The second controversy would be over whether universal or selective screening should be done. The selective principle means that the screening model would be limited only in the selected group of pregnant women with various risk factors for GDM that existed before pregnancy or are developed during pregnancy. It has been shown that a consequence of risk factor screening is that a significant number of women with impaired carbohydrate metabolism would remain unrecognized and at risk of perinatal complications [6, 7]. There is no properly conducted randomized control trial with a sufficient number of patients examined for the benefit of selective or universal screening for GDM compared with no screening. Moreover, performances of screening would also vary according to the population studied, frequency of risk factors and the thresholds used. Most publications show that between 3 and 10 % of women with GDM are not diagnosed by selective screening but this can go to as high as 30–50 % in some series.

Currently, the ACOG, SOGC and NICE recommend routine risk-factor-based screening, whereas the USPSTF, ADA, IADPSG, ADIPS, CDA and ATLANTIC DIP network recommend universal screening in asymptomatic pregnant women at 24–28 weeks of gestation, followed by definitive testing only in those women who are labeled as high-risk population.

Timing of screening most commonly mentioned in the literature is between 24 and 28 weeks, but it has to be emphasized that this recommendation specifically refers to healthy women without anamnestic risk factors. GDM classically occurs in the third trimester of pregnancy (from 24 weeks) due to physiopathological changes in glucose metabolism during pregnancy, including progressive insulin resistance caused by increased placental secretion of diabetogenic hormones growth hormone (GH), corticotropin releasing hormone (CRH) which drives release of adrenocorticotrophic hormone (ACTH) and cortisol, human placental lactogen and progesterone and a postreceptor defect. It is not logical to screen after 28 weeks of pregnancy because the initial phase of fetal growth acceleration has already begun.

Early GDM accounts for between 15 and 20 % of cases, although a high proportion of early GDM cases are probably undiagnosed type 2 diabetes mellitus. Screening for overt diabetes not diagnosed before pregnancy before 20 weeks of gestation is in general consensus recommended for populations at high risk for type 2 diabetes mellitus. These include women with a previous diagnosis of gestational diabetes, a first-degree relative with diabetes, a family origin with a high prevalence of diabetes, polycystic ovarian syndrome (PCOS), a previous large for gestational age baby and severe obesity. The NICE guidelines recommend screening for women with a previous diagnosis of GDM at 16–18 weeks [8]. Other guidelines by the ADA recommend screening either fasting or random glucose levels, 2-h glucose post OGTT or the HbA1c in such populations at first booking [9].

For healthy pregnant women, energy needs are no higher than the estimated energy requirement for nonpregnant women until the second trimester [10]. The extra energy need is 340 kcal in the second and 450 kcal in the third trimester. Prepregnancy body mass index, rate of weight gain, maternal age, and appetite must be considered when tailoring this recommendation to the individual. Moreover, for obese women with GDM, caloric restriction of 30 % may improve glycemic control without increasing ketonuria.

There is scant level 1 evidence to support most aspects of nutritional prescription for GDM. However, it does make clinical sense that the goals of medical nutritional therapy should be to achieve normoglycaemia, prevent ketosis, provide adequate weight gain based on their body mass index and contribute to fetal well-being. Nutrition therapy has been shown to improve glycemic control for people living with overt diabetes and for women with gestational diabetes [11]. Hence, nutritional therapy is widely recommended as an integral part of the treatment of women with gestational diabetes mellitus.

The acceptable macronutrient distribution ranges for consumption of carbohydrates, protein and fat as a percentage of total energy intakes for healthy pregnant women are estimated to be 45–65 % for carbohydrate, 10–35 % for protein and 20–35 % for fat.

The conventional diet approach to gestational diabetes mellitus (GDM) advocates carbohydrate restriction (35–40 % of total calories) [12–14], with carbohydrate intake distributed across meals and snacks in order to blunt postprandial glucose and mitigate glucose-mediated fetal macrosomia.

For women with type 1 diabetes, a lower proportion of carbohydrate during pregnancy such as 40 % of energy (moderate-low carbohydrate diet) has also been suggested to result in normoglycaemia and excellent pregnancy outcome [15].

Few studies have been done in pregnant women with type 2 diabetes mellitus with regards to which dietary treatment would give the best outcome for mother and fetus. A low carbohydrate diet with 35 % of the calories as carbohydrates was described to reduce the incidence of hyperglycemia in women with type 2 diabetes mellitus during pregnancy, with a reduced need for insulin treatment and a significant reduction in the incidence of macrosomia [16]. Nonetheless, the same principles of avoiding excessive weight gain and controlling postprandial glucose levels apply, as in gestational diabetes mellitus and type 1 diabetes mellitus, and again, the long term fetal and maternal outcomes of these interventions are not yet conclusively known.

It is worth noting that although carbohydrate restriction is often useful in the management of diabetes in pregnancy, it should not be done excessively. The IOM (Institute of Medicine) clinical guidelines recommend a minimum intake of 175 g carbohydrate/day to ensure sufficient supplementation of glucose to the mother and fetus including 33 g/day for the fetal brain and avoid the possibility of low carbohydrate diet induced ketogenesis that may be associated with a decrease in intelligence and fine motor skills in the offspring [17].

On the other hand, perhaps carbohydrate restriction is not the only component in the equation when working towards better glycemic control in women with diabetes in pregnancy. Carbohydrate restriction often results in higher fat intake, given that protein intake is remarkably constant at 15–20 %. Outside of pregnancy, a high-fat diet typically increases serum free fatty acids, promoting insulin resistance. In nonhuman primates and in some human studies, a maternal high-fat diet increases fetal fat accretion and infant adiposity, promotes hepatic steatosis, increases inflammation and oxidative stress, and impairs skeletal muscle glucose uptake. This might explain why in a recent randomized controlled trial of carbohydrate restriction to 40 % (with 40 % fat) versus 55 % (with 25 % fat) in women with gestational diabetes, there was no significance difference in the need for insulin treatment or pregnancy outcomes [18]. Perhaps, also, the choice of carbohydrate, in particular the glycemic index of carbohydrate, plays an equally important role in attaining good glycemic control.

Several studies have revealed that a diet that is high in carbohydrates of low glycaemic index improves overall glucose control and reduces postprandial glucose excursions in pregnant women with no diabetes [19], gestational diabetes mellitus [20], nonpregnant individuals with type 1 diabetes mellitus and type 2 diabetes mellitus [21, 22] and reduces the need for insulin in gestational diabetes [23]. However, evidence for obstetric and fetal advantages are still lacking. Currently, only three published randomized controlled trials have studied fetal and obstetric outcomes between pregnant women with gestational diabetes treated with a diet with low-moderate glycaemic index and controls on moderate-high glycemic index carbohydrates. These have shown no significant differences in rates of macrosomia, large for gestational age infants, caesarean section, operative vaginal birth and normal vaginal birth [23–25]. The ROLO study of 800 women without diabetes but with a previous infant weighing greater than 4 kg showed that a low glycaemic index diet in pregnancy did not reduce the incidence of large for gestational age infants, but did have a significant positive effect on gestational weight gain and maternal glucose tolerance [26]. At this moment, there are no studies or data with regards to long-term fetal and maternal health outcome data in women with gestational diabetes prescribed a pregnancy diet high in carbohydrates of low glycaemic index.

Carbohydrate counting is a very essential strategy for the improvement of glycemic control in non-pregnant patients with type 1 diabetes and is frequently used together with insulin pump treatment for pregnant as well as for non-pregnant patients. Although data on the effect in pregnancy are lacking, it makes clinical sense for the continuation of accurate carbohydrate counting with intensive reviews of these patient’s postprandial glucose levels as these will intuitively impact on macrosomia.

Regular insulin, the rapid-acting insulin analogues aspart and lispro, the long-acting insulin analogue levemir and human insulin are licensed for use in gestational diabetes mellitus and pre-existing diabetes mellitus in pregnancy. Metformin is not licensed for use in gestational diabetes mellitus and pre-existing diabetes mellitus in pregnancy but certain national institutes like the England and Wales National Institute for Health and Clinical Excellence included metformin and glyburide as GDM and type 2 diabetes in pregnancy treatment options, with the proviso that it is not licensed for these indications and that there should be informed consent on the use of metformin or glyburide if the managing diabetologist would like to use it due to strong evidence for its effectiveness and safety so far [8].

The lure of using metformin in pregnancy is its low cost, easy administration without need for much patient training (unlike insulin), low risk of hypoglycaemia, especially when used as monotherapy, and potential for prevention of excessive weight gain, a useful advantage when used in patients requiring insulin, or if used instead of insulin for glycaemic control. Its disadvantage would be its failure rate of about 40–50 % of patients (the proportion being dependent on the population studied), and the lack of long-term safety data.

The use of metformin in early pregnancy in several randomized controlled trials of pregnant women with polycystic ovary syndrome has revealed no increased risk of major congenital malformations and in fact showed reduction in adverse pregnancy outcomes, reduction in the incidence of gestational diabetes and higher pregnancy and live birth rates [27–30].

The non-inferiority in terms of effectiveness and safety of metformin compared to insulin in the treatment of gestational diabetes was also proven in the Metformin in Gestational Diabetes (MiG) trial [31]. In this randomized control trial of 751 women with gestational diabetes randomly allocated to open-label treatment with metformin (1,000–2,000 mg daily) or to insulin alone, there was no significant difference in the primary outcome (composite of neonatal complications including hypoglycaemia, respiratory distress, phototherapy, birth trauma, low APGAR and prematurity). In fact, severe neonatal hypoglycaemia occurred more commonly in the insulin group than in the metformin group. There were no differences in the number of congenital abnormalities between groups. There was also less maternal weight gain (metformin 0.4 vs. insulin 2.0 kg, p = 0.001) and more treatment satisfaction in favour of metformin. Seven more recent but smaller trials have also largely confirmed the safety and efficacy of metformin in gestational diabetes compared to insulin [32–35].

There also seems to be potential in the addition of metformin to insulin therapy in women with type 2 diabetes mellitus with pregnancy. A recent randomized controlled trial on 90 women with either gestational or pre-existing diabetes mellitus having poor glycaemic control at a daily dose of insulin of ≥1.12 units/kg (defined as having insulin resistance in this study) showed that adding metformin to insulin therapy in these women achieved good glycemic control in 75 % of without needing an increased insulin dose [36]. Addition of metformin was also associated with a reduction in hospital stay, maternal hypoglycemia, neonatal hypoglycemia, NICU admission and neonatal respiratory distress syndrome. We await the results of the Metformin in Women with Type 2 Diabetes in Pregnancy (MiTy) trial of 500 pregnant women with type 2 diabetes recruited from 25 centres in Canada who will be randomized to receive metformin or placebo in addition to their usual regimen of insulin. This study will clarify whether adding metformin to insulin in women with type 2 diabetes will be beneficial to the mothers and infants (Clinical Trials Registry No; NCT 01353391).