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


Meta-analysis: Insulin plus a sulfonylurea agent only slightly improves glycemic control in NIDDM

ACP J Club. 1991 Nov-Dec;115:75. doi:10.7326/ACPJC-1991-115-3-075

Source Citation

Peters AL, Davidson MB. Insulin plus a sulfonylurea agent for treating type 2 diabetes. Ann Intern Med. 1991 Jul 1;115:45-53.



To review randomized trials comparing insulin plus sulfonylurea agents (SA) with insulin alone in patients with non-insulin-dependent diabetes mellitus (NIDDM).

Data sources

The key words "insulin" and "sulfonylurea agents" were included in a computerized search of the MEDLINE database for the years 1976 to 1990. Additional articles were retrieved using references from articles, a manual search of Index Medicus from 1986 to 1990, and an investigator's files.

Study selection

Randomized trials and crossover studies (with treatment order randomized) of SA plus insulin compared with insulin alone (or with placebo) were included if patients had unsatisfactory blood glucose control with insulin or had recently started insulin. In addition, fasting blood glucose or glycated hemoglobin, or both, were measured before and after therapy for randomized trials and at the end of each treatment phase for crossover studies. 8 concurrent randomized trials and 14 crossover studies were reviewed.

Data extraction

Data were extracted by one investigator, on patient characteristics, duration of follow-up, mean fasting blood glucose concentration, and fasting or stimulated C-peptide levels.

Main results

The number of patients included in individual studies ranged from 6 to 37; mean ages ranged from 51 to 73 years. Mean follow-up in the randomized trials was 27 weeks; mean duration of each arm of the crossover studies was 10.6 weeks. Results across studies were calculated separately for randomized trials and crossover studies. In the randomized trials, insulin plus SA lowered the mean fasting blood glucose concentration from 11.3 to 9.4 mmol/L (glycated hemoglobin fell from 10.7% to 10.0%), whereas insulin plus placebo lowered the mean fasting blood glucose concentration from 11.6 to 10.7 mmol/L (glycated hemoglobin increased from 10.8% to 11.1% in this study arm). Similar results were obtained from the crossover studies, with a mean fasting blood glucose concentration of 9.0 mmol/L and glycated hemoglobin concentration of 9.8% at the end of the insulin plus SA period and 11.0 mmol/L and 10.6%, respectively at the end of the insulin plus placebo period. In most studies, fasting or stimulated C-peptide levels, or both, were higher and insulin doses were lower in patients receiving insulin plus SA compared with those receiving insulin plus placebo.


Combination therapy with insulin and a sulfonylurea agent only slightly improves glycemic control in patients with non-insulin-dependent diabetes mellitus. Other approaches, such as increased dosage of insulin, are preferred to achieve nearly normal blood glucose control in these patients.

Source of funding: Not stated.

Address for article reprint: Dr. M.B Davidson, Division of Endocrinology, Room B-131, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, CA 90048, USA.


Combination insulin and sulfonylurea therapy has been gaining clinical acceptance for the treatment of NIDDM despite minimal evidence of its efficacy over insulin alone. It is usually used in patients who, despite "large" doses of insulin, have poor glycemic control. Peters and Davidson reported a thorough search for randomized trials of combined therapy and correctly summarized the data as weighted means. They concluded that the observed improved glycemic control is clinically unimportant and, given the possible side effects and cost, combination therapy should not be used.

I might conclude otherwise. First, combination therapy improved glucose levels in all of the parallel-group trials and in 11 of 13 of the crossover trials. All but 1 study showed improved glycosylated hemoglobin percentages. If one combines these effect sizes statistically, they are significant. Second, the consistency of the effect is striking, given such small patient groups and such diverse settings, patient populations, and medication dosages. Finally, some patients had excellent responses, although no clear predictors of response were identified. Therefore, overall, these small "preliminary" studies show a consistent, albeit clinically small, improvement in glucose control. Given the difficult patient populations studied (for which innovative treatments are needed), combination therapy needs to be studied in greater detail, with more emphasis on predicting who will respond.

I use combination therapy only when patients have not achieved adequate control despite approaching 100 units for either of their insulin doses (which means they will have to use 2 syringes for 1 dose of insulin). If they respond favorably, they are kept on the sulfonylurea. If they do not achieve a response with maximal dose by 3 months, I discontinue it. In this way, I can explore whether they may respond while avoiding exposing them to long-term side effects or costs if they do not respond.

Jacqueline A. Pugh, MD
Audie L. Murphy Memorial Veterans HospitalSan Antonio, Texas, USA