The Glycaemic Index of Potatoes: the Effect of Variety, Cooking Method and Maturity

The glycaemic index of potatoes: the effect of variety, cooking method and maturity

NL Soh1 and J Brand-Miller1*

1Human Nutrition Unit, Department of Biochemistry, the University of Sydney, NSW, 2006, Australia

Objective: The aim of this study was to determine the impact of variety, cooking method and maturity on the GI of potatoes, it was hypothesised that new potatoes may have a relatively lower GI. Design and subjects: Ten healthy volunteers were recruited as subjects through advertising on the campus of the University of Sydney. Equal (50 g) carbohydrate portions of eight potato meals (three varieties, four cooking methods, two states of maturity) and two reference white bread meals were fed in random order to each of the subjects over a period of 10 weeks. Capillary blood samples were taken in the fasting state and then at 15, 30, 45, 60, 90 and 120 min from the start of each meal. Samples were analysed for plasma glucose concentrations and incremental areas under plasma glucose curves were calculated. The GI of the potato was calculated as the AUC of the potato expressed as a percentage of the individual's average AUC of the white bread. This was then multiplied by 0.7 to index the GI to glucose as the reference food. Results: GI values (meanÆ s.e.m.) ranged from 65Æ 9 (canned new potatoes) to 101Æ 15 (boiled Desiree potatoes), glucose 100. No signi®cant difference was found among the three varieties of potato tested (P 0.38) or among the four different cooking methods (P 0.55). The GI values of the canned new potato and boiled Desiree potato were signi®cantly different (P 0.047). The average size of the tuber was found to correlate with the GI (r 0.83, P < 0.05). Conclusions: Potatoes, regardless of variety, cooking method and maturity, have exceptionally high GI values. New potatoes have relatively lower GI values which is attributed to differences in starch structure. Sponsors: University of Sydney. Descriptors: glycaemic index; food; starch; digestion; potatoes

Introduction Low GI diets have been shown to decrease fasting and postprandial insulin responses (Wolever et al, 1991; DAA, The glycaemic index (GI) is a method of ranking foods 1997) and improve insulin sensitivity in patients with according to their postprandial blood glucose response with advanced coronary heart disease (Frost et al, 1996). respect to that of an equicarbohydrate portion of a reference The potato (Solanum tuberosum) generally has one of food (Wolever et al, 1994). The GI is relevant in both the highest GI values of any food, although some types preventing and managing diabetes mellitus. Two six-year such as `new' potatoes appear to be lower than others. The cohort studies, one in men (SalmeroÂn et al, 1997a) and one published GI values for potatoes vary from as low as 56 to in women (SalmeroÂn et al, 1997b) have demonstrated that as high as 85 (on a scale where glucose 100) for reasons diets with high glycaemic load=low cereal ®bre content are which are not clear and when centres of study are com- linked with more than twice the risk of non-insulin depen- pared, the potato is the food which yields the most variable dent diabetes mellitus (NIDDM) when compared to diets glycaemic response (Wolever, 1990). Furthermore, the with low glycaemic load=high cereal ®bre content. Hence, variety of the potato investigated is often not speci®ed, a diet with a low glycaemic index may help prevent leading to some confusion over the impact of variety vs NIDDM. Cross-over studies in NIDDM patients have preparation method. shown that decreasing dietary GI leads to improved gly- The GI of potatoes needs to be clari®ed because they caemic control (Wolever et al, 1992; Brand et al, 1991), make a major contribution to total starch intake. In indus- similar to that achieved by hypoglycaemic drugs. trialised countries, potatoes contribute about 15 ± 20% of Studies have also shown improved lipidaemic control total starch intake compared with bread, another staple, with low GI diets in NIDDM and hyperlipidaemic subjects which contributes about 35% (English et al, 1987). Given (Wolever et al, 1992). Establishing and maintaining accep- the popularity of the potato, a low GI variety and=or table plasma lipid levels is important in diabetes manage- cooking method could signi®cantly lower the overall gly- ment as the disease carries a greater risk of developing caemic load of the western diet, this in turn may decrease cardiovascular disease (Anderson & Geil, 1994). Cardio- the risk of NIDDM. vascular disease has also been directly linked to serum The aim of this present study was to determine the insulin levels (DespreÂs et al, 1996; Salonen et al, 1998). differences in GI between the more popular varieties of potatoes commonly sold on international markets and the *Correspondence: Prof J Brand-Miller, Human Nutrition Unit, Department effects of different cooking methods on GI. Emphasis was of Biochemistry G08, University of Sydney NSW 2006, Australia. Received 30 June 1998; revised 27 October 1998; accepted 6 November placed on determining the true available carbohydrate 1998 content of the different potatoes studied, as opposed to The glycaemic index of potatoes NL Soh and J Brand±Miller 250 reliance on tables of food composition. It was hypothesised Available carbohydrate contents of the fresh potatoes that new potatoes may have a relatively lower GI due to were analysed using a previously described method (Wills differences in starch structure affected by maturity. et al, 1980). Brie¯y, a homogenised food sample is extracted with methanol and ®ltered, the extract is analysed for sugar content by HPLC while the starch in the residue is Materials and methods hydrolysed with amyloglucosidase before also being analysed for sugar content by HPLC. The available carbo- Subjects hydrate content of the food is provided by the sum of the Ten healthy volunteers (eight female and two male) were sugars in the extract and the residue. This method excludes recruited through advertisements placed around The Uni- most resistant starch. Carbohydrate contents for the refer- versity of Sydney campus. The study was approved by the ence white bread and the canned new potatoes were taken Medical Ethics Committee of The University of Sydney from manufacturers' information. Table 1 shows the carbo- and all volunteers gave written informed consent. Mean hydrate contents of the foods and the weight of 50 g ageÆ s.d. was 25Æ 7 y and mean body mass indexÆ s.d. carbohydrate portions. The average weights per tuberÆ s.d. was 22.8Æ 2.8. Before each test, subjects fasted for 10 h (unpeeled) for Sebago, Desiree, Pontiac and new potatoes overnight, during which time consumption of water was were 196Æ 66 g, 188Æ 64 g, 91Æ 28 g and 73Æ 20 g. The permitted. To avoid the `second meal effect' (Wolever et average weight in the canned product was 33 g, the range al, 1988), they were instructed not to eat legumes in the being 22 ± 45 g. meal preceding the fast. Alcohol was limited to two drinks Fresh potatoes were peeled, weighed out, dipped for 30 s the day before each test and a similar meal was eaten before in a 0.1% sodium metabisulphite solution to prevent each fast. browning and stored covered in the refrigerator overnight. On the morning of each test, two ®nger-prick capillary To avoid starch retrogradation, potatoes were freshly blood samples were collected ®ve minutes apart to deter- cooked each morning. Potatoes were cooked until soft mine baseline glucose levels. Foods were eaten at 0 min and large potatoes were halved before cooking. Potatoes time and over the 2 h following the start of each test meal, for boiling were heated to 100C in an excess of boiling 1 ml capillary blood samples were collected at 15, 30, 45, water and then boiled for 35 min. New potatoes were left 60, 90 and 120 min. To enhance peripheral blood circula- unpeeled, the skins pricked with a fork and added to an tion to the ®ngers, subjects warmed their hands with hot excess of already boiling water and boiled for 20 min. water bottles for approximately 5 min before each blood 1 Potatoes for mashing were cut into 1 ± 2 cm cubes, heated sampling. Blood samples were taken using an Autoclix to 100C in an excess of boiling water and boiled for device (Boehringer Mannheim, Australia), collected into 15 min. The potato was then drained and mashed with a 1.5 ml microcentrifuge tubes coated with heparin (10 IU fork. In all the potato meals subjected to the boiling heparin sodium salt, Sigma Chemical Co., St Louis, USA) process, the cooking liquid was included in the water and immediately centrifuged at 12 000 g for 30 s. The balance of the meal. plasma was removed and stored at 720 C for later Potatoes for baking were prepared by wrapping the analysis. peeled potatoes individually in aluminium foil and baking them at 190C in a gas oven for 25 min. Potatoes for Test foods microwaving were placed in a covered ceramic dish and Eight test meals and two reference food meals (white microwaved at full power (650 W) for 6 ± 7.5 min. The bread) were given to each subject in randomised order canned new potatoes were drained, weighed out on the over a 10 week period. White bread (Tip Top1 Sunblest1, morning of each test and microwave-heated at full power Sydney, Australia) was chosen as the reference food over for 3 min before serving. the original glucose as it was considered to be a more physiological standard (Wolever, 1990). Fresh potatoes Plasma glucose analysis were purchased from Sydney retail supermarkets, in bulk Plasma samples, stored at 720C, were thawed at room quantities suf®cient to conduct all tests; canned new pota- temperature, vortexed and centrifuged before analysis for toes (EdgellTM Mint Tiny TatersTM, Edgell-Birdseye, Mel- glucose concentration. Photometric analyses were con- bourne, Australia) were donated by Cowra Export Packers ducted in duplicate on a Cobas Fara centrifugal analyser Ltd. The three varieties of potato chosen for testing were (Roche Diagnostica, Basle, Switzerland) using a hexokina- the major varieties grown for the Australian fresh market se=glucose-6-phosphate dehydrogenase method (Unimate 5 (Kirkham, 1995) and also featured in international markets Gluc HKTM, Roche Diagnostic Systems, Frenchs Forest, in the United Kingdom and United States, namely, Sebago, Australia). All plasma samples pertaining to the one food Desiree and Pontiac. Fresh new potatoes and canned new potatoes were tested because immature potatoes are known to contain starch with different amylose to amylopectin Table 1 Carbohydrate (CHO) content of the foods tested based on the ratios (Brunt and Zinsmeester, unpublished data). Four raw product with the exception of canned potatoes methods of cooking were assessed using one variety, that Food Available Weight of 50 g CHO is Pontiac: boiled, boiled and mashed, oven-baked and CHO (g=100 g) portion (g) balance (ml) Water microwaved. All tests were conducted on the peeled Tip Top1; Sunblest1 43.8 114 620 potato, except in the case of fresh boiled new potatoes White bread which are generally eaten with the skin intact. Therefore, in Sebago, peeled 11.0 454 315 total, eight different potato meals, each containing 50 g Desiree, peeled 11.2 446 325 available carbohydrate (starch and sugars) were fed to all New, unpeeled 14.0 358 415 Pontiac, peeled 12.1 414 355 10 subjects. Water was used to make up the meal volume to Canned new, drained 12.0 416 360 750 ml in each case. The glycaemic index of potatoes NL Soh and J Brand±Miller 251 were analysed at the same time. Inter- and intra-assay coef®cients of variance were 1.16% and 1.45% respectively.

Calculation of GI The incremental areas under the plasma glucose curves (AUC) were determined for each food according to stan- dardised criteria (Wolever et al, 1991), ignoring any area below the baseline. The average AUC for the two white bread tests was used as the reference value and each subject's individual GI for each food was calculated. The GI for each food was taken as the average of all 10 individual values. This was then multiplied by 0.7 to convert the value to one indexed to glucose, GI 100 (Foster-Powell & Brand Miller, 1995).

Statistical analyses The glycaemic responses of the two reference white bread tests were compared using a paired t-test. Two-way analysis of variance was used to compare GI values between potato varieties (boiled Sebago, Pontiac and Desiree), between cooking methods (baked, boiled, mashed and microwaved Pontiac potatoes) and between the new potatoes and the potato variety with the highest GI. The correlation coef®cient between GI value and tuber weight was determined, the level of signi®cance was taken as P < 0.05.

Results The GI of the products tested is shown in Table 2, no statistically signi®cant difference was found among the GI of the different varieties of potato (Sebago, Pontiac and Desiree) (P 0.38, Figure 1) nor among the various cooking methods (boiled, baked, microwaved and mashed Pontiac potato) (P 0.55, Figure 2). The only signi®cant Figure 1 Top: Incremental glycaemic response to potatoes by variety. difference was between the canned new potato and boiled Bottom: Glycaemic index of potatoes by varietyÆ s.e.m. (glucose 100), Desiree potato (P 0.047, Figure 3). The GI of the three none of the differences is statistically signi®cant. mature varieties and two `new' potato products correlated with their average tuber weight (r 0.83, P < 0.05, Figure values, with one case equivalent to that of a 50 g glucose 4). load (Desiree, boiled: GI 101Æ 15). Current nutritional advice to increase the intake of starchy foods such as Discussion potatoes may therefore lead to a diet with increased glycaemic load and hence greater risk of NIDDM (Sal- This study con®rms that potatoes, irrespective of variety, meroÂn et al, 1997a, 1997b). cooking method or maturity, have exceptionally high GI In the literature, there is a wide range of GI values for boiled potatoes, from as low as 56 to as high as 85. Because Table 2 GI values of test foodsÆ s.e.m. (n 10, except n 9 for new the variety is usually not speci®ed, it is not possible to potatoes, canned and microwave-heated) determine whether this is the source of the variation. In this present study, the GI of boiled Pontiac potatoes was 88, GIÆ s.e.m. GIÆ s.e.m. Food (white bread 100) (glucose 100) which is 32 units higher than a previously published value of 56 (Brand et al, 1985). Such discrepancies may be Variety largely attributed to the use of different food composition Sebago, peeled and boiled 124Æ 10 87Æ 7 data when calculating the serving size of the food (Jenkins Desiree, peeled and boiled 144Æ 22 101Æ 15 Pontiac, peeled and boiled 125Æ 13 88Æ 9 et al, 1988). For example, published carbohydrate content Cooking method for peeled, boiled new potatoes varies from 12.8 g=100 g Pontiac, peeled and boiled 125Æ 13 88Æ 9 (English & Lewis, 1991) to 17.8 g=100 g (Holland et al, Pontiac, peeled, boiled 130Æ 13 91Æ 9 1991). This results in 50 g carbohydrate portion sizes of and mashed Pontiac, peeled and microwaved 112Æ 13 79Æ 9 391 g and 281 g respectively, a difference of nearly 30%. In Pontiac, peeled and baked 133Æ 15 93Æ 11 this present study we used 414 g of Pontiac potato, fresh Maturity weight while the previous study speci®ed 294 g of boiled New, unpeeled and boiled 112Æ 17 78Æ 12 Pontiac potato based on data from food tables (Brand et al, New, canned and microwave 93Æ 13 65Æ 9 1985), this difference alone can account for the discrepancies heated in GI values. The glycaemic index of potatoes NL Soh and J Brand±Miller 252

Figure 2 Top: Incremental glycaemic response to Pontiac potato by cooking method. Bottom: Glycaemic index of Pontiac potato by cooking methodÆ s.e.m. (glucose 100), none of the differences is statistically Figure 3 Top: Incremental glycaemic response to canned new potato, signi®cant. boiled new potato and Desiree potato. Bottom: Glycaemic index for canned new potato, boiled new potato and Desiree potatoÆ s.e.m. (glucose 100). Authorities differ in their conclusions regarding the impact of cooking methods on the GI of potatoes. In this present study, there were no signi®cant differences in GI of Pontiac potatoes whether they were boiled, oven-baked, microwaved or mashed. In contrast, Lunetta et al (1995) found that baked potatoes produced a signi®cantly lower incremental glycaemic response compared with boiled potatoes, even when the same amounts of fat were added to each. They concluded that the baking process led to less cooking of the internal part of the potato and reduced the potato's digestibility. Wolever et al (1994), like us, found no signi®cant differences between baked, boiled and canned potatoes. However, they found that mashing signi®cantly increased the glycaemic response (by 15 ± 20%). On the other hand, Englyst & Cummings (1987) found that starch digestibility of cooled potato was identical when eaten as large lumps or ®nely sieved twice, implying that the glycaemic response is unlikely to be affected. Our study showed no effect of cooking method on the GI of just one variety of potato (Pontiac). This does not necessarily imply that other potato varieties will behave in the same way. However, we can think of no scienti®c Figure 4 Correlation between average weight of potato tuber and GI reason why there might be differences. Cooking methods (Æ s.e.m.). The glycaemic index of potatoes NL Soh and J Brand±Miller 253 involving `moist heat' (mashing, baking, microwaving) are status as a high carbohydrate food, the potato is relatively likely to affect different varieties in similar ways. Potato low when compared with most cereal foods: close 0.4 kg starch gelatinises at 55 ± 66C (Crapo et al, 1981) which is was needed to achieve the 50 g carbohydrate portions used well below the temperatures involved in baking and boiling in this study. It is unlikely that such an amount would potatoes. Furthermore, fresh potato tubers contain suf®cient normally be consumed in one sitting and this would affect water to fully gelatinise their starch contents when heat- the glycaemic response in the usual setting. Nonetheless, to treated (Kingman & Englyst, 1994) and resistant starch achieve a prescribed goal in carbohydrate (200 g) or granules are unlikely to remain in potatoes cooked using number of exchanges (12), potatoes could well contribute conventional domestic methods. To avoid differences in to an exceptionally high glycaemic load, this, in turn, may resistant starch content as a result of cooling, we served all increase insulin demand and exacerbate insulin resistance. the products immediately after cooking (or heating in the It is possible that `wild' or less domesticated varieties of case of the canned product). potato have lower GI values than current ones. Starchy Both boiled new potatoes and canned new potatoes had vegetables which were important foods in the diets of some the lowest GI values of the potatoes tested, the difference Australian Aborigines (for example, cheeky yam (Discorea reaching statistical signi®cance when compared with the bulbifera), pencil yam (Vigna lanceolata), bush potato highest GI potato. In absolute terms, the average GI of (Ipomoea costata)) and Paci®c Islanders (for example, canned potatoes was almost 36% less, a difference that is sweet potato (Ipomoea batatas)) have been shown to be probably biologically important. The relatively lower digested more slowly than Western starchy staples (Thor- values of the new potatoes may be due to differences in burn et al, 1987a). and to produce signi®cantly lower starch structure. As potatoes mature, the quantity of amy- glycaemic responses when compared to Sebago potatoes lose increases but the difference is small and not likely to (Thorburn et al, 1987a, 1987b). Given the popularity of the affect the glycaemic response (Brunt & Zinsmeester, potato and its extraordinarily high GI, it may be valuable to unpublished data). On the other hand, the degree of manipulate the genotype of wild or cultivated varieties of amylopectin branching signi®cantly increased with the potato to yield a commercially viable low GI potato. maturity of the potato. Amylopectin has an irregular, branching structure and is more readily gelatinised than the linear amylose molecule leading to a higher glycaemic Conclusions effect (Wolever, 1990). We speculate that the lower GI of Neither variety nor cooking method appears to have new potatoes may be due to reduced amylopectin branching marked effects on the GI of potatoes tested in this current and hence greater resistance to gelatinisation. Canned new study. Apparent differences among varieties in the litera- potatoes may be more immature again as the typical size is ture may be methodological artifacts resulting from differ- smaller: the average weight of the fresh new potato was ences in assumed carbohydrate content. Apart from the 73 g while that of the canned was only 33 g. To test the canned variety, all the potatoes tested in the present study hypothesis that the size of the potato in¯uenced the GI (via had high GI values (GI > 78). The lower GI of canned new differences in maturity and starch structure), we determined potato (GI 65) may result from differences in starch the correlation between GI and the average size of the structure in immature potatoes but may also be due to potato tuber in our study (Figure 4). The r value of 0.83 unintentional inclusion of some resistant starch in the 50 g (P < 0.05) suggests that the starch of more mature potatoes carbohydrate portion. The lower GI of canned new potatoes is easier to digest, leading to a higher GI. is probably of suf®cient magnitude to be clinically impor- The unintentional inclusion of some resistant starch in tant in the dietary management of diabetes. The correlation the 50 g available carbohydrate portion of canned new between the average size of the potatoes and their GI needs potatoes might also have contributed to its low GI. The to be con®rmed in further studies. carbohydrate content of this product was provided by the manufacturer who used the `carbohydrate by difference' Acknowledgements ÐWe are grateful to Cowra Export Packers Ltd who method (calculated by substracting the percentages of donated the canned new potatoes, Edgell-Birdseye, Australia who pro- water, protein, fat, ash and dietary ®bre from 100). The vided nutritional information and the volunteers who undertook the ten manufacturer used the AOAC method (Prosky et al, 1988) weeks of GI testing. of measuring dietary ®bre which includes some but not all resistant starch (RS). In the published literature, the RS of cold canned new potatoes is 6.8% of total starch but re- References heating reduces this to 1.4% (Kingman & Englyst, 1994). Anderson JW & Geil PB (1994): Nutritional management of diabetes The canned new potatoes in our study had a total starch mellitus. In Modern Nutrition in Health and Disease. In Vol 2, 8th edn, content of 11.6 g=100 g drained weight. Hence, of the 50 g ME Shils, JA Olson & M Shike (eds.) Philadelphia: Lea and Febiger. pp carbohydrate portion served, 48.3 g was starch of which 1259 ± 1286. Brand JC, Colagiuri S, Crossman S, Allen A, Roberts DCK & Truswell AS 1.4% may have been RS 0.7 g. This small amount would (1991): Low-glycemic index foods improve long-term glycemic control not account for all of the difference between the GI of the in NIDDM. Diabetes Care 14, 95 ± 101. canned new potatoes and the mature potatoes. Brand JC, Nicholson PL, Thorburn, AW & Truswell AS (1985): Food The very high GI values of the potatoes tested begs the processing and the glycemic index. Am. J. Clin. Nutr. 42, 1192 ± 1196. Crapo PA, Insel J, Sperling M & Kolterman OG (1981): Comparison of question: is the potato an `undesirable' food? Foods con- serum glucose, insulin and glucagon responses to different types of taining glucose are not recommended to people with complex carbohydrate in noninsulin-dependent diabetic patients. Am. J. diabetes because of their assumed potential to cause hyper- Clin. Nutr. 34, 184 ± 190. glycaemia. Potatoes are still recommended in diabetes due DespreÂs JP, Lamarche B, MaurieÁge P, Cantin B, Dagenais GR, Moorjani S to their low fat and high micronutrient content. Their high & Lupien PJ (1996): Hyperinsulinemia as an independent risk factor for ischemic heart disease. N. Engl. J. Med. 334, 952 ± 957. satiety (Holt et al, 1995) may also bene®t NIDDM patients Dietitians Association of Australia (1997): Glycaemic index in diabetes as weight control is often part of their treatment. Despite its management. Aust. J. Nutr. Diet. 54, 57 ± 63. The glycaemic index of potatoes NL Soh and J Brand±Miller 254 English R, Cashel K, Bennett S, Berzins J, Waters A & Magnus P (1987): SalmeroÂn J, Ascherio A, Rimm, EB, Colditz GA, Spiegelman D, Jenkins National Dietary Survey of Adults: 1983ÐNo. 2 Nutrient Intakes. 29. DJ, Stampfer MJ, Wing AL & Willett WC (1997a): Dietary ®ber, Canberra: Australian Government Publishing Service. glycemic load, and risk of NIDDM in men. Diabetes Care 20, 545 ± English R & Lewis J (1991): Nutritional Values of Australian Foods. 550. Canberra: Australian Government Publishing Service, pp 31 ± 32. SalmeroÂn J, Manson JE, Stampfer MJ, Colditz GA, Wing AL & Willett Englyst HN & Cummings JH (1987): Digestion of polysaccharides of WC (1997b): Dietary ®ber, glycemic load, and risk of non-insulin- potato in the small intestine of man. Am. J. Clin. Nutr. 45, 423 ± 431. dependent diabetes mellitus in women. JAMA 277, 472 ± 477. Foster-Powell K & Brand Miller J (1995): International tables of glycemic Salonen JT, Lakka TA, Lakka HM, Valkonen VP, Everson SA & Kaplan index. Am. J. Clin. Nutr. (suppl) 62, S871 ± S890. GA (1998): Hyperinsulinemia is associated with the incidence of Frost G, Keogh B, Smith D, Akinsanya K & Leeds A (1996): The effect of hypertension and dyslipidemia in middle-aged men. Diabetes 47, low-glycemic carbohydrate on insulin and glucose response in vivo and 270 ± 275. in vitro in patients with coronary heart disease. Metabolism 45, Thorburn AW, Brand JC & Truswell AS (1987a): Slowly digested and 669 ± 672. absorbed carbohydrate in traditional bushfoods: a protective factor Holland B, Unwin ID & Buss DH (1991): Vegetables, Herbs and Spices. against diabetes? Am. J. Clin. Nutr. 45, 98 ± 106. 5th Supplement to McCance and Widdowson's The Composition of Thorburn AW, Brand JC, O'Dea K, Spargo RM & Truswell AS (1987b): Foods 4th Edition, 10 ± 15. Cambridge: Royal Society of Chemistry and Plasma glucose and insulin responses to starchy foods in Australian Ministry of Agriculture, Fisheries and Food. Aborigines: a population now at high risk of diabetes. Am. J. Clin. Nutr. Holt SHA, Brand Miller JC, Petocz P & Farmakalidis E (1995): A satiety 46, 282 ± 285. index of common foods. Eur. J. Clin. Nutr. 49, 675 ± 690. Wills RBH, Balmer N & Green®eld H (1980): Composition of Australian Jenkins DJA, Wolever TMS & Jenkins AL (1988): Starchy foods and foods. 2. Methods of analysis. Food Technol. Australia 32, 198 ± 204. glycemic index. Diabetes Care 11, 149 ± 159. Wolever TMS (1990): The glycemic index. World Rev. Nutr. Diet. 62, Kingman SM & Englyst HN (1994): The in¯uence of food preparation 120 ± 185. methods on the in-vitro digestibility of starch in potatoes. Food Wolever TMS, Jenkins DJA, Jenkins AL & Josse RG (1991): The Chemistry 49, 181 ± 186. glycemic index: methodology and clinical implications. Am. J. Clin. Kirkham R (1995): Potatoes. In: Horticulture Australia: The Complete Nutr. 54, 846 ± 854. Reference of the Australian Horticultural Industry. B Coombs (ed.) Wolever TMS, Jenkins DJA, Ocana AM, Rao VA & Collier GR (1988): Hawthorn East, Victoria: Morescope Publishing, pp.250 ± 256. Second-meal effect: low-glycemic-index foods eaten at dinner improve Lunetta M, Di Mauro M, Crimi S & Mughini L (1995): In¯uence subsequent breakfast glycemic response. Am. J. Clin. Nutr. 48, of different cooking processes on the glycaemic response to 1041 ± 1047. potatoes in non-insulin dependent diabetic patients. Diab. Nutr. Wolever TMS, Jenkins DJA, Vuksan V, Jenkins AL, Wong GS & Josse Metab. 8, 49 ± 53. RG (1992): Bene®cial effects of low-glycemic index diet in overweight Prosky L, Asp N-G, Schweizer TF, DeVries JW & Furda I. (1988): NIDDM subjects. Diabetes Care 15, 562 ± 564. Determination of insoluble, soluble and total dietary ®ber in foods Wolever TMS, Katzman-Relle L, Jenkins AL, Vuksan V, Josse RG & and food products interlaboratory study. J. Assoc. Off. Anal. Chem. 71, Jenkins DJA (1994): Glycaemic index of 102 complex carbohydrate 1017 ± 1023. foods in patients with diabetes. Nutr. Res. 14, 651 ± 669.