Previous Next

Question: Who should postcoordinate?

Just-in-time postcoordination

Postcoordination as described above offers the ability to dynamically extend a compositional terminology. But where should this extending take place, and who should do it?

One approach is to continue offering the terminology to end-users only as static term-picking lists, similar to an enumerated scheme. When no term in the picking lists matches what needs to be recorded, the clinician either selects a best fit from the available options, or sends a request for a new code to a local curation service, who may then use postcoordination and a local namespace to add it.

This just-in-time postcoordination essentially exploits the compositional nature of a terminology to devolve and distribute the curation of a static, pre-enumerated terminology. This has obvious advantages, principally a dramatically improved response time for feature requests when compared to the normal enumerative scheme release cycle. It also has disadvantages, in particular the problem of reliably detecting and reconciling when the same new concept has been created by different just-in-time centres but assigned different local identifiers.

Such reconciliation should be easier with just-in-time postcoordination than for the more familiar scenario where data has been encoded using entirely different schemes: the potential differences between any two schemes are already known, as the set of local namespace extensions to be compared and contrasted is readily identifiable. Without this reconciliation, however, sharing data between different sites will remain unreliable, and the behaviour of centrally authored decision support algorithms deployed over multiple centres will be unpredictable.

If just-in-time back-office postcoordination gains increased local expressivity and responsiveness whilst minimising the need for users to learn significant new skills, it may be doing so only by deferring much of the necessary expense to secure wider interoperability.

Postcoordination at the clinical coalface

Just-in-time postcoordination also presupposes that picking lists of potentially ever-increasing length, or an ever-increasing library of bespoke static data entry forms, all comprising terms of arbitrary coverage and detail, remain either the only or the best paradigm for clinical data acquisition.

As an alternative to picking lists, offering postcoordination directly to the end-user may be considered.

Previous Next
Making the impossible very difficult, ©OpenGALEN.org, All rights reserved