P02.07: New logistic regression model compared to the type of miscarriage alone for the prediction of successful expectant management of miscarriage
Ishwari Casikar, Chuan Lü, J. Riemke, Dalya Alhamdan, S. Reid, Tommaso Bignardi, Max Mongelli, George Stanley Condous · Ultrasound in Obstetrics and Gynecology · 2010
We have developed a new logistic regression model for the prediction of successful expectant management of first trimester miscarriage. We aimed to compare its performance to the type of miscarriage alone for the prediction of successful expectant management. Prospective observational study from November 2006 to December 2009. The new logistic regression model, which was developed on 186 women with successful expectant management of miscarriage, comprised of diagnosis of incomplete miscarriage at primary scan, vaginal bleeding (no/yes without clots/yes with clots) and maternal age. The performance of this model was compared to the type of miscarriage alone i.e. (incomplete miscarriage, empty sac miscarriage or missed miscarriage). The performances of the new logistic regression model and the type of miscarriage were evaluated using receiver operating characteristic (ROC) curves. Complete data from another 126 women with miscarriage were included in the test set. The new logistic regression model gave an area under the ROC curve of 0.803 with an accuracy of 82.5%. Compared to the type of miscarriage alone, gave an area under the ROC curve of 0.725 with an accuracy of 82.5%. The difference between the performance of these models was significant (P-value = 0.0443). We have developed a new mathematical model to predict successful management of first trimester miscarriage. This model outperforms the type of miscarriage alone for the prediction of successful expectant management.