Glycaemic control apps for diabetes: lifting the lid

The rapidly expanding diabetes app market presents new challenges for patients, health care providers and regulators

The medical technology industry’s explosive growth has set the stage for medical mobile applications (apps) to assist with the management of many chronic diseases. Diabetes mellitus is the archetypal example: its management requires optimising diet, physical activity, blood glucose self-monitoring, and safe medication use. As a multifaceted condition affecting so many people, diabetes is the perfect target for developers of medical apps to realise the promise of the digital revolution.

There are over 1500 diabetes apps available on iTunes and Google Play, the two largest app marketplaces, a number growing faster than any other health care sector.1 App functionality ranges from the simple — information provision, social networking, reminders, documentation of clinical statistics, and data sharing with health care professionals — to the more complex — analysis of glycaemic patterns to derive insulin dose recommendations and blood glucose predictions.2

App marketers make various claims regarding clinical outcomes; anecdotally, the most common claims for diabetes apps involve improved glycated haemoglobin (HbA_1c) levels and reduced hypoglycaemia. Despite the multitude of commercially available apps, recent meta-analyses identify few randomised controlled studies of safety and efficacy.1,3 These analyses found an overall significant difference of just under 0.5% reduction in the HbA_1c level over a range of 1–12 months of follow-up with app use, greater in participants with type 2 (−0.7%; 95% confidence interval [CI], −1.0% to −0.3%) than type 1 diabetes (−0.4%; 95% CI, −0.8% to +0.1%).1 Such a difference is clinically similar to the reduction seen with the use of non-insulin hypoglycaemic agents.4 The clinical relevance of the effect seen in patients with type 1 diabetes is more questionable. Severe hypoglycaemia data were limited: five studies reported results and only one showed significantly lower rates in type 1 diabetes.5 App features that significantly improve glycaemic control have been shown to include complication prevention modules and structured displays. Of note, improved glycaemic control has not been associated with apps providing medication management advice.1

Therapeutic functionality is of particular concern in diabetes apps. Although apps increasingly advise on insulin doses, there is minimal published information on safety and efficacy, despite these apps effectively providing drug treatment recommendations without health care professional oversight. Only three randomised controlled studies of two apps have been published to date.5,6 It is thus clear that most diabetes app manufacturers make claims that are unsubstantiated by appropriate scientific evidence. The largest systematic assessment of insulin dose calculation apps reached alarming conclusions: of the 46 apps reviewed, only one was without safety concern.7 The majority (67%) carried a risk of inappropriate dosing, including dose suggestions that would result in hypoglycaemia. It is unclear who, if anyone, is medico-legally responsible for adverse effects related to app-derived therapeutic recommendations.

The approval of new drugs requires the provision of scientific safety and efficacy evidence followed by ongoing post-marketing surveillance, but medical apps with the potential for harm make it to the market without these rigorous processes. Under section 41BD of the Therapeutic Goods Act 1989, medical software that encompasses the “diagnosis, prevention, monitoring, treatment or alleviation of disease,” including all diabetes apps that go beyond simple information provision, is regulated as a medical device.8 When applying for approval, manufacturers determine a device’s risk category from I (lowest risk), II, III or AIMD (active implantable medical device; highest risk) based on the Therapeutic Goods Administration’s (TGA) rules. Thus, clinicians can rely on TGA approval of a glucometer with integrated bolus calculator capabilities, a Class IIb device, to mean significant malfunction resistance as the TGA has independently assessed the device.8 However, physicians cannot be sure of relying on insulin dosing recommendations made by a glycaemic control app, a Class I device, despite it having the same approved registration status, as it did not even require a conformity assessment — the most basic assessment of fault resistance.

Diabetes apps, in particular therapeutic apps with automated insulin dose calculation technology, promise significant clinical benefits. However, their rapid uptake and expansion of features has outpaced regulation. Without appropriate oversight, these medical interventions have the potential, in the absence of efficacy, accuracy and safety data, to compromise clinical care and detract from the proven quality improvement that health systems diabetes management already enjoys. Anticipated reforms to the TGA medical device registration framework following the Sansom Review9 and alignment to the new European Union medical device regulations10 will likely modernise regulation to encompass these new device classes.

It is important for all health care professionals who manage people with diabetes using therapeutic apps to be aware of the potential limitations to their use due to scientifically unsubstantiated medical information and management advice provided by the majority of these apps. Furthermore, the regulation of therapeutic apps and the claims made by their developers do not currently provide patients with sufficient protection, which the Australian public can reasonably expect from our mature regulatory bodies. Despite this, there is an increasing number of diabetes apps being developed and marketed, and an increasing uptake of these apps by patients seizing the opportunity to realise this promise. An opportunity exists to establish practical guidelines and a framework to ensure the development of effective, safe and relevant health apps, with appropriate regulatory oversight, if relevant, which are also financially feasible to developers. This is an important challenge to maintain the interest of app developers in solving relevant and practical clinical problems so that patients may safely benefit from novel technological solutions in chronic disease management. In the meantime, health care professionals must, at this stage, remain circumspect with regards to recommending and endorsing the majority of diabetes apps and, in particular, those apps with therapeutic functionality.