Re: [Gnumed-devel] state of test results handling

[James Busser]

On 1-Apr-08, at 1:28 AM, Karsten Hilbert wrote:
> > - they *may* (in some jurisdictions) need to be signed by the  
> > ordering
> > physician, even if already signed by someone else
> No problem. Each physician can have their own review.

But I would make the case against preserving *multiple* reviews per  
test result (even if it is only one per user) and, for each result,  
would instead obsolete all but the most recent review.

The associated table clin.reviewed_test_results has been designed to  
record, for each test, one or multiple rows each of which includes  
all of the following:

- acknowledgment that each has been "reviewed" i.e. the "signing"  
function
- agreeing or correcting whether results had been correctly  
identified by the test_org as technically_abnormal
- interpreting whether each is noteworthy (clinically significant)

As far as signing, even in a multi-clinician practice, there will be  
cases where a single signing "row", when the signing was done by the  
clinician who ordered of the test, will be enough. In a circumstance  
where --- in the absence of the ordering doctor --- results were  
signed by a colleague, then at most only two signings may be  
required. The second signing would be by the doctor who ordered the  
test. This would complete any statutory or jurisdictional  
requirement, where this existed, and could (in fact) obsolete a  
signing row that had been created by the previous doctor (a copy  
would exist in the audit table).

I am still trying to understand the rationale behind maintaining more  
than one non-obsolete signing row per test. Presently, health issues  
carry an attribute as to whether or not they are clinical significant  
and for each item there exists only a *single* active value  
reflecting the decision of whichever clinician last defined or  
changed it. There is no provision for each of multiple reviewers  
deciding and keeping active in the EMR whether the individually (and  
differently) believe it to be clinically significant therefore I  
don't understand the reasoning behind supporting multiple different  
judgments by different clinicians whether a lab test is clinically  
significant. I am now wondering how this is framed for the document  
archive items. I think clinical relevance should be managed under the  
same approach across all three.

Working through some test result use cases:

1. say an in-praxis colleague (of a patient's usual doctor) orders  
tests, and becomes (by default) the intended reviewer. If this  
ordering doctor is the first to sign, they could (as part of signing)  
"refer" responsibility to the usual doctor, who would become the new  
intended reviewer. The requirement for the ordering doctor to have  
signed the results would have already been satisfied, and would be  
verifiable from the audit log. Any required action on these results  
would remain with the ordering doctor, until other arrangements are  
(in the meantime) made or another doctor in the group (e.g. the usual  
doctor) intercepted and took action. The intended reviewer should  
have to remain the ordering doctor until they had been able to sign.  
Therefore instead of users "taking" responsibility for results, it  
should be the intended reviewer who, if it would be the agreed  
procedure within a praxis (as part of signing), can redirect the role  
of "intended reviewer" to the patient's usual doctor.

2. next suppose the ordering and patient's usual doctors (whether or  
not they are the same person) have "both" signed and a colleague sees  
the patient and, in examining the already-signed results, notices one  
or more that warrant editing of their test_abn and clin signif  
values. Altering these values could only be done as part of signing,  
making evaluable and identifiable any results that are the  
responsibility (based on fk_intended_reviewer) of, yet not currently  
(re)signed by, the current user

Technically_abnormal will most often already be correct as supplied  
by the test org. The first review by any clinician can confirm what  
was received, or can compensate for situations where test  
abnormalities had been unaccompanied by flags. A second review, for  
example by the ordering clinician, could carry-forward everything  
that was correct from the first signing as well as fixing any  
disagreement or failure in what the lab had provided. But we are  
really here talking about carrying forward into a new state the most- 
correct picture of technical abnormality. There is no question of  
merit... we are not aiming to weigh the first clinician's judgement  
against the second. It is not a question of opinion. Therefore any  
new reviewed value for technically_abnormal should deprecate the  
previous. I cannot understand why different clinicians should want to  
re-examine results through the lens of "personal definitions of  
technically_abnormal".

It seems to me that for any test result, the newest signing should  
obsolete not only the current user's row but any user's row.

BTW clin.reviewed_test_results needs a foreign key "dem.identity.pk"  
defined for fk_reviewer