Medical Devices

STPA looks at how a device is controlled, and how unsafe interactions can arise with the patient, other equipment, the staff, and the clinical setting.

A Recall Where No Part Failed

A bedside blood analyzer reported a patient’s electrolyte level as dangerously low. The real level was normal. If staff gave electrolytes through an IV to raise the level, the patient could suffer arrhythmia, seizure, or death. Reports like this led to a global recall.

The sensor had not failed. It detects electrolyte ions in the blood sample, but a deposit of foreign material on its surface kept the ions from reaching it. The designers knew a sensor’s readings could drift in exactly this way, so they built in a calibration run after each sample to check the sensor. But they also designed the analyzer to send each result before the calibration run, so staff get it quickly. Each calibration run takes as long as the blood test, and the analyzer repeats a failed run before giving up. By the time the analyzer added a warning to the result it had originally reported, staff at the bedside may already have acted on that result.

The device’s FMECA, accepted by regulators, looked for parts that could fail, such as a degrading sensor membrane. It had no place for a working sensor covered by a deposit, or for the conflict of reporting fast versus finishing the calibration run first. An MIT analysis built on STAMP, the model behind STPA and CAST, found both. In six of the analyzer’s roughly 20 control loops it found 175 hazards, versus 70 in the FMECA, and it produced the requirement the design lacked: hold each result until the calibration run is done.

Read the MIT thesis
Control loop in the analyzer recall Clinical staff, who believe the level is low, treat the low result. The patient's electrolyte level is actually normal. A blood sample goes to the analyzer, where a deposit blocks the sensor; the analyzer sends the result before its calibration run, so the low result reaches staff with no warning yet. Clinical staff Believe the level is low Patient Electrolyte level actually normal Blood analyzer Deposit blocks sensor Sends result before calibration run Treat the low result Blood sample “Low” result, no warning yet
The control loop in this recall. The analyzer is the staff’s only feedback on the patient’s electrolyte level.

What Is STPA?

System-Theoretic Process Analysis (STPA) is a hazard analysis method developed at MIT. It finds unsafe interactions among a device, its software, the people using it, and the clinical setting, including interactions where every component works as designed.

STPA can identify hazards caused by:

  • Unsafe interactions between functions (with or without a failure)
  • Missing requirements or functions needed for safety
  • Intended functionality that inadvertently introduces new hazards
  • Control algorithms that may contain complex flaws (regardless of failures)
  • Human interactions, like mode confusion, caused by clumsy automation
  • Poor coordination between multiple controllers

Why It Matters for Medical Devices

  • Medical devices today are increasingly software-driven, connected, and interoperable. They function within complex socio-technical systems (e.g., networks of devices, clinicians, and environments that interact dynamically). Hazards can and do emerge from the dynamic interactions within these systems, beyond what traditional failure-based analyses typically reveal.
  • At the same time, failure-based techniques require significant effort and cost to complete. These methods can require extensive testing and documentation for each individual component, yet still miss systemic hazards that lead to costly recalls, regulatory penalties, or patient harm. For many years, organizations have allocated substantial resources of time and capital, often without fully mitigating essential risks.
  • In contrast, STPA more efficiently exposes critical system flaws by grounding them in real-world operational context.

At Massachusetts General Hospital, STPA of the Varian Ethos found 496 unsafe control actions, most from how staff and software interact rather than from equipment failure. On one IMRT process, it found hazards an earlier HFMEA study had missed.

Examples of Areas Targeted by STPA

Human Factors

Identifies potential for user errors, specific device behaviors that can cause or prevent user errors, solutions to prevent harm despite user errors, and ways to design interfaces or workflows that minimize confusion and improve safety.

Software & Control Logic

Reveals unsafe control actions in embedded or closed-loop control systems where timing, environmental factors, available feedback, and the way decisions are made matter.

Connected & Interoperable Devices

Identifies risks arising from inadequate coordination and misinterpretations among networked devices or hospital systems.

Operational & Postmarket Contexts

Extends beyond design to evaluate how the user context, clinical environment, maintenance, configuration changes, and updates can introduce new risks.

Fits Medical Device Standards and Submissions

ISO 14971

Hazards and Risk Controls

STPA identifies hazards and the scenarios that lead to them, for your hazard identification. The safety requirements it derives become risk control measures.

IEC 62304

Software Safety Requirements

Unsafe control actions by software controllers show how software can contribute to a hazardous situation, which shapes software safety requirements.

IEC 62366-1

Use-Related Hazards

STPA treats the user as a controller, so it finds use errors along with the device behavior and feedback that lead to them, for your hazard-related use scenarios.

Regulatory submissions

Traceability from Hazard to Requirement

Each requirement traces back through a scenario and an unsafe control action to a hazard, the traceability a submission needs.

SAE J3187, the published STPA recommended practice, applies to any industry, and a medical-device supplement, J3187-5, is in development. An STPA of cross-vendor interoperability fed Annex J of ANSI/AAMI/UL 2800-1.

Why Adoption Is Growing

Regulators and manufacturers are recognizing that system-level safety requires more than failure analysis. The U.S. FDA, medical device companies, hospital systems, and other industry leaders are increasingly applying STPA.

Why Regulators and Manufacturers Apply STPA

  • Support risk management by identifying hazards not visible through traditional failure-based analysis.
  • Reduce cost by identifying hazards and risks early, preventing costly redesigns, post-market fixes, and regulatory delays.

Used By

Medical device leaders that have used STPA to set the gold standard for system safety and device reliability include:

  • Medtronic
  • 8i Robotics

Who Uses STPA in Medical Devices

  • Published STPA work on insulin pumps, defibrillators, radiotherapy systems, surgical robots, and AI-enabled devices
  • Studies run side by side with FMEA, where STPA found hazards FMEA had missed
  • STPA in ISO 14971 risk management for a device headed for certification
  • The FDA’s own STPA work, and how one manufacturer rolled STPA out across its engineering organization

Each one links to its source.

See who uses STPA

The Bottom Line

STPA provides a structured, engineering-driven way to uncover hazards early, define explicit safety constraints, and design more resilient medical devices that take into account the full operational context.

By moving beyond failure-based thinking, STPA helps teams address the safety challenges of modern healthcare technology systematically, with a better understanding of how their devices fit into the wider clinical setting.

Where to Go Next

Keep Learning

The introductions to the methods, the evidence for using them, and the industry pages are all free.

STAMP Resources

Take a Class

Self-paced online courses and live classes, with certification for those who want it.

See the classes

Get Help on a Project

We work alongside teams applying STPA or CAST to their own systems.

About project support

Questions? Email [email protected].