Former Visiting Professor Patrick Holford is Head of Science and Education at Biocare. Like proud parents who insist that their off-spring should entertain visitors with songs and recitations that would be better honoured in silence, Biocare proudly publicise their belief in Holford’s scientific acumen despite the faux pas and errors that have been highlighted in his work. From time to time, one wonders whether Biocare keeps up to date with Holford’s work and whether they notice, or even care about, the subtle and not-so-subtle non-sequiturs and distortions that crop up on a regular basis. Today’s example is no exception to Holford’s grim pattern of imagineering others’ research to support his own entrepreneurial needs.
In yet another post about the glycaemic load (GL) Diet, Holford claims that people who follow his low GL diet (he wants you to buy his seminars, book and recommended supplements) find it easy to adhere to it because they lose their desire to over-eat: Why a low GL diet makes you feel full. When HolfordWatch read the following, we were unsurprised when Holford failed to give any references for this assertion as this is par for the course.
While many studies have also reported this finding with low GL diets exactly why this happens has not been clear.
However, HolfordWatch was looking forward to Holford’s discussion of a study that he claims in support of his GL Diet and its appetite suppressing qualities because we like playing Bad Science Bingo.[a]
Researchers at Kings College London University may have discovered why and are presenting their findings today at the annual conference of the Society of Endocrinology. They fed volunteers a high or low GL meal and measured the release of a key hormone involved in appetite control, called gut hormone glucagon-like peptide 1 (GLP-1). GLP-1 levels went up by 20% after the low GL meal which is consistent with the reports of reduced appetite. These results show for the first time that eating a low GI meal increases GLP-1 production and suggest a physiological mechanism as to why a low GI meal makes you feel fuller than a high GI meal.
This is an important discovery as a natural desire to eat less is a great ally for those wanting to lose weight. [Emphasis added.]
Holford didn’t supply a reference or link but we identified the study that currently only exists in abstract form for anyone who did not attend the meeting: Effect of single high vs low glycemic index (GI) meal on gut hormones. We have emphasised the references to the meal in Holford’s account because we want you to notice that Holford switches between low GI and low GL as if they are equivalent – this is not necessarily true:[b] it is also not what the authors reported in the abstract.
That’s the problem with an abstract, there just isn’t enough detail but what there is indicates that it is not acceptable for Holford to have done this. The researchers actually reported that they had assessed the impact of a low GI meal:
Twelve healthy individuals with BMIs of 18–25 kg/m2 took part in a randomized cross-over study. The subjects had an identical medium GI dinner in the evening prior to the study and fasted overnight. On the morning of study each subject consumed a single meal of low (46) or high (66) GI and subsequently had blood samples taken every 30 min for a total of 150 min.
Holford is usually quite eager to differentiate between GI and GL but he hasn’t on this occasion. Indeed, Holford adds to his misrepresentation of this study by claiming:
This research specifically looked at the effect of low GL carbs versus high GL carbs.
The researchers do not mention that in the abstract. It is not particularly plausible, but, based on the abstract and in the absence of further information, the reader does not know whether the high GI meal actually had a lower GL than the low GI meal: this is feasible if (say) the high GI breakfast had been made up of a little ripe banana and a portion of watermelon because both have high GI but, depending on the quantity, might have delivered a low GL meal. Watermelon has a GI of 72 (which is high). However, in a serving size of 120 grams there are only 6 grams of available carbohydrate so the GL is low, 72/100*6=4.32, rounded to 4. Exaggerating the point, a Twix® Cookie Bar (item 1501) comes it at a possible (low) GI of 44±6 and (medium to high) GL of 17 (item 387 porridge oats are GI 49±8 but item 390 oats have a GI of 52±4 with a GL of 17).
A conference paper is does not have the same standing as something that has been published in an appropriate journal after adequate peer-review but it can alert interested parties to some research that might be worth following up. Now, we don’t believe that the researchers provided meals of the sort we discuss above but it does illustrate why a careful researcher or responsible reporter is wary of extrapolating from an abstract to support a point. And, we should recall, Holford sells his subscription service to people with the promise that he is an adequate intermediary between them and scientific research they need to know about in order to improve their health. In this case, Holford has conflated GI and GL and then used a paper about GI to support his contention that his GL diet is desirable because following this way of eating stabilises people’s blood sugar and therefore their hunger and they don’t over-eat.[c]
Added to the above, it is difficult to be excited about a study about a single meal rather than a way of eating that is sustained over weeks or months:[d] it is not always true that something that is true for a particular type of meal in a particular metabolic state is readily generalisable to the wider population and other metabolic states. Crudely, it may not be true that people who normally eat a mixed, unrestricted diet and who have replete glycogen stores, will respond in the same way to a meal as (say) somebody who eats a restricted diet and may have depleted their glycogen stores (say, by following a low carbohydrate diet for some weeks). It is not even necessarily true that men and women respond the same way to particular diets or that the physiological response to diet is the same for different levels of BMI. See, e.g., the sex differences in response reported in: Sex differences in energy homeostatis following a diet relatively high in protein exchanged with carbohydrate, assessed in a respiration chamber in humans.
Effects of a diet relatively high in protein exchanged with carbohydrate, versus an adequate protein diet are a stronger increased energy expenditure, fat oxidation, protein anabolism in men, and a stronger increased satiety in women.
A study suggests that there may be different consequences for overweight women: Glycemic index, cholecystokinin, satiety and disinhibition: is there an unappreciated paradox for overweight women? This small study of single meals reports:
The clinical utility of a low glycemic index (LGI) diet for appetite and food intake control is controversial…The HGI meal resulted in greater satiety overall, suppressing hunger, desire to eat and prospective consumption compared with the LGI (P<0.01) meal…A LGI diet may not be suitable for optimal satiety and appetite control in overweight women.
In a free-living group of 19 overweight and obese women (it should be emphasised that the reported difference in dietary GI levels was only 8.4 units) Aston et al reported: No effect of a diet with a reduced glycaemic index on satiety, energy intake and body weight in overweight and obese women.
This study provides no evidence to support an effect of a reduced GI diet on satiety, energy intake or body weight in overweight/obese women.
An 18-mo randomized trial of a low-glycemic-index diet and weight change in Brazilian women reported no difference in mean weight loss or reported hunger levels between women on a low or high GI diet (difference of 20 GI units).
As a further indication of confounding variables that influence this area of research, for some foodstuffs, it may be necessary to consider how well the food is chewed before presuming to forecast its impact on hormone levels, nutrient absorption etc.: Mastication of almonds: effects of lipid bioaccessibility, appetite, and hormone response. Again, a very small study (13 participants) in healthy, non-obese subjects but some results that are in line with what one might expect about greater chewing exposing more surface area of a food and promoting availability of its nutrients.
Hunger was acutely suppressed below baseline (P < 0.05), and fullness was elevated above baseline longer (P < 0.05) after 40 chews than after 25 chews….Initial postingestive glucagon-like peptide-1 concentrations were significantly lower after 25 chews than after 40 chews (P < 0.05), and insulin concentrations declined more rapidly after 25 and 40 chews than after 10 chews (both P < 0.05)…The results indicate important differences in appetitive and physiologic responses to masticating nuts and likely other foods and nutrients.
Despite the optimism of Holford, there is, as yet, no clear clinical picture that connects or predominantly associates satiety and GLP-1. This was recently highlighted in a study involving another one of Holford’s enthusiasms – cinnamon. 15 healthy subjects experienced increases in GLP-1 but no impact on satiety following consumption of rice pudding: Effects of 1 and 3 g cinnamon on gastric emptying, satiety, and postprandial blood glucose, insulin, glucose-dependent insulinotropic polypeptide, glucagon-like peptide 1, and ghrelin concentrations in healthy subjects.
The addition of 1 or 3 g cinnamon had no significant effect on GER, satiety, glucose, GIP, or the ghrelin response….Ingestion of 3 g cinnamon reduced postprandial serum insulin and increased GLP-1 concentrations without significantly affecting blood glucose, GIP, the ghrelin concentration, satiety, or GER in healthy subjects.
It is not fully established that a low GL diet is associated with greater satiety and less over-eating and the paper that Holford refers to in support of his claims does not support them: predictably it doesn’t even discuss a low GL diet or even (necessarily) a single low GL meal. Just to be absolutely clear about this, it’s not even as if the mistake lies with the King’s College press release upon which so many commentators seem to have relied.
Despite the hyperbole that surrounds GI and GL, there is remarkably little supportive evidence for its usefulness as a dietary recommendation for the general population. There are substantial and well-founded concerns about the inter-individual variation in glycaemic response to individual foodstuffs[d] as well as the variability of the foodstuffs and the evaluation of its GI.[b] Irritatingly, although there are some suggestions that appropriate dietary modification (at least for people with insulin resistance or diabetes) depends upon the knowledge of the Insulin Index or Fructose Index of a food, in addition to its GI, there is not a sufficient corpus of research on either.
The idea that people who want to lose weight while never feeling hungry is a seductive one, however, despite the rhetoric, there is no acceptable body of evidence to support this assertion over the long term. Evidence for clinical, sustained benefit from adopting a low GL way of eating for the general population isn’t available: it is not entirely clear that there is strong and compelling evidence for this way of eating for the chronic conditions such as diabetes type II for which is it so frequently recommended. We’re not sure how obvious that is when reading the books and promotional literature of people who want to sell particular diet plans, diet clubs, diet seminars and diet-supporting supplements to the general population.
As for Holford helping his readers to ‘stay informed’ and keep abreast of the latest in science…Time and time again, Holford’s scholarship is demonstrated to be shoddy and inaccurate. As Professor Tom Sanders remarked, nutritional therapists tend to “feed on the detritus that comes out from the scientific community” – but it doesn’t mean that they assimilate it correctly.
[a] Our particular version of Bad Science Bingo – the Holford edition includes failing to cite the paper under discussion, failing to link to it and misreporting some vital part of the paper’s contents (see the discussion of Khan): other favourites include making up references that don’t exist or citing it in a completely inappropriate context; plus the always exciting accusations of incompetence or a lack of integrity when Holford’s accusations are grounded in his own misunderstandings of the data.
[b] The glycaemic index (GI) measures the increase in circulating blood sugar a fixed amount of carbohydrate triggers relative to a reference food in the same person over a set time period–crudely, the higher the number, the greater the blood sugar response soon after eating it. This is typically characterised as a low GI food triggering a small increase, and a high GI food returning a dramatic spike. For general purposes a GI of 70 or more is high, a GI of 56 to 69 inclusive is medium, and a GI of 55 or less is low.
The glycaemic load (GL) is an appropriately nuanced way to evaluate the impact of carbohydrate consumption that uses GI and facilitates a better theoretical understanding of the impact of a foodstuff on blood sugar levels because it takes into account the amount of carbohydrate in a serving of a particular foodstuff. As in the above example, watermelon has a high GI but a low GL in the quantities in which it is typically eaten. A low GL foodstuff will usually have a low GI but it doesn’t automatically follow that a low GI foodstuff is eaten in sufficiently small quantities that it is always low GL.
It should be remembered that the GI of a foodstuff can vary with season, age, country of origin, preparation method etc. and that there are substantial inter-subject differences in response even though the database of GI values is drawn from 8–12 healthy subjects. Recently, the problem of considerable variation in assigning GI values to foods was highlighted: Methodological Challenges in the Application of the Glycemic Index in Epidemiological Studies Using Data from the European Prospective Investigation into Cancer and Nutrition. The authors acknowledge that such difficulties impede researchers who need meaningful GI/GL estimates for nutritional epidemiology.
[c] Unfortunately, even when it is clear that there are statistically significant outcomes from modifying the GI or GL of a diet, it is not necessarily true that these differences are biologically or clinically relevant.
Despite Holford’s assertions to the contrary, there is no automatic equivalence between post-prandial glucose and insulin surges and chronically elevated glucose and insulin. Although chronically elevated insulin may be a marker of metabolic dysfunction (in the absence of known side-effects from various drugs, disease etc.), it does not follow that post-meal insulin surges are (of course, in the absence of exculpatory factors, regular blood sugar spikes in excess of 140 mg/dL raise a strong index of suspicion for glucose intolerance). Marmot’s Whitehall Study reported that for healthy, middle-aged men and women: Dietary glycemic index and glycemic load are associated with high-density-lipoprotein cholesterol at baseline but not with increased risk of diabetes in the Whitehall II study.
Higher dietary GI and GL were not associated with an increased risk of incident diabetes in this population after 13 y of follow-up. On the contrary, high-dietary GL was associated with decreased risk of diabetes…These findings are consistent with a possible protective role of dietary factors on disease risk, but they do not support the hypothesis that high-dietary GI and GL may be a risk factor for type 2 diabetes. In this population, participants with high-dietary GI and GL had slightly lower BMIs, but this did not appear to mask a possible increased risk of diabetes associated with such diets.
[d] It is unavoidable that researchers need to examine the impact of single meals. However, for free-living and healthy individuals, it must be remembered that fat, protein, acidity may delay gastric emptying or slow carbohydrate absorption in a mixed meal. There are far too few data collections of the impact of individuals eating mixed meals over extended periods of time.
Interestingly, a recent study did report some small scale results from assessing in Role of glycemic index and glycemic load in the healthy state, in prediabetes, and in diabetes.
postprandial plasma glucose concentrations were evaluated in a group of patients with type 2 diabetes in response to isoglucidic portions (50 g available carbohydrate) of 3 starch-rich foods: 65 g spaghetti, 90 g white bread, or 285 g potatoes; each was consumed on different days, in random order, as part of a standard meal (46). This study showed that when spaghetti was replaced by a portion of bread containing the same amount of available carbohydrate, the glycemic response (evaluated for 5 h after the meal) increased by as much as 68%. When spaghetti was replaced by potatoes, the glycemic response was 48% higher. These differences cannot be accounted for by variations in any of the food constituents known to influence blood glucose metabolism, because the amount of simple and complex carbohydrate as well as the amount of dietary fiber were similar in the 3 test meals. The results of this study are particularly relevant because, unlike most other studies on this subject, the evaluation of the glycemic response of the different foods was performed within a composite meal. This guarantees that the results are applicable to everyday life.
Significant differences in insulin score were found both within and among the food categories and also among foods containing a similar amount of carbohydrate…However, protein-rich foods and bakery products (rich in fat and refined carbohydrate) elicited insulin responses that were disproportionately higher than their glycemic responses.
Is the fructose index more relevant with regards to cardiovascular disease than the glycemic index? The authors suggest that the G.I. is better aimed at identifying foods that stimulate insulin secretion rather than foods that stimulate insulin resistance. They argue that
it is the ingestion of foods that induce insulin resistance that carries the increased risk for obesity and cardiovascular disease and not eating foods that stimulate insulin secretion.