Derivation and external validation of prediction model for hypertensive disorders of pregnancy in twin pregnancies: a retrospective cohort study in southeastern China.

OBJECTIVE

We aimed to develop and validate an effective prediction model for hypertensive disorder of pregnancy (HDP) in twin pregnancies after 28 weeks of gestation.

DESIGN

Retrospective cohort study.

SETTING

Maternity hospital.

PARTICIPANTS

We recruited twin pregnancies who delivered in Fujian Maternity and Child Health Hospital from January 2014 to December 2019 as a training cohort. Besides, we included twin pregnancies delivered at Fujian Maternity and Child Health Hospital; Women and Children's Hospital of Xiamen University from January 2020 to December 2021 as temporal validation set and geographical validation set, respectively.

MAIN OUTCOME MEASURES

We performed univariate analysis, the least absolute shrinkage and selection operator regression and Boruta algorithm to screen variables. Then, we used multivariate logistic regression to construct a nomogram that predicted the risk of HDP in twin pregnancies. We employed the bootstrap resampling method for internal validation, used the receiver operating characteristic (ROC) curve to evaluate the predictive performance of the model and constructed decision curve analysis to assess the clinical benefit of the model. Thereafter validated the nomogram through the index of concordance (C-index) and calibration curves in the temporal validation set and geographical validation set.

RESULTS

Multivariate logistic regression showed that primipara (OR=1.284, 95% CI=1.016 to 1.622), the higher pre-pregnancy body mass index (OR=1.077, 95% CI=1.039 to 1.116), the higher uric acid (OR=1.004, 95% CI=1.002 to 1.005), the higher urea nitrogen (OR=1.198, 95% CI=1.087 to 1.321), the higher creatinine (OR=1.011, 95% CI=1.002 to 1.020), the higher lactate dehydrogenase (OR=1.001, 95% CI=1.000 to 1.002), the higher ratio of large platelets (OR=1.034, 95% CI=1.020 to 1.048), the lower albumin (OR=0.887, 95% CI=0.852 to 0.924), the lower calcium (OR=0.148, 95% CI=0.058 to 0.375) are influencing factors of HDP in twin pregnancies. The area under the ROC curve of the prediction model was 0.763. The C-index were 0.842 and 0.746, respectively, on the temporal validation set and geographical validation set.

CONCLUSIONS

The new model for predicting HDP in twin pregnancies constructed by clinical characteristics and laboratory indicators had high clinical application value. It can be used to individually evaluate the occurrence of HDP in twin pregnancies after 28 weeks of gestation.