Friday, 23 December 2011

From pharmaceuticals to interactive devices

Badly designed medical devices have great potential to cause harm – whether by providing inaccurate readings that result in poor diagnoses, or by administering inaccurate doses of drugs or radiation. In CHI+MED, we've been looking at how medical devices are assessed for safety before being deployed. At least as a first approximation, they're treated in the same way as drugs, for which the protocol was originally developed. Both drugs and devices are, after all, classed as "medical technologies".

For drugs, there are well developed protocols, involving randomised controlled trials (RCTs) to initially test efficacy and side-effects, and careful monitoring of production to ensure the quality of every single pill, etc. It is relatively easy to define how a drug is intended to be administered, and hence to draw up a set of assumptions about use (a pharmacist might take exception to this statement, but let's go with it for now). The protocols are, on the whole, fit-for-purpose.

A very similar protocol is applied to medical devices. Where these devices are passive (e.g., stethoscopes, walking frames), it is still relatively easy to draw up a reasonable set of assumptions about how a device is intended to be used. However, when they are interactive (e.g. infusion devices for delivering drugs, or glucometers for recording blood sugar levels), there is much greater scope for variation in use. For example, the way a glucometer is held, or the times of day at which its user chooses to take readings, can significantly affect its performance. The way a nurse has been trained to use an infusion device, or the workload in the ward at the time, or the location and purpose of use (e.g. chemotherapy or pain relief) may have a big influence on how it's used. This introduces many dimensions of variability that are almost impossible to predict or define at the time of design. And yet this is what is expected – even required – of manufacturers.

It's surely time to stop trying to produce one-size-fits-all solutions to the challenge of designing safe medical technologies, when the things that are classified as medical technologies have such different properties. It's reminiscent of the days of the 'Red Flag Act' when cars were classified as being like horses, and were limited to behaving in similar ways – except that that limited the potential of cars, whereas the approach of treating interactive medical technologies like drugs risks compromising their safety because it doesn't recognise the potential for variability in use.

No comments:

Post a Comment