Share this article on:

Metformin Compared With Glyburide in Gestational Diabetes: A Randomized Controlled Trial

Moore, Lisa E. MD; Clokey, Diana RPH, CDE; Rappaport, Valerie J. MD; Curet, Luis B. MD

doi: 10.1097/AOG.0b013e3181c52132
Original Research
Annual Awards

OBJECTIVE: To compare the efficacy of metformin with glyburide for glycemic control in gestational diabetes.

METHODS: Patients with gestational diabetes who did not achieve glycemic control on diet were randomly assigned to metformin (n=75) or glyburide (n=74) as single agents. The primary outcome was glycemic control. Secondary outcomes were drug failure rate and neonatal and obstetric complications.

RESULTS: In the patients who achieved adequate glycemic control, the mean fasting and 2-hour postprandial blood glucose levels were not statistically different between the two groups. However, 26 patients in the metformin group (34.7%) and 12 patients in the glyburide group (16.2%) did not achieve adequate glycemic control and required insulin therapy (P=.01).

CONCLUSION: In this study, the failure rate of metformin was 2.1 times higher than the failure rate of glyburide when used in the management of gestational diabetes (95% confidence interval 1.2–3.9).



In women with gestational diabetes, the failure rate of metformin to control glucose levels adequately is 2.1 times greater than that of glyburide.

From the Department of Obstetrics and Gynecology, Division of Maternal–Fetal Medicine, University of New Mexico School of Medicine, Albuquerque, New Mexico.

Corresponding author: Lisa E. Moore, MD, Department of Ob-Gyn, Division of Maternal–Fetal Medicine, University of New Mexico School of Medicine, MSC 10 5580, 1 University of New Mexico, Albuquerque NM 87131-0001; e-mail:

Financial Disclosure: The authors did not report any potential conflicts of interest.

Gestational diabetes mellitus (GDM) affects approximately 4% of all pregnancies. These pregnancies are at risk for a variety of fetal complications, including macrosomia and hypoglycemia. Maternal complications include risk of cesarean delivery and hypertensive disease.1–3 Tight control of blood glucose has been shown to reduce the occurrence of these complications.4,5

The standard of treatment for GDM not controlled by diet and exercise is insulin. Recent studies have looked at the feasibility of using oral agents for GDM treatment. In 2000, Langer et al6 published a randomized trial using glyburide in patients with GDM. This study demonstrated that glyburide, a second-generation oral sulfonylurea, controlled blood glucose as well as insulin in a selected population, with no increase in adverse neonatal outcomes.6

Another oral agent commonly used for diabetic control in the nonpregnant population is metformin. Metformin is an oral agent in the biguanide class and works by increasing sensitivity to insulin. It is a logical choice for the treatment of gestational diabetes which is characterized by relative insulin insensitivity. Rowan et al7 compared metformin alone, or supplemented with insulin as necessary to control blood glucose, with insulin alone. They found no increase in perinatal complications due to metformin and additionally demonstrated that women preferred metformin to insulin injections.7

We hypothesized that metformin and glyburide would show equivalent efficacy in controlling blood glucose in a head-to-head comparison. The goal of this study was to compare blood glucose levels in patients taking metformin to blood glucose levels in patients taking glyburide for management of gestational diabetes. The primary outcome was achievement of glycemic control defined as a fasting blood glucose of 105 mg/dL or less and a 2-hour postprandial blood glucose of 120 mg/dL or less.

Back to Top | Article Outline


This study was conducted at the University of New Mexico in Albuquerque, New Mexico, between July 2003 and May 2008. Pregnant women receiving prenatal care were screened using a 50-g glucose load. Women with a 1-hour glucose level of 130 mg/dL or more were then given a 3-hour 100-g glucose tolerance test. Using Carpenter and Coustan guidelines, two or more abnormal values were considered diagnostic of GDM.8 All women were given counseling on diet and exercise and were initially treated with diet. Women who did not maintain fasting blood glucose less than 105 mg/dL or 2-hour postprandial blood glucose less than 120 mg/dL with no other exclusion criteria were offered participation in the study. Exclusion criteria were a history of significant renal or hepatic disease, chronic hypertension necessitating medication, or substance misuse. Patients were between 11 and 33 weeks at the time of randomization. Women were assigned to receive either metformin or glyburide by sequentially labeled, opaque sealed envelopes that were arranged by a computer-generated random list using as parameters a two-arm study with equal numbers in each arm. After written informed consent was obtained, the next envelope was selected by the diabetes educator. Each participant received a copy of the signed consent form, which included contact numbers to report questions or concerns about the study. Study participants and care providers were not blinded to group assignments. This study was approved by the institutional review board at the University of New Mexico (Human Research Review Committee number 03–277).

All women were instructed on a diet designed to provide 30 kcal/kg of normal body weight and 25 kcal/kg of body weight in obese women. The diet was also designed so that 40% of calories came from carbohydrates, 20% came from protein, and 30–40% came from fats. Calories were distributed 10% at breakfast, 20–30% for lunch and dinner, and 30% for snacks. The importance of exercise in the management of blood glucose was emphasized, and 30 minutes of walking per day was recommended. Patients received instruction from a nurse educator or a certified diabetes educator regarding diet and the use of the glucometer. All glucometers were memory based. Meter correlation was performed with serum blood glucose each trimester. Meters were expected to read within 10% of serum glucose readings. Patients tested blood glucose in the morning (fasting) and 2 hours after each meal. Compliance was assessed by polling the meter at each visit and by meeting with the diabetes educator at each visit in which medication use, diet, and amount of exercise were reported by the patient.

Women in the glyburide group received an initial dose of 2.5 mg twice daily, which was increased as necessary to a maximum dose of 20 mg/d (10 mg twice daily). The initial dose of metformin was 500 mg/d, taken in divided doses and increased as necessary to a maximum dose of 2 g/d. Glucose levels were reviewed weekly. Medication was increased if two or more glucose levels in the same meal exceeded target values by 10 mg/dL or greater for 2 consecutive weeks. Patients taking the maximum dose of either metformin or glyburide with two or more glucose values in the same meal exceeding target glucose values by 10 mg/dL or greater for 2 consecutive weeks were considered treatment failures and were started on insulin. The oral medication was discontinued when insulin therapy was initiated. The decision to increase medication or to initiate insulin was made by group decision based on the study protocol.

Prenatal care was provided by the University of New Mexico Pregnancy Diabetic group, which includes maternal–fetal medicine specialists, ob-gyn residents, diabetes nurse educators, and dieticians. All patients received monthly ultrasonography for growth, and twice weekly antenatal testing was initiated at 28 weeks. Elective delivery was planned at 38 weeks by induction of labor or repeat cesarean delivery.

The primary outcome of interest was glycemic control. Secondary outcomes were medication failure rate, macrosomia, admission to the neonatal intensive care unit (NICU), 5-minute Apgar score less than 7, birth trauma, preeclampsia, maternal and neonatal hypoglycemia, and route of delivery.

The study was powered to have an 80% probability of detecting a 10-mg/dL difference in blood glucose between the two groups with a standard deviation of 20 mg/dL and an α of .05. This is reasonable because Langer et al6 reported a standard deviation of less than 20 mg/dL in both arms of his study comparing insulin with glyburide. The Fisher exact test was used to compare categorical data, and the Student t test was used to compare mean numerical data. An intent-to-treat analysis was performed.

Back to Top | Article Outline


A total of 149 patients were randomized in the study, 74 in the glyburide arm and 75 in the metformin arm. All randomized patients were included in the analysis. In the glyburide arm, three patients never took the medicine and three relocated before delivery. In the metformin arm, five patients had only two prenatal visits, two patients relocated, and one patient could not tolerate the gastrointestinal effects of metformin and took only two doses of the medication.

There was no statistical difference between groups in age, ethnicity, weight, body mass index (BMI), or gestational age at randomization. Thirteen patients (17%) in the metformin group and eight patients (11%) in the glyburide group were at less than 24 weeks of gestation at the time of randomization (Table 1). Twenty-six patients (34.7%) in the metformin group and 12 patients (16.2%) in the glyburide group did not meet glycemic goals and required insulin therapy (P=.01). The failure rate of metformin was 2.1 times higher than the failure rate of glyburide (95% confidence interval 1.2–3.9, odds ratio 2.7). Of the patients who achieved glycemic control on either medication, there was no difference in mean fasting or 2-hour postprandial blood glucose values between the two treatment arms, nor was the time course different (Table 2; repeated-measures analysis of variance, arm P=.34, time P=.41, and interaction P=.42).

Table 1

Table 1

Table 2

Table 2

There was no statistical difference in the mean values of the 1-hour 50-g glucose challenge test between patients who failed to respond to metformin or glyburide, or between patients who had success with either drug. The value of the 1-hour glucose challenge was also not predictive of success or failure of metformin. The incidence of maternal hypoglycemia and preeclampsia was not different between the two groups (Table 3). Two patients in the metformin group and one patient in the glyburide group experienced hypoglycemia defined as blood glucose less than 60 mg/dL (P=1.0). Three patients (4%) in the glyburide group and two patients (2.7%) in the metformin group developed severe preeclampsia (P>.5). There was one shoulder dystocia in the glyburide group and one third-degree tear in the metformin group. Excluding elective repeat cesarean deliveries, there were 11 cesarean deliveries in the metformin group compared with two cesarean deliveries in the glyburide group (P=.02).

Table 3

Table 3

The mean birth weight of babies in the metformin group was smaller than the mean birth weight of babies in the glyburide group (P=.02). Other neonatal outcomes did not differ between the two groups. Macrosomia occurred in four patients (5.4%) in the glyburide group and one patient (1.3%) in the metformin group. There were no 5-minute Apgar scores less than 7 in either group. There were four NICU admissions in the metformin group (two for iatrogenic prematurity due to preeclampsia and two for transient tachypnea) and one NICU admission in the glyburide group (prematurity due to preeclampsia). One infant in the metformin group experienced hypoglycemia with blood glucose less than 40 mg/dL.

Back to Top | Article Outline


This randomized study compared glycemic control with glyburide and metformin as single agents for glucose control in gestational diabetes. We found that the failure rate to achieve adequate glycemic control was 2.1 times higher for the metformin group in comparison with the glyburide group. We also found a higher rate of nonelective cesarean deliveries and smaller birth weights in the metformin group. The maternal and neonatal outcomes were otherwise similar between the two groups.

The rate of failure of glyburide in this study was 16.2%. Although this is significantly higher than the 4% reported in the initial study by Langer et al, this rate is in accord with subsequent reports of failure rates up to 20%.9–11 The failure rate of metformin in this study was 34.7%. This is in line with the study reported by Rowan et al7 in which 46.3% of patients receiving metformin received supplemental insulin to achieve euglycemia. It is vastly different from preliminary results that we previously reported of 34 patients receiving metformin in which all patients had adequate glycemic control while using the medication.12 We speculate that one possible explanation may be an ethnic difference in response to metformin. The participants in the previous study were predominantly African American (55%) or Native American (35%), with a mean BMI of 39.7±9, compared with the participants in the current study, who are mostly Hispanic (86%), with a mean BMI of 32.8±5.8.

In the metformin group, 11 nonelective cesarean deliveries were performed: three for breech presentation and eight for nonreassuring fetal status. This is compared with two nonelective cesarean deliveries in the glyburide group: one for failure to progress and the other for nonreassuring fetal status (P = .02). This difference in the nonelective cesarean delivery rate is statistically significant, although the manner in which the drug may have affected this outcome is unclear. In addition, the number of patients in the cesarean delivery group is too small to look at other obstetric or neonatal factors which may have resulted in a bias toward the higher operative delivery rate in the metformin group.

Mean birth weights were smaller in the metformin group than in the glyburide group, although the rate of macrosomia was not different between the two groups. Smaller birth weight has been suggested as a benefit of metformin due both to its action as an insulin sensitizer and its known ability to traverse the placenta.

The major limitation of our study is the small size. It is underpowered to address secondary outcomes. The increase in cesarean deliveries and the smaller mean birth weight in the metformin group have not been found in other studies. Strengths are that the patients encompass the spectrum of gestational diabetes and that the study was conducted in a regular clinic environment.

Stringent control of gestational diabetes has traditionally been achieved through use of intensive insulin therapy. However, this type of therapy is difficult for pregnant women to adhere to, often requiring four injections a day. This results in decreased compliance and suboptimal control. In addition, hypoglycemia can be an issue with intensive insulin therapy. Oral therapeutic agents are better tolerated and are increasingly used in pregnancy; however, questions remain regarding their efficacy and safety. In this study, we compared glyburide with metformin as single agents for glycemic control in women who did not respond to diet and exercise. Although both medications were well-tolerated, the metformin group had more than twice the failure rate of glyburide, with 34.7% of patients eventually requiring insulin therapy, compared with a little more than 16% who required insulin therapy in the glyburide group.

This study addressed the use of glyburide and metformin as single-agent therapies. Future studies are needed to determine whether there is a benefit in continuing these agents as adjuvant therapies in patients requiring insulin. In addition, the role of oral agents in patients with pregestational diabetes either as single agents or in combination with insulin needs to be addressed in future studies.

Back to Top | Article Outline


1. Kjos SL, Buchanan TA. Gestational diabetes mellitus. N Engl J Med 1999;341:1749–56.
2. Xiong X, Saunders LD, Wang FL, Demianczuk NN. Gestational diabetes mellitus: prevalence, risk factors, maternal and infant outcomes. Int J Gynaecol Obstet 2001;75:221–8.
3. Yogev Y, Xenakis EM, Langer O. The association between preeclampsia and the severity of gestational diabetes: the impact of glycemic control. Am J Obstet Gynecol 2004;191:1655–60.
4. Langer O, Yogev Y, Most O, Xenakis EM. Gestational diabetes: the consequences of not treating. Am J Obstet Gynecol 2005;192:989–97.
5. Langer O. Management of gestational diabetes. Clin Obstet Gynecol 2000;43:106–15.
6. Langer O, Conway DL, Berkus MD, Xenakis EM, Gonzales O. A comparison of glyburide and insulin in women with gestational diabetes mellitus. N Engl J Med 2000;343:1134–8.
7. Rowan JA, Hague WM, Gao W, Battin MR, Moore MP; MiG Trial Investigators. Metformin versus insulin for the treatment of gestational diabetes [published erratum appears in N Engl J Med 2008;359:106]. N Engl J Med 2008;358:2003–15.
8. Coustan DR, Carpenter MW. The diagnosis of gestational diabetes. Diabetes Care 1998;21(suppl 2):B5–8.
9. Rochon M, Rand L, Roth L, Gaddipati S. Glyburide for the management of gestational diabetes: risk factors predictive of failure and associated pregnancy outcomes. Am J Obstet Gynecol 2006;195:1090–4.
10. Jacobson GF, Ramos GA, Ching JY, Kirby RS, Ferrara A, Field DR. Comparison of glyburide and insulin for the management of gestational diabetes in a large managed care organization. Am J Obstet Gynecol 2005;193:118–24.
11. Conway DL, Gonzales O, Skiver D. Use of glyburide for the treatment of gestational diabetes: the San Antonio experience. J Matern Fetal Neonatal Med 2004;15:51–5.
12. Moore LE, Briery CM, Clokey D, Martin RW, Williford NJ, Bofill JA, et al. Metformin and insulin in the management of gestational diabetes mellitus: preliminary results of a comparison. J Reprod Med 2007;52:1011–5.


© 2010 by The American College of Obstetricians and Gynecologists.