Re: [Gnumed-devel] About systems of codification

[Jim Busser]

On 2011-09-29, at 1:58 AM, Karsten Hilbert wrote:

> On Thu, Sep 29, 2011 at 10:22:38AM +0200, Karsten Hilbert wrote:
> 
>>> Had a suitable coding been available from the start, there
>>> could (in many cases) have been *no need* for the free text.
>> 
>> I TOTALLY disagree with this idea and there's been books
>> written on the subject.
> 
> Such as: 0-9615255-2-5

While I have covered some of this off in the interim replies, here are key 
paragraphs from a JAMA review of the above book

        JAMA, November 22/29, 2000—Vol 284, No. 20

with which I concur, and on which I'd comment:

> The authors propose a solution called entity coding, which allows for the 
> capture of clinical detail, yet supports classification of the data for such 
> purposes as research, statistics, and billing and reimbursement. The authors 
> even outline how the transition to entity coding should occur and recommend 
> that the wholesale adoption of International Classification of Diseases, 10th 
> Revision (ICD- 10) in the United States be delayed until entity coding can be 
> implemented nationwide.

It could be worthwhile (however difficult) to convince the agencies to delay. 
They might argue that failings in ICD-9 demand that improvements not be 
delayed. Which is a pity, if a better solution would be to derive understanding 
(and policy decisions) from better medical terminology which is being 
inadequately developed and implemented.

The problem gets exacerbated by the enforcement of adoption, by upstream 
clinicians, of systems that are not fit for purpose.


> Once implemented, entity codes could then be mapped algorithmically to 
> classifications such as ICD- 10, preventing the need for any direct, human 
> classification of patient cases. The advantage of this approach is the 
> elimination of variability in classification, the subsequent inaccuracy and 
> variability of data, and the consequent ill-informed policy, funding, and 
> reimbursement decisions.

Agree. Why perform manually that which could be computed from better-quality 
upstream information (except when no code adequately captured a clinical 
condition, which then and only then would justify manual adjudication).


> The fact that the first author helped the government develop the clinical 
> modifications to ICD-9 lends credibility to the description of the problem 
> and its consequences for the health care industry and to the solution that is 
> set forth.

ok


> The argument for entity coding is logical and clearly stated—the book is 
> comprehensible to the average reader; one need not be a clinician or a coder. 
> However, entity coding is unproven —at least the authors present no evidence 
> of its success in health care—thus an argument against the wholesale adoption 
> of such a system nationwide all at once, as the authors propose, can be made.

An important point. Every attempt at implementation has met with challenges. It 
is not yet any 'slam dunk', and so any hopes vested in wide-scale, all-at-once 
implementation may be misplaced. 


> Perhaps a trial of entity coding and an evaluation of its benefits would be a 
> more rational step.

yes


> Furthermore, there is a large body of medical informatics literature about 
> medical terminologies and coding sys- tems, and a different solution is pro- 
> posed in this literature: concept-based reference terminologies with a rich 
> set of hierarchical and semantic relation- ships among them.

e.g. SNOMED-CT

> both camps could learn from one another to create a solution that draws on 
> the strengths of both approaches, providing an even better solution

yes

-- Jim