A prediction model of superimposed preeclampsia in women with chronic hypertension

Yongjun Lu^1,2Li Yang²Xiaoyan Li¹Dan Kuai¹Tian Wenyan^1*Huiying Zhang^1*

• ¹Tianjin Medical University General Hospital, Tianjin, China
• ²Tongzhou Maternal＆Child Health Hospital of Beijing, Beijing, China

Objective:The purpose of this study is to establish and validate a prediction model for CH with superimposed PE. Methods: A retrospective case-control study was conducted on 471 CH patients admitted to the Tongzhou Maternal and Child Health Care Hospital of Beijing from January 2020 to December 2023. The patients were divided into superimposed preeclampsia (SPE) group (161 cases) and non-preeclampsia (NPE) group (310 cases) based on whether they had complicated PE. The patients were randomly divided into training set and validation set in a 7:3 ratio. General and clinical data were collected, and single-and multi-factor logistic regression analysis were used to screen for independent factors affecting PE in the training set. Based on the screening results, the diagnostic efficacy of PE was evaluated using the receiver operating characteristic curve. Risk prediction nomogram model was constructed using R language. The Bootstrap method was used to validate and produce calibration plots; DCA was used to assess the clinical benefit rate of the model. Results: The results of single factor and further multi-factor analysis showed that peripheral blood levels of DBIL, GGT, GLU, CysC and pre-pregnancy body mass index (BMI) were independent influences on the occurrence of PE (P < 0.05). The area under the curve of the combined of DBIL, GGT, GLU, CysC and pre-pregnancy BMI was 0.789, with a sensitivity of 0.593 and a specificity of 0.830, which is better than a single clinical diagnostic indicator. The results of multifactor analysis were constructed as a nomogram model, and the mean absolute error of the calibration curve of the modeling set was 0.035, suggesting that the predictive probability of the model was generally compatible with the actual value. DCA showed the predictive model has clinical utility value. Conclusions: The occurrence of PE in women with CH is related to the peripheral blood levels of DBIL, GGT, GLU, CysC and pre-pregnancy BMI, and the combination of these indexes has a better clinical diagnostic value than a single index. The nomogram model constructed by using the above indicators can be used for the prediction of PE and has high predictive efficacy.