Narrative Review

Gynecol Obstet Invest 2021;86:218–230 Received: May 13, 2020

Accepted: March 14, 2021
DOI: 10.1159/000515893 Published online: May 12, 2021

Comparison of Insulin, Metformin, and Glyburide

on Perinatal Complications of Gestational Diabetes
Mellitus: A Systematic Review and Meta-Analysis
Xian Wang a Wanting Liu b Huizhen Chen a Qiu Chen c
aSchool
of Clinical Medicine, Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, China;
bDepartment
of Chinese Medicine, The Fourth People’s Hospital of Chengdu, Chengdu, China; cDepartment of
Endocrinology, Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, China

Keywords 0.50–0.90, p = 0.002), shorter gestation age at delivery (MD:

Gestational diabetes · Metformin · Glyburide · Insulin · −0.22, 95% CI: −0.34 to −0.10, p = 0.0002), lower maternal
Perinatal complications weight gain (MD: −1.41, 95% CI: −2.28 to −0.55, p = 0.001),
less incidence of caesarean section delivery (RR: 0.86, 95% CI:
0.78–0.95, p = 0.0004), lower maternal postprandial blood
Abstract glucose (SMD: −0.41, 95% CI: −0.72 to −0.11, p = 0.008), and
Aim: This systematic and meta-analysis was conducted to lower incidence of pregnancy-induced hypertension (RR:
evaluate the efficacy and safety of insulin, metformin, and 0.47, 95% CI: 0.27–0.83, p = 0.01). However, glyburide, com-
glyburide on perinatal complications for gestational diabe- pared with insulin, was associated with higher BW (MD:
tes mellitus (GDM). Methods: Medline (PubMed), EMBASE, 54.95, 95% CI: 3.87–106.03, p = 0.03) and increased the inci-
The Cochrane Library (Cochrane Database of Systematic Re- dence of neonatal hypoglycemia (RR: 1.52, 95% CI: 1.12–2.07,
views, Cochrane Central Register of Controlled Trials [CEN- p = 0.007). Meanwhile, compared to glyburide, metformin
TRAL], and Cochrane Methodology Register), Web of Science was associated with higher maternal fasting blood glucose
(Science and Social Science Citation Index), and ClinicalTrials (SMD: 0.20, 95% CI: 0.05–0.36, p = 0.01) and lower incidence
(Clinicaltrials.gov) were searched, as well as manual search- of induction of labor (RR: 0.76, 95% CI: 0.59–0.97, p = 0.03).
ing. We included randomized controlled trials comparing ef- Conclusions: This review suggests that metformin can de-
ficacy and safety of metformin versus glyburide, metformin crease the incidence of perinatal complications, and it should
versus insulin, and glyburide versus insulin in patients with be considered as a generally safe alternative to insulin.
GDM. Results: We included 32 articles including 5,964 pa- © 2021 S. Karger AG, Basel
tients published from inception to July 2020. Compared with
insulin, metformin was more effective at lower incidence of
macrosomia (RR: 0.66, 95% CI: 0.50–0.88, p = 0.005), lower Introduction
incidence of neonatal intensive care unit admission (RR: 0.78,
95% CI: 0.67–0.91, p = 0.002), less neonatal hypoglycemia Pregnancy is associated with insulin resistance (IR)
(RR: 0.67, 95% CI: 0.56–0.80, p < 0.0001), decreased birth and hyperinsulinemia that may predispose some women
weight (BW) (SMD: −0.37, 95% CI: −0.62 to −0.12, p = 0.004), to develop diabetes [1]. Gestational diabetes mellitus
lower incidence of large for gestational age (RR: 0.76, 95% CI: (GDM) is a pathological state of glucose intolerance rec-
130.209.6.61 - 8/13/2021 10:39:25 PM

karger@karger.com © 2021 S. Karger AG, Basel Correspondence to:

www.karger.com/goi Qiu Chen, chenqiu1005 @ cdutcm.edu.cn
Glasgow Univ.Lib.
Downloaded by:
ognized for the first time during pregnancy, which is not highest probability of being the most effective treatment
clearly overt diabetes. With the increasing incidence of in reducing the risk of most outcomes compared with in-
obesity, unhealthy lifestyles, and rise in maternal age of sulin or glyburide.
women, the occurrence of diabetes keeps an upward ten- Glyburide is a sulfonylurea that increases insulin se-
dency [2]. Meanwhile, in addition to these factors, envi- cretion by binding to pancreatic β-cell adenosine triphos-
ronmental and psychosocial factors and genetic polymor- phate calcium channel receptors. Balsells et al. [12]
phisms may contribute to the development of GDM, such showed that glyburide is inferior to either insulin or met-
as transcription factor 7 like 2, tumor necrosis factor-α formin and therefore should not be employed for treating
[3], and estrogen receptor α [4]. The prevalence of GDM women with GDM if insulin or metformin is available.
varies widely worldwide, ranging from 1 to >30%. In the Nevertheless, in the metformin in gestational diabetes
Middle East and some North African countries, the prev- (MiG) trial [8], half of the women randomized to metfor-
alence of GDM is the highest, with a median of 15.2%. On min required insulin to achieve glucose goals. In an RCT
the other hand, the prevalence of GDM in Europe is the of glyburide versus metformin in GDM [13], 34.7% of the
lowest with a median of 6.1% [5]. Women with GDM are women randomized to metformin required insulin and
at increased risk for severe perinatal complications for around 16% randomized to glyburide required insulin.
maternal and neonatal, such as macrosomia, hyperbiliru- This demonstrates that glyburide may be superior to met-
binemia, respiratory distress, and cesarean delivery [6]. formin in GDM.
Long-term complications also appear in GDM mothers Although several systematic reviews and meta-analy-
and their offspring, who are at higher risk of diabetes, ses have assessed insulin and oral hypoglycemic agents,
obesity, and premature cardiovascular disease [5]. there are still some inconsistent maternal or neonatal out-
Diet modification is normally used as the first-line comes. The study by Jiang et al. [14] demonstrated that
treatment. When lifestyle interventions fail to manage both metformin and glyburide are suitable for use in the
blood glucose at the goals of women, the most recom- management of GDM due to good maternal glycemic
mended therapy for GDM is insulin according to the control. However, glyburide is associated with increased
American Diabetes Association (ADA). However, insulin neonatal hypoglycemia, high maternal weight gain, high
treatment has some limitations due to women’s anxiety, neonatal birth weight (BW), and macrosomia. Neverthe-
injection skills, dose adjustment, and the risk of hypogly- less, Amin et al. [15] conducted a meta-analysis to com-
cemia [7]. The American College of Obstetricians and pare the efficacy and safety of glyburide with metformin
Gynecologists (ACOG) considers insulin and oral medi- in treating GDM. They suggested that metformin seems
cations equivalent in efficacy with either being an appro- to be a superior choice to glyburide because a significant
priate choice for first-line therapy [8, 9]. Two options of increase in the risk of macrosomia and large for gesta-
oral agents are available in the management of GDM and tional age (LGA) births was observed in the glyburide
include metformin and glyburide. Unlike insulin, metfor- group, whereas a nonsignificant increase in the risk of
min and glyburide can cross the placenta and thus can neonatal hypoglycemia. Furthermore, Balsells et al. [12]
potentially affect the developing fetus. As a result, a series concluded that metformin performs slightly better than
of recent randomized controlled trials (RCTs) compared insulin, but uneven for fetal outcomes according to more
metformin, glyburide, and insulin for the treatment of preterm birth and less severe neonatal hypoglycemia. As
GDM, especially about the safety and reverse effect on a result, we performed this meta-analysis and review with
women and children. Despite the proven efficacy of insu- updated data to assess comparative efficacy and safety be-
lin, metformin, and glyburide, the perinatal complica- tween oral hypoglycemic drugs and insulin, expecting to
tions are still controversial. provide sufficient and valid evidence for the management
Metformin, the first-line treatment for type 2 diabetes of GDM.
mellitus [2], belongs to the biguanide class. The primary
target tissue of metformin is the liver, and its major effect
is decreasing hepatic glucose output, mainly due to the Materials and Methods
suppression of gluconeogenesis, which leads to lower
fasting blood glucose levels without insulin stimulation Following the Cochrane Handbook for Systematic Reviews of
Interventions [16], we conducted this systematic review and meta-
and weight gain [10]. Although insulin is a first-line ther- analysis. This systematic review protocol was registered
apy in most countries, Farrar et al. [11] conducted a net- (CRD42018117434) on PROSPERO, and the results were present-
work meta-analysis suggesting that metformin had the ed according to the Preferred Reporting Items for Systematic Re-
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 219

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
views and Meta-Analyses (PRISMA) statement [17]. This study Statistical Synthesis
aims to analyze insulin, metformin, and glyburide on perinatal According to the following treatments, the experiment was di-
complications of GDM, as well as the efficiency of these treatments. vided into 3 categories: (1) metformin versus glyburide; (2) insulin
versus metformin; and (3) insulin versus glyburide. For dichoto-
Study Selection mous variables, the relative risk (RR) of each group compared with
Studies were included if they fulfilled the following criteria: (1) its 95% CI was calculated based on the number of result events in
subjects were women with GDM who were not well controlled each randomized group and the number of randomized events for
with lifestyle interventions and needed pharmacological modifica- each group. For continuous results, the mean values between
tion to lower blood glucose levels; (2) study design was an RCT; (3) groups were calculated according to the average values of the re-
compared the efficacy and safety parameters between 2 drugs: met- sults and standard deviation.
formin and glyburide, metformin and insulin, or glyburide and By using Review Manager 5.3 software, the initial study results
insulin; (4) report one or more maternal or neonatal outcomes of were aggregated into the relative risk of dichotomous outcomes or
interest; and (5) were published as a full text. Exclusion criteria the average difference of numerical results. Heterogeneity was ex-
were as follows: (1) studies published with incomplete original plored for all the meta-analyses. A fixed-effects model was used to
data; (2) retrospective studies, observational studies, case series, analyze the data. If heterogeneity exists (I2 ≥50%), the random-
cross-sectional studies, reviews, letters, and animal experiments; effects model was used. Pooled analyses were also used to compare
(3) studies including patients with pre-existing diabetes; (4) and in baseline characteristics. We attempted to obtain the missing data
the case of repeated reports in one study cohort, only the recent from the original researchers. If not successful, we only analyzed
article was selected, excluding other literature. Furthermore, the the available data.
individual study definition of GDM and the diagnostic criteria
used to identify such patients were accepted. No restriction was
imposed in terms of language.
Results
Search Strategy
We searched studies from inception to July 2020 through the After searching, a total of 1,644 studies were identified
following electronic databases: Medline (PubMed), EMBASE, The from databases. Endnote X9.3.3 software was used for
Cochrane Library (Cochrane Database of Systematic Reviews,
Cochrane Central Register of Controlled Trials [CENTRAL],
­ managing the data. Then, 1,180 of those were removed
and Cochrane Methodology Register), Web of Science (Science for duplicates. Based on the title and abstract, we exclud-
and Social Science Citation Index), and ClinicalTrials (Clinicaltri- ed 373 articles. Afterward, 91 of full-text articles were as-
als.gov). In addition, a manual search of journals and conference sessed for eligibility. However, 59 of them were excluded
proceedings was also conducted. The key search terms used were for reasons. Finally, 32 studies were screened and en-
gestational diabetes, glyburide, metformin, hypoglycemic agents,
and randomized controlled trial. The search strategy is shown in rolled in this meta-analysis, shown in online suppl. Fig-
online suppl. Table 1 (for all online suppl. material, see www.­ ure 3.
karger.com/doi/10.1159/000515893).
Characteristics of the Included Studies
Data Collection and Risk of Bias Assessment We included 32 articles published from inception to
The screening of titles and abstracts was performed and se-
lected independently by 2 investigators (W.X. and L.W.T.). The July 2020. The characteristics of these studies comprise
articles that did not fit the eligibility criteria were excluded. If the the location, number of participants, screening strategy,
title or abstract appeared to meet the eligibility criteria, the full diagnostic criterion, intervention, and outcomes, shown
texts of the articles were obtained for further evaluation. The dis- in online suppl. Table 2. A total of 5,964 female patients
agreements were resolved by consensus or by the third reviewer participated in these studies and were treated with insu-
(C.H.Z.).
The risk of bias among the included studies was evaluated ac- lin, glyburide, or metformin. Among them, 4 articles
cording to the risk of bias assessment tool by the Cochrane Col- compared glyburide with metformin, 17 articles com-
laboration [10]. The specific evaluation criteria were taking into pared metformin with insulin, and 11 articles compared
account 7 items, including sequence generation, allocation con- glyburide with insulin. Outcomes of perinatal complica-
cealment, blinding of participants and medical staff to treatment tions include macrosomia, NICU admission, neonatal
allocation, blinding of assessors, loss to follow-up, selective report-
ing of outcomes, and other sources of bias. Each study was assigned hypoglycemia, BW, LGA, small for gestational age
a high risk, low risk, or unclear risk. All studies were assessed by 2 (SGA), neonatal hyperbilirubinemia, phototherapy, fe-
reviewers independently (W.X. and L.W.T.), and the divergences tal or neonatal death, Apgar 5′ <7, congenital anomaly,
were resolved by discussion with the third author (C.H.Z.) until respiratory distress, shoulder dystocia, preterm birth,
consensus was reached. The double-blinded assessment remained GA at delivery, maternal weight gain, cesarean delivery,
at high risk of performance, for example, the administration mode
of insulin and oral hypoglycemic drugs. The risk of bias graph and maternal blood glucose, glycosylated hemoglobin, preg-
a summary were performed according to the Cochrane Risk of Bias nancy-induced hypertension, preeclampsia, and induc-
Tool [16], shown in online suppl. Figures 1 and 2, respectively. tion of labor.
130.209.6.61 - 8/13/2021 10:39:25 PM

220 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
Table 1. Meta-analysis of RCTs comparing outcomes from treatment with metformin vs. glyburide, metformin vs. insulin, and glyburide
vs. insulin in women with GDM

Outcomes Treatment Studies, Sample size, Overall effect

N N
statistical effect p value I2 effect
method estimate model

Neonatal outcomes
BW (g) Metformin vs. glyburide 3 453 MD −133.23 [−289.26, 22.80] 0.09 64% RE
Metformin vs. insulin 14 2,791 SMD −0.37 [−0.62, −0.12] 0.004* 90% RE
Glyburide vs. insulin 8 2,549 MD 54.95 [3.87, 106.03] 0.03* 29% FE
LGA Metformin vs. glyburide 2 304 RR 1.16 [0.62, 2.16] 0.64 0% FE
Metformin vs. insulin 10 1,340 RR 0.76 [0.63, 0.90] 0.002* 31% FE
Glyburide vs. insulin 3 515 RR 1.66 [0.57, 4.86] 0.35 50% RE
Metformin vs. glyburide 0 – – – – – –
SGA Metformin vs. insulin 12 2,775 RR 1.15 [0.87, 1.53] 0.32 40% FE
Glyburide vs. insulin 0 – – – – – –
Hyperbilirubinemia Metformin vs. glyburide 1 104 RR 0.77 [0.43, 1.36] – – –
Metformin vs. insulin 11 2,530 RR 0.81 [0.61, 1.06] 0.12 18% FE
Glyburide vs. insulin 5 1,549 RR 1.18 [0.76, 1.84] 0.46 0% FE
Phototherapy Metformin vs. glyburide 2 263 RR 0.84 [0.51, 1.38] 0.49 0% FE
Metformin vs. insulin 6 1,389 RR 1.07 [0.77, 1.50] 0.69 0% FE
Glyburide vs. insulin 2 119 RR 1.17 [0.61, 2.24] 0.63 9% FE
Fetal or Metformin vs. glyburide 2 359 RR 0.92 [0.06, 14.55] 0.95 – FE
neonatal death Metformin vs. insulin 4 1,311 RR 0.34 [0.05, 2.15] 0.25 0% FE
Glyburide vs. insulin 3 491 RR 1.41 [0.28, 7.07] 0.68 0% FE
Apgar 5′ <7 Metformin vs. glyburide 1 104 RR 0.35 [0.01, 8.31] – – –
Metformin vs. insulin 7 1,041 RR 1.26 [0.73, 2.18] 0.41 0% FE
Glyburide vs. insulin 2 832 RR 0.33 [0.09, 1.17] 0.09 – FE
Congenital Metformin vs. glyburide 0 – – – – – –
anomaly Metformin vs. insulin 6 2,554 RR 0.68 [0.38, 1.22] 0.20 0% FE
Glyburide vs. insulin 6 933 RR 0.87 [0.39, 1.95] 0.74 0% FE
Respiratory distress Metformin vs. glyburide 1 159 RR 0.51 [0.10, 2.69] – – –
Metformin vs. insulin 10 1,197 RR 0.73 [0.50, 1.05] 0.09 0% FE
Glyburide vs. insulin 2 832 RR 0.88 [0.36, 2.15] 0.77 – FE
Shoulder dystocia Metformin vs. glyburide 2 253 RR 0.34 [0.04, 3.20] 0.34 0% FE
Metformin vs. insulin 3 454 RR 0.83 [0.26, 2.69] 0.76 1% FE
Glyburide vs. insulin 2 832 RR 0.60 [0.05, 6.61] 0.68 – FE
Preterm birth Metformin vs. glyburide 2 263 RR 1.86 [0.65, 5.33] 0.25 0% FE
Metformin vs. insulin 11 2,611 RR 1.22 [0.93, 1.60] 0.15 23% FE
Glyburide vs. insulin 3 896 RR 1.34 [0.79, 2.26] 0.28 43% FE
Gestational age Metformin vs. glyburide 4 612 MD −0.10 [−0.32, 0.13] 0.40 40% FE
at delivery Metformin vs. insulin 10 2,963 MD −0.22 [−0.34, −0.10] 0.0002* 6% FE
Glyburide vs. insulin 4 290 MD −0.07 [−0.38, 0.23] 0.64 22% FE
Maternal outcomes
Maternal Metformin vs. glyburide 2 304 MD −1.44 [−2.99, 0.10] 0.07 11% FE
weight gain (kg) Metformin vs. insulin 5 738 MD −1.41 [−2.28, −0.55] 0.001* 89% RE
Glyburide vs. insulin 2 455 MD −0.91 [−2.87, 1.06] 0.37 0% FE
Caesarean delivery Metformin vs. glyburide 3 412 RR 1.34 [0.77, 2.35] 0.31 56% RE
Metformin vs. insulin 11 1,693 RR 0.86 [0.78, 0.95] 0.004* 46% FE
Glyburide vs. insulin 4 1,025 RR 0.93 [0.79, 1.10] 0.40 19% FE
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 221

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
Table 1 (continued)

Outcomes Treatment Studies, Sample size, Overall effect

N N
statistical effect p value I2 effect
method estimate model

Fasting blood glucose Metformin vs. glyburide 4 612 SMD 0.20 [0.05, 0.36] 0.01* 0% FE
Metformin vs. insulin 10 1,148 SMD −0.23 [−0.52, 0.07] 0.13 90% RE
Glyburide vs. insulin 5 666 SMD 0.14 [−0.01, 0.30] 0.06 0% FE
Postprandial blood Metformin vs. glyburide 4 612 SMD 0.10 [−0.06, 0.26] 0.21 43% FE
glucose Metformin vs. insulin 10 2,148 SMD −0.41 [−0.72, −0.11] 0.008* 91% RE
Glyburide vs. insulin 6 689 SMD −0.07 [−0.82, 0.69] 0.86 94% RE
Glycosylated Metformin vs. glyburide 0 – – – – – –
hemoglobin Metformin vs. insulin 7 2,053 MD −0.02 [−0.13, 0.10] 0.79 79% RE
Glyburide vs. insulin 4 584 MD −0.45 [−1.01, 0.12] 0.12 93% RE
Pregnancy-induced Metformin vs. glyburide 2 263 RR 0.66 [0.29, 1.46] 0.30 0% FE
hypertension Metformin vs. insulin 5 822 RR 0.47 [0.27, 0.83] 0.01* 0% FE
Glyburide vs. insulin 0 – – – – – –
Preeclampsia Metformin vs. glyburide 1 149 RR 0.66 [0.11, 3.82] – – –
Metformin vs. insulin 9 2,000 RR 0.79 [0.60, 1.04] 0.10 0% FE
Glyburide vs. insulin 1 64 RR 1.50 [0.47, 4.82] – – –
Induction of labor Metformin vs. glyburide 2 263 RR 0.76 [0.59, 0.97] 0.03* 0% FE
Metformin vs. insulin 4 693 RR 0.89 [0.74, 1.08] 0.24 41% FE
Glyburide vs. insulin 1 23 – – – – –
RCTs, randomized controlled trials; GDM, gestational diabetes mellitus; MD, mean difference; RE, random effect; SMD, standard
mean difference; FE, fixed effect; RR, risk ratio; BW, birth weight; LGA, large for gestational age; SGA, small for gestational age.

Bias of Risk for the Included Studies 0.01). Nevertheless, several outcomes only exist in one of
The risk of bias was measured for each of the 32 stud- those studies, for example, neonatal hyperbilirubinemia,
ies included. Although all the trials mentioned random- preeclampsia, Apgar 5′ <7, congenital anomaly, respiratory
ization, only 17 studies described their specific random- distress, and maternal glycosylated hemoglobin.
ization strategies [13, 18–33]. About half of the studies
did not specify the strategy of allocation concealment [22, Effect of Metformin versus Insulin
23, 31, 32, 34–48]. Twenty-one studies were judged to be Seventeen studies including 3,384 participants com-
high risk of performance bias, of which 6 studies reported pared metformin with insulin [20–22, 24–26, 31, 32, 34,
open-label [18, 20, 23, 35, 37, 42], and 16 studies were un- 38, 40–42, 44, 45, 47, 48]. For neonatal outcomes, there
likely to have been blinded due to differences in mode of was evidence of decreased incidence rates of macrosomia
administration of the interventions [13, 21, 22, 24, 30–32, (RR: 0.66, 95% CI: 0.50–0.88, p = 0.005), neonatal hypo-
36, 39–41, 43, 45, 47, 48]. The graph and summary of the glycemia (RR: 0.67, 95% CI: 0.56–0.80, p < 0.0001), NICU
risks of bias are provided in online suppl. Figures 1 and 2. admission (RR: 0.78, 95% CI: 0.67–0.91, p = 0.002), LGA
(RR: 0.76, 95% CI: 0.50–0.90, p = 0.002), and reduction
Effect of Metformin versus Glyburide of BW (SMD: −0.37, 95% CI: −0.62 to −0.12, p = 0.004)
Four studies, including 612 participants, compared met- between women who received metformin versus insulin,
formin with glyburide [13, 28, 37, 46]. There is no signifi- shown in Table 1 and Figures 1–3, respectively. However,
cant difference between metformin and glyburide for neo- there was no difference in SGA between the metformin
natal outcomes, as shown in Table 1 and Figures 1–3. For and insulin group. For maternal outcomes, it is shown
maternal outcomes, metformin can lower the incidence of that metformin is more effective at reducing the risk of
induction of labor (RR: 0.76, 95% CI: 0.59–0.97, p = 0.03); maternal weight gain (MD: −1.41, 95% CI: −2.28 to
however, glyburide has the efficacy of controlling maternal −0.55, p = 0.001), cesarean delivery (RR: 0.86, 95% CI:
fasting blood glucose (SMD: 0.20, 95% CI: 0.05–0.36, p = 0.78–0.95, p = 0.0004), maternal postprandial blood glu-
130.209.6.61 - 8/13/2021 10:39:25 PM

222 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
 Color version available online
Macrosomia
Metformin Glyburide
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Moore, 2010 1 75 4 74 80.4 0.25 [0.03, 2.16]
Nachum, 2017 2 51 1 53 19.6 2.08 [0.19, 22.22]

Total (95% CI) 126 127 100.0 0.61 [0.15, 2.47]

Total events 3 5
Heterogeneity: χ2 = 1.70, df = 1 (p = 0.19); I2 = 41% 0.01 0.1 1 10 100
Test for overall effect: Z = 0.70 (p = 0.48) Metformin Glyburide

Metformin Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Ashoush, 2016 2 47 5 48 4.8 0.41 [0.08, 2.00]
Eid, 2018 3 113 6 116 5.8 0.51 [0.13, 2.00]
Hague, 2003 2 16 2 14 2.1 0.88 [0.14, 5.42]
Hassan, 2012 8 75 14 75 13.7 0.57 [0.25, 1.28]
Ijas, 2010 9 47 11 50 10.4 0.87 [0.40, 1.91]
Mesdaghinia, 2013 11 100 18 100 17.6 0.61 [0.30, 1.23]
Niromanesh, 2012 3 80 8 80 7.8 0.38 [0.10, 1.36]
Ruholamin, 2014 0 50 0 50 Not estimable
Spaulonci, 2013 0 47 3 47 3.4 0.14 [0.01, 2.69]
Tertti, 2012 22 110 16 107 15.9 1.34 [0.74, 2.40]
Wasim, 2019 7 137 19 141 18.3 0.38 [0.16, 0.87]

Total (95% CI) 822 828 100.0 0.66 [0.50, 0.88]

Total events 67 102
Heterogeneity: χ2 = 10.26, df = 9 (p = 0.33); I2 = 12% 0.01 0.1 1 10 100
Test for overall effect: Z = 2.82 (p = 0.005) Metformin Insulin

Glyburide Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, random, 95% CI M-H, random, 95% CI
Behrashi, 2016 4 120 17 129 19.7 0.25 [0.09, 0.73]
Lain, 2009 9 41 1 41 9.8 9.00 [1.19, 67.85]
Langer, 2000 14 201 9 203 23.3 1.57 [0.70, 3.55]
Masoomeh, 2015 2 37 4 59 12.8 0.80 [0.15, 4.14]
Senat, 2018 33 367 28 442 28.2 1.42 [0.87, 2.30]
Tempe, 2013 1 32 1 32 6.3 1.00 [0.07, 15.30]

Total (95% CI) 798 906 100.0 1.13 [0.53, 2.42]

Total events 63 60
Heterogeneity: τ = 0.48, χ = 13.24, df = 5 (p = 0.02); I = 62%
2 2 2 0.01 0.1 1 10 100
Test for overall effect: Z = 0.31 (p = 0.76) Glyburide Insulin

Fig. 1. Main outcomes of macrosomia between metformin, glyburide, and insulin in neonates.

cose (SMD: −0.41, 95% CI: −0.72 to −0.11, p = 0.008), and 35, 36, 39, 43]. For neonatal outcomes, there was a re-
pregnancy-induced hypertension (RR: 0.47, 95% CI: duced rate of neonatal hypoglycemia (RR: 1.52, 95% CI:
0.27–0.83, p = 0.01), as well as a shorter GA at delivery 1.12–2,07, p = 0.007) and a decrease of BW (MD: 54.95,
(MD: −0.22, 95% CI: −0.34 to −0.10, p = 0.0002). Further- 95% CI: 3.87–106.03, p = 0.03) shown between women
more, there is no difference between the other outcomes, who received insulin versus glyburide (Table 1; Fig. 3,
shown in Table 1. respectively). However, there is no obvious difference
between neonatal hyperbilirubinemia, macrosomia,
Effect of Glyburide versus Insulin and other outcomes, shown in Table 1 and Figures 1
Eleven studies including 1,958 participants com- and 2.
pared glyburide with insulin [18, 19, 23, 27, 29, 30, 33,
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 223

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
 Color version available online
NICU Admission
Metformin Glyburide
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, random, 95% CI M-H, random, 95% CI
Moore, 2010 4 75 1 74 52.4 3.95 [0.45, 34.48]
Silva, 2012 0 104 7 96 47.6 0.06 [0.00, 1.06]

Total (95% CI) 179 170 100.0 0.54 [0.01, 36.77]

Total events 4 8
Heterogeneity: τ2 = 7.59, χ2 = 5.55, df = 1 (p = 0.02); I2 = 82% 0.01 0.1 1 10 100
Test for overall effect: Z = 0.28 (p = 0.78) Metformin Glyburide

Metformin Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Ainuddin, 2015 7 16 69 100 6.6 0.63 [0.36, 1.12]
Ashoush, 2016 4 47 5 48 1.7 0.82 [0.23, 2.86]
Eid, 2018 12 113 19 116 6.5 0.65 [0.33, 1.27]
Ghomian, 2018 29 143 27 143 9.3 1.07 [0.67, 1.72]
Hassan, 2012 14 75 19 75 6.6 0.74 [0.40, 1.36]
Ijas, 2010 7 47 11 50 3.7 0.68 [0.29, 1.60]
Mesdaghinia, 2013 14 100 33 100 11.4 0.42 [0.24, 0.74]
Niromanesh, 2012 5 80 2 80 0.7 2.50 [0.50, 12.51]
Rowan, 2008 68 363 78 370 26.7 0.89 [0.66, 1.19]
Ruholamin, 2014 0 50 2 50 0.9 0.20 [0.01, 4.06]
Saleh, 2016 10 67 12 70 4.1 0.87 [0.40, 1.88]
Silva, 2016 10 250 3 106 1.5 1.41 [0.40, 5.03]
Tertti, 2012 34 110 39 107 13.7 0.85 [0.58, 1.23]
Wasim, 2019 9 137 20 141 6.8 0.46 [0.22, 0.98]

Total (95% CI) 1,598 1,556 100.0 0.78 [0.67, 0.91]

Total events 223 339
Heterogeneity: χ2 = 13.78, df = 13 (p = 0.39); I2 = 6% 0.01 0.1 1 10 100
Test for overall effect: Z = 3.15 (p = 0.002) Metformin Insulin

Glyburide Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Behrashi, 2016 4 120 8 129 20.2 0.54 [0.17, 1.74]
Langer, 2000 12 201 14 203 36.5 0.87 [0.41, 1.83]
Masoomeh, 2015 4 37 7 59 14.1 0.91 [0.29, 2.90]
Reynolds, 2017 4 13 1 10 3.0 3.08 [0.40, 23.44]
Senat, 2018 10 367 11 442 26.2 1.09 [0.47, 2.55]

Total (95% CI) 738 843 100.0 0.93 [0.60, 1.45]

Total events 34 41
Heterogeneity: χ2 = 2.35, df = 4 (p = 0.67); I2 = 0% 0.01 0.1 1 10 100
Test for overall effect: Z = 0.31 (p = 0.75) Glyburide Insulin

Fig. 2. Main outcomes of NICU admission between metformin, glyburide, and insulin in neonates.

Discussion pregnant women’s fasting and postprandial glucose are

generally lower than in nonpregnant women due to in-
In 1957, Carrington first put forward the term “gesta- sulin-independent glucose uptake by the fetus and pla-
tional diabetes” [49]. The precise numeric cutoff values centa [52]. Meanwhile, in the late stage of gestation, pe-
of maternal glucose levels used for diagnosis are still con- ripheral insulin sensitivity decreases by about 50%, and
troversial. However, the general consensus of diagnosis the ability of insulin to increase glucose uptake in skeletal
is the IADPSG (International Association of Pregnancy muscle and adipose tissue declines [53]. Therefore, ex-
Study Group) [50] and WHO criteria [51], which is fast- cessive peripheral IR and insufficient insulin production
ing plasma glucose ≥7.0 mmol/L (126 mg/dL) and/or 2-h contribute together to the hyperglycemia of pregnant
glucose ≥11.1 mmol/L (200 mg/dL). In early pregnancy, women. Although increased IR and β-cell defects are the
130.209.6.61 - 8/13/2021 10:39:25 PM

224 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
 Color version available online
Neonatal hypoglycemia
Metformin Glyburide
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, random, 95% CI M-H, random, 95% CI
George, 2015 0 79 10 80 19.0 0.05 [0.00, 0.81]
Moore, 2010 1 75 0 74 16.4 2.96 [0.12, 71.52]
Nachum, 2017 5 51 1 53 25.1 5.20 (0.63, 42.96]
Silva, 2012 11 104 13 96 39.5 0.78 [0.37, 1.66]

Total (95% CI) 309 303 100.0 0.92 [0.18, 4.76]

Total events 17 24
Heterogeneity: τ = 1.63, χ = 7.82, df = 3 (p = 0.05); I = 62%
2 2 2 0.01 0.1 1 10 100
Test for overall effect: Z = 0.10 (p = 0.92) Metformin Glyburide

Metformin Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Ainuddin, 2015 4 16 30 100 3.4 0.83 [0.34, 2.05]
Ashoush, 2016 6 47 7 48 2.9 0.88 [0.32, 2.41]
Eid, 2018 9 113 14 116 5.7 0.66 [0.30, 1.46]
Ghomian, 2018 12 143 17 143 7.1 0.71 [0.35, 1.42]
Hassan, 2012 10 75 20 75 8.3 0.50 [0.25, 1.00]
Ijas, 2010 4 47 7 50 2.8 0.61 [0.19, 1.94]
Mesdaghinia, 2013 10 100 15 100 6.2 0.67 [0.31, 1.41]
Niromanesh, 2012 3 80 2 80 0.8 1.50 [0.26, 8.74]
Rowan, 2008 55 363 69 370 28.4 0.81 [0.59, 1.12]
Ruholamin, 2014 0 50 2 50 1.0 0.20 [0.01, 4.06]
Saleh, 2016 7 67 15 70 6.1 0.49 [0.21, 1.12]
Spaulonci, 2013 6 47 10 47 4.1 0.60 [0.24, 1.52]
Tertti, 2012 18 110 18 107 7.6 0.97 [0.54, 1.77]
Wasim, 2019 13 137 38 141 15.5 0.35 [0.20, 0.63]

Total (95% CI) 1,395 1,497 100.0 0.67 [0.56, 0.80]

Total events 157 264
Heterogeneity: χ2 = 10.80, df = 13 (p = 0.63); I2 = 0% 0.01 0.1 1 10 100
Test for overall effect: Z = 4.27 (p < 0.0001) Metformin Insulin

Glyburide Insulin
Weight, Risk ratio Risk ratio
Study or subgroup events total events total % M-H, fixed, 95% CI M-H, fixed, 95% CI
Behrashi, 2016 2 120 7 129 11.2 0.31 [0.07, 1.45]
Langer, 2000 18 201 12 203 19.8 1.51 [0.75, 3.06]
Mukhopadhyay, 2012 4 30 3 30 5.0 1.33 [0.33, 5.45]
Oguntemi, 2007 12 48 6 49 9.9 2.04 [0.83, 5.00]
Reynolds, 2017 2 13 0 10 0.9 3.93 [0.21, 73.71]
Senat, 2018 45 367 32 442 48.2 1.69 [1.10, 2.61]
Tempe, 2013 4 32 3 32 5.0 1.33 [0.32, 5.49]

Total (95% CI) 811 895 100.0 1.52 [1.12, 2.07] 0.01 0.1 1 10 100
Total events 87 63 Glyburide Insulin
Heterogeneity: χ2 = 5.21, df = 6 (p = 0.52); I2 = 0%
Test for overall effect: Z = 2.68 (p = 0.007)

Fig. 3. Main outcomes of hypoglycemia between metformin, glyburide, and insulin in neonates.

main metabolic abnormalities of GDM, these defects are The primary management of GDM is lifestyle inter-
almost asymptomatic and are usually detected due to ex- vention, including dietary modification and physical ac-
tensive testing of glucose levels during pregnancy. There- tivity. The RADIEL (Finnish Gestational Diabetes Pre-
fore, the ADA (American Diabetes Association) recom- vention) study revealed that diet intervention and moder-
mended that all pregnant women without diabetes ate physical activity could reduce the incidence of GDM
should take a 75-g OGTT at 24–28 weeks of gestation among the high-risk population [55]. Diet and physical
[54]. activity are sufficient to control the blood glucose status
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 225

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
of approximately 70–85% of women with GDM [56]. >4,000 g [13, 21, 22, 25, 28, 30, 33, 34, 39, 43], 7 studies
However, when glycemia remains increased after ≥1–2 defined as greater than or equal to 4,000 g [18, 24, 27, 38,
weeks of lifestyle interventions, pharmacological treat- 42, 45, 48], and 2 studies defined as >4,500 g [26, 40], lead-
ment should be initiated. In this review, we summarized ing to heterogeneity. However, metformin showed a sig-
the safety and efficacy of 2 oral antidiabetic drugs (OADs) nificant decrease in macrosomia incidence compared to
(metformin and glyburide) and insulin. When lifestyle insulin treatment. This result is consistent with the previ-
interventions, including dietary intervention and exer- ous studies [11, 59, 60]. Compared with insulin, glyburide
cise and self-monitoring of blood sugar, fail to meet the was associated with higher neonatal BW, while metfor-
required blood sugar levels within 1–2 weeks, GDM pa- min was with lower BW of infants, in which the results
tients begin to take medication [8, 18]. Most people agree were the same as a previous study [14]. Zhou et al. [61]
on insulin use if drug treatment is to be started. Although suggested that the level of triglyceride in maternal blood
insulin is recommended as the first-line therapy, the use was positively correlated with neonatal BW. Metformin
of OADs has been increasing among female patients with is also confirmed to reduce the level of triglyceride in
GDM [8, 57]. Metformin is a reasonable alternative for nonobese type 2 diabetes mellitus patients [62]. As a re-
women who refuse insulin therapy or who, in the opinion sult, the potential mechanism for metformin to reduce
of their obstetrician or obstetric caregiver, are unable to neonatal BW is improved maternal lipid metabolism.
use insulin safely, or who are unable to pay for it. Accord- Newborn weight, especially for macrosomia, is associ-
ing to the ACOG, glyburide is considered to be a third ated with the risk for delivery complications such as pro-
option for GDM because it failed to develop equivalent longed labor, operative vaginal delivery, cesarean section,
outcomes to insulin or metformin. Although it is sug- and NICU admission [63]. Similarly, our study confirmed
gested that there are no short-term adverse effects of that metformin decreases the incidence of cesarean sec-
OADs on maternal or neonatal health during pregnancy, tion and the risk of LGA compared with insulin, while it
there is a lack of long-term outcomes for adults and chil- failed to show difference in SGA. Furthermore, our study
dren who were exposed to these medications in utero confirmed that glyburide was connected with higher in-
[58]. The aim of our meta-analysis is to distinguish the duction of labor than metformin. Gui et al. [64] per-
efficacy and perinatal complications of metformin, gly- formed an analysis from 6 RCTs and reported that aver-
buride, and insulin. age gestational ages at delivery were significantly lower in
metformin groups than insulin groups. From 10 RCTs,
Main Findings we confirmed this result.
A total of 32 studies comprising 5,964 patients were In addition, the nonoptimal fetal growth environment
included for the present meta-analysis on metformin ver- has been shown to have long-term effects on cardiovas-
sus glyburide versus insulin therapy in GDM patients. cular and metabolic disorders, partly because of its im-
Our study confirmed that being compared with insulin, pact on birth age [65]. A previous review showed that
metformin has advantages on neonatal outcomes, espe- macrosomia is associated with the incidence of shoulder
cially on reducing the incidence of neonatal hypoglyce- dystocia [66]. However, there was no significant differ-
mia, macrosomia, and NICU admission. Meanwhile, ence in shoulder dystocia between the 3 treatments mutu-
metformin is more effective than insulin in reducing ges- ally. More and more RCTs recently demonstrated a sig-
tational age at delivery, maternal postprandial blood glu- nificantly reduced NICU admission rate in the metfor-
cose, and maternal complications including weight gain min group than in the insulin group [11, 59, 60, 67–69].
and pregnancy-induced hypertension. However, gly- It is considered that the rate of NICU admission is associ-
buride is associated with lower maternal fasting blood ated with neonatal hypoglycemia and hyperbilirubine-
glucose. mia. In our study, the treatment of metformin has a low-
er NICU admission rate than insulin. However, there was
Comparison with Other Studies no statistically significant difference with respect to the
According to Poolsup et al. [57], macrosomia is likely incidence of neonatal hyperbilirubinemia and photother-
to be linked with delayed motor development, premeno- apy between the 3 treatment groups.
pausal breast cancer, obesity, and diabetes later in life in Neonatal hypoglycemia frequently occurs in new-
the infants, which is defined as BW >4,000 g in this study. borns of women with GDM and is one of the most com-
In our meta-analysis, the definition of macrosomia varied mon diagnoses which requires NICU admission. How-
among the included studies, with 10 studies defined as ever, due to the various definition of neonatal hypoglyce-
130.209.6.61 - 8/13/2021 10:39:25 PM

226 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
mia, the incidence of it varies significantly among ficient data to determine whether glyburide affects the
different studies. Balsells et al. [12] reported that the inci- incidence of pregnancy-induced hypertension. A de-
dence of neonatal hypoglycemia in the metformin group crease in the rate of gestational hypertension is likely to
was lower than in the insulin group. Song et al. [70] and reduce the incidence of preeclampsia [57]. However, in
Jiang et al. [14] revealed that the incidence of neonatal our study, there is no significant statistical difference be-
hypoglycemia in the glyburide group was higher than in tween metformin and insulin in terms of preeclampsia.
the insulin group. Our analysis confirmed these results. Due to insufficient studies, we cannot report the differ-
However, there was a nonsignificant difference between ence in preeclampsia of metformin versus glyburide and
metformin and glyburide treatment. This result was like- glyburide versus insulin.
ly due to insufficient statistical power, as only 4 RCTs In the previous meta-analysis, Su et al. [73] reported
were included in this analysis. As glucose can pass through that metformin was associated with slightly lower ma-
the placenta, maternal hyperglycemia leads to increasing ternal weight gain compared with insulin, which is the
glucose in the fetus, causing excess fetal insulin produc- same as our study. In our study, insulin was associated
tion. Nonetheless, due to the cease of glucose supply from with higher maternal weight gain compared with met-
mothers after birth, the insulin level of newborns contin- formin. Current evidence suggests that weight changes
ues to rise, causing neonatal hypoglycemia and inhibition associated with metformin are more likely to be due to
of the metabolic compensation mechanism [71]. In addi- reduced caloric intake rather than increased energy ex-
tion, stress response at birth and hypersensitivity of pan- penditure. Due to its gastrointestinal side effects, met-
creatic β cells contribute to a higher neonatal insulin lev- formin seems to directly or indirectly affect appetite
el. It is reported that persistent or prolonged hypoglyce- regulation [74].
mia may result in permanent brain damage and adverse However, there were no differences in the prevalence
neurodevelopment outcome [72]. As a result, antihypo- of macrosomia, NICU admission, maternal weight gain,
glycemia treatment should be considered to prevent these cesarean section, and gestational ages in the group of met-
complications. formin and glyburide, as well as glyburide and insulin. In
The management of GDM is primarily aimed at glyce- this meta-analysis, some outcomes failed to show the dif-
mic targets to reduce the incidence of adverse pregnancy ference between the 3 treatments, including fetal or neo-
outcomes. Our study revealed that glyburide was better at natal death, Apgar 5′ <7, congenital anomaly, respiratory
controlling maternal fasting blood glucose compared distress, shoulder dystocia, maternal glycosylated hemo-
with metformin, which is consistent with a few previous globin, and preterm birth.
studies [12, 15]. However, there was no significant differ- In our study, metformin shows significant advantages
ence in postprandial blood glucose and glycosylated he- in many respects compared to insulin and glyburide, in-
moglobin. In addition, we confirmed there was no sig- cluding neonatal hypoglycemia, macrosomia, NICU ad-
nificant difference between glyburide and insulin. These mission, BW, maternal weight gain, cesarean section de-
findings demonstrate that both glyburide and insulin are livery, GA, LGA, maternal postprandial blood glucose,
equally effective in managing blood glucose in GDM. The and pregnancy-induced hypertension. Therefore, it
probable reason is that glyburide has a longer half-life, should be considered as a generally safe alternative to in-
whose continuous glucose control effect is better than sulin. Our study confirmed that glyburide has the efficacy
metformin. Still, there is no difference in long-term blood of controlling maternal fasting blood glucose, which has
glucose levels. a lower rate of requiring insulin. However, compared
Women with GDM are at higher risk of gestational with insulin, glyburide is associated with an increased in-
hypertension and preeclampsia. Metformin treatment cidence of neonatal hypoglycemia and higher BW. Over-
showed a statistically significant reduction in gestational all, it suggests that metformin can decrease the incidence
hypertension incidence compared with insulin in our of perinatal complications, while glyburide has advantag-
study. Gui et al. [64] reported that the incidence of preg- es in controlling maternal blood glucose. At last, more
nancy-induced hypertension in the metformin group was evidence is needed to explore the perinatal complications
significantly lower than that in the insulin group. The of glyburide in patients with GDM.
possible explanation is that metformin has complex prop-
erties on endothelial function and reactive oxygen species Strength and Limitation
production, thereby reducing endothelial activation and Our study had some important strengths. It is the me-
maternal inflammatory response to IR. There are insuf- ta-analysis of metformin, glyburide, and insulin on neo-
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 227

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
natal adverse outcomes, which has included the most up- Acknowledgement
dated studies in recent years. All the original studies used
We would like to appreciate Professor Li Jing from the Depart-
a randomized controlled study design, which greatly re- ment of Evidence-based Medicine and Clinical Epidemiology,
duced the likelihood of recall and selection biases. Sev- West China Hospital, for assistance of methodology.
eral limitations involved in this study should be consid-
ered. First, in order to reduce heterogeneity, we included
only trials comparing metformin with glyburide or insu- Statement of Ethics
lin, excluding, for example, trials comparing different in-
sulin regimens, insulin with diet, and OADs with diet. This review was conducted in accordance with the World Med-
Therefore, only 32 RCTs fulfilled the set eligibility criteria ical Association Declaration of Helsinki. Ethical approval was not
required for this study in accordance with local guidelines. Written
for inclusion. Furthermore, not all important outcomes informed consent to participate in the study was not required in
of interests were included in each study. Second, the dif- accordance with local guidelines. The protocol for this systematic
ferent doses of the drug used may lead to heterogeneity in review was registered on PROSPERO (CRD42018117434), and the
the study. We found that the dose of glyburide in the results were present­ed according to the Preferred Reporting Items
Moore et al. [13] study was 5 mg and Behrashi et al. [43] for Systematic Reviews and Meta-Analyses (PRISMA) statement.
was 1.25 mg. Of the included studies, two of them did not
report specific doses of OADs and insulin [19, 34], and 6 Conflict of Interest Statement
did not report specific doses of insulin [20–22, 29, 40, 47].
As a result, macrosomia and cesarean sections both The authors declare that they have no conflicts of interest to
showed greater heterogeneity. Due to the different types disclose.
and doses of insulin, we found that there is more signifi-
cant heterogeneity in the maternal fasting blood glucose,
postprandial blood glucose, and glycosylated hemoglo- Author Contributions
bin, which is difficult to adjust. Third, most of the includ- Study design: Xian Wang and Wanting Liu; project administra-
ed studies were not designed with blind methods due to tion: Xian Wang; study implementation and data collection: Xian
the different administration of OADs and insulin, caus- Wang, Wanting Liu, and Huizhen Chen; data analysis: Wang Xian
ing a relatively high risk of performance bias and detec- and Liu Wanting; writing of the manuscript: Wang Xian, Liu
tion bias, which may influence the results. Last, potential Wanting, and Chen Huizhen; all authors read and provided critical
comments on the manuscript. Chen Qiu supervised the study and
publication bias might influence the findings, yet little is a guarantor.
evidence of publication bias was observed. Therefore, fur-
ther studies with sufficient evidence should be conducted
to explore meaningful differences in maternal and neona- Data Availability Statement
tal-perinatal complications.
The datasets used and/or analyzed during the current study are
available from the corresponding author on reasonable request.

References
1 Alfadhli EM. Gestational diabetes mellitus. betes mellitus. Nat Rev Dis Primers. 2019; 5: 10 Pryor R, Cabreiro F. Repurposing metformin:
Saudi Med J. 2015;36(4):399–406. 47. an old drug with new tricks in its binding
2 Ramachandran A, Snehalatha C, Nanditha A. 6 Metzger B, Lowe L, Dyer A, Trimble E, Chao­ pockets. Biochem J. 2015;471(3):307–22.
Classification and diagnosis of diabetes. varindr U, Coustan D, et al. Hyperglycemia 11 Farrar D, Simmonds M, Bryant M, Sheldon
Hoboken, NJ: John Wiley & Sons, Ltd; 2016. and adverse pregnancy outcomes (HAPO). N TA, Tuffnell D, Golder S, et al. Treatments for
3 Wang B, Xue X. Investigations of associations Engl J Med. 2008;358:1991–2002. gestational diabetes: a systematic review and
between seven gene polymorphisms and ges- 7 Kimber-Trojnar Z, Marciniak B, Leszczyńska- meta-analysis. BMJ Open. 2017; 7(6):
tational diabetes mellitus: evidence from a Gorzelak B, Trojnar M, Oleszczuk J. Gly- e015557.
meta-analysis. Gynecol Obstet Invest. 2020; buride for the treatment of gestational diabe- 12 Balsells M, García-Patterson A, Solà I, Roqué
85(3):229–36. tes mellitus. Pharmacol Rep. 2008;60(3):308– M, Gich I, Corcoy R. Glibenclamide, metfor-
4 Li C, Qiao B, Zhou Y, Qi W, Ma C, Zheng L. 18. min, and insulin for the treatment of gesta-
Association of estrogen receptor α gene poly- 8 Practice bulletin No. 137: gestational diabetes tional diabetes: a systematic review and meta-
morphism and its expression with gestational mellitus. Obstet Gynecol. 2013; 122(2 Pt 1): analysis. BMJ. 2015;350:h102.
diabetes mellitus. Gynecol Obstet Invest. 406–16. 13 Moore LE, Clokey D, Rappaport VJ, Curet LB.
2020;85(1):26–33. 9 Introduction: standards of medical care in di- Metformin compared with glyburide in gesta-
5 McIntyre HD, Catalano P, Zhang C, Desoye abetes: 2018. Diabetes Care. 2018;41(Suppl 1): tional diabetes: a randomized controlled trial.
G, Mathiesen ER, Damm P. Gestational dia- S1–S2. Obstet Gynecol. 2010;115(1):55–9.
130.209.6.61 - 8/13/2021 10:39:25 PM

228 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
14 Jiang YF, Chen XY, Ding T, Wang XF, Zhu ized clinical trial. Acta Med Iran. 2015;53(2): 41 Ainuddin JA, Karim N, Zaheer S, Ali SS,
ZN, Su SW. Comparative efficacy and safety 97–103. Hasan AA. Metformin treatment in type 2 di-
of OADs in management of GDM: network 28 Nachum Z, Yefet E. Glyburide versus metfor- abetes in pregnancy: an active controlled, par-
meta-analysis of randomized controlled tri- min and their combination for the treatment allel-group, randomized, open label study in
als. J Clin Endocrinol Metab. 2015; 100(5): of gestational diabetes mellitus: a randomized patients with type 2 diabetes in pregnancy. J
2071. controlled study. Diabetes Care. 2017; 40(8): Diabetes Res. 2015;2015:325851.
15 Amin M, Suksomboon N, Poolsup N, Malik e116. 42 Ashoush S, El-Said M, Fathi H, Abdelnaby M.
O. Comparison of glyburide with metformin 29 Reynolds RM, Denison FC, Juszczak E, Bell Identification of metformin poor responders,
in treating gestational diabetes mellitus: a sys- JL, Penneycard J, Strachan MWJ, et al. Gliben­ requiring supplemental insulin, during ran-
tematic review and meta-analysis. Clin Drug clamide and metfoRmin versus stAndard care domization of metformin versus insulin for
Investig. 2015;35(6):343–51. in gEstational diabeteS (GRACES): a feasibil- the control of gestational diabetes mellitus. J
16 Higgins J, Green S. Cochrane handbook for ity open label randomised trial. BMC Preg- Obstet Gynaecol Res. 2016;42(6):640–7.
systematic reviews of interventions. 2009. nancy Childbirth. 2017;17(1):316. 43 Behrashi M, Samimi M, Ghasemi T, Saberi F,
17 Moher D, Liberati A, Tetzlaff J, Altman DG, 30 Sénat MV, Affres H, Letourneau A, Cous- Atoof F. Comparison of glibenclamide and
Tugwell P. Preferred reporting items for sys- tols-Valat M, Cazaubiel M, Legardeur H, et insulin on neonatal outcomes in pregnant
tematic reviews and meta-analyses: the PRIS- al. Effect of glyburide vs subcutaneous in- women with gestational diabetes. Int J Prev
MA statement. Rev Espanola Nutr Hum Diet. sulin on perinatal complications among Med. 2016 Jun;7:88.
2014;18(3):172–81. women with gestational diabetes: a ran- 44 da Silva AL, do Amaral AR, de Oliveira DS,
18 Langer O, Conway DL, Berkus MD, Xenakis domized clinical trial. JAMA. 2018; 319(17): Martins L, Silva MRE, Silva JC. Neonatal out-
EM, Gonzales O. A comparison of glyburide 1773–80. comes according to different therapies for
and insulin in women with gestational diabe- 31 Ghomian N, Vahed SHM, Firouz S, Yaghoubi gestational diabetes mellitus. Jornal De Pedia-
tes mellitus. N Engl J Med. 2000; 343(16): MA, Mohebbi M, Sahebkar A. The efficacy of tria. 2017;93(1):87–93.
1134–8. metformin compared with insulin in regulat- 45 Eid SR, Moustafa RSI, Salah MM, Hanafy SK,
19 Ogunyemi D, Jesse M, Davidson M. Compar- ing blood glucose levels during gestational Aly RH, Mostafa WFG, et al. Is metformin a
ison of glyburide versus insulin in manage- diabetes mellitus: a randomized clinical trial. viable alternative to insulin in the treatment of
ment of gestational diabetes mellitus. Endocr J Cell Physiol. 2019;234(4):4695–701. gestational diabetes mellitus (GDM)? Com-
Pract. 2007;13(4):427–8. 32 Galal M, El Bassioune WM, Sherif L. Metfor- parison of maternal and neonatal outcomes.
20 Rowan JA, Hague WM, Gao W, Battin MR, min versus insulin in treatment of gestational Egypt Pediatr Assoc Gaz. 2018;66(1):15–21.
Moore MP. Metformin versus insulin for the diabetes mellitus: a randomized controlled 46 George A, Mathews JE, Sam D, Beck M, Ben-
treatment of gestational diabetes. N Engl J trial. Res J Obstet Gynecol. 2019;12(1):23–7. jamin SJ, Abraham A, et al. Comparison of
Med. 2008;358(19):2003–15. 33 Lain KY, Garabedian MJ, Daftary A, Jeyaba- neonatal outcomes in women with gestation-
21 Ijas H, Vaarasmaki M, Morin-Papunen L, lan A. Neonatal adiposity following maternal al diabetes with moderate hyperglycaemia on
Keravuo R, Ebeling T, Saarela T, et al. Metfor- treatment of gestational diabetes with gly- metformin or glibenclamide: a randomised
min should be considered in the treatment of buride compared with insulin. Am J Obstet controlled trial. Aust N Z J Obstet Gynaecol.
gestational diabetes: a prospective ran- Gynecol. 2009;200(5):501–6. 2015;55(1):47.
domised study. BJOG. 2011;118(7):880–5. 34 Hague WM, Davoren PM, Oliver J, Rowan J. 47 Saleh HS, Abdelsalam WA, Mowafy HE, Abd
22 Hassan JA, Karim N, Sheikh Z. Metformin Contraindications to use of metformin. Met- ElHameid AA. Could metformin manage ges-
prevents macrosomia and neonatal morbidity formin may be useful in gestational diabetes. tational diabetes mellitus instead of insulin?
in gestational diabetes. Pak J Med Sci. 2012; BMJ. 2003;326(7392):762. Int J Reprod Med. 2016;2016:3480629.
28(3):384–9. 35 Bertini AM, Silva JC, Taborda W, Becker F, 48 Wasim T, Shaukat S, Javaid L, Mukhtar S,
23 Mukhopadhyay P, Bag TS, Kyal A, Saha DP, Lemos Bebber FR, Zucco Viesi JM, et al. Peri- Amer W. Comparison of metformin and in-
Khalid N. Oral hypoglycemic glibenclamide: natal outcomes and the use of oral hypoglyce- sulin for management of gestational diabetes
can it be a substitute to insulin in the man- mic agents. J Perinat Med. 2005;33(6):519–23. mellitus: a randomized control trial. Pak J
agement of gestational diabetes mellitus? A 36 Anjalakshi C, Balaji V, Balaji MS, Seshiah V. Med Health Sci. 2019;13(3):823–7.
comparative study. J SAFOG. 2012; 4(1): 28– A prospective study comparing insulin and 49 Carrington ER, Shuman CR, Reardon HS.
31. glibenclamide in gestational diabetes mellitus Evaluation of the prediabetic state during
24 Niromanesh S, Alavi A, Sharbaf FR, Amjadi in Asian Indian women. Diabetes Res Clin pregnancy. Obstet Gynecol. 1957;9(6):664.
N, Moosavi S, Akbari S. Metformin com- Pract. 2007;76(3):474–5. 50 Metzger BE, Gabbe SG, Persson B, Buchanan
pared with insulin in the management of ges- 37 Silva JC, Fachin DR, Coral ML, Bertini AM. TA, Catalano PA, Damm P, et al. Internation-
tational diabetes mellitus: a randomized clin- Perinatal impact of the use of metformin and al association of diabetes and pregnancy study
ical trial. Diabetes Res Clin Pract. 2012;98(3): glyburide for the treatment of gestational diabe- groups recommendations on the diagnosis
422–9. tes mellitus. J Perinat Med. 2012;40(3):225–8. and classification of hyperglycemia in preg-
25 Mesdaghinia E, Samimi M, Homaei Z, Saberi 38 Spaulonci CP, Bernardes LS, Trindade TC, nancy. Diabetes Care. 2010 Mar; 33(3): 676–
F, Moosavi SG, Yaribakht M. Comparison of Zugaib M, Francisco RP. Randomized trial of 82.
newborn outcomes in women with gestation- metformin vs insulin in the management of 51 WHO. Diagnostic criteria and classification
al diabetes mellitus treated with metformin or gestational diabetes. Am J Obstet Gynecol. of hyperglycaemia first detected in pregnan-
insulin: a randomised blinded trial. Int J Prev 2013;209(1):34–7. cy. Geneva: World Health Organization;
Med. 2013;4(3):327–33. 39 Tempe A, Mayanglambam RD. Glyburide as 2013.
26 Safura R, Safieh E, Zahra A. Neonatal out- treatment option for gestational diabetes mel- 52 American Diabetes Association. 13. Manage-
comes in women with gestational diabetes litus. J Obstet Gynaecol Res. 2013; 39(6): ment of diabetes in pregnancy: standards of
mellitus treated with metformin in compare 1147–52. medical care in diabetes-2018. Diabetes Care.
with insulin: a randomized clinical trial. J Res 40 Tertti K, Ekblad U, Koskinen P, Vahlberg T, 2018;41(Suppl 1):S137.
Med Sci. 2014;19(10):970–5. Rönnemaa T. Metformin vs. insulin in gesta- 53 Catalano PM, Tyzbir ED, Roman NM, Amini
27 Masoomeh M, Zahra H, Ziba F, Zahra N, Ra- tional diabetes. A randomized study charac- SB, Sims EA. Longitudinal changes in insulin
zyeh J. Comparison of glyburide and insulin terizing metformin patients needing addi- release and insulin resistance in nonobese
in women with gestational diabetes mellitus tional insulin. Diabetes Obes Metab. 2013; pregnant women. Am J Obstet Gynecol. 1991;
and associated perinatal outcome: a random- 15(3):246–51. 165(6 Pt 1):1667–72.
130.209.6.61 - 8/13/2021 10:39:25 PM

Comparison of Treatment for GDM Gynecol Obstet Invest 2021;86:218–230 229

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by:
54 American Diabetes Association. Diagnosis 61 Zhou J, Bai J, Guo Y, Fu L, Xing J. Higher lev- 68 Genxia Li SZ, Cui S, Li L, Xu Y, Li Y. Effect
and classification of diabetes mellitus. Diabe- els of triglyceride, fatty acid translocase, and comparison of metformin with insulin treat-
tes Care. 2014;37:1935–5548. toll-like receptor 4 and lower level of HDL-C ment for gestational diabetes: a meta-analysis
55 Koivusalo SB, Rn K, Klemetti MM, Roine RP, in pregnant women with GDM and their close based on RCTs. Arch Gynecol Obstetr. 2015;
Stach-Lempinen B. Gestational diabetes mel- correlation with neonatal weight. Gynecol 292(1):111–20.
litus can be prevented by lifestyle interven- Obstet Invest. 2021 [Online ahead of print]. 69 Zhu B, Zhang L, Fan YY, Wang L, Li XG, Liu
tion: the Finnish gestational diabetes preven- 62 Lund SS, Tarnow L, Frandsen M, Smidt UM, T, et al. Metformin versus insulin in gesta-
tion study (RADIEL). Diabetes Care. 2016; Pedersen O, Parving HH, et al. Impact of met- tional diabetes mellitus: a meta-analysis of
39(1):24–30. formin versus the prandial insulin secreta- randomized clinical trials. Ir J Med Sci. 2016;
56 Gunderson EP. Gestational diabetes and nu- gogue, repaglinide, on fasting and postpran- 185(2):371–81.
tritional recommendations. Curr Diab Rep. dial glucose and lipid responses in non-obese 70 Song R, Chen L, Chen Y, Si X, Liu Y, Liu Y, et
2004;4(5):377–86. patients with type 2 diabetes. Eur J Endocri- al. Comparison of glyburide and insulin in the
57 Poolsup N, Suksomboon N, Amin M. Effect nol. 2008;158(1):35–46. management of gestational diabetes: a meta-
of treatment of gestational diabetes mellitus: 63 Mocanu EV, Greene RA, Byrne BM, Turner analysis. PLoS One. 2017;12(8):e0182488.
a systematic review and meta-analysis. PLoS MJ. Obstetric and neonatal outcome of babies 71 Desoye G, Nolan CJ. The fetal glucose steal:
One. 2014;9(3):e92485. weighing more than 4.5 kg: an analysis by par- an underappreciated phenomenon in diabetic
58 ACOG practice bulletin No. 190: gestational ity. Eur J Obstet Gynecol Reprod Biol. 2000; pregnancy. Diabetologia. 2016; 59(6): 1089–
diabetes mellitus. Obstet Gynecol. 2018; 92:229–33. 94.
131(2):e49–64. 64 Gui J, Liu Q, Feng L. Metformin vs insulin in 72 Alsaleem MA-O, Saadeh L, Kamat D. Neo­
59 Bao L-X, Shi W-T, Han Y-X. Metformin ver- the management of gestational diabetes: a natal hypoglycemia: a review. Clin Pediatr
sus insulin for gestational diabetes: a system- meta-analysis. PLoS One. 2013;8:e64585. ­(Phila). 2019 Nov;58(13):1381 -6.
atic review and meta-analysis. J Matern Fetal 65 Gluckman PD, Hanson MA, Cooper C, 73 Su DF, Wang XY. Metformin vs insulin in the
Neonatal Med. 2019(1):1–13. Thornburg KL. Effect of in utero and early- management of gestational diabetes: a sys-
60 Guo L, Ma JA-O, Tang J, Hu D, Zhang W, life conditions on adult health and disease. N tematic review and meta-analysis. Diabetes
Zhao X. Comparative efficacy and safety of Engl J Med. 2008;359:61–73. Res Clin Pract. 2014 Jun;104:353–7.
metformin, glyburide, and insulin in treating 66 Araujo Júnior E, Peixoto AB, Zamarian AC, 74 Yerevanian A, Soukas AA. Metformin: mech-
gestational diabetes mellitus: a meta-analysis. Elito Júnior J, Tonni G. Macrosomia. Best Pract anisms in human obesity and weight loss.
2019;2019:9804708. Res Clin Obstet Gynaecol. 2017 Jan;38:83–96. Curr Obes Rep. 2019 Jun;8:156–64.
67 Brown J, Grzeskowiak L, Williamson K,
Downie MR, Crowther CA. Insulin for the
treatment of women with gestational diabetes.
Hoboken, NJ: John Wiley & Sons, Ltd; 2016.

130.209.6.61 - 8/13/2021 10:39:25 PM

230 Gynecol Obstet Invest 2021;86:218–230 Wang/Liu/Chen/Chen

DOI: 10.1159/000515893
Glasgow Univ.Lib.
Downloaded by: