Low-carb vs High-fiber diet

COMPARISON OF A LOW CARBOHYDRATE VERSUS HIGH FIBER DIET
IN CATS WITH DIABETES MELLITUS

From the 2001 American College of Veterinary Internal Medicine Forum in Denver, Colorado, May 23-26. See abstracts at http://www.acvim.org/wwwfp/forum2001/Abstracts.pdf

Abstract #103

COMPARISON OF A LOW CARBOHYDRATE VERSUS HIGH FIBER DIET IN CATS WITH DIABETES MELLITUS, N Bennett, DS Greco, ME Peterson, Colorado State University, Fort Collins, CO and The Animal Medical Center, New York, NY.

The purpose of this study was to determine whether a low carbohydrate (LC) or a high fiber (HF) diet was more effective in the management of diabetes mellitus in cats. Thirty client-owned cats with naturally- occurring diabetes mellitus were randomly assigned to receive a LC diet (canned Hill's feline growth, n=17) or a HF diet (canned Hill's w/d, n=13) for 4 months. None were ketotic and all were previously treated with insulin (NPH, median dose 5 U). A complete physical examination, history, CBC, chemistry profile, serum fructosamine, TT4 and urinalysis were evaluated on a monthly basis. Mean body weight prior to entrance in the study was 5.85 kg. Mean±SE serum fructosamine (487±45 µmol/L) and fasting blood glucose concentrations (343±39 mg/dl) were increased prior to treatment in most cats. Four cats were subsequently switched from HF to LC at the owner's request or because of continued clinical signs. Eight cats (4 LC, 4 HF) were eliminated from the study because of concurrent disorders (acromegaly-LC, heart failure-LC, leukemia-HF, pancreatitis-LC) or dietary non-compliance (1LC, 3 HF). Cats were treated with protamine zinc insulin at an initial dosage of 1-3U/cat BID.

All cats in the LC group and 3 cats in the HF group showed improvement in clinical signs (PU/PD, plantigrade stance). Responders tended to lose body weight during the study and non-responders or cats remaining on insulin tended to maintain or increase body weight. Response, defined as a normal serum fructosamine <400 µmol/L and fasting blood glucose <170 mg/dl respectively, occurred in 12/13 cats in the LC group and in 3/9 cats in the HF group. By the end of the study, 4 cats in the LC group were able to discontinue insulin entirely, 6 cats experienced a reduction in insulin dosage to 1 unit BID and three cats remain on 1.5-3 units BID. None of the cats in the HF group were able to discontinue insulin entirely and median insulin dose for the HF group was 3 units BID.

My comments as a statistician:

Defining the response to treatment as a yes/no variable (such as "fasting blood glucose less than 170") rather than a continuous variable (such as "mean fasting blood glucose") is a recipe for failure with such a small sample size, unless the difference is dramatic. The fact that the researchers were able to see such strong differences (p=.0066 with Fisher's exact test, without even taking into account the 4 cats who were taken off the HF diet due to owner request or continued symptoms and the 3 cats who didn't like the diet) is pretty impressive.

These results are consistent with the Purina DM experiment, which showed that of 14 cats, 6 went off of insulin entirely, and another 7 had their insulin requirements reduced.

More comments: I have received communications from two people whose cats participated in the study, and whose experience does not match the description in the abstract. Specifically, neither cat was on Humulin NPH at the start of the study, and one cat never had urinalysis done. More serious, since it's related to the outcome, is that they reported that fasting blood glucose was not taken: they were not instructed to fast their cats, and took them in for testing after the insulin shot was already given, generally during the working day. Presumably the reported results of reduced insulin requirement and reduced fructosamine still stand, which would support the results that blood glucose in general was lower, but these reports from participants are disturbing.

-- Janet (& Binky)