Re: [Gnumed-devel] Challenges in lab test aggregation - panels

[Karsten Hilbert]

On Sat, Aug 31, 2013 at 12:03:08AM +0000, Jim Busser wrote:

> This query
> 
>       SELECT T.pk, fk_test_org, T.name,  T.comment, conversion_unit, T.loinc, 
> T.abbrev, fk_meta_test_type, M.name
>       FROM clin.test_type T
>       INNER JOIN clin.meta_test_type M
>       ON T.fk_meta_test_type = M.pk
>       ORDER BY T.pk;
> 
> returns
> 
>  pk | fk_test_org |           name           |   comment   | conversion_unit 
> | loinc | abbrev  | fk_meta_test_type |        name        
> ----+-------------+--------------------------+-------------+-----------------+-------+---------+-------------------+--------------------
>   6 |           4 | leukocytes (EML)         | EDTA sample | Gpt/l           
> |       | WBC-EML |                 1 | leukocytes
>   7 |           4 | erythrocytes (EML)       | EDTA sample | Tpt/l           
> |       | RBC-EML |                 2 | erythrocytes
>   8 |           4 | platelets (EML)          | EDTA sample | Gpt/l           
> |       | PLT-EML |                 3 | platelets
>   9 |           4 | C-reactive protein (EML) | blood serum | mg/l            
> |       | CRP-EML |                 4 | C-reactive protein
> 
> leading me to wonder whether
> 
> 1) when creating any new test_type, the requirement for
> the 'name' to be unique

There's no such requirement.

> should by convention inherit an
> abbreviation for the lab test org and

I only did this to clearly distinguish those types from any
types the user may want to create for real.

But employing such a convention may prove a wise decision in general.

> 2) when creating a panel or profile, will we be able to input
> 
>       meta_test_types

Not ATM.

> or will the individual test_types each have to be
> discretely defined and will this result in discrete rows
> that do not take advantage of the aggregation?

For the time being, yes.

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