Mind the gap: principles and practices in the use of interactive medical devices

Interactive medical devices, including those for monitoring vital signs and treatment administration, are essential in modern health care. Such instruments have transformed clinical practice and the patient experience. For example, programmable infusion devices (see Figure 1) can deliver drugs and fluids more accurately and reliably than their predecessors, leading to improved treatment outcomes—particularly for people in intensive care, infants and those undergoing chemotherapy.[1] These devices have become smaller and lighter, with improved battery life, which renders them increasingly mobile, improving patients’ sense of freedom. Furthermore, as these devices continue to evolve, the delivery of care at home, partly through patients and lay carers having a role in self-management and partly through telehealth, is gaining prominence. Telehealth is the delivery of health care (including monitoring of vital signs as well as treatment) through the use of communications technology, enabling care to be delivered remotely. This approach requires patients to take some responsibility for any interactive devices that are used in monitoring or treatment.

Figure 1.

Figure 1. A programmable infusion device.

We have been conducting studies on the use of interactive medical devices by both clinicians and patients. To gain an understanding of actual practices, and the reasoning behind those practices, we use qualitative research methods, including in-depth observations and interviews focusing on device use, in both hospitals and homes. Data has been analysed using thematic analysis,[2] with a focus on patient safety and user experience.

Patient safety depends both on the reliability of such devices (do they perform as designed?) and on them being used within safe bounds (are they used as intended?). Premarket approval processes—particularly those defined by the FDA,[3] which are currently the most stringent—require that manufacturers define a product’s ‘intended use’ in detail. This definition includes both what it is designed to be used for, in what environments and the procedures that should be followed to use it correctly, which requires manufacturers to conduct tests to check both the reliability and the usability of devices. For example, if an infusion device is designed for use in air ambulances, its performance characteristics will need to be tested across a range of temperatures and on test rigs that reproduce the forces and vibrations that are likely to be experienced in a moving helicopter. Usability testing typically involves recruiting nurses as test users who will be required to set up and manage a set of infusions in a simulation laboratory (see Figure 2) that is configured to mimic a typical context of use, such as a hospital ward or intensive care bay, with a dummy patient, to check that they can use the devices as intended.

Figure 2.

Figure 2. Simulation hospital ward.

In practice, real use diverges from intended use for many reasons, such as poor understanding of intended use, achieving other goals that conflict with intended use or troubleshooting when something unexpected happens. Grissinger[4] reported a case of a neonatal baby being fed through a nasogastric tube, using a syringe driver to regulate the flow of milk. This setup was not a use for which the device had been designed: it was designed to deliver medication. Indeed, the baby was also receiving medication delivered via an identical pump. At some point, “a nurse mistakenly connected a syringe containing breast milk to the wrong line”—resulting in the baby receiving milk intravenously.[4] Within the CHI+MED (Computer-Human Interaction for Medical Devices) programme,[5] a project to improve the safety of interactive programmable medical devices, one of our lines of research is how interactive medical devices are actually used, and why. The long-term aim of the programme is to improve the safety of interactive medical devices, which includes achieving greater convergence between intended and actual use.

Convergence has two aspects: bringing actual use closer to intended use and bringing intended use closer to actual use. Actual use can be brought closer to intended use through improved design (so that it is easier to do the right thing), improved clinical processes, and improved training. Conversely, intended use can be brought closer to actual use by better understanding of that use, as well as the underlying causal factors that influence it, and designing devices to support those uses and address those underlying factors.

Our studies have highlighted ambiguities around the nature of ‘error’. Furniss et al.[6] studied the use of infusion pumps in an oncology day care unit. They identified many ‘unremarkable errors’—the little incidents that take place many times every day, that are quickly recovered from, but that increase staff workload and draw their attention away from patient care and from engaging with patients as people. For example, a nurse might start to programme an infusion pump then realise that the battery charge is low, so has to go and find a fully charged pump and start the procedure again. On other occasions, the nurse might not notice that the charge is low, and only discover it when the pump alarms, at which point it becomes necessary to find and set up a replacement pump to deliver a volume of fluid that is now not accurately known. These problems could be mitigated by having more power outlets readily available, or by improving the display of information about remaining battery life on the infusion device. Each such difficulty is minor, and has become normalized in health care, but they accumulate to affect performance and outcomes. Importantly, many can be addressed through changes in design or policy once they are recognised.

As the example of an infusion pump being used for feeding illustrates, professionals sometimes use systems in ways that were not intended by their manufacturers. When this happens, it is important to understand the causes and consequences of such ‘workarounds’ to redesign systems to reduce negative—and exploit positive—effects. O’Connor[7] studied how clinicians used a blood gas analyser (BGA) in two different settings. In one, he noted several occasions in which a false patient ID was entered; in the other, no such incidents were observed. Although the study was small, analysis of the observational and interview data highlighted factors that contributed to these behaviours. Most of the false patient IDs related to patients who had just arrived at the hospital as emergency cases: they needed treatment (informed by BGA results) before they had been formally admitted to the hospital and allocated an ID. The BGA device did not provide an obvious method to specify that this was an emergency patient to provide a record that could subsequently be linked to the correct patient. Although this workaround was motivated out of concern for patient care, the quality of the care record was subsequently compromised; this highlights the need to formally support this behaviour.

Not all difficulties result in errors or workarounds. We observed the use of infusion devices in an intensive care unit,[8] and found functions that were used frequently that could only be accessed by navigating deeply into the menu hierarchy. For example, we noted: “Every hour, to record the hourly intake of a drug with the volumetric pump, the nurse needs to access a Status menu from the Main Menu of the pump interface, choose an Intermediate Parameters option, read the volume infused, and then reset the counter to zero. The operation takes 8 key presses.” Such a mismatch between localized data monitoring and recording practice and the design (or configuration) of the infusion device highlights the need to review the practice or propose design changes.

We are exploring ways to inform design[9] and training[10] to close the gap between actual and intended use of medical devices. The focus is on interactive medical devices that are used by people (typically nurses and patients) for whom device use is only a small part of their daily activity. As well as studying infusion device use in a range of clinical settings (oncology, haematology and intensive care), to better understand the variabilities in practices across clinical settings, we have ongoing and planned studies on the use of blood glucose meters (in both hospitals and for home use by people with diabetes) and of home haemodialysis systems (used by people with renal disease). As more care moves from the hospital to the home, designing medical devices that are safe, reliable and usable, and that deliver the best possible user experience across a range of settings, is becoming increasingly important. The first step to achieve this goal is building a deep understanding of how they are used and why.

Ann Blandford


  1. N. Sims, M. E. Kinnealey, R. Hampton, G. Fishman and H. DeMonaco, Drug infusion pumps in anesthesia, critical care, and pain management, Churchill Livingstone, 2010.
  2. V. Braun and V. Clarke, Using thematic analysis in psychology, Qual. Res. Psychol. 3, pp. 77–101, 2006.
  3. FDA Device Advice.
  4. D. Furniss, A. Blandford, and A. Mayer, Preventing accidental infusion of breast milk in neonates, Pharm. Therapeutics, p. 127, 2010.
  5. CHI+MED: Computer-Human Interaction for Medical Devices.
  6. D. Furniss, A. Blandford, and A. Mayer, Unremarkable errors: low-level disturbances in infusion pump use, Proc. 25th BCS Conf. Hum. Comp. Interact., pp. 197–204, 2006.
  7. L. O'Connor, Workarounds in Accident and Emergency and Intensive Therapy Departments: Resilience, Creation and Consequences, MSc thesis.
  8. A. Rajkomar and A. Blandford, Understanding infusion administration in the ICU through distributed cognition, J. Biomed. Informatics 45} pages = {580-590, 2012.
  9. C. Vincentand and A. Blandford, Integration of human factors and ergonomics during medical device design and development: it's all about communication, Appl. Ergonom..
  10. I. Iacovides, A. L. Cox, and A. Blandford, Supporting learning within the workplace: device training in healthcare, Proc. 31st Eur. Conf. Cognitive Ergonomics, p. 127, 2010.

Cookies on IBME website

Our website uses cookies to provide us with important information about visitors. By continuing to browse the site we'll assume that you are happy to receive all cookies set by IBME.