This piece was originally posted as a Birmingham Law School Research Blog Post
This post summarises a paper we published a few months ago in Diabetic Medicine: ‘#WeAreNotWaiting DIY Artificial Pancreas Systems and the Challenges for the Law’*
Authors: Rachael Dickson, Jessica Bell, Amber Dar, Laura Downey, Victoria Moore, Muireann Quigley
Post written by: Rachael Dickson
OS AID Systems
Although more commercial hybrid closed-loop insulin delivery systems are becoming available, the number of people using DIY open source automated insulin delivery (OS AID) systems (also called DIY artificial pancreas systems – DIY APS) continues to rise. As their name suggests, such DIY systems have not undergone the usual regulatory approvals procedures and thus present some legal difficulties for actors involved in diabetes care, including clinicians, parents building OS AID systems for their children, and adult users themselves. We argue in the paper that these issues have received insufficient attention. Although the OS AID systems ecosystem is complex in terms of the actors involved and the actions they take which makes determining liability more difficult, unless clear guidance is issued by relevant bodies or a case is presented in court which clarifies the situation then existing legal principles apply. We give an overview of some of the issues of clinician liability in relation to adult and child users, as well as the liability of pump manufacturers and software developers.
Clinicians and Adult Users
Overall, we recognise the concern from some clinicians that their involvement in care using OS AID systems could create liability if the person experiences harm from using the system. Although OS AID systems are an emerging technology, the actions of the clinician would be subject to the usual laws of negligence. This means the rules are no different to all other aspects of the clinician’s practice and it must be proven that the clinician owed the patient a duty of care, breached that duty of care, and that breach resulted in the harm to the patient. Overall we argued that it would be unlikely that a clinician would be negligent for failing to tell a patient about OS AID systems as a treatment option. This is because OS AID systems would not meet the threshold of a ‘reasonable alternative treatment’. Nevertheless, given the rising popularity of these systems amongst persons with diabetes (PWD), it may be expected that a competent diabetes specialist would have knowledge about these systems.
We also consider the question of whether clinicians could face negative consequences from discussing OS AID systems with a patient. This depends on the nature of the discussion. Merely providing information would likely not count as prescribing, but part of good care, whereas recommending or advising patients to use OS AID systems is different. Central to any interaction is ensuring that patients are appropriately informed of the material risks and doing so in an understandable manner. If this is followed, it is unlikely the clinician would be found to be negligent.
If a doctor makes a negligent statement which a PWD relies upon, then the doctor might be held liable for any resulting harm. Due to the emerging nature of OS AID technology and its user base, it is not clear what doctors would be expected to know. However, what is clear is that doctors must be scrupulously honest with patients about the limits of their knowledge and engage in open communication if they are unsure or unable to advise appropriately.
The third question considered, in relation to doctors treating PWD, is whether any liability could arise from prescribing the component devices required to create a OS AID systems. Here, clinicians would only be liable if they breach their duty of care in such a way that results in reasonably foreseeable harm to the patient. Liability of this nature relies on accepted practice in the field, so a clinician would likely only be found negligent if no other responsible clinician would have acted in a similar manner. The specific circumstances would be highly relevant; for example, withholding prescriptions from a patient already using the requisite technologies because they intend to loop could be negligent if this results in harm to the patient.
Children, Best Interests, and Innovative Treatment
The paper then turns to consider the legal implications of looping on children. In the absence of specific legal provisions on ‘innovative treatment’, non-binding professional standards/guidance and the general principles of civil and criminal provide the basis for navigating this issue. This lack of clarity can be concerning and stressful to all parties involved. Whether OS AID systems are within the bounds of parental discretion over children’s medical treatment has not been tested in court. The particular circumstances will thus be important in determining the bounds of this.
If the health care professionals (HCPs) involved disagree with parents’ choice to use an OS AID system for their child, case law provides some indications of how the courts could approach such a dispute. Overall, good communication is emphasised, including explaining risks. From the perspective of parents, in any decisions about a child’s upbringing the child’s best interests is the primary consideration. The courts adopt a wide approach to determining what is in a child’s best interests and this would be scrutinised to determine if OS AID systems was in the best interests of the child. Overall, while case law can provide some insight on assessing best interests, updated ethical guidance on conducting such assessments would be welcome and reduce the likelihood of conflict between parents and HCPs.
Hardware Manufacturers and Software Developers
Other actors with potential liability include insulin pump manufacturers and software developers producing the OS software code. Product liability law holds manufacturers strictly liable for damage caused by a defective product. However, it is likely that insulin pump manufacturers would dispute liability for any harm resulting from OS AID systems use on the basis that they are being used in a way not originally intended by the manufacturer. Nevertheless, some manufacturers knowingly leave open their pumps’ communication protocols which enables looping to take place. Whilst it is possible that the courts may view such pumps as defective under product liability law, warnings against ‘off-label’/outside of intended use could be considered sufficient to discharge this legal obligation. In terms of a claim of pump manufacturer negligence, the effort which OS AID systems users go to in setting up their systems could be persuasive in arguing that they have consented or voluntarily accepted the potential risk of harm.
For the software developers, the difficulty is that software as a medical device is a poor fit for current law and regulation which was created with tangible products in mind. Lack of clarity in the application of regulations is not limited to software in OS AID systems. The MHRA states that uncompiled software (software not downloadable in its final complete form), including uncompiled OS software, fall within the regulations so long as all installation information is provided. A hurdle to determining liability is the fact that OS AID systems development and maintenance involves multiple possible ‘producers’ and ‘manufacturers’ of code and how one could (or should) distil or apportion responsibility is unclear. We suggest that identifying a final producer, as has been suggested should be the case when it comes to 3D bioprinting, may not be helpful or appropriate here, given the diffuse, global, and collaborative nature of OS AID systems development.
Relying on the law relating to ‘defective’ products would also be challenging. Questions about the clarity of the instructions for intended use and provision surrounding adequate safety warnings would be considered. The OS AID systems community has developed comprehensive development tools, instructions, and warnings that users do so at their own risk, which could help argue that even if they are the ‘producer’ of a ‘product’, the necessary steps to ensure its safety have been taken. This might be especially challenging given that, for the law, software is often viewed as a ‘service’ and not a ‘product’.
A Lack of Clarity
On the whole, the article highlights that the boundaries of legal responsibility in relation to OS AID systems are unclear. In the absence of specific legal provision, settled case law, or explicit guidance from regulators and professional bodies, existing legal principles apply. The DIY approach to diabetes care seems unlikely to disappear soon, thus clearer guidance should be a priority to allay the concerns of clinicians, users, and developers alike.
*This paper was part of a special issue following on from the ‘Do-It-Yourself Diabetes Technology – Opportunity or Threat for Diabetes Research’ workshop hosted by the Danish Diabetes Academy. The workshop brought together clinicians, persons with diabetes, diabetes technology developers, academics, and DIY users to share insights from their work on DIY diabetes and consider the future directions for research in this area. We provided a legal perspective and examined the main challenges that OS AID systems raise for the law.
Funding: Work on this was generously supported by a Wellcome Trust Investigator Award in Humanities and Social Sciences 2019-2024 (Grant No: 212507/Z/18/Z) and a Research England QR Enhancing Research and Knowledge Exchange Funding Programme award.