Martha’s Rule: what are we detecting?

Martha’s Rule began rolling out across English acute NHS trusts in 2024, reaching all acute inpatient sites from August 2025. The accompanying article in this issue [1] makes the case for caution. It is a policy introduced across the country ahead of a prospective evaluation, with questions of mechanism and workforce still open. This is an argument well made, but given that Martha’s Rule is national policy, and therefore unlikely to be removed or significantly changed in the near future, we need to engage with it and, ideally, use it to the benefit of our perioperative patients.

For those of us caring for patients after surgery, the rationale for Martha’s Rule sits within the work on failure to rescue. This is when death occurs following a complication, which, if it had been recognised and treated sooner, the patient might have survived. Much of this burden falls in the days after a patient leaves recovery or intensive care for a general ward, where surveillance is thinner and deterioration can go unnoticed. Hospitals differ surprisingly little in how often complications occur. However, they differ a great deal in whether the patient survives [2]. Sheetz and colleagues found an up to 11-fold variation in failure-to-rescue rates for individual procedures, with structural features such as teaching status, hospital technology, nurse staffing and intensive care capacity explaining only part of it [3]. Most of the rest lies in how teams behave. A study comparing high and low failure-to-rescue hospitals found near-uniform agreement that the principal drivers were delayed recognition and an inability to voice concern, shaped by psychological safety and communication [4]. Martha’s Rule is aimed squarely at this, at making it easier and faster for a concern to be heard and acted upon.

We now have some early data. Between September 2024 and October 2025, trusts logged 8171 Martha’s Rule calls. Most, 64%, were not for acute deterioration at all. They were concerned with communication, coordination and other matters, and it is this group that lies behind the accompanying article’s legitimate concern about the load on critical care outreach. The remaining 2942 calls did relate to acute deterioration. In the most recent period broken down in the figures, June to October 2025, an early warning score had been recorded before 914 of 989 such calls. In 746 of these (82%) the patient’s vital signs alone would not have met the escalation threshold, yet in 515, just over two-thirds, the call led to a change in treatment or transfer to higher-level care [5]. Someone, often a patient or family member, had raised an alarm but the early warning score had not. In these patients, something is being detected, but what?

If treatment changed in over two-thirds of the activations in which the early warning score sat below the threshold, the score was not what prompted the change. It may be a signal that precedes measurable physiological change, the patient sensing that something is wrong before it registers on a chart. It may be subtler physiology than early warning scores capture: a shift in mental state, in colour, or in the way a person is moving or talking. Martha’s Rule is not the first system of its kind. For over a decade, other health systems have let patients and families call an urgent, independent review when they fear deterioration is being missed, among them Ryan’s Rule and REACH in Australia [6], and Condition Help in the United States [7]. Patient-reported wellness has been shown to track with later early warning scores [8], and in a large paediatric cohort, caregiver concern was independently associated with intensive care admission, more strongly than any single abnormal vital sign [9]. Whether the same holds on the adult surgical ward has not yet been established.

If structured patient-reported assessment adds something to physiological monitoring, a fifth vital sign of sorts, one question is where else this could be deployed. Growing numbers of patients are managed in virtual wards and step-down settings, where the aim should be not thinner monitoring but good monitoring paired with a structured account of how the patient feels and a route for families to raise concerns. This combination might give clinicians the confidence to manage and discharge more patients safely into those settings than monitoring alone would allow. Cancellation of planned surgery remains common: the national PACE2024 evaluation found almost one operation in 10 had been cancelled within 24 hours, at a real cost to patients, theatres and waiting lists [10]. Bed availability is only one driver, but we are unlikely in the coming years to see bed numbers grow as dramatically as the requirement for surgery. Safer step-down and virtual-ward care may offer an opportunity to increase surgical volumes without increasing the number of inpatient beds.

The programme statistics cannot yet tell us whether Martha’s Rule is reaching surgical patients at all, because these have not been broken down by admitting speciality. Reporting that distinguished surgical from medical activations would be a straightforward and valuable first step. These data could then be linked to our perioperative research infrastructure, including the National Emergency Laparotomy Audit and the Perioperative Quality Improvement Programme, letting us ask of the surgical cohort what the patient-reported signal is detecting and whether it adds discrimination to the monitoring we already have [11].

References