Predictive models for gestational diabetes ranked
medwireNews: Dutch researchers have assessed the ability of 12 published models to predict gestational diabetes in a large cohort of women.
The models contained between three and eight predictors – 15 in total – and were designed to be applied in the first trimester.
“Prognostic models have the advantage of being cheap and easy to implement and could avoid the need to perform an oral glucose tolerance test in women with a low risk of developing gestational diabetes mellitus, which relieves both burden and costs”, say researcher Marije Lamain-de Ruiter (University Medical Centre Utrecht) and team.
The four top ranked models all included body mass index and ethnicity, three included maternal age, two included family history of diabetes and two history of gestational diabetes, and one each included blood pressure, parity and history of macrosomia.
When recalibrated to a cohort of 3723 pregnant Dutch women, these models were between 74% and 78% accurate for discriminating between those who did and did not develop gestational diabetes. The original versions were between 70% and 81% accurate and they were between 72% and 76% accurate in the 1655 women who were nulliparous.
“We recommend that these four models be further investigated for implementation in clinical practice”, writes the team in The BMJ.
The two prognostic models with the best overall performance based on discrimination, calibration and accuracy in nulliparous women were those developed by Teede et al in 2011 and Van Leeuwen et al in 2010.
“Predictors in these particular prognostic models (that is, maternal age, body mass index, ethnicity, parity, history of gestation diabetes mellitus, and history of macrosomia) are easy to measure and widely applicable”, say the researchers.
But they add: “The decision on which of the four best models to implement in clinical practice might also depend on population characteristics, availability of predictors, and the incidence of gestational diabetes mellitus, and could therefore be country or region specific.”
All 12 models had reasonably good predictive ability, however, with the poorest performer, when recalibrated, discriminating between women with and without gestational diabetes with an accuracy of 67% in the whole cohort and 69% in nulliparous women. In general, models containing the fewest predictors had the worst accuracy, with the poorest performer having just three predictors.
By Eleanor McDermid
medwireNews is an independent medical news service provided by Springer Healthcare Limited. © Springer Healthcare Ltd; 2016