
Roti, dosa, upma, sundal: a team in Chennai picked twelve commonly eaten traditional Indian foods like these and put each one through a standard glycaemic index (GI) test. Their numbers make the first chart on this page. The second chart lists five tamar-stage dates measured in a separate study at the United Arab Emirates University. Read them side by side with care, because they come from different volunteers in different countries.
What a GI number means here
Both studies used the same basic test. After an overnight fast, a volunteer ate an amount of the food that supplied 50 g of available carbohydrate, and capillary blood glucose was followed for two hours. The rise was then set against the rise after 50 g of glucose, which scores 100. The UAE paper spells out the bands: a GI of 55 or less counts as low, 56 to 69 as medium, and 70 or more as high.
A GI value says nothing about portion size. The Chennai team also gives a glycaemic load (GL), but only for that 50 g carbohydrate portion, so the GL column below rises and falls with GI.
Chart 1: twelve traditional Indian foods
Shobana and colleagues in Chennai, India, cooked every dish in their own test kitchen and tested it between 2010 and 2013. Fifteen adults aged 20 to 45 took part, seven men and eight women. They were screened as healthy, but the study only enrolled people with a BMI above 23, and the group’s average BMI was 26.2. A few results were dropped as outliers, so each food has 12 to 15 people behind it. Values are from the paper’s Table III, sorted from lowest to highest GI.
| Food | Cooked weight for 50 g carbohydrate | GI (mean ± SEM) | Band | GL |
|---|---|---|---|---|
| White chickpea sundal | 224.5 g | 24.1 ± 2.6 | Low | 12.1 |
| White peas sundal | 322.6 g | 29.8 ± 4.2 | Low | 14.9 |
| Broken wheat upma | 189.8 g | 51.7 ± 7.7 | Low | 25.9 |
| Methi paratha | 117.9 g | 60.2 ± 6.7 | Medium | 30.1 |
| Sorghum idli | 181.8 g | 61.3 ± 5.1 | Medium | 30.7 |
| Wheat dosa | 154.0 g | 61.8 ± 5.7 | Medium | 30.9 |
| Adai (rice and pulse pancake) | 155.6 g | 66.2 ± 3.5 | Medium | 33.1 |
| Pearl millet roti | 104.6 g | 70.1 ± 3.0 | High | 35.1 |
| Wheat flakes chivda | 94.6 g | 72.5 ± 3.4 | High | 36.3 |
| Maize roti | 116.1 g | 74.8 ± 5.5 | High | 37.4 |
| Sorghum roti | 113.9 g | 84.1 ± 6.7 | High | 42.1 |
| Finger millet stiff porridge (ball) | 187.3 g | 98.2 ± 5.5 | High | 49.1 |
The surprise sits at the bottom of the table. Finger millet balls, a whole-grain food, scored highest of all at 98.2. The authors explain that the flour is boiled in water into a stiff porridge, so the grain structure is lost and the starch is easy to reach. Their own conclusion is blunt: being whole grain does not by itself give a food a lower GI.
Two pairs are worth a second look. Methi paratha came out medium while the plain rotis were high, and the authors suggest the curd and oil in the paratha dough may explain the gap. Sorghum turned up twice, at 84.1 as a roti and 61.3 as an idli, where the batter also holds black gram. The two pulse dishes were the only foods under 30, and the paper ends by suggesting that low-GI legumes served alongside a high-GI food may help shape the response to the whole meal.
Chart 2: five tamar dates from the UAE
Alkaabi and colleagues tested five varieties popular in the UAE, all at the tamar stage and all from the same packaged batches, supplied by a date factory in Al Ain. Thirteen healthy volunteers, average age 40.2 and average BMI 27.4, ate each date once. The weights below are the amount of date flesh that gave 50 g of available carbohydrate (the paper’s Table 4).
| Variety | Flesh eaten | GI (mean ± SEM) |
|---|---|---|
| Bo ma’an | 72.7 g | 46.3 ± 7.1 |
| Dabbas | 76.2 g | 49.1 ± 3.6 |
| Lulu | 73.6 g | 53.5 ± 8.6 |
| Fara’d | 72.5 g | 54.0 ± 6.1 |
| Khalas | 72.6 g | 55.1 ± 7.7 |
The authors found no statistically significant difference among the five (p = 0.83). Khalas at 55.1 sits just above the 55 cut-off, so call it borderline. Every portion fell between 72.5 and 76.2 g, less food by weight than any dish in Chart 1, because the date flesh carried more available carbohydrate per gram than any of the cooked dishes.
Reading the two charts together
On the numbers alone, the five dates land near broken wheat upma and well below the rotis. That is as far as the comparison should go. Different people ate the foods in Chennai and Al Ain, each food was eaten on its own rather than as a meal, and neither study tested Sayer or Mazafati. Sayer is a semi-dry variety from Khuzestan in Iran, and these figures cannot be borrowed for it.
This page reports published test results and is not dietary or medical advice. Sources: Shobana S et al., Indian J Med Res 2022 (PMID 35859429); Alkaabi JM et al., Nutr J 2011 (PMID 21619670). Other articles are on the blog.
More about this site is on the About Us page and the contact page.