Dietary glycaemic index and glycaemic load in the European Prospective Investigation into Cancer and Nutrition



To describe dietary glycaemic index (GI) and glycaemic load (GL) values in the population participating in the European Prospective Investigation into Cancer and Nutrition (EPIC) study according to food groups, nutrients and lifestyle characteristics.


Single 24-h dietary recalls (24-HDRs) from 33 566 subjects were used to calculate dietary GI and GL, and an ad hoc database was created as the main reference source. Mean GI and GL intakes were adjusted for age, total energy intake, height and weight, and were weighted by season and day of recall.


GI was the lowest in Spain and Germany, and highest in the Netherlands, United Kingdom and Denmark for both genders. In men, GL was the lowest in Spain and Germany and highest in Italy, whereas in women, it was the lowest in Spain and Greece and highest in the UK health-conscious cohort. Bread was the largest contributor to GL in all centres (15–45%), but it also showed the largest inter-individual variation. GL, but not GI, tended to be lower in the highest body mass index category in both genders. GI was positively correlated with starch and intakes of bread and potatoes, whereas it was correlated negatively with intakes of sugar, fruit and dairy products. GL was positively correlated with all carbohydrate components and intakes of cereals, whereas it was negatively correlated with fat and alcohol and with intakes of wine, with large variations across countries.


GI means varied modestly across countries and genders, whereas GL means varied more, but it may possibly act as a surrogate of carbohydrate intake.


Carbohydrates are traditionally classified according to their saccharide chain length as ‘simple sugar’ or ‘complex carbohydrate’. However, Jenkins et al. (1981) developed a more physiological classification on the basis of post-prandial glycaemia. The so-called glycaemic index (GI) of a food is determined by comparing blood glucose response to ingestion of 50 g available carbohydrate (i.e., total carbohydrate minus fibre) from a test food with that of a reference food (either glucose or bread). The GI ranks carbohydrates in foods on a scale of 0–100 according to this blood glucose response, with <55 considered low, 56–69 considered medium and >70 considered high. Later, the term ‘glycaemic load’ (GL) was introduced, which takes into account differences in portion sizes of carbohydrate-containing foods, meals or diets. Diets based on low-GI carbohydrates have been shown to improve (serum) triglyceride levels, total cholesterol levels and the ratio of low- to high-density lipid proteins (Jenkins et al., 1987; Riccardi et al., 2008), whereas hyperglycaemia, hyperinsulinaemia, as well as faster digestion and absorption are characteristic post-prandial effects of high-GI foods (Wolever, 2000). A high habitual dietary GI seems to favour overweight (Ludwig, 2003), and direct associations between high GI and/or GL have been found for different types of diseases, including non-insulin-dependent diabetes mellitus, obesity, cardiovascular disease and cancer (Du et al., 2006; Barclay et al., 2008), but results are still contradictory. It seems that for certain diseases, the association with dietary GL is only present among certain subgroups, such as overweight (Liu et al., 2002; Beulens et al., 2007) and sedentary individuals (Michaud et al., 2002) or in pre-menopausal women with a body mass index (BMI) >25 kg/m2 (Cho et al., 2003).

So far, GI and GL values have been difficult to compare across studies and populations because of differences in the application of the GI database and because of a lack of comparability and detail in dietary data. In addition, the correlates of GI and GL with foods, nutrients and lifestyle characteristics have been studied only once in an American population (Schulz et al., 2005).

To help understand the potential health impact of dietary GI and GL, we endeavoured to gain an insight into their distribution across various lifestyle strata, and their correlation with foods and nutrients in a free-living population, using a unique dataset of highly detailed, standardized 24-h dietary recalls (24-HDRs) from the European Prospective Investigation into Cancer and Nutrition (EPIC) study and a recently developed ad hoc standardized GI and GL database (van Bakel et al., 2009).

Materials and methods

The EPIC study is a prospective cohort study originally involving 519 978 subjects in 23 centres in 10 western European countries. All study subjects filled out a country- and/or centre-specific dietary questionnaire and, to calibrate these data, a subsample of the cohort was approached for a single 24-HDR. This calibration study was undertaken between 1995 and 2000 to improve the comparability of dietary data across centres and to correct partially for dietary measurement error caused by centre-specific bias and random and systematic within-person errors (Willett, 1998; Ferrari et al., 2004).

Study population

This study involved the population subsample that completed a 24-HDR interview for calibration purposes in the EPIC study. It was based on an age- and gender-stratified random sample of 5–12% (United Kingdom 1.5%) obtained from each of the EPIC cohorts, weighted according to the cumulative numbers of cancer cases expected over 10 years of follow-up per gender and 5-year age strata. Most centres were population based, but some included particular populations, such as blood donors (centres in Italy and Spain), subjects in cancer screening programmes (Utrecht, The Netherlands) and a ‘health-conscious’ cohort, mainly lacto-ovo vegetarians or vegans, recruited by the Oxford centre in the United Kingdom (Riboli and Kaaks, 1997). Men and women were represented in most centres, except in France, Norway, Utrecht (The Netherlands) and Naples (Italy), from where only women were recruited (Slimani et al., 2002).

The initial number of subjects was 36 994, which was reduced to 36 034 after the exclusion of subjects below 35 or above 74 years of age. For the present analyses, 2468 diabetic subjects were also excluded (6.8%, too few to perform stratified analyses), because they might have changed their diet in favour of low-carbohydrate foods, leaving 11 978 eligible men and 21 588 women.

Dietary data and non-dietary variables

A single 24-HDR was obtained by face-to-face interview, except in Norway where it was obtained by telephone, a validated alternative approach (Brustad et al., 2003). Data were collected between March 1995 and June 2000. The interviewers used EPIC-SOFT, a specially designed computer programme, to enter dietary data in a standardized manner with several control steps, for example, with regard to food portion sizes and missing quantities (Slimani et al., 1999). Nutrient values were derived from the ENDB (EPIC Nutrient Database) and were thus standardized among the participating countries (Deharveng et al., 1999; Slimani et al., 2007). Missing carbohydrate components in the original national datasets (e.g., Sweden) or definitions that differed from the reference ENDB definitions (e.g., Denmark) were corrected in the final national datasets (Slimani et al., 2007). The definitions of all nutrients (including carbohydrates) and the methods used to determine their values in each centre and standardize values across centres have also been described in detail (Slimani et al., 2007).

Data on other lifestyle factors, including education level, total physical activity and smoking history considered in this analysis, were collected at baseline through standardized questionnaires and clinical examinations, and have been described for the calibration sample elsewhere (Haftenberger et al., 2000a, 2000b; Riboli et al., 2002). Data on age as well as body weight and height were self-reported by participants during the 24-HDR interview. The mean time interval between these baseline questionnaire measures and the 24-HDR interview varied by country, from 1 day to 3 years later (Slimani et al., 2002).

Calculation of dietary GI and GL

Glycaemic index values with glucose as the reference scale were assigned to the items reported in the 24-HDRs in a standardized manner as described in detail elsewhere (van Bakel et al., 2009). Briefly, foods reported in the 24-HDRs were selected on the basis of the GI value of the food while considering aspects of the food that might influence GI (e.g., cooking method, preservation method, sugar content and country-specific types of food). GI values obtained from the Foster-Powell table (Foster-Powell et al., 2002), British values (Henry et al., 2005), internet updates ( and some communicated values from GI experts (J Brand-Miller and T Wolever, personal communication) were then assigned to individual food items. Food items containing no or negligible amounts of carbohydrates or that do not increase blood glucose levels (chiefly meat and fish, fats, eggs) were not assigned a value. Average dietary GL was calculated by adding up the products of digestible carbohydrate of each food (quantity per day) and its GI. Average dietary GI was calculated as GL, but by dividing by the total amount of digestible carbohydrate in 1 day (Wolever et al., 1994).

Statistical analysis

The 23 original centres of the EPIC study were redefined into 27 centres and geographical regions as described elsewhere (Slimani et al., 2002). Data were presented by study centre or country, separately for men and women when justified, and ordered geographically on the basis of a south–north gradient.

Dietary GI and GL values were calculated as means and s.e. using generalized linear models. Minimally adjusted means were adjusted for age and weighted for day of the week and season to control for different distributions of participants across seasons and days of 24-HDR. We examined the independent effect of adjustment of several potential confounders—including height, weight, total energy intake, BMI, smoking status, highest education level and physical activity—on mean GI and GL values, centre rankings and on the R2 of the model as an estimation of the percentage variability of GI/GL that is explained by the potential confounder. In ‘fully adjusted’ models, we retained total energy intake (strong influence on mean GL), as well as height and weight (considered a priori potentially important predictors of carbohydrate intakes and thus influencing GL), in addition to weighting by season and day of the week. Although none of the confounders explained a sufficient variation for GI, for the sake of comparability, the same confounders as those for GL were retained. Separate analyses in which these confounders were not adjusted for showed no significant differences in the results.

Stratified analyses were carried out to determine differences in intakes of dietary GI/GL according to BMI category (<25, 2530, 30 kg/m2), smoking status (never, former, current), education level (primary, secondary/technical college, university), physical activity (inactive, moderately inactive, moderately active, active), season and day of the week of 24-HDR.

The relative contribution of individual food groups to overall dietary GL was calculated using a food group classification adapted from the original EPIC-SOFT programme (Slimani et al., 2000; Wirfalt et al., 2002) with more details at the subgroup level. The food groups ‘soups’ and ‘sauces’ were combined to ‘soups and sauces’ as few items had a GI value, whereas food groups with no GI values were excluded (e.g., eggs, fat).

A stepwise linear regression was performed to determine, within centres, which food groups contributed most to the inter-individual variation of dietary GI and GL. As only one 24-HDR was available per subject, conclusions on intakes at the individual level cannot be drawn, only insights into which food groups have the most effect when counselling on GI and GL reduction can be obtained. Furthermore, together with the results on the relative contribution of food groups to GL, linear regression can help to identify food groups for which it is important to analyse country-specific GI values to discriminate better between people. The model contained 25 food groups and age. The significance for entry into the model was set at P<0.1. The five food groups explaining most of the within-centre differences in dietary GI and GL were listed, as well as other food groups that contributed at least 1% of the total variation, the partial and model R2s.

The linear relationships between daily food group as well as nutrient intakes and dietary GI and GL were evaluated overall and at the country level in correlation analyses. As gender differences were minor, results were not presented separately for men and women. As the distributions of most food groups and nutrients were positively skewed, log-transformed values were used to calculate correlation coefficients (for food groups, 1 was added to each value to account for some individuals with zero intakes in these variables). Results were adjusted for gender, age and centre, and weighted for day of the week and seasons. Food group and nutrient intakes were adjusted for total energy intake using the residual method (Willett and Stampfer, 1986).

Analyses were performed using SAS (version 9.1, SAS Institute, Cary, NC, USA). A P-value <0.05 was considered statistically significant.


Mean dietary GI and GL

Mean dietary GI ranged from 54.0 in San Sebastian (Spain) to 59.3 in Bilthoven (The Netherlands) in men and from 52.4 in Asturias (Spain) to 57.3 in Bilthoven (The Netherlands) in women (Figures 1a and b and Table 1a). A full adjustment did not appreciably change these figures nor did it affect centre ranking. No clear north–south gradient could be distinguished, but the highest values were observed among the Danish, Dutch and British centres and the lowest among the Spanish centres for both genders.

Figure 1

Fully adjusted glycaemic index (GI) and glycaemic load (GL) means by country and centre (a) men and (b) women.

Table 1a Minimally and fully adjusted mean glycaemic index (GI)a by centre ordered from south to north and gender

The range of mean values for dietary GL was larger than for GI (Figures 1a and b and Table 1b). Again, no clear north–south gradient could be distinguished. For men, the lowest mean was found in Greece (119.8) and the highest in Ragusa, Italy (184.5). For women, the lowest values were also observed in Greece (88.7) and the highest in the UK health-conscious population (133.6). Full adjustment changed the ranking of some centres, particularly Italy, the United Kingdom and Greece. No relationship with age was observed for either of the two variables (results not shown).

Table 1b Minimal and fully adjusted mean glycaemic load (GL)a by centre ordered from south to north, and gender

Relative contribution of each food group to overall dietary GL

Of all the food groups, bread contributed the most to dietary GL in all centres and for both genders, ranging from 15% for both genders in Umeå (Sweden) to 40% in Aarhus (Denmark) for women and 45% in Navarra (Spain) for men (Figures 2a and b). Supplementary information is available on the EPIC website ( In women, the second highest contributor to dietary GL was fruit in the southern centres (except Varese and Naples), in the UK health-conscious cohort, in Germany and in South and East Norway (9–17%), whereas cakes scored second in the UK general population, in Bilthoven, Sweden, Aarhus and Varese (10–14%), potatoes in Utrecht, Copenhagen and in North and West Norway (9–11%) and pasta in Naples (12%).

Figure 2

Relative contribution (%) to overall dietary glycaemic load (GL) according to main GL-contributing food (sub) groups (a) men and (b) women.

In men, fruit was the second highest contributor in Greece and—except for Asturias—in Spain (11–15%). Pasta was the second highest contributor in the Italian centres (12–15%), potatoes in Asturias, Potsdam and in both centres in Sweden and Denmark (10–14%), sugar and confectionery in Bilthoven (13%) and cakes in both British centres and in Heidelberg (9–13%). Large differences were observed for the contribution of pasta, breakfast cereals, crispbread and vegetables. Among both alcoholic and non-alcoholic beverages, clear north–south gradients could be distinguished, with higher contributions in the northern than in the southern countries. Of the alcoholic beverages, beer was the main contributor to dietary GL but the contribution was significant only in men in northern European centres.

Stratified analyses according to lifestyle factors

None of the trends for adjusted dietary GI or GL differed statistically between BMI category, educational level, physical activity and smoking status, day of the week and season of the year (data not shown but available on the EPIC website: However, higher GI values were generally associated with higher physical activity and current smoking in men. Higher GL values were generally observed in subjects with lower BMI, in never-smokers of both genders and in physically active men with a technical school level education, but were lower in women with a university degree. Both UK centres showed a strong GL peak on Saturday. Across seasons, GI differences were observed for Greece and for the United Kingdom only, whereas GL differences were observed in the UK health-conscious cohort only.

Inter-individual variation of GI and GL

For both genders, bread explained most of the inter-individual variation in dietary GI and GL, especially in Southern Europe (Tables 2a and b). The inter-individual variation in both GI and GL was explained by fewer food groups in the southern than in the northern centres.

Table 2a The five principal food groups (log (foodgroup+1)) by gender contributinga to inter-individual variation in glycaemic index for most centres and deviating food groups by centre, identified by stepwise linear regression
Table 2b The five principal food groups (log(foodgroup+1)) by gender contributinga to inter-individual variation in glycaemic load for most centres and deviating food groups by centre, identified by stepwise linear regressiont

Individual correlations between dietary GI, GL and nutrients and food groups

Tables 3a and b show the crude and adjusted overall and adjusted country-specific linear Pearson's correlations of individual dietary GI and GL with nutrients, and Tables 4a and b show the crude and adjusted overall and adjusted country-specific linear Pearson's correlations of GI and GL with food groups, for men and women combined.

Table 3a Pearson's correlation coefficients between dietary glycaemic index (GI)a and nutrients in non-diabetic subjects
Table 3b Pearson's correlation coefficients between dietary glycaemic load (GL)a and nutrients in non-diabetic subjects
Table 4a Pearson's correlation coefficients between dietary glycaemic index (GI)a and food (sub)groups (log (food group+1) in non-diabetic subjects
Table 4b Pearson's correlation coefficients between dietary glycaemic load (GL)a and food (sub)groups (log (food group+1) in non-diabetic subjects


Except for starch, unadjusted dietary GI correlated very weakly with total energy, total carbohydrates, sugar (monosaccharides and disaccharides negatively), fat, protein, alcohol and fibre (Table 3a). After adjustment, only the correlation for starch remained positive, whereas lower sugar intakes were associated with increased GI. Across countries, starch and sugar correlates showed clear differences, but no evident north–south tendencies could be distinguished. Gender differences were minor.

A different picture was observed for the correlations between nutrients and dietary GL (Table 3b). All crude correlation coefficients, except for alcohol, were significantly positive. However, when total energy was taken into account, significant negative correlations were observed for fat and alcohol and, to a lesser extent, for protein. Across countries, clear differences were found for all correlates, with north–south tendencies for total carbohydrates, starch and alcohol. Marked differences in correlation coefficients between genders were found only in the UK health-conscious cohort for most nutrients (data not shown).

Food groups

Overall, higher GI values were related to higher intakes of white bread and potatoes, and to lower intakes of fruit, dairy products and legumes (Table 4a). These figures did not change after adjustment. Correlations with the other food groups were of minor importance. Across countries, adjusted correlates differed widely. The most interesting observations were the GI correlates for non-white bread, which were negative for Sweden (r=−0.18) and Germany (r=−0.16) but were positively correlated in all other countries. Gender differences were minor except in the UK general population for protein (men: −0.39, women: −0.01) (results not shown).

Increased GL values were related to higher intakes of cereal (products), in particular bread (irrespective of the type), sugar/honey/jam, and sweet buns/cakes/pies, but the correlation values were attenuated considerably after adjustment, eliminating cakes as an important correlate of dietary GL (Table 4b). For meat, fat and wine, correlations inversed, indicating that higher GL diets were lower in these foods under iso-energetic conditions.

At the country level, large differences were found for most food groups. Interesting observations were the absence of a correlation with bread in Germany (r=0.09) and Sweden (r=0.03), the fact that France scored the highest for the correlation with sugar and confectionery (r=0.29) and that Italy scored the lowest for fruits (r=0.11). Differences between genders were minor, apart from the UK health-conscious cohort for potatoes (men: 0.24, women: 0.42), fruits (men: −0.03, women: −0.28), bread (men: 0.11, women: 0.31) and breakfast cereal (men: 0.31, women: −0.02) (results not shown).


This study was undertaken to examine the complex associations of GI and GL with food groups, nutrients and lifestyle factors across free-living populations with large differences in dietary intakes.

Mean GI in both genders and GL values in women were in the same range as those recently reported in a meta-analysis by Barclay et al. (2008), but for men in more than half of the EPIC centres, mean GL values were higher than the range reported in that study, perhaps because the meta-analysis covered mostly studies of American origin. GL values for male subjects in Ragusa (Italy) exceeded the values for men in all other centres, which is probably mainly because of the high consumption of pasta. A similar pattern was observed for total carbohydrate intakes (Cust et al., 2009), which suggests that GL is a surrogate of carbohydrate intake. The correlation analysis between GL and total carbohydrate also showed a very high correlation for the unadjusted value, but this figure improved on adjustment for energy. In an earlier study by our group using dietary questionnaire data, the high correlation between these two variables could not be improved on adjustment for energy (van Bakel et al., 2009).

The contribution of some of the food groups to dietary GL gave a slightly different picture from that of carbohydrates (Cust et al., 2009). Bread was a more important contributor to overall dietary GL than to total carbohydrates because of its relatively high GI values, whereas fruit, pasta and dairy products contributed less because of their relatively low GI values (Foster-Powell et al., 2002). Food groups contributing principally sucrose as sugar, such as ‘sugar and confectionery’, ‘cakes’ and ‘carbonated drinks’, contributed fractions to overall GL similar to those for carbohydrates.

Bread contributed the most in terms of quantity to dietary GL, but was also the main determinant of inter-individual variation in dietary GI and GL. This knowledge can help dieticians when counselling on a diet low in GI and GL, and it shows that for this food it is important to further analyse country-specific GI values to improve the discrimination of dietary GI and GL between individuals.

From the stratified analyses it seemed that, in women, a higher GL corresponded to lower BMI (<25 kg/m2), higher education level and non-smoking, whereas in men it corresponded to lower BMI, higher physical activity, technical school education and non-smoking. Although high dietary GL has been associated with several diseases (Barclay et al., 2008), in our study, a higher GL was associated with a healthy BMI. Apparently women with a higher education have healthier lifestyles and generally follow dietary guidelines to consume more carbohydrates, which consequently leads to a high dietary GL, whereas in men GL seems to be more related to a physically active job. On the other hand, in men only, higher GI values were related to higher physical activity and current smoking status, partly confirming results from an earlier study that found an association between higher GI in subjects with no high school education and current smoking status (Schulz et al., 2005).

In both unadjusted and adjusted models, a high GI diet was characterized by high starch and low sugar intake at the overall European level, whereas there was no relation with total carbohydrate, protein, fibre and alcohol. Across countries, however, larger differences were found, particularly for fibre and alcohol; negative GI correlations were observed in Italy and Spain for fibre, and in the Netherlands, in the UK general population and Norway for alcohol. Apart from the inverse result for sugar that we shared with Schulz et al. (2005), and the absence of a relation with fibre found by Livesey et al. (2008) (although only true at the overall level), our results were quite different from those of earlier studies on nutrient correlates that reported inverse associations with total carbohydrates (Jonas et al., 2003; Scholl et al., 2004; Schulze et al., 2004; Schulz et al., 2005) and protein (Schulz et al., 2005). With regard to fat, some studies have found positive trends with total fat (Jonas et al., 2003; Scholl et al., 2004; Schulz et al., 2005), as well as a negative trend with animal fat (Buyken et al., 2001), across GI quintiles. The negative correlation with alcohol found by Sahyoun et al. (2008) was observed in our study at the country level only (The Netherlands, the UK general population and Norway) and thus seems to be a population-dependent food pattern effect.

In terms of foods, a higher intake of potatoes and white bread, which are rich in carbohydrates, was related to increasing GI, probably explaining the positive correlation of GI with starch, whereas a decreased intake of fruit and dairy products was related to increased GI, probably explaining the negative correlation with simple sugars present in these foods. The higher intakes of regular soft drinks and non- or low-carbohydrate foods (meat, fat, beer) and lower intakes of legumes are probably dietary food pattern effects, that is, people who consume potatoes and white bread generally also consume more meat, fat, beer and soft drinks and fewer legumes. Only in the northern European countries did we observe generally higher correlates for soft drinks, beer and dairy products. Similar results were found in the study by Schulz et al. (2005), except for legumes, which were negatively related to GI in their study. In addition, they found negative relations for adjusted GI with vegetables, pasta, fruit juices, fish and wine, and positive relations with eggs (Schulz et al., 2005), which we did not find at the overall level, although some of these correlations were observed at the country level. Our most surprising finding was the correlation with non-white bread, which was negative in Germany and Sweden, while being close to zero in Spain and in the UK health-conscious population and positive in other countries, with the highest correlation in Greece. Perhaps this is because of the fact that this food comprises relatively heterogeneous items, including wholemeal and brown breads, which contain very different amounts of wholemeal flour across countries. A possible explanation for the negative correlations in Germany and Sweden may be the prevalent cereal source and/or bread leavening processes in non-white breads (i.e., in rye and sourdough fermentation), which are both factors that markedly reduce the GI of bread (Liljeberg et al., 1995). However, non-white bread in Denmark is also mainly made from rye flour and sourdough.

In the unadjusted model, GL was highly correlated with proteins, lipids, fibre and total carbohydrate and with its individual components. After energy adjustment, the correlations with fat and protein inversed and became negative for alcohol, with the northern European countries generally having stronger negative correlations. This means that under iso-energetic conditions, high-GL diets are lower in fat, protein and alcohol. Although similar results for fat and protein were obtained earlier by Schulz et al. (2005), they reported a positive association with alcohol, whereas we, along with others (Salmeron et al., 1997; Michaud et al., 2002; Higginbotham et al., 2004; Schulze et al., 2004), found a negative relationship. However, this may be because of the fact that beer was assigned a hypothesized value of 95 and wine of 61 by Schulz et al. (2005), whereas we assigned a communicated value of 66 to beer and no value to wine, and in the other studies no value was assigned to alcoholic beverages.

In terms of food groups, GL is strictly correlated with carbohydrate-rich foods. Interestingly, after bread, the food group honey/sugar/jam correlated the most, although its contribution to total GL was only between 3.7 and 8.6%.

At present, no recommendations on dietary GI and GL exist. However, results from a recent meta-analysis have shown that an average dietary GI below 49 and an average dietary GL below 92 are associated with reduced risk of chronic disease (Barclay et al., 2008). At the population level, all EPIC centres exceeded these thresholds for GI and GL.

In conclusion, bread seemed to be the highest contributor to GL but it also contributed the most to the variation in dietary GL across subjects. The most stable contributor to GL across subjects was fruit. Both the relative contribution and daily variance provide important information when it comes to counselling on dietary GI and GL, that is, to opt for low GI foods within a high contributing group; for example, in the case of bread, choosing rye bread or sourdough-leavened bread and in the case of fruit, choosing apples or oranges instead of bananas. In addition, to improve discrimination between people's GI and GL, future foods to be analysed should be selected from these food groups.

As high GI/GL values are positively related to poor lipid and glycaemic blood profiles (Jenkins et al., 1987; Wolever, 2000; Riccardi et al., 2008), and may be related to different types of diseases (Du et al., 2006; Barclay et al., 2008), population GI/GL recommendations are urgently required. They may in turn have an influence on current carbohydrate recommendations, which do not differentiate between high- and low-GI carbohydrates.

Supplementary information

Supplementary information is available on the EPIC website (

Conflict of interest

F Brighenti has received grant support from Parmatecninnova SRL and Soremartee SPA. JW Beulens has received grant support from the Dutch Heart Foundation. S Bingham has received grant support from MRC Centre. The remaining authors have declared no financial interests.


  1. Barclay AW, Petocz P, McMillan-Price J, Flood VM, Prvan T, Mitchell P et al. (2008). Glycemic index, glycemic load, and chronic disease risk–a meta-analysis of observational studies. Am Clin Nutr 87, 627–637.

  2. Beulens JW, de Bruijne LM, Stolk RP, Peeters PH, Bots ML, Grobbee DE et al. (2007). High dietary glycemic load and glycemic index increase risk of cardiovascular disease among middle-aged women: a population-based follow-up study. J Am Coll Cardiol 50, 14–21.

  3. Brustad M, Skeie G, Braaten T, Slimani N, Lund E (2003). Comparison of telephone vs face-to-face interviews in the assessment of dietary intake by the 24 h recall EPIC SOFT program–the Norwegian calibration study. Eur J Clin Nutr 57, 107–113.

  4. Buyken AE, Toeller M, Heitkamp G, Karamanos B, Rottiers R, Muggeo M et al. (2001). Glycemic index in the diet of European outpatients with type 1 diabetes: relations to glycated hemoglobin and serum lipids. Am J Clin Nutr 73, 574–581.

  5. Cho E, Spiegelman D, Hunter DJ, Chen WY, Colditz GA, Willett WC (2003). Premenopausal dietary carbohydrate, glycemic index, glycemic load, and fiber in relation to risk of breast cancer. Cancer Epidemiol Biomarkers Prev 12, 1153–1158.

  6. Cust AE, Skilton MR, van Bakel MME, Halkjær J, Olsen A, Agnoli C et al. (2009). Total dietary carbohydrate, sugar, starch and fibre intakes in the European Prospective Investigation into Cancer and Nutrition. Eur J Clin Nutr 63 (Suppl 4), S37–S60.

  7. Deharveng G, Charrondiere UR, Slimani N, Southgate DA, Riboli E (1999). Comparison of nutrients in the food composition tables available in the nine European countries participating in EPIC. European Prospective Investigation into Cancer and Nutrition. Eur J Clin Nutr 53, 60–79.

  8. Du H, van der AD, Feskens EJ (2006). Dietary glycaemic index: a review of the physiological mechanisms and observed health impacts. Acta Cardiol 61, 383–397.

  9. Ferrari P, Kaaks R, Fahey MT, Slimani N, Day NE, Pera G et al. (2004). Within- and between-cohort variation in measured macronutrient intakes, taking account of measurement errors, in the European Prospective Investigation into Cancer and Nutrition study. Am J Epidemiol 160, 814–822.

  10. Foster-Powell K, Holt SH, Brand-Miller JC (2002). International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr 76, 5–56.

  11. Haftenberger M, Lahmann PH, Panico S, Gonzalez CA, Seidell JC, Boeing H et al. (2002a). Overweight, obesity and fat distribution in 50- to 64-year-old participants in the European Prospective Investigation into Cancer and Nutrition (EPIC). Public Health Nutr 5 (Suppl), S1147–S1162.

  12. Haftenberger M, Schuit AJ, Tormo MJ, Boeing H, Wareham N, Bueno-de-Mesquita HB et al. (2002b). Physical activity of subjects aged 50–64 years involved in the European Prospective Investigation into Cancer and Nutrition (EPIC). Public Health Nutr 5 (Suppl), S1163–S1176.

  13. Henry CJ, Lightowler HJ, Strik CM, Storey M (2005). Glycaemic index values for commercially available potatoes in Great Britain. Br J Nutr 94, 917–921.

  14. Higginbotham S, Zhang ZF, Lee IM, Cook NR, Giovannucci E, Buring JE et al. (2004). Dietary glycemic load and risk of colorectal cancer in the Women's Health Study. J Natl Cancer Inst 96, 229–233.

  15. Jenkins DJ, Wolever TM, Collier GR, Ocana A, Rao AV, Buckley G et al. (1987). Metabolic effects of a low-glycemic-index diet. Am J Clin Nutr 46, 968–975.

  16. Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JM et al. (1981). Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 34, 362–366.

  17. Jonas CR, McCullough ML, Teras LR, Walker-Thurmond KA, Thun MJ, Calle EE (2003). Dietary glycemic index, glycemic load, and risk of incident breast cancer in postmenopausal women. Cancer Epidemiol Biomarkers Prev 12, 573–577.

  18. Liljeberg HG, Lonner CH, Bjorck IM (1995). Sourdough fermentation or addition of organic acids or corresponding salts to bread improves nutritional properties of starch in healthy humans. J Nutr 125, 1503–1511.

  19. Liu S, Manson JE, Buring JE, Stampfer MJ, Willett WC, Ridker PM (2002). Relation between a diet with a high glycemic load and plasma concentrations of high-sensitivity C-reactive protein in middle-aged women. Am J Clin Nutr 75, 492–498.

  20. Livesey G, Taylor R, Hulshof T, Howlett J (2008). Glycemic response and health–a systematic review and meta-analysis: the database, study characteristics, and macronutrient intakes. Am J Clin Nutr 87, S223–S236.

  21. Ludwig DS (2003). Dietary glycemic index and the regulation of body weight. Lipids 38, 117–121.

  22. Michaud DS, Liu S, Giovannucci E, Willett WC, Colditz GA, Fuchs CS (2002). Dietary sugar, glycemic load, and pancreatic cancer risk in a prospective study. J Natl Cancer Inst 94, 1293–1300.

  23. Riboli E, Hunt KJ, Slimani N, Ferrari P, Norat T, Fahey M et al. (2002). European Prospective Investigation into Cancer and Nutrition (EPIC): study populations and data collection. Public Health Nutr 5 (Suppl), S1113–S1124.

  24. Riboli E, Kaaks R (1997). The EPIC Project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 26 (Suppl 1), S6–S14.

  25. Riccardi G, Rivellese AA, Giacco R (2008). Role of glycemic index and glycemic load in the healthy state, in prediabetes, and in diabetes. Am J Clin Nutr 87, S269–S274.

  26. Sahyoun NR, Anderson AL, Tylavsky FA, Lee JS, Sellmeyer DE, Harris TB (2008). Dietary glycemic index and glycemic load and the risk of type 2 diabetes in older adults. Am J Clin Nutr 87, 126–131.

  27. Salmeron J, Manson JE, Stampfer MJ, Colditz GA, Wing AL, Willett WC (1997). Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women. J Am Med Assoc 277, 472–477.

  28. Scholl TO, Chen X, Khoo CS, Lenders C (2004). The dietary glycemic index during pregnancy: influence on infant birth weight, fetal growth, and biomarkers of carbohydrate metabolism. Am J Epidemiol 159, 467–474.

  29. Schulz M, Liese AD, Mayer-Davis EJ, D'Agostino Jr RB, Fang F, Sparks KC et al. (2005). Nutritional correlates of dietary glycaemic index: new aspects from a population perspective. Br J Nutr 94, 397–406.

  30. Schulze MB, Liu S, Rimm EB, Manson JE, Willett WC, Hu FB (2004). Glycemic index, glycemic load, and dietary fiber intake and incidence of type 2 diabetes in younger and middle-aged women. Am J Clin Nutr 80, 348–356.

  31. Slimani N, Deharveng G, Charrondiere RU, van Kappel AL, Ocke MC, Welch A et al. (1999). Structure of the standardized computerized 24-h diet recall interview used as reference method in the 22 centers participating in the EPIC project. European Prospective Investigation into Cancer and Nutrition. Comput Methods Programs Biomed 58, 251–266.

  32. Slimani N, Deharveng G, Unwin I, Southgate DA, Vignat J, Skeie G et al. (2007). The EPIC nutrient database project (ENDB): a first attempt to standardize nutrient databases across the 10 European countries participating in the EPIC study. Eur J Clin Nutr 61, 1037–1056.

  33. Slimani N, Ferrari P, Ocke M, Welch A, Boeing H, Liere M et al. (2000). Standardization of the 24-h diet recall calibration method used in the European Prospective Investigation into Cancer and Nutrition (EPIC): general concepts and preliminary results. Eur J Clin Nutr 54, 900–917.

  34. Slimani N, Kaaks R, Ferrari P, Casagrande C, Clavel-Chapelon F, Lotze G et al. (2002). European Prospective Investigation into Cancer and Nutrition (EPIC) calibration study: rationale, design and population characteristics. Public Health Nutr 5 (Suppl), S1125–S1145.

  35. Van Bakel MME, Slimani N, Feskens EJM, Du H, Beulens JWJ, van der Schouw YT et al. (2009). Methodological challenges in the application of the glycemic index in epidemiological studies using data from the European Prospective Investigation into Cancer and Nutrition. J Nutr 139, 1–8.

  36. Willett W (1998). Nutritional Epidemiology. Oxford University Press: New York.

  37. Willett W, Stampfer MJ (1986). Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 124, 17–27.

  38. Wirfalt E, McTaggart A, Pala V, Gullberg B, Frasca G, Panico S et al. (2002). Food sources of carbohydrates in a European cohort of adults. Public Health Nutr 5 (Suppl), S1197–S1215.

  39. Wolever TM (2000). Dietary carbohydrates and insulin action in humans. Br J Nutr 83 (Suppl 1), S97–S102.

  40. Wolever TM, Nguyen PM, Chiasson JL, Hunt JA, Josse RG, Palmason C et al. (1994). Determinants of diet glycemic index calculated retrospectively from diet records of 342 individuals with non-insulin-dependent diabetes mellitus. Am J Clin Nutr 59, 1265–1269.

Download references


The work described in this paper was carried out with the financial support of the European Commission: Public Health and Consumer Protection Directorate 1993–2004; Research Directorate-General 2005; Ligue contre le Cancer (France); Société 3M (France); Mutuelle Générale de l’Education Nationale; Institut National de la Santé et de la Recherche Médicale (INSERM); Institut Gustave Roussy; German Cancer Aid; German Cancer Research Center; German Federal Ministry of Education and Research; Danish Cancer Society; Health Research Fund (FIS) of the Spanish Ministry of Health; Spanish Regional Governments of Andalucía, Asturias, Basque Country, Murcia and Navarra and the Catalan Institute of Oncology; and ISCIII RETIC (RD06/0020), Spain; Cancer Research UK; Medical Research Council, UK; the Stroke Association, UK; British Heart Foundation; Department of Health, UK; Food Standards Agency, UK; the Wellcome Trust, UK; Greek Ministry of Health; Hellenic Health Foundation; Italian Association for Research on Cancer; Italian National Research Council, Regione Sicilia (Sicilian government); Associazione Iblea per la Ricerca Epidemiologica—ONLUS (Hyblean association for epidemiological research, NPO); Dutch Ministry of Health, Welfare and Sport; Dutch Prevention Funds; LK Research Funds; Dutch ZON (Zorg Onderzoek Nederland); World Cancer Research Fund (WCRF); Swedish Cancer Society; Swedish Research Council; Regional Government of Skane and the County Council of Vasterbotten, Sweden; Norwegian Cancer Society; the Norwegian Research Council and the Norwegian Foundation for Health and Rehabilitation. We thank Sarah Somerville, Nicole Suty and Karima Abdedayem for their assistance with editing and Kimberley Bouckaert and Heinz Freisling for technical assistance.

Author information

Correspondence to N Slimani.

Additional information

Guarantor: MME van Bakel.

Contributors: MMEvB wrote the paper and prepared the tables and figures, taking into account comments from all co-authors. MMEvB and CB conducted the statistical analyses. NS was the overall coordinator of this project and was responsible for the EPIC nutritional databases. AEC, JH, JWB, YTvdS, EJMF, DvdA, HD, FB, VP, KA, MJT, PF, CB and NS were members of the ‘GI/GL working group’ and provided input on the statistical analyses, drafting of the paper and interpretation of results. The other co-authors were local EPIC collaborators involved in the collection of dietary and other data and in documenting and compiling the EPIC Nutrient Database (ENDB). ER is the overall coordinator of the EPIC study. All co-authors provided comments and suggestions on the paper and approved the final version.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

van Bakel, M., Kaaks, R., Feskens, E. et al. Dietary glycaemic index and glycaemic load in the European Prospective Investigation into Cancer and Nutrition. Eur J Clin Nutr 63, S188–S205 (2009).

Download citation


  • glycaemic index
  • glycaemic load
  • 24-h dietary recall
  • EPIC
  • ENDB
  • standardization

Further reading