Dietary Treatment of Diabetes Mellitus in the Pre-Insulin Era

Frederick Allen, Elliott Joslin, Louis “Harry” Newburgh, and Phil Marsh were pioneers in the treatment of diabetes before the discovery of insulin. This 2006 review analyzes their archived case histories to provide insight into the makeup and effectiveness of the diets they prescribed to their patients.

Allen first studied insulin in dogs, where he observed that in an animal whose pancreas had been removed (1), a high-carbohydrate diet would lead to glycosuria — the excretion of glucose in the urine (2) — while a high-fat diet would not. His diet for human patients followed accordingly. Diabetic patients were instructed to fast for as long as was required to remove all glycosuria. Carbohydrate was then slowly introduced until glycosuria returned; once this carbohydrate “ceiling” was discovered, protein was introduced, again until glycosuria returned. Allen then added fat and observed that fat had little effect on glycosuria.

Allen’s patients were prescribed highly variable diets when he discharged them. This was because he found some patients were able to tolerate very large amounts of carbohydrate and protein while others were able to consume no carbohydrate at all without the return of glycosuria. The mean discharge diet, however, fit squarely within the diets that would later be prescribed by Dr. Robert Atkins and other low-carbohydrate advocates — 38.7 grams of carbohydrate per day, 85.3 grams of protein, and 156.4 grams of fat (3). This mean diet provided 1,955 daily calories, 70% of which came from fat, 18% from protein, 8% from carbohydrate, and the remainder from alcohol (4).

The diet Joslin recommended to his patients was very similar. Joslin saw carbohydrates as “of no use to the body” in the diabetic patient. He encouraged patients to eat low-carbohydrate vegetables and “all the cream, butter and fatty food possible.” This diet, which included green leafy vegetables and other low-carb greens, similarly derived 70% of its calories from fat, 20% from protein, and 10% from carbohydrate (5).

The “Michigan Diet” promoted by Newburgh and Marsh prescribed 1,400 calories per day, 980 of which (70%) came from fat and only 60-80 of which (< 6%) came from carbohydrate. In other words, they recommended a diet with a similar relative composition to those recommended by Joslin and Allen (6).

The authors of this review note that, while the discovery of insulin was miraculous and saved many lives (7), the modern use of exogenous insulin as a primary therapy for diabetes may be misplaced. Standard-of-care treatment for diabetes has failed to prevent high rates of diabetic complications (8). It also often fails to even normalize blood glucose levels in many patients (9). Moderate alternatives to carbohydrate restriction, such as low-glycemic-index diets, have been only modestly effective in most patients (10). The historical evidence from Allen, Joslin, Newburgh, and Marsh’s work — particularly when placed alongside modern evidence — is unequivocal. It provides strong support for the use of low-carbohydrate diets to establish glycemic control in diabetic patients without the use of insulin.