Development and validation of a predictive model for adverse pregnancy outcomes of non-standard antiphospholipid antibody-positive
Objective Development and validation of a predictive model for the risk of adverse pregnancy outcomes in non-canonical antiphospholipid antibody (NC-aPL)-positive pregnant women. Methods 4000 NC-aPL-positive pregnant women attending the centres from May 2021 to June 2022 were selected as the study subjects,and were divided into a modelling group ( 2800 cases) and a validation group (1200 cases) using the random number table method;the modelling group was divided into the adverse pregnancy group and the no-adverse-pregnancy group according to the presence or absence of an adverse pregnancy outcome after the pregnancy. A prediction model was established and a column chart was drawn using the R software through Logistic regression. The differentiation and calibration of the predictive models were verified and evaluated by using receiver operating characteristics (ROC) curves,calibration curves,and the Hosmer-Lemeshow goodness-of-fit test.Results In this study,831 cases of adverse pregnancy outcomes were seen in the modelling group with an incidence rate of 29.68% and 357 cases were seen in the validation group with an incidence rate of 29.75%. Multifactorial Logistic regression analysis showed that a history of smoking,poor lifestyle,antiphospholipid antibody-anti-beta2 glycoprotein I antibody (aPL-αβ2GPI) positivity,and unregulated use of medication were the risk factors for adverse pregnancy outcomes in NC-aPL-positive pregnant women (P<0.05). The validation results showed that the area under the ROC curve (AUC) of the modelling and validation groups were 0.853 and 0.936,respectively,with good differentiation;the H-L test showed good agreement (x2=7.208,7.163,P=0.503,0.510).Conclusion A history of smoking,poor lifestyle,aPL-αβ2GPI positivity,and unregulated medication use are risk factors for adverse pregnancy outcomes in NC-aPL-positive pregnant women,and the column-line graphical model constructed in this way has good differentiation and consistency,and is able to intuitively predict the risk of adverse pregnancy outcomes in NC-aPL-positive pregnant women.