Two new glucose meter standards are nearing completion: CLSI POCT 12 for hospital based glucose meters and ISO 15197 for home use glucose meters. The current versions of these standards use the same performance specification – the CLSI standard defaults to the ISO standard.
The latest drafts of the new standards have tighter performance limits. The current ISO standard states that performance limits are based on medical acceptability. This sounds good and who can argue with that. But one can ask, has medical acceptability in diabetes changed this much in 10 years to warrant tighter performance limits? Another possible explanation for the spec change is that performance limits are based on glucose meter performance capability and this capability has improved.
To base standards on the performance capability of an assay is not so strange. The 510(k) process uses the principle of a candidate assay’s performance being substantially equivalent to a predicate assay.
Moreover, medical acceptability is a difficult concept. One could argue that patient harm increases with increasing assay error and the only condition that is benign is the case of no error. So it is difficult to draw a line which would equate patient harm as being medically acceptable.
Assay error occurs so there will always be some patient harm but the alternative – to keep an assay off the market and prevent clinicians from obtaining the information that the assay provides will invariably be more harmful than patient harm from assay error.
So the basis of a spec to set limits according to the performance capability of existing assays makes sense. And it is not based on medical acceptability.