Improving patient safety by reducing sample identification errors

The 1999 US Institute of Medicine report "To Err is Human: Building a Safer Health Care System" provided an influential wake-up call for healthcare workers concerning the frequency of preventable medical errors. Initiatives over the past decade aimed at improving the safety of patient-sample testing represent just one aspect of the welcome legacy of that report. 

A recent paper describes one such initiative at the Emergency Department (ED) of John Hopkins Hospital in Baltimore that sought to reduce the incidence of patient-specimen labeling errors. In May 2008 a new system, which combined electronic test ordering with barcode specimen labeling, was implemented in the department. 

This new system was presumed to be less prone to error than the manual specimen ordering and labeling process that it was replacing. The paper provides the results of a study aimed at verifying this presumption. The total study period was 5 years, and comprised a pre-implementation period from September 2004 through April 2008, and a post-implementation period, May 2008 through September 2009. 

A little over 1 million patient samples were collected in the ED of John Hopkins during this study period. Of the 724,465 specimens collected during the period before implementation of the new system, 3007 (0.42 %) involved an error in sample labeling or patient identification. 

This compared with 379 (0.11 %) of the 334,039 samples collected after implementation of the new system. The authors of the study were able to conclude that implementation of the new electronic/barcode system had resulted in a 74 % relative reduction and 0.31 % absolute reduction in the number of sample-labeling/patient-identification errors.

The success of the new system depended crucially on barcode technology being a reliable means of patient identification. As another separate study has revealed, barcode technology is not infallible. The authors of this second US study were alerted to possible barcode-reading errors when several point-of-care test results at their hospital had not been transmitted to the correct patient records. 

On the basis of the results of their study to investigate possible barcode-reading anomalies, the authors estimate that in 1 in 84,000 barcode-scanning events generate an incorrect patient identifier. This is a significantly higher error rate than was previously supposed. The fallibility of barcode technology and significance of that fallibility for patient safety is discussed by the authors, and is the subject of an accompanying editorial.