Soapbox: Why an app isn’t like a book

The suggestion has been made recently at a couple of events that this editor attended that there is an unnecessary fuss over regulation of medical apps because they are just like medical books; as there is no regulation of books, why the need to regulate medical apps? . In order to try to move to a consensus, this post puts the opposite point of view, to stimulate debate. In summary the arguments of why they are different are:

  • We are familiar with books and have worked out how to deal with them;
  • Books give formulae and leave users to compute; apps do it all, often without showing their working;
  • Tablets and, especially, smartphones have screens that are smaller than books so require a different design.

This issue of course only relates to serious medical apps – something like 99.5% of all health apps available are very unlikely to do serious harm, helping people as they do record things like their fitness and their weight, and so do not require such detailed scrutiny. It is the ones that get close to, or meet, the test of being a medical device that are of particular interest here. The goal is that once clinicians are comfortable prescribing medical apps, and patients are comfortable using them, the NHS will save substantial sums by, for example replacing drugs with apps for a range of diseases where both are effective and apps are far cheaper. There are also huge benefits for clinician-facing apps – properly certified medical apps like Mersey Burns and Mersey Micro are already massively improving patient outcomes and significantly reducing NHS costs.

In more detail, books have been with us for many centuries so we are familiar with their structure, with the processes for their removal from publication if they give dangerous advice, and with the idea of specialist publications accessible by appropriate experts only – the same is not true of apps. In the event that advice in a book was dangerously wrong, as technical literature does not typically fly off the shelves, it would likely only harm patients one at a time, and the authorities would be alerted early, whereas apps that can be downloaded by many, instantly, so could quickly affect thousands. Also, we know where to find and check references on books, though we typically don’t on apps (indeed many apps don’t bother to include them at all, although they should).

A second, separate, point is that books don’t compute, they only give the formulae, then leave it to the user to add the appropriate variables and compute the result. Many medical apps involve computations, for example of insulin dosage for those with diabetes, or to tricky drug dosage calculations for clinicians, that could potentially have flaws in them unless carefully checked. Worse, many just give the result without showing the calculation. There is therefore much more out of the user’s control or view that can go wrong with, for example, a dosage calculation on an app than if using a book. Richard Brady has demonstrated this admirably in his paper examining opioid converter apps where he showed a difference of a factor of 13 in the results these apps produced for the same input. One of the outliers in this review could therefore result in serious harm being done to a patient.

(Lest anyone consider the opioid experience exceptional, Richard has also mentioned in passing articles on potentially harmful melanoma apps here, here & here, a blood pressure app, and to a paper identifying deficiencies in some smartphone calculator apps.)

A final point, made to this editor by Prof Jeremy Wyatt of Leeds University, is that smartphone screens (especially) are much smaller than books, so if a book page contains, say, an exclusion at the bottom of the page (eg “this calculation is not appropriate for pregnant women”, or “if you have heart failure, you should ignore this advice and turn to page xxx”, etc.) the app needs careful design to ensure that, if the person receiving the advice is not covered by the app, they know they aren’t, because it might be impossible to fit exclusions on the same page as the advice.

For regular readers of Telehealth & Telecare Aware, it goes without saying of course that there is also a European law that requires certain medical apps to be treated as medical devices, which is not the case for paper books. This has been covered in exhaustive detail previously. There also seems to be little disagreement among those working with medical apps that there are some potentially extremely harmful apps around – perhaps the uncertified dosage calculators used by hospital clinicians is the group that poses the highest risk. It is therefore important that agreement is reached in high places soon as to how medical apps should be both regulated and promoted, so that the role of agencies such as the MHRA, which is understood to have insufficient manpower allocated to the area to police potentially harmful unregulated apps, is clarified.

Categories: Latest News and Soapbox.