Re: [Gnumed-devel] where are we at?

[Hilmar Berger]

Hi,
On Sat, 14 Dec 2002, Ian Haywood wrote:

> gmclinical.sql is also lagging behind (AFAIK, no-one has touched it
> since the conference) I have added some tables for recording
> prescriptions (I do not pretend what I have is ideal, hopefully it will
> spark debate!)
You asked for it :) - so here are my comments:

Table script_drug: 
 1. IMHO every prescription should be stored completely as text, too. 
   Reason: There are drugs, that are prepared by the pharmacist and
therefore are not listed in any drug database. Other examples might be
prescriptions that for some reason do not match the structure given in the
table (this might be the case for additional entries, drug timings
not covered etc.). 

 2. I would suggest storing the brand name (if known) instead of
having a default entry "GENERIC". 
 Reason: 
  a) Brands containing the same generic drug often have different adjuvants
that might cause adverse drug effects. We should store information as
precise as possible about the drugs prescribed.
  Since we don't check for the presence of a related entry in
"constituents" we might end up having an entry in script_drug without any
information on the type of drug we prescribed. Maybe there should be a
rule on inserting an entry in table script_drug ?
 
If, however, the prescription is stored as text this problem will somewhat
disappear.

3. Frequency/time of drug application seems to be a rather complex issue.
I'm not quite sure if you can really cover all possible/meaningful cases
within a simple structure. Think of different dosages at different
timepoints, more than 3 timepoints a day or drugs applied only at 22.00
(that is, once a day, but not at the usual 'nocte' timepoint about
18-19.00. I Germany we use a scheme like 1-0-1/2 or 1-0-0-1 to denote
dosages and timepoints of an applied drug. I feel that it would rather be
difficult to come up with an easy structure to store this. 

I would suggest storing all that information in just one text field in a
predefined way. You can still parse the text if you really want to
know what exactly was done.

Hilmar