EDN 818 Module 7 Six Sigma DMAIC Project Example

Forty-Two Wrong or Missing Labels: A Six Sigma DMAIC Project to Reduce Specimen Labeling Errors

Student Name

Doctor of Education Program, Aspen University

EDN 818: Innovation and Technology in Health Care

Instructor Name

Month Day, Year

Forty-Two Wrong or Missing Labels: A Six Sigma DMAIC Project to Reduce Specimen Labeling Errors

Last year Kingsmere Regional Hospital's laboratory rejected or corrected 42 specimens because they were mislabeled or unlabeled, out of about 186,000 inpatient and emergency specimens collected by nursing and phlebotomy staff. Each defect meant a repeat collection, a delayed result or, in the worst case, a result attached to the wrong patient. One mislabeled blood bank specimen was caught only at the final bedside check before a transfusion. This paper describes a Six Sigma project, using the DMAIC method of define, measure, analyze, improve and control, that set out to reduce those defects.

Six Sigma and Its Measures

Six Sigma is an improvement approach that relies on statistical measurement to make a process more consistent and less prone to error. It measures performance as defects per million opportunities and converts that figure to a sigma level, with 3.4 defects per million conventionally described as six sigma performance. Where Lean focuses on waste and flow, Six Sigma focuses on consistency and error, and many organizations combine the two.

Define

The project charter defined a defect as any specimen received without a label, with a label for the wrong patient, or with a label missing one of the two required patient identifiers. The customer requirement, or critical-to-quality characteristic, was a correctly labeled specimen at the moment it leaves the bedside. The scope covered inpatient units and the emergency department, and the team included two nurses, two phlebotomists, a laboratory supervisor, an information systems analyst and a Six Sigma black belt from the quality department.

Measure

At baseline, 42 defects in 186,000 specimens is about 226 defects per million opportunities, or roughly 5.0 sigma using the conventional 1.5 sigma shift. That is better than the national picture from a study of 147 laboratories reviewing more than 3.3 million labels, where errors occurred at 0.92 per 1,000 labels, about 920 per million (Wagar et al., 2008). But a comparison with the average offers little comfort when a single error can lead to a transfusion reaction. The team set a target of no more than 10 defects a year, about 54 per million, or roughly 5.4 sigma.

Analyze

A Pareto analysis sorted the 42 defects by circumstance. Seventeen involved labels printed before collection and applied later, often to several tubes from different patients carried together; 11 occurred in the emergency department during periods of high volume; 8 involved labels with only one identifier; and 6 had other causes. The pattern matched national findings. In an analysis of 227 root cause analysis reports in the Veterans Health Administration, Dunn and Moga (2010) found that mislabeling during collection was linked to batching of specimens with printed labels and to failures of two-source patient identification. Three units at Kingsmere lacked bedside label printers, so nurses printed labels at the station before going to the room.

The DMAIC Project at a Glance

The table summarizes the project's phases.

Improve

The improvements targeted the causes directly. The hospital installed bedside label printers on the three units that lacked them and configured the collection workflow so that no label will print until the patient's wristband barcode has been scanned at the bedside, which makes preprinting impossible. A written rule prohibits carrying specimens from more than one patient at a time. In the emergency department, phlebotomy coverage was extended into the evening peak, and the label layout was changed to show two identifiers in large print. The changes were piloted on the three units for six weeks, adjusted after staff feedback about printer placement, and then extended hospital-wide.

Control

To hold the gains, the laboratory plots monthly defects on a control chart, with a reaction plan that triggers a review within one week of any defect and a team meeting if two occur in a month. Nurse managers receive their units' defect counts monthly. Printer downtime is tracked, since a broken printer tempts staff back to preprinting. In the first six months after full implementation, the hospital recorded 4 defects in about 93,000 specimens, about 43 per million, consistent with the annual target.

Limits of the Evidence

The project's results should be read with care. Six months is short for a rare event, and a few defects either way change the rate substantially. The wider evidence on Six Sigma in health care is also thinner than its popularity suggests: DelliFraine et al. (2010) identified 177 articles; only 34 reported any outcomes, and statistical testing appeared in less than a third of those 34. Kingsmere will continue to measure for at least two years before claiming success and will report the results with their uncertainty.

Leadership Lessons

Three lessons stand out. The most effective improvement was a forcing function, printing only after a wristband scan, rather than more training. The Pareto analysis focused effort on the causes that mattered instead of spreading it across every possible cause. And the project depended on cooperation among nursing, phlebotomy, the laboratory and information systems, which only leadership sponsorship could secure.

The Human Side of a Statistical Method

Six Sigma's language of defects and sigma levels can make it sound impersonal, but the project depended on people. Nurses on the pilot units initially resisted bedside printing because the printers slowed them down in crowded rooms; the team moved printers to the room entrance after two weeks of feedback. Phlebotomists suggested the large-print identifiers. And the laboratory staff who had quietly corrected mislabeled specimens for years were asked to stop fixing errors downstream and report them instead, which made the problem visible enough to solve.

Conclusion

Kingsmere's specimen labeling errors were already less frequent than the national average, but each one carried a risk of serious harm. A Six Sigma DMAIC project defined the defect precisely, measured it as about 226 per million, traced most errors to preprinted labels and batching, and replaced that practice with bedside printing triggered by a wristband scan. Early results suggest a rate near the target of 54 per million. Continued measurement will show whether the process holds, which is exactly what the control phase exists to do.

References

DelliFraine, J. L., Langabeer, J. R., II, & Nembhard, I. M. (2010). Assessing the evidence of Six Sigma and Lean in the health care industry. Quality Management in Health Care, 19(3), 211-225. https://doi.org/10.1097/QMH.0b013e3181eb140e

Dunn, E. J., & Moga, P. J. (2010). Patient misidentification in laboratory medicine: A qualitative analysis of 227 root cause analysis reports in the Veterans Health Administration. Archives of Pathology & Laboratory Medicine, 134(2), 244-255. https://doi.org/10.5858/134.2.244

Wagar, E. A., Stankovic, A. K., Raab, S., Nakhleh, R. E., & Walsh, M. K. (2008). Specimen labeling errors: A Q-Probes analysis of 147 clinical laboratories. Archives of Pathology & Laboratory Medicine, 132(10), 1617-1622. https://doi.org/10.5858/2008-132-1617-SLEAQA