Multivariate Logistic regression analysis of factors affecting the occurrence of pregnancy complications in older pregnant women and the construction and efficacy verification of the prediction model

Abstract

To explore the influencing factors of pregnancy complications in older parturients and the construction and efficacy verification of prediction models. Methods: A total of 141 older parturients from June 2023 to January 2025 were selected as the subjects and divided into an occurrence group and a non-occurrence group according to whether complications occurred during pregnancy. The types of pregnancy complications of the patients were counted, and the medical records of the two groups were reviewed. The possible influencing factors of pregnancy complications in older parturients were analyzed by univariate and multivariate analysis. R software was used to construct a prediction model for the occurrence of pregnancy complications in older parturients, and the prediction efficacy was verified. Results: A total of 42 older parturients had pregnancy complications (accounting for 29.79%). The top three were anemia, hypertensive disorders complicating pregnancy, and gestational diabetes. The results of univariate and multivariate Logistic regression analysis showed that age, number of prenatal examinations, whether there was excessive nutrient intake during pregnancy, bad living habits, psychological pressure, and mood swings were risk factors for complications during pregnancy in older mothers (P<0.05). A prediction model for complications during pregnancy in older mothers was successfully constructed, and the slope of the calibration curve was high (close to 1), and the curve area was 0.817 (95%CI was 0.759-0.871). The calibration and standard curves had a good fit and good consistency. Conclusion: The incidence of complications during pregnancy in older mothers is high, and there are many influencing factors. Based on the multivariate results, a prediction model for complications during pregnancy in older mothers was successfully constructed, which has high prediction sensitivity, specificity, discrimination, and calibration.