Analysis of prognostic factors and construction of prediction model for pregnancies with liver failure

Abstract

Background and objectives: Liver failure during pregnancy adversely affectsmothers. However, it is not thoroughly found which its prognostic factors are. In the present study, we explored some factors which could affect the short-term prognosis. Furthermore, a logistic regression model (LRM)was constructed to predict the outcomes of mothers. Method: One hundred and twenty-nine pregnant women with liver failure were enrolled in this study. The mothers were treated at several hospitals in Guangzhou from January 2008 to September 2022. Ninety-six patients were divided into an effectual group (n=76) and an ineffectual group (n=20). Retrospective and logistic regression analyses were performed to screen for possible prognostic factors and to construct LRM. The remaining 33 cases, combined with the original 96 cases, were used to validate the model. Results: Age, |Na-135|, and INR are independent risk factors for liver failure. The area under the curve (AUC) for LRM and MELD are 0.896 and 0.780, respectively. Thesensitivity of the two models was 95.83% and 70.83%, respectively. The specificity was 71.43% and 75.24%, respectively. The total prediction accuracy rate was 75.97% and 74.42%, respectively. Conclusion: Age, |Na-135|, and INR were independent risk factors for pregnant women with liver failure with poor prognosis, both the LRM and the MELD could predict the prognosis, however, the LRM was superior to the MELD in terms of sensitivity.