Tot voor kort werd het gebruik van orale antidiabetica ontraden in de zwangerschap vanwege met name angst voor of risico op aangeboren aandoeningen bij de foetus. De voordelen van deze medicatie met betrekking tot de bloedglucoseregulatie en gebruiksgemak ten opzichte van insuline nopen toch om het gebruik hiervan tijdens de zwangerschap bij vrouwen met diabetes mellitus te onderzoeken.

Het aantal orale glucoseverlagende middelen is de laatste jaren flink uitgebreid maar voor gebruik in de zwangerschap zijn er alleen gegevens over metformine, glibenclamide en glyburide. Dit laatste middel wordt echter in Nederland niet gebruikt.

NICE gebruikt in 3.8 de volgende evidence levels (EL):

Levels of evidence for intervention studies

Level Source of evidence

1++ High-quality meta-analyses, systematic reviews of randomised controlled trials (RCTs), or RCTs with a very low risk of bias

1+ Well-conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias

1− Meta-analyses, systematic reviews of RCTs, or RCTs with a high risk of bias

2++ High-quality systematic reviews of case-control or cohort studies; high-quality case-control or cohort studies with a very low risk of confounding, bias or chance and a high probability that the relationship is causal

2+ Well-conducted case-control or cohort studies with a low risk of confounding, bias or chance and a moderate probability that the relationship is causal

2− Case-control or cohort studies with a high risk of confounding, bias or chance and a significant risk that the relationship is not causal

3 Non-analytical studies (for example, case reports, case series)

4 Expert opinion, formal consensus

3.8 Safety of medicines for diabetes before and during pregnancy: Oral hypoglycaemic agents

Oral hypoglycaemic agents are used to maintain blood glucose control in people with type 2 diabetes. There are four main categories of oral hypoglycaemic agents: sulphonylureas (chlorpropamide, glibenclamide (also known as glyburide), gliclazide, glimepiride, glipizide, gliquidone and tolbutamide), biguanides (metformin), α-glucosidase inhibitors (acarbose) and thiazolidinediones (pioglitazone and rosiglitazone). Two other medications are used to stimulate insulin release (nateglinide and repaglinide). Each works in a different way and is suitable for different clinical situations.

A systematic review and meta-analysis examined the relationship between first-trimester exposure to oral hypoglycaemic agents and subsequent congenital abnormalities and neonatal mortality (Gutzin, 2003). The review sought to account for the potential confounding effect of maternal glycaemic control. The meta-analysis included ten studies which reported on 471 women exposed to oral hypoglycaemic agents in the first trimester and 1344 women not exposed. There were three prospective cohort studies, three retrospective cohort studies, three case series and one case– control study. The oral hypoglycaemic agents used in the studies included chlorpropamide (eight studies), tolbutamide (six studies), glibenclamide (four studies), metformin (five studies) and phenformin (three studies). Six studies were rated as of ‘poor’ quality, two were ‘fair’ and two were ‘good’. Most women in the studies had type 2 diabetes. Women with type 1 diabetes and gestational diabetes were also present in some studies. There was no significant difference in the rate of major malformations between those exposed to oral hypoglycaemic agents and those not exposed (10 studies, OR 1.05, 95% CI 0.65 to 1.70) [EL=2++].

A further systematic review included seven additional studies that looked at the use of metformin in the first trimester of pregnancy (Hawthorne, 2006). All the studies were small observational studies, six in women with polycystic ovary syndrome and one in women with type 2 diabetes. There were no reports of birth defects, increased incidence of pre-eclampsia or other adverse maternal or neonatal outcomes. The use of metformin during pregnancy in women with polycystic ovary syndrome is associated with a reduction in miscarriage in early pregnancy, weight loss, a reduction in fasting serum insulin levels and in the incidence of gestational diabetes [EL=2+].

Another systematic review evaluated the safety of metformin in pregnancy for women with diabetes and polycystic ovary syndrome (Gilbert, 2006). The review reported that metformin use during the first trimester of pregnancy was not associated with an increased risk of major malformations [EL=2+].

An RCT compared glibenclamide with insulin for the treatment of gestational diabetes (Langer, 2000). The study involved 404 women with gestational diabetes. Approximately 83% were Hispanic, 12% were white and 5% were black. Women were randomly assigned between 11 and 33 weeks of gestation to receive either glibenclamide (n=201) or insulin (n=203). The mean blood glucose concentrations during treatment were 5.9 ± 0.9 mmol/litre in the glibenclamide group and 5.9 ± 1.0 mmol/litre in the insulin group (P=0.99). Eight women (4%) in the glibenclamide group required insulin treatment. There were no significant differences between the glibenclamide and insulin groups in the percentage of babies who were LGA (12% versus 13%), macrosomic (7% versus 4%), had lung complications (8% versus 6%) or neonatal hypoglycaemia (9% versus 6%), were admitted to a NICU (6% versus 7%), or had fetal anomalies (2% and 2%). Cord-serum insulin concentrations were similar in the two groups and glibenclamide was not detected in the cord serum of any baby in the glibenclamide group [EL=1++].

An RCT of metformin for the treatment of gestational diabetes (the Metformin in Gestational Diabetes (MIG) trial) is due to report soon. A pilot study randomised 14 women to insulin and 16 to metformin. There were no differences in perinatal outcomes (Elder, 2003) [EL=1−].

A retrospective cohort study examined the effects of oral hypoglycaemic agents in 379 singleton pregnancies of women with type 2 diabetes (Ekpebegh, 2007). The women were subdivided into: oral hypoglycaemic agents alone (n=93 pregnancies); converted from oral hypoglycaemic agents to insulin (n=249); and insulin alone or converted from diet alone to insulin (n=37). The oral hypoglycaemic agents assessed were metformin and glibenclamide. Fetal anomaly rates were similar across the three groups, whereas perinatal mortality rates (per 1000 births) were higher in the group that used oral hypoglycaemic agents alone (125, P=0.003). Conversion from oral hypoglycaemic agents to insulin was protective for perinatal mortality compared with oral hypoglycaemic agents alone (OR 0.220, 95% CI 0.061 to 0.756, P=0.024). The data suggest that metformin and glibenclamide are not teratogenic [EL=2+].

A reference guide to medicines in pregnancy and lactation reported that there were limited data on the use of metformin, acarbose, nateglinide, glimepiride, glipizide and glibenclamide in pregnant women and suggested they present a low risk to the fetus. No data were found on the use of repaglinide or pioglitazone in pregnant women, but it was suggested that they present a moderate risk to the fetus. No comparative studies were found on the use of rosiglitazone in pregnant women, but it was suggested that it presents a risk to the fetus. Evidence suggested that chlorpropamide and tolbutamide present a risk to the fetus if taken by women in the third trimester of pregnancy. Gliclazide and gliquidone were not reviewed (Briggs, 2005) [EL=3].

The reference guide reported that 14 small observational studies had investigated the use of metformin in pregnant women with diabetes. Metformin has been shown to cross the placenta. The observational studies identified congenital malformations in some babies of women taking metformin, but the rate was not compared with the expected rate of congenital malformations in babies born to women with diabetes who were not taking metformin. The reference guide suggested that use of metformin may decrease fetal and infant morbidity and mortality in developing countries where the proper use of insulin is problematic, but that insulin is still the treatment of choice (Briggs, 2005) [EL=3].

…

The reference guide reported that several observational studies had investigated the use of glibenclamide in pregnant women. Small amounts of glibenclamide may cross the placenta. The observational studies suggested that use of glibenclamide may decrease fetal and infant morbidity and mortality in developing countries where the proper use of insulin is problematic, and the reference guide suggested that it may be an acceptable alternative to insulin for women with gestational diabetes (Briggs, 2005) [EL=3].

…

The British National Formulary recommends that metformin, acarbose and repaglinide should be avoided during pregnancy and are normally substituted with insulin (Joint Formulary Committee, 2007). The manufacturers of nateglinide, pioglitazone and rosiglitazone advise pregnant women to avoid them, and insulin is normally substituted in women with diabetes. Sulphonylureas can lead to neonatal hypoglycaemia, and insulin is normally substituted in women with diabetes. If oral hypoglycaemic medicines are used then therapy should be stopped at least 2 days before birth.

4.5 Interventions in women with gestational diabetes: pharmacological interventions (metformin, glibenclamide and insuline)

4.5.3 Introduction

The purpose of this review is to determine the effectiveness of interventions for gestational diabetes mellitus, alone or in combination. The search included randomised controlled trials (RCTs) and systematic reviews of RCTs. The same search criteria were used to identify results for each of the interventions addressed by this review. Interventions included:

• metformin – a biguanide which lowers hepatic glucose output and improves insulin sensitivity: this has a low risk of hypoglycaemia and does not cause weight gain
• glibenclamide – a sulfonylurea which works by stimulating insulin release from the pancreas: this can cause hypoglycaemia and is associated with weight gain.

The following combinations of intervention were identified by the guideline development group as being potentially important for reducing maternal and fetal adverse outcomes in women with gestational diabetes mellitus:

• pharmacological interventions:

• metformin versus Insulin;
• glibenclamide versus insulin;
• glibenclamide versus metformin.

4.5.4 Description of included studies

4.5.4.3 Pharmacological interventions

Fifteen randomised controlled trials met the inclusion criteria for this review (Bertini, 2005; Hague, 2003; Ijäs, 2010; Lain, 2009; Langer, 2000; Mesdaghinia, 2013; Moore, 2007; Moore, 2010; Mukhopadhyay, 2012; Niromanesh, 2012; Ogunyemi, 2007; Rowan, 2008; Silva, 2012; Spaulonci, 2013; Tertti, 2013).

Trial locations included Finland (Ijäs, 2010; Tertti, 2013), Australia (Hague, 2003; Rowan, 2008), the USA (Lain, 2009; Langer, 2000; Moore, 2007; Moore,2010; Ogunyemi, 2007), Brazil (Bertini, 2005; Silva, 2012; Spaulonci, 2013), India (Mukhopadhyay, 2013) and Iran (Mesdaghinia, 2013; Niromanesh, 2012). The number of participants ranged from 30 to 404 women.

All included women had gestational diabetes. An oral glucose tolerance test (OGTT) was used to diagnose women with gestational diabetes, although different glucose loads and diagnostic criteria were used in the trials and women could be diagnosed as early as 11 weeks of gestation in some studies.

4.5.4.3.1 Metformin versus insulin

Eight trials compared metformin with insulin (Hague, 2003; Ijäs, 2010; Mesdaghinia, 2013; Moore, 2007; Niromanesh, 2012; Rowan, 2008; Spaulonci, 2013; Tertti, 2013).

Two trials did not provide treatment details with regard to the doses of metformin and insulin that were used, nor were further details given with regard to any concurrent dietary or exercise interventions or monitoring techniques used (Hague, 2003; Spaulonci, 2013). Women in the remaining six trials received dietary advice/instruction and glucose levels were monitored (Ijäs, 2010; Mesdaghinia, 2013; Moore, 2007; Niromanesh, 2012; Rowan, 2008; Tertti, 2013). Women in three trials also received lifestyle counselling/exercise advice (Ijäs, 2010; Niromanesh, 2012; Rowan, 2008).

The initial dose of metformin was 500 mg/day (Mesdaghinia, 2013; Moore, 2007; Niromanesh, 2012; Rowan, 2008; Tertti, 2013) or 750 mg/day (Ijäs, 2010) increasing to a maximum dose of 2000 mg/day (Moore, 2007; Tertti, 2013), 2250 mg/day (Ijäs, 2009) or 2500 mg/day (Mesdaghinia, 2013; Niromanesh, 2012; Rowan, 2008). In the seven trials that described treatments, treatment failure with metformin occurred when normoglycaemia or targets were not reached on a maximum dose (Niromanesh, 2012; Tertti, 2013), or were not reached on a maximum dose over a one to two week period (Ijäs, 2010). One trial did not report criteria for starting insulin therapy when treatment with metformin failed but did report the use of supplementary insulin (Spaulonci, 2013). In four trials (Ijäs, 2010; Niromanesh, 2012; Rowan, 2008; Tertti, 2013) when failure of treatment with metformin occurred, supplementary insulin was added as required. In one study (Moore, 2007) metformin was stopped and insulin started. In one trial women who failed treatment with metformin were excluded from the study and from analyses and were replaced with women who did not fail treatment (Mesdaghinia, 2013).

Insulin treatment consisted of long-acting and rapid-acting insulin in one trial (Ijäs, 2010), split doses of regular insulin and NPH insulin in three trials (Mesdaghinia, 2013; Moore, 2007; Niromanesh, 2012) and NPH insulin and/or rapid acting insulin lispro or aspart in one trial (Tertti, 2013). One trial (Rowan, 2008) stated that insulin was prescribed according to usual practice.

Maternal outcomes reported were mode of birth, treatment failure (need for additional treatment with insulin) and acceptability of treatment. Outcomes related to mode of birth were spontaneous vaginal birth, induction of labour, operative vaginal, vacuum extraction and caesarean section. Acceptability of treatment was assessed using a questionnaire in one trial (Rowan, 2008). Reported neonatal outcomes were large for gestational age, NICU admission and length of stay, shoulder dystocia, neonatal hypoglycaemia fetal death and perinatal mortality. One composite outcome (Rowan, 2008) was included owing to its use in the health economic analysis for this review.

4.5.4.3.2.Glibenclamide versus insulin

Five trials compared glibenclamide with insulin (Bertini, 2005; Lain, 2009; Langer, 2000; Mukhopadhyay, 2012; Ogunyemi, 2007).

One trial (Langer, 2000) prescribed a diet for normal weight and obese women and confirmed all women performed self-monitoring of blood glucose. One trial prescribed a diet with caloric intake calculated according to pre-gestational BMI and asked women to self-monitor seven times a day; women included in the trial were those who had failed dietary treatment (Mukhopadhyay, 2012). However, three trials (Bertini, 2005; Lain, 2009; Ogunyemi, 2007) did not provide details of any concurrent dietary or exercise interventions, although one of these (Ogunyemi, 2007) was conducted in a population of women in whom diet therapy had failed. One trial reported that blood glucose was tested in the clinic weekly. The two remaining trials (Lain, 2009; Ogunyemi, 2007) did not detail the monitoring techniques used.

One trial did not provide treatment details with regard to the doses of metformin and insulin that were used (Ogunyemi, 2007). The initial dose of glibenclamide in the remaining four trials was 2.5 mg/day (Lain, 2009; Langer, 2000; Mukhopadhyay, 2012) or 5 mg/day (Bertini, 2005), increasing weekly to a maximum dose of 20 mg/day (Bertini, 2005; Lain, 2009; Langer, 2000; Mukhopadhyay, 2012). In all four trials, glibenclamide treatment failure occurred when glucose control was not achieved despite taking the maximum dose. Oral medication was stopped and insulin therapy started. Two trials stipulated that glucose control was assessed over a two week period for a decision of treatment failure to be made (Langer, 2000; Mukhopadhyay, 2012).

Insulin doses started at 0.7 U/kg (Bertini, 2005; Langer, 2000; Mukhopadhyay, 2012) or 0.8 U/kg (Langer, 2000) administered in multiple daily injections.

Maternal outcomes reported were caesarean section, treatment failure (need for additional treatment with insulin) and acceptability of treatment. Acceptability of treatment was assessed using reporting of maternal hypoglycaemic episodes. Reported neonatal outcomes were large for gestational age, NICU admission, shoulder dystocia, neonatal hypoglycaemia, intrauterine and neonatal mortality.

4.5.4.3.3 Glibenclamide versus metformin

Two trials compared metformin with glibenclamide (Moore, 2010; Silva, 2012).

In both trials all women were given instructions for a diet prescribed according to normal or obese weight and performed self-monitoring of blood glucose. In addition, in one trial the importance of exercise in controlling blood glucose was stressed and 30 minutes of walking per day was recommended (Moore, 2010).

In both trials, the initial dose of glibenclamide was 2.5 mg two times a day which, following a weekly review, was increased as necessary to a maximum dose of 20 mg/day.

In both trials, the initial dose of metformin was 500 mg/day in divided doses and was increased as necessary following a weekly review until glucose control was achieved or until a maximum dose was reached of 2000 mg/day (Moore, 2010) or 2500 mg/day (Silva, 2012).

Treatment failures were defined in one trial (Moore, 2010) as occurring when two or more glucose values in the same meal exceeded target glucose values by 10 mg/dl (0.56 mmol/litre) or more for two consecutive weeks while the maximum dose was taken. In this trial, oral medication was stopped and insulin therapy started. In the second trial (Silva, 2012), insulin therapy was started at 0.7 IU/kg/day regular insulin preprandially and neutral protamine hagedorn (NPH) insulin at bedtime when glycaemic goals were not met. Of the guideline development group’s priority outcomes, maternal data were available for caesarean section and treatment failure (need for additional treatment with insulin). Acceptability of treatment was assessed using reporting of maternal hypoglycaemic episodes.

Reported neonatal outcomes were large for gestational age, NICU admission, shoulder dystocia, neonatal hypoglycaemia, stillbirth and neonatal mortality.

A summary of the details of the studies of pharmacological interventions is shown in Table 36.

4.5.5 Evidence profile

The GRADE profiles for this review question are presented in tables 42 to 44:

Table 42 GRADE profile for comparison of metformin and insulin in women with gestational diabetes

Table 43 GRADE profile for comparison of glibenclamide and insulin in women with gestational diabetes

Table 44 GRADE profile for comparison of metformin and glibenclamide in women with gestational diabetes 

4.5.6 Evidence statements

4.5.6.3 Pharmacological interventions

4.5.6.3.1 Metformin versus insulin (table 42)

Maternal outcomes

Mode of birth

Seven outcomes pertaining to mode of birth were reported in seven RCTs that compared metformin with insulin treatment in women with gestational diabetes. Heterogeneity prevented full meta-analysis of data for spontaneous vaginal birth (two trials), induction of labour (four trials) and elective caesarean section (two trials) and only one trial presented data on vacuum extraction, emergency caesarean section and assisted vaginal delivery.

Spontaneous vaginal birth

Data were conflicting in two trials that reported spontaneous vaginal birth. One small trial (n=30) found fewer spontaneous vaginal births among women who received metformin compared to those who received insulin (RR 0.4, 95% CI 0.18 to 0.86) and the evidence was of low quality, while another trial (n=97) provided low quality evidence which found no difference (RR 0.98, 95% CI 0.67 to 1.45).

Induction of labour

A meta-analysis of data from 3 trials (n=343) found that the risk of requiring induction of labour among women who received metformin compared with insulin treatment was lower (RR 0.67, 95% CI 0.54 to 0.83) and the evidence was of low quality. One further trial (n=733) found no difference in labour induction rates between treatment groups (RR 0.96, 95% CI 0.84 to 1.09) and the evidence was considered of moderate quality.

Vacuum extraction

One trial (n=97) provided low quality evidence of no difference between treatment groups in the number of births requiring vacuum extraction (RR 1.86, 95% CI 0.58 to 5.95).

Caesarean section

A meta-analysis of data from seven trials (n=1391) found no difference in the risk of caesarean section among women who received metformin compared with those receiving insulin (RR 1.00, 95% CI 0.87 to 1.15) and the evidence was of moderate quality.

Elective caesarean section

Two trials (n=30; n =733) reported no difference in rates of elective caesarean section (RR 3.5, 95% CI 0.89 to 13.82 and RR 0.89, 95% CI 0.64 to 1.24, respectively) and the evidence was of low quality.

Emergency caesarean section

One trial (n=41) provided low quality evidence which demonstrated no difference in the risk of emergency caesarean section between groups (RR 1.6, 95% CI 0.9 to 2.7).

Assisted vaginal delivery

One further trial (n=216) found no difference in the risk of assisted vaginal delivery (RR 1.10, 95% CI 0.44 to 2.74) and the evidence was of very low quality.

Need for additional insulin

In terms of need for additional insulin, 36% of women who received metformin in 5 trials (206 out of 568 participants) had a treatment failure requiring insulin. The evidence for this finding was of moderate quality.

Acceptability

One trial (n=664) assessed the acceptability of metformin compared with insulin. More women forgot to take metformin compared with insulin (p<0.001) during the course of the study. However, more women would choose metformin over insulin in a subsequent pregnancy (p<0.001). If given the choice in a subsequent pregnancy, more women would start with metformin and add insulin if needed (p<0.001).

In terms of assessing the easiest part of the diabetes treatment, more women in the insulin group said that doing the finger-prick test was easiest. Less women in the metformin group (n=63 versus n=95) said that being careful with diet was the easiest aspect. More women in the metformin group (n=197 versus n=117) said that taking the medication was the easiest.

Neonatal outcomes

NICU admission

A meta-analysis of five trials (n=736) found a reduction in the risk of admission to NICU in babies of women with gestational diabetes who received metformin compared with babies of women who received insulin (RR 0.69, 95% CI 0.52 to 0.92). The quality of the evidence for this outcome was very low. An additional trial (n=733) found no difference between groups for a NICU stay of more than 24 hours for babies of women with gestational diabetes who received metformin compared with babies of women who received insulin (RR 1.04, 95% CI 0.71 to 1.53).

Large for gestational age/macrosomia

There were no differences in any other neonatal outcomes when metformin was compared with insulin treatment. Outcomes examined included large for gestational age, which included a meta-analysis of data from five trials (n=1406) that found no difference between treatment groups (RR 0.85, 95% CI 0.68 to 1.05) and was of very low quality. One trial (n=92) assessed macrosomia as a proxy for large for gestational age and the very low quality evidence found no difference between groups (RR 0.14, 95% CI 0.007 to 2.64).

Neonatal morbidity

One trial (n=733) assessed a composite neonatal morbidity outcome (RR 0.99, 95% CI 0.8 to 1.23) and found moderate quality evidence of no difference between groups.

Shoulder dystocia

Data from four trials (n=1156) were combined in a meta-analysis for shoulder dystocia (RR 0.76, 95% CI 0.36 to 1.59) which found low quality evidence of no difference between groups.

Neonatal hypoglycaemia

Six trials (n=828) combined in a meta-analysis examined neonatal hypoglycaemia and found very low quality evidence of no difference between groups (RR 0.71, 95% CI 0.48 to 1.04).

Supplemental feeding

One trial (n=733) found no difference between groups using supplemental feeding (RR 0.91, 95% CI 0.75 to 1.09); evidence was moderate quality.

Intravenous dextrose

Two trials (n=7636) that meta-analysed data for intravenous dextrose found low quality evidence of no difference between groups (RR 1.27, 95% CI 0.75 to 2.15).

Fetal death

One trial (n=733) reported no difference between groups for fetal death (RR 0.34, 95% CI 0.01 to 8.31); evidence was very low quality.

Perinatal mortality

There were no cases of perinatal mortality in one trial (n=97) which was considered of very low methodological quality.

4.5.6.3.2 Glibenclamide versus insulin (table 43)

Maternal outcomes

Caesarean section

A meta-analysis of two trials (n=139) found no difference in risk of caesarean section for women with gestational diabetes who received glibenclamide compared with those who received insulin (RR 0.87, 95% CI 0.61 to 1.23). The evidence for this outcome was of very low quality.

Need for additional insulin

In four trials, 5.9% (19 out of 322) of women who received glibenclamide experienced a treatment failure that required insulin. The evidence for this finding was of low quality.

Hypoglycaemia

One trial (n=33) reported no difference in the number of women with gestational diabetes experiencing hypoglycaemia when treated with glibenclamide compared with insulin treatment (RR 1.23, 95% CI 0.7 to 2.14). The quality of evidence for this finding was very low.

Neonatal outcomes

Large for gestational age

A meta-analysis of three trials (n=190) found that women with gestational diabetes who had been treated glibenclamide had a greater risk of giving birth to babies that were large for gestational age compared with women who had received treatment with insulin (RR 3.62, 95% CI 1.54 to 8.49). The evidence for this outcome was of low quality. A fourth trial (n=404) found no difference between groups in the risk of giving birth to babies that were large for gestational age (RR 0.93, 95% CI 0.55 to 1.57). The evidence for this outcome was of low quality.

Neonatal hypoglycaemia

A meta-analysis of five trials (n=702) found higher risk of neonatal hypoglycaemia in the babies of women who received glibenclamide compared with those who received insulin (RR 2.13, 95% CI 1.32 to 3.43). The evidence for this outcome was of low quality.

Other outcomes

There were no differences in any other reported neonatal outcomes when glibenclamide was compared with insulin treatment in women with gestational diabetes. Data from three trials (n=554) were combined in a meta-analysis (RR 1.22, 95% CI 0.63 to 2.37) and provided low quality evidence for NICU admission; one trial (n=99) provided very low quality evidence for the outcome of shoulder dystocia (RR 0.51, 95% CI 0.05 - 5.45) and one trial (n=404) for intravenous glucose therapy (RR 1.29, 95% CI 0.76 to 2.17) which was of moderate quality. One trial (n=90) reported very low quality evidence finding no difference in the risk of intrauterine death between groups (RR 3.73, 95% CI 0.16 to 89.21). One trial (n=404) reported on risk of stillbirth (RR 1.01, 95% CI 0.06 to 16.04) which was of low quality evidence. Three trials reporting data for neonatal death (n=554) were combined in a meta-analysis (RR 1.01, 95% CI 0.06 to 16.04) giving very low quality evidence.

4.5.6.3.3 Glibenclamide versus metformin (table 44)

Maternal outcomes

Caesarean section

One trial (n=149) reported a lower risk of non-elective caesarean section in women with gestational diabetes who received glibenclamide compared with those who received metformin (RR 0.18, 95% CI 0.04 to 0.8). The evidence for this outcome was of moderate quality.

Need for additional insulin

Two trials presented conflicting findings with regard to treatment failure. One trial (n=149) found lower risk of treatment failure requiring insulin in women who received glibenclamide compared with those who received metformin (n=0.47, 95% CI 0.26 to 0.86), while the other trial (n=200) found no difference between treatment groups (RR 1.38, 95% CI 0.85 to 2.24). The evidence for both of these outcomes were of moderate quality.

Maternal hypoglycaemia

One trial (n=149) reported no difference in the risk of maternal hypoglycaemia between women with gestational diabetes who received treatment with glibenclamide compared with those who received metformin (RR 0.51, 95% CI 0.05 to 5.47). The evidence for this outcome was of low quality.

Neonatal outcomes

Large for gestational age

One trial (n=200) provided low quality evidence that demonstrated an increased risk of delivering a baby that was large for gestational age associated with glibenclamide compared with metformin (RR 2.29, 95% CI 1.09 to 4.81).

Other outcomes

There were no differences between groups for any other neonatal outcomes when treatment with glibenclamide was compared with treatment with metformin in women with gestational diabetes. Outcomes reported were neonatal hypoglycaemia, which included one trial (n=149) which was of low quality (RR 0.34, 95% CI 0.01 to 8.16) and another trial (n =200) which was also of low quality (RR 1.28, 95% CI 0.6 to 2.72). One trial (n=149) provided very low quality evidence for shoulder dystocia (RR 3.04, 95% CI 0.13 to 73.44). Data from two trials (n=346) assessing NICU admission was combined in a meta-analysis and assessed as being of very low quality (RR 0.66, 95% CI; 0.28 to 1.55). One trial (n=200) reported very low quality evidence which found no difference between groups for the risk of neonatal death (RR 1.08, 95% CI 0.07 to 17.08).

Om de uitgangsvraag te kunnen beantwoorden is er een systematische literatuuranalyse verricht naar de volgende wetenschappelijke vraagstelling:

1. Wat is de effectiviteit van metformine, glibenclamide en insuline bij vrouwen met GDM?
2. Zijn er ongewenste effecten voor de foetus/neonaat en de zwangere vrouw bij het gebruik van orale glucoseverlagende middelen (metformine en glibenclamide) bij een vrouw met diabetes mellitus type 2 of GDM die niet of onvoldoende reageert op een dieet?

De klinisch relevante vergelijkingen zijn die van metformine versus insuline (= gouden standaard), glibenclamide versus insuline (=gouden standaard) en metformine vs glibenclamide.

Relevante uitkomstmaten

De werkgroep achtte de items aangeboren afwijkingen bij de foetus/pasgeborene, macrosomie (P>95) van de foetus/pasgeborene, geboortetrauma bij de pasgeborene en hypoglykemie bij de moeder voor de besluitvorming als kritieke uitkomstmaten en de items Apgar-score <7 na 5 minuten, prematuur zijn van de pasgeborene; NICU-opname; noodzaak voor fototherapie, hypoglykemie bij de foetus/pasgeborene; complicaties aan de longen en afwijkingen op lange termijn bij de pasgeborene en gastro-intestinale bijwerkingen en therapiefalen bij de moeder voor de besluitvorming belangrijke uitkomstmaten.

Zoeken en selecteren (Methode)

In de databases Medline (OVID), Embase and Cochrane is met relevante zoektermen gezocht naar systematische reviews en RCTs die orale bloedglucose verlagende middelen vergelijken met insuline bij zwangere vrouwen met diabetes mellitus type 2 of GDM. De zoekverantwoording is weergegeven onder het tabblad Verantwoording. De literatuurzoekactie leverde 319 treffers op. Studies werden geselecteerd op grond van de volgende selectiecriteria:

1. relevant voor beantwoording van de uitgangsvraag;
2. betreft primair (origineel) onderzoek of systematische review;
3. studiepopulatie komt overeen met patiëntengroep waarvoor richtlijn bedoeld is;
4. uitkomstmaten komen overeen met gekozen uitkomstmaten.

Op basis van titel en abstract werden in eerste instantie 77 studies voorgeselecteerd. Na raadpleging van de volledige tekst, werden vervolgens 67 studies geëxcludeerd (zie exclusietabel) en tien studies definitief geselecteerd.

Direct na de selectieperiode bleek een NICE-richtlijn (februari, 2015) met onder meer dezelfde uitgangsvragen te zijn verschenen. De werkgroep heeft deze richtlijn vergeleken met de eigen search en uitkomsten en heeft besloten dat deel van de NICE-richtlijn wat betreft literatuursamenvatting integraal over te nemen (NICE guideline Diabetes in Pregnancy, February, 2015). De werkgroep heeft ervoor gekozen de samenvatting en beoordeling van de literatuur over te nemen en niet de paragrafen waarin de mening of de overwegingen van de Britse werkgroep zijn opgenomen. De tekst, evidence tabellen en beoordeling van individuele studiekwaliteit zijn derhalve integraal overgenomen en kunt u onder het tabblad Onderbouwing vinden.

De NICE-richtlijn is op onderdelen herzien en geactualiseerd. Het gedeelte over ‘Safety of medicines for diabetes before and during pregnancy’ (hoofdstuk 3.8) is niet veranderd en bevat de tekst uit 2008. Het hoofdstuk ‘Interventions in women with gestational diabetes’ (hoofdstuk 4.5) is wel geactualiseerd (met behulp van GRADE). Als gevolg hiervan is de manier van beoordelen en beschrijven van de literatuur verschillend voor de twee hoofdstukken.

Uit de NICE-richtlijn zijn overgenomen:

Uit hoofdstuk 3.8: Safety of medicines for diabetes before and during pregnancy:

3.8.1 Description of the evidence

3.8.1.1 Oral hypoglycaemic agents: alleen de tekst over orale bloedsuikerverlagende middelen

Uit hoofdstuk 4.5: Interventions in women with gestational diabetes:

4.5.3 Introduction (alleen de tekst over pharmacological interventions)

4.5.4 Description of included studies

4.5.4.3. Pharmacological interventions

4.5.4.3.1 Metformin vs insulin

4.5.4.3.2 Glibenclamide vs insulin

4.5.4.3.3 Glibenclamide vs metformin

4.5.5 Evidence profile

4.5.5.3 Pharmacologial interventions

4.5.6 Evidence statements

4.5.6.3 Pharmacological interventions

4.5.6.3.1 Metformin vs insulin

4.5.6.3.2 Glibenclamide vs insulin

4.5.6.3.3 Glibenclamide vs metformin