Predictive value of cervical length for preterm birth in women with antepartum haemorrhage of unknown origin: a retrospective cohort study

When a pregnant woman arrives at a hospital with vaginal bleeding in the second or third trimester and no obvious cause can be found, clinicians face one of obstetrics’ most unsettling diagnostic puzzles. A new study from Hong Kong now suggests that a routine ultrasound measurement of the cervix, taken at the moment of that unexplained bleed, may quietly reveal which of these women are on course to deliver their babies dangerously early. The research, published in BMC Medical Imaging, analyzed 321 women admitted with antepartum haemorrhage of unknown origin between 24 and 32 weeks of gestation at a tertiary hospital from 2019 to 2025, and found that a shortened cervix on transvaginal ultrasound was one of the strongest predictors of preterm birth ever reported in this population.

Antepartum haemorrhage of unknown origin, abbreviated APHUO in the medical literature, refers to vaginal bleeding after 24 weeks of pregnancy that persists after the usual culprits, such as placenta previa or placental abruption, have been ruled out by ultrasound and clinical examination. It is a common reason for antenatal admission, yet its management remains largely supportive because doctors have had no reliable way to distinguish women whose bleeding is an isolated nuisance from those whose bleeding is the first visible sign of a cervix beginning to fail. Preterm birth, defined as delivery before 37 completed weeks, remains the leading cause of death and disability in newborns worldwide, and every additional week in utero translates into measurable gains in survival and neurodevelopment for the infant.

The research team, led by Lorraine Chan of the Department of Obstetrics and Gynaecology at Queen Elizabeth Hospital in Hong Kong, took advantage of a measurement that many of these women were already receiving: transvaginal ultrasonography of the cervix. Unlike abdominal ultrasound, the transvaginal approach places the probe close to the cervix, allowing precise visualization of the closed cervical canal from the internal os at the top to the external os at the bottom. Cervical length in this study was measured as the distance between these two landmarks, and a cervix was classified as short when it measured less than 25 millimeters, the threshold long used in guidelines from bodies such as the National Institute for Health and Care Excellence in the United Kingdom.

Among the 321 women studied, 62, or 19.3 percent, had a short cervix at presentation, while the remaining 259, or 80.7 percent, had a cervix of normal length. The differences in outcomes between the two groups were stark. Women with a short cervix delivered at a median of 33 weeks of gestation, compared with 38 weeks in women with a normal cervix, and the interval from admission to delivery was dramatically shorter, at a median of 7 days versus 67 days. The rate of preterm birth before 37 weeks was significantly higher in the short cervix group, with a p value below 0.001, indicating that the association was extremely unlikely to have arisen by chance alone.

Because women with unexplained bleeding differ from one another in age, body mass index, smoking status, prior obstetric history, and other characteristics that independently influence the risk of early delivery, the researchers used multivariable logistic regression to isolate the effect of cervical length. After adjusting for these confounders, a short cervix remained powerfully associated with preterm birth, with an adjusted odds ratio of 25.16 for delivery before 37 weeks, 41.67 for delivery before 34 weeks, and 60.39 for delivery before 32 weeks, all with 95 percent confidence intervals well above 1. In practical terms, women with a cervix shorter than 25 millimeters at the time of their bleed faced odds of delivering before 32 weeks that were roughly sixty times higher than those of women whose cervix remained normal, a magnitude of association rarely encountered in clinical prediction research.

The study also examined spontaneous preterm birth, meaning delivery that begins on its own through labor or rupture of membranes rather than being initiated by doctors for maternal or fetal indications. Cervical length and a maternal history of preterm birth in a previous pregnancy were both independently associated with spontaneous preterm birth, suggesting that in women with unexplained bleeding, the same biological processes that drive spontaneous early labor, including cervical ripening and inflammation, may be at work. The researchers correlated cervical length with maternal and fetal outcomes as secondary endpoints, and the admission-to-delivery interval served as a proxy for how quickly the bleed heralded the arrival of delivery.

Beyond the binary short-versus-normal classification, the team used receiver operating characteristic analysis, a statistical technique that plots the trade-off between sensitivity and specificity across every possible cut-off value, to determine the optimal cervical length threshold for predicting preterm birth at different gestational ages. The area under the curve, a measure of overall discriminative ability ranging from 0.5, equivalent to a coin flip, to 1.0, equivalent to perfect prediction, indicated good predictive performance for cervical length. The optimal cut-off for predicting delivery before 37 weeks was 24 millimeters, a threshold that achieved 95 percent specificity, meaning that when the cervix measured below this value, the finding almost always signaled true elevated risk rather than a false alarm.

Perhaps the most clinically consequential conclusion of the study is that a single universal threshold may not serve all women equally. The authors argue that different cervical length cut-offs should be applied according to a woman’s background risk of preterm birth, taking into account factors such as previous preterm delivery, multiple gestation, and other known risk factors. In women whose baseline risk is already elevated, a less stringent cut-off may suffice to trigger interventions such as corticosteroids for fetal lung maturity, magnesium sulfate for fetal neuroprotection, or transfer to a center with a neonatal intensive care unit. In women whose baseline risk is low, a more conservative threshold may prevent unnecessary anxiety and overtreatment.

The retrospective design of the study, in which investigators analyzed records of women already admitted rather than prospectively enrolling them, carries inherent limitations, including the possibility of unmeasured confounders and the reliance on the accuracy of documentation. The waiver of individual patient consent was granted by the Central Institutional Review Board of Hospital Authority in Hong Kong precisely because of the retrospective nature of the research, and the study was conducted in accordance with the principles of the Helsinki Declaration as revised in 2024. The single-center setting at a tertiary hospital in Hong Kong may also limit generalizability to populations with different demographic profiles and healthcare systems.

Nevertheless, the findings carry immediate implications for the management of one of the most common and frustrating presentations in obstetrics. Transvaginal cervical length, a measurement that takes minutes to perform and involves no radiation, no contrast agents, and no invasive procedures, may transform unexplained antepartum bleeding from a diagnostic dead end into an actionable risk stratification opportunity. For the roughly one in five women with unexplained bleeding whose cervix measures below 25 millimeters, the measurement may justify intensified surveillance, timely administration of antenatal corticosteroids, and delivery planning at a facility equipped to care for premature infants. For the majority whose cervix remains normal, the finding may offer reassurance that the bleed, however alarming, is less likely to presage an early delivery. As the study authors conclude, cervical length below 25 millimeters is a useful predictor of preterm birth in women with unexplained antepartum haemorrhage, and the next step is to translate that prediction into interventions that keep babies in utero longer and healthier.

Subject of Research: Predictive value of cervical length for preterm birth in women with antepartum haemorrhage of unknown origin: a retrospective cohort study

Article Title: Predictive value of cervical length for preterm birth in women with antepartum haemorrhage of unknown origin: a retrospective cohort study

Article References: Chan, L., Yu, F. N. Y., & Ma, T. W. L. (2026). Predictive value of cervical length for preterm birth in women with antepartum haemorrhage of unknown origin: a retrospective cohort study. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02731-3

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02731-3

Keywords: Predictive, value, cervical, length, preterm, birth, women, antepartum, haemorrhage, unknown, origin, retrospective