Gestational diabetes in Victoria in 1996: incidence, risk factors and outcomes

Considerable controversy surrounds the diagnosis,1 natural history, management and outcomes2 of women whose initial diagnosis of glucose intolerance occurs during pregnancy (ie, gestational diabetes mellitus [GDM]). Opinions regarding the impact on the infant range from only minor to highly significant.3 The true incidence of GDM is unknown because of the lack of universal testing, inconsistent diagnostic criteria and the absence of a uniform reporting system to a centralised register. Some of this uncertainty arises from the paucity of true population studies.3,4

In Australia, previous attempts to determine the true incidence have been based on surveys of prenatal clinics consisting of hospital patients and/or private patients,5-8 but these are subject to bias due to referral patterns, local population variations and local clinical practice.

Using information from routinely collected datasets and epidemiological methods, we aimed to estimate the incidence of GDM in all singleton pregnancies for Victoria in 1996 and compare the risk factors for GDM and the maternal and infant outcomes with those in non-diabetic pregnancies.

The estimated incidence of gestational diabetes for 1996 of 3.6% is very close to the average of 3.3% estimated for New South Wales for 1991 to 1994,14 despite the different methods used to determine the level of under-reporting

In 1991, and again in 1998, the Australian Diabetes in Pregnancy Society guidelines recommended that all women should be screened for GDM.15,16 Until there is universal screening, the percentage of pregnant women diagnosed as having GDM will underestimate the incidence. Previous reports have suggested that in the early 1990s only 50% or less of pregnant women were screened for GDM.14 However, the women who are tested are more likely to be in the higher-risk groups, so the incidence might be lower in the untested, undiagnosed group.

Previous estimates of GDM incidence worldwide vary from 1% to 10%. Previous Australian reports may have over-estimated the incidence.5-8 Our population-based study showed that the incidence was higher in women attending the larger hospitals or living in the metropolitan regions — sites where earlier studies have been conducted.

A limitation of this study is the lack of information on some risk factors for GDM, such as previous history of GDM, previous birth of a large-for-gestational-age baby, a family history of type 2 diabetes and the woman's body mass index before conception. However, we have been able to examine in a logistic regression many other potential risk factors at a population level.

In Victoria, about 25% of the women giving birth in any one year were not born in Australia. Almost all of the non-Australian-born groups have a higher frequency of GDM. These groups also tend to be older when they give birth,17 which may further contribute to the higher frequency. Generally, the incidence of GDM in different racial groups reflects their propensity for developing type 2 diabetes.7

The link between gestational diabetes and poor maternal and fetal outcomes has been recognised for a long time.3,4,14,15,18,19 The increased risk of hypertension and pre-eclampsia was a significant finding and is consistent with previous reports.4,20-23 Women with GDM have been described as fitting into the "metabolic syndrome" because of their tendency to increased body mass index, higher serum triglyceride and free fatty acid levels and insulin resistance,24 and this may partially explain the link with hypertension and pre-eclampsia.

Women with diagnosed GDM had a much higher risk of induced delivery (37.3% v 22.7%). This has been reported previously overseas20,23 and locally,25 and is ascribed to the perceived increased risk of macrosomia, shoulder dystocia and delivery complications in these women.

The increased odds ratio for caesarean section is supported by numerous studies,3,4,20-23,26,27 and occurred after adjustment for the presence of macrosomia. It has been noted for some time that the link between GDM and interventional outcomes may become a self-fulfilling prophecy, where the decision to intervene may be influenced by the diagnosis of GDM itself rather than other indications.28 We cannot be sure if this was happening in Victoria in 1996 without linking birthweight data and glucose levels with the caesarean outcome. Some local centres have not found an increased rate of caesarean section in GDM pregnancies.29

Operative delivery and episiotomy were also more likely in GDM pregnancies — even after adjustment for the higher prevalence of macrosomia in this group.

For the infant, hyaline membrane disease and neonatal jaundice were more common in GDM pregnancies, and this is consistent with overseas experience.30 Macrosomia, whether defined as ≥ 4000 g birthweight or birthweight ≥ 90th centile, is one of the major infant morbidities associated with GDM. Some studies have reported normalised rates of macrosomia and other major neonatal and maternal outcomes with intensive antenatal treatment.31-33 These are predominantly reports from large tertiary centres, where common unit policies and rigorous guidelines may have helped to achieve the outcomes. Our data and other large studies3,4 suggest that, on a population level, women identified as having GDM still have an increased rate of macrosomia.

Recent reports have found increased perinatal mortality associated with GDM.4,23,30 We did not find raised perinatal mortality, but our study did not have the power to detect a difference of the order previously reported.

A strength of this study is the large numbers involved and the fact that it examines an unselected population. Retrospective studies do rely on the accuracy of the recorded data, however, and there is a potential for recorder bias: women who develop complications may be more likely to be identified as having GDM. Their infants are also more likely to be tested for hypoglycaemia and jaundice. In addition, in this analysis, the diagnoses of hypertension, pre-eclampsia, neonatal jaundice and hyaline membrane disease are as recorded in the datasets and are not subject to a single standard definition or validation of entry. Finally, while many centres in Victoria used criteria similar to the Australian Diabetes in Pregnancy Society criteria for diagnosing GDM, they were not adopted by all,34 so our population of women have not all necessarily been subject to the same diagnostic criteria.

There are several different criteria worldwide and the diagnosis of GDM has been subject to much controversy,2 as there is no clear level of glucose above which the maternal and fetal outcomes become distinctly worse. Instead, there seems to be a continuously increasing risk as maternal glucose levels rise.35

The important question that remains to be resolved in the management of GDM is: at what level of hyperglycaemia do the adverse perinatal and maternal outcomes justify treatment and interventions? This cannot be answered on the basis of retrospective observational data, but the ongoing HAPO (Hyperglycaemia and Adverse Pregnancy Outcomes) study, taking place in Australia and around the world, will go some way to address this question.

In our study, women identified as having gestational diabetes had increased rates of hypertension, pre-eclampsia, induced labour, and interventional delivery. Their offspring had a higher risk of macrosomia, neonatal jaundice and hyaline membrane disease. Thus, it does seem important to identify all women with GDM and to continue to monitor their outcomes.

1: Capture–recapture method of estimating gestational diabetes