[Gnumed-devel] More on HL7 in BC CA (was the Open Source thread)

[Jim Busser]

On 2010-08-16, at 7:48 AM, Jim Busser wrote:

>> Major messaging companies now send specialists letters via hl7 (eg Medical 
>> Objects - you can  check their web  site if interested).  The only problem 
>> is 
>> not being able to receive images. The hl7 received in this format I'd say is 
>> 100% reliable.
>> 
>> I looked at Mirth and gave it a miss personally.

A local "data broker/delivery" company (Medinet, different than the one which 
is sending the path lab granular results) distributes a lot of local text based 
results.

These are old-fashioned, narrative-type text results... ASCII… anatomical 
pathology, consultations each is a "report". Also included in this transmission 
type are the reports of some hold-out path labs who did not HL7 encode 
individual per-test results, and who supply only a blob of test results for 
example

Sodium     128   L   135 -- 145 mEq/L

where, each Windows CR/LF wrapped line within a report (below its header -- see 
below -- plus various miscellaneous information) you get a plain text "test 
name" (uncoded) and to the right of which is a usually-numeric result, then in 
some cases a "flag" maybe H for High, L for low, and at far right a "reference 
range" of {normal / therapeutic}.

Medinet previously provided both text and HL7 reports, either because the 
upstream lab in some cases already provided HL7 or maybe Medinet did a 
conversion, however factors unknown to me caused the HL7-providing labs to 
instead flow their results through Excelleris.

As a result Medinet only provides text (making some of their web page content 
obsolete) so the HL7 "bits" on this page need ignoring

What Medinet does is to parse each source and to create a HL7 header which has 
only 10 fields as per

        http://medi.net/jjjj.htm

For Field 7, "Lab" where the example offers CAP^3577 I've asked them to define 
this (unacknowledged) ^3577 minor field

For Fields 9, 10 "PHSA" and "VCH" refer to two public system, mostly 
hospital-based lab sources

The sample header happens to end after Field 7. It was explained to me that HL7 
specifications define that each message ends as soon as it has run out of data, 
i.e. that when trailing fields would be empty, they are simply not even 
delimited empty. In any case of "empty" fields that would be *followed* by a 
value, then placeholder pipes "|" will have been supplied up until the last 
value that is supplied.

I expect that each header is separable from the actual start of the report by 
Windows CR/LF and that the report is separable from the next "JJJJJ" header by 
a Windows CR/LF?

I would postpone by at least several weeks any consideration of how to get this 
into GNUmed, so that what Luke has come up with can have its tires better 
kicked, but I imagine there is enough there.

-- Jim