FOUND IN THE LITERATURE
Barcode Scanning Errors, a Threat to Patient Safety
Snyder ML, Carter AB, Jenkins KJ, Fantz CR. Patient Mis-identification Caused by Errors in Standard Barcode Technology. Clin Chem. 2010 Oct; 56(10):1554–60.
By Nancy Sasavage, PhD
Editor, Clinical Laboratory News, AACC
We are pleased to welcome Corinne Fantz, PhD, to the Patient Safety Focus Editorial Board. Dr. Fantz is associate professor of pathology and laboratory medicine at Emory University and the director of point-of-care testing at Emory Medical Laboratories in Atlanta, Ga. She and her colleagues recently published a study on how the laboratory can take steps to avoid barcode errors on patient samples.
“Clearly, barcodes play a vital role in many facets of healthcare today, from pharmacy and nutrition, to inventory and POCT. With the increases in healthcare workload forcing more providers to use barcode technology, the errors discovered here and their implications strongly suggest that the healthcare industry needs to move to more reliable and safer barcode technology. Although Code 128 is the current laboratory standard, it is certainly not the safest and most reliable symbology in use today. There are 2D-barcodes, including matrix type barcodes, that contain more sophisticated algorithms for error correction and redundant data embedded within the barcode. In fact, in some cases up to 30% of the barcode can be destroyed and an accurate scan can still be obtained. Interestingly, these symbologies are in widespread use in a number of industries, including banks, the pharmaceutical industry, and mail delivery services. For the safety of our patients, healthcare systems should consider transitioning toward similar, more reliable identification technology.”
—Corinne Fantz, PhD
Hospitals and other healthcare providers are relying more than ever before on barcodes for identification and tracking of patients, medications, orders, and equipment. Compared with manual data-entry error rates of one error per 300 entry events, the practice of using barcodes to rapidly link patient information to the laboratory information system is thought to be essentially error proof. Despite the improved efficiency and accuracy of barcode-based identification, a recent report in the October 2010 issue ofClinical Chemistry demonstrates that linear barcode identification methods are not fail-safe and misidentification of patient specimens can still occur.
Over a 1-year period, researchers at Emory University identified 10 patient wristband barcodes that generated incorrect patient identification numbers when scanned for glucose point-of-care testing (POCT). With approximately 840,000 glucose POCTs performed annually in their institution, the authors estimated an error rate of at least one barcode misidentification error observed per 84,000 patient wristbands scanned. This finding was surprising given that the generally accepted error rate for the laboratory barcode standard, Code 128, is thought to be 15–440-fold lower. Overall, the researchers identified four major sources of the observed errors: 1) subtle printing defects in the barcodes; 2) suboptimal wristband barcode orientation; 3) lack of adequate error detection in Code 128 barcode symbology; and 4) failure to control for specified scanner resolution requirements.
Four Sources of Barcode Errors
- Printing defects in the barcode
- Suboptimal barcode orientation on wrist bands
- Lack of error detection
- Problems adhering to scanner resolution requirements
The study is important because it documents how failure to adhere to the scanner manufacturer’s resolution requirements led to barcode misread errors. Even after the researchers reprinted the barcodes in “pristine” condition to control for the printer error, they still observed substitution error rates as high as 13%. This finding emphasizes the importance of ensuring that the widths of and spaces in the bars meet the manufacturer’s specifications and are linked to the resolution capacity of the scanner. To confirm this observation, the researchers reprinted the bar codes in a size that was compatible for the scanners tested, and no errors were observed.
Code 128, like the majority of barcode symbologies in use, is encoded with built-in integrity check systems that are designed specifically to prevent substitution errors. Barcode scanners are programmed to use the symbology-specific algorithm to decode the check digit from the barcode data and compare it to a check character encoded in the barcode. The scanner should only generate an identifier when the calculated and encoded check characters match. But in the study, the incorrect patient identifiers coincidentally generated the same check characters as were encoded in the barcode. Clearly, the inability of the Code 128 integrity check system to detect such substitution errors is a major drawback of the current laboratory barcode standard.
This paper not only provides insight into limitations of linear barcode technology, but also offers laboratories some helpful strategies either for implementing barcode systems or improving the accuracy of existing barcode systems (Figure 1). In order to minimize the risk posed by relatively common barcode and print-head malfunctions, the authors rotated the orientation of the barcodes by 90 degrees, so that the barcode bars and spaces were printed perpendicular to the printing axis. With this change, relatively common printer errors that produce a white streak running across the width of the bar code were easier to detect, and the integrity of the barcode was preserved. When the barcode orientation cannot be changed, the authors recommend printing a black bar perpendicular to the barcode so that printer failures are easy to recognize. The latter solution, however, still depends on visual recognition of a white streak running through the printed black line, making it still possible to place flawed barcodes into service. For high-volume healthcare applications in which printer failures are likely to occur more frequently, the institution may want to consider installing barcode verifiers to check the barcodes before they are put into service.
Error Detection Solutions
Arrows pointing to white spaces indicate a printer failure. The example on the left reveals the error by printing a thick black line below the barcode. This is useful when orientation of the barcode relative to the print media cannot be changed. The barcode on the right has been rotated 90º so that the resulting printer defect is perpendicular to the lines and printer failures can be detected without compromising the integrity of the barcode.
The bottom line is that barcode scanners are not a one-size-fits-all solution. Scanners also have varying resolution requirements, and laboratorians should make sure that barcodes and scanners are compatible based on the scanners’ specifications. Some scanner models in this study were more error-prone than others, but all of them generated substitution or rejection errors. Scanners that are more likely to reject the barcode are obviously preferred, because these errors pose minimal threat to patient safety. Algorithms built into middleware applications may also be useful in identifying and/or preventing the transmission of incorrect data. When possible, labs should restrict scanners to read-only for one barcode symbology.