Current issues of ACP Journal Club are published in Annals of Internal Medicine


Insulin led to lower glucose levels, less combination therapy, and more weight gain in type 2 diabetes mellitus

ACP J Club. 1998 Jul-Aug;129:12. doi:10.7326/ACPJC-1998-129-1-012

Source Citation

United Kingdom Prospective Diabetes Study 24: a 6-year, randomized, controlled trial comparing sulfonylurea, insulin, and metformin therapy in patients with newly diagnosed type 2 diabetes that could not be controlled with diet therapy. United Kingdom Prospective Diabetes Study Group. Ann Intern Med. 1998 Feb 1;128:165-75.



In patients with newly diagnosed type 2 diabetes mellitus whose disease cannot be controlled by diet therapy, which treatment is most effective: sulfonylurea, insulin, or metformin?


Randomized controlled trial with 6-year follow-up.


Outpatient diabetes clinics of 15 hospitals in the United Kingdom.


From a study of 4075 diabetic patients, 560 patients (mean age 50 y, 53% men) with "diet failure" were analyzed if they had newly diagnosed type 2 diabetes mellitus with fasting plasma glucose levels > 15 mmol/L or persistent symptoms of hyperglycemia after 3 months of diet therapy. Exclusion criteria included ≥ 1 previous major vascular episode, serum creatinine level > 175 µmol/L, severe retinopathy, malignant hypertension, an uncorrected endocrine abnormality, and extensive systemic therapy. Follow-up was 82%.


After stratification for obesity (defined as body weight > 120% of ideal), patients were allocated to intensive treatment with insulin (n = 178, 29% were obese); sulfonylurea (chlorpropamide or glyburide; n = 231, 31% were obese); or metformin (n = 49, 100% were obese). The doses were adjusted to obtain a fasting plasma glucose level < 6.0 mmol/L; additional therapy was given if that level could not be achieved with monotherapy.

Main outcome measures

Changes from baseline in fasting plasma glucose levels, hemoglobin A1c levels, and body weight; hypoglycemic episodes; and the number of patients requiring additional therapy at 6 years (i.e., failing monotherapy).

Main results

Patients allocated to insulin had a greater decrease in fasting plasma glucose levels over 6 years than did those allocated to sulfonylurea (mean decrease [MD] 7.9 mmol/L vs 6.1 mmol/L, P < 0.001) or metformin (MD 7.9 mmol/L vs 5.2 mmol/L). Changes in hemoglobin A1c levels did not differ between treatments. Compared with sulfonylurea, insulin led to a greater increase in body weight (mean increase 9.9 kg vs 5.3 kg, P < 0.001) and to more hypoglycemic events (52% vs 16% for chlorpropamide and 21% for glyburide, P < 0.001). More patients allocated to insulin than to sulfonylurea (49% vs 34%, P < 0.01) or metformin (49% vs 35%, P < 0.01) were receiving monotherapy at 6 years.


In patients with newly diagnosed type 2 diabetes that could not be controlled by diet, initial insulin therapy led to lower fasting glucose levels, greater weight gain, more hypoglycemic events, and less need for additional therapy at 6 years than initial therapy with sulfonylurea or metformin.

Sources of funding: United Kingdom Medical Research Council; British Diabetic Association; United Kingdom Department of Health; National Institutes of Health; British Heart Foundation; Health Promotion Research Trust; Clothworkers' Foundation; Charles Wolfson Charitable Trust; Alan and Babette Sainsbury Trust; Oxford University Medical Research Fund Committee; in part, Novo Nordisk, Bayer, Bristol-Myers Squibb, Hoechst, Lilly, Lipha, Farmitalia Carlo Erba.

For correspondence: United Kingdom Prospective Diabetes Study Group, Diabetes Research Laboratories, Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE, England, United Kingdom. FAX 44-1865-723884.


The United Kingdom Prospective Diabetes Study (UKPDS) provides an opportunity to answer several fundamental questions about the management of type 2 diabetes. In this report, the authors describe the metabolic responses to and adverse effects of 3 common therapies. The patients included in this analysis are the diet failure group and represent only 14% of the originally recruited UKPDS cohort. Thus, they are a distinct subset and many are probably patients with type 1 diabetes that presented in adulthood. Nevertheless, this report provides useful comparisons among different therapies for patients whose diabetes cannot be controlled by diet and shows both the benefits and problems of insulin therapy.

It is surprising that the improvement seen in fasting plasma glucose levels in the patients allocated to insulin was not mirrored by a similar improvement in hemoglobin A1c levels. This discrepancy probably reflects the insulin protocol used: ultralente insulin once daily. Of note, the lack of weight gain in patients allocated to metformin and the increased risk for hypoglycemia in patients allocated to insulin are consistent with current clinical experience. Ultimately, the effect of glucose control on the long-term complications of diabetes will be the determining factor for evaluating the risk-benefit ratio for each of these therapies and for deciding which is the best therapy for patients with type 2 diabetes. This question is being addressed by the full UKPDS study, the reports from which are due in the fall of 1998.

Bernard Zinman, MD
University of TorontoToronto, Ontario, Canada