Vol. 24 • Issue 7 • Page 18
Errors in the pre-analytical phase of laboratory testing have been reported to be anywhere between 1 percent and 68.2 percent depending on each laboratory’s definition of a specimen error.1-3 However, even at 1 percent, which constitutes the predominant source of laboratory induced challenges to patient safety,1-3 it is not close to the achievable Six Sigma rate of 0.00034 percent. Most laboratory errors occur during the most manually intensive aspect of lab testing, namely at the front end of the process where specimen collection and labeling, specimen transportation, quality inspection, aliquoting and sorting for analysis takes place. As laboratories increase their capacity for testing through the use of automation, it results in increased errors due to the greater burden on the pre-analytical processes. Therefore, there is increasing interest in automating as many pre-analytical steps as possible.
Automation currently exists as mobile robots delivering specimens from the phlebotomy areas to the laboratory, conveyor systems moving specimens from the receiving area to the accessioners and automated racking and sorting workstations placing specimens in racks destined for each analytical area of the laboratory.
When designing automation for specimen quality inspection, one has to take into account the many kinds of mistakes and/or attributes that a medical specimen may have to be considered low quality. Some issues may be operator generated, such as mislabeling, selecting the ðinappropriate vial (e.g., the correct anticoagulants), wrong time of collection, failing to refrigerate labile specimens and excessive hemolysis. Some issues may result from the patient’s health condition, such as ðexcessive lipids or bilirubin (lipemia or icterus). Other issues may result from transportation-induced quality factors, such as not getting the specimen to the laboratory within a necessary analytical time window or excessive accelerations and decelerations in a pneumatic tube system that can damage fragile cells releasing enzymes into the plasma.4
To automate the process of quality control, pre-analytical robotics must have inspection systems that can read and verify label content with both machine printed and human printed information. Fortunately, as more laboratories adopt barcode labeling technology, the demographics of each specimen can be checked in the information system after a simple barcode scan. Two-dimensional bar codes can contain all the patient information obviating the need to check the laboratory database. However, many state regulations require the presence of printed patient information, which necessitates having scanners that can perform optical character recognition in addition to barcode scanning. Each tube must be turned and scanned to obtain all its information.
However, it would be ideal if the label scanner could read the entire circumference of the tube without having to turn it on its long axis to obviate the need for yet another mechanical device – a task that can be achieved with multiple cameras or the use of mirrors. A second sensing system should detect the specimen temperature for those that should have been refrigerated. A third sensing system should be able to detect the volume of blood or other liquid (e.g., cerebral spinal fluid or urine) in the vial. Even more sophisticated sensing could determine the volume of serum or plasma and the quality of these fluids with regard to icterus, hemolysis and lipemia. Since some specimens are labeled multiple times with overlaid labels, the sensing system should be able to perform these quality tasks through these paper layers. Finally, the cap color should be determined to validate the use of the appropriate sample tube.
Some early pre-analytical systems were engineered with quality inspection stations. The first published specimen inspector was filed as a patent in 1996 by Cadell and Samsoondar.5 Their device identified specimen volume, hemoglobin, lipids and bilirubin using light-emitting diodes. It wasn’t until 2006 when the first attempts were made to use machine vision and spectrophotometry together to read bar codes, text and determine the serum/plasma quality factors.6 In 2001, Markin’s developed a fiber optic-based, charge-coupled device to measure hemolysis through specimen labels.7 In the 2000s, pre-analytical processing robots contained some form of quality determination devices (e.g., Tecan FE500,8 Pathfinder (Ai Scientific, Australia), and systems from Beckman Coulter and Olympus). Hawker et al. developed a specimen inspector and quantified its ability to read specimen label text using optical character recognition.
Even with today’s technology, the reader still had a 25 percent error rate compared to human operators when over 1 million reads were performed.9 Much of the error centered on the diversity in labels and ways to print the patient information. Adhering to established specimen label standards (CLSI Standard AUTO12-A) would have significantly reduced the error rate.10
Cost justifying the purchase of automation focused on quality improvement is often ðchallenging since the return on investment for quality can be an ephemeral number. Laboratories who document their errors and their reduction by rectifying the cause can justify the purchase of “front-end” solutions since there is often a direct impact of reducing pre-analytical errors on medical costs. Once an improper specimen is generated, they can take up valuable health worker time by requiring the process be repeated. In addition, delays in the completion of diagnostic testing can lengthen patient hospital stays, upset surgical schedules and require ill patients to travel back to the hospital for a second blood draw. The deeper into the analytical process a poor quality specimen travels causes increased costs for processing time, delays in the analysis of high quality specimens and overuse of valuable diagnostic reagents.
Preanalytical automation will have many unintended benefits, such as reducing operator fatigue and repetitive use injuries. Individuals involved in the accessioning process may perform other, more productive laboratory tasks or be trained to do so. Automation generally takes up less space than manually performed tasks and, thus, laboratories will be able to reduce their overall footprint. When an automation process is engineered into a laboratory, the laboratory can be assured that the quality metrics that the automation will meet will be met consistently for many years without the usual quality erosion seen with many human based Six Sigma efforts. Many of the pre-analytical automation systems installed in the early 2000s are still yielding the same quality performance over a decade later. However, as quality control automation becomes available, these systems will need to be upgraded or replaced.
Reduction in errors and improved quality and patient safety will be the underpinning of a future health system that will be more affordable and accessible to all Americans. Automation will be the means by which the U.S. laboratories will improve their quality, streamline their procedures, reduce their waste and provide better care to those in need.
Dr. Felder is professor of pathology, associate director of clinical chemistry, The University of Virginia, Charlottesville, VA.
1. Plebani M, Carraro P: Mistakes in a stat laboratory: Types and frequency. Clin Chem 1997;43:1348-1351
2. Holman JW, Mifflin TE, Demers LM, Felder RA. Evaluation of an automated preanalytical robotic workstation at two academic health centers. Clin Chem 2002;48:540-8.
3. Markin RS. Challenging the future. Clinical laboratory automation: a paradigm shift. Clin Lab Manage Rev 1993;7:243-51.
4. Felder RA. Automated specimen inspection, quality analysis, and its impact on patient safety: beyond the bar code. Clin Chem. 2014 Mar;60(3):433-4. PMID: 24407911
5. Cadell TE, Samsoondar J. Apparatus and method for rapid spectrophotometric pre-test screen of specimen for a blood analyzer. US Patent 6,195,158 B1, Filed Nov. 21, 1996, issued Feb. 27, 2001.
6. Felder RA. Automation of preanalytical processing and mobile robotics. In: Kost GJ, ed. Handbook of clinical laboratory automation and robotics. New York: Wiley and Sons; 1996. p 252-82.
7. Stickle DF, Moss PS, Markin RS. An approach to automated detection of hemolysis in capped, labeled specimens without sampling using a fiber optics charge-coupled device (CCD) array spectrophotometer. Clin Chem 2001;47:A166.
8. FE500proTM [product brochure]. Tecan. http://www.tecan.com/platform/apps/ product/index.asp?MenuID1195&ID579&Menu1&Item184.108.40.206&OrderBy&ActionLogin&file71,29 (Accessed June 2014).
9. Hawker CD, McCarthy W, Cleveland D, Messinger BL. Invention and validation of an automated camera system that uses optical character recognition to identify patient name mislabeled samples. Clin Chem 2014;60:xxx-xxx.
10. CLSI. Specimen labels: content and location, fonts, and label orientation; approved standard. Wayne (PA): CLSI; 2011. CLSI document AUTO12-A.