Risk-based computational models can predict spontaneous preterm childbirth

By James DuckerReviewed by Emily Henderson, B.Sc.Jul 5 2021

A newly developed predictive model could help predict the risk of impending preterm birth, avoiding the adverse and costly effects of overtreatment or misdiagnosis.

Premature Baby. Image Credit: M.Moira/Shutterstock.com

Improving the prediction of impending preterm childbirth

Preterm labor is a challenging incident to diagnose due to the non-specific nature of symptoms. Current methods of diagnosis are inadequate as overtreatment of symptoms is common and costly to patients. However, the health impacts incurred if preterm labor is not treated effectively also poses a concern for the successful delivery of the child as well as long-term effects for the mother.

A new study published by Sarah Stock at the University of Edinburgh, United Kingdom, and colleagues in the journal PLOS Medicine, provides a new way of addressing the challenge of diagnosing preterm labor symptoms. This was accomplished by developing a risk prediction computational model that improves the prediction of impending preterm births.

First, the researchers identified the clinical risk factors for preterm labor by analyzing individual participant data from five European prospective cohort studies, including 1,783 pregnant European women, and used these to develop a model to predict the risk of spontaneous preterm birth.

Then, the model was validated in a prospective cohort study of 2,924 women with signs and symptoms of preterm labor from 26 consultant-led obstetric units in the United Kingdom. This allowed researchers to validate the accuracy of the model by comparing predicted and observed outcomes between datasets.

Findings showed that the best predictive model included key symptom combinations. Superficially, including vaginal fluid fetal fibronectin concentration analysis alongside clinical risk factors outperformed predictions of impending spontaneous preterm birth and was more cost-effective in comparison to fetal fibronectin alone.

The risk prediction model showed promising performance in the prediction of spontaneous preterm birth within seven days of testing and can be used as part of a decision support tool to help guide management decisions for women at risk of preterm labor. It is readily implementable, with potential for immediate benefit to women and babies and health services, through avoidance of unnecessary admission and treatment"

Predicting the unpredictable – model shows potential to limit adverse health effects during spontaneous preterm birth

The implementation of such predictive models could help identify the most important factors leading up to the clinical diagnosis of adverse incidents, such as spontaneous preterm childbirth.

Nonetheless, the study noted limitations in data collection and subsequent analysis that may limit the representative implications of this study.

The data collected was focused primarily on Caucasian European and British patients, with few non-White participants providing limited insight into ethnic or regional differences. Including other populations may therefore help improve model accuracy. Moreover, missing data in the risk predictor development cohort also requires further study, as the risk prediction model improves clinical outcomes in practice.

Developing more accurate and applicable predictive models could provide key guidelines for clinical treatments to be conducted successfully, avoiding potential misdiagnosis or overtreatment, that may lead to harmful health consequences.

The vast majority of women with signs and symptoms of preterm labor don't give birth early, but many receive unnecessary hospital admission just in case of preterm birth. The risk predictor developed by our research team will help women to understand their chance of giving birth early, so they can decide whether or not to have admission and treatment.

We are now working towards linking the predictor to maternity records, so it can easily be used as part of women's care and be continually improved as more women use it."

Dr. Stock

Extending predictive models and the use of computational data-based risk predictions beyond preterm childbirth may also provide support for other clinical treatments. This may include the diagnosis of neurodegenerative and cardiovascular diseases, which can display a range of symptoms, and often require specific treatments to be addressed successfully.