Plotting the ROC curves for various adverse perinatal outcomes will give the threshold for hyperglycemia
The research article says-
Considering all eligible evidence, it is clear that the association between glucose concentration and a wide range of clinically relevant adverse perinatal outcomes is linear and that there is no threshold above which odds substantially increase in high income countries...........Treatment seems to reduce the risk of these outcomes, but the optimal glucose threshold to define gestational diabetes is unknown......... Data were extracted on induction of labour; caesarean and instrumental delivery; pregnancy induced hypertension; pre-eclampsia; macrosomia; large for gestational age; preterm birth; birth injury; and neonatal hypoglycaemia.
Here we want to point out that the way linear association is determined between the glucose concentration and adverse outcome, in the similar way an ROC (a receiver operating characteristic) curve can be plotted, which is a graphical plot that illustrates the performance of a binary classifier system as its discrimination threshold is varied. We can take various thresholds , and
the curve is created by plotting the true positive rate (TPR) or sensitivity against the false positive rate (FPR) or specificity at various threshold settings and is a trade off between sensitivity and specificity. The left corner of curve is the optimum sensitivity and specificity (e.g. it is100 mmol/L) then it can be considered as threshold or cut-off. The gold standard for calculation of TPR and FPR, the gold standard would be various perinatal outcomes (presence versus absent) for which data were extracted (as mentioned in the above paragraph) and the least cut-off or threshold among the various outcomes will be considered as the threshold of hyperglycemia which is dangerous.
Rapid Response:
Plotting the ROC curves for various adverse perinatal outcomes will give the threshold for hyperglycemia
The research article says-
Considering all eligible evidence, it is clear that the association between glucose concentration and a wide range of clinically relevant adverse perinatal outcomes is linear and that there is no threshold above which odds substantially increase in high income countries...........Treatment seems to reduce the risk of these outcomes, but the optimal glucose threshold to define gestational diabetes is unknown......... Data were extracted on induction of labour; caesarean and instrumental delivery; pregnancy induced hypertension; pre-eclampsia; macrosomia; large for gestational age; preterm birth; birth injury; and neonatal hypoglycaemia.
Here we want to point out that the way linear association is determined between the glucose concentration and adverse outcome, in the similar way an ROC (a receiver operating characteristic) curve can be plotted, which is a graphical plot that illustrates the performance of a binary classifier system as its discrimination threshold is varied. We can take various thresholds , and
the curve is created by plotting the true positive rate (TPR) or sensitivity against the false positive rate (FPR) or specificity at various threshold settings and is a trade off between sensitivity and specificity. The left corner of curve is the optimum sensitivity and specificity (e.g. it is100 mmol/L) then it can be considered as threshold or cut-off. The gold standard for calculation of TPR and FPR, the gold standard would be various perinatal outcomes (presence versus absent) for which data were extracted (as mentioned in the above paragraph) and the least cut-off or threshold among the various outcomes will be considered as the threshold of hyperglycemia which is dangerous.
Competing interests: No competing interests