A.C.TAccelerating

Change Today

F O R A M E R I C A ’ S H E A L T H

F E B R U A R Y 2 0 0 0

Reducing Medical Errorsand Improving Patient Safety

T H E N A T I O N A L C O A L I T I O N O N H E A L T H C A R ET H E I N S T I T U T E F O R H E A L T H C A R E I M P R O V E M E N T

Success Stories from theFront Lines of Medicine

THE NATIONAL COALITION ON HEALTH CARE

NCHC is the nation’s most broadly representative alliance working to improve America’shealth and health care. It is comprised of 90 member organizations. They include someof the nation’s largest businesses, labor unions, health care providers, consumers groupsand religious organizations. The Coalition was founded in 1990. It is non-profit andnon-partisan. Its members are united in the belief that America needs better, moreaffordable health care and that all Americans should have health insurance. FormerPresidents George Bush, Jimmy Carter and Gerald R. Ford serve as the Coalition’sHonorary Co-Chairs. Former Iowa Governor Robert D. Ray and former CongressmanPaul G. Rogers of Florida are the Coalition’s Co-Chairmen. NCHC is in Washington, DC.Founder and President Henry E. Simmons, M.D., M.P.H., F.A.C.P., is a widely respectedpioneer in the field of health quality assessment and improvement.

THE INSTITUTE FOR HEALTHCARE IMPROVEMENT

IHI is an independent, non-profit education and research organization based in Boston,MA. It was founded in 1991 with the goal of fostering collaboration among health careorganizations to improve the quality of health care. IHI each year holds a wide array ofeducational forums, symposia and workshops, and demonstration projects for medicalprofessionals and health care administrators. IHI’s co-founder and President, Donald M.Berwick, M.D., M.P.P., a practicing pediatrician and clinical professor at Harvard MedicalSchool, is one of the nation’s leading authorities on health care quality.

Accelerating Change Today (A.C.T.)F O R A M E R I C A ’ S H E A L T H

A.C.T. is a collaborative initiative of the National Coalition on

Health Care and the Institute for Healthcare Improvement. It

aims to improve the quality of health care in the United States

through the identification of “best practices” and administrative

and clinical innovations that are: (1) yielding better patients out-

comes; (2) making the delivery of care more efficient; (3) increas-

ing access to timely medical care; (4) making the health system

easier to use; (5) lowering costs, and (6) reducing medical errors

and inappropriate care. The initiative seeks to accelerate the

spread of best practices and innovations throughout the health

system by publishing them and through presentations at medical

meetings and health care and business symposia. A central pur-

pose is to make a broad range of health care stakeholders,

including consumers and those who pay the health care bill,

more aware of cutting-edge efforts to improve the quality of

health care. The initiative will actively encourage the replication

of best practices in health care facilities.

About This Publication

This report presents the stories of

institutions and organizations that

made a commitment to change and

innovation in the field of patient

safety and medical error reduction.

The profiles herein reflect some of

the most pioneering and innovative

efforts underway in this field. The

health facilities and organizations

discussed in this report were identi-

fied by a team of six experts (see

credits and acknowledgements).

Their choice is in no way meant to

imply that other institutions and

health care facilities are not under-

taking meaningful and laudable

efforts to reduce errors and

enhance patient safety.

Profiles ofinstitutions andorganizationsthat madea commitmentto change…and a difference

A.C.TAccelerating

Change Today

F O R A M E R I C A ’ S H E A L T H

F E B R U A R Y 2 0 0 0

Reducing Medical Errorsand Improving Patient Safety

T H E N A T I O N A L C O A L I T I O N O N H E A L T H C A R E

T H E I N S T I T U T E F O R H E A L T H C A R E I M P R O V E M E N T

LUCIAN L. LEAPE, M.D., (Team Leader) is an Adjunct Professor of HealthPolicy at the Harvard School of Public Health and a member of the Health SciencesDivision at RAND. At RAND, Dr. Leape has directed several studies of overuse andunderuse of cardiovascular procedures and the development of medical practiceguidelines. He is a member of the Board of Directors of the National Patient SafetyFoundation and the Committee on Quality of Care in America at the Institute ofMedicine. In the past, he has served on the Health Services Research ReviewCommittee of the Agency for Healthcare Research and Quality. Dr. Leape’s researchhas focused on medical error prevention, quality of care, unnecessary surgery, andthe development of practice guidelines. He recently led the Institute for HealthcareImprovement’s initiative in prevention of adverse drug events.

DAVID BATES, M.D., is Chief of the Division of General Medicine at Brighamand Women’s Hospital, Medical Director of Clinical and Quality Analysis for Partner’sHealthcare Systems and an associate professor of medicine at Harvard MedicalSchool. Trained as a clinical epidemiologist, Dr. Bates’ main interest and area ofresearch is the use of computer systems to improve patient care. He has also doneextensive work on evaluating the incidence and prevention of adverse drug events.

DONALD M. BERWICK, M.D., M.P.P., is President and CEO of theInstitute for Healthcare Improvement in Boston, MA. He is a practicing pediatrician,a clinical professor of pediatrics and health care policy at Harvard Medical School,and an associate professor of health policy and management at the Harvard Schoolof Public Health. Dr. Berwick currently serves as Chairman of the National AdvisoryCouncil at the Agency for Healthcare Research and Quality. He was one of 32appointed members in 1997-98 to the President’s Advisory Commission on ConsumerProtection and Quality in the Health Care Industry. He is a member of the Committeeon Quality of Care in America at the Institute of Medicine. Dr. Berwick is a co-authorof Curing Health Care: New Strategies for Quality Improvement (Jossey-Bass, 1990).

MARILYN SUE BOGNER, PH.D., is President and Chief Scientist at theInstitute for the Study of Medical Error in Bethesda, MD. She is Chair of theInternational Workgroup on Medical Error and an advisor to the National PatientSafety Foundation of the American Medical Association. She has served as a con-sultant to numerous groups and worked with organizations including the Instituteof Medicine of the National Academy of Sciences and the Food and DrugAdministration. She edited and contributed to the book Human Error in Medicine(Lawrence Erlbaum Associates, Inc., 1994).

STEVEN MEISEL, PHARM.D., is Assistant Director of Pharmacy for PatientCare at Fairview Southdale Hospital in Edina, MN. He is Co-Chair of Fairview’sIntegrated Formulary & Drug Use Committee and serves as the Chair of the Institutefor Healthcare Improvement’s Breakthrough Series on Adverse Drug Events andMedical Errors. Prior to his affiliation with Fairview, Dr. Meisel was ClinicalPharmacy Coordinator at St. Joseph’s Hospital in St. Paul, MN, and Chief ofPharmacy, Keams Canyon Indian Hospital in Keams Canyon, AZ. Dr. Meisel is amember of the American Society of Health System Pharmacists, and is a contribut-ing editor for the Institute for Safe Medication Practices’ Medication Safety Alert.

ELLISON C. PIERCE, JR., M.D., is Executive Director of the AnesthesiaPatient Safety Foundation in Boston, MA. Dr. Pierce founded the group in 1985 andserved as its president until 1996. He is an executive committee member of theNational Patient Safety Foundation. From 1968 to 1996 he was Chairman of theDepartment of Anesthesia at New England Deaconess Hospital in Boston. Prior tothat, from 1960 to 1968, he was the Vice Chairman of the Department of Anesthesiaat the Brigham Hospital in Boston. He has been an adjunct professor of anaesthesi-ology at Harvard Medical School from 1960 to the present.

EDITORSteven Findlay

ASSISTANT EDITORAlyssa Keefe

WRITERSCaralee Adams, Lisa Krieger,Lani Luciano, Christine Paul,Kris Rebillot, Shelley Reese,Jane Roessner, Jan Ziegler

DESIGNPensaré Design Group, LTD

This report was funded by a grantfrom The W.K. Kellogg Foundationof Battle Creek, Michigan.

© 2000 by the National Coalitionon Health Care and the Institutefor Healthcare Improvement

Credits and Acknowledgements

Table of Contents

Can We Make Health Care Safe?By Lucian L. Leape, M.D. 2

Back from the Brink: Making Chemotherapy SaferThe Dana-Farber Cancer Institute 4

A Government Health System Leads the WayThe VA Health System 9

Making Doctors Computer LiterateBrigham and Women’s Hospital 12

What’s in a Name? When it Comes to Drugs—LotsBristol-Myers Squibb 16

Tackling Medication Errors Head OnLuther-Midelfort Hospital 18

Making Mistakes Harder to MakeFairview Health Services 21

A Dummy Helps Pave the Way to Better MedicineVA/Stanford Simulation Center 24

A Medical Specialty Blazes a TrailAnesthesiology was first in addressing medical errors 26

Resources and Suggested Reading29

1

MEDICAL ERRORS—the very termsounds ominous. And well it should. Whenyou enter a hospital or a nursing home, orgo to a doctor’s office, you don’t alwaysexpect to be cured, but you certainly don’texpect to be hurt! Yet, when the presti-gious Institute of Medicine (IOM) inNovember 1999 released the landmarkreport, To Err Is Human: Building a SaferHealth System, many Americans wereshocked. The report contained this sober-ing, if not new, statistic: each year morethan one million people in the U.S. sufferfrom preventable medical injuries and100,000 die from them. As the media wide-ly reported at the time, that’s more thandied in 1998 from breast cancer, AIDS andmotor vehicle accidents combined.

The IOM panel, on which I served,called on health care organizations, doctors’ groups, regulators, governmentagencies and Congress to make patientsafety a priority. Within weeks, Congresslaunched hearings, President Clinton held a Rose Garden ceremony to appointa task force, and a host of other groupslaunched initiatives. Health insurer Aetnasaid it would spend an extra $1 millionthis year on patient safety research. TheAmerican Hospital Association said itwould join forces with the Institute forSafe Medication Practices to provide hos-pitals with proven strategies to reduceerrors. The Joint Commission on Accredi-tation of Healthcare Organizations (JCAHO)announced it will begin making randominspections at hospitals to probe errors.And a business group led by, among oth-ers, General Motors, said it would pres-sure hospitals and doctors it contractswith to reduce errors.

This new momentum is profoundlywelcome. After all, numerous past warn-ings, based on research results, went rela-tively unheeded. A critical mass appearsnow to have been reached. Indeed, it maynot be hyperbole to say that a deep cultur-al shift on this issue is underway. At its

core, this shift means redefining how toaccomplish one of the basic tenets of med-ical practice: “First, do no harm.”

The vital first task is to recognize that,as in other complex fields of science, technology, engineering, and human en-deavor, errors in medicine are caused by failures in the systems and organizationsthat human beings build.

The evidence is overwhelming. Medicalerrors most often result from a complexinterplay of multiple factors. Only rarelyare they due to the carelessness or mis-conduct of single individuals. Yet, in thepast, rather than addressing those under-lying system design faults, error preven-tion strategies have relied almost exclu-sively on enhancing the carefulness ofcaregivers, heavily reinforced by fear ofpunishment for failure. But punishmentdrives reporting of errors underground,preventing the very systems examinationthat is needed to discover and correct theunderlying causes. As the stories that fol-low show, when punishment is removed,reporting of errors sharply increases—often by 10 or even 20-fold.

Discontented, even horrified, by thestatus quo, a small band of health careleaders and organizations began to coa-lesce in the mid-90s around the missionof dramatically reducing medical errors.Their persistence and willingness to trynew solutions has borne fruit, createdmodels that have reduced some types ofmedical errors by 80 to 90 percent in justa couple of years, and yielded manyimportant lessons.

The first is that leadership is essential.In its absence, efforts will be fragmentary,uncoordinated, and have only minoreffects. The leaders of the institutions pro-filed in this publication understand that.They made highly visible commitments tomake their institutions safer.

Second, change is difficult. Even withthe most enlightened leadership, creating a non-punitive atmosphere is a major

challenge. The urge to punish is deeplyentrenched. And it is reinforced by the sin-cere belief of many doctors and nurses thaterrors result from individual carelessness.If there is any lesson to be learned fromsafe industries, it is that fear, reprisal, andpunishment produce not safety but defen-siveness, secrecy and personal anguish.

Third, health care professionals andstaff are anxious to deliver safe care.When shown effective methods forchanging systems to make them safer,doctors, nurses, and pharmacists eagerlyand energetically embrace them.

Fourth, the problem of medical errorsis not due primarily to a lack of knowl-edge. Rather, the chief culprit is in-adequate dissemination and implemen-tation of ideas and practices we know work. Recently, for example, the NationalPatient Safety Partnership (NPSP) releaseda list of 16 proven and accepted best prac-tices in medication safety. Many havebeen known for years. Yet, most are not inplace in the vast majority of hospitals, andno hospital in the country uses them all.

Fifth, safety pays. A single, preventableadverse drug event in a teaching hospitalhas been estimated to cost over $4,000. Aserious injury can cost hundreds of thou-sands of dollars. Reducing injury rates byeven 10 to 15 percent can yield large-scalesavings. Most systems-level changes forsafety require few extra resources. Andeven those that do require a significantup-front investment produce a full pay-back rapidly. For example, the mostexpensive system to allow doctors to writetheir orders at computer terminals mayrequire an initial investment of $2 millionto $3 million. But the proven ability ofsuch a system to reduce medication errorsby 50 to 80 percent within a year or twomeans most hospitals recoup the invest-ment in a few years.

Finally, there are significant externalbarriers to improving patient safety.Hospital efforts to restructure theirapproaches to error prevention may beoverruled by state boards of nursing,pharmacy or medicine, departments ofhealth, or the courts. Reason: these insti-tutions are still locked into a blame andpunish approach to errors and a focus onindividual culpability. In the name ofaccountability, a nurse loses her licensebecause of a medication error or a doctor

2

BY LUCIAN L . LEAPE, M.D.

Can we make health care

SAFE?SAFE?

is sued for an error-induced injury. Thefear of malpractice litigation thusbecomes a major barrier to openly dis-cussing or reporting errors.

WHAT WE NEED TO DO

Reducing the risk of error in health carewill require a substantial and sustainedeffort at all levels of the health care sys-tem. It must become a priority goal wher-ever care is given—the doctor’s office, thehospital, and the nursing home. That goalmust be supported by the commitment ofboth human and financial resources. Thechief executive officer—or other leader—must articulate the vision of safe care thathe or she calls upon others to work toward.

Leadership is also needed to redefineaccountability. Some have been concernedthat not punishing people for makingerrors relieves them of responsibility.Actually, the converse is true. In a safetyoriented organization, every employeefeels and accepts a personal responsibilityto identify unsafe conditions, report them,and work toward their elimination.

Health care organizations must holdworkers accountable to perform theirduties carefully, conscientiously, compe-tently, and safely. But they will stillmake errors. Who is responsible? Theorganizations and systems. Who isresponsible for the systems? Managers.Consider a nurse who makes a seriousmedication dose error. If one of the fac-tors leading her to make the error is thatshe is working a second shift, or has adoubled patient load, or is inadequatelytrained for her responsibilities, whoseresponsibility is that?

Of course, if a doctor or nurse hasinjured a patient through an error causedby egregious misconduct, neglect, orcriminal activity, he or she must be pun-ished. But if such a person had a priorhistory of reckless behavior and disregardof safe practices, why has he or she beenpermitted to continue working?

Simply put, management must “man-age” for patient safety just as they man-age for efficiency and profit maximiza-tion. And safety must become part ofwhat a hospital or health care organiza-tion prides itself on.

But accountability needs also to workin reverse. Regulators and accrediting bod-

ies should be held accountable to the pub-lic and the professions to set and enforcesafety standards. The plain fact is thatmany hospitals do not initiate error reduc-tion programs because no one demands it.Inertia prevails. If, for example, the JCAHOor the Health Care Financing Administra-tion (which runs Medicare) were to requirefull implementation of the NPSP safe med-ication practices for accreditation or pay-ment, I’m quite sure nearly 100% compli-ance would occur within a short period of time. Likewise, if the Food and DrugAdministration required pharmaceuticalcompanies to label medications in a stan-dardized, readable manner, in bar-codedpackages, and if they only approved newdrug names that would not be confusedwith other medicines, medication errorswould decline substantially.

Safe industries don’t shy away fromstandards that reduce the risk of injury.They embrace them, recognizing the onlyway to achieve safe practice is to clearlyspell out what it is and require it. We havestrict rules governing working hours, pro-cedures, equipment, and maintenance forour airlines and railroads. But we havefew such rules in health care. The need isobvious, and opportunities abound: stan-dards for maximum working hours, workloads, staffing ratios, expertise (pharma-cy, intensivists, ER specialists, etc.), safemedication practices, and many others.

The IOM panel called for the Agencyfor Healthcare Research and Quality to

establish a Center for Patient Safety. Oneof the key functions of such a Centerwould be to convene groups to set suchstandards.

Finally, regulators and accreditingbodies need to become much more posi-tive forces for safety in health care, bothin terms of preventing errors and in the way they react to them. When mis-takes occur, they need to shift their focusfrom individuals to organizations, requir-ing both a thorough investigation of theunderlying systems failures and a planthat addresses those failures.

The publication you hold in yourhands presents case studies of hospitals,health facilities and organizations thathave made the commitment to systemati-cally reduce the risk of patient harm inhealth care. Their stories show the dra-matic results that can be achieved, oftenin a short period of time. The profiles alsoshed light on the obstacles that get in the way of reducing errors, the mecha-nisms by which change occurs most effi-ciently in organizations, and the veryhuman process of shifting the prevailingparadigm on this issue. The stories willenlighten you. But more than that, I hopethey will motivate you to act—to pushyour institutions to get involved in thestruggle to make sure that health care,first and foremost, does no harm.

The author is an adjunct professor of healthpolicy at the Harvard School of Public Health.

3

• Congress should create a federal Center for Patient Safety that sets national goalsfor medical error reduction and tracks progress in meeting those goals.

• Congress should authorize the creation of a nationwide mandatory medical errorreporting system to collect data on errors that result in death or serious harm to patients.

• Federal and state laws should encourage the development of voluntary medicalerror reporting systems in all health care facilities, to include reporting of all eventsthat could have harmed patients.

• Data and information on medical errors should be legally protected when used byprofessional peer review organizations to improve health care quality.

• Medical professional societies and health care licensing bodies should focus greaterattention on patient safety.

• The Food and Drug Administration should increase attention to the safe use of drugs.

• Health care facilities and organizations should make continually improved patientsafety a declared and serious aim and implement proven medication safety practices.

T H E I N S T I T U T E O F M E D I C I N E ’ SK E Y R E C O M M E N D A T I O N S

Every medical facility’s worst nightmarebecame a reality for Boston’s prestigiousDana-Farber Cancer Institute in the earlywinter months of 1995. Two patientsbeing treated in an experimental drugstudy for advanced breast cancer, theInstitute belatedly discovered, hadreceived massive overdoses of chemother-apy in November 1994. On December 3,one patient had died as a direct result ofthe error. The other suffered permanentheart damage. Complicating a tragic situ-ation for all, the patient who died—BetsyLehman, 39—had been a medical reporterfor the Boston Globe. The paper trumpet-ed the death when it learned that a med-ical error was the cause. And it dug intoevery crevice to ferret out what had hap-pened and affix blame. The case quicklyattracted national media attention andwas followed closely for months. Indeed,the Dana-Farber incident became thehighest profile medical mistake in the U.S.in decades, perhaps ever.

Stephen Sallan, M.D., a senior oncolo-gist at Dana-Farber when the overdosesoccurred and now chief of staff, recallsthat difficult time: “It shocked and sad-dened us in very profound ways, bothprofessional and personal. For months, apall hung over the Institute. Everyonewas deeply affected, because after all, weare caring and conscientious people in thebusiness of saving lives.”

As in the wake of all tragedies, lifegoes on for the survivors—but it is also

changed forever. Reeling from the badpublicity and glare of the spotlight, Dana-Farber undertook a painstaking internalexamination of what went wrong. Stateauthorities also probed the case. The JointCommission on Accreditation of Health-care Organizations (JCAHO) placed theInstitute on temporary probation. Fiveyears later the investigations and lawsuitsare only beginning to wind down.

But Dana-Farber didn’t stop at justuncovering what went wrong. Seeking towrest from the tragedy something of last-ing value not only for Dana-Farber butfor health care facilities and patientseverywhere, the Institute’s leadershipboldly committed themselves to makingDana-Farber a model of patient safetyand error prevention.

Today, many experts in the field thinkthey have succeeded. Dana-Farber haspioneered new systems to prevent mis-takes in the delivery of chemotherapyagents—systems that are being emulatedby others. And in making error preven-tion a top priority throughout, Dana-Farber has raised the profile—in a positiveway—of the very issue that brought thefacility to its knees just a few short yearsago. “We wanted to find and fix the rootcauses of our mistakes. But we also want-ed to make our systems the best in theworld at preventing errors while provid-ing the latest in cancer care,” says JamesConway, M.B.A., Dana-Farber’s chiefoperations officer.

4

BY CHRISTINE PAUL

Dana-Farber Cancer InstituteBoston, Massachusetts

Stunned by

the death of a

patient because

of an error,

a prestigious

cancer center

commits to a bold

new approach to

error prevention

Making Chemotherapy Safer

BACK FROMTHE BRINKBACK FROMTHE BRINK

High-tech safeguardsLike many hospitals that have traced the“root cause” of a fatal medical error,Dana-Farber’s doctors, nurses and admin-istrators discovered that behind thehuman error that caused the death were ahost of “systems errors.” The saga beganwhen Betsy Lehman and MaureenBateman, 52, were admitted on November14 and 16, respectively, for treatment ofrecurrent breast cancer. The physician incharge of their care was fully aware of theprotocol of the clinical study under whichthey were to be treated. It was designed totest whether cimetidine, a common anti-ulcer drug, could boost the tumor-killingeffects of cyclophosphamide, a powerfulanticancer drug. The drug therapy was tobe administered over four days, followedby a bone marrow transplant to restorethe patients’ blood-forming and immunecells, which unavoidably are destroyed byhigh doses of chemotherapy.

Unfortunately, the doctor wrote thedrug’s total four-day dose to be givendaily for four days. The backup system inplace at the time was not designed to

catch such an error. Nurses involved withthe patients’ care were able to check theorder against the protocol to ensure thatthe right drug, dosage, frequency, routeand time were being followed per thedoctor’s written order. In this case, how-ever, a protocol written and approved byresearch scientists, rather than the clini-cians administering it, seemed to confirmthat the high dosage level was appropri-ate. “The fact that physicians and scien-tists think of dose in terms of course oftherapy, and nurses and administratorsthink in terms of daily dose, played a sig-nificant role in the overdose,” saysConway. At the time, there was no com-puter system in place to monitor theamount of the drug ordered—and signal apossible error. As a result, both womenreceived four times the daily dose. It wasa toxic blow to their hearts. Lehman diedthree weeks later. Bateman survived, butwith a damaged heart, and died of cancerseveral years later.

Dana-Farber’s first corrective actionwas to put in place a computerized systemfor drug and chemotherapy orders. It alsolaunched a series of backup systems andset in motion a many-pronged effort tostrengthen its “team approach” to treat-ment and care delivery. Finally, it sought

to create an institutional culture thatencouraged the reporting of errors, nomatter how small.

When you walk into Dana-Farber’sclinical areas today, you’re very likely tosee something you would not have seenjust a few short years ago: a physician at a computer terminal. In many hospi-tals and clinics, computers are, of course,indispensable for administrative purpos-es. But doctors and nurses still ordertreatments and make care decisionsbased on “hard copy”—written patientcharts and other paper-based forms.Dana-Farber is changing all that.

“We basically asked which pieces ofour system could be taken out of humanhands and given to a computer with built-in safeguards,” says Lawrence Shulman,M.D. Shulman oversees Dana-Farber’sclinical service for adult patients, forwhich drug ordering is now completelycomputerized. (The hospital’s pediatricservice is reviewing a transition to a sim-ilar system.) “As for the components thatremain in human hands, we designed acomprehensive system of double andtriple cross-checks,” he added.

Gone are slips of paper, handwrittenscribbles and lined paper that can easilyhide a critical decimal point. Instead, foradult cancer patients the doctors must fillout an electronic form. After typing inthe patient’s name, weight, and height(critical to proper dosing), the physicianindicates the intended drug, dose, andnumber of days it is to be dispensed. Thecomputer compares this data with a pre-programmed set of drug-specific guide-lines with established high-dose limitsand route of administration. Dana-Farber’s staff developed these guidelinesby searching the scientific literature anddrawing on their own expertise. The pri-mary focus is on the “five rights”: givethe right drug to the right patient at theright dose through the right route at theright time. If the doctor makes a mistake—for example, typing in 300 milligramsinstead of 30 milligrams—the computersignals a possible error. The only way tooverride a dosage that exceeds the pre-established limit is for a senior physicianto provide a reason for the deviation.Such changes are then double-checked bynursing and pharmacy staff. The systemhandles more than anticancer drugs. It

5

A pharmacist rechecks the drug order with a technician as part of Dana-Farber’sdrug order entry system. (Photo by Steven Gilbert, StudioFlex Productions)

also automatically signals for auxiliarytreatments, such as anti-nausea drugs, tobe included in the patient’s treatment.

Dana-Farber’s customized computersystem cost $1.7 million. It’s an integra-tion of two computer checking systems—a pharmacy checking system developedat Dana-Farber and an order-entry sys-tem adopted from Brigham and Women’sHospital, a sister institution (See storypage 12). Since 1996, through a jointventure titled Dana-Farber/PartnersCancerCare, three premier Boston insti-

tutions—Dana-Farber, Brigham andWomen’s, and Massachusetts GeneralHospital—provide collaborative servicesin adult oncology. Today, the majority ofcare provided at Dana-Farber consists ofchemotherapy done in an outpatient set-ting. Surgery, radiation therapy, andother procedures requiring hospitaliza-tion are done at its sister hospitals. Tofurther integrate the drug order-entrysystem, an even more sophisticated com-puter system is now being phased ingradually at all three institutions.

Double andtriple checksWhen it’s time to dispense medicines, the checking continues. Before ordering adrug from the pharmacy, a nurse reentersthe patient’s weight and height into thecomputer. If they differ by more than 5percent, the computer returns the orderto the senior physician for reverificationand signature. The discrepancy may becaused by a measuring or reporting error,or it may be due to a change in thepatient’s weight, which is not uncommonin cancer therapy.

At the pharmacy, there’s anothercomputerized review—for drug-drug,drug-food, or drug-allergy reactions. Thedrug order is checked against thepatient’s medical history, includingallergies, current lab results, verificationof registration in a clinical study, andthe protocol itself—anything that mightsignal possible error.

Once okayed and prepared by thepharmacy, the drug is then delivered tothe nursing station. Two nurses check thedrug’s label, as well as the patient’s wrist-band, to ensure it’s going to the rightpatient. For experimental therapies,there’s a triple-check. The patient is thenclosely monitored to detect and respondto any adverse reactions such as weak-ness and nausea—which often becomeeven more pronounced with higher dosesof chemotherapy. Betsy Lehman, forexample, had voiced complaints. But suchside effects were not unexpected. And inthe absence of the comprehensive check-ing system that is now in place, thephysician’s error in writing the orderwent undetected. In fact, her overdosewasn’t discovered until three months laterwhen an assistant data manager wasrecording research statistics.

Cultural shiftThat lapse got Dana-Farber’s doctors andadministrators to ask some bold ques-tions. Were nurses, pharmacists, and tech-nicians reluctant to question a physician’sorders or treatment plan? Did the profes-sional culture encourage their input? Was

6

Dana-Farber’s staff and patients as well as academic experts in medical-error preventionare often asked to share their insights at local, regional and national conferences. Beloware key messages directed to health care stakeholders.

PAT I E N T S“The best thing you can do is let your health care providers know when something isn’tgoing well and how they might help. Once it’s on their radar screen, they can start deal-ing with it.”

Judie Beard, cancer survivor, member of Patient-Family Advisory Council

P H Y S I C I A N S“We need to acknowledge that we’re all human beings and that error is always a possi-bility. We need to work together to create an environment and culture that minimizes thelikelihood of error.”

Lawrence Shulman, M.D., Vice Chair of Clinical Services, Adult Oncology

N U R S E S“Listen to the users of the system—the nurses, doctors, pharmacists and patients—andencourage them to identify ways to simplify and standardize the system.”

Susan Grant, R.N., Chief of Nursing and Patient Care Services

P H A R M A C I S T S“Don’t just focus on the individual person or individual departments. Instead, work togeth-er to look at the whole patient-care system.”

Sylvia Bartel, R.Ph., Director of Pharmacy

R I S K M A N A G E R S“Promote a non-punitive environment for reporting error. It encourages the reporters toidentify system problems and assists in preventing errors.”

Maureen Connor, R.N., Director of Risk Management

E X E C U T I V E O F F I C E R S“The effort to prevent medical errors can only be successful if executive leaders makepatient safety one of their priorities.”

James Conway, M.B.A., Chief Operations Officer

T R U S T E E S“The function of Dana-Farber’s board of trustees has changed in fundamental ways. Theynow see error prevention and quality control as central to their role.”

Richard Bohmer, M.B.Ch.B., M.P.H., Senior Lecturer, Harvard Business School

A W A K E - U P C A L L

everyone working together to ensure thatcare was both optimal and safe? The dis-couraging answer was that communica-tion between these various health profes-sionals around the issue of medical errorsand patient safety was less than optimal.That led Dana-Farber to commit to nur-turing a “team approach” to the preven-tion of errors similar to that emphasizedin, say, doing an operation.

To that end, Dana-Farber set up aPharmacy Performance ImprovementTeam. The team meets twice a month andis now part of the 15-member Pharmacyand Therapeutics Committee. Says SylviaBartel, Dana-Farber’s pharmacy director,“Each of us can freely question any aspectof a proposed treatment plan.” Beyondthat, management has emphasized that allpersonnel should feel free to raise ques-tions if they see anything that doesn’tlook right. Physicians are also exhorted towork with nurses and pharmacists to con-tinuously review treatment decisions.And as new cancer treatment protocolsare developed, they are evaluated by amultidisciplinary team which puts safetyhigh on its criteria list.

This questioning spirit is especiallyimportant in shaping the guidelines forexperimental therapies, which often carryhigher risks. Certain patients, whose dis-ease progression affords them no otheroptions, are willing to receive new drugs,drugs at new strengths, or in novel com-binations with other medications or pro-cedures. Betsy Lehman, for example, hadvolunteered to be part of a breast cancerstudy involving high doses of one drugwhose powers were enhanced by another.

The team’s focus is on anticipatingproblems and preventing errors in the firstplace, not just trying to fix things afterthey occur. Pooling their expertise oftenleads to unanticipated solutions. “I couldnever go back to our old clinical practicemodel,” says Bartel. “Sometimes I wonderwhy we didn’t think of this before; itmakes so much sense. Then I realize it’sbecause we didn’t have to think about it;the culture wasn’t there to support it.”

Patients are also considered “teammembers.” “The more informed the patientis, the safer we all are—the patient and thesystem. We encourage them to speak out,”says Susan Grant, R.N., Dana-Farber’schief of nursing and patient care services.

Indeed, Dana-Farber has made ongoingpatient education a top priority. A newPatient-Family Advisory Council advisesstaff on ways to enhance care. SomeCouncil members accompany the medicalstaff on hospital rounds to get patientfeedback. “We are truly welcomed onthese rounds,” says cancer survivor JudieBeard. “When mistakes are uncovered, thestaff genuinely wants our input.”

Reporting errorsBut by far the toughest job Dana-Farberundertook was to ensure errors would infact be uncovered—and reported. It setabout to implement a “non-punitive”system of error reporting. But it waslaunching such an effort amid a tenseenvironment of suspicion in the wake ofthe overdose events.

Although Institute leaders realized sys-temic errors were the major factors in thecases of Betsy Lehman and MaureenBateman, they had in fact dismissed theresident physician who had made the orig-inal error. The senior attending physician incharge at the time was also fired. In addi-tion, 15 nurses recently have reluctantlyaccepted reprimands by the Massachusetts

Board of Registration in Nursing for theirroles in the incident. Three other nurses do not agree with the Board’s position andwill soon face hearings.

The Institute’s current leadership isunderstandably reluctant to comment onthese personnel matters. They focusinstead on the future and the importanceof establishing a system that will makefiring people unnecessary (except incases of intentional or truly negligentbehavior that leads to patient harm orputs a patient at risk of harm).

“Removing the threat of judgment ofthe individual frees people up to discusssystem failures. But you have to be careful to not decrease the incentives forpersonal vigilance, accountability andresponsibility,” says Richard Bohmer,M.B.Ch.B., M.P.H., who teaches healthcare management at Harvard BusinessSchool. At the same time, Bohmer andother management experts say healthfacilities have held a wrong-headednotion of what would keep errors incheck. “The general assumption [before]at Dana-Farber was, if you hire the best people, give them the best trainingand provide the best equipment, thatexcellence in care is a happy coalition of these organizational elements,” says

7

Rather than a handwritten order that can be misread, physicians enter drug orders via a computerized ordering system. (Photo by Steven Gilbert, StudioFlex Productions)

lines for the use of the anesthetic wereestablished within weeks as were a newset of guidelines for performing bone mar-row aspirations and biopsies. In addition,a new procedure was put in place to moreextensively train and monitor physicians-in-training who are learning the proce-dure. At the conclusion of the investiga-tion, in a memo to his colleagues, the doc-tor expressed his gratitude for the “profes-sional respect and treatment he receivedin a potentially embarrassing situation.”

Maureen Connor, R.N., who overseesDana-Farber’s risk-management program,says the new reporting climate producesresults. “Health care providers andpatients are more willing to report everyerror, down to the smallest mistake. Withthis information, we can move forwardand deal with it,” she added.

The Next StepsDana-Farber officials say they will sustainthe commitment to error reduction in otherways as well. For example, with fundingfrom the Risk Management Foundationin Cambridge, MA, Institute personnel will

participate in a Harvard study of error pre-vention in the ambulatory setting. Thestudy’s goal is to define—and ultimatelyprevent—errors that can occur in outpatientclinics, doctors’ offices, nursing homes andcenters that do outpatient surgery.

Dana-Farber physicians, nurses, phar-macists and administrators are also often onthe road, sharing their message with thou-sands of people at meetings and confer-ences nationwide. “We realized that we hadthe burden of our action and the responsi-bility to correct it,” says David Nathan,M.D., Institute president since 1995. “But wealso realized we had the power to help oth-ers learn from our experience.”

And others are learning. Several can-cer centers have switched to a safer wayof handling medication orders since theoverdose tragedy at Dana-Farber. Some,including Memorial Sloan-Kettering inNew York, are implementing computer-ized order-entry systems.

But no system, whatever its advan-tages, will eliminate the need to be con-stantly on guard, watching for every pos-sible crack in the safety shield. Says Dr.Nathan, “Even with the best effort, errorprevention requires eternal vigilance.”

8

Bohmer. “For years, they got away withthis line of thinking—lots of hospitalshave gotten away with it.”

Dana-Farber’s shift to non-punitiveerror reporting is still evolving and farfrom perfect. But “compared to the past,”says Robert Soiffer, M.D., co-chair of thePharmacy and Therapeutics Committee,“there’s been a dramatic increase in staffreporting of errors—about 60 to 80 eachquarter.” Most have been minor and notaffected patients directly, he says.Several “near misses” have occurredsince the overdose incident, althoughnothing of that magnitude. All werecaught by the pharmacy staff who ques-tioned ordered prescriptions.

The Institute’s response to a March1999 incident is telling. A physician-in-training (a fellow) injected a patient witha local anesthetic in preparation for abone-marrow aspiration. The patient suf-fered a seizure but later fully recovered. Acommittee was immediately convened toreview the incident and interview thephysician. They probed the incident to seeif any processes should be changed. Itturned out the fellow had used a doseslightly higher than standard. New guide-

C H A N G I N G T H E P A R A D I G MF R O M TO

All is Fine! Endless Opportunities

Errors Rare Errors Everywhere

Tell as Little as You Can Tell Whatever You Can

Keep Board Out Actively Involve Board

MD’s Don’t Participate Doc’s Actively Involved

Middle Management Must Fit Exec’s Will Lead Staff

Error Rate is Average No Threshold for Errors

F R O M TO

Just Give Me a Name Look for the Root Cause

Penalize the Reporter Thank the Reporter

Add More Checks Simplify the System

It Will Pass You Will Carry the Burden

Fix as They Happen Set Reduction Priorities

“ W H O D I D I T ! J U S T F I X I T !I W A N T N O M O R E ! ”

“ E V E R Y T H I N G ’ S F I N E ”

When Kenneth Kizer became UnderSecretary for Health in the U.S. Depart-ment of Veterans Affairs (VA) in 1994, heknew he was taking on a major challenge.Long-maligned for shoddy care, the VAsystem is also famous for being anentrenched and highly politicized bureau-cracy resistant to change.

“I saw Born on the Fourth of July likeeveryone else,” the 48-year-old physicianadministrator says, “but I also knew thatthe VA’s problems were no worse thanaverage. They were just more visible.”Kizer set about making a series of changesat the VA that shook the institution.Among them were decentralizing the VA’spower and emphasizing outpatient-basedprimary care. In the process, more thanhalf of all acute care hospital beds havebeen eliminated. But it was a commit-ment he made in 1997 that has recently catapulted the VA into the limelight, and could be judged Kizer’s most endur-ing legacy. (He stepped down as the VA’sadministrator in July 1999 and now headsthe National Quality Forum, a privatenon-profit group.)

After the St. Petersburg Times docu-mented a series of fatal medical errors at

VA hospitals—a report that led toCongressional hearings—Kizer decided tomake a virtue of visibility. He testified in1997 that such errors not only occurredbut were far more common than mostpeople understood—both at VA hospitalsand other health care facilities. And hepledged to Congress that he would wagean all-out campaign against medicalerrors in the VA system. “I thought, let’sput all the problems out there and useheightened awareness to correct them, notjust in the VA but throughout health care.”

It was a bold move that by mostaccounts has paid off. Media attention inlate 1999 surrounding a report from theInstitute of Medicine (IOM) on medicalerrors included numerous favorable men-tions of the VA’s effort to address theproblem. And the VA’s own report—tally-ing medical errors in its system for thefirst time—was positively perceived as anexercise in government openness. The VAreport documented nearly 3,000 errorsthat resulted in some 700 patient deathsbetween June 1997 and December 1998.

About half (277) of the deaths weresuicides. Of these, an undetermined num-ber—but perhaps most—can be considered

“medical errors” by virtue of being pre-ventable if proper care had been rendered.The other half were associated with anassortment of medical mishaps, includ-ing improper insertion of catheters andfeeding tubes, botched surgery, falls, andpatients receiving the wrong drug or toomuch of the right drug.

The VA report was an almost instantbeacon for other, private health facilitiesand systems to begin tracking errorsmore systematically and preventingthem. Indeed, the watchword now in thehospital industry is “if the VA can do it,we certainly can.”

The $18.3 billion VA is the nation’slargest integrated hospital and health caresystem. It includes 172 hospitals, 600 out-patient clinics, 132 nursing homes, 206counseling centers, 73 home health careagencies, and assorted other programs.The VA employs 200,000 people, andmore than three million veterans a yearseek medical services at VA hospitals.Indeed, to Kizer, the VA’s national pres-ence, size, and status as a governmententity—plus its vital role in medicalresearch and training—made it ideal forsetting the national example he had in

9

Two physician leaders with a deep commitment

to safe medicine helped make the sprawling

VA health network a model for others

The VA Health SystemBY LANI LUCIANO

Leads the WayA Government Health System

mind. “We capture a huge amount ofdata, we’re too big to ignore and, sincegovernment employees can’t be individu-ally sued, our system is freer than most toadmit mistakes.”

The first steps came in January 1998.Kizer gathered the VA’s effort into a singleadministrative entity, the Patient SafetyImprovement Initiative. Later that year,Kizer recruited former astronaut JamesBagian to head the VA’s newly formedNational Center for Patient Safety, head-quartered in Ann Arbor, MI. Bagian, aphysician, mechanical engineer, and vet-eran of two space flights, was one of thelead investigators on the 1986 Challengeraccident. His new mandate was both sim-ple and staggering: identify errors andpotential errors in the VA system and fig-ure out how to prevent them. “I heard thetheme from Mission Impossible playing inmy head,” jokes Bagian, whose playfuldemeanor belies his rock-hard resolve.

The initiative aims to reduce medicalerrors through changes in everything fromadministrative procedures to contracting,technology support, nurse training, andmedical practice itself. A few parts of themulti-faceted effort are well underway—such as the new program to track andreport errors. But Bagian is quick to pointout that most are still evolving.

The first big test of a systemwidechange, he says, will be a requirement thatall hospitals put in place a bar-coding sys-tem similar to, but more sophisticated than,the kind grocery store check-out clerks useto read the price of foods. All VA hospitalsmust have the system by June 2000.Bagian says the decision was acceleratedbecause an initial test at two VA hospitalsin Kansas proved the technology dramati-cally successful. It reduced the medicationerror rate 70 percent over a 5-year period.

The Colmery-O’Neil VA MedicalCenter in Topeka, KS, began testing anearly version of the bar-coding system in1994. Now they have 35, $2,700 hand-held battery-operated computer termi-nals, two per 30-bed ward.

The system works this way: patientsand staff wear bar-coded identificationstrips, and all medications also have theirown ID strips. Before medicating apatient, a nurse or staff member laser-scans all three strips—from the patient,nurse, and medication—into the computer.

In a few seconds, the software verifiesthat the right person is receiving the rightdrug in the right dose at the right time.The program screens for half a dozenother potential problems, such as druginteractions. If everything checks out, thesoftware simply records the event—pre-serving an electronic record. If not, itflashes an immediate warning.

The 350-bed hospital—like almost allother U.S. hospitals—previously had noformal system of verifying medicationprescriptions at the “point of care.”Doctors and nurses were simply exhort-ed to double check, says Edgar Tucker,director of the VA’s Eastern Kansas divi-sion. And that was not proving to begood enough, Tucker says.

Many of the hospital’s nurses worriedat first that the new system was mainly away to track efficiency and boost workloads. But the very quick decline in errorsdispelled skepticism, Tucker says. Medi-cation error rates at the hospital plungedfrom an average 34 per month in 1993 tosix per month by mid-1999. Similarresults were obtained at the VA hospitalin Leavenworth, KS.

A similar bar-coding system eventual-ly will be required for blood productsused in transfusions. Says Kizer: “This isa relatively low-cost, high-benefit safetyimprovement that all hospitals can andshould invest in.”

Other InitiativesA second innovation the VA undertookwas to revise its medication storage proce-dures. Following the advice of the NationalPatient Safety Partnership (NPSP)—a groupthe VA founded in 1997 with 11 otherorganizations—all VA facilities mustremove a host of particularly hazardousmedications from areas where patients arecared for. Heading the list is the drug mostfrequently involved in medication errors—concentrated potassium chloride. It cancause cardiac arrest when injected in undi-luted amounts. Also banished are concen-trated epinephrine, digoxin, insulin, lido-caine, pancuronium and verapamil. Suchdrugs can cause significant harm or evendeath if delivered to the wrong patient orgiven in inappropriate doses.

Physical restraints, which cause an esti-mated 100 deaths a year, most due to

strangulation, are a third patient safety tar-get. The VA is conducting intensive work-shops to teach staff alternative ways tosafeguard patients without using restraints.The training is paying off. According toJudith Salerno, the physician who is theagency’s chief consultant for geriatrics andextended care, two-thirds of all VA nursinghomes are currently restraint-free.

Beefing up the science base of errorreduction is also a priority. Four VA cen-ters—in Palo Alto, CA, Cincinnati, OH,Tampa, FL, and White River Junction, VT—have been designated Patient SafetyCenters of Inquiry. Each will get $1.5 mil-lion over the next three years to conductresearch and identify safety techniquesand technologies other industries are usingwhich may have health care applications.

But both Kizer and Bagian say thebiggest challenge in reducing medicalerrors and making hospitals safer places ischanging the culture of medicine. As Kizerencapsulates it: “Can you imagine a nurseyelling ‘Doctor, stop!’ or even mentioning‘Are you sure you want to do that?’ No,you can’t, and that needs to change.”

Bagian says the initiative aims toreplace medicine’s fault-finding hierarchywith the institutional trust and sense ofpersonal responsibility typical of “high-reliability” organizations—such as his for-mer employer, NASA.

Though brimming with ideas he’d loveto execute right away, Bagian knows theeffort will take time and that he mustmove deliberately and cautiously. To thatend, he is field-testing several projectsbefore trying to implement them at all VAfacilities. One is revamping of VA’s errordata repository, known as the SafetyEvents Registry (SER). All VA personnelcan report to the SER so-called “sentinel”

10

events—incidents or situations that causedsignificant harm to a patient. But Bagianwants to make the system a repository forclose calls, too—events or situations whichcould have resulted in an accident, illness,or injury, but didn’t.

“Close-calls show us weaknesses in thesystem just as harmful events do, but at amuch lower cost in human misery andwasted resources,” Bagian says. To identi-fy problems that require the most urgentattention, the SER uses a safety assessmentcode. It rates individual incidents on boththeir severity of harm to the patient andtheir frequency of occurrence. Incidentsthat cause great harm or occur often aretargeted for immediate intervention.

Paradoxically, Bagian says, making SERa more useful tool for improving patientsafety will depend a great deal on makingVA doctors, nurses, and technicians feelsafer using it. “It’s the front-line folks, theclinical and risk management staff, thatwill make this initiative work. We needtheir trust and their input,” says Bagian.

To foster that trust, he has rewrittenthe VA’s Patient Safety Improvementhandbook. The new version stresses institutional and interpersonal coopera-tion, not personal consequences. And itemphasizes that reported errors are sub-ject to analysis leading to correctiveaction and preventive measures ratherthan punitive action against staff mem-bers. The sole exception is a criminal ordeliberately unsafe act.

The VA’s efforts to uncover mistakesare aided by its exemption from the legalliability cloud that hangs over privatehospitals. Neither the VA nor its doctors,as government employees, can be sued by

patients or their family members if a mis-take that resulted in patient harm isuncovered. This has made it easier for theVA to encourage error reporting. ButBagian doubts that this protection hasplayed that big a role. Rather, he says, fearof job loss or punishment is the obstacle.

“We don’t favor the beatings-will-con-tinue-until-morale-improves approach,”smiles Bagian. “We want people to under-stand that all problems merit analysisbecause they’re opportunities to inter-vene. Everyone on the team has to buyinto that crucial belief.” The new hand-book and the system it promotes are cur-rently being tested in Florida, PuertoRico, California and Nevada. This effortalso will test on-site “root-cause” analysistraining at all VA facilities by September2000. The goal of the three-day sessions,he says, is to foster new attitudes, notimpose new hard and fast rules or formu-las. “The old model was, you stressed whowas at fault. The new model is that weunderstand you do not come to work tomake errors, and we want to minimize therisk that you will do so,” says Bagian.

Getting information on errors andpotential errors is but half the calculus inimproving safety, however. The other halfis responding. Under a new set of direc-tives from Bagian, error analysis mustculminate in an action plan that moves tosolve the problem. Simplistic remedies,such as “nurses must pay more attention”are no longer acceptable, and eventhoughtful, plausible solutions are sub-ject to ongoing scrutiny for effectiveness.Says Bagian: “We’ll get monthly progressreports and we’ll be able to tell by thecontent whether the staff really got the

message or not. If not, we’ll head backthere and figure out where we missed theboat and how to do it better.”

The VA also has not been afraid to usesome tried and true methods of inducingchange to reduce errors. Recognitionthrough awards is important. But puttingsome money on the line appears to payoff, too. Individuals who suggest broadlyapplicable safety improvements qualifyfor bonuses of up to $5,000. Institutionscan garner up to $25,000. Topeka’s med-ication scanning system won the maxi-mum prize in 1999.

Continuing education for staff mem-bers is also a critical component of theinitiative. In 1999, the VA became thecountry’s first health system to requirethat all permanent employees completethirty hours of continuing education eachyear, ten of which focus on quality im-provement and patient safety. Bagian alsowants to launch a magazine devoted tosafety, such as those published by theNavy and Air Force. “They’re filled withinspiring you-are-there type stories thatmake you consider how you’d handle asudden problem. They’re absorbing toread and they stimulate you to think.”

Yet another idea is to increase the useof simulators for medical procedures.Such devices allow doctors and nurses topractice procedures and emergency drills,and make mistakes, using equipment—including computerized human dum-mies—before they ever perform them onreal patients. (See story page 24.) VAfacilities are also under orders to doublethe number of autopsies they perform.Currently, 15 percent of patients who diein a VA hospital get an autopsy. Medicalexperts have agreed for years that autop-sies are very useful tools for learningabout medical mistakes.

Bagian is far from blasé about the challenge ahead. To build a system wheresafety permeates the culture will takeyears, he acknowledges. “We will needeverybody up and down the chain of command, from Congress to doctors andpatients to believe in this new approach,”he says. “We have to encourage andpraise and reward all the steps forward,including baby steps, so that people arecomfortable taking bigger steps. But oncewe reach critical mass, I think themomentum will be unstoppable.”

11

“Close calls show us weaknesses in

the system just as harmful events do,

but at a much lower cost in human

misery and wasted resources.”

Even as the e-health boom gainsmomentum, many health facilities arestill far behind in adapting to the bravenew world of information technology.And though they have seen computer-ized machines—like CAT scans andMRIs—revolutionize medicine, manyhospitals have not yet put computers touse to increase productivity, stream-line administrative functions, enhanceresearch, and improve quality. Somehealth care institutions, though, areblazing trails in this form of computeri-zation that the rest of the industry is cer-tain to follow. Brigham and Women’sHospital, a 726-bed academic medicalcenter in Boston, MA, is one such facili-ty. Among the most powerful changes:Brigham’s physicians today write alltheir orders for treatment and medica-tions on a computer screen, renderingpaper orders a thing of the past.

The payoffs have been many. Doctorsand nurses say it saves time and makesthem feel more professional, efficient, andconfident. And hospital administrators saythe system has saved them between $5 and$10 million a year. But it’s the direct ben-efit to patients—in reduced errors—that’slikely to compel other hospitals to follow

suit quickly. Serious medication errors atthe hospital have decreased by 55 percentover the past few years.

Brigham and Women’s began to em-brace the computer age as early as 1989. Itbuilt the Brigham Integrated ComputingSystem (BICS), a clinical information sys-tem running on a network of over 6,000computers. Physicians, nurses, and admin-istrators used the system to access labresults, discharge summaries, and otherclinical data. By 1991, the hospital’sadministrators were casting about for morespecific ways to use computers to improveboth the delivery and quality of care.

A key target emerged soon enough:“iatrogenic injuries”—injuries to patientsthat occur during hospitalization—and,more specifically, injuries due to medica-tion errors. Studies at the time estimatedthat “adverse drug events” were far andaway the most common medical error andcost U.S. hospitals over $20 billion peryear. In addition, the problem seemed tobe on the rise as the health systembecame more complex and medicinesbecame more plentiful and powerful.With 36,000 inpatients and 610,000 out-patients per year seen at four differentsites, the Brigham’s administrators knew

12

Brigham and Women’s HospitalBoston, Massachusetts

BY JANE ROESSNER

Coaxing physicians to

give up pen and paper

cut errors sharply

and saved $5 million

to $10 million at one

inner city hospital.

Making DoctorsComputerLiterate

they had a significant opportunity toimprove quality and save money.

David Bates, a fellow in general medi-cine at the Brigham at the time, couldn’thave agreed more. A pioneer in qualityresearch, Bates had found that both theexisting processes by which physiciansordered medications as well as the waythey were administered were fraught withproblems and invited errors. His researchfound that 56 percent of adverse drugevents at the Brigham occurred becauseof errors at the time of ordering. Batesdiscerned early on that the primarysource of the problem was the age-oldtradition of handwritten orders. Doctorswere scribbling their drug and test orderson forms with triplicate copies, one ofwhich would be torn off and taken to thepharmacy (or shot through pneumatictubes). Also, about half of the medica-tions issued by the pharmacy wererecorded as having been written by “doc-tor unknown,” and in an astounding 80percent of lab orders, the doctor’s namecould not even be determined.

And despite exhortations to writemore clearly, physicians’ handwritingstayed notoriously illegible.

Bill Churchill, director of the Brigham’spharmacy since 1990 and a pharmacistthere since 1974, says deciphering theordering physician’s name was the least ofthe problems. “We simply couldn’t readthe doctors’ orders,” he says. “Differentpeople would use different symbols andabbreviations for medicines, and the phar-macy didn’t know what they meant. Youcouldn’t tell if the order was for 0.5 mil-ligrams or 5 milligrams.”

The problem was compounded by dra-matic advances in the pharmaceuticalindustry. “New drugs were coming on-line every day,” Churchill says, “Thesewere potent drugs with profound abilityto help, but also to harm. And these drugsinteract differently on each patient.Asking physicians to keep all of thisinformation in their heads is a lot. Evenfor pharmacists who specialize in this, itcan be overwhelming.”

At the same time that the data werebeing gathered, a potential solution wasbeing developed. Bates and others rea-soned that a comprehensive computerized“order entry” system combined—and herewas the trick—with a requirement that

physicians write their medication orders onthe computer would go a long way to solv-ing the problem. In addition, Bates becameconvinced that the computer and newinformation technology could become apowerful “decision support” tool for physi-cians—giving them pertinent informationat the time they were writing orders.

DESIGNINGTHE SYSTEMBrigham and Women’s set out buildingthe order entry system in 1991. JonathanTeich, then director of the hospital’sInformation Systems group, believed thatthe most potent application—the one thatwould have the biggest impact on care—was “the ability to challenge and critiqueand guide the physician‘s order at thetime it was placed.” At the same time, heand other designers recognized that acomputerized order entry system wouldimpose formalism onto what had been—rather bizarrely in retrospect—a largelyinformal process. In essence, the new sys-tem would be asking physicians to makea major change in the way they practicedmedicine. Physicians are required to doc-ument all of their orders—not just formedications, but for lab and radiologytests, food, and activity—essentiallyeverything they do in the process of car-ing for a patient. At the Brigham, some16,000 orders are written each day, 40percent of them for medications.

The team’s first decision was to getphysicians to buy into the system bydesigning it for them. Teich decided thatif one aim was to promote decision sup-port, doctors had to use the system them-selves. They couldn’t rely on nurses oradministrative staff to enter their ordersinto the computer. To get doctors to usethe system, Teich knew he had to make iteasy to use and not time consuming.Physicians told the design team in nouncertain terms that if it took a lot moretime, it simply would not fly. The teammade the computer format resemble thewritten forms doctors were used to, butthey also imposed standards on how thescreen should be filled out. The final con-ceptual challenge was to find the rightbalance of decision support information.

Too much would overwhelm doctors. Toolittle would undermine physician buy-inand the hoped-for reduction in errors.

In the end, the team settled on a menuof medications. For each medication,software offered a “pick list” of appropri-ate doses. Before any medication orderwould be accepted, physicians wererequired to enter dosage, method ofdelivery (for example, by mouth or intra-venously), and frequency.

But the key new element was this: assoon as the physician completed an order,it would be checked for errors. For exam-ple, if the patient was found to be allergicto the drug the doctor was ordering orwas already getting another drug thatdidn’t mix with the new one, the comput-er would alert the physician. If the doctorchose to override such an alert, he or shewould have to enter the rationale for theoverride. Orders were then sent directly tothe pharmacy. There, they underwent asecond check for errors.

THE ROLLOUTThe commitment of the hospital’s leader-ship proved to be a vital ingredient tosuccessfully implementing the new sys-tem. CEO Dick Nesson was a persuasivechampion. He personally mustered thesupport of the hospital’s departmentchiefs. Indeed, when one chief threat-ened to instruct his residents to refuse touse the system, Nesson made it clear thatevery department was required to atleast give it a try.

After a one-month pilot trial on thebone marrow transplant unit in January1993, initial rollout began in May 1993and continued for the next 18 months.The rollout took place, says John Glaser,the Brigham’s chief information officer, in“punctuated bursts”—first in Medicine,including general medicine, cardiology,and oncology; and then, after a two-month pause, in Surgery, and then inObstetrics and Gynecology.

Like the design itself, the rollout put thedoctor front and center. The operative prin-ciple in the short run was to minimizeproblems for the physician, even if it meantputting other departments (for example,the lab or radiology) and the nursing staffin an awkward position—between the paper

13

world and the electronic world—for a peri-od of time. Indeed, when the new systemwent into effect, a variety of questionsabout workflow bubbled up. For example,if a physician writes orders for postopera-tive care before surgery, how and when arethose orders activated? What happens ifthose orders say “continue all medication,”but the medications change in the interim?How is it indicated that orders have beentaken care of?

“Lots of things had to be handledexplicitly that were handled implicitly inthe paper world,” explains Gil Kuperman,a physician and one of the system’sdesigners. “Humans can be ambiguous.Computers aren’t.”

Physicians were invited to use a “feed-back” button on the screen to air griev-ances and suggest improvements in thesystem. Teich recalls that in the first weekof rollout, his department received 122pieces of “hate mail” via the feedback but-ton. “We felt better,” he says, “when werealized that 99 of them came from thesame person.” The feedback mechanismalso invited physicians to interact with thesystem; they became active participants inmaking the system better, not passive vic-tims of its intrusion into their lives.

In time most physicians respondedpositively, Teich says. They liked the basicinformation about proper doses and fre-quency for medications. And they likedthe computer’s ability to calculate theright dose of potassium for a particularpatient. But the single greatest perceivedbenefit by physicians, and the one thatreally saved time, was the ability to writeorders from remote locations—for exam-ple, writing an order for a patient on the12th floor when they were on the 9thfloor; or even writing orders from home.

But, as anticipated, there were com-plaints. The major one involved the addedtime commitment. Before computerizedorder entry, doctors spent approximatelytwo percent of their time writing orders;

the new system was using 10 to 12 per-cent of their time, according to atime/motion study. That translated intoan average 40 minutes a day longer writ-ing orders. Fortunately, as they adapted tothe system, doctors recovered an averageof 20 minutes of that time. But it wasn’tenough. So the designers set about tryingto make the process more efficient.

Their first discovery was that a signif-icant amount of time could be saved ifdoctors wrote orders in groups ratherthan singly. While writing single orderscould take as much as five times as longon the computer as by hand, writingorders in groups was much faster. Forexample, writing orders one by one for

patients who needed a blood thinner wastime-consuming—at five minutes perpatient, for forty patients, taking overthree hours. To facilitate “group orders,”the system’s designers built one screenwith a spreadsheet listing all patients incertain categories. Doctors could thenwrite orders for blood thinner for manypatients at once—a process that turnedout to require only seven seconds perpatient. It proved to be a critical change,improving efficiency all around.

CONTINUOUSIMPROVEMENTGetting the initial bugs out and weavingcomputerized order entry into the day-to-day fabric of the hospital took about ayear. The next step was to add the “deci-sion support” component.

Decision support was designed withtwo goals in mind: reducing errors (e.g.,“patient X is allergic to medication Y”)and reducing costs (e.g., “would you liketo switch to this equivalent, cheaperdrug?”). The basic notion behind “realtime” decision support is that the comput-er can modify the physician’s decisions by

displaying information at the time thephysician is ordering medications.

The Brigham’s system offers fivemodes of decision support:

• It alerts physicians to potential druginteractions, patient contraindications orallergies;

• It calculates proper doses—determin-ing the right amount of a particular med-ication for a particular patient at a partic-ular time. The system’s designersresearched the hospital’s database todetermine the most commonly ordereddoses of particular drugs. Using thisinformation, they constructed “pick lists”of doses, highlighting the dose most com-monly ordered.

• It presents drug substitution infor-mation. The designers recognized thepower of the system as a tool to changephysicians’ behavior. When physiciansordered the common ulcer treatment andprevention drugs Zantac® or Tagamet®,the system suggested they consider sub-stituting Axid®, a less expensive butequivalent medication. The interventionallowed physicians to make the switch by simply pressing an “OK” button. In thevery first week the intervention wasintroduced, physicians elected to substi-tute the less expensive medication anastounding 94 percent of the time.

• It can alert physicians to potentiallydangerous situations. For example, if alab test for a patient indicated a lowpotassium level, the computer wouldquery if the patient was also takingdigoxin—a potentially dangerous combi-nation. If so, the computer would identifyand page the patient’s physician, at anyhour of the day, informing the physicianof the reason for the page and asking foraction. Physicians were overwhelming intheir support of this feature, giving it a 96percent approval rating.

• It can track patients’ test results forkidney function and alert doctors andstaff if medication doses exceed recom-mended levels for people with abnormalkidney function. Dubbed the NephrosProject, this feature was made a part ofthe decision support because so manyadverse drug events involved patientswith kidney problems. Doses of quite afew common drugs that would be appro-priate for a patient with normal kidneyfunction could be dangerous or even fatal

14

“Lots of things can be handled explicitly thatwere handled implicitly in the paper world.

Humans can be ambiguous. Computers aren’t.”

in a patient with low kidney function. Asa result of Nephros, the percentage oforders with dosing that is appropriate forkidney function has increased from 30percent to 70 percent.

RESULTS—AND THE FUTUREComputerized physician order entry at theBrigham is now six years and some 35million orders old. It has become an inte-gral part of providing care at the hospital.Paper orders are a rapidly fading memory.Thirteen thousand orders are entered dailyat the Brigham, 88 percent of them by aphysician. On an average day, 386 of theseorders are changed by a physician becausethe computer suggested doing somethingdifferent. An average of four times a week,a physician cancels a chemotherapy orderbecause the dose is too high.

Any way it’s cut, computerized orderentry has earned its keep. Research byDavid Bates concludes that the systemhas yielded a 55 percent reduction inserious medication errors—those thateither had the potential for harm or actu-ally resulted in injury. In all, the systemsaves $5–10 million per year, with mostsavings coming from use of less expen-sive drugs and tests.

As well as it has performed, the systemat the Brigham has been due for an upgradefor several years. That is now underway.The biggest issue has been the network,which has been at capacity for some time,with about 3,200 linked PC’s. This is beingreplaced with a faster network. Also, allapplications are being moved to a faster,more flexible software program.

In addition, plans are underway toextend the drug order entry system to the outpatient setting—the 610,000patients treated annually at theBrigham’s outpatient sites. Even thoughthe primary care sites already use anelectronic medical record, computerizingordering will be challenging. Primarycare physicians see many more patients,each for a much shorter time. Althoughthe range of orders is narrower in the pri-mary care setting, there are many moreorders per day. Moreover, it will be moredifficult to “require” independent prima-

15

ry care physicians to use such a systemthan the hospital’s own physicians.

The hope is that physicians will buy intothe system because it can help them navi-gate the complex rules of managed care—the varying rules plans have that specifywhich drugs and which tests are covered,under which conditions. These can changefrom month to month and it’s almostimpossible for physicians to keep up.According to John Glaser, an order entrysystem with decision support informationon every plan’s formulary—updated regu-larly—could be highly attractive. Brighamand Women’s researchers project that sucha system could reduce outpatient medicalerrors by as much as 80 percent and cut the

cost of medications under capitated con-tracts by as much as 25 percent for a sav-ings of an estimated $20 million per year.

As for the more distant future, Brighammanagement is looking closely at adapt-ing computer assistance and decision sup-port to emergency room operations and toinvolving patients more directly in theirown care. For example, on-site computers,with internet links, could help peopleunderstand health issues and treatmentchoices for their condition. They couldalso help people communicate directlywith doctors and other health providers.It’s the brave new world that manyexperts envision. And the Brigham isalready heading in that direction.

Building a computerized order entry system cost Brigham and Women’s Hospital about

$1.4 million in 1993; thereafter, annual maintenance has cost about $500,000 a year.

How many hospitals have the resources—financial, medical, and technical—to build such

a system? How does a home-grown system compare with a ready-made one?

According to the Brigham’s Information Systems team, building a system from scratch

makes sense for a hospital if three conditions are met:

• The hospital believes it can move faster than the market.

• The hospital is uncertain as to the best configuration of such a system, and wants

to learn from rapid cycles of testing and fixing features.

• Many functions within the hospital are intertwined.

Hospitals that meet these conditions can usually develop a system that meets their

needs better. The system is also likely to be cheaper in the long run than if purchased from

a vendor. Doing it yourself also gives an organization deeper knowledge on using tech-

nology to improve care.

However, most community hospitals and smaller hospitals will not have the human

resources to develop a system on their own, and will likely be better off purchasing one

from a vendor. Although off-the-shelf systems won’t have all the features or the flexibil-

ity of Brigham’s system, most vendors in this sector—such as Cerner, Eclipsys, SMS, and

HBOC—are making important improvements. The cost of off-the-shelf systems varies,

depending in part on the computer networks already in place.

BUILD YOUR OWN ORBUY OFF THE SHELF?

P harmaceutical giant Bristol-MyersSquibb Co. (BMS) had a problem.

Reports were filtering into the company’sPrinceton, NJ, headquarters that one of itspremier drugs—Platinol® (cisplatin), usedto treat testicular, ovarian and advancedbladder cancer—was being confused bypharmacists and doctors with anotheranticancer drug it makes called Para-platin® (carboplatin), used mostly to treatadvanced ovarian cancer.

The issue was serious because theappropriate dose of cisplatin is about fiveto six times less than that for carboplatin.Mix-ups can be fatal. Indeed, the poten-tial danger of a mix-up between the two drugs had been noted as early as1991. The Institute for Safe MedicationPractices (ISMP), based in Pennsylvania,had received a report of a woman aboutto get a bone marrow transplant who wasto receive cisplatin. Because of a pharma-cy dispensing error, she received cisplatinat the carboplatin dosage. She later died.

In another case that year, a nurse tran-scribed a telephone order wrong, writingcisplatin instead of carboplatin. The phar-macy dispensed the drug as written, andthe patient received the higher dose. Thenext morning the physician saw the tran-scribed order, and the error was discov-ered. The patient died a week later.

Confusion was apparently occurringwith the generic names (cisplatin and car-boplatin) as well as the brand names(Platinol® and Paraplatin®) because theyall shared the same “stem:” “platin.”

But BMS at that point still was unsurewhether they had a systemic problem ontheir hands—or whether the reports wereisolated events due to fluke mix-ups.Even so, it took a precautionary first stepin 1992—putting letters to the editor invarious oncology medical, nursing andpharmacy journals warning of possiblecisplatin mix-ups.

R eports of the error kept mounting,though. By the end of 1995, some 32

medication errors related to cisplatin hadbeen reported to the Food and DrugAdministration (FDA), and eight peoplehad died, according to the FDA. Thatprompted BMS to add information to thelabel for Platinol® highlighting the poten-tial mix-up. It also distributed letters tooncology health care professionals toalert them to the problem and to the dis-continuance of one form of Platinol®, apowder for injection that had the highestpotential to be confused with Paraplatin®.

Despite these steps, Frank Pasqualone,then director of policy and planning forBMS Oncology-Immunology, was still con-

cerned. He worried that the inserts andwarnings would go unread—as such insertsoften are. And he was concerned that morepatients might be harmed if BMS did notmove aggressively. The ISMP also was urg-ing the company to take stronger measures.

The Platinol®-Paraplatin® mix-up wasthen and still is symptomatic of a seriousand growing problem: health profession-als make drug dispensing and administra-tive errors every day—some of them life-threatening, even fatal—because thenames of many drugs are so similar. In the most recent example, the FDA received53 reports of dispensing errors involvingthe three sound-alike drugs Celebrex® (anarthritis drug), Cerebyx® (to treat seizures)and Celexa® (an anti-depressant).

The precise magnitude of the problem isdifficult to gauge. There is no comprehen-sive record-keeping or data base that tracksit. According to the FDA, roughly half ofthe 6,000 medication errors reported to theagency between 1992 to 1997 were due tolabeling or packaging issues. Of that half,some 27 percent were caused by generic ortrade name confusion. But Jerry Phillips,the agency’s associate director of medica-tion error prevention, says studies indicatethe number of medication errors reportedto FDA—including mix-ups—represent onlya small fraction of those that occur.

16

Bristol-Myers Squibb

BY CARALEE ADAMS

A pharmaceutical gianttackles a drug mix-upand sets a precedentLOTSWHAT’S IN A NAME? When it Comes to Drugs—

I t’s a significant and persistent prob-lem,” concurs Bruce Lambert, an asso-

ciate professor in the Department ofPharmacy Administration at the Univer-sity of Illinois at Chicago. He says phar-macists and other health professionalshave been concerned about it for yearsand have pressed the pharmaceuticalindustry to change their practices.

Magnifying the problem today,Lambert notes, is the growth in the num-ber of medications. There are about13,000 prescription drugs on the marketnow, according to IMS Health, a healthcare information company. And in thepast decade, the number of new registeredtrademark drug names has quadrupled,from 744 new trademarks registered in1988 to 3,038 in 1998. This creates amajor challenge for an industry alwaysstriving to come up with catchy drugnames—such as Prozac® and Viagra®.

Not surprisingly, with so much invest-ed in brand names and the marketing thatsurrounds them, drug makers are usuallyvery reluctant to change the name orpackage design of their products if aproblem arises.

That ultimately was the problem fac-ing BMS. But first the company sought togather the facts on Platinol® mix-ups andreach out to experts. Pasqualone pressedhis superiors to convene an advisoryboard of outside experts including fivepharmacists, two nurses and one physi-cian. The panel’s central mission was toreview the data and advise on a responseto the Platinol®-Paraplatin® issue. But itwas also charged with recommendingways to make the company’s productssafer in the future.

Pasqualone’s mounting concern,assertive approach, and BMS’s commit-ment to change their practices eventuallyset a precedent for the industry thatreverberates today.

The independent advisory group gath-ered in Naples, FL, in January 1996. Foreight long hours, and with company rep-resentatives on hand, it engaged in aprocess dubbed “failure mode and effectsanalysis.” The group examined, in detail,the distribution of the two drugs from thetime they were loaded onto trucks aftermanufacture to the moment they wereinjected into a patient in a hospital.

T he vast majority of the mix-ups withPlantinol® and Paraplatin® were

linked to the wrong container being cho-sen as it was dispensed by a pharmacist.

“It was clear that it was a problembeing contributed to by their packaging,”says Dominic Solimando, a pharmacistand oncology consultant in Arlington,VA, and a panel member. Indeed, thepackaging designs of the originalPlatinol® and Paraplatin® boxes wereextremely similar—which turned out notto be an accident. The packaging of allthe oncology products were intentionallysimilar for purposes of product line con-sistency and brand recognition, a BMSspokesperson says.

Working with company staff, thepanel came up with a number of recom-mendations:

• Redesign the cartons of both drugsto de-emphasize the generic name andeliminate the confusing similarities—putthe letters “cis” in bold uppercase and redletters with the “platin” in smaller type.

• Add a graphic of a stop sign to thepackage to warn pharmacists and otherusers to verify the drug name and dose.

• Redesign the outer carton so the crit-ical information is on more than one sideof the box, making the warnings visibleno matter how the box is placed on thepharmacy shelf.

• Put warnings of the maximumdosage of cisplatin on the vial.

• Put a set of three stickers on the car-ton to help certify the drug had been pre-pared at the proper dosage.

• Put the same stickers on I.V. bags;one is used by the pharmacist and anoth-er by the nurse.

Pasqualone set out to convince thecompany to make all the changes. It tooktime. But Pasqualone was determined.

Indeed, as 1996 progressed, the num-ber of cisplatin-related problems mount-ed. Twenty-nine cisplatin medicationerrors were reported in that year, withanother eight deaths—bringing the totalsince 1991 to 61 incidents and 16 deaths.

By spring, Bristol-Myers’ managementwas firmly on board with the panel’s andPasqualone’s suggestions. They gave thego-ahead for rapid action. Some of theoutside experts were consulted as packag-ing changes were made. By October 1996

nearly everything the advisory panel hadrecommended had been put into place.

The company won’t reveal the cost ofthe changes, but Pasqualone says it was a“significant investment.” The costliestitem was the advertising campaign toeducate health care professionals aboutthe product changes.

Notably, one change the panel did notrecommend was changing the name ofthe drugs. Doing so would have added$5–10 million to the cost of the response,estimates Michael Cohen, president of theISMP. Although Cohen favored a namechange, the panel fully considered thatoption and the consensus was that chang-ing the name would not have added tothe prevention of the mix-ups. Instead, itwould have confused pharmacists andperhaps even led to other drug mix-ups.

In the end, most agree that the compa-ny did the right thing. “All in all, we wantpeople to act like Bristol-Myers Squibb didand fix the problem,” says pharmacy man-agement expert Lambert. “Whether they didso quickly enough is another question.…When errors come to light, we’d like to see(companies) be much more aggressive.”

“Bristol-Myers Squibb set a goodexample of acting responsibly and usinginnovation to correct a problem that ledto serious patient injury,” says WilliamEllis, executive director of the AmericanPharmaceutical Association Foundation,a research group affiliated with a nation-al professional society of pharmacists.

Cohen says BMS’s actions shouldserve as a model to others. It was one ofthe first companies willing to reach out tooutside experts and analyze what couldbe done to correct a problem with itsproduct, rather than blaming outsideforces, he says. But, he adds, the moreglobal response to drug name mix-ups iscareful testing of drug names, labels, andpackaging before they are launched.

After the packaging changes weremade in 1996 and the advertising cam-paign in 1996 and 1997, cisplatin-relatedmedication errors and deaths declineddramatically. In 1997, six cisplatin errorswere reported, with one patient death. In1998, only one incident was reported. Andby October 1999, two incidents had beenreported. No deaths related to cisplatinoverdoses were reported in 1998 or 1999.

17

“

Like hundreds of hospitals in hundreds of similar towns, 310-bed Luther Hospi-tal and Midelfort Clinic sit placidly amid the tree-lined streets of Eau Claire, WI, acity of 60,000 at the confluence of theEau Claire and Chippewa Rivers. And likemost doctors, Roger Resar thoughtLuther-Midelfort—hospital and clinic—were pretty safe places for the 2,000patients a day that pass through theirdoors. As far as Dr. Resar could tell, a ter-rific team of physicians and nurses,backed up by a crackerjack administrativestaff, were delivering top quality care in awell-greased system. Resar knew, ofcourse, that errors sometimes occurred, asthey do at every hospital. In his 22 yearsas a pulmonologist, he’d witnessed hisshare. But he felt Luther-Midelfort had asolid program to prevent serious errorsbefore they occurred, catch them in atimely manner when they did, andaddress what went wrong.

Resar has a very different view today.And so do his bosses. After authorizingResar in 1997 to assess errors and effortsto reduce them at Luther-Midelfort, thehospital’s leaders came to a soberingreassessment of the success of their exist-ing programs.

“I was astounded,” says Resar. “Wewere finding 200 to 230 actual or poten-tial medication errors for every 100 chartswe reviewed. That means if you went toour hospital, you’d be at risk of two tothree errors during your stay with us.”

Fortunately, the vast majority of theerrors uncovered at Luther-Midelfort wereminor. But the findings—from a painstak-ing year-long study of charts and proce-dures—frightened Resar, his colleagues,and the hospital’s administrators. All real-ized that a process which allows too manysmall mistakes is the same process thatleads inevitably to a big mistake. “Maybewe gave a sleeping pill to someone whoshouldn’t have gotten one. That’s not a bigdeal. But what if someone got penicillinand they were allergic to it,” Resar says.

Resar was put in charge of dramatical-ly reducing errors at Luther-Midelfort. Hisfirst task was to take on medication errors—where most of the errors were occurring.Resar was also asked to build a new systemof error reporting, which proved to be oneof his toughest challenges.

The results so far have been highlyrewarding, Resar says. Since mid-1998,medication errors at Luther-Midelforthave declined by 82 percent at the keypoints of patient care—admission, transferfrom one department to another withinthe hospital, and discharge. Insulin errorshave been cut by 50 percent after a stan-dard dosing regimen was implemented.And 90 percent of patients on the blood-thinning drug Coumadin® are now get-ting the proper dose, up from 75 percentbefore improvements were made.

As part of the Mayo Regional HealthSystem, and thus partner to the famedMayo Clinic in Rochester, MN, Luther-

Midelfort’s initiative is being watched close-ly and eventually could have even broaderapplication. Mayo runs 13 hospitals in 54communities in and around Rochester.

The Mayo Systemand ChangeTwo factors have fueled the innovations atLuther-Midelfort—the sheer force of Resar’spersonality and an organization that cameto embrace the cause and open itself to change. “You have to recognize thatimprovement is hard work,” says Dr.Terrance Borman, the hospital’s medicaldirector. “It has to be someone’s job. We’verecognized that in order to help our doctorsimprove, we need doctors who are changeagents.” Ron Hitzke, Director of InpatientPharmacy Services, says Resar has been“the spark plug that keeps us going.”

Dr. Patrick Macken, a nephrologist in-volved in many of Resar’s projects, concurs.“He’s recognized as an excellent doctor anda team player. He’s very passionate abouthis work.” Macken says the Mayo organiza-tion’s commitment to let Resar do the nutsand bolts work of quality improvement hasmade all the difference. “Resar creates thenew forms, he looks at the data, he’s theone who works closely with the nurses.”

Resar says getting physician buy-in is akey to the success of the initiative so far. “Ifyou’ve been around a hospital longenough, you know the guys who carry the

18

Tackling Medication Errors

Luther-Midelfort HospitalEau Claire, Wisconsin BY KRIS REBILLOT

Putting one determineddoctor in charge of reducing errors madeall the difference at oneMidwest hospital

HEAD ON

big stick are the physicians. They have tobe in on it from the beginning,” he says. Healso acknowledges the approach used atLuther-Midelfort to implement change:built a project in small steps—steering clearof making wholesale changes too quickly—and test changes first on a smaller scale.“We spend five minutes coming up with aform,” says Resar. “We don’t do a lot ofplanning because we know it probablywon’t be right the first time. Some organi-zations take months to design a form, butspending too much time thinking about itmakes things too complicated.” This systemalso allows physician “champions” to takepart in short, 20-minute meetings aimed atimprovements, instead of involving themin knock-down, drag out review sessionsthat are common in many organizations.

TacklingMedication ErrorsThe efforts at Luther-Midelfort started in1997 when the hospital sent a number ofstaff members, including Resar, to a sym-posium on reducing medication errors.Resar and hospital leaders became convertsto the emerging—but still minority—viewthat errors were a more serious problem athospitals than generally appreciated.

Resar’s first step was simply to starttalking to staff about the issue. He queriednurses, doctors, pharmacists and purchas-ing agents, asking them blunt questions ininformal and off-the-record conversations.His most pointed question was always,“Are there accidents waiting to happen?”

As these conversations progressed, itbecame clear to Resar that, one, errors wereprobably more common than believed atthe hospital and, two, employees had beenwithholding information about them.Retribution had never been a big part ofthe corporate culture at Luther-Midelfort.But even in this positive work environ-ment, the threat of punishment held sway.Nurses, faced with the choice of “telling”on colleagues (who could face disciplinaryaction) were not reporting medication mis-takes—even the most minor ones.

Resar’s informal survey yielded a listof key drug delivery issues—and specificmedications—of biggest concern. That listcame eventually to form the basis for athree-year, 11-part project. Resar’s chart

review further shaped the initiative aspatterns emerged. Data showed, forinstance, that 56 percent of the hospital’smedication errors occurred at the “inter-faces of care”—in other words, duringadmission, in-hospital transfers, and atdischarge. Armed with this information,Resar and his team began their first for-mal improvement project in 1998 on a40-bed medical unit. They committed tothe creation of a “drug reconciliationprocess” to address medication errorsoccurring during the interfaces of care.They set a target: they would reduce actu-al and potential adverse drug events by60 percent in three months.

They met the goal. The team beganfirst by tackling medication mistakesoccurring during the admission process.“We started with just five patients and twonurses on one shift,” said Resar. “We tookan idea and tried it. We didn’t refine it todeath. If it wasn’t right, we let it go. If itworked we kept moving it to a largercycle.” The first step, he says, involved thesimple act of moving the “admission med-ication notes”—typically buried in a pile ofpaper—to the front of the chart. “It took alot of work just to get that done,” saidResar. “We had to figure out what form touse, and we had to get people to move it.”

Equally painstaking was the develop-ment of new protocols for admittingnurses who became accountable forunraveling the snarl of drug informa-tion. Elderly patients, for example, oftenare taking 12 to 20 different medications a day. “It was a nightmare,” says JaneJustesen, a nurse manager.

During their initial chart study, forexample, Resar discovered one case involv-ing a disoriented elderly woman who tookthe anti-coagulant drug Coumadin® on adaily basis. The patient was admitted overa weekend by her primary physician’s part-ner, who was unaware that the patient tookthe blood-thinner. The patient’s regulardoctor didn’t catch the omission onMonday and as a result, the patient had astroke during her stay.

Justesen’s admitting nurses began tosit down with patients and their familiesto detail and verify every drug they weretaking, its dosage, frequency, and when itwas last taken. If needed, clinic recordswere pulled, and pharmacists were con-sulted. Everything was compared to adoctor’s orders. If something didn’t makesense and was time-sensitive, the doctorwas called. The nurses had 24 hours to getthe medication record in order.

Once the new admission process wasestablished, Resar tackled medicationerrors associated with transfers from thehospital’s intensive care unit. That effortinvolved the formidable task of getting twocomputer systems to work with eachother—the Intensive Care Unit had a systemseparate from the rest of the hospital. Now,a process is in place allowing those systemsto work in coordination. In addition, newforms make it easy for doctors to detailmedications for patients leaving the ICU.

The discharge process was next.Typically, hospital pharmacists were spend-ing 30 minutes to two hours per patient todiscern precisely what the doctor’s “dis-charge on home medication” orders meant.

19

Measures to prevent medication errors at Luther-Midelfort have involved the entirehealth care team and the patient, as well. (Photo by Rick Gregerson)

Working closely with the pharmacy staff,Resar built a new process within a coupleof months. Now at discharge, the pharma-cy’s computer generates a record of everymedication the patient has taken while inthe hospital. All the doctor has to do ischeckmark the medications he or she wantscontinued. Every drugstore in Eau Claireaccepts the watermarked document as aprescription form.

“Not only does this save the doctors’time,” says Hitzke, “It also prevents thepotential for mistakes in transcription.”Pharmacists also have more time to helppatients. “We’ve always wanted to bemore clinically involved, and this is away to do that,” says Hitzke.

Resar and the patient safety team alsotargeted specific drugs they found associ-ated with errors and delivery problems.Chief among these were insulin for diabet-ics and Coumadin®, a blood thinner com-monly prescribed for clotting disorders andheart rhythm irregularities. Resar’s researchfound that about 20 episodes of low bloodsugar were being identified among diabet-ics each week in the hospital. The problem:practically every doctor on staff used his orher own “sliding scale” to adjust insulindosages to keep patients’ blood sugar lev-els stable. “We were all trained differently;everybody had his or her own way ofdoing things,” says Macken. “We identified14 patients who were being cared for bynine doctors, who used 12 different scalesfor insulin. How were the nurses supposedto know what to do?”

Testing different options with a hand-ful of patients under Macken’s care, Resarworked with pharmacists to come up witha standard scale for insulin dosage. Thatscale is now being used for all diabeticsadmitted to the hospital. “The nurses arehappy with it,” said Macken. “The phar-macists are thrilled.” The patients alsowon: insulin medication errors declinedby 50 percent.

Problems involving blood-thinningdrugs appeared even more life threaten-ing. Clinic patients taking Coumadin®were being admitted to the hospital foreither excessive bleeding, resulting fromtoo much of the drug, or strokes, a poten-tial result from getting too little of thedrug. Adverse drug reactions also werebeing noted in patients taking Coumadin®in combination with certain antibiotics.

Once again, the problem stemmed fromtoo many options. There are a number ofways to calculate Coumadin® dosage, and there are a wide variety of tabletsavailable. In addition, Resar and Mackenfound that patients were coming in for labtests when their primary physician wasnot available to read results. Partnerphysicians, unfamiliar with the patients,would make adjustments and order addi-tional tests as a precautionary measure.Resar worked with Macken to come upwith a standard nurse-run protocol foroutpatients on Coumadin®. The result: areduction in unacceptable blood clottingmeasurements from 25 percent to 10 per-cent in patients taking the drug. Today,nurses trained in drug interactions writeall Coumadin® orders in the majority ofclinic departments, and the patients comein for lab tests at a set time. Nurses haveguidelines allowing them to adjust pre-scriptions with a requirement that theyconsult with physicians when lab resultsgo too far out of range.

Affecting theBottom LineThe payoff for these improvements is notonly better quality care but cost savings.While Luther-Midelfort administratorshave yet to measure the actual magnitudeof the savings, there is little doubt theyare significant. “We know at least 12 peo-ple were admitted for bleeding complica-tions from Coumadin® last year,” says Dr.William Rupp, the hospital’s presidentand CEO. “By making the changes in ouroutpatient clinics, we’ve had no admis-sions for that this year. Coagulation mea-surements among patients are more stablebecause of our nurse-run protocols. We’vebeen able to cut back the need for addi-tional blood tests by 30–40 percent.”

Rupp says other hospital CEOs ask himhow he can afford to have a pulmonolo-gist do administrative work. “I say wecan’t afford not to have him do it. Basedon all of his improvement projects, I esti-mate Roger saves us four to six timeswhat we pay him each year.”

Rupp and Borman also champion the“pilot project” approach to change atLuther-Midelfort. The strategy is to testthings on a smaller scale and let improve-

ment projects grow naturally as the suc-cesses pile up and hospital staff adapts.That way, says Borman, “we get very littlestaff resistance.” The change happensalmost by osmosis, he adds. “Peer pressurekicks in because the ideas really work.”

Innovations in patient safety are mov-ing faster and faster through the Luther-Midelfort system, and momentum isbuilding for larger systemwide qualityimprovements. The Coumadin® projecttook 18 months while the sliding scaleinsulin project took only two months toroll out. The “medication reconciliation”process aimed at the interfaces of care isnow in place in 80 percent of the hospi-tal. The process has evolved through 14iterations. Most recently, Resar has givenhimself three months to implement a listof innovations recently devised by theMassachusetts Coalition for the Preven-tion of Medical Errors. These involve set-ting up educational review policies fororal medications given intravenously andthe color coding of different drug deliverysystems, among others.

And after a year-long effort that start-ed out in just one department, a “non-punitive” error reporting system is inplace hospital-wide. As predicted, reportsof errors from nurses and technicianshave risen dramatically—up seven-fold inthe first month after the policy wasimplemented in the spring of 1998. SaysRupp, “We have worked hard at changingthe cultural fear of punishment. Now wetell people they’ll get in trouble if theydon’t report an error—and they have a 48-hour period to let us know if somethinghas gone wrong.”

Resar says he has the proverbial “jobthat is never finished.” But his enthusiasmfor the work—and the mission—is un-diminished. In years past, “we werebenchmarking ourselves against medioc-rity,” Resar says. “Now, we are trying tocreate an organization that’s among thesafest in the country. I think we can do it.”Then he launches into a list of future pro-jects: standardized protocols for infusionpumps, better post-operative pain proto-cols, and simplified forms to track “nearmisses” on medication errors. “Some ofmy colleagues think I’m wasting my edu-cation because I’m not ‘doing’ medicine,”says Resar. “But I know I’m just takingcare of a larger population of patients.”

20

Like a growing number of hospitalsnationwide, staff and administrators atFairview Health Services, a system ofseven hospitals based in Minneapolis,became increasingly aware in the mid-1990s that too many patients were gettingthe wrong drugs or the wrong dose of adrug. Steve Meisel, assistant director forclinical pharmacy services at FairviewSouthdale Hospital, says it was not somuch an epiphany as a dawning con-sciousness and conviction that the prob-lem had to be made a priority.

In 1996, the hospital’s leadershiplaunched a program to reduce the risk ofdrug errors. After three years, the result hasbeen a series of innovations aimed at cre-ating an integrated medications manage-ment system with much tighter controls,fewer handoffs, and increased accountabil-ity. To do this, Fairview brought togetherdoctors, nurses, pharmacists, and qualityimprovement experts to focus on the entirecontinuum of drug delivery.

“The chief thing we had to communi-cate to people at first was that they maybe doing their job very, very well but thatthe whole process was just not working,”says Denise White, a quality consultant atFairview. “We had to get everyone to stepback far enough to see how they work inthe whole context.”

The first step, both White and Meiselsay, was to identify clinical areas, diseases,and medicines where the problems loomedlargest and where pilot projects would

have the most impact. A multi-disciplinaryteam ended up creating 15 separate pro-jects. All started off on a small scale sothey could be implemented quickly andefficiently. The primary testing groundbecame Fairview Southdale, a 300-bedhospital in the twin cities suburb of Edina.

The projects involved medications asdiverse as anticoagulants, chemotherapy,insulin and sleeping pills. They alsoexplored ways to adjust drug doses forpeople with abnormal kidney functionand took patients out of harm’s way bysimply removing some of the most dan-gerous drugs from areas where they werereadily available.

All the projects were grounded in acommon—and common sense—premise:make it more difficult for staffers at everystage of the process to make a mistake.

A crucial choice was made early onand may well prove a popular model forother hospitals. The hospital system’spharmacy was selected to spearhead andoversee the initiative. The choice is not anobvious one. While pharmacists have longplayed a key role in hospitals, their exper-tise has often been underutilized, andsome doctors are still reluctant to cedepharmacists control over patient care.

The decision in this case was based onthe judgement that pharmacy had boththe needed expertise and organizationalskills and a strong rapport with variousdepartments throughout the system. Thoseties had been forged over 20 years by the

21

Fairview Health ServicesMinneapolis, Minnesota

BY SHELLEY REESE

Focusing on themanagement ofa few high-riskdrugs, a Minnesotahospital systemcut medicationerrors and sharplyimproved care

MISTAKESMISTAKESHarder to Make

Making

presence of pharmacists at nursing sta-tions throughout Fairview hospitals—apractice common in Minnesota but notthroughout the country.

The Warfarin Project Among the first changes FairviewSouthdale initiated was to reorganize themanagement of a drug called warfarinamong cardiology patients. Warfarin, ananticoagulant, is prescribed for a varietyof conditions to prevent and treat bloodclots. It’s a difficult drug to manage be-cause numerous factors—including diet,alcohol use, and other medications—canaffect the amount of drug in the body.

Patients must be carefully educatedabout warfarin and factors that can affectits use. Frequent monitoring of bloodtests is also necessary to assure the drugis working properly and to prevent poten-tial problems. These problems includesevere and potentially life-threateningbleeding and clotting.

The warfarin project was prompted inthe fall of 1996 when data showed that anaverage of 1.7 patients per week—almost90 a year of the roughly 4,000 taking thedrug—were being admitted for bleeding.The majority, 70 percent, of the admis-sions were thought to be due to an errorin monitoring or dosing, or to inadequatefollow-up. Warfarin was thus an idealcandidate for more intense management.

The project was first launched as a six-month pilot for 135 patients of theMinnesota Heart Clinic, an independentcardiology practice affiliated with thesuburban hospital. A Fairview pharmacistwas assigned to educate patients on war-farin, oversee dosing, and follow-up withthose patients who had been started onthe drug in the hospital.

The program yielded benefits from dayone. Cardiology patients with insufficientmonitoring fell from 22 percent before thepilot to less than one percent. Blood testsshowed the share of cardiology patientswith the correct dosage of warfarin in theirbloodstream increased from 35 percent to65 percent. As a result, warfarin complica-tions per 100 cardiology patients fell from12 to two. In addition, patient knowledgeabout the drug, measured by correctanswers on a test, improved from a base-line of 30 percent to over 80 percent.

“The results were dramatic to the pointthat I almost didn’t believe them,” saysDr. William Hession, president of theMinnesota Heart Clinic. “It was prettyhumbling to see.”

The initiative’s success is largelyattributable, says Meisel, to tracking andfollow-up of patients. But he says it’s alsodue to the staff’s acceptance and embraceof guidelines that were developed in con-junction with the Minnesota Heart Clinic.Two full-time pharmacists, Melissa VanHolland and Caren Allivato, ran the pro-gram from within the Minnesota HeartClinic. (A third pharmacist has since beenadded but one is part-time.) Patients areseen face-to-face at each visit to identifypotential problems, make adjustments intheir therapy, and to provide ongoingeducation about their drug therapy.

Says Hession: “If you look at why peo-ple fail on warfarin, it’s because they getlost. Patients figure, ‘if I’m not bleedingand the doctor is not bugging me, I mustbe OK’ and they don’t get their tests. Nowwe have a database, and we know whoshould be coming in for blood studies.Patients are getting watched more closely,and that is really paying off.” He sayspatients also better understand the effectsdiet and drug interactions can have onthem and learn to avoid potentially dan-gerous situations.

Both Fairview and the MinnesotaHeart Clinic were so impressed withresults that they extended the warfarinprogram to other patient populations. Thehospital made the service available to itsorthopedic patients and vascular patients,for example. These patients are often pre-scribed warfarin after surgery. Resultshave been similarly promising. After thepharmacists took over the orthopedicpatients’ warfarin management, the num-ber of blood clot complications for thegroup fell from an average of 7.6 per yearto 4. Today, Van Holland and Allivato,and new colleague Kim Graupe, managewarfarin therapy for about 400 patients.

But warfarin is not the only drug tocome under Fairview’s microscope. Thehospital also established dosing protocolsfor sleeping pills and for patients withimpaired kidney function. And the projectteam developed a chart with a slidingscale for determining correct insulindosage. All three programs were piloted

over a period of months and then adopt-ed at some or all of Fairview’s hospitals.

Reorganizing drug management ofpatients with impaired kidney functionproved particularly successful. Usingdosage guidelines developed by the hospi-tal’s Pharmacy & Therapeutics Committee,pharmacists were given permission toindependently change physician-pre-scribed doses of drugs for patients withimpaired kidney function. The purpose: toreduce the risk of toxicity associated withtoo high a dosage, a common problem.About 70 doses now are being changedeach month. Physicians are notified.

Fairview Ridges Hospital developed asimilar protocol for sleeping pills, such as triazolam (Halcion®) and temazepam(Restoril®), to reduce the risk of overmed-ication. Too high a dose of such drugs canheighten the risk of falls and accidents.Under the new protocol, if a physicianprescribes a dose that is too high for thepatient, the pharmacist is empowered tochange the dose.

In a related step at Fairview Ridges,all patients considered to be at risk offalling are identified on admission.Nurses then develop plans to preventfalls, for example, by assisting at-riskpatients to the bathroom.

Meanwhile, the team at FairviewSouthdale tackled the issue of drug errorsdue to miscommunication. It developed a

22

pre-printed chart with a sliding scale forinsulin dosing and a corresponding pre-printed form for recording medicationadministration information. The aim was toreduce errors caused by illegible handwrit-ing, improper abbreviations, overlappingblood-sugar ranges, and failure to treathigh blood sugar levels in a timely fashion.

The hospital system also sought totighten controls on drug prescriptions by changing the way information isrecorded. For example, at Fairview Southdale chemotherapy order formswere redesigned to prompt practitionersto record the level of detail necessary tosafely prescribe, dispense, and admin-ister these powerful drugs. Under the previous system, doctors could placeorders verbally or over the telephone. The new template provides consistencyand reduces the opportunities for poten-tial errors. Fairview Ridges Hospital hassince adopted the system, as well.

“We tried to design the template toreflect exactly the way people thoughtabout dosage,” said Paula Welford, anurse manager formerly in the oncologyunit. “So there is no chance of mistakinga one-time only dose for a daily dose.”

The same logic was used to developnew procedures for copying informationonto patient medication records. Byestablishing criteria to govern the wayinformation is recorded—such as the useof abbreviations and brand versus gener-ic names—Fairview Ridges reduced thenumber of places information might bemisinterpreted by 60 percent, says Meisel.

Finally, Fairview implemented a keychange to reduce the likelihood that high-

hazard drugs might mistakenly be admin-istered. At all Fairview hospitals, suchmedications were simply removed fromfloor drug carts stock. The move reducedthe number of stock items in intensivecare units by about 30 percent. In addi-tion, to prevent patients from sufferingdangerous reactions to drugs such asastemizole, an antihistamine, and oralketorolac and metformin, which are usedto treat arthritis and diabetes, respectively,they were simply removed from the hospi-tal system’s pharmaceutical drug formula-ry. With the exception of metformin, lesshazardous medications were substituted.

So far, the success of Fairview’s vari-ous programs are confined to distinctpockets of its complex care systems. Andthere is no single, overarching statisticdescribing how much patient safety hasimproved, says Meisel. It is impossible tomeasure progress in such a tidy way, hesays, because there is no baseline figureto show how often mistakes are made.

“If you ask me what’s the overall med-ical error rate in our hospital, I would nothave a clue,” Meisel says. “It’s not likemeasuring the mortality rate after bypasssurgery. You don’t always know when anerror occurs, and people don’t alwaysagree on what constitutes an error.”

Fairview’s successes have not beenwithout accompanying failures. Somepilot programs have been droppedbecause of lack of staff support. Forexample, a measure intended to ensurepatients are not being over-medicatedwould have required doctors to reviewtheir orders prior to a patient’s transferfrom a critical care unit. That initiative

was dropped because staff objected to theadditional responsibility.

More troubling, the long-term viabili-ty of some programs now in placeremains uncertain. The warfarin project,for example, relies on tenuous financing.It costs $175,000 a year to run, Meiselsays, with expenses comprised almostentirely of the two pharmacists’ salariesand a portion of a secretary’s salary.Initially, the hospital funded the entireprogram. But in January 1998, theMinnesota Heart Clinic agreed to begincovering the costs of its increased usage.The cardiology component of the serviceruns about $100,000 per year.

Still, the doctors, patients and Fair-view end up paying for some of the cost:Medicare only pays a portion for face-to-face visits and does not reimburse forphone consults or follow-up. The hospitaland Minnesota Heart Clinic bear this cost.“The onus falls on the wrong parties,”said Hession. “Right now, none of theinsurance companies are willing to cometo the table to fully fund this kind ofservice.” He says he hopes insurancecompanies will eventually come to see thecost effectiveness of warfarin manage-ment as a preventive service.

Despite such financial uncertainties,Meisel is brimming with other ways toreduce medical errors. He would like, forexample, to expand the more intensivemedical management of warfarin to neu-rology and primary care patients and,eventually, to the entire patient popula-tion. And he’d like to apply similar proce-dures to drugs such as amiodarone, whichis prescribed for cardiac rhythm disorders,and digoxin, which is prescribed for heartfailure, and other high-hazard drugs.

Finally, Meisel would like to see thehospital invest in technology to automateand streamline drug dispensing with com-puters and lower the risk of errors with abar-coding system that tracks all drugorders. Having witnessed the results ahandful of small-scale initiatives can yield,he’s eager to tackle more ambitious mea-sures. “When you think of the contributionpharmacy can make to improving safetyand how far things can go, it’s enormous,”Meisel says. “There are several hundredmore things we’d like to try. We’d like toget to the point where no patient againwould be the victim of a medication error.”

23

Dr. William Hession, left, Minnesota Heart Clinic’s president, reviews a patient’s chart, while CarenAllivato, above, a Clinic Pharmacist, works with a patient taking warfarin. (Photos by Jill Neu)

In the operating room, things are notgoing well. An electronic monitor record-ing a patient’s heart rate registers a sharprise from 70 to 100 and then to 160 beatsper minute. His blood pressure is crash-ing. His airways spasm. Warning lightsflash red. Alarms sound.

All in a few minutes, routine kneesurgery has turned lethal. The anesthesi-ologist reaches for a syringe, fumbles,then drops the vial. Time is running out.The doctors scramble to figure out what isgoing wrong. They are acutely aware thatthe combination of rapid heart rate andplummeting blood pressure could kill aperson in minutes.

But fortunately, not this patient. Lyingbefore this medical team is literally adummy, a simulated emergency carepatient made of plastic, wires and comput-er chips. His formal name is “PatientSim,”but he usually goes by Sam, Johnny, orJessica (add a wig and some other featuresand he becomes a she). Even if Sam’s doc-tors made all the wrong decisions, hewould survive; a simple touch of a “reset”button brings him back to life.

But the doctors who are tending toSam are keenly aware that they’ll nothave that luxury with a flesh and bloodhuman being. They are “practicing” onthe simulated patient to learn how toavoid and overcome medical errors and torespond in crisis situations—where studiesconsistently show the risk of medicalerrors is particularly high.

Since its creation in 1986, based onwork by Dr. David M. Gaba, Director ofthe Patient Safety Center of Inquiry at the

VA Medical Center in Palo Alto, CA, anda professor of anesthesia at Stanford, thePatientSim simulator has become a cen-tral learning tool at the VA/StanfordSimulation Center for Crisis ManagementTraining in Health Care. Both PatientSimand the Center were inspired by Gaba’sfirm belief that the failure of doctors tolearn to cope with urgent crises and reactin a systematic, timely way is a majorfactor in preventable anesthesia andoperating room medical mishaps.

Early forerunners of PatientSim at theVA/Stanford Center were the nation’s onlyhi-tech human simulators until just a fewyears ago. But PatientSim now has some150 “friends” in hospitals around the worldmade by two competing manufacturers.The growing number has occurred becausesimulator-based training has graduallygained an important foothold in medicaleducation. And with interest in reducingmedical errors and improving safety risingrapidly, medical safety experts predictmedical simulation training to spread morewidely in coming years.

The technology is modeled to a largedegree on the experience in aviation.After World War II, the rapidly expandingcommercial airline industry knew it had asafety and liability problem on its hands.Simulators were one way the industrysought to enhance training and reducethe risk of pilot error—which had risensharply as planes got more and morecomplex. Today, airline pilots are requiredto learn to fly in a simulator before theytake the controls of a real plane. Hours ofadditional simulator time are required to

keep skills sharp. And pilots frequentlylearn to use new equipment and fly unfa-miliar planes in a simulator.

Simulators are credited with playing akey role in the enormous reduction in aircrashes and fatalities over the past fewdecades. Simulation-based training is alsoused in automobile driving, shipping,military command and control, and theoperation of nuclear power plants. Gababelieves simulator training in health carecan and should play a similar role.

Like pilots, he says, surgeons and anes-thesiologists can face crisis events that areinfrequent but potentially lethal. Bothpilots and doctors must absorb a prodi-gious amount of technical knowledge. Andthe bias in their education and trainingleans heavily towards mastering a definedand complex set of skills. But unliketoday’s pilots, doctors are often poorlytrained in “crisis management”—how torespond when something goes wrong,sometimes terribly wrong, Gaba says.

Many anesthesia mistakes, for example,are caused by technical problems, such asincorrect administration of a drug. Butmany others are caused by behavioralproblems and poor communication amongdoctors, nurses, and technicians. For exam-ple, a doctor barking an order into thin airin an operating room could go unheeded ifit is not addressed to a specific individual.

Gaba also thinks the system trains doc-tors poorly to do new tasks. It is notuncommon for a doctor performing a newtechnique or using a new piece of technol-ogy to try it out for the very first time ona real patient. That, Gaba says, is unwise.

24

VA/Stanford Simulation CenterPalo Alto, California

Simulation technology, like that

used to train pilots, can help

doctors learn from their mistakes

BetterMedicine

A DummyBetter

Medicine

A DummyHELPS PAVE THE WAY TO

BY L ISA KRIEGER

In the practice of anesthesia, Gabanotes that deaths from errors havedeclined sharply over the last 30 years(see story page 26). But “we need to, andcan, make it even safer,” he says. Deathsfrom anesthesia errors have declined to as low as one per 100,000 to 200,000“healthy” patients (those whose physicalcondition can not have contributed to thedeath). But the risk of dying in a domes-tic commercial jet flight in the U.S. is far,far lower—about one in 8 million.

Gaba says the idea for the patient sim-ulator came to him when he was readingthe 1985 book Normal Accidents by Yalesociologist Charles Perrow. The book ana-lyzes the social and human side of tech-nological risk. It concluded that systemcomplexity, and the human response toemergency, make error and failureinevitable in many high-risk industries.

The relevance to anesthesia and medi-cine was clear to him immediately, Gabasays. “I wanted to test how anesthesiolo-gists respond to unexpected events. Andit dawned on me that we could buildsomething to measure and monitor this.”

But building a patient simulator posedchallenges for Gaba not faced by thedesigners of flight simulators. An artifi-cial environment that mimics the humanbody is enormously more complicatedthan simulating flight. The human bodypresents thousands of variables that makeeven fickle weather conditions seem easy.

The first electromechanical anesthesiasimulator had been developed in the late1960s. Though ahead of its time, themachine was flawed and drifted into obliv-ion. In 1986, Gaba and a colleague begandeveloping the first generation of realisticsimulators, using new computer capabili-ties to provide signals to actual clinicalinstruments. This model was improvedwith a modified mannequin which allowedmask ventilation, intubation and monitor-ing of breath sounds. But it was still rela-tively crude, with no pulse, spontaneousbreathing or human-like physiology.

A major redesign in 1989 incorporat-ed a physiologic model of the cardiovas-cular system. Then in 1992, the CAE LinkCorporation, a manufacturer of militaryaviation and space flight simulators,licensed the technology from Gaba’steam to develop a commercial patientsimulator. Its mannequin contained a

complete model of cardiovascular, pul-monary, fluid, acid-base electrolyte andthermal physiology. It includes computer-controlled electromechanical lungs, heartand breath sounds, changeable airwayanatomy, and a palpable pulse.

HOW THES IMULATOR WORKSThe VA/Stanford Simulation Center,opened in 1995, looks and feels like a fullyfunctional operating room, complete withassorted high tech medical equipment,surgical lights, and tension in the air. Anadjacent control room contains a comput-er that executes all Sam’s “physiologic”functions—from the cardiovascular andpulmonary systems, to metabolism, fluidand electrolyte balance, and thermal reg-ulation. For example, the system moves“mathematical blood” from one heartchamber to the next and from blood ves-sel to blood vessel—a process requiringthousands of computations every second.

The simulator can mimic dozens ofcomplex and realistic clinical crisis sce-narios. And instructors can modify thesescenarios in hundreds of ways. Thus, Samcan be programmed to act like a healthyyoung truck driver or a frail 80-year-oldwoman. Instructors can induce a heartattack, respiratory trouble, an allergicreaction to medicines, and other kinds ofdistress. For each event, myriad settingsare possible to mimic severity and respon-siveness to treatment. And reactions todrugs and procedures vary with each com-puter-programmed change in physiology.

The simulator is programmed to showthe same distress symptoms as a humanpatient. For example, a working bloodpressure band is strapped to Sam’s bicepsand an IV is placed in one arm. His pulsecan be felt at both wrist and neck. Theheart beat can be heard with a stethoscopeand the heart rhythm can be made to beabnormal. Inside Sam’s mouth are teeth,tongue, uvula and windpipe. Electro-mechanical, computer-controlled lungs areembedded in its chest; they “breathe” spon-taneously, as well as by hand or mechani-cal ventilation. Incoming gases are detect-ed and quantified and their concentrationsfed to the physiologic and pharmacologicalcomputer models. Oxygen sensors ensurethat if Sam quits breathing, alarms go off.

Most amazingly, Sam is programmed torespond as a healthy human would to morethan 70 medications. When he gets an anes-thetic, for example, his eyelids stop movingand his breathing slows; doctors mustquickly stick a tube down his windpipe togive him air. If the tube does not go downcorrectly and he fails to get enough oxygen,his heartbeat speeds up and carbon dioxidemakes his blood pressure rise. By mimickingan actual patient during surgery as much aspossible, doctors know quickly if they’veinjected the wrong drug, erroneously inter-preted vital signs, or made other errors.

After every simulator session, doctorsand nurses can watch their performance onvideo monitors in a debriefing room. Thesesessions, Gaba says, provide profoundinsight into glitches in the system that canlead to patient harm.

“It’s not about finding bad doctors,”Gaba says. “It’s about helping doctors learnto work together better to prevent prob-lems and errors.”

The simulator also is being used forresearch. Gaba and his team have studieddecisions made by intensive care unit clin-icians, comparing their responses to thoseof a prototype computer-based decisionsupport system. They also are measuringthe influence of fatigue on cognition.

“Within the next few years, we canexpect to see continued, incrementalimprovement of simulator technology,”Gaba says. Beyond that—in 10–20 years—the next generation of simulators willlikely use virtual reality technology toreplicate the operating room’s environ-ment. And beyond that...well, anything ispossible. It’s likely that simulators will beused broadly in training and perhaps evenbecome so commonplace that they allowdoctors to practice particularly trickyoperations before they do them.

Already the crisis management trainingpioneered in anesthesiology has beenextended to delivery room crisis andneonatal resuscitation, emergency roomand trauma care, intensive care medicine,cardiac arrest response (“code”) teams, andother health care settings. Although Gababelieves that simulators will improve med-ical safety, he’s quick to say they are com-plementary to, not a replacement for, con-ventional means of medical education.After all, for all his marvelous technicalsophistication, Sam is still just a dummy.

25

It’s just a small device, resembling aplastic clothespin with some wires thatconnect to a monitor about the size of aportable phone. Without looking closelyyou might think it was part of an elec-tronic game. But the elegantly simpledevice is among the most important in medicine. It determines whether apatient under general anesthesia isreceiving enough oxygen—oxygen thatwill keep the patient alive through theprocedure and, if all goes as expected,permit her to awaken in the recoveryroom with no ill effects.

Oxygen, of course, is vital to all livingtissue. In decades past, oxygen depriva-tion while a patient was under anesthesiawas one of the most common errors inmedicine. Lack of oxygen, or hypoxia,occurred most frequently because thetube that was supposed to deliver the pre-cious gas through the trachea to the lungs

was placed by mistake into the esopha-gus. It’s an “operator error,” and when ithappens, irreversible brain damage anddeath can occur in minutes. In the past,by the time the mistake was detected, itwas usually too late. But the clothespin-like device, developed in the early 1980s,discerns the levels of oxygen circulatingin the patient’s bloodstream.

The oxygen pulseoximeter, as thedevice is called, is one of dozens ofimportant advances in the field of anes-thesiology in the past 20 years. Whatmakes the field unique is that many ofthe advances were either specificallyaimed at reducing the risk and rate oferrors, or did so as an important sec-ondary benefit. As a result, undergoinganesthesia has gone from being one ofthe most notoriously dangerous proce-dures in medicine to one of the safest.

In the 1950s, the death rate from anes-

26

Deeply disturbed by patient deaths and facing

skyrocketing malpractice premiums, the field of

anesthesia became the first to seriously tackle

medical errors. It’s now a model for others.

BY JAN ZIEGLER

A MED ICAL SPEC IALTY

BLAZESA TRAIL

thesia was roughly one in 3,000 to 4,000.By the 1970s, it had come down to rough-ly one in 10,000, says John Eichhorn,M.D., professor and chairman of anesthe-siology at the University of MississippiMedical Center. But by 1990, Eichhorn’sown meticulous research among patientsin Boston found just one non-fatal mis-take among 392,000 patients undergoinganesthesia. Other recent data indicate arate of one death per 200,000 to 300,000anesthetics administered.

“It has been a true revolution in healthcare,” says Eichhorn.

That revolution and the field of anes-thesia is now being held up as an exam-ple for other medical specialties, andmedicine in general.

Most notable, say experts, is that anes-thesiologists—led by their professionalgroups—went about making the necessarychanges themselves. They set strict inter-nal standards and organized initiatives.No outside government body—state orfederal health authorities, Congress, orthe American Medical Association—madethem do it. There was of course another,baser motive: malpractice liability. Andeventually, some in the profession alsopushed for state regulations.

The early yearsAmong the early leaders of this effort,Ellison “Jeep” Pierce is widely consideredto have been particularly indispensable.His Spencer Tracy-like appearance andmanner inspired allegiance and trust. His occasional gruffness hid compassion-ate commitment to the improvement ofmedical practice. What’s more, Pierce isstill involved in the fight. He servestoday, at 71, as executive director of theAnesthesia Patient Safety Foundationbased in Boston.

Pierce began his residency in 1954 atthe Hospital of the University ofPennsylvania. “Our practices back thenwere abysmal,” he says. “I can rememberwe had one cardiac arrest a week as a direct result of the anesthesia. Wewould go to the family and say, `We’resorry, old Joe just couldn’t take it.’ Thatwas it. They accepted it.”

Pierce says it was common in thosedays, for example, for anesthetists to be

allowed to leave the operating room for acup of coffee after the patient was“under.” Ether was the primary agentused, despite the availability of othergases. They monitored the patient bysimply observing signs such as breathingrate and pulse. “It was really very rudi-mentary,” Pierce says.

In 1960, Pierce became vice-chairmanof anesthesia at Peter Bent BrighamHospital in Boston. “As I collected casesover the next 10 years,” he says, “itbecame obvious there were a lot of deathsthat could be prevented.”

The public was beginning to surmisethat as well. No longer willing to acceptthe “old Joe” explanation, people beganto sue anesthesiologists for malpracticein record numbers in the late 1970s andearly 1980s. As a result, malpracticepremiums for anesthesiologists sky-rocketed, becoming among the mostexpensive in medicine.

The media poured fuel on this fire. The risks of anesthesia became a commonstory in print media, and television wasnot far behind. In 1982, the ABC news-magazine show 20/20 aired a now-famous segment highlighting shoddyanesthesia practices at several institu-tions. The show warned viewers that6,000 would die or experience brain damage that year because of avoidableanesthesia error.

Pierce and others knew a major effortwas now in order or the speciality wouldsuffer long-lasting damage. In 1983Pierce became the vice-president of theAmerican Society of Anesthesiologists(ASA). He used his position there—bolstered by the gathering outrage overmalpractice premiums—to establish theASA Committee on Patient Safety andRisk Management. The group set stan-dards of care and safety and used everymeans available to get the word out—including the production of a series ofeducational videos that remain popular.Two years later in 1985, Pierce and otherleaders organized the first InternationalSymposium on Preventable AnesthesiaMortality and Morbidity. The meetingwas a turning point for the wider adop-tion of more sophisticated patient safetymeasures in the field. The creation in1984, of the Anesthesia Patient SafetyFoundation also added momentum.

At the same time, outside forces wereimpinging. Malpractice insurers were tak-ing a proactive stance towards anesthesiaerrors. Eichhorn says the major malprac-tice carrier for the nine Boston hospitalsaffiliated with Harvard Medical Schoolmade the rounds of anesthesia depart-ment chiefs in the mid-1980s. Their mes-sage was simple and blunt, he says: “Youguys are costing us too much money.”

Spurred by the insurer’s warning,Pierce, Eichhorn and others helped formthe Harvard Department of AnaesthesiaRisk Management Committee. The groupembarked on creating a comprehensiveset of practice standards for the field.They would be the first ever. The stan-dards were issued in 1985 and publishedin the Journal of the American MedicalAssociation in 1986. They quickly gainedbroad currency and remain today thecore set of practice standards in thefield—widely adhered to. Regular updateshave taken place. For example, pulseoximetry, the mechanism that uses asmall clip or bandage-like device on fin-gertip, ear, nose or toe to measure oxy-gen in the blood, became part of thestandards in 1990. Capnography, a com-panion device that measures blood levelsof carbon dioxide, was added to the stan-dards in 1991.

Redesigningthe machineryEven as doctors were finding better andsafer ways to use the technology at theirdisposal, engineers like Peter Schreiberand Jeffrey Cooper were trying to makethe technology more user-friendly.

Schreiber, who moved to the U.S. fromGermany in 1970 to start his own compa-ny, says the equipment even in the 1970swas “very imperfect.” It left too muchroom for operator error and machine fail-ure. The machines consisted of a meteredgas cylinder and a vaporizer for the othergas or gases that would be administeredconcurrently, such as nitrous oxide oroxygen. There was no way to measure thepatient’s responses to the anesthesia.Anesthetists simply observed their breath-ing and appearance and took their pulse,much as as they had in the 1850s.

27

Schreiber knew that monitors wereneeded to keep track of cardiac, lung andother functions. But he says there wastremendous initial resistance to this idea.Anesthetists at the time considered whatthey did an art form. Many, he says, flatly told him: “I do not trust any mechanism or electrical equipment that isbetween me and my patient.”

Schreiber’s most important insight—and one that resounded later throughmuch of medical equipment design—wasthat humans were not wired well to mon-itor “boring events” for hours at a time.And he defined surgery as a boringevent—from the anesthetists point of viewat least. Attention can easily slip duringhours of watching a patient.

His argument slowly began in the1970s to make a dent. Manufacturersrolled out a plethora of monitors thatwould provide information on bodilyfunctions. Then, realizing the sheer num-

ber of these devices was impractical, theydesigned a system that contained all thenecessary monitoring components.

Cooper, meanwhile, was focused onanother element of anesthesia technol-ogy. As a bio-engineer at MassachusettsGeneral Hospital in the 1970s, he beganto use a method popular in the field ofaviation safety—called critical incidenttechnique—to evaluate the interfacebetween machine and human in anesthe-sia. He did not like what he saw.

“The equipment was no longer thebiggest problem; it was all this otherstuff,” says Cooper, now chairperson ofbio-engineering at Massachusetts Gen-

eral Hospital and Brigham and Women’sHospital in Boston. The “other stuff” wasthe human error involved in using thetechnology. His research demonstratedthat inadequate experience and familiar-ity with equipment were major factors inanesthesia errors. His ground-breakingpapers, published in 1978 and 1984, ledto training programs that are widelybelieved to have played a major role inanesthesia error reduction.

It’s the lawThe latest step in this evolution began adecade ago. Even as anesthesiologistswere eagerly embracing the new equip-ment and adhering to voluntary stan-dards, some felt more was needed. Oneof these people was Ervin Moss, M.D., apracticing anesthesiologist in New Jerseyand New York for 45 years. Moss

believes voluntary standards are com-mendable and useful, but lack the teethneeded to go the rest of the way towardsreducing anesthesia errors.

He began lobbying heavily for staterules in the mid-1980s. His first opportu-nity came when his own state of NewJersey set about changing its medicallicensing regulations in 1988. Mossinsisted that anesthesiology be governedseparately. The result was passage of theAnesthesia Regulation Bill. The new law,which covered any facility with two ormore surgical operating rooms, had swiftimpact. Three hospital chiefs of anesthe-siology were forced to resign because one

new rule in the law required board certi-fication for the post. Hospitals weregiven six months to junk out-of-datemachines. If they didn’t, they had toclose the operating room. All anesthesia-related deaths and other adverse eventshad to be reported to the state healthdepartment by hotline immediately, andin writing in 30 days. Failure to complycarried a number of penalties, includingloss of license.

But Moss did not stop there. In theearly 1990s, he pushed for regulationsin New Jersey that would govern hospi-tals and facilities—primarily doctor’soffices—with just one operating room.“They were hiding behind the idea thatthey were simply offices,” Moss says.“But a lot of surgical procedures aredone in a doctor’s office.”

It was a long fight but Moss, again,ultimately prevailed. The New Jersey leg-islature in June 1998 passed a law regu-lating office-based anesthesia and surgery.

But New Jersey is so far the only statewith such a law on the books. Californiahas regulations but they are not nearly asstrong, Moss says. A few other states arelooking at the issue. Moss believes amajor push from anesthesiologists andconsumers will be needed to get the issueon to the front burner.

Meanwhile, the anesthesia professioncontinues to wage an aggressive cam-paign to further reduce errors. One initia-tive has yielded important data. It’s calledthe “Closed Claims Study.” It originated atthe University of Washington in Seattle in1985. The project uses data from settledmalpractice claims to analyze anesthesiaaccidents. Among its findings to date arethat death and brain damage represent adeclining proportion of malpracticeclaims—31 percent in the late 1990s,down from 56 percent in the 1970s.Inadequate ventilation represented 22percent of all claims in the 1970s but justseven percent in the 1990s. Pierce saysthe data are more concrete documenta-tion that the most serious, and fatal, kindsof anesthesia accidents have declined.

That these event still occur at allthough, is very disturbing, Pierce says.“We can not rest on our laurels. That’s thebottom line,” he says. “There will alwaysbe room for more improvement, and wesimply must achieve it.”

28

“We can not rest on ourlaurels. That’s the bottom line. There will always be room formore improvement, and we simply must achieve it.”

SELECTEDORGANIZATIONS

National Patient SafetyFoundation (NPSF)515 N. State Street, 8th FloorChicago, IL 60610PHONE 312-464-4848FAX 312-464-4154EMAIL npsf@ama-assn.orgWEBSITE www.ama-assn.org

The NPSF maintains a website that provides links toa variety of sites with information on patient safety.

Institute for Safe MedicationPractices (ISMP)300 West Street RoadWarminster, PA 18974PHONE 215-956-9181FAX 215-956-9266

ISMP provides an independent review of medica-tion errors that have been voluntarily submitted by practitioners to a national Medication ErrorsReporting Program (MERP) operated by the U.S.Pharmacopeia (USP).

US Pharmacopeia (USP)12601 Twinbrook ParkwayRockville, MD 20852PHONE 800-4-USP-PRN (800-487-7776)WEBSITE www.usp.org

USP Medication ErrorsReporting ProgramPHONE 800-23-ERROR (800-233-7767)

National Coordinating Councilfor Medication Error Reportingand PreventionPHONE 800-822-8772

USP has a Medication Errors Reporting Program.USP’s Practitioner Reporting Network (PRN) publishes the USP Quality Review, which often discusses medication error topics.

National Patient Safety PartnershipOffice of the Under Secretary for HealthDepartment of Veterans Affairs810 Vermont Avenue NWWashington, DC 20420PHONE 202-273-5807FAX 202-273-7090

VHA National Center for Patient Safety2215 Fuller RoadAnn Arbor, MI 48105PHONE 734-930-5890FAX 734-930-5899

BOOKS

Bogner, M.S., ed. Human Error inMedicine. (Hillsdale, NJ: LawrenceErlbaum) 1994.This collection of presentations from a symposiumon errors in medicine provides a good overview ofmany aspects of health care errors. A number of thekey contributors to current thought concerningerror in medicine are represented.

Cohen, M. Medication Errors. (Washing-ton, DC: The American PharmaceuticalAssociation Foundation) 1999.This book provides an overview of the types of medication errors, processes that lead to errors,research on medication errors, and suggestionsfor minimizing the potential for future errors.

Corrigan, J.; Kohn, L.; and Donaldson,M., eds. To Err Is Human: Building aSafer Health System. Committee onQuality of Health Care in America,Institute of Medicine. (Washington, DC:Institute of Medicine) 1999.The report lays out a comprehensive strategy for government, industry, consumers, and healthproviders to reduce medical errors, and it calls onCongress to create a national patient safety centerto develop new tools and systems needed toaddress persistent problems. Each chapter of thereport contains a reference list, allowing the readerto select additional material based on a specificarea of interest.

Kahneman, D.; Slovic, P.; and Tversky,A., eds. Judgment Under Uncertainty:Heuristics and Biases. (New York, NY:Cambridge University Press) 1982.The classic collection of studies of how humansmake judgments under (usual) conditions of uncertainty. The authors coined some of the keyterms in behavioral science.

Leape, L. L.; Kabcenell, A.; Berwick,D.M.; and Roessner, J. Reducing AdverseDrug Events: Breakthrough Series Guide.(Boston, MA: The Institute for HealthcareImprovement) 1998.Institute for Healthcare Improvement Guides arebased on the real-life experiences of health careorganizations that have made dramatic changeswhile participating in a Breakthrough SeriesCollaborative. Each contains a synopsis of theCollaborative goals and results, the model for accelerating improvement, change concepts thatCollaborative participants used successfully, addi-tional resources, key contacts and a bibliography.The book is based on a 1996-97 Collaborative.

Perrow, C. Normal Accidents. (NewYork, NY: Basic Books) 1984.A fascinating book that opens the world of complex systems and introduces the reader toconcepts of human error. The contrasts betweenthe effectiveness of error prevention in aviation and the merchant marine are striking.

Reason, J. Human Error. (New York,NY: Cambridge University Press) 1990.The “bible” of error theory, written by the leadingauthority. Reason reviews the literature and estab-lishes a theoretical framework for understandingerror. A treasury of important concepts.

ARTICLES

Alibhai, S.M.H.; Han, R.K.; and Naglie,G. “Medication education of acutelyhospitalized older patients.” J GenIntern Med, 1999, 14:610-616.The results of this study indicate that even thoughpatients perceive the education they receive fromtheir physician or pharmacist to be satisfactory,many patients still make errors when they taketheir medications.

Bates, D. W.; Cullen, D. J.; Laird, N.;Petersen, L. A.; Small, S.; Servi, D.;Laffel, G.; Sweitzer, B.; Shea, B. F.;Hallisey, R.; Vander Vliet, M.; Nemeskal,R.; and Leape, L. L. “Incidence of adversedrug events and potential adverse drugevents: Implications for prevention.”JAMA, 1995, 274:29-34.A study of 4301 inpatient admissions of medicaland surgical intensive care units at two tertiarycare hospitals. The authors concluded that adversedrug events (ADEs) were common and often pre-ventable and that serious ADEs were more likely to be preventable.

Bates, D.W.; Leape, L.L.; Cullen, D.J.; etal. “The impact of a physician computer-ized order entry system on prevention of serious medication errors.” JAMA,1998, 280:1311-1316.To evaluate the impact of computerized physicianorder entry (POE), patients admitted to three medicalunits were studied for seven-to ten-week periods infour different years. The baseline period was beforeimplementation of POE, and the remaining three wereafter implementation. During the study, the non-missed-dose medication error rate fell 81 percent.

29

Resources andSuggested Reading

Bates, D.W.; Spell, N.; Cullen, D.J.; et al. “The costs of adverse drug eventsin hospitalized patients.” JAMA, 1997,277:307-311.Analysis of the costs of adverse drug events identi-fied in Adverse Drug Event Prevention Study (seeBates et al., JAMA 1995 above), using a case-controlmethodology. Preventable ADEs cost twice as muchas nonpreventable ADEs and averaged $4,685 eachat one teaching hospital. Authors estimate thathospitals in the study are losing $2.8 million eachyear on preventable ADEs.

Bedell, S.E.; Deitz, D.C.; Leeman, D.; andDelbanco, T.L. “Incidence and character-istics of preventable iatrogenic cardiacarrests.” JAMA, 1991, 265:2815-2820.Landmark study that demonstrates that most cardiacarrests that occur in hospitals are caused by misuse of drugs and are, therefore, potentially preventable.

Bordage, G. “Why did I miss the diag-nosis? Some cognitive explanations andeducational implications.” Acad Med,1999, 74(10):S138-S143.A review article that has several purposes: 1) todescribe diagnostic errors in general; 2) presentdetails about two specific types of diagnosticerror; and 3) apply lessons learned to the educa-tional process so medical students will learn howto avoid similar errors.

Chassin, M.R.; Galvin, R.M.; et al. “TheUrgent Need to Improve Health CareQuality.” JAMA, 1998, 280:1000-5.This article explained the framework of health carequality defined as “underuse, overuse, and misuse,”and concluded that these problems exist in largeand small health care organizations in all regionsof the country.

Classen, D.C.; et al. “Adverse DrugEvents in Hospitalized Patients.” JAMA,1997, 277:301-6.This study examined the excess length of stay, extracosts, and mortality attributable to adverse drugevents (ADEs) in hospitalized patients. The authorsfound that ADEs are associated with a significantlyprolonged length of stay, increased economic bur-den, and an almost two-fold increased risk of death.

Classen, D.C.; Pestotnik, S.L.; Evans, R.S.;and Burke, J.P. “Computerized surveil-lance of adverse drug events in hospitalpatients.” JAMA, 1991, 266:2847-51.Description of the continuous monitoring systemat LDS Hospital in Salt Lake City, Utah. This is animpressive system that integrates clinical, laborato-ry, drug, and cost information to identify ADEs bothprospectively and retrospectively.

Cullen, D.J.; Bates, D.W.; Small, S.D.; etal. “The incident reporting system doesnot detect adverse drug events: A prob-lem in quality assurance.” Jt Com J QualImprov, 1995, 21:541-548.The yield of spontaneous reporting of ADEs by thehospital incident reporting system was compared tointensive, confidential data collection by means ofself-report and daily record review. Ninety-four per-cent of ADEs discovered by record review and confi-dential report were not reported as incident reports.

Dansky, K.H.; Gamm, L.D.; Vasey, J.J.;and Barsukiewicz, C.K. “Electronic medical records: are physicians ready?”J Healthcare Manage, 1999, 44:440-455.Electronic medical records are often presented as onesolution for reducing medical error. This article out-lines some of the barriers in successful acceptance of this mode of operation within the clinical setting.

Ko, R. “Adverse reactions to watch forin patients using herbal remedies.” WestJ Med, 1999, 171:181-186.This review article looks at patient safety issuesregarding the use of alternative medicines andoffers several advisory lists for clinicians whosepatients use herbal remedies.

Kronz, J.D.; Westra, W.H.; and Epstein,J.I. “Mandatory second opinion patholo-gy at a large referral hospital.” Cancer,1999, 86:2426-2435.Johns Hopkins Medical Center was the site of thisstudy. A mandatory second opinion program wasput in place, and it was found that a significantportion of initial diagnoses reviewed within thestudy had discrepancies with the conclusionsdrawn through the second review.

Leape, L. L.; Bates, D. W.; Cullen, D. J.;Cooper, J.; Demonaco, H. J.; Gallivan, T.;Hallisey, R.; Ives, J.; Laird, N.; Laffel, G.;Nemeskal, N.; Petersen, L. A.; Porter, K.;Servi, D.; Shea, B. F.; Small, S.; Sweitzer,B.; Thompson, B. T.; and Vander Vliet,M. “Systems analysis of adverse drugevents.” JAMA, 1995, 274:35-43.The landmark article on systems ananlysis ofadverse drug events. The authors define theunderlying problems and identify the “proximalcauses” of medication errors.

Leape, L.L.; Cullen, D.J.; Clapp, M.D.;Burdick, E.; Demonaco, H.J.; Erickson,J.I.; and Bates, D.W. “Pharmacist partici-pation on physician rounds and adversedrug events in the intensive care unit.”JAMA, 1999, 282:267-270.To measure the effect of pharmacist participationon medical rounds in the ICU on the rate of pre-ventable ADEs caused by ordering errors, theauthors compared between phase 1 (baseline) andphase 2 (after intervention implemented) and phase2 comparison with a control unit that did notreceive the intervention. The rate of preventableordering ADEs decreased by 66 percent from 10.4per 1000 patient-days before the intervention to 3.5after the intervention. In the control unit, the ratewas essentially unchanged during the same timeperiods: 10.9 and 12.4 per 1000 patient-days.

Lesar, T.S.; Briceland, L.L.; Delcoure, K.;et al. “Medication prescribing errors ina teaching hospital.” JAMA, 1990,263:2329-34.One of the first descriptions of prescribing errors in hospitals.

Lesar, T.S.; Briceland, L; and Stein, D.S.“Factors related to errors in medicationprescribing.” JAMA, 1997, 277(4):312-7.This study quantified the type and frequency ofidentifiable factors associated with medication pre-scribing errors. Systematic evaluation of every thirdprescribing error over a one-year period at a tertiarycare teaching hospital revealed total errors of 2103,with 696 errors meeting a study criteria of errorswith the potential for adverse patient effects.

McDonald, C.J. “Protocol-based comput-er reminders, the quality of care and thenon-perfectibility of man.” N Eng J Med,1976, 295:1351-55.Pioneering paper on the potential use of computer-ized reminders to aid medical decision-making.

McNally, K.M.; Page, M.A.; andSunderland, V.B. “Failure mode andeffects analysis in improving a drugdistribution system.” Am J Health-SystPharm, 1997, 54:171-177.Description of use of such analysis in pharmacypractice.

Melmon, K.L. “Preventable drugreactions—causes and cures.” Sem MedBeth Israel Hosp, 1971, 284:1361-7.One of the earliest papers on preventable adversedrug events. A classic.

Thomas, E. J.; Studdart, D.M.; Newhouse,J.P.; et al. “Costs of medical injuries inUtah and Colorado.” Inquiry, 1999,36(3):255-64.The authors estimate the direct and indirect costsof adverse events. Extrapolation to all hospitaladmissions in the U.S. indicates national costsfor preventable adverse events of $17 billion.

30

PR E PA R E D BY LA U R E L S I M M O N S , IN S T I T U T E F O R HE A LT H C A R E IM P R O V E M E N T

Back from the Brink:Making Chemotherapy SaferThe Dana-Farber Cancer Institute

James Conway, M.B.A.Chief Operations OfficerDana-Farber Cancer Institute44 Binney StreetBoston, MA 02115PHONE 617-632-2158FAX 617-632-2161EMAIL james_conway@dfci.harvard.edu

A Government Health SystemLeads the WayThe VA Health System

James P. Bagian, M.D., P.E.Director VHA National Center forPatient Safety (10X) 2215 Fuller RoadAnn Arbor, MI 48105PHONE 734-930-5890FAX 734-930-5899EMAIL james.bagian@med.va.gov

Making DoctorsComputer LiterateBrigham and Women’s Hospital

Gilad Kuperman, M.D.Instructor in MedicineHarvard Medical School850 Boylston Street, Suite 202Chestnut Hill, MA 02467PHONE 617-732-9069FAX 617-731-3690

David Bates, M.D., M.Sc.Division of General Internal MedicineBrigham and Women’s Hospital75 Francis StreetBoston, MA 02115PHONE 617-732-7063FAX 617-732-7072EMAIL dbates@partners.org

What’s in a Name? When itComes to Drugs—LotsBristol-Myers Squibb

Frank PasqualoneVice President, MarketingP.O. Box 4500Princeton, NJ 08543PHONE 609-897-2511FAX 609-897-6399EMAIL frank.pasqualone@bms.com

Tackling Medication ErrorsHead OnLuther-Midelfort Hospital

Roger Resar, M.D.Change AgentLuther-Midelfort—Mayo Health System1400 Bellinger StreetEau Claire, WI 54703PHONE 715-838-3566FAX 715-838-3812EMAIL resar.roger@mayo.edu

Making MistakesHarder to MakeFairview Health Services

Steven Meisel, Pharm.D.Assistant Director of PharmacyFairview Health Services6401 France Avenue SouthEdina, MN 55435PHONE 612-924-5894FAX 612-924-1511 EMAIL smeisel1@fairview.org

Caren Allivato, Pharm.D.Clinical Pharmacy SpecialistFairview Health Services6401 France Avenue SouthEdina, MN 55435PHONE 612-924-5840EMAIL calliva1@fairview.org

A Dummy Helps Pave the Wayto Better MedicineVA/Stanford Simulation Center

David Gaba, M.D.Director, Patient Safety Center of Inquiryat VA Palo Alto Health Care SystemProfessor of AnesthesiaStanford University School of MedicinePatient Safety Center of Inquiry, 112A/PSCVA Palo Alto HCS3801 Miranda AvenuePalo Alto, CA 94304PHONE 650-849-0421FAX 650-849-0285EMAIL gaba@stanford.edu

A Medical SpecialtyBlazes a TrailAnesthesiology was firstin addressing medical errors

Ellison C. Pierce, Jr., M.D.Associate Clinical Professorof AnaesthesiaHarvard Medical School770 Boylston Street, Apartment 10CBoston, MA 02199-7709PHONE 617-450-0242FAX 617-450-0247EMAIL epierce500@aol.com

FOR ADDIT IONAL INFORMATION ON THE

QUALITY IMPROVEMENT PROJECTS PROFILED

IN THIS REPORT PLEASE CONTACT:

The National Coalition on Health Care1200 G Street NW, Suite 750Washington, DC 20005

PHONE 202-638-7151FAX 202-638-7166

EMAIL info@nchc.orgWEBSITE www.nchc.org

The Institute for Healthcare Improvement135 Francis StreetBoston, MA 02215

PHONE 617-754-4800FAX 617-754-4848

EMAIL info@ihi.orgWEBSITE www.ihi.org