Developing a risk stratification tool for postpartum haemorrhage

Brisbane’s maternity system brings together tertiary public hospitals, private obstetric services, midwifery clinics and general practices, creating both a rich testing environment and a patchwork of workflows. A collaborative model supported by Brisbane Diamantina Health Partners could align these groups around a shared tool while keeping decisions close to the woman and the maternity team.

Postpartum haemorrhage is common enough to affect many Australian families and serious enough to require rapid, coordinated treatment. The proposed risk stratification tool would estimate baseline risk from antenatal and birth factors, then support repeated reassessment as bleeding evolves. Its purpose is not to label a birth as safe, but to help teams anticipate escalation, blood loss, theatre review and specialist support.

Why risk stratification matters

A useful tool would organise risk into clear tiers linked to predefined responses, such as enhanced observation, senior obstetric review, anaesthetic involvement, blood-product preparation and access to operative and critical care. Queensland Health’s maternity and neonatal guidance on postpartum haemorrhage should provide the clinical baseline, with the tool acting as decision support rather than a substitute for judgement. Prediction must not obscure underlying causes such as uterine atony, retained placental tissue or birth trauma, because each requires a different clinical response.

National and state guidance should also agree on case definitions, estimated versus measured blood loss and time-to-treatment measures. Australia’s low maternal mortality overall can obscure the burden of severe haemorrhage, recurrent blood loss and psychological trauma. Risk stratification helps services identify those harms before they become emergencies while avoiding unnecessary intervention for people whose risk remains low.

Designing variables and clinical workflows

Potential predictors include previous postpartum haemorrhage, caesarean birth, placenta accreta spectrum, placenta previa, uterine fibroids or overdistension, multiple pregnancy, induction or augmentation, operative delivery, birth trauma and coagulation problems. Some factors are known before labour; others emerge during the third and fourth stages. The model should therefore combine an admission risk assessment with dynamic triggers based on bleeding, vital signs, ultrasound and response to initial care.

A transparent points-based matrix may suit early implementation better than a complex prediction model. Candidate variables must be available at the point of care and tested for missingness, bias and calibration across public and private hospitals, urban and regional services, and people using continuity-of-care or home-birth pathways. Aboriginal and Torres Strait Islander data governance and community involvement are essential; race should never be used as a biological shortcut that conceals unequal access to care.

The tool should prompt review of evidence-based treatment, not automatically prescribe uterotonics, tranexamic acid or transfusion. A Norwegian account of prescription medicine substitution illustrates why prescribing data need context: substitution may sometimes be appropriate, but treating every change as misuse can distort decisions. Any Australian medicines component should follow current obstetric guidance, local drug protocols and time-sensitive indications such as tranexamic acid within three hours of birth.

Validating the tool and governing data

Development should begin with retrospective data from several Queensland sites, followed by prospective silent validation before clinical use. Teams should compare discrimination, calibration, sensitivity for major haemorrhage and net clinical benefit, while also examining whether the score changes decisions or merely adds paperwork. A simple model that performs as well as a complex algorithm is usually easier to govern and explain.

Once piloted, the evaluation protocol, outcomes, analysis plan and handling of missing data should be registered before recruitment. The lessons from clinical trial transparency are directly relevant: selective reporting and unclear site performance can make an unsafe tool appear dependable. Full reporting should include protocol deviations, subgroup performance and negative findings, with a clear route for suspending use if safety signals emerge.

Identifiable information remains sensitive. Queensland’s Health Records Act 2001, the federal Privacy Act 1988 and applicable privacy principles should shape collection, linkage, access and retention, while public-sector accountability sits under the Hospital and Health Boards Act 2011 (Qld). A steering group should include an ethics specialist, data custodian, midwives, obstetricians, anaesthetists, blood-bank staff, Aboriginal health representatives and consumers, with documented approval for every model update.

Implementing risk tiers in Queensland

A staged pilot could begin in a large Brisbane public maternity unit, then test transferability through public hospitals in regional centres such as Townsville and Cairns, smaller rural services and participating private hospitals. Queensland’s geography means retrieval times, severe-weather disruption and limited on-site blood stocks can vary substantially. Planned home births and midwifery continuity models also need explicit escalation and transfer criteria rather than a score designed only for conventional hospital labour.

In practice, the score should appear in the electronic medical record where possible, with a concise paper version available during outages. Trigger packages should connect risk tiers to tasks: accurate blood-loss measurement, vital signs, intravenous access and pathology, appropriate medicines, blood-bank notification, theatre preparation and critical-care transfer. Regular simulation should rehearse the complete pathway across shifts and disciplines, not just the screen prompt.

Procurement should consider both usability and environmental impact. Digital prompts can reduce printed checklists, while a plain-language explanation of recycling after collection can help purchasing teams ask suppliers about packaging destinations and disposal pathways. Clinical waste must remain separate from council recycling, and cost savings should never be allowed to compromise infection control or emergency access to equipment.

Funding collaboration and knowledge sharing

Building and validating the tool requires stable funding for data engineering, statisticians, clinical time, training and independent evaluation. Although most haemorrhage is common, placenta accreta spectrum and some recurrent or complex cases are less frequent. Lessons from collaborative funding models can support shared protocols, multicentre recruitment and consistent reporting when the commercial and clinical value of individual services is limited.

Queensland could convene a grant consortium across health services, universities, research institutes and community groups, with clear governance for intellectual property, data ownership, publication rights and patient benefit. Outputs should include an open implementation manual, peer-reviewed evaluation, education modules and a living dashboard of outcomes. Those assets would allow research findings to travel from one hospital to another without depending on a single local champion.

Acting now for safer maternity care

Success should be judged by outcomes and experience: time from recognition to treatment and escalation, severe haemorrhage, operative intervention, intensive-care admission, blood-product use, readmission, psychological support and equity between population groups. The tool should also be retired if it creates delay, encourages unnecessary intervention or performs unevenly. Public reporting can then distinguish a genuinely safer pathway from one that merely records more activity.

Clinical leaders, service managers, researchers, consumers and Aboriginal health practitioners can act now by joining a cross-disciplinary working group. The first shared actions are to agree the risk variables and outcome set, nominate pilot units, secure seed funding, consult ethics and governance partners, and publish the evaluation protocol before clinical deployment. A transparent, collaborative pathway can then progress from local testing to Queensland-wide practice, with every revision checked for safety, usefulness and fairness.