Dramatic Reductions In Medication Errors Reported At International Patient Safety Conference

RANCHO MIRAGE, CA, November 9, 1998 -- If you are one of the 30 million Americans admitted to a hospital each year, getting your medicine should become a lot safer once the findings of research reported today at an international patient safety conference in Palm Springs, California, are implemented nationwide. And, computers will play a major role according to the research, which included:

• Replacing hand-written medication orders with a computerized physician order entry system at Brigham and Women's Hospital in Boston that led to an 86 percent reduction in serious medication errors that could cause injury,

• Implementing a new computerized system for detecting potentially dangerous drug combinations at Barnes-Jewish Hospital in St. Louis that resulted in a 66 percent reduction in potentially unsafe combinations with cisapride, a gastrointestinal drug,

• Reconfiguring a machine used by patients to self-administer their pain medication, thereby eliminating 55 percent of the programming mistakes made by hospital staff, according to University of Toronto researchers, and

• Developing a first-of-its-kind computer program that predicts error potential -- or "cognitive toxicity" -- due to medication names sounding or looking alike, the cause of one in four reported medication errors in the U.S.

Still, computers used to catch mistakes may not be mistake-free themselves, as researchers at the Cleveland Clinic Foundation discovered. They used computers with radiation therapy equipment to verify compliance with pre-set parameters and essentially eliminated serious errors. Of remaining minor errors, about 15 percent were related to the safety system itself. That's because some human operators appeared to trust the computer safety system's accuracy more than their own common sense. In some cases, minor data mistakes were allowed to persist in the computer safety system that might have been caught on hand-written data entry forms.

Demonstrating that humans are just as important as computers in making health care safer are the findings of two studies also reported today:

• Placing a clinical pharmacist on an intensive care unit patient care team that resulted in a 77 percent reduction in preventable adverse drug events at Massachusetts General Hospital, and

• Standardizing medication orders, eliminating "look-alike" medications, and revamping medication processes that led to a 50 percent reduction in medication errors at Hermann Hospital in Houston.

Opening yesterday and continuing through tomorrow, the conference -- Enhancing Patient Safety and Reducing Errors in Health Care -- is being held at the Annenberg Center for Health Sciences at Eisenhower Medical Center in Rancho Mirage. It was convened by the American Association for the Advancement of Science, the Annenberg Center for Health Sciences, the Joint Commission on Accreditation of Healthcare Organizations, the National Patient Safety Foundation at the AMA, and the U.S. Department of Veterans Affairs.

Additional support for the conference is coming from the Agency for Health Care Policy and Research, the American Hospital Association, the American Society of Health-System Pharmacists Research and Education Foundation, Pfizer Inc., Pharmacia & Upjohn Inc. and the Robert Wood Johnson Foundation. The conference comes after a pathbreaking patient safety meeting held at the Annenberg Center in October 1996.

The conference closely follows the October 28, 1998, publication by the Journal of the American Medical Association of an editorial calling for new approaches in promoting patient safety. The authors wrote: "To solve a problem, that problem must first be recognized. Unfortunately, error in medicine is real and common. The good news is that we have recognized the problem of error in medicine. The next step has been to go public with the problem, to study it, and to create methods to solve it. Many groups are now addressing this. It will then be necessary to implement methods that are found likely to reduce the probability of error." (http://www.ama-assn.org/sci-pubs/sci-news/1998/snr1028.htm#jed80080)

Additional information on the research findings reported today follows:

• More than 80 Percent of Medication Errors Eliminated by Computerized Physician Order Entry System

  Replacing hard-to-read, sometimes illegible written medication orders with a computerized physician order entry system has enabled Brigham and Women's Hospital in Boston to achieve an 81 percent reduction in medication errors (excluding missed dosages). Non-intercepted serious medication errors -- those with the potential to cause injury -- fell even further, by 86 percent. All patients admitted to three of the hospital's medical units were studied for 7 - 10 week periods in four different years, with the first year serving as a control period prior to the installation of the system in May 1993. The last measurement period occurred as recently as 1997. Initially, the system enabled physicians to write orders on-line, but decision support features were added soon thereafter to suggest appropriate doses and frequencies, and display relevant laboratory data and screen orders for allergies and drug-drug and drug-laboratory interactions. During the study, the number of admissions varied among the study periods from 379 to 492, patient days varied from 1704 to 2619 and medication orders varied from 10,070 to 15,025. The non-missed dose error rate per 1,000 patient days fell from 142 during the baseline period to 26.6 in the final period. The non-intercepted serious medication error rate fell from 7.6 in the baseline period to 1.1 in the final period. (Corresponding Author: David W. Bates, M.D., M.Sc., Brigham and Women's Hospital and Harvard Medical School, Boston, MA; Media Contact: Suzanne O'Brien, 617-732-5008, sobrien@bics.bwh.harvard.edu)

• Potentially Dangerous Drug Interactions Caught by Computerized Safety Net

  Potentially dangerous drug interactions are often identified by computer software screening packages when medication orders are entered, but they are not always prevented. Others may not be identified at all. That is why the pharmacy department at Barnes-Jewish Hospital and researchers at Washington University in St. Louis erected a more comprehensive, computerized safety net that forces a "second look" at 130 potentially dangerous drug interactions. Accounting for 71 percent of the potentially dangerous drug interactions identified were interactions involving cisapride, a gastrointestinal drug. Using daily patient lists with detailed clinical information generated by the system, pharmacists alerted patients' physicians who discontinued at least one of the interacting medications in 92 percent of patients taking cisapride with another contraindicated medication. Following implementation of the system in January 1996, the rate of potentially dangerous drug combinations with cisapride has decreased by 66 percent. Another computerized system at Barnes-Jewish checks orders involving 41 medications for a variety of potential dosage errors, particularly those related to renal dysfunction. The number of potentially inappropriate dosages detected by the system has declined from 13.7 percent of drug orders screened in 1994 to 7.8 percent in 1997, suggesting greater awareness of the importance of adjusting dosages on the basis of renal dysfunction.
  (Corresponding Author: S. Troy McMullin, Pharm.D, BCPS, Barnes-Jewish Hospital, St. Louis, MO; Media Contact: Kim Kitson, 314-286-0300, kbk2591@bjcmail.carenet.org)

• Redesign of Patient Controlled Pain Relief Machine Reduces Programming Mistakes

  With the press of a button, patients can inject their bodies with pain reliever, as needed, using a computerized patient-controlled analgesia machine. But, programmed incorrectly, these machines can be the instruments of a patient's own injury or death. Researchers at the University of Toronto report that setting incorrect drug concentrations when the machines are programmed for specific patients is the most common cause of serious adverse incidents involving a machine they studied. However, in a first, they were able to reduce programming errors by 55 percent under laboratory conditions after redesigning the machine. They used human factors principles such as making the functions of various controls clear and obvious. Other human factors design principles include providing users with prompt, useful feedback after each action, making displayed messages easy to understand and minimizing the load on users' memory. (Corresponding Author: Kim J. Vicente, PhD, University of Toronto, Toronto, Canada; Media Contact: Bruce Rolston, 416-978-6974, bruce.rolston@utoronto.ca)

• New Computer Program Predicts, Prevents Look Alike, Sound Alike Medication Errors

  Everyday experience tells us that when two words sound or look alike, we have a good chance of confusing them. When the names are on medications, getting them confused can cause injury or even death. In fact, one in four reported medication errors in the U.S. is caused by drug names that look or sound too much alike (LASA). But, predicting with a high degree of confidence which drug name combinations are more likely to cause an error has not been possible until now. Researchers at the University of Illinois at Chicago have developed a computer program that compares drug names, assesses their similarity and then predicts the chances the name combination will be involved in a LASA error. They measured degrees of similarity by comparing how many two- or three-letter combinations in a pair of drug names are the same, and counting how many editing steps are needed to change -- or "morph" -- one name into the other. In a first, they were able to show that the more alike two drug names are the more likely an error would result -- a "dose-response" relationship in medical terms. Using the new computer program, the researchers say that the vast majority of medication name-pairs are not candidates for a LASA error. However, the very small fraction that are LASA error candidates -- less than one percent of the 44 million name-pairs possible -- results in nearly 100,000 medication name pairs they consider "cognitively toxic" because of their propensity to cause LASA errors. As a safeguard, they suggest inc orporating their new computer program into computerized systems physicians use for entering medication orders. The Institute for Safe Medication Practices and a variety of regulatory agencies are testing versions of the University of Illinois at Chicago computer program to screen proposed new drug names for confusion potential.
  (Corresponding Author: Bruce L. Lambert, PhD, University of Illinois at Chicago; Media Contact: Jodie Oesterreicher, 312-996-8277, joest@uic.edu)

• Computerized Safety Systems Prevent Major Errors But Can Create Their Own Unique Mistakes

  During 1995, the Cleveland Clinic Foundation gave 1,925 cancer patients more than 93,000 individual radiation treatments using machines that were electronically linked via a "record and verify" computer controlled treatment check system. Treatment commenced only when the radiation therapy machine settings agreed with a pre-programmed set of treatment prescription parameters set for that particular treatment day. Although no serious errors occurred, 59 minor errors were detected upon detailed review of treatment transcripts and self-reported incidents logged by treatment technologists. Nine of these errors -- about 15 percent -- were themselves related to the use of the electronic check system. Most were minor digit transpositions that took place upon data entry into the electronic check system. It appeared that the information programmed into the automated verification system was sometimes given priority over other conflicting data sets despite the fact that in some cases common sense should have suggested that the verification system itself was incorrectly programmed. In some cases, minor mistakes that would probably have been questioned and caught if they appeared on hand-written data entry forms were allowed to persist because they had the "authority" of the computer check system format. Thus, the implementation of electronic treatment coordinate verification and error suppression systems may itself contribute to certain classes of errors. These errors may affect a long series of treatments given over a period of weeks without additional recalculation or mid-course independent validation.
  (Corresponding Author: Roger M. Macklis, M.D., Cleveland Clinic Foundation, Cleveland, OH; Media Contact: Holli Birrer, 216-444-0142, birrerh@cesmtp.ccf.org)

• Putting Pharmacist on Patient Care Team Reduces Medication Errors

  Typically, hospital pharmacists remain in their pharmacies, dispensing medications. However, Massachusetts General Hospital added a clinical pharmacist as a fully functioning member of a medical intensive care unit patient care team when the first phase of the Adverse Drug Event Study revealed that drug information access and management was the most common system failure causing adverse drug events. The result was a dramatic 77 percent reduction in preventable adverse drug events when compared to a similar intensive care unit that did not include a clinical pharmacist on the patient care team. During a nine month period in 1995-1996, the pharmacist team member intervened 398 times to clarify or correct an erroneous proposed or previous medication order, provide drug use information, recommend alternative therapy or identify potential drug interaction and allergy problems. If this reduction in preventable adverse drug events were achieved for all patients admitted to this intensive care unit, the hospital would save $270,000 annually. For all intensive care units in the hospital, the saving would be $1.9 million. The pharmacist cost per intensive care unit is approximately $50,000 annually for half a pharmacist's time, although not all intensive care units will require such an extensive commitment.
  (Corresponding Author: David J. Cullen, M.D., M.Sc., St. Elizabeth's Medical Center of Boston, Boston, MA; Media Contact: Sonya Hagopian, 617-789-2329, shagop@aol.com)

• Houston Hospital Achieves Big Reductions in Medication Errors

  When the performance improvement staff at Hermann Hospital in Houston "diagnosed" adverse events -- those that can lead to death or disability -- they traced two thirds to communication problems and nearly half to not following standard practice or using equipment correctly. They also discovered that mistakes tended to occur on weekends and holidays; when there were more patients in the hospital, and particularly acute or emergency medical conditions were being treated; or when staff members were unfamiliar with their areas. Armed with these findings, Hermann Hospital achieved a 50 percent reduction in errors since August 1996, even as the number of patient admissions increased, by standardizing medication orders, eliminating "look-alike" medications, and revamping medication processes such as ordering, dispensing, administration and monitoring. In August 1998, the six-month cumulative adverse drug event rate per 100,000 patient days stood at just 3.5, down from nearly 9 two years previously.
  (Corresponding Author: Joanne E. Turnbull PhD, Memorial Hermann Healthcare System, Houston, TX; Media Contact: Patty Riddlebarger, 713-704-1600, patty_riddlebarger@mhhs.org)

  ###

  ---

  
  

  ---

  [ | E-mail Article ]