dm+d: consider filtering out AMPP noise

[sebbacon]

There are a few categories of AMPPs that are "not really" AMPPs, and constitute noise when browsing the dm+d as a "normal" user.

As well as being noisy, they prevent us from being able to identify "single-supplier generics", which are drugs where market forces are unable to keep costs low - an important area that is likely to be costing the NHS a lot of money.

In addition to filtering by the "availability" field by default, we should consider if we can further filter on the following bases:

Things from manufacturers rather than wholesalers

For example, Ramipril 10mg tablets (A A H Pharmaceuticals Ltd) has the wholesaler in brackets, but Ramipril 10mg tablets (Teva UK Ltd) has the manufacturer.

Normally it would never make any sense to order a wholesaler-specific version. Very occasionally it makes sense to order a manufacturer-specific version.

We've been told the reason wholesalers are in there at all is to support pharmacists in ordering from wholesalers. Which is a domain we don't care about, and in fact the confusion has led to the whole Ghost Branded Generics situation.

The dm+d currently has no way of distinguishing these. We could potentially maintain lists of wholesalers and manufacturers ourselves, but this sounds a bit insane, unless in the context of some meaningful research.

There is also an argument for hiding all non-branded AMPPs (except non-biosmilars - which do have a dm+d flag, and also have specific guidance regarding including a manufacturer in the name).

Usually, wholesalers are literally just selling on a manufacturer's boxed product. Sometimes they repackage the original product, still in generic packaging, but with their own branding. In these cases the product licensing and authorisation are all identical but the box looks different.

Things get marketing authorisation but not going to market

We've been told this is a thing, but don't know much about it yet. We think we might be able to identify them thus:

• Try filtering out AMPPs without GTINs
• Try filtering out AMPPs without prices

Things no longer sold for other unknown reasons

• Try filtering on AMPPs with last price dates a long time in the past
• Try filtering on AMPPs without BNF codes
• Try filtering on AMPPs without prescribing in the last 5 years

[brianmackenna]

I have just been speaking to BSA colleagues on a related manner and asked about wholesalers. Their use case is relevant to pricing in the drug tariff. in part 8 and non part 8 items. The BSA uses manufacturers and supplier (i.e. wholesaler) data for example in Cat A where the price is based on AAH, Alliance Healthcare (Distribution) Ltd, Teva UK and Actavis/Accord basket. Also when a pharmacist claims for something that is not in part 8 they are asked to supply as much detail as possible, this may involve the manufacturer and the medicine being reimbursed against a wholesaler dm+d as opposed to a manufacturer dm+d.

On discussion of solutions BSA colleagues advise that the GTIN strategy is a good way forward.

Coverage of GTINs is circa 90% complete and getting better according to them. For more bakground (we may need to look into this more) GTINs are issued by GS1 but they don't require any details of the product they are being issued to for example a manufacturer could request 20 GTINs and assign them in any manner to any of their products. T

[sebbacon]

So does that mean AAH and Alliance Healthcare are the only distributors in dm+d...?

[sebbacon]

Ah - from dm+d implementation guide, it seems we can use licensed route to identify wholesalers...

Wholesaler Items

It should be noted that there are generic products as AMPs and AMPPs in dm+d where the supplier (appearing as part of the description) is a wholesaler and not a true manufacturer (E.g. Kent Pharmaceuticals Ltd). However such wholesalers may also have products that they market within their own livery. In this situation it is not possible to distinguish whether the AMPs or AMPPs represent the products that wholesalers supply or market as their own. These products can be identified by the lack of licensed route information populated at AMP and are present only to facilitate the ordering and reimbursement of products against a wholesalers price point. They should not be available in prescribing pick lists.

But from eyeballing the data, it seems like nonsense. The only AMPs without licensed routes are things like imports and specials (see dmd2.vw__available_ampps_with_pricing)

[sebbacon]

I have just noticed that step 6 of the DMDProduct pseudo-code in the implementation guide has something to say on availability restrictions:

The AVAIL_RESTRICTCD in the AMP entity identifies any restrictions on the availability of the AMP. These examples supply this information for the related VMPs. The first part of the example sets the AVAIL_RESTRICTCD for all VMPs where one or more associated AMPs are indicated as being available. The second part of the example sets AVAIL_RESTRICTCD for all VMPs where the associated AMPs have some form of restriction but are not set as “not available”. The final part of the example looks for a situation where all AMPs related to a VMP are discontinued. Note: There may still be stock available in the supply chain. These examples have to be executed in this order or a different result will occur

[brianmackenna]

NHS Digital have contacted us to say they have added a new appendix 1 to the primary care dm+d implementation guidance with a new methodology to filer out wholesale products.

It looks like they have used @sebbacon approach of utilising GTINs. I don't think we have investigated the GTIN strategy yet so leaving open for future use. This will have relevance to OpenSAFELY and how dm+d is implemented in primary (and secondary care) EHR vendor systems.