What is the "magic number" or percentage of patient records that need to be verified when implementing a new computer system in a blood bank?

A California colleague asks if there is a “magic number” or percentage of patient records that need to be verified in a new system, when validation is performed to check the transfer of data from the previous system to the new one. What has been the experience of others?

ADDENDA Oct. 27, 2007

1. The colleague who initiated this discussion has provided the following three sample responses to a similar question that was posted on another website:
   
   Posting #1: "I think you will find that anything less than 100% would be contested by a certain number of people in the QA Industry. The sampling would have to be quite large to be statistically significant, since the probability of an error would be very low. I'd be more concerned that the important records are accurately transferred: antibodies, special transfusion requirements, transfusion reactions, and the like.
   
   Posting #2: "I just validated a software upgrade. We used 100 donors from 10 different collection dates (10 from each collection date). We selected the date our system first went live, the date after each prior upgrade, and some dates in between, and selected 10 donors from the collection list. We (I) printed off each page of their record, including all visits, memos, deferrals, etc. We compared that to their record after the upgrade. We selected 100 units to compare all records (processing, testing, inventory tracking). The print screens for these comparisons stacked 8 feet high. I pity the poor fool who thinks they are going to find a discrepancy when reviewing our validation, because we didn't."
   
   Posting #3: (note: same author as posting #1 above)- I didn't say that I agreed with checking every record-after the first hundred, your eyes will start playing tricks on you. I've just heard QA-types say "prove to me you don't have to…." I checked all of the “important” ones listed above, plus a sampling of others, then called it a day."

ADDENDA Dec. 3, 2007

2. An IT Systems Compliance Manager in Los Angeles reports that in her opinion it is an industry standard to use ANSI standard Z 1.4, or Military Standard 105E upon which the ANSI standard was based, to determine the minimum number of records based on the total number or records, and the concept of 'Reduced, Normal and Tightened' inspections based on identified risk levels. ANSI standards can be purchased from ASQ, or colleagues may download the Mil-STD-105E HERE.

Submit comments to the e-Network Forum at enetworkforum@cbbsweb.org

Ira A. Shulman, MD
CBBS e-Network Forum Senior Editor & Moderator

W. Tait Stevens, MD
CBBS e-Network Forum Editor & Moderator

Elizabeth M. St. Lezin, MD
CBBS e-Network Forum Associate Editor & Moderator