Editorials

Screening for gestational diabetes mellitus

A simple test may make it easier to study whether screening is worthwhile

Papers p   812

Gestational diabetes mellitus is a concept that arouses considerable controversy. It is defined as "carbohydrate intolerance of varying degrees of severity with onset or first recognition during pregnancy."¹ Rather than predicting the development of diabetes later in life, as proposed originally,² the main purpose of identifying gestational diabetes is to detect women at risk of adverse perinatal outcomes, such as macrosomia, neonatal metabolic abnormalities, birth trauma, and caesarean section. ^{1 3 4} Evidence of the effectiveness of universal screening for gestational diabetes on these outcomes is still lacking.⁵ However, recent randomised studies indicate that women who are intensively managed can achieve near normal rates of macrosomia and neonatal hypoglycaemia.^5-7

Those who do not favour screening for gestational diabetes claim, among other things, that the current screening and diagnostic strategies are cumbersome. In this issue of the BMJ Perucchini et al propose a protocol which could counter this argument: they suggest using a fasting glucose value as a screen for gestational diabetes (p 812).⁸ This protocol differs from the two currently recommended procedures. The first, mostly used in North America, is a two step scheme: a screening test consisting of a one hour 50 g glucose challenge test at 24-28 weeks of pregnancy followed, if positive, by a diagnostic three hour 100 g or two hour 75 g oral glucose tolerance test. ^{1 3} Recent guidelines do not recommend the screening test in women under 25 years, with normal weight, with no personal or family history of diabetes, with no history of poor obstetric outcomes, and who do not belong to an ethnic group predisposed to diabetes. ^{1 3} The second strategy, a one step procedure using a two hour 75 g tolerance test as proposed by the World Health Organisation,⁹ is mostly used in Europe.¹

Perucchini et al performed a one hour 50 g glucose challenge test followed, whatever the result, by a tolerance test.⁸ The challenge test result and the tolerance test fasting glucose value were analysed for their ability to predict gestational diabetes, which was diagnosed on a three hour 100 g glucose tolerance test using Carpenter and Coustan criteria. The authors calculated the sensitivity and specificity of the two tests and determined the thresholds with the best sensitivity-specificity association by the receiver operating characteristic (ROC) curves. For the challenge test this cut off was determined to be 7.0 mmol/l, with a sensitivity of 68% and a specificity of 82%. For the fasting glucose value the best threshold was 4.8 mmol/l (sensitivity 81%, specificity of 76%). Sensitivity is the probability of a positive test result if gestational diabetes is present and specificity the probability of screening negative if it is absent. A high sensitivity decreases the number of women with gestational diabetes who are missed by the screening test. As specificity increases the number of women without gestational diabetes who are incorrectly classified as positive decreases.

The results of Perucchini et al imply that if a fasting glucose threshold of 4.8 mmol/l is used as a screening test 70% of women do not need a diagnostic tolerance test and 19% of cases of gestational diabetes are undetected (1.9% of their population). For a challenge test threshold of 7.0 mmol/l 77% of women do not require a fasting tolerance test and 32% of cases are missed (3.3% of their population). Should a two step strategy be used then the fasting glucose value is preferable to the challenge test as the slight increase in the number of diagnostic tolerance tests needed overcomes the high number of undetected cases of gestational diabetes. In a one step procedure, performing a diagnostic tolerance test only in women with a fasting glucose value higher or equal to 4.8 mmol/l appears to consume fewer resources, human and financial, than submitting all subjects to a tolerance test. However, 19% of the women with gestational diabetes would be missed compared with none with the tolerance test. We do not know the clinical impact of not detecting these cases. Pre-existing but undiagnosed diabetes is unlikely to be missed with a fasting glucose value of 4.8 mmol/l cut off. Anyhow, data on pregnancy outcomes in the undetected cases of gestational diabetes are needed.

A fasting glucose value offers many advantages: it is easy to administer, well tolerated, inexpensive, reliable, and reproducible.¹⁰ However, more studies are required before endorsing the fasting glucose value as the screening test for gestational diabetes. Its validity has to be established with the World Health Organisation and Sacks criteria. Its has to be compared with the 50 g selective screening strategy. The glucose fasting value has to be validated in different populations, especially those with a lower prevalence of gestational diabetes. The threshold of 4.8 mmol/l may need to be revised if screening is done in an office or surgery setting with glucose meters. Meters are generally accurate, but their precision varies. They may not be subject to the same quality control as laboratory assays.¹¹

In conclusion, screening for gestational diabetes mellitus with a fasting glucose value is an attractive strategy. What we need now is an assessment of its effectiveness in decreasing the adverse perinatal outcomes associated with gestational diabetes as part of an intervention programme.

Evelyne Rey, head of obstetric medical unit. 

Department of Obstetrics and Gynaecology, Sainte-Justine Hospital, Montreal, Quebec, Canada H3T 1C5