Apps for apps sakes?

The EU’s Innovative Medicines Initiative WEB-RADR project is a €6bn project looking at how public reporting of adverse reactions can be improved, interrogating social media and providing better information. A lot of focus has been the provision of an app epitomised by the MHRA’s Yellow Card app.

A recent paper in Drug Safety(1) looked at some public reactions to this.

Outside of a small focus group sample, the important issue is whether a pure ADR reporting app would be widely adopted. The main problem I foresee is that people are unlikely to download such an app in the anticipation that they might get (and recognise) an ADR. Space on smartphones is often at a premium and granting memory for an app with a low perceived need is questionable. Low awareness of this class of app, for a situation that might rarely happen (at the individual level), to report a circumstance that would be difficult to recognise, is how I would characterise this. A case of an app for apps sake?

However.

Many of the comments made in the paper, along with the clinical and research necessity for better ADR detection, lead me to look at two alternate scenarios.

The first addresses the issue that voluntary reporting of ADRs by clinicians is a low level priority. The Gold Standard has a low bar. The 2006 BMA report ‘Reporting Adverse Drug Reactions’ (2) makes this very clear. The underreporting of ADRs is deemed to be significant and needs to be addressed.

One way to do this is to become less dependent on voluntary reporting and start taking advantage of the technology that is widely utilised in healthcare systems. By directing our efforts to the background analysis of electronic health records in medical record systems we can look for newly reported problems or side effects, providing hypothesis generation and strengthening. By collating locally processed data centrally in a national system, it would be possible to identify potential signals automatically and then direct further research.

Secondly, for patients, the greatest potential harvest of ADR data is from social media because that’s where the people are. There is a huge amount of comment in social media platforms and fora about health and health problems often related to treatments. Using clinical natural language processing (CNLP) it is possible to identify patients’ comments on medication issues.

Rather than creating apps, surely it would be more productive to embed ADR reporting functions into the largely untapped social media community (web and mobile). By embedding very simple filters into these platforms we can identify potential problems and avoid submerging users into full blown reporting. Thus, we add ADR reporting into social media services rather than depending on a large range of variable quality health information services.

Finally, software developments such as machine learning, conditional logic and neural network systems could help us to greatly resolve patient safety issues no matter what the cause (even better where we can connect to a patient’s EHR). Hardware progress provides collaborative opportunities; voice activated data communication systems such as Amazon's Echo could provide a potential mechanism for the public to engage in triage, outcomes and symptom reporting.

"Alexa I don't feel well"!

(1)   https://link.springer.com/article/10.1007/s40264-016-0494-x

(2)   https://www.isoponline.org/wp-content/uploads/2015/01/BMAreport.pdf