Intended for healthcare professionals

Clinical Review

Should we screen for gestational diabetes? “The concept of gestational diabetes was popularised before considerations of evidence based medicine came on the scene”

BMJ 1997; 315 doi: (Published 20 September 1997) Cite this as: BMJ 1997;315:736
  1. R J Jarrett, emeritus professor of clinical epidemiologya
  1. a 45 Bishopsthorpe Road, London SE26 4PA
  • Accepted 4 March 1997


Much confusion surrounds the topic of screening for glucose intolerance-hyperglycaemia during pregnancy in terms of who should be screened, how to screen, and the management of those with positive results.2 3 4 Confusion arises from lack of or poor quality evidence, compounded in this instance by a concept (gestational diabetes mellitus) founded on risk of subsequent non-insulin dependent diabetes mellitus rather than outcome of the index pregnancy.5 In addition the criteria for gestational diabetes prescribe a minimum, but not a maximum, level of glucose intolerance, so that any group of women labelled as having gestational diabetes might contain some with glycaemia in the range that qualifies for a diagnosis of non-insulin dependent diabetes, rendering comparisons of different series impossible. Coustan, whose comment is reproduced in my title, suggested four questions which required answers to achieve resolution1:

  1. How severe must maternal hyperglycaemia be to measurably worsen pregnancy outcome?

  2. Can we intervene to prevent adverse outcomes?

  3. Is such intervention cost effective?

  4. If so, what is the most appropriate way of screening and detecting the problem?

Severity of maternal hyperglycaemia

Women with pre-existing diabetes, either insulin dependent or non-insulin dependent, undoubtedly have an increased risk of bearing a child with a congenital abnormality and this risk is related to the degree of hyperglycaemia during embryogenesis.6 7 However, this is not relevant to screening at typical booking times as embryogenesis is complete by week seven of gestation. By contrast, gestational diabetes as defined5 is not associated with risk of congenital abnormalities,8 despite the presence of some women with glucose intolerance sufficient to qualify them for a diagnosis of non-insulin dependent diabetes.

The only commonly (though not absolutely consistently) reported “complication” of gestational diabetes is macrosomia, a rather emotive description of a newborn infant with a birth weight in the upper centiles (variously defined) of the distribution. To what extent birth weight is determined by maternal glycaemia is debated, but the relation is confounded by maternal fatness.9 A very large baby is more likely to give rise to obstetric problems and to acquire a birth injury, but one estimate suggested that about 4% of women with untreated gestational diabetes would deliver infants weighing 4500 g or more compared with about 2% of the general obstetric population.10 While there is some evidence that treatment can reduce fetal weight,11 this cannot be automatically assumed to be justified given the data showing an inverse association between birth weight and the incidence of disorders in later life.12

Are there any adverse effects of the diagnosis of gestational diabetes? Women with gestational diabetes are more likely to be delivered by caesarean section. This has been attributed to their higher proportion of large babies, but in a recent study the section rate was higher even though the proportion of large babies was not,1 supporting the view that the diagnostic label sensitises obstetricians. The gestational diabetes label also leads to the necessity of self monitoring of blood glucose and possibly insulin injections. The possible distress due to screening and treatment in someone who previously thought herself to be healthy has not been investigated.


There is only one clinical trial of any merit.11 In this 66 women, mostly Hispanic and including an appreciable (though unstated) number with undiagnosed non-insulin dependent diabetes, were treated at random with either a more or a less intensive insulin regimen. Birth weights were, on average, nearly 400 g less in the intensively treated group, but caesarean section rates were not significantly different. Whether treatment influences any outcome of pregnancy in women discovered to have non-insulin dependent diabetes during pregnancy has not otherwise been subject to clinical trial. Indeed, pregnancy associated with non-insulin dependent diabetes has attracted little research interest.

Screening for hyperglycaemia is bedevilled by the lack of a suitable screening test.13 If sensitivity is important then some kind of glucose tolerance test is essential to identify gestational impaired glucose tolerance. Oral glucose tolerance tests are, however, tedious to perform and poorly reproducible. Single blood tests, such as glycated haemoglobin and fructosamine, cannot even identify the lower range of non-insulin dependent diabetes glycaemia, let alone gestational impaired glucose tolerance, though they could identify more florid hyperglycaemia.

The only relevant data available on cost effectiveness concern the yield of screening for non-insulin dependent diabetes using the World Health Organisation's “epidemiological” criterion—a plasma glucose value over 11.0 mmol/l two hours after a 75 g oral glucose load. Two studies provide minimal estimates of incidence of 4/10 000 for Europid women14 and 18/10 000 for south Asian women.15 As stated earlier, there is no good evidence of any undoubted benefit for the index pregnancy, though there are putative (but uncosted) benefits in having one's diabetes diagnosed early.13


The ethics of screening require the screener to show the likelihood of benefit from screening. No clear benefit has been shown from screening for glucose intolerance-hyperglycaemia (at least for the woman being screened) during pregnancy, and there are disadvantages, which include the acquisition of disease status and an increased risk of caesarean section. It is argued that screening to identify someone at risk of subsequent non-insulin dependent diabetes or with undiagnosed non-insulin dependent diabetes is a good thing. If so, it should be available to all adults, not only pregnant women. However, the most recent review of population screening for non-insulin dependent diabetes13 advocates extensive and varied further research on all aspects of the question. This was in the context of non-pregnant adults, but the same requirements apply to screening in pregnancy before it can be regarded as justified.

Since submitting this article I have noted two American groups which do not recommend screening for gestational diabetes. The American College of Obstetricians and Gynecologists, which in 1986 recommended selective screening, in 1994 noted the absence of data to support screening and did not make a specific recommendation.16 The US Preventive Services Task Force cites insufficient evidence for or against screening for gestational diabetes.17 In contrast, an expert committee of the American Diabetes Association continued to recommend screening, though no longer without some degree of selection.18


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
View Abstract