ROC curves to determine optimal points
The ROC curves for the overall and low and high risk cohorts are shown in Figures 1-3 and their respective area under the curve (AUC) and diagnostic accuracy is summarised in Table 3. The overall cohort had an AUC of 62.3% with optimal centile 11.3, with a sensitivity of 36.8% and specificity of 84.1% for detection of stillbirth. For the low-risk cohort, the AUC was 61.3% with optimal point 15.2, sensitivity 35.1% and specificity 84.6%, while for the high risk cohort the AUC was 61.2%, optimal centile point 2.7, sensitivity 34.2% and specificity 88.9% (Table 3). With false positive rate fixed at 10% i.e. specificity 90%, the sensitivity was - overall cohort: 30.2% (cut-off point 6.6), low risk: 29.1% (10.1) and high risk 33.0% (2.3).