Classification of diabetes mellitus

The two most common types of diabetes are known as type 1 and type 2 diabetes mellitus (Expert 2000).

Normal pregnancy and insulin resistance

Normal pregnancy is associated with insulin resistance similar to that found in type 2 diabetes. Physiological resistance to insulin action during pregnancy becomes apparent in the second trimester, and insulin sensitivity declines progressively to term. These changes facilitate transport of glucose across the placenta to ensure normal fetal growth and development. Transfer of glucose across the placenta stimulates fetal insulin secretion, and insulin acts as an essential growth hormone. However, if resistance to maternal insulin action becomes too pronounced, fetal hyperinsulinaemia, accelerated growth and possible organ damage may occur. In normal pregnancy after delivery, maternal pregnancy‐associated insulin resistance returns rapidly to normal, and glucose tolerance usually returns to normal within six weeks of delivery.

Gestational diabetes mellitus

Gestational diabetes mellitus (GDM) is defined as "carbohydrate intolerance resulting in hyperglycaemia of variable severity with onset or first recognition during pregnancy" (Expert 2000; WHO 1999). GDM is defined in this way for women with undiagnosed pre‐existing diabetes and for those in whom the first onset of hyperglycaemia is identified during pregnancy. To identify early in pregnancy women with previously undiagnosed diabetes, the International Association of Diabetes and Pregnancy Study Groups (IADPSG) (Metzger 2010) recommends assessment of high‐risk populations using glycated haemoglobin, fasting or random plasma glucose at first pregnancy booking. Other groups including the American Diabetes Association (ADA) (ADA 2017) and the UK National Institute for Health and Care Excellence (NICE 2015) also recommend early testing of high‐risk groups to identify women with pre‐existing diabetes.

In the past GDM has been categorised by severity of glucose impairment, with less severe cases referred to as impaired glucose tolerance, and more severe cases as GDM. Hyperglycaemia is now viewed as a continuum, with levels along the complete distribution of glycaemia in pregnant women associated with adverse effects (HAPO Study Cooperative Research Group 2008). Because of this, the term 'impaired glucose tolerance' is no longer used, and all abnormalities of glucose intolerance in pregnancy are now referred to as 'GDM'. GDM necessarily includes women who first develop diabetes during pregnancy, as well as those with previously undiagnosed type 1 or 2 diabetes mellitus (frank or pre‐existing diabetes) recognised for the first time in pregnancy, because it is impossible at the time of first recognition to distinguish clearly between these different diagnoses. GDM complicates between 1% and 24% of pregnancies (Farrar 2015a; Farrar 2016; NIH 2013) and varies with the population characteristics and the diagnostic glucose thresholds used. It is the most common metabolic disorder of pregnancy and is associated with increased maternal and perinatal morbidity (Bitó 2005; Crowther 2005; Kim 2002; Landon 2009; Schytte 2004). Reducing modifiable risk factors for developing GDM by eating a healthy balanced diet, achieving and maintaining a healthy normal weight and exercising before and throughout pregnancy may reduce the risk of development of GDM or its severity, but the evidence is limited (Ceysens 2006; Tieu 2017). Maternal hyperglycaemia leads to transfer of excessive glucose to the fetus, resulting in fetal hyperinsulinaemia. Fetal hyperinsulinaemia causes increased risk of fetal adiposity, macrosomia (increased birthweight) and perinatal complications (Jovanovic 2001; Kjos 1999). Specifically, GDM is associated with overgrowth of insulin‐sensitive tissue such as adipose tissue, especially around the chest, shoulders and abdomen, which increases the risk of shoulder dystocia, perinatal death and birth trauma and makes the need for caesarean section more likely (Jovanovic 2001; Kjos 1999). Risk of hypoxaemia is increased in utero, and this can lead to increased risk of intra‐uterine death, fetal polycythaemia (too many red blood cells), hyperbilirubinaemia (jaundice) and renal vein thrombosis (blood clot in the kidney vein). Evidence also suggests increased risk of neonatal metabolic complications, including hypoglycaemia, in babies of mothers experiencing GDM (Jovanovic 2001; Kjos 1999; Kostalova 2001; Pribylová 1996; Silverman 1995). Observational data suggest that babies of women with GDM are at increased risk of obesity (Dabelea 2000; Gillman 2003; Pettitt 1983) and hypertension (Akisü 1999) later in life. However, data from trials of treatment of women with mild to moderate GDM suggest similar adiposity in infants, regardless of their mother's group allocation (Gillman 2010; Landon 2015), data are few and length of follow up limited.

Oral glucose tolerance test and diagnostic criteria

Until 2013 the World Health Organization recommended that diagnosis of GDM should be based on the same criteria used for non‐pregnant adults (WHO 1999). The WHO recommended that with the 75 g OGTT, GDM (including impaired glucose tolerance) was diagnosed if the fasting plasma glucose level was greater than 7.0 mmol/L, or if the two‐hour level was greater than 7.8 mmol/L. The UK National Institute for Health and Care Excellence (NICE) recommendations (NICE 2015) are based on these criteria; however, although still recommending the 75 g OGTT, NICE recommend that GDM should be diagnosed when either fasting plasma glucose level is equal to or greater than 5.6 mmol/L, or when the two‐hour postload level is equal to or greater than 7.8 mmol/L.

The National Diabetes Data Group (NDDG) recommendations for GDM diagnosis were first published by the American Diabetes Association in 1997 (NDDG 1997); these criteria are based on initial screening with an oral glucose challenge test (OGCT). This test did not require women to fast; they were given a 50 g glucose drink and had blood glucose checked after one hour. If the one‐hour value was equal to or greater than 7.8 mmol/L, a three‐hour 100 g OGTT followed. Subsequently, the NDDG revised its recommendations, suggesting a one‐step approach for high‐risk women, using a 100 g three‐hour OGTT (Expert 2000). For low‐risk women, the NDDG continued to recommend a two‐step approach with an initial 50 g glucose challenge test, followed by a 100 g three‐hour OGTT for women who screen positive with the one‐hour OGCT threshold of 7.8 mmol/L. Blood glucose levels for diagnosis of GDM on the three‐hour test are equal to or greater than 5.3 mmol/L for fasting blood sugar, equal to or greater than 10.0 mmol/L at one hour, equal to or greater than 8.6 mmol/L at two hours and equal to or greater than 7.8 mmol/L at three hours (Expert 2000).

As discussed, the IADPSG has published guidance recommending the 75 g OGTT with lowered thresholds (compared with WHO 1999 criteria) in the light of HAPO study findings (Metzger 2010). Lowering the diagnostic threshold will result in the diagnosis of GDM in more women who have less severe hyperglycaemia. Evidence indicates that treatment of women with mild to moderate GDM reduces the risk of adverse perinatal outcomes; however the beneficial effects of treatment based on diagnosis determined using one criterion may not be the same as those of treatment based on diagnosis of GDM using different criteria.

Current recommendations on the criteria for diagnosis are presented in Table 1, Table 2, Table 3 and NICE 2015. It is clear that the choice of test and diagnostic criteria varies both within and between countries (Kanguru 2014).

Other tests for gestational diabetes

The main disadvantages of the OGTT are that it is costly and unpleasant for women. Simpler, cheaper and more acceptable tests for initial screening and even diagnosis would be advantageous, provided they were accurate in identifying women with GDM or its complications or both. Random plasma glucose testing has been popular in the past, as it is convenient to perform. However, it fails to identify a large proportion of women who have GDM (low sensitivity) (McElduff 1994). Fasting plasma glucose has greater sensitivity and specificity than random glucose tests, but it relies on women fasting overnight, and it does not perform as well as the OGTT in identifying those with GDM (Perucchini 1999) because worsening insulin resistance in women with GDM may be compensated for by increased insulin production, which means that fasting glucose can be normal, with the early abnormality becoming apparent only after meals.

Glycated haemoglobin, which reflects glucose levels over the past four weeks, does not require an overnight fast and needs only a single blood sample; it is increasingly recommended as a potentially valuable tool for diagnosing diabetes in non‐pregnant people (Fonseca 2009). Use of glycated haemoglobin to diagnose GDM has not been well studied. In countries where fasting or random glucose or glucose challenge tests are not used to universally screen for GDM, the presence of glycosuria has been used as a screening method. Assessment of glycosuria is relatively cheap, non‐invasive and acceptable to pregnant women (Hanna 2008). However, guidance from NICE 2008 and ADA 2017 no longer recommend its use as a screening method, instead suggesting that risk assessment (e.g. based on ethnicity, family history, obesity) should be used as the primary screening method in the absence of a universal blood test for glycaemia. Evidence indicates that glycosuria has low sensitivity for identifying hyperglycaemia/GDM determined by glucose tolerance testing (Scott 2002). However, its sensitivity has not been directly compared with that of risk factors used as the preliminary screening tool in the UK, and limited evidence suggests an association between glycosuria and macrosomia and risk of obesity for future children (Scott 2002).

When to test for gestational diabetes

Views on the most appropriate time to test women for GDM are varied. Timing is important because consideration of the burden on pregnant women and on healthcare service resources means that ideally the test should be performed only once during pregnancy. At the same time, earlier diagnosis and treatment in any woman with GDM may prevent adverse perinatal outcomes. Administering the test during the first trimester results in identification of women with previously undiagnosed type 1 or type 2 diabetes, as well as those with early‐onset newly occurring diabetes. Women at particularly high risk, such as those who had GDM in a previous pregnancy, are often offered an OGTT in the first trimester. However, as most physiological pregnancy‐related changes in insulin sensitivity (sometimes referred to as a glucose challenge test of pregnancy) begin during the second trimester, testing in the first trimester may miss a large proportion of women who will develop GDM. For this reason it has been suggested that high‐risk women who have a normal test in the first trimester will usually require a second test later in pregnancy.

To avoid the need for repeated testing, most women are offered a single test in the late second trimester. This timing serves as a balance between testing too early (thus missing some women who do not develop gestational diabetes until after the test) and testing too late (leaving little time to intervene to reduce plasma glucose and improve outcomes).

Universal versus selective testing

The question of whether to offer universal or selective screening remains controversial and is the topic of another Cochrane Review (Screening for gestational diabetes mellitus to improve maternal and infant health) (Tieu 2014). In the past, screening has been targeted at those with increased risk of developing the condition, for example, women who have had GDM in a previous pregnancy, those with a family history, women of Asian or Carribean ethnicity and those who are obese; however some women without risk factors may develop GDM and may remain untreated. If all women are offered an OGTT, more women with hyperglycaemia will be identified and more women may benefit from treatment (Farrar 2014). There are however no robust trial data to suggest which approach (universal or selective) is best for improving maternal and infant outcomes, even though treatment trials (Crowther 2005; Landon 2009) and reviews (Brown 2017; Farrar 2016) have demonstrated an improvement in several important perinatal outcomes when hyperglycaemia is reduced (a Cochrane Protocol also details a review that will provide an overview of Cochrane Reviews of interventions (Martis 2016)).