How can we compare equivalent products which don't have generic equivalents coded?

[sebbacon]

This issue is the inverse of #9.

There are over 1000 products without a generic equivalent listed in the BNF (identified by codes ending in A0).

Some of these can be meaningfully compared with each other, despite not having a formally coded generic equivalent. For example, Blood Glucose Test Strips are clinically equivalent, but are coded as A0. In this case, we can group them together by BNF subparagraph -- everything within 0601060D0% is a blood glucose test strip, and we believe that the quantity field consistently refers to number of strips.

We need to come up with a list of products without generic equivalents which we can add as special cases to the code that computes price-per-pill variability.

As these are all special cases, and we might not be able to solve them all with the "subparagraph" trick that applies to BGTS, we may need to prioritise only the ones with higher spending.

The approach discussed is to

• make a list of A0 products ordered by total spend
• exclude foods from the list
• pick an arbitrary cut-off point below which we don't care about (e.g. £20k per month?)
• for products above that point, pick things where we think meaningful comparisons could be made, and record here (with reasons)

An example of a product we don't think we can handle is ThickenUp cream -- because it is recorded (like glycopyrronium) inconsistently in quantities of 1 (box) or 75 (sachets in a box). See this analysis for a discussion

[sebbacon]

The vast majority of A0 presentations that look promising from eyeballing them are those in:

• Urine Testing Reagents (0601060U0%) and
• Glucose Blood Testing Reagents (0601060D0%)

In both cases, the subparagraph only contains comparable products.