analytics

More on Duplicates

Duplicate and Hierarchy Problems

  • Roger Weeks; MBBS
  • Dec 16, 2016



I am working on coding for an Electronic Patient Record system converting concepts from Read 2 and CTV3 to SNOMED CT and ICD10 CM but am getting bogged down because I keep finding duplicate concepts sprinkled with the addition of seemingly pointless (0 points for TV addicts), wretched 'Observable entities'. All we need is a single (i.e. unique concept) to use for our conversion but what we get is duplicates (my record is 6, yes six) often from different hierarchies or sometimes the same such as a child concept being for all practical purposes identical to its parent. This problem is exceedingly common throughout SNOMED CT – and yes, I have alerted IHTSDO.


This problem is exceedingly common throughout SNOMED CT – and yes, I have alerted IHTSDO.

Here is my latest bête noir: I need a 'header' for the set of child concepts for the simple process of taking/measuring the oral temperature. Now the purists might say 'Ah, we need a concept for the procedure and a concept for oral temperature but throughout Read and SNOMED these are often combined - there is no consistency to this so to avoid confusion put them all together.

Looking for simple 'oral temperature' concepts in SNOMED CT I can find four:

example

(NB The version 2 Read code(s) are first, then the SNOMED CT code, then the Terms and synonymous terms separated by ';' and finally the CTV3 code(s))

My answer to this problem is to amalgamate the whole lot like so:

example

This is a lot of work which should be sorted out by IHTSDO as according to them:

"Membership and national use of SNOMED CT have a number of advantages in terms of reduced costs and improved care, such as:
  • Cost savings in terminology development, licensing fees, and transcription costs of mapping to ICD or other coding systems;
  • Reductions in inappropriate or duplicate testing
  • Reduction in adverse drug events
  • Improved management of key disease groups, such as cardiac care and oncology

Encoded health information and decision support tools enable these benefits. Consistently codified information enables comparison, exchange and reuse throughout the healthcare system. Therefore, an open SNOMED CT-based ecosystem presents an opportunity for exponential increases in benefits"

One rarely sees products or services described as saving costs, time AND improving operational performance when the purchasers are expected to fix the known errors. The Kano model describes the outcome of failure to meet basic customer needs , see Kano's Model.

Lack of Consistency is the norm for SNOMED CT and can time and time again easily be demonstrated.

The major other problems are:

  1. Downloaded file structure does not follow data base update rules - i.e. files do not appear as New, changed and dropped. Where concepts are deprecated no clear indication of true maps are given, frequently there are NONE., MAYBE A, and rarely (what we really want/need) ISA.
  2. This makes SNOMED CT really, really unstable because the apparent tinkering that goes an all the time in a seemingly random sort of way means that the developer (me) suddenly discovers that a set of concept he thought were sorted and stable have been changed, deprecated or gone to the black hole.


NEXT TIME

I will be asking questions about substances and the Substance hierarchy problems which have been known about by IHTSDO for several years. This involves the important areas of drug ingredients unfortunately called (Substance and Product)

387461009 | Digoxin (substance)
796001| Digoxin (product)

This is really important for sorting out adverse drug events which I thought had been abandoned by IHTSDO as I was told at their April London meeting this year that it was impossible to get/find/make a perfect or even workable drug prescribing model and because of a 'lack of resources’ they were not going to attempt sorting out the nonsenses, duplicates, missing concepts (e.g. most so-called inactive ingredients) and the fact that many non-drug substances and elements (e.g. Potassium) are in the pharmaceutical hierarchy!

But that's next time.

Comments

Work in progress

Roger Weeks

The aim of this blog is share my findings with the community. Uneasy at the lack of objective discussion of the issues around the uncritical adoption of SNOMED CT I am hoping to support the community by sharing my findings. As I am developing a comprehensive thesaurus of medical terms much of my focus, naturally, is on SNOMED CT. The bottom line is that SCT will be used in critical clinical systems and unless there is a rational dialogue about the issues much of the anticipated beneficial deliverables will not be met.

Blog & Comments

Roger Weeks

Notifying IHTSDO. I am not acting as an unpaid QA consultant for the well funded IHTSDO but prefer to draw your attention, dear reader, to the prolific types of error that you should be aware of before implementing SNOMED CT and come to your own conclusions about how the process is being managed. My own solution is to create a thesaurus that accommodates all the terms under a single entity that cluster like terms for data entry and retrieval. If you would like further information about the Thesaurus of Medical Terms please contact me at the email address shown below.


The aim of this blog is share my findings with the community. Uneasy at the lack of objective discussion of the issues around the uncritical adoption of SNOMED CT I am hoping to support the community by sharing my findings. As I am developing a comprehensive thesaurus of medical terms much of my focus, naturally, is on SNOMED CT. The bottom line is that SCT will be used in critical clinical systems and unless there is a rational dialogue about the issues much of the anticipated beneficial deliverables will not be met.


Previous Blogs
Intestinal Intussusception - Duplicate and Hierarchy Problems

For Further information

Please contact:

Dr Roger Weeks - roger@medicalintelligency.com

or

Steve Mott - steve@medicalintelligency.com

or

Use this contact form