Re: [Gnumed-devel] GNUmed support for coding systems

[Karsten Hilbert]

On Wed, May 04, 2011 at 11:17:11AM +0100, Adrian Midgley wrote:

> Storing codes with an extra field to indicate which code family or
> system it came from makes it much easier to import medical records
> from elsewhere in good order.

Sure. Our structure for storing reference data on coding
systems goes like this:

create table ref.data_source (
        pk serial
                primary key,
        name_long text
                not null,
        name_short text
                not null,
        version text
                not null,
        description text,
        source text
                not null,
        unique(version, name_long),
        unique(version, name_short)
) inherits (audit.audit_fields);


create table ref.coding_system_root (
        pk_coding_system serial
                primary key,
        code text
                not null,
        term text
                not null,
        fk_data_source integer
                not null
                references ref.data_source(pk)
                on update cascade
                on delete restrict,
        comment text,
        unique(fk_data_source, code),
        unique(fk_data_source, term)
);

        -> all coding system tables inherit from this one

> It is an approach we thought well of toward the end of last century.
> I think it is pretty much unused, possibly due to copyright
> contentions by coding system suppliers.
> 
> It seems clearly to me the correct way to go.

Absolutely.

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346