How can I initiate Six Sigma project in Medical Transcription which is totally inherent variability of input quality and human skills.  How to stop the wrong text errors which is totally based on the listening skills of a person and even medical errors like entering wrong drug other than what is dictated due to sound alike.  How can we avoid demographics error which is based on discretion of MT or different specifications by different clients.  To what extent we can initiate Poka-Yoka or other control methods.

Try making a detailed list of all the various types of Errors occuring i.e. errors happening due to :

1. Listening

2. Typing.

3. Due to sounds alike etc etc.

Work on each type of error seperately by identifying the factors responsible for those errors . for example:

Factors for errors due to listening can be :

a) week vocabulary

background noise

c) Skill/ expertise/ no. of years experience of the listener

etc etc..

Once you jot down all the possible errors & its causes ( C&E matrix), try regression or DOE to find out which factors you should work on...

Poka-Yoke wouldn't really be possible since this involves human interface and for this same reason, the medical transcription is being outsorced to us.