New prediction tool may help identify early birth risk in women with ICP
Model combines lab and clinical data to support care decisions
Written by |
Researchers in China have developed and validated a new prediction model that can help distinguish pregnant women with intrahepatic cholestasis of pregnancy (ICP) who are at higher risk of delivering their babies early from those who are not.
Using data from 370 women, the model incorporated several factors, including immune cell counts, bile acid levels, placental maturity, and medication use.
“This study established the first preterm birth prediction tool for intrahepatic cholestasis of pregnancy integrating multidimensional indicators” and “provides an objective basis for early clinical identification of high-risk patients,” the researchers wrote.
Researchers develop model to predict early birth risk in ICP
The study, “Establishment and validation of clinical prediction models for preterm birth in patients with intrahepatic cholestasis of pregnancy: Single-center retrospective observational study,” was published in Medicine.
ICP is a form of cholestasis, a condition marked by the slowing or stalling of bile flow out of the liver that develops during pregnancy. Bile is a digestive fluid produced in the liver that is transported to the intestines, where it helps break down fats and proteins.
The condition increases the risk of complications for both the mother and the baby. These include preterm birth, defined as delivery before 37 weeks of gestation, which is linked to a higher risk of short- and long-term complications compared with full-term birth.
“However, the lack of standardized tools to accurately predict preterm birth risks has limited early intervention for high-risk patients,” the researchers wrote.
With this in mind, researchers in China set out to develop and validate a model to predict preterm birth in women with ICP. They used data from 370 pregnant women with ICP who were hospitalized at Yichang Central People’s Hospital between 2017 and 2024.
Study uses patient data to build and validate prediction model
A total of 259 women comprised the training set used to develop the predictive model, and the remaining 111 women made up the validation set. In the validation group, about 80% of the women were younger than 35. In both groups, about 2% had a previous history of ICP, and nearly half worked during pregnancy.
The rate of preterm birth was higher in the training set than in the validation set (17.37% vs. 9.91%), but this difference did not reach statistical significance.
In the model developed from the training group, working during pregnancy, medication use, and placental maturity grades II and III were significantly associated with a lower risk of preterm birth. Placental maturity grades II and III are generally seen later in pregnancy, typically in the third trimester (weeks 28 to 40).
The placenta is an organ-like structure that connects to the unborn baby via the umbilical cord and helps deliver oxygen and nutrients to the fetus and remove waste products.
In contrast, higher blood levels of certain immune cells (lymphocytes and neutrophils) and bile acids, a main component of bile, were significantly associated with a higher risk of preterm birth.
Model shows good accuracy in identifying higher-risk pregnancies
Based on these risk factors, the team developed a model to predict preterm birth and calculated the area under the curve (AUC), a statistical measure of how well a test distinguishes between two groups (in this case, women with or without preterm birth). AUC values range from zero to one, with higher values indicating higher predictive accuracy.
The model’s AUC was 0.85 in the training group and 0.72 in the validation group, indicating that it was able to distinguish women at higher risk of preterm birth from those without. Further analysis showed that the predicted risk was in good agreement with the observed risk values, and decision-curve analysis suggested the model may have clinical usefulness.
To facilitate use of the model, researchers presented it as a visual prediction tool. Each factor corresponds to a distinct score, and “by summing all variables’ scores, we obtain a total score,” where higher values reflect a higher chance of preterm birth, the researchers wrote.
“Although the prediction model developed in this study demonstrated good performance,” at discriminating women at a higher risk of preterm birth, further studies “in diverse populations and clinical settings [are] still needed to evaluate its generalizability and robustness.”
