Applying a robust approach to digital clinical safety in diagnosis

Summary

Diagnostics and digital go hand in hand. Digital healthcare has brought so many advancements in diagnostics and we are at the point of another paradigm shift with the advancements in artificial intelligence (AI), with some early and convincing diagnostic-use cases. New things also bring (new) risks. Some we can predict and plan for, perhaps some we haven’t, and some we can’t yet. In this blog, Ben Jeeves, Associate Chief Clinical Information Officer and Clinical Safety Officer, looks at the digital clinical safety aspects in relation to diagnostic safety.

This blog has been published as part of a series for World Patient Safety Day 2024 and the theme of Improving diagnosis for patient safety. #WPSD24, World Patient Safety Day 2024, WPSD 2024.

Content

Humans can make mistakes. We always have and we always will. In principle, computers are more consistent and reliable (cue reliability debate with NHS computers!) than humans. They certainly hold the potential for the productivity gains that are so drastically needed in healthcare. No breaks. No annual leave. No sick leave. But computers and systems break too, they have both planned downtime (annual leave) and unplanned downtime (sick leave). The impact of that can be far bigger, especially when these systems run at great scale. We have been reminded about the impact of system interruption with the Synnovis cyber attack.[1] There are multiple other examples.

What is digital clinical safety?

So how can we assess and, importantly, mitigate the digital risk that if it occurs can impact patient safety? Well, in short there are two standards that effectively relate to different stages of the health IT lifecycle; the DCB0129 for the 'manufacturer' of a product and DCB0160 for the 'deploying, maintaining and decommissioning' organisation.[2],[3] According to the standards (NHS DCB0160), the purpose is to “promote and ensure that effective clinical risk management is carried out by those health organisations that are responsible for deploying, using, maintaining or decommissioning Health IT Systems within the NHS.”[4]

While digital clinical safety is not a new concept, and despite digital clinical safety standards in the UK adopted by the NHS in 2009 (then published as DSCN 14/2009 and DSCN 18/2009, now published as DCB0129 and DCB0160), they remain an evolving concept in some healthcare organisations, and indeed to manufacturers of healthcare systems. Therefore, this poses some gaps in terms of identifying and also attempting to mitigate risk. As new digital solutions for diagnostic purposes are rolled out, this comes with potential patient safety consequences.

What are the risks?

It depends on the solution. It could be an AI solution that interprets an image for a dermatology problem or AI that supports mammography results reporting. The AI could give the wrong diagnosis, either a false positive or a false negative. Or, if the system goes down, it produces no results and could delay care. How long it is down for will likely change the severity of the impact at a personal level for those waiting for results. Each scenario is different and needs different steps to mitigate.

How is digital clinical safety 'done'?

There are multiple models, but a common method is to apply SWIFT— a 'Structured What IF Technique'. For example, if we applied this process to a cyber attack that takes a system offline, one question you would ask is “what if the system went down?”. You would consider the cause, the effect of that cause, what hazard(s) might arise and, if those hazards were to come to fruition, what harm could that result in. It always comes back to the impact on the service user. Then come the mitigations. Prevention is first and foremost. But don’t stop there. What if it DOES happen, what then? Then come your reactive controls.

The same applies for an AI-driven diagnostic algorithm. What if it stops working? How do we prevent it from not working. But when/ if it does, how do we minimise the impact of that? What if the outcome is wrong? How is that checked? That could be a human control, but the error has already occurred. The human has to spot it.

That brings us to the current position in relation to AI. The sense check is often the human in the loop; i.e., the clinician, who is left to sense check the outcome. An interesting article by Lawton et al. highlights how clinicians' roles are becoming sense checkers.[5] However, in the early stages of AI-driven diagnostic solutions, maybe this would be comforting to our end users?

A new model for diagnoses

As discussed in Digital diagnosis—what the doctor ordered?, with new tools comes new models of working and, with that, new risks not previously considered, assessed or mitigated. Have we considered, for example, how a person in their own home might use the new tech inadvertently in the wrong way (affecting data in and therefore data out), and what impact will that have on the results and the subsequent long-term health outcome?

But you don’t know what you don’t know. Sometimes these things will go wrong. Humans get things wrong, either in the design or in the use. Risk mitigation can go a long way to reduce risk, but you cannot eliminate it. But when it happens, we must learn from it and importantly share that learning.

Conclusion

There is an exponential growth in technology that will directly impact upon diagnostic safety, but with this technology new risks are introduced that need to be accounted for and mitigated against. We absolutely need to embrace new technology and embed it into our healthcare systems. People make mistakes, and computers can too, but we can reduce the likelihood and impact by applying a robust approach to digital clinical safety, because, first and foremost, it is people who will be impacted.

References

1. NHS England. Update on cyber incident: Clinical impact in south east London, Friday 19 July 2024.
2. NHS Digital. DCB0129 - DCB0129: Clinical Risk Management: its Application in the Manufacture of Health IT Systems. 2018.
3. NHS Digital. DCB0160 - DCB0160: Clinical Risk Management: its Application in the Deployment and Use of Health IT Systems. 2018.
4. NHS Digital. Applicability of DCB 0129 and DCB 0160. Background. 2022. 
5. Lawton T, Morgan P, Porter Z, et al. Clinicians risk becoming “liability sinks” for artificial intelligence. Future Healthcare Journal, 2024:11(1). https://doi.org/10.1016/j.fhj.2024.100007.

Share your experience

Have you been affected by a late diagnosis? Or perhaps you have insights to share on diagnostic safety through the work that you do. If you would like to write a blog or share your thoughts, experiences or resources through the hub please get in touch with our team at [email protected] or add your comments to our community forum page.

About the Author

Ben is currently an Associate Chief Clinical Information Officer (ACCIO) with Clinical Safety Officer (CSO) duties and remain clinically active as an Advanced Practice Physiotherapist (APP) in an integrated musculoskeletal (MSK) service at Midlands Partnership University NHS Foundation trust. He also holds a position on the Digital Health Networks CCIO advisory panel and is the current chair for the Digital Health Networks CSO Council.

Ben qualified in 2008 as a physiotherapist and specialised in MSK. He undertook a Masters in Sport & Exercise Medicine, leading to a role as an advanced practitioner in an Emergency department. Later he moved to a community MSK service where he had the privilege to lead the digital work stream to integrate the MSK service within which he worked with podiatry, physiotherapy and community pain services. He undertook clinical safety officer (CSO) duties as part of this project which allowed him to later secure his ACCIO post.

Ben truly believes digital holds the central role in meeting the healthcare needs of the future but sees daily the challenges and frustrations that can come with digital and digital transformation.

When not immersed in the worlds of digital, clinical practice and professional leadership you can find Ben out on his bike, in the garden, and clowning about with his kids or somewhere in between.