Patient Safety with Blood Products Administration Using Wireless and Bar-Code Technology

1. Patient Safety with Blood Products Administration Using Wireless and Bar-Code Technology  AHRQ Patient Safety and Health Information Technology Conference June 9, 2005 I appreciate the opportunity today to share the experiences of the University of Iowa Hospitals and Clinics as we implemented new technology to improve patient safety during the blood administration process. This work was funded with a grant from AHRQ, and was the result of a great deal of work by a large team of individuals, including my co-PI on the grant, Dr Loreen Herwaldt, and my associate Kristy WalkerI appreciate the opportunity today to share the experiences of the University of Iowa Hospitals and Clinics as we implemented new technology to improve patient safety during the blood administration process. This work was funded with a grant from AHRQ, and was the result of a great deal of work by a large team of individuals, including my co-PI on the grant, Dr Loreen Herwaldt, and my associate Kristy Walker

2. University of Iowa Hospitals and Clinics 831 Bed Tertiary Care Teaching Facility Annual Statistics: acute inpatient visits: 41,000 ambulatory visits: 845,000 blood product administrations: 43,000 The University of Iowa Hospitals and clinics is a large, academic tertiary care center located in Iowa City, Iowa. A relevant metric associated with this project, we deliver approximately 43,000 blood product administrations annuallyThe University of Iowa Hospitals and clinics is a large, academic tertiary care center located in Iowa City, Iowa. A relevant metric associated with this project, we deliver approximately 43,000 blood product administrations annually

3. Project Goals Use wireless mobile computing device, bar code technology, and a new online tool to Prevent patient identification errors that occur in the blood transfusion process When blood samples are collected When Blood Bank evaluates blood samples When Blood Bank dispenses blood product When blood product is administered Our objective for this project was to leverage a number of planned IT initiatives to attempt to address patient identification errors in the blood transfusion process. We had started to deploy an institutional wireless data network, and we were planning to role out bar coded patient ID bands. We decided that these projects, combined with the capabilities we had with our internally developed clinical information system, provided us with a foundation to provide support for improving the patient identification process and the specimen management process associated with blood product administration . Our objective for this project was to leverage a number of planned IT initiatives to attempt to address patient identification errors in the blood transfusion process. We had started to deploy an institutional wireless data network, and we were planning to role out bar coded patient ID bands. We decided that these projects, combined with the capabilities we had with our internally developed clinical information system, provided us with a foundation to provide support for improving the patient identification process and the specimen management process associated with blood product administration .

4. Expectations Based upon our extensive analysis of errors we experienced in 2003 relating to blood product administration, we felt our new bar-coded processes offered us the potential to reduce our error rate by over 73%. We also felt that removing these sources of error offered us the opportunity to reduce a number of other undesirable consequences, such as needing to re-collect samples from patients when a mismatch occurs.Based upon our extensive analysis of errors we experienced in 2003 relating to blood product administration, we felt our new bar-coded processes offered us the potential to reduce our error rate by over 73%. We also felt that removing these sources of error offered us the opportunity to reduce a number of other undesirable consequences, such as needing to re-collect samples from patients when a mismatch occurs.

5. Actual Experience Expectations surpassed Overall sample rejection decreased from average 1.82% in the manual system to 0.17% in the automated system. An Improvement of 91% The automated system rejection percentage reflects the time period of Feb 7 to April 21, 2005. The percentage from previous years is the existing average and is not time period-specific. I am happy to report that our actual results exceeded our original goals. Sample rejection rates in our new system, when compared to our previous manual process for the same time periods, indicate we have lowered our rejection rate to 0.17%, an improvement over our manual system of 91%I am happy to report that our actual results exceeded our original goals. Sample rejection rates in our new system, when compared to our previous manual process for the same time periods, indicate we have lowered our rejection rate to 0.17%, an improvement over our manual system of 91%

6. Project Time Line Awarded AHRQ Grant Sept 2003 Developed Initial Online Tracking Function Dec 2003 Completed Replacement Patient ID Bands Feb 2004 Progressive Pilot with Dual Process Apr 2004 – Dec 2004 Enabled Wireless Network House-wide Dec 2004 Trained 1,800 Clinicians Jan 2005 House-wide Implementation Feb 7, 2005 Dropped Dual Process Feb 14, 2005 Our project timeline began in Sept of 2003 when we learned we had been awarded our AHRQ implementation grant. By Dec of 2003 we had completed the initial programming needed to support this project. We completed a house wide conversion to the new bar coded wristbands by Feb 04, we were in position to start a series of pilots on different units by April 04, which we concluded by Dec of 04. During the pilot phase, we maintained the 2 person identification process while we were using the new bar code ID process, to ensure patient safety. At the same time we completed the installation and testing of our wireless data infrastructure in all clinical areas, and we began training of all remaining units in January, aiming for a housewide implementation in Feb. On Feb 7, we went live housewide, still maintaining the manual 2 person identification process for one more week, while we continued to collect data on the process and confirmed that the process we were moving to was safer then the standard 2 person identification process. By Monday, Feb 14, we were confident in the safety of the new process, and we dropped the dual process.Our project timeline began in Sept of 2003 when we learned we had been awarded our AHRQ implementation grant. By Dec of 2003 we had completed the initial programming needed to support this project. We completed a house wide conversion to the new bar coded wristbands by Feb 04, we were in position to start a series of pilots on different units by April 04, which we concluded by Dec of 04. During the pilot phase, we maintained the 2 person identification process while we were using the new bar code ID process, to ensure patient safety. At the same time we completed the installation and testing of our wireless data infrastructure in all clinical areas, and we began training of all remaining units in January, aiming for a housewide implementation in Feb. On Feb 7, we went live housewide, still maintaining the manual 2 person identification process for one more week, while we continued to collect data on the process and confirmed that the process we were moving to was safer then the standard 2 person identification process. By Monday, Feb 14, we were confident in the safety of the new process, and we dropped the dual process.

7. Patient ID Bands Here is a picture showing what our previous patient ID wristbands looked like, and what our new, bar coded wristbands look like. As you can see., not only do the new bands provide a bar code, but they also offer a far more legible text for the clinicianHere is a picture showing what our previous patient ID wristbands looked like, and what our new, bar coded wristbands look like. As you can see., not only do the new bands provide a bar code, but they also offer a far more legible text for the clinician

8. Patient ID Wristband Issues Computer generates bar-coded labels for wristband, requisitions, other needs Should the wristband label be unique ? Should we limit the number of labels printed ? What number should be bar-coded on the label ? The move to a new patient ID band was not without its challenges. Besides the expected challenges of selecting a label that could retain image quality when wet, did not present any allergy or sensitivity risks, and could be printed using our existing installed base of label printers, we had to decide if we should configure the wristband so it would have a unique code that would tell the computer system it was a wristband ID, or if it would be a standard label that we would use for a variety of purposes. If the wristband was unique, it would allow us to program checks that would prevent staff from scanning non-wristbands for a workaround to avoid going to the patients bedside. But, it would great operational challenges that staff would have to follow a different process to printout a wristband label vs any other bar coded ID label. We decided to have a non-unique label for the wristband, but we still debate that issue. We had to decide if we should limit the number of labels someone could print in one step. If we printed out too many labels, we might make it too easy for staff to develop workarounds to going to the patient for identification, and we might also experience problems with unused labels from one patient becoming mixed in with labels from another patient. In the end, we decided to program the application to let staff choose how many labels they print out at any one time. We also had to decide what number to bar code on the ID – we initially were confident that we should use the MRN, but after we started down that path and we began looking at EMAR systems, we found that many of the leading EMAR vendors did not use MRN on wristband for bar coding.The move to a new patient ID band was not without its challenges. Besides the expected challenges of selecting a label that could retain image quality when wet, did not present any allergy or sensitivity risks, and could be printed using our existing installed base of label printers, we had to decide if we should configure the wristband so it would have a unique code that would tell the computer system it was a wristband ID, or if it would be a standard label that we would use for a variety of purposes. If the wristband was unique, it would allow us to program checks that would prevent staff from scanning non-wristbands for a workaround to avoid going to the patients bedside. But, it would great operational challenges that staff would have to follow a different process to printout a wristband label vs any other bar coded ID label. We decided to have a non-unique label for the wristband, but we still debate that issue. We had to decide if we should limit the number of labels someone could print in one step. If we printed out too many labels, we might make it too easy for staff to develop workarounds to going to the patient for identification, and we might also experience problems with unused labels from one patient becoming mixed in with labels from another patient. In the end, we decided to program the application to let staff choose how many labels they print out at any one time. We also had to decide what number to bar code on the ID – we initially were confident that we should use the MRN, but after we started down that path and we began looking at EMAR systems, we found that many of the leading EMAR vendors did not use MRN on wristband for bar coding.

9. Mobile Computing and Wireless Issues Troubleshooting wireless infrastructure Computer cart / scanner selection Cleaning of Computer Equipment Different scanner requirements Rapid I.T. support Our deployment of cart-based mobile computers and our move to using our wireless data infrastructure present a number of issues. Although we had tested all the access points in our wireless system, we occasionally found areas with poor coverage, and we had access points with manufacturing problems that failed shortly after installation. Because wireless coverage exists in 3 dimensions, troubleshooting problems was often difficult because we would find that the closest access point had failed, but users were able to connect via an access point on the floor above or the floor below, often without knowing it. We found, however, that after an initial shakedown period the architecture proved reliable and quite functional. Computer cart selection was, and continues to be, an issue. Our initial cart selection was widely supported by Nursing and Pathology, but after deployment of a number of these units, we discovered that the locking drawers on the cart could not be easily interchanged with the drawers our inpatient medications were deliver in – this would not impact the blood product project, but it would create operational problems in the future when we went to deploy our eMAR. In addition, we have found that different configurations were needed for some ICU areas, for the Operating Rooms, and for other unique clinical areas. In addition, the reliance on these technologies have required us to develop better processes for requesting and receiving immediate technical support – it is not sufficient to have clinical staff call the IT Help Desk.Our deployment of cart-based mobile computers and our move to using our wireless data infrastructure present a number of issues. Although we had tested all the access points in our wireless system, we occasionally found areas with poor coverage, and we had access points with manufacturing problems that failed shortly after installation. Because wireless coverage exists in 3 dimensions, troubleshooting problems was often difficult because we would find that the closest access point had failed, but users were able to connect via an access point on the floor above or the floor below, often without knowing it. We found, however, that after an initial shakedown period the architecture proved reliable and quite functional. Computer cart selection was, and continues to be, an issue. Our initial cart selection was widely supported by Nursing and Pathology, but after deployment of a number of these units, we discovered that the locking drawers on the cart could not be easily interchanged with the drawers our inpatient medications were deliver in – this would not impact the blood product project, but it would create operational problems in the future when we went to deploy our eMAR. In addition, we have found that different configurations were needed for some ICU areas, for the Operating Rooms, and for other unique clinical areas. In addition, the reliance on these technologies have required us to develop better processes for requesting and receiving immediate technical support – it is not sufficient to have clinical staff call the IT Help Desk.

10. Monitoring On-line function: “Transfusion: Blood Product History” authorized user may select at any time to review date/time and user involved in all blood transfusion tracking events for a given patient or all events for all patients for a specified time period. Daily review of any aborted incidents The internally developed application was programmed to allow us authorize users to review all activity associated with administration of blood, and project staff were assigned to monitor this information daily during the pilot, and during the early weeks of housewide deployment. Any time unexpected events were identified, a team met to review and assess the situation, and, if necessary, make needed modifications to the process. Staff continue to review this data on a regular basis, and follow-up on any resulting incidents.The internally developed application was programmed to allow us authorize users to review all activity associated with administration of blood, and project staff were assigned to monitor this information daily during the pilot, and during the early weeks of housewide deployment. Any time unexpected events were identified, a team met to review and assess the situation, and, if necessary, make needed modifications to the process. Staff continue to review this data on a regular basis, and follow-up on any resulting incidents.

11. Identification Errors Caught During 8-Month Pilot Data collected during the 8 month pilot supports the objective that not only did the bar code process catch identification errors that the manual 2 person process missed, but it caught these errors earlier in the blood administration process.Data collected during the 8 month pilot supports the objective that not only did the bar code process catch identification errors that the manual 2 person process missed, but it caught these errors earlier in the blood administration process.

12. Comparison of Manual to Bar Code System Post House wide Implementation Comparing scanner-detected prevented errors to voluntarily-reported incident reports from same reporting period in previous year (2/7 – 4/21 in 2003, 2004) Relative risk of finding a misidentification event ( 95% confidence) - At SAMPLE PROCESSING increased 10x (RR=9.98 (2.9 – 34.5 )) Total activity = 6,953 - In ANY STEP OF THE PROCESS increased 30x ( RR=30.6 (9.5 – 98.4)) Total Activity = 22,569 We estimate we are 30 x more likely to catch an identification error using our bar code system compared to our previous manual system Reviewing the data post go-live house-wide shows that, when compared to the same reporting period in previous years using the manual 2 person process, the new system is 10 times more likely to catch a misidentification event during sample processing, and 30 times more likely to catch an identification error in any step of the process. ****************** Any one involved in Healthcare IT knows some of the toughest critics of an IT implementation are the Nursing Staff. I would now like to share with you insights on our project from our lead Nurse. Reviewing the data post go-live house-wide shows that, when compared to the same reporting period in previous years using the manual 2 person process, the new system is 10 times more likely to catch a misidentification event during sample processing, and 30 times more likely to catch an identification error in any step of the process. ****************** Any one involved in Healthcare IT knows some of the toughest critics of an IT implementation are the Nursing Staff. I would now like to share with you insights on our project from our lead Nurse.

14. On-Going Challenges Work-arounds Study why Work to reduce the temptation Establish repercussion standards Critical fore-runner to use of bar-code technology in medication administration checks  As we become more familiar with the technology, we are able to spend more time looking are other critical issues relating to this project. We are actively looking at the issue of work-arounds, trying to identify why staff try to work around the system, and trying to modify the process to make it more convenient to follow the process then to circumvent the process. We are also trying to define appropriate responses for staff when it is discovered they tried to short-cut the system. In addition, we are very excited that the experiences we have acquired during this project have better positioned us for the challenges we will encounter during the implementation of a bar-coded electronic medication administration record. And, our clinical staff are feeling far more confident with the technology, and, as a result of their experiences with this project, are actually anxious to begin using the emar tool.As we become more familiar with the technology, we are able to spend more time looking are other critical issues relating to this project. We are actively looking at the issue of work-arounds, trying to identify why staff try to work around the system, and trying to modify the process to make it more convenient to follow the process then to circumvent the process. We are also trying to define appropriate responses for staff when it is discovered they tried to short-cut the system. In addition, we are very excited that the experiences we have acquired during this project have better positioned us for the challenges we will encounter during the implementation of a bar-coded electronic medication administration record. And, our clinical staff are feeling far more confident with the technology, and, as a result of their experiences with this project, are actually anxious to begin using the emar tool.

15. Thank you for your attention In conclusion, I would like to thank AHRQ for providing capital dollars to help support us on this project. Recently, the physician director of our laboratories commented to me that he has never in his career been involved in a project with federal financial support that has generated such positive results in such a short period of time. Again, I thank AHRQ for helping to make that happen.This concludes my remarks.I thank you for your attention today.In conclusion, I would like to thank AHRQ for providing capital dollars to help support us on this project. Recently, the physician director of our laboratories commented to me that he has never in his career been involved in a project with federal financial support that has generated such positive results in such a short period of time. Again, I thank AHRQ for helping to make that happen.This concludes my remarks.I thank you for your attention today.

16. Implementation Challenges Inaccessible wristbands in OR OR rapid infusion (‘blood bath’) situations Separate equipment requirement for perfusionists Introduction of new network, hardware and software together Dual process until rollout complete Idiosyncratic workflow in same day surgery Surprise barcode symbology Duel scanners Management of labels

17. 2. Sample Arrival in Labperformed by Blood Bank personal at the time the sample and requisition are received in the Blood Bank And here is the screen the staff access when they receive the sample in the blood bank.And here is the screen the staff access when they receive the sample in the blood bank.

18. 3. Blood Product Dispensedperformed by the Blood Bank at the time the blood product is dispensed from the Blood Bank Here is the screen the staff access when they dispense the blood from the Blood BankHere is the screen the staff access when they dispense the blood from the Blood Bank

19. And here is the screen staff access when they are administering the blood product to the patient.And here is the screen staff access when they are administering the blood product to the patient.

20. 1. Sample Collection performed by a phlebotomist or nurse at the time the specimen is collected at the patient’s bedside/point-of-care Here you can see the application as we scan for patient identification when a sample is acquired.Here you can see the application as we scan for patient identification when a sample is acquired.

21. Identification Error History -Comparison of Manual to Bar Code System Pilot If we compare the data obtained during the pilot with our historical data, we feel we can state with statistical confidence that the new bar code enabled process is 3 to 10 times more likely to catch an identification error then our previous, 2 person manual system. This data, combined with the very positive feedback from our Nursing staff and the Blood Bank staff, convinced us we had a system in place that we could use without relying on the manual, 2 person identification protocol.If we compare the data obtained during the pilot with our historical data, we feel we can state with statistical confidence that the new bar code enabled process is 3 to 10 times more likely to catch an identification error then our previous, 2 person manual system. This data, combined with the very positive feedback from our Nursing staff and the Blood Bank staff, convinced us we had a system in place that we could use without relying on the manual, 2 person identification protocol.

22. System Caught Identification Errors Since House-Wide Go-Live Since we have gone house-wide, we have identified an additional 43 errors that the system has identifiedSince we have gone house-wide, we have identified an additional 43 errors that the system has identified