Paper to computerized T.S. and validation

[amblanki]

Okay, I need some help/guidance/advice.  We are currently a paper Transfusion Service going to Cerner Millennium.  I apparently need to do my own validation.  I have printed the FDA guidance and started looking at it, but I have questions.  I understand the general layout of making a plan, but how detailed does the plan have to be and how much do you test it?  I know that you have to try to "break it" with things like interpretations not matching testing results, etc, but how much is enough.  Do the SOP's for procedures need to be written before validation?  We dont' have a migration from an old software program specific to blood bank (because we were paper), but our eMR will be used to import patient data across the system to which I will be manually entering blood bank information( antibodies, blood types, etc).  What sort of layout would you use for verifying information is all entered correctly?  I'm really in a small panic mode as I have this on top of benchwork (gen lab too) and all of the other supervisory stuff that goes on a day to day basis.  Oh, and, no overtime allowed.  Thanks in advance. Hope it makes sense.

[Dr. Pepper]

1. I would bet Cerner has a validation plan. Start with that. Document your checks (print a lot of screen shots). You will need this documentation for inspection purposes.

2. Order and result all the tests you've built, every way you can, with all patient types. Look at all the reports and inquiries and make sure they look right. Make sure the proper billing goes across to Fiscal.

3. By all means "break it" every way you can. This is the only way to test the accuracy of the background truth tables that drive the interpretations and QC aspects of your system.

4. I don't think SOPs are necessary for validation, but they sure are for staff training. Cop them from Cerner and individualize them, or, as you learn the system doing your validation, prepare them yourself. Your staff will love you if you have prompt by prompt instructions. My LIS manual is task-oriented ("Entering units into inventory", "resulting ABO/Rh typings", "how to find an old specimen" etc).

5. Manually entering all your old data and results into your new BB history file, may God have mercy on your soul! We did an electronic transfer of data from old LIS to new and audited 100 entries for accuracy (Meditech said audit it all - a couple of million pieces of data. Fat chance.) With manual entry, however, you might want to do a 100% audit. Lots more room for data entry error. Can you get secretarial help for this chore? Documentation of the accuracy audit could be as simple as a check mark, initial and date on whatever the hard copy was that was used to enter the data.

6. Having been through this three times, I realize that it's a mammoth, daunting undertaking. It's all there, though, to do the job well and to ensure patient safety. There are great dangers in not thoroughly validating. Your superiors should be aware of this if you are not given adequate time or resources to do so. Let it be on their heads. Prioritize if you must - there may be functionality not totally necessary for routine work that you can introduce after you go live and the dust has settled.

 

Best of luck - Phil

[tbostock]

Phil's response is excellent advice. I always start with their test scripts, then make some of my own based on my specific workflow. Do normal, abnormal results as well as really bizarre scenarios that would never normally happen.

This will require an extensive amount of time, you will never find the time, you will have to make it (meaning everything else is pushed to the side, unfortunately). It will all be worth it in the end, you will know Cerner inside and out and feel comfortable about going live.

For manual entry of historical data, have a process that a second person verifies the accuracy of what you entered...before go-live. Double work, but well worth the effort.