A blood banker at a large mid-western laboratory in a major medical center is curious about how other institutions assure patient identification. The blood banker's institution transfuses more than 35,000 RBCs per year and an equivalent number of non RBC products each year. Currently they are using three points of identification on each patient from the point of specimen collection to transfusion. These points are patient's name, hospital registration number, and birth date. The hospital registration number changes upon each admission. The registration number refers back to a permanent medical record number that is not readily available to the laboratory or the patient care unit. In certain instances, they find discrepancies due to hospital registration number changes, e.g. patient transferred from Emergency Department to inpatient status, they use the patient's Social Security Number (SSN) as a fourth identifier since it appears on the patient's wrist ID band. There have been recent concerns with the use of SSNs due to potential identity theft and confidentiality issues. They have been asked to find an alternative to the use of the SSN. Do other facilities use the SSN in their patient identification process? What points of identification do other facilities use and how do they resolve discrepancies?"

The following replies were submitted in response to the above:

ADDENDA April 11, 2002

1. Editor's note: The following links to previous issues in this forum may be of interest in relationship to the above questions:

• Review calls for improved patient identification systems for blood
• Are two people needed to verify that the correct blood products are issued and transfused?
• Prevention of Mislabelling of Pre-transfusion Compatibility Specimens
• Who should draw blood specimens for compatibility testing?

(Links to the articles on this problem published in Newsday in early April, 2002 are no longer available.)

2. A state laboratory inspector (who works in a different state than the mid-western blood banker) reported that many older and non user-driven laboratory information systems (LIS's) and hospital information systems (HIS's) use patient identifications that, as the blood banker from a large mid-western laboratory pointed out, use a non-unique, non-permanent patient registration number for each hospital stay. This must be considered unacceptable (at least in the inspector's state) as a laboratory ID for any reason, much less for transfusion-related purposes. The laboratory inspector lamented "Why do laboratorians continue to allow hospital information systems personnel and other personnel to dictate patient ID's, and should we address this from regulatory standpoint?"

ADDENDA Feb. 24, 2006

3. A transfusion medicine colleague in Texas reports that her hospital has a flawed system for assigning hospital ID numbers in that from time to time a patient with a pre-existing ID number is issued a new (second, sometimes a third) hospital ID number. When patients are issued more than one ID number it creates a situation where a patient with an unexpected red cell antibody might have their historical antibody documented under one ID number, but not the other. If the patient with a historical antibody (such as anti-Jk(a) is admitted using an ID number that is not associated with that antibody history, it creates a risk of delayed hemolytic transfusion reaction, should the strength of the historical antibody have dropped below the level of detection, and the patient is transfused with RBCs that are positive for the antigen corresponding to the historical antibody. The inquiring Texan would like to know what other colleagues have done to address this problem.

ADDENDA Feb. 25, 2006

4. A transfusion medicine physician in Vermont comments in response to the ADDENDUM of February 24, 2006, that from time to time a patient who is treated at his institution may inadvertently be assigned two (or more) different ID numbers (unbeknownst to the Blood Bank). His technologists tend to pull up records by ID number because it's faster than pulling up records by name. However, they have noticed that if they pull up records by patient last name only, they are more likely to spot a "duplicate" because they will see two listings with an exact date of birth, but slightly different name. (i.e. J. Tiberius Kirk vs. James T. Kirk vs. Tiberius Kirk with the "J" left off). He acknowledges that the aforementioned procedure is only likely to discover patients who have multiple ID numbers if the laboratory does not have too many similar names on the system. For example, a common name like John Smith might not trigger the discovery of a patient with multiple ID numbers. In his opinion, a better solution to hoping that Blood Bank technologists will spot (by chance) patients with multiple ID's, is for the hospital's Patient Registration Department to employ algorithms that prevent such occurrences at the beginning of the process of generating an ID number. For instance, while people's last names and addresses change, their social security number probably does not. This problem affects more areas than just the Blood Bank, including Pharmacy (drug history/allergies) and general patient care (important notes or other information that might be in the system under the different ID number). Consequently, this problem should be addressed by hospitals as a global patient safety issue.

Editor's note: I agree fully with the Vermont physician about the need to prevent the occurrence of patients with multiple ID numbers by using better 'algorithms' at the time an ID number is assigned to each patient. However, even if an institution uses great algorithms to prevent patients from receiving more than one hospital ID number, this problem will still occur, because patients may present to a hospital without any identifying information. Such a patient may be given an ID at the time of emergency admission, only to have it be discovered later that s/he was a patient previously under a different ID number. Consequently, there should also be a mechanism to uncover (discover) patients who have been assigned more than one ID number, and upon discovery to alert the blood bank, pharmacy, and other areas that 'need to know' this information. There should also be a mechanism (when appropriate) to cross reference all of a patient's different hospital ID numbers so that health care providers can have access to a complete set of the patient's information.

ADDENDA Mar. 2, 2006

5. A Laboratory Clinical Coordinator at a Florida hospital reports that they require patients to wear two different ID bands, when it comes time to collect a specimen for pre-transfusion testing; the patient must wear a generic hospital ID band AND a blood bank specific "Secureline^®" wristband. However, upon investigating a recent NEAR MISS event, they discovered that the information appearing on the blood bank specific ID band may not match exactly with the information on the generic hospital ID band. For example, one band might include a patient's middle name (or initial) while the other ID band did not. In such a case, a computer generated test order label might match one but not the other ID band. She wonders if hospitals are strict with regards to requiring that there be 100% matching of all information that appears on any ID bands that a patient might wear, including middle name or initials, when they appear. Do institutions require 'redraws' if a middle name (or initial) is listed in a draw label but not on a both a blood bank specific and a hospital generated ID band?

ADDENDA Mar. 5, 2006

6. A colleague in California reports that her medical center often treats patients who have been issued a new identification number that is different from a historical hospital ID number, in situations where the admitting area is unaware of a prior ID number, either by mistake, or when a patient is treated for trauma and is identified with a "Doe name and number". However, as soon as a pre-existing "true medical record number" is discovered, both the new and historical ID numbers are displayed (cross referenced) in the hospital computer information system and on any requisition slips that are generated for Blood Bank tests/products. At that point, the Blood Bank staff can check their computer history for any prior special patient needs or antibody problems, etc. She comments that their system is good, but not perfect, because it requires human awareness at several levels. Another problem is the timeframe between blood transfusion needs and realization of pre-existing records.

7. Editor's note: The following link to a previous issue in this forum may be of interest in relationship to the above questions:

Protocols for 'identifying' unidentified trauma patients who need emergency transfusion

8. A colleague at a 1000+ bed hospital in the Midwest reports that they use a separate unique blood bank identifier imprinted onto the patient wristband. The patient is tracked through the computer systems by their medical record number. The unique blood bank identifier is transcribed onto tubes drawn for the Blood Bank and the unique identifier is applied to the blood products and paperwork by the Blood Bank. The unique identifier is only present on the patient wristband, creating a "force function" of the transfusionist checking the blood prior to spiking with the patient wristband. It is not a fool-proof system, but has served them well through the years. They are investigating RFID as the ideal patient identification system and continue to ask their administration for financial backing. However, she acknowledges that if a patient is unconscious with no identification at the time of arrival at their hospital, there would be no way to access a historical record (since the patient's previous record number would not yet be known). However, as soon as the patient's identification is known, they could access the patient's blood bank record by name and birthdate. That would bring up any previous history in their Blood bank electronic record, such as the history of an unexpected red cell antibody or any special needs. As far as reconciling 2 records belonging to the same patient, it occurs regularly as they reconcile their John Doe's and Jane Doe's when their identities become known to them. They recently modified their system to include a fictitious middle name such as John Alexander Doe or Jane Gertrude Doe to prevent mixups in the busy emergency setting.

Please submit comments to the e-Network Forum.

Ira A. Shulman, MD
CBBS e-Network Forum Editor & Moderator