HeÂrault deÂpartement, southern France, for the Mediterranean Diet and Health Study (MEDHEA) (Gerber et al, 1997, 1999). The MEDHEA study is cross-sectional and investigates current eating behaviour and the socio-cultural in¯uences of food in Mediterranean countries. The study is co-ordinated from Montpellier, and includes 967 individuals in HeÂrault, southern France. Data on food frequency servings were available on 635 respondents, which are included in this analysis. Strati®ed sampling from electoral lists by age group was conducted within the age range 20 ± 74. Respondents were interviewed in their own homes using an intervieweradministered, structured questionnaire (available from the ®rst author). The ®nal response rate was 48%. Food frequency and portion size of 162 individual food items were recorded. Questions were also included to measure vitamin and mineral supplement intake, breakfast cereal consumption, type of spreading fat used and alcohol intake (questionnaire available from the ®rst author). Comparison study Regrouping food frequency responses. The investigation involves analysis and comparison of data on dietary behaviour: the frequencies of consumption of food types are compared, eg how often fruit and vegetables are eaten. For the purposes of this analysis the 162 food items in the French food frequency questionnaire (FFQ) were arranged into groups that corresponded with those in the UK study. For example, the UK `Bread, rice, pasta and potatoes' group included the frequency of: white bread, wholegrain bread, other breads, pasta, rice, wholegrain rice, wholegrain pasta and potatoes. All the ®nal categories are described in Table 2. The response categories used for the FFQ in each country were regrouped to facilitate comparison. Three of the frequency responses used were identical, ie never, once a day and > once a day. The remaining two categories were regrouped as follows: < once a week (UK) 3 times a month=twice a month=once a month (French). The frequency category of 1 ± 6 times a week used was rather broad but was unavoidable and resulted from the original categories used in both studies (1 ± 3 times a week=1 ± 6 times a week (UK) and once a week=2 ± 4 times a week (French). There was no response category for 5 ± 6 times a week in the French study. The frequencies of individual foods in the French study were summed to form the ®nal response categories by ®rstly converting the frequency of response into a function of the frequency of consumption per day. The original categories were therefore transformed as follows: European Journal of Clinical Nutrition A comparison of dietary behaviour M Holdsworth et al 532 never 0; once a month 1=30 (0.03); twice a month 2=30 (0.07); three times a month 3=30 (0.1); once a week 1=7 (0.14); 2 ± 4 times a week 3=7 (0.43); once a day 7=7 (1); and at every meal 163 (3). Frequency of food items were then summed to represent the total frequency of consumption of foods in the ®nal response categories. Responses were grouped so that ®nal response categories matched the original function of days, ie group 0 0; group 1 > 0 but 0.1; group 2 > 0.1 but < 1; group 3 1; group 4 > 1. The variables for vitamin and mineral supplement intake, breakfast cereal consumption, type of spreading fat used and alcohol intake were recorded as follows: use of vitamin and=or mineral supplements at least once a month; subjects who did=did not eat cereal at breakfast; type of spreading fat with four possible responses of `butter', `polyunsaturated margarine', `hard margarine' or `low fat spread'; alcohol `drinkers' and `non-drinkers'; and number of units of alcohol drunk in a typical week. In both studies the following demographic variables were measured: age, gender, body mass index (BMI), marital status and employment status. Age was reclassi®ed into three age groups of 20 ± 34 y; 35 ± 54 y and 55 ± 74 y. BMI was assessed from self-reported weight=height and data was classi®ed into four groups based on the Garrow scale (Garrow, 1983). Data analysis. Similarities and differences between France and the UK are highlighted and compared with dietary recommendations for the prevention of cancer (World Cancer Research fund=American Institute for Cancer Research Recommendations, 1997). The Mann ± Whitney U test was used to test the tendency for respondents in one country to exceed respondents in the other country (Bland, 1996). The chi-squared test for association=cross-tabulations was used to test the null hypothesis that there was no relationship between two independent variables within a country. Analyses by age and gender were performed. When a difference in frequency of consumption was found between the countries, data were analysed by age groups, to ensure the differences between the two countries did not result from a difference in age pro®le. Analysis by social class or educational level was not possible as we were unable to form comparable de®nitions of social class or educational level. Developing dietary indices. In addition to describing dietary behaviour in terms of frequency of consumption, indices were developed to facilitate a description of eating behaviour in broader terms. The purpose of the indices was to summarise the data more succinctly and to compare the ®ndings with dietary recommendations for the prevention of cancer. Indices were developed incorporating similar principles to those used previously in three studies (Smith & Smith, 1994; Davenport et al, 1995; Dowler & Calvert, 1995). Indices were developed for fat, dietary ®bre, fruit and vegetables and meat. These were chosen as they relate to current cancer recommendations. The foods contributing to the indices were: Fat index processed meats; chicken, turkey and ®sh; beef, lamb and pork; cheese; sweets= chocolates; sweet biscuits, cakes, pies, puddings and pastries; crisps, fried snacks and peanuts; Dietary ®bre European Journal of Clinical Nutrition index beans and pulses; fresh=stewed fruit; vegetables (fresh= frozen) or salad; tinned fruit and vegetables; cereals and potatoes; Fruit and vegetable index fresh=stewed fruit; vegetables (fresh=frozen) or salad; tinned fruit and vegetables; Meat index processed meats; chicken, turkey and ®sh; beef, lamb and pork. A maximum of two points were given for a response that met the nutrition guidelines based on recommendations in the World Cancer Research Fund Report (1997). For example, two points were given if the response met current guidelines, one point was given when a response indicated a shift towards healthier eating but not reaching the guidelines, and no points were given if the response indicated that the individual had a long way to go before meeting current guidelines. Foods that current guidelines recommend decreasing (processed meats, cheese, sweets and chocolate, puddings=desserts, crisps, fried snacks and peanuts and beef, pork and lamb) were all given two points if eaten < once a week=never; 1 point if eaten 1 ± 6 times a week and zero points if eaten once a day. For foods that current guidelines recommend increasing (fresh=stewed fruit, vegetables (fresh=frozen) or salad, tinned fruit and vegetables, chicken and ®sh, and lastly cereals and potatoes), zero points were given if eaten < once a week=never, 1 point was given if eaten 1 ± 6 times a week and 2 points if eaten once a day. The exception was for beans and pulses, which were considered to have met the guideline if eaten 1 ± 6 times a week (2 points allocated), whereas zero points were given if they were eaten less frequently. For ease of presentation and interpretation, the scales were adjusted so that a higher score indicates eating in line with current guidelines for cancer prevention. Indices were investigated within socio-demographic groups. Indices were transformed into continuous variables by scaling them up to 100, which meant that indices could be compared with each other. Results Sociodemographic pro®le Demographic characteristics are shown in Table 1. The UK sample comprised more women compared with the French sample, which had equal representation. Around a ®fth of both samples were aged less than 34 y. The French sample comprised more subjects in the > 54 age group compared with the UK sample, re¯ecting the ®nding that there were almost twice as many retired respondents in the French sample. In terms of BMI, there were more ideal weight respondents (41.6%) in the French sample compared with the UK sample, whereas there were more overweight=obese respondents (41.6%) in the UK sample compared with 35.7% in the French sample. However, these differences in BMI pro®le did not reach signi®cance (P 0.068). Almost three-quarters of the UK sample was working compared with just over half of French respondents. Unemployed respondents and students=those not seeking work were similarly represented in both surveys. Dietary behaviour The frequency of consumption of the food groups described above are illustrated in Table 2. Fruit and vegetables. The French sample ate signi®cantly more fruit and vegetables than the UK sample (P 0.000). A comparison of dietary behaviour M Holdsworth et al Table 1 Sociodemographic characteristics of the UK and French samples Dietary dilemmas (UK) MEDHEA (France) Age 20 ± 34 y 35 ± 54 y 55 ± 74 y Gender Females Males BMI statusa Group 0 (underweight) Group 1 (ideal weight) Group 2 (overweight) Group 3 (obese) Employment status Working Unemployed Retired Students=not working but not seeking work n % n % 91 217 110 21.8 51.9 26.3 124 237 274 19.5 37.3 43.1 242 176 57.9 42.1 318 317 50.1 49.9 22 221 139 34 5.3 53.1 33.4 8.2 37 371 195 32 5.8 58.4 30.7 5.0 291 18 52 46 71.5 4.4 12.8 11.0 338 25 194 78 53.2 3.9 30.6 12.3 a Where Group 0 (underweight) BMI < 20; group 1 (ideal weight) BMI 20 ± 25; group 2 (overweight) BMI 26 and < 30; group 3 (obese) BMI 30. Over 90% of the French sample ate fruit more than once a day, compared with only a ®fth of UK respondents. The difference was even more marked for vegetables. However, UK respondents were signi®cantly more likely to eat tinned fruit and vegetables (P 0.000). Sociodemographic differences. Older respondents ate fresh vegetables=salad and fresh=stewed fruit more often than younger respondents in both countries (Table 3). These trends reached signi®cance when consumption was grouped into respondents consuming them at least once a day and those consuming less than once a day. In the UK this reached signi®cance for fresh vegetables=salad (w2 7.31df 2; P 0.026) and fresh=stewed fruit (w2 12.56df 2; P 0.051, Table 3). Similarly in France, this reached signi®cance for both fresh vegetables=salad (w2 7.98df 2; P 0.018) and fresh=stewed fruit (w2 19.12df 2; P 0.000). Analysis of consumption of fresh vegetables=salad and fresh=stewed fruit by age group illustrated that the differences between the two countries were not due to an age effect, as French respondents in all age groups had signi®cantly higher consumption of fresh vegetables=salad and fresh=stewed fruit (results not shown). In both countries, females ate fruit and vegetables (except tinned) more often than males. In France, this reached signi®cance for fresh vegetables=salad (w2 4.64df 1; P 0.031) with 95.0% (n 302) females compared with 90.5% (n 287) males being high consumers. In addition, 95.9% (n 305) of females compared with 90.5% (n 287) males were high consumers (at least once a day) of fresh=stewed fruit (w2 7.27df 1; P 0.007). In the UK these gender differences in consumption of fresh vegetables=salad and fresh=stewed fruit were also signi®cant, with 71.8% (n 173) of females compared with 52.9% (n 91) of males being moderate consumers (1 ± 6 times=week was the highest response modality) of fresh vegetables=salad (w2 15.51df 1; P 0.000). Half of females (49.3%; n 134) were high consumers (at least once a day) of fresh=stewed fruit compared to 40.5% (n 68) of males (w2 15.67df 3; P 0.001). 533 Poultry and ®sh. The French sample ate poultry and ®sh more often than the UK sample (P 0.000), as 11.7% of the French sample ate these foods at least once a day compared with 1.9% of the UK sample. Sociodemographic differences. No difference in consumption by age or gender was detected in either country. Red meat. The French sample ate red meat more often than the UK sample (P 0.001), with 28.0% of the French sample and 1.9% of the UK sample eating red meat at least once a day. Sociodemographic differences. In both countries, men ate red meat more often than women. In France, 18.6% (n 59) of females compared with 28.1% (n 89) of males reported eating red meat at least once a day (w2 9.24df 2; P 0.010). The difference in gender also reached signi®cance in the UK, with 67.4% (n 163) of females compared with 81.8% (n 144) of males eating red meat 1 ± 6 times a week (w2 12.10df 2; P 0.020). Processed meats. French respondents ate processed meats more often than their UK counterparts (P 0.000), with 67.0% eating processed meats at least once a week compared with 45.4% of the UK sample. Sociodemographic differences. In both countries, men ate processed meats more often than women. In France, 15.8% (n 50) of males compared with 4.4% (n 14) of females reported eating processed meat at least once a day (w2 35.14df 3; P 0.000). The difference in gender also reached signi®cance in the UK, with 61.9% (n 109) of males compared with 33.5% (n 81) of females eating processed meat 1 ± 6 times a week (w2 36.94df 2; P 0.000). In the French sample the under 34s (15.3%; n 19) were more likely to eat processed meats than the older age groups (35 ± 54 y, 7.6%; n 18; > 54 y, 9.9%; n 27); however this did not reach signi®cance. Beans and pulses. Beans and pulses were eaten more often in the UK sample (P 0.000), with 71.5% eating beans and pulses at least once a week compared with 48.8% of the French sample. Sociodemographic differences. In France, men ate beans and pulses more often than women, although this did not reach signi®cance, with 51.7% (n 164) of males compared to 45.9% (n 146) of females reporting eating beans and pulses at least once a day (w2 2.84df 2; P 0.242). There was no trend for gender in the UK or for age in either country. Cereals and potatoes. The vast majority of the French sample (95.4%) ate bread, potatoes, pasta or rice more than once a day compared with under half (46.9%) of the UK European Journal of Clinical Nutrition A comparison of dietary behaviour M Holdsworth et al 534 Table 2 A comparison of food frequency between French and UK respondents Dietary Dilemmas (UK) Fresh or stewed fruit Never 0 less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Salad or vegetables (fresh or frozen) Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Tinned fruit and vegetables Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Chicken, turkey and ®sh Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Beef, lamb, pork Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Processed meats Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Beans and pulses Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Cereals and potatoes Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Sweet biscuits, cakes, pies, puddings or pastries Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Sweets or chocolates Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Crisps, fried snacks, peanuts Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 Cheese Never 0 Less than once a week 1 1 ± 6 times a week 2 Once a day 3 > Once a day 4 European Journal of Clinical Nutrition n % 9 33 174 111 91 2.2 7.9 41.6 26.6 21.8 5 149 264 0 0 1.2 35.6 63.2 Ð Ð 84 191 133 10 0 20.1 45.7 31.8 2.4 Ð 16 35 359 7 1 3.8 8.4 85.9 1.7 0.2 33 78 299 0 8 7.9 18.7 71.5 Ð 1.9 70 158 187 3 0 16.7 37.8 44.7 0.7 Ð 23 96 290 8 1 5.5 23.0 69.4 1.9 0.2 0 1 89 132 196 Ð 0.2 21.3 31.6 46.9 11 84 174 5 144 2.6 20.1 41.6 1.2 34.4 33 177 165 37 6 7.9 42.3 39.5 8.9 1.4 52 144 185 30 7 12.4 34.4 44.3 7.2 1.7 36 69 272 39 2 8.6 16.5 65.1 9.3 0.5 MEDHEA (France) Median 2 2 1 2 2 1 2 3 2 1 2 2 n % 9 4 30 4 588 1.4 0.6 4.7 0.6 92.6 1 2 43 0 589 0.2 0.3 6.8 0 92.8 280 182 160 8 5 44.1 28.7 25.2 1.3 0.8 1 34 526 45 29 0.2 5.4 82.8 7.1 4.6 4 18 429 32 152 0.6 2.8 67.6 5.0 23.9 78 131 383 4 39 12.3 20.6 60.3 0.6 6.1 57 268 307 0 3 9.0 42.2 48.3 Ð 0.5 0 0 27 2 606 Ð Ð 4.3 0.3 95.4 80 144 214 33 164 12.6 22.7 33.7 5.2 25.8 363 65 81 73 53 57.2 10.2 12.8 11.5 8.3 188 103 269 50 25 29.6 16.2 42.4 7.9 3.9 20 23 284 36 272 3.1 3.6 44.7 5.7 42.8 Median P value (Mann ± Whitney) 4 0.000 4 0.000 1 0.000 2 0.000 2 0.001 2 0.000 1 0.000 4 0.000 2 0.000 0 0.000 2 0.111 2 0.000 A comparison of dietary behaviour M Holdsworth et al 535 Table 3 Differences between French and UK respondents in fruit and vegetable consumption by age Dietary Dilemmas (UK) Salad or vegetables (fresh or frozen) Never=less than once a week 1 ± 6 times a week > Once a day Fresh=stewed fruit Never=less than once a week 1 ± 6 times a week Once a day > Once a day MEDHEA (France) < 34 y 35 ± 54 y > 54 y < 34 y 35 ± 54 y > 54 y n % n % n % 41 46.1 48 53.9 Ð 78 36.3 137 63.7 Ð 30 27.5 79 72.5 Ð 1 0.8 19 15.3 104 83.9 1 0.4 14 5.9 222 93.7 1 0.4 10 3.6 263 96.0 n % n % n % n % 13 14.6 38 42.7 23 25.8 15 16.9 15 7.0 98 46.0 55 25.8 45 21.1 5 4.7 38 35.5 33 30.8 31 29.0 4 3.2 9 7.3 1 0.8 110 88.7 5 2.1 15 6.3 1 0.4 216 91.1 4 1.5 6 2.2 2 0.7 262 95.6 sample. The difference between the two samples reached signi®cance (P 0.000). Sociodemographic differences. In the UK, younger respondents ate cereals and potatoes less often than older respondents (w2 12.28df 4; P 0.015). Around a third of under 34s (38.5%; n 35) ate cereals and potatoes more than once a day compared with 43.5% (n 95) of those aged 35 ± 54 y and 60.0% (n 66) of those aged > 54 y. In France, there was no difference in starchy food consumption by age. Sweet biscuits, cakes, pies, puddings or pastries. Over a third (34.4%) of the UK sample reported eating these foods more than once a day compared with 25.8% of those in the French sample. The difference between the two countries reached signi®cance (P 0.000). Sociodemographic differences. In both countries, men ate these foods more often than women. In France, 29.3% (n 93) of males compared with 23.6% (n 75) of females ate these foods daily, although this did not reach signi®cance (w2 5.44df 3; P 0.143). However the difference in gender reached signi®cance in the UK, with 42.0% (n 74) of males compared with 31.0% (n 75) females eating these foods daily (w2 7.95df 3; P 0.047). Sweets or chocolate. Over half (57.2%) of the French sample did not eat sweets or chocolate compared with only 7.9% of UK respondents (P 0.000). However there were more `high consumers' in France, with 19.8% of the sample eating sweets or chocolate at least once a day compared with 10.3% of UK respondents. Sociodemographic differences. There was no trend for age in France but in the UK the younger age group ate more sweets and chocolate, as 16.5% (n 15) of < 35s ate sweets and chocolate daily compared with 11.1% (n 24) of those aged 35 ± 54 and 3.6% (n 4) of > 54 y olds (w2 24.19df 6; P 0.000). Crisps, fried snacks or peanuts. These foods were eaten less often in the French sample, with 29.6% of respondents reporting that they never ate these foods compared to 12.4% of UK respondents (see Table 2). However, there was a similar distribution in frequency in both French and UK samples for consumption of crisps, fried snacks or peanuts at least once a week. The differences did not reach signi®cance (P 0.111). Sociodemographic differences. In both countries men and younger respondents ate crisps and fried snacks more often than their counterparts. In the UK, more men (14.8%; n 26) than women (4.5%; n 11) ate crisps=fried snacks daily (w2 15.46df 3; P 0.001). In France, the trend was similar with 14.8% (n 47) of men compared with 8.8% (n 28) of women eating crisps=fried snacks daily, although this did not reach signi®cance at the 5% level (w2 6.46df 3; P 0.091). In relation to age, < 35 y olds in the UK were almost twice (14.3%; n 13) as likely to eat crisps every day as those in the 35 ± 54 y age group (8.8%; n 19) and almost three times as likely as the over 54s (4.5%; n 5) (w2 35.07df 6; P 0.000). Similarly in the French sample, 16.9% (n 21) of < 35s ate crisps=fried snacks every day compared with 9.9% (n 27) of > 54 y olds (w2 32.66df 6; P 0.000). Cheese. Cheese was eaten more frequently in the French sample, with around half (48.5%) of French respondents eating cheese at least once a day compared to only 9.8% of UK respondents (P 0.000). Sociodemographic differences. In the French sample men ate cheese more often, as 54.3% (n 172) of men compared with 42.8% (n 136) of women reported eating cheese every day (w2 9.51df 3; P 0.023). In the UK there was no trend in cheese consumption with gender. Dietary indices Median dietary indices are compared in Table 4, indicating that the French sample scored signi®cantly better for fat, dietary ®bre, fruit and vegetables. Although median meat European Journal of Clinical Nutrition A comparison of dietary behaviour M Holdsworth et al 536 Table 4 A comparison of median dietary indices (of percentage) between French and UK respondents Dietary Dilemmas (UK) Fat index Fibre index Fruit and vegetable index Meat index MEDHEA (France) Median Interquartile range Median range Interquartile range P value (Mann ± Whitney) 57.14 60.00 33.33 50.00 50.00 ± 64.29 50.00 ± 70.00 33.33 ± 50.00 50.00 ± 66.67 64.30 70.00 66.67 50.00 57.14 ± 71.43 60.00 ± 80.00 66.67 ± 83.33 33.33 ± 66.67 0.000 0.000 0.000 0.000 A higher score indicates eating habits in line with current recommendations. indices were the same in each country, the inter quartile range in the French sample was larger, re¯ecting higher French intakes of red meat and processed meats. Use of vitamin and mineral supplements. Vitamin and mineral supplements were used twice as often in the UK sample (P 0.000), with 32.3% (n 135) of the UK sample taking them at least once a month compared with 15.3% (n 97) of the French sample. Sociodemographic differences. In both countries, women were more likely to use vitamin and mineral supplements than men. This was more marked in France, with 22.6% (n 72) of women compared with 7.9% (n 25) of men (w2 26.71df 1; P 0.000) in France and 39.3% (n 95) of women compared with 22.7% (n 40) of men (w2 12.73df 1; P 0.000) in the UK. In France, younger respondents ( 54 y) used vitamin and mineral supplements more than older respondents ( > 54 y), with 19.7% (n 71) of those aged 54 y using vitamin and mineral supplements compared with 9.5% (n 26) of those aged > 54 (w2 12.47df 1; P 0.000). In the UK sample the trend was in the opposite direction, as older respondents (> 54 y) were more likely to use vitamin and mineral supplements than younger respondents ( 54 y): 37.3% (n 41) of those aged > 54 y used supplements compared with 30.5% (n 94) of those aged 54 y; although this did not reach signi®cance. Breakfast cereals. These were eaten twice as often in the UK, with 51.2% of the UK sample eating them compared with 24.4% of the French sample. Sociodemographic differences In the French sample, younger respondents (37.1%, n 46 of < 35s and 16.4%, n 45 of > 54s) and females (28.0%, n 89 of women and 20.8%, n 66 of men) were most likely to eat breakfast cereals, which reached signi®cance for both age (w2 21.15df 2; P 0.000) and gender (w2 4.42df 2; P 0.036). However in the UK sample, there was no trend in consumption with age or gender. Type of spreading fat. The French sample (59.7%; n 379) used butter almost three times as often as the UK sample (19.5%; n 71). UK respondents preferred instead to use low fat spread (UK, 30.4% (n 111) of respondents vs France, 18.5% (n 117) of respondents), polyunsaturated margarine (UK, 36.7% of respondents vs France, 8.8% of respondents), and hard margarine (UK, 13.4% (n 49) of respondents vs France, 0.9% (n 6) of respondents). UK respondents (50.1%; n 183) were therefore ®ve times more likely to use margarine than French respondents (9.7%; n 62). Sociodemographic differences. In the UK butter was used signi®cantly more by older respondents (Table 5), whereas low fat spread was used less by older respondents ( > 54 y). In France, the trend was for butter to be used more by the younger age groups, whereas low fat spread was more likely to be used by the older respondents, although the latter did not reach signi®cance in the sample. In the French sample there were no age differences for consumption of hard margarine or polyunsaturated margarine. In both countries women were more likely to use low fat spread, which reached signi®cance in the UK (see Table 5). In the UK, men used butter more than women, although this did not reach signi®cance, but there was no gender difference for butter use in France. In both countries there was no Table 5 Gender and age differences for spreading fat use in French and UK respondents Gender Low fat spread UK France Butter UK France European Journal of Clinical Nutrition Age Men Women w2df 1 P-value < 34 y 35 ± 54 y > 54 y w2df 2 P-value n % n % 34 21.9 51 16.1 77 36.7 66 20.8 9.15 0.002 23 25.6 59 21.5 0.480 0.129 64 32.7 39 16.5 1.47 2.30 24 30.4 19 15.3 3.17 0.205 n % n % 35 22.6 190 59.9 36 17.1 189 59.4 1.68 0.195 27 30.0 145 52.9 0.010 0.897 34 17.3 151 63.7 9.27 0.02 10 12.7 83 66.9 9.52 0.090 A comparison of dietary behaviour M Holdsworth et al gender difference for consumption of polyunsaturated margarine. In the UK, men used hard margarine more than women, with 20.0% (n 31) of men compared with 8.6% (n 18) of women (w2 10.02df 1; P 0.002), this trend was not found in the French sample. Alcohol. A third (33.5%; n 140) of UK respondents reported not drinking alcohol compared with a quarter (23.8%; n 151) of the French sample. Almost threequarters (UK, 71.9%, n 200; France, 73.9%, n 355) of drinkers in each sample reported drinking less than 14 units a week. Sociodemographic differences. In both countries there were no signi®cant differences in age distribution of nondrinkers. Gender differences were marked in both countries, as signi®cantly more men than women were drinkers, with 76.7% (n 135) of UK men and 59.1% (n 143) of UK women (w2 14.19df 1; P 0.000) reporting they drank, compared with 83.6% (n 265) of French men and 68.9% (n 219) of French women (w2 19.00df 1; P 0.000). In both countries, signi®cantly (P 0.000) more men than women exceeded alcohol recommendations of > 14 units=week for women and > 21 units=week for men. This was more marked in the UK, where 34.1% (n 46) of UK men compared with 26.0% (n 69) of French men drank more than 21 units=week. The difference between women was not so marked with 7.7% (n 11) of UK women compared with 6.9% (n 15) of French women drinking, more than 14 units=week. In the UK, the younger two age groups ( 54 y) drank larger quantities than those aged > 54 y. Of those that drank, almost a quarter (21.5%; n 44) of 54 y olds compared with 9.6% (n 7) of those aged over 54 drank heavily (> 21 units=week) (w2 7.86df 3; P 0.049). In France, this trend was the reverse, as more older subjects reported drinking heavily. Of those that drank, almost a ®fth (17.0%; n 67) of 34 y olds compared with 7.8% (n 7) of those aged < 34 y drank heavily (> 21 units=week) (w2 9.98df 3; P 0.019). Discussion This study has highlighted that, although there are positive and negative features in food consumption in each country, the diet in the French Mediterranean is healthier overall as French respondents scored signi®cantly better for fat, dietary ®bre, fruit and vegetables. However, the French sample scored poorer for the meat index, re¯ecting higher French intakes of red meat and processed meats. On the positive side, the southern French ate signi®cantly more fruit, vegetables, cereals and potatoes, poultry and ®sh, and fewer sweets and chocolates, cakes, pastries, biscuits and puddings. The higher consumption of fruit and vegetables in the French sample was not a result of differences in the age pro®les of each sample. However, on the negative side they ate more cheese, processed meats, red meat and butter, all of which are rich sources of saturated fat. In addition, French respondents ate fewer beans and pulses. They also ate less breakfast cereals, but as data is not available on type, we are unable to say if they ate less of the higher ®bre=lower sugar varieties. Although the public health message to eat more breakfast cereal is common to both countries, the emphasis is different. In the UK, the population is encouraged to eat lower sugar=higher ®bre varieties, whereas the French population are encouraged to increase consumption of all breakfast cereals. On the other hand, UK respondents reported eating more beans and pulses, low fat spread, and less cheese, red meat and processed meats. However, on the negative side, they ate less fruit and vegetables, ®sh and poultry, cereals and potatoes and more sweets and chocolates, cakes, pastries, biscuits and puddings, crisps and fried snacks. This habit of snacking on energy-dense foods could contribute to the higher incidences of overweight and obesity seen in the UK sample. Data from the Food and Agricultural Organization from 10 y ago also suggested that the French diet contains more cereals, vegetables and ®sh than the UK diet (Hill and Caygill, 1994). Diets containing substantial amounts of fruit and vegetables may on their own reduce the overall incidence of cancer by over 20%. This is therefore enormous potential for improving diet-related cancer in the UK by focusing primarily on increasing fruit and vegetable intake. Cummings and Bingham (1998) have suggested a need for fruit and vegetable intake in the UK to at least double if the recommended consumption of non-starch polysaccharides is to be achieved. Recent campaigns (Health Education Authority, 1992; Sharp, 1997) are a start in raising public awareness, but it has been proposed that public health nutrition measures need to emphasize the potential for reduced cancer risk (Cox et al, 1996), which may be more motivating for individuals. The barriers to consuming more fruit and vegetables need to be addressed, particularly the cost of fruit and vegetable dishes. One of the main limitations of this study is that the UK food frequency data were already grouped, whereas in the French survey the consumption of individual food items was recorded. This introduces an error as frequency in the French sample may be overestimated as individuals may overestimate their consumption more when recording individual food items than a total (Nelson & Bingham, 1996). However, the converse may also be true as the French study (Bonifacj et al, 1997) found that the FFQ could underestimate consumption due to insuf®cient questions on some foods. Although there were more alcohol drinkers in the French sample, there was a similar distribution of units drunk in each country. For age, the trend was worrying in the UK from a public health perspective, as younger age groups drank more than older ones. This was the opposite in France, with older respondents taking larger quantities of alcohol, which re¯ects the traditional behaviour of regular drinking of wine with meals by older men. It would have been useful to have had data on daily alcohol consumption to compare intakes with current daily guides (Health Education Authority, 1996). Unfortunately, there were no data in the UK study on type of alcohol taken, hence a comparison of this aspect was not possible. UK respondents used vitamin and mineral supplements more than French respondents. They were primarily used by women in both countries, re¯ecting the fact that women tend to be more health conscious and possibly more in¯uenced by media messages. Although vitamin and mineral supplementation may be regarded as a healthy behaviour by the public, it is recommended that individuals choose a healthier diet rather than supplement their diets 537 European Journal of Clinical Nutrition A comparison of dietary behaviour M Holdsworth et al 538 arti®cially (Department of Health, 1991; Kirk et al, 1999). Again, age trends differed in both countries, with older respondents using supplements more in the UK, and younger respondents more in France. The Dietary and Nutritional survey of British adults (Gregory et al, 1990) also reported higher use of dietary supplements among women and older people. Although response rates were lower than anticipated, both samples were representative for age pro®le, BMI and eating behaviour. The UK sample was comparable with age pro®le statistics from the General Household Survey in Britain in 1994 (Of®ce of Population Censuses and Surveys, 1996a). The weight pro®le of the sample was comparable with the UK average (Of®ce of Population Censuses and Surveys, 1996b) for mean BMI for men (UK mean 25.8 kg=m2; study mean 25.53 kg=m2). Mean BMI for women in the UK study sample was, however, lower than the UK average for women (UK mean 25.5 kg=m2; study mean 24.08 kg=m2). There was no particular regional variation in eating habits in the area of study that made it atypical of the UK (Ministry of Agriculture, Fisheries and Food, 1992). In the French study the sampling technique was designed so that a balance of three equal age groups were represented, which resulted in a slight over-representation of older subjects compared to the age distribution of the population of the region. The BMI pro®le of the sample was comparable with the national distribution of the French population. There was no particular regional variation in eating habits in the area of study that made it atypical of the French southern region. We support the recommendations made by Assmann et al (1998) to adopt a Mediterranean style diet in northern European countries and to encourage the preservation of the traditional Mediterranean diet in countries that already have this diet. This is not a new idea and was proposed as far back as 1614 (Haber, 1997). Such a diet could address consumer concerns about taste and palatability (eg Glanz et al, 1998), however, any recommendations need to be culturally acceptable, affordable and convenient. The traditional Mediterranean diet necessitates a large variety of fresh products, some of which are expensive and time consuming to prepare (Gerber & Corpet, 1998). Effective strategies for health promotion need to be developed that promote the positive aspects of eating patterns in France and the UK. As already stated these need to focus on increasing fruit and vegetable consumption in the UK, as well as increasing consumption of ®sh and poultry and cereals. The increasing use of olive oil in the UK (International Olive Oil Council, 1998) is encouraging, but needs further promotion and preferably subsidizing on a national level so that it is affordable to all socioeconomic groups. Unfortunately there were no data from the UK study on olive oil use. The British habit of inbetween meal snacking on energy dense foods is a large contributor to saturated fat intake. Encouraging fruit as a snack is the ideal solution, but a realistic approach needs to acknowledge the place of these foods in the UK diet. Encouraging food manufacturers to use monounsaturated fat (in the form of olive oil preferably) in place of saturated fat in these foods would contribute to improving the fatty acid pro®le of the average British diet. For the French Mediterranean population it is essential to promote the continuation of good dietary behaviour. Increased consumption of beans and pulses should be European Journal of Clinical Nutrition encouraged, particularly in traditional dishes such as casseroles, soups and salads. A reduction in consumption of saturated fat could be achieved by discouraging high consumption of processed meats, cheese and butter. Having investigated the differences in food consumption between the two countries, the next logical step seems to be to investigate the differences in attitudes to food and cultural in¯uences on food choice between the UK and Mediterranean France. This is essential in the development of appealing nutrition education messages that are both culturally sensitive and affordable. Acknowledgements ÐThe Dietary Dilemmas study was funded from a grant from the Leverhulme Trust. The MEDHEA study was ®nancially supported by the ArdeÁche and HeÂrault Committees of the `Ligue contre le cancer', the Fondation PreÂvot and the Regional Council of LanguedocRoussillion. References Askegaard S (1993): A European analysis of selected food consumption statements. Adv. Consumer Res. 20, 1 ± 7. Assmann G, de Backer G, Bagnara S, Betteridge J, Crepaldi G, FernendezCruz A et al (1998): International Consensus Statement: Olive oil and the Mediterranean Diet: Implications for health in Europe. Olive oil Medical Information Newsletter January, 20 ± 24. Bland M (1996): An Introduction to Medical Statistics. New York: Oxford Medical Publications. Bonifacj C, Gerber M, Scali J & Daures JP (1997): Comparison of dietary assessment methods in a southern French population: use of weighed records, estimated-diet records and a food-frequency questionnaire. Eur. J. Clin. Nutr. 51, 217 ± 231. Corpet DE & Gerber M (1997): Alimentation meÂditerraneÂenne et SanteÂ. IcaracteÂristiques. Maladies cardio-vasculaires et autres affections. MeÂd. Nutr. 4, 129 ± 142. Cox DN, Anderson A, McKellar S, Reynolds J, Lean MEJ & Mela DJ (1996): Vegetables and fruits: barriers and opportunities for greater consumption. Nutr. Food Sci. 5, 44 ± 47. Cummings JH & Bingham SA (1998): Diet and the prevention of cancer. Br. Med. J. 317, 1636 ± 1640. Davenport M, Roderick P, Elliott L, Victor C & Geissler C (1995): Monitoring dietary change in populations and the need for speci®c food targets; lessons from the North West Thames Regional Health Survey. J. Hum. Nutr. Diet 8, 119 ± 128. Department of Health (1991): Dietary Reference Values for Food Energy and Nutrients in the UK. London: HMSO. Dowler E & Calvert C (1995): Nutrition and Diet in Lone Parent Families in London. London: Family Policy Studies Centre. Garrow JS (1983): Indices of adiposity. Nutr. Abstr. Rev. 53, 697 ± 708. Gerber M (1999): Les beÂne®ces sante de l'alimentation meÂditerraneÂennne. Etat de l'Art. VIII eÁme entretiens d'Agropolis, Montpellier. Gerber M & Corpet DE (1997): Alimentation meÂditerraneÂenne et SanteÂ. IIcancers. MeÂd. Nutr. 4, 143 ± 154. Gerber M & Corpet D (1998): Food, lifestyle and cardio-vascular disease in Europe. Riv. Antropologia 76, 419 ± 430. Gerber M, Scali J, Avallone MH et al (1997): A nutritional survey in Mediterranean countries. Reprod. Nutr. Devl. 37, 373 ± 374. Gerber M, Siari S, Michaud A & Scali J (1999): Alimentation meÂditerraneÂennne et santeÂ. MEDHEA, les reÂsultats de l'HeÂrault. ActualiteÂs DieÂteÂtique 35, 1391 ± 1396. Gibney MJ (1997): Eur. J. Clin. Nutr. 51(Suppl), S1 ± S59. Glanz K, Basil M, Maibach E, Goldberg J & Snyder D (1998): Why Americans eat what they do: taste, nutrition, cost, convenience, and weight control concerns as in¯uences on food consumption. J. Am. Diet. Assoc. 98, 1118 ± 1126. Goode J, Beardsworth A, Haslam C, Keil T & Sherratt E (1995): Dietary Dilemmas: nutritional concerns of the 1990's. Br. Food J. 97, 3 ± 12. Goode J, Beardsworth A, Keil T, Sherratt E & Haslam C (1996): Changing the nation's diet: a study of responses to current nutrition guidelines. Health Educ. J. 55, 285 ± 299. Haber B (1997): The Mediterranean diet: a view from history. Am. J. Clin. Nutr. 66, 1053S ± 1057S. Health Education Authority (1992): Enjoy Fruit and Vegetables Campaign. London: HEA. A comparison of dietary behaviour M Holdsworth et al Health Education Authority (1996): Think about Drink Ð There's More to Drink Than you Think. London: HEA. Hill MJ & Caygill CPJ (1994): Diet in Europe Ð the general picture. In: Epidemiology of Diet and Cancer, ed, MJ Hill, A Giacosa & CPJ Caygill. Chichester: Ellis Horwood Press, pp 189 ± 199. Holdsworth M & Spalding D (1997): The development, implementation, monitoring and evaluation of a food and nutrition policy within a local health plan. J. Hum. Nutr. Diet. 10, 209 ± 217. Holdsworth M, Haslam C, Raymond NT & Leibovici D (1997): An evaluation of the customers' perspective of the HBA scheme in public eating places. J. Nutr. Educ. 29, 231 ± 236. International Olive Oil Council (1998): International Olive Oil Council Information Sheet. 2. James WPT (1995): Nutrition Science and Policy Research: Implications for Mediterranean diets. Am. J. Clin. Nutr. 61, 1324S ± 1328S. Kearney M, Gibney MJ, Martinez JA, de Almeida MDV, Freibe D, Zunft HJF, Widhalm K & Kearney JM (1997): Perceived need to alter eating habits among representative samples of adults from all member states of the European Union. Eur. J. Clin. Nutr. 51, S30 ± S35. Kemm KR & Booth D (1992): Promotion of Healthier Eating. How to Collect and use Information for Planning, Monitoring and Evaluation. London, HMSO. Kirk SFL, Cade JE, Barrett JH & Conner M (1999): Diet and lifestyle characteristics associated with dietary supplement use in women. Public Health Nutr. 2, 69 ± 73. LennernaÈs M, Fjellstrom C, Becker W, Giachetti I, Schmitt A, Remaut de Winter AM & Kearney M (1997): In¯uences on food choice perceived to be important by nationally-representation samples of adults in the European Union. Eur. J. Clin. Nutr. 51, S8 ± S15. Mennell S (1996): Eating and Taste in England and France from the Middle Ages to the Present, 2nd edn. Illinois Press. Ministry of Agriculture, Fisheries and Food (1992): Household Food Consumption and Expenditure. London: HMSO. Nelson M & Bingham S (1997): Assessment of food consumption and food intake. In: Design Concepts in Nutritional Epidemiology, ed. BM Margetts & M Nelson, 2nd edn. Oxford: Oxford University Press, pp 123 ± 169. Of®ce of Population Censuses and Surveys (1996a) Living in Britain. Results from the 1994 General Household Survey. London: HMSO. Of®ce of Population Censuses and Surveys (1996b) Health Survey for England 1994. A Survey Carried out by the Social Survey Division of OPCS. London: HMSO. Sharp I (1997): At Least Five a Day: Strategies to Increase Fruit and Vegetable Consumption. London: HMSO. Smith AM & Smith C (1994): Dietary intake and lifestyle patterns: correlates with socio-economic, demographic and environmental factors. J. Hum. Nutr. Diet. 7, 283 ± 294. Wardle J & Solomons W (1994): Naughty but nice: a laboratory study of health information and food preferences in a community sample. Health Psychol. 13, 180 ± 183. WHO (1996): Support to food and nutrition policy making. Nutr. Pol. Infant Feeding Food Security July, 3 ± 5. World Cancer Research Fund (1997): Food, nutrition and the prevention of cancer: a Global perspective. Washington: American Insitute for Cancer Research. 539 European Journal of Clinical Nutrition