California State University, Fresno Equine Program

Glycemic Index and Low Glycemic Horse Feeds

Anne Rodiek, Ph.D.

Californa State University, Fresno

Needs for low or moderate glycemic diets in humans

The glycemic index (GI) for foods was developed in the 1980s (Foster-Powell et al, 1981) to reduce the compromise of glucose homeostasis in many people advised to use low fat (hence high carbohydrate) diets advocated by the American Heart Association and other public health agencies (Kronfeld et al, 2004).  Chemical analyses of food did not accurate predict the glycemic response to the food, so glycemic indices were determined based on physiological responses to the foods, that is, changes in blood glucose concentration over time after ingestion of a specific food.  The area under the blood glucose response curve for a test food was measured and compared to the area under the glucose response curve for a standard food (originally a determined amount of glucose but later white bread).  The GI of any given food was calculated and expressed as a percentage of the glucose response to the standard food.

Research has shown, in population studies, a relationship between high glycemic diets and obesity, insulin resistance, coronary heart disease and possibly cancer, particularly in older people who are overweight and maintain a sedentary lifestyle (Jenkins et al.  2002).

Glycemic load (GL) is more useful than GI when formulating low glycemic diets as it quantifies the glucose “load” of a single food or an entire mixed meal.

Human dieticians calculate the glycemic load of a food by the following equation: 

GL = GI x available carbohydrate (g) in a single serving of a given food

The GL of a mixed meal or daily diet is calculated by simple addition.  It may be divided by the total available carbohydrate to yield a GI for meal or diet.

Needs for low or moderate glycemic diets for horses

Formulating equine diets that produce attenuated glycemic and insulemic responses may have health benefits to the horse in avoiding or ameliorating insulin resistance and its undesirable effects related to laminitis and other metabolic conditions (Kronfeld et al, 2005).

Less research has been done to show similar relationships between high glycemic diets and comparable equine conditions as in humans, but limited research and anectodal observations indicate that older, overweight (Hoffman et al, 2003), sedentary horses show a higher incidence of insulin resistance, laminitis and other metabolic maladies than horses that are young, active and of normal (not overweight) weight.  Therefore, it appears that low glycemic diets are more indicated for horses in the former group than in the latter. Whether such rations will meet nutrient requirements for horses with high energy requirements seems likely but remains to be demonstrated.

Glycemic index and limitations in GI of horse feeds

Glycemic responses to different horse feeds from several studies were compiled by Kronfeld et al (2004) and are adapted here in Table 1.  Notable in Kronfeld’s table, but not shown here is the large variation in glycemic indices for similar foods, which is also seen in the human GI data.  Some of the variation is, no doubt, caused by differences in research protocols for determining GI and other variation is caused by different feed compositions, feed processing and likely also different physiological responses of the research horses. 

Small differences in GI between different feeds may not be physiologically meaningful and it may be most useful, from a practical standpoint, to divide feeds into only 3 categories (and perhaps not make differentiations between feeds within categories) based on GI.  In human nutrition, foods with GIs greater than 70 are considered to have a high GI, foods with GIs between 55 and 69 and medium and foods with GIs less than 55 are considered to have a low GI.

Given the variation in GI in horse feeds, ration formulation based on GI requires taking huge liberties with the data and ignoring the physiological as well as statistical differences in the data.   

While pre-cecal carbohydrate digestion in horses and humans is similar, the main dietary ingredients of the two species are not.  Most horses eat, as the main component of their diet, highly fibrous feeds that are primarily fermented in the large intestine.  These feeds (hay, pasture, silage) are much higher in insoluble carbohydrate than even the most fibrous of (human) vegetarian diets and therefore are likely lower in total soluble carbohydrate, fat or protein content.  As such, it may be conceivable that quite low GI diets may be formulated for horse, at least lower than 55, although very large variation in GI for some forages has been seen.

¹ adapted from Kronfeld et al, 2004

² arithmetic mean from values to left

^a Stull and Rodiek, 1988

^b Pagan et al, 1999

^cGroff et al, 2001

^d Rodiek, 2003

^e Williams et al, 2001

^f Jose-Cunilleras et al, 2004

^g Rodiek and Stull, 2005

Carbohydrate analysis and calculating Glycemic loads of horse feeds

In order to use the GL formula described above, human nutritionists must know the glycemic index of a food, its available carbohydrate content and the amount of food in a single serving.  Horse nutritionists struggle to implement this formula, as while there is now initial, albeit variable data on the GI for various common horse feeds, there are no consistent, available values for the available carbohydrate for feeds, and there is the practical matter of determining the weight of a “single serving” of a feed for a horse.

Available carbohydrate, as used in the glycemic load formula, means the amount of carbohydrate that will be digested in the small intestine and will contribute to the amount of glucose that enters the circulation from the small intestine.  A specific assay for the determination of available carbohydrate in horse feeds is not commercially available at this time. Soluble carbohydrate (non-structural or non-fiber carbohydrate) is sometimes used to estimate “available” carbohydrate, but present assays do not separate the carbohydrate that is “available” for enzymatic (small intestine) digestion or for microbial (large intestine) digestion. Both contribute energy to the horse, and under circumstances may increase or decrease the incidence of laminitis, but only enzymatically digested carbohydrate will contribute to the blood glucose concentration of a food or meal.  Enzymatic laboratory assays that separate rapidly from slowly available glucose have been shown to be highly correlated with blood glucose response, but these assays are not commercially available at this time (Englyst et al, 1999).

The determination of GI of a mixed meal as a weighted average of the glycemic load contributions of other feeds is also problematic. Glycemic loads are not additive in a linear fashion   Different nutrients interact with each other in non-linear ways; the extent of non-linearity is confounded by many factors.  The food factor that likely most powerfully confounds mathematical calculation of glycemic loads in mixed diets is the addition of fats and oils.  Fats diminish glycemic effects of feeds to a variable degree, presumably by their effect to slow rate of passage of the meal through the stomach and small intestine, thereby decreasing the rate at which glucose enters the blood.  On the other hand, molasses substantially increases the glycemic index of feeds, even when added at relatively small amounts (ie, 10%).

GL_en and GL_total of horse feeds

Many equine nutritionists use digestible energy (DE, Mcal/kg or lb) as a basis for ration formulation, meeting energy requirements first in the formulation process as this is the largest requirement.  DE is not a measure of carbohydrate content of feed, but there is merit in its use for determination of GL of horse feeds:  1. it is a widely recognized measure of energy, and 2. it solves the problem of “serving size” as it measures energy/kg of feed.  This allows the calculation of glycemic load/kg or lb of feed or ration instead of glycemic load per ambiguous unit of “serving size”. 

Using DE, a calculation of GL_en for a food would be accomplished as:

GL_en = GI x DE (Mcal/kg). 

The GL for a total ration (GL_total) could be calculated as:

GL_total = GI x DE (Mcals) contributed by each component feed in a ration.

Formulating low GI diets for horses

Low glycemic feeds such as hays (grass and legume), rice bran and beet pulp may be used, with proper supplementation, to formulate adequate diets for horses with low nutrient requirements, primarily adult horses in maintenance condition or performing only light, recreational work.  This population comprises a large percentage of the equine population.  These same feeds may also meet nutrient requirements for gestating mares under most conditions.

Using a generic DE value of 2.0 Mcal/kg for hay and a mean GI value for alfalfa (mean of 29, but range from 12 to 45 in Table 1) or bermudagrass hay (23, only one value in Table 1), one could calculate GL values for these feeds of approximately .45 to .6 units of “glycemia”/kg of feed (GL_en = ~ .23 or .29 x 2.0 Mcal/kg).  Compared to other feeds, these GL values are low, but so are the energy contributions to meeting the DE requirement of the horse.  Using a DE requirement of 16.4 Mcals/day (NRC, 1989) for a 500 kg horse in maintenance condition, the energy requirement would be met with 8.2 kgs (1.6% of body weight) of feed and approximately 4 glycemic units in the total ration (GL_total = ~.25 x 16.4 Mcals DE)  Crude protein requirements would also likely be met.

A 500 kg mare in the last third of gestation requires 19.7 Mcals DE (NRC, 1989).  This DE requirement would be met with slightly less than 10 kg of hay (2 Mcal/kg) and a GL_total of 5 glycemic units.

As DE requirements increase, so does glycemic load and also likely energy density of the ration.  A 500 kg mare in early lactation requires 28.3 Mcals (NRC, 1989), which would require 14 kgs of hay (2.8% of body weight), an amount even a voracious horse would be challenged to eat.  Protein requirements would be met by alfalfa hay but not by bermudagrass (or most grass) hays.  Glycemic load would be approximately 7 glycemic units.

How can energy density be increased while keeping GL low to meet the DE requirements of the lactating mare or hard working horse?  Traditional high energy diets would include grain supplementation above a basal level of hay, provided as a constant weight, energy level or percentage of body weight.  If, for example, hay was provided at the rate of 1% of body weight (5 kgs), hay could provide, on average, 10 Mcals DE to the daily ration.   In the case of the lactating mare, the additional 18.3 Mcals would be provided by grain.  Oats, for example, provides 3.2 Mcals DE/kg and has a glycemic index of 100.  18.3 Mcals of DE would be provided by 5.7 kgs of oats.  The hay portion of the diet produces 2.5 glycemic units.  The oats portion provides 18.3 glycemic units and GL_total becomes 20.8 glycemic units.  How can this ration be reformulated to provide the same amount of energy but with a lower glycemic effect?

A replacement of oats with rice bran with a mean GI of approximately 16 (range of 11 to 21) and 2.9 Mcal DE/kg would provide 18.3 Mcals of DE in 6.3 kgs with a GL of 3 glycemic units (vs. 18) and a GL_total of 5.5 instead of  20.8 glycemic units.  Partial replacement of oats with rice bran would reduce GL in ways that are not linear, but in a downward direction. .  Most equine nutritionists would not advocate a ration composed of solely hay and rice bran as being nutritionally adequate.  Certainly, other nutrients must be considered in the formulation of this ration (protein, vitamins, minerals), but this example shows how glycemic effects can be changed with feed choice.

A more commonly seen ration would replace a portion of the grain in the ration with fat or oil.  Vegetable oil contains approximately 9 Mcals DE/kg, almost three times the energy density of oats.  One-half kg of vegetable oil could replace 1.4 kgs of oats on an energy equivalent basis.  A ration of 5 kgs hay, 3.7 kgs oats and .5 kg vegetable oil would meet the DE requirement of the early lactating mare with a GL_total of 6.2 glycemic units, not accounting for the depression of glycemic response due to the addition of fat to the ration.

Grain based feeds may be altered to decrease their glycemic effects by the addition of fat or low glycemic feeds.  Partial replacement of grains with beet pulp or forage pellets and partial replacement of molasses with oil maintains the “sweet feed” palatability and energy density of grain mixes while producing a lower glycemic load.  Feeding small meals frequently will decrease the glycemic load of any one meal and help stabilize glucose and insulin concentrations over time.

Feeds commonly used to produce diets with low or moderate glycemic and insulinemic effects may also confer other health benefits, outside of direct effects on glucose and insulin, including:

1. decreased likelihood of soluble carbohydrate overload in the large intestine.
2. increased bulk in the diet which may have beneficial effects on gastrointestinal health through stabilization of pH and improved physical and nutritive environment for cells lining the intestines as well as increased water holding capacity of the digestive tract contents, and
3. rations that are more “natural”, being higher in fiber, less nutrient dense and more similar to diets consumed by undomesticated, non-confined horses.

 REFERENCES

Englyst, K. N., Englyst, H. N., Hudson, G.J., Cole, T.J., and Cummings, J.H. 1999.  Am J. Clinical Nutr 69(3):448-454).

Foster-Powell, K. S.H.A. Holst, J.C. Brand-Miller.  International table of glycemic index and glycemic load values. 2002.  Am J Clin. Nutr. 1981;76:5-56

Groff, L, J. Pagan, K. Hoekstra, et al.  2001.  Effect of preparation method on glycemic response to ingestion of beet pulp in Thoroughbred horses.  Proc. Equine Nutr. Physiol. Soc. 17:125-6.

Hoffman, R.M., R.C. Boston, D. Stafanovski, D.S. Kronfeld  and P.A. Harris.  2003.  Obesity and diet affect glucose dynamics and insulin sensitivity in Thoroughbred geldings.  J. Anim. Sci. 81:2333-2342

Jenkins et al.  2002.  Glycemic index:  overview of implications in health and disease. Am. J. Clin. Nutr. 75(1):266s-273s

Jose-Cunilleras, E., L.E. Taylor, K.W. Hinchcliff.  2004. Glycemic index of cracked corn, oat groats and rolled barley in horses.  J Anim. Sci 82:2623-2629  

Kronfeld, D.S., K.H. Treiber, T.M. Hess and R.C. Boston.  2005.  J. Anim. Sci 83(E. Suppl.):E22-E31.

Kronfeld, D., A. Rodiek and C. Stull.  2004.  Glycemic indices, glycemic loads and glycemic dietetics.  2004.  J. Equine Vet. Sci. 24(9): 399 – 404. 

National Research Council. 1989. Nutrient Requirements of Horses.  5^th Revised Ed. National Academy Press.  Washington, D.C.

Pagan, J.D., P.A. Harris, M.A.P. Kennedy et al.  1999. Feed type and intake affects glycemic response in Thoroughbred horses.  Proc. Equine Nutr. Physiol. Soc. 16:174-5. 

Rodiek, A. and C. Stull. 2005.  Glycemic index of common horse feeds.  Proc. Equine Sci. Soc. P. 153.

Rodiek, A.V.  Glycemic index of practical horse feeds.  2003. ARI Project No. 00-2-024 (Research focus area:  Production and cultural practices.  California State University, Fresno, CA

Stull, C.L and A.V. Rodiek.  1988.  Responses of blood glucose, insulin and cortisol concentrations to common equine diets.  J. Nutr. 118:206-22

Williams, C.S., D.S. Kronfeld, W.B. Staniar, P.A. Harris.  2001. Plasma glucose and insulin responses in Thoroughbred horses fed a meal high in starch and sugar or fat and fiber.  J. Anim. Sci. 79:2196-201.

UAL Home | Calendar | UAL Directory | Maps | Comments  | CAST Home  |  University Home

2385 E. Barstow  · MS AG85
Fresno, CA  93740
559/278-2011 · Fax: 559/278-6998