The effect of handwritten vs computer-generated labels on the rate of misidentified samples received for compatibility testing

A colleague in Florida reports that nurses at her hospital would like to switch from handwritten to computer-generated labels for blood samples intended for compatibility testing. Their current policy prohibits the use of computer-generated labels and requires hand-labeling of pre-transfusion blood samples. The Nursing Department believes that by changing to computer-generated labels they will see a reduction in the rate of sample labeling errors that result in delays and recollections. The blood bank believes that this change will increase the possibility of misidentified samples.

What has been the experience with using computer-generated labels versus handwritten labels at institutions that have used both approaches? Of those colleagues who do allow computer-generated labels, what safeguards are in place to assure proper identification?

The following responses have been received.

ADDENDA Oct. 25, 2003

1. A colleague in Southern California reports that one of the things they did to help reduce labeling errors when printing computer generated labels for pre-transfusion samples was to ORDER custom label stock with yellow highlight in the area where the patient name and medical record number print. This strategy highlights the most critical information that needs to be checked against the armband, making it more visible and readable. The responding colleague's institution also requires a confirmatory ABO/Rh from a separate phlebotomy on all patients whose initial blood grouping result is not group O and do not have a previous ABO/Rh on file. Since the policy of requiring verification of a patient's ABO by a separate sample went into effect in 1996, they reporting having received approximately 59,000 blood samples for pretransfusion testing, and discovered that 7 of these samples were mislabeled and contained the WRONG BLOOD IN THE TUBE. Thus, without ABO verification by a separate sample, they would have a risk of at least 1 in 8429 of using a mislabeled sample for pretransfusion testing in their institution (where they have had a blood label CQI team in place for over 5 years). This represents a much higher risk than that of most of the infectious diseases. The responding colleague firmly believes that confirmatory blood typing for non-group O patients should be made mandatory. It guards against both a labeling error and a laboratory testing error. Many times it is not necessary to redraw the patient, as there are samples available in the other Laboratory sections that were drawn at a different time. The cost of this program is low and the added measure of patient safety is high.
   
   
2. The Editor believes the information at: A network dialogue on the question of using a second blood sample to verify a patients ABO/Rh type is worth reading, in light of the comments made by the colleague from Southern California in posting #1 above.

ADDENDA Oct. 26, 2003

3. A second colleague in Southern California reports that they use computer generated labels for pretransfusion blood samples, but require that the patient's birthdate be handwritten onto the label, using the patient's armband as the source of the birthdate information. The blood bank personnel are supposed to routinely check the handwritten birthdate against the birthdate information in the laboratory computer before testing any specimen. It is the institution's belief that this safeguard should ensure that if a wrong person is phlebotomized or a wrong label is attached to a specimen tube, the birthdate double-check by the laboratory will catch such a labeling error. However, the responding colleague points out that from time to time the patient's birthdate is handwritten on the specimen label away from the bedside, usually at a nurse's station, and this defeats the entire purpose of the safeguard. The responding colleague believes that if their labeling procedure is followed to the letter, the chance of a labeling error slipping through their safety net is remote, however, the failure of humans to follow the procedure as written increases the risk that an error will not be caught before work is done on the specimen.
   
   
4. A colleague in Vellore, India reports that at his hospital, samples for pretransfusion testing are labeled with handwritten labels that include the patient's hospital number and the date. Overwriting/corrections are not permitted. Labeling mistakes, when they occur, depend on the rank and experience of staff. When 'fresh' interns come on service, they notice a slight increase in specimen mislabeling that tapers off. The following policy is in use at his hospital blood bank to limit erroneous ABO/Rh results: For new patients, initial ABO/Rh determination is performed on the same sample, but by two different people, one person uses a tube method and the other person uses a column agglutination method. The results of these two determinations are compared for concordance. If a subsequent sample is submitted for the same patient, the ABO/Rh is determined only once and that result is compared against historical records. Whenever there is a discrepancy (infrequent), a fresh sample is requested, to resolve the discordance.

ADDENDA Oct. 27, 2003

5. A colleague at the Mayo Clinic reports that they used to require handwritten labels in addition to barcoded labels generated by their laboratory information system for all pretransfusion samples. The thought was that the need to write the labels would force the phlebotomist to look at the wristband for patient identification. However, people could simply copy the information from the computer label or the requisition and they still did not eliminate identification errors. In the end they decided to take the approach that correct patient identification was critical for all samples since a patient can be harmed with the wrong potassium result as well as a wrong transfusion. Therefore, they standardized to a single identification process using Laboratory Information System (LIS)-generated labels and instituted an extensive continuous improvement effort which included non-punitive reporting and group discussion of all errors (not just those involving pre-transfusion samples). As a result, they report that their error rate was significantly reduced (currently <1 per 20,000 processes). Their future plan is to provide the phlebotomist with wireless access to the LIS so that they can go to the patient's bedside, use the patient's barcoded wristband for identification, print the tube labels at the bedside and collect the sample. The Mayo Clinic colleague believes that the key to their success was the standardization of the ID process, better knowledge of the incidence and in particular the consequences of misidentifications, and open discussion. However, virtually all of their samples are collected by their own phlebotomy team - this approach may not be generalizable to the nursing practice. In that circumstance, it is recommended to coordinate efforts with initiatives on medication errors since patient ID is a critical component of that as well. And no process will be successful if the appropriate wristband is not on the patient, which has also been a focus of improvement efforts at the Mayo Clinic over the last 2 years.

ADDENDA Oct. 28, 2003

6. A second Mayo Clinic colleague provides additional information on patient identification systems at the Mayo Clinic. At the 2003 AABB meeting in San Diego their group is presenting a poster on "Blood Orders for Wrong Patients" which deals with changes in their required patient identification information on the order form (now electronic) which led to an increase in such errors; (AP 113, Transfusion ,2003; 43, page 167A ). They also have the full paper to appear in Mayo Clinic Proceedings, probably in November 2003.

ADDENDA Dec. 19, 2003

7. The article from the Mayo Clinic Proceedings (referred to in 6. above) is now available.

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