I very much like the down-to-earth approach to unique patient
ID's that GNUmed uses. I am not sure how to handle the
following situations, though:
> 8 characters d.o.b,
According to what calendar ? [eg. Thailand: CE (Common Era) or
BC (Buddhist Calendar) ?]
What about people who don't know ? (eg. Thailand: 50+ years
old hill tribe people can hardly tell you their age in years)
> followed by 1 character code for each gender & ethnic group & eye colour
Which character set (unicode) and language ?
Ethnic group == race in [asian, caucasian, black, ...] ?
What does Native American count as ?
Ambiguity of eye color judgement ?
> two characters initials of mother at time of birth as
> known or as the patient remebers it,
Which character set (unicode) ?
What about not-so-decided names like in some arabic parts of
the world (different ways to construct the "true" name according
to purpose).
> 3 characters international country code of birth,
According to ISO-something, I assume.
> 2 characters first two letters of city/town of birth.
Again, which character set ?
I really think that GNUmed's approach is at least Good Enough
but I did get to know people where most if not all criteria
were not easily if at all filled in.
May I add two suggestion for dealing with those kinds of
situations ? Those shouldn't/needn't be mandatory but they
might help in difficult situations.
1) a picture of the person in question, difficulties with
ageing or facial injury arise, though
2) Something like the "secret answer" approach that quite a
few freemailers use when you forget your web mail password:
During registration you type in a question that you'll always
remember the answer to but that isn't too obvious either.
By answering that question correctly (in case of a medical
office semantic correctnes should be sufficient since
usually a human being is going to deal with matching the
answer you provided during registration and what you now
claim to be the answer) you can be identified to some
degree of certainty. This requires no more than two
additional strings in the database.
And, indeed, have I seen ink fingerprints being used as
signatures and means of identification (if need arises, not on
a daily basis) in Thai hospitals.
Maybe it is worth a thought to issue physical IDs with
arbitrarily assigned numbers in those cases that map into the
grander scheme of a unique patient ID at the domain=="this GP
practice" level. After all we'll be able to identify whatever
card we issued. But, again, problems arise with people unable
to produce that physical ID when needed. It also seems
necessary to complement this with other evidence as to prevent
identity hijacking.
A can of worms, truly. Let's get back to have a good look at
PIDS, again.
Regards,
Karsten Hilbert
--
GPG key ID E4071346 @ certserver.pgp.org
E167 67FD A291 2BEA 73BD 4537 78B9 A9F9 E407 1346
PGP signature
