|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology Multidisciplinary Systems Approach to Chemotherapy Safety: Rebuilding Processes and Holding the GainsByFrom the Department of Pediatrics and School of Nursing, University of Pennsylvania, and Division of Oncology and Departments of Pharmacy and Nursing, The Children’s Hospital of Philadelphia, Philadelphia, PA. Address reprint requests to Richard B. Womer, MD, Division of Oncology, The Children’s Hospital of Philadelphia, 34th St and Civic Center Blvd, Philadelphia, PA 19104; email: rwomer{at}mail.med.upenn.edu
PURPOSE: The problem of medication safety came to public attention largely through a chemotherapy error, and the high toxicity and low therapeutic index of anticancer medications make safety in their prescription and administration critical. We have undertaken a thorough revision of our systems for inpatient chemotherapy. METHODS: We participated in a multi-institutional collaborative effort of the Institute for Healthcare Improvement, and used their rapid cycle change method. Particularly powerful systems change concepts were driving out fear, "trapping" errors and learning from them, focusing on outcome rather than on input, simplifying and standardizing, using constraints and "forcing functions," reducing handoffs, and paying attention to human factors. RESULTS: Applying these concepts to our chemotherapy delivery system, we have achieved an 84% decrease in the number of chemotherapy errors that actually reach patients per 1,000 chemotherapy doses, and have sustained that improvement for 5 years. CONCLUSION: Factors contributing to our success include the rapid cycle change method, strong support from hospital administration, grassroots participation, and a tradition of interdisciplinary cooperation. Computerized direct physician order entry and cooperative group participation have had mixed effects. Continued efforts at improvement have been key to holding our gains. Although specific problems and changes may not be relevant to other organizations, the concepts and methods we used are generally applicable.
TEN YEARS AFTER the first systematic studies of the incidence of medication errors, and 7 years after the death of Boston Globe medical writer Betsy Lehman of a chemotherapy overdose, intense public, press, and legislative attention is focused on the problem of medical errors. Oncologists have no greater or lesser chance of erring than most other physicians, but the high toxicity and small therapeutic index of the drugs they prescribe make chemotherapy errors potentially catastrophic. After the Lehman incident, many hospitals and practices examined their chemotherapy ordering practices and put new safeguards in place, and by 1996 the vast majority of comprehensive cancer centers, university hospitals, and community hospitals had begun chemotherapy safety programs.1 The public and the legal profession tend to view medical errors as failures of individuals, and accordingly expect regulation, litigation, punishment, and shame to prevent errors. However, this "person approach" has been ineffective in preventing errors and may impede safety improvements. A "systems approach" recognizes that all humans are fallible and will err, and then develops mechanisms in which errors are less likely to occur and are identified and corrected when they do occur.2 At The Children’s Hospital of Philadelphia, we undertook a thorough examination and overhaul of our chemotherapy delivery system, beginning as part of a Breakthrough Series collaborative of the Institute for Healthcare Improvement (IHI). Lucian Leape, the leader of the collaborative, has described its overall structure, goals, and results.3 Our multidisciplinary effort, now in its sixth year, has resulted in a variety of systems improvements and substantial reduction in chemotherapy errors that reach the patient. We present here a summary of our efforts and results. Because the project was large and included many changes that did not result in improvements, we focus here on strategies that we found productive. Although specific changes may or may not work at all institutions, we believe that the strategies used are widely applicable to the delivery of chemotherapy and other medications.
Definitions The literature on patient safety usually considers adverse drug events, in which a patient experiences harm as a result of medication, whether as a result of an error or an idiosyncratic reaction (an adverse drug reaction). Potential adverse drug events occur when there is a medication error but no harm occurs, either because the error is intercepted before it reaches the patient or because of chance; for example, a patient with a history of penicillin allergy may receive a cephalosporin in error, but not react. We did not consider these terms suitable for our project, however, because we wanted to learn about all errors in the ordering and administration of chemotherapy, whether or not they reached the patient and whether or not they caused the patient harm. Accordingly, we divided chemotherapy errors into two categories: intercepted errors, in which mistakes occurred but were caught and corrected before they reached the patient; and actual errors, in which the mistake was not caught and reached the patients. Although we considered only serious errors (those with the potential to cause harm), we have not distinguished between actual errors that caused harm and those that did not.
The Setting The nurses on the oncology inpatient and outpatient units work exclusively with pediatric oncology patients. Before they may give chemotherapy, they have 8 to 10 weeks of supervised orientation and must pass a locally devised certification test. Approximately 10% of the staff are certified pediatric oncology nurses.
Ordering and Safety Procedures Already in Place In the pharmacy, two pharmacists, or a pharmacy technician and a pharmacist, independently check the orders against a copy of the patient’s "roadmap" (treatment schedule) and perform independent dose and dilution calculations. Finally, the nurse administering the medication checks the bag or syringe from the pharmacy against both the order and the roadmap.
The IHI Collaborative and Approach to Change The IHI approach to improvement emphasizes rapid cycle change (Fig 1). Briefly, one begins with a flow chart of the process under study and identifies the critical points where change is necessary. In a complex process such as prescribing and delivering chemotherapy, the flow chart is several pages long and a variety of critical points are addressed. The critical points are each approached with a goal-directed "ramp" consisting of a series of plan-do-study-act cycles. In each cycle, the minimum necessary amount of data is collected, a change is tried on a small scale, the results are analyzed with another small data collection, and the change is generalized if it results in an improvement. The emphasis is on making small changes quickly, with major systems changes flowing from the accumulation over time of many small change cycles along several simultaneous ramps.4,5
Error Reporting The pharmacy department had established a mechanism for identifying medication errors in the pharmacy several months before the project began. All unused medications are returned to the pharmacy so that charges to patients can be corrected. Medication errors are identified by analyzing the reasons for medication returns or for supplemental orders, examining incident reports, recording discrepancies between the pharmacist’s dose calculations and those of the physician, recording the pharmacist’s interventions, and examining the use of antidote or "rescue" medications such as naloxone. Although no method of capturing errors is perfect, the pharmacy’s system was established 6 months before the project began and has not changed during the project; thus it provides a true baseline and a consistent measure across time.
The First Steps
Principles of Medication Safety, Illustrative Changes, and Results
Driving Out Fear The surest way to eliminate reports of errors at any institution would be regular public humiliation of errant staff in the main lobby (the "blame-and-shame" approach). People would never report their own errors, nor would they report the errors of others. However, this approach would not reduce the errors themselves; it would only make their discovery more difficult. Conversely, the surest way to increase reporting of actual and intercepted errors is to eliminate punishment, or even the threat of punishment, from the process. This requires the realization that people involved in ordering and administering medications are well-trained, highly motivated, hard-working professionals who do not want to make mistakes, who already feel bad when they do err, and whose errors are usually the result of flawed systems rather than flawed individuals. Our establishment of nonpunitive reporting is described in detail in a separate article (Tracy et al, manuscript in preparation). Before the project began, nurses received verbal warnings after a first chemotherapy error and written warnings after a second error, and further errors or a serious error led to suspension or even termination. During the first week of the project, one of us (E.T.) announced the change of approach, met with each shift of nurses to hear about actual and near-miss errors they had witnessed or been involved in, and quickly gathered many ideas for improvements. Among the many systems issues uncovered were over 40 programming problems in the Eclipsys system; order output (the patient care summary) which differed significantly from the physician’s input; many ambiguous or conflicting orders; procedures requiring error-prone handoffs; and confusing or inconsistent treatment roadmaps or schedules. After the initial meetings, nurses were encouraged to report errors by dropping either an incident report or anything relating to an error into a specifically designated box. This box accumulated scribbled notes, copies of orders, and labels from bags and syringes, in addition to formal incident reports. As the staff’s confidence increased that formal reporting would not lead to punishment and would lead instead to system improvements, the submitted materials shifted almost entirely to incident reports. Managers often fear that without punishment their personnel will become careless and errors will increase. We found exactly the opposite: although reports of intercepted errors increased dramatically, the actual errors detected declined (Fig 2).
Trap Errors and Learn From Them The actual and intercepted errors we collected, sorted, and analyzed formed the foundation of our series of changes. Analyzing errors points to problems in systems; analyzing errors that are intercepted before reaching the patient (intercepted errors) also indicates what parts of systems are working effectively. We interviewed participants in actual or intercepted errors, often with the flow chart of the system at hand so that problems in systems could be pinpointed. We did not keep written records of our conversations, nor did we assign blame or responsibility for actual or intercepted errors, as the focus was on the systems. We also did not record whether an actual error led to an adverse drug event, as our goal was preventing errors. Sometimes fairly simple changes in the system resulted from our error analyses (such as correcting coding in a computerized order), and sometimes errors pointed to new ramps of change cycles. Virtually all of the changes discussed below flowed from error trapping and analysis. Focus on Output Rather Than Input As we analyzed reports of potential and actual errors, it became clear that many arose from orders that were clear to the prescriber but not clear to the pharmacist, the nurse, or both. Sometimes this lack of clarity arose from coding problems in the computer system and sometimes from ambiguous wording. When a computerized order repeatedly caused confusion, we asked a small group of nurses to rewrite the order in a fashion that they found clear and evaluate the new order as a group, and then we asked the information systems analyst to program Eclipsys so that the order appeared that way on the patient care summary. Occasionally, this involved some changes in the physicians’ input screens, but these did not prove problematic. Simplify and Standardize These changes have proven the most difficult and controversial of all that we have made, but also the most powerful. As in most pediatric oncology centers, three fourths of our patients are treated on clinical trials, most supplied by the Children’s Cancer Group (CCG) (now the Children’s Oncology Group [COG]), a National Cancer Institute–funded cooperative group with more than 200 institutional members. These protocols contain instructions for administering chemotherapy that usually reflect the habits or preferences of the protocols’ authors or their institutions. As a result, there is wide variation between (and sometimes within) protocols in the administration of common drugs. For example, pediatric oncology nurses and pharmacists at Children’s Hospital in Boston found 23 different cyclophosphamide regimens in only 14 protocols, varying in pre- and postcyclophosphamide fluid administration, diluent, infusion time, mesna use, and dose and administration of mesna.6 In 1996, we found 29 different CCG protocols using mesna, with 10 different methods of administration. Similarly, protocols using cisplatin specified infusion times ranging from 1 to 6 hours, in an array of electrolyte solutions containing (or not containing) a range of concentrations of sodium, magnesium, calcium, and mannitol. Also troublesome are the rules for calculating drug doses for infants and toddlers: some protocols change from weight-based to surface area-based dosing at 1 year, 2 years, or 3 years of age; or 10 or 12 or 30 kg weight; or consistently use one dose basis but divide the dose by two for infants. One brain tumor protocol had different instructions in different sections. We knew from our error trapping that these differences—and many others, affecting almost all commonly used agents—were leading to many potential and actual errors. Accordingly, we convened a chemotherapy standards committee, consisting of physicians, pharmacists, nurse practitioners, nurses, and a clinical research associate to develop standard methods of chemotherapy administration for our hospital. These standards apply to all uses of the drugs in oncology, except when a different protocol-directed method is central to the objectives of a clinical trial (eg, evaluation of toxicity or pharmacokinetics). The committee reviewed the current protocols, the available literature, and current practice and arrived at a method that deviated least from those in use and was practical for our personnel, systems, and equipment. For example, for high-dose doxorubicin (> 60 mg/m2), we chose a 48-hour continuous infusion, because the protocols in use were using 24-, 48-, and 72-hour infusions; and we chose to administer the drug in 96 mL at a rate of 2 mL/h, because this matched the capabilities of our inpatient and portable syringe pumps. We have updated the standards several times as new information has become available and practices and equipment have changed. This requires a large commitment from the members, estimated at 1 to 2 hours of work for every hour of meeting time, with meetings varying between weekly and monthly. Because deviations from protocol-directed methods of administration could cause problems in an institutional audit, we gained support for a change to the CCG audit rules so that institutional chemotherapy standards developed for purposes of improving patient safety and approved by the institutional therapeutic standards committee override protocol instructions, except when the instructions are essential to a specific aim of the clinical trial. Currently, the COG pharmacy committee is developing chemotherapy administration standards for the entire group, modeled on ours. Adoption of standard methods of administration promotes safety in several ways. First, it reduces confusion in ordering and makes the checking and cosigning of orders easier and more accurate. Adopting standard orders allowed us to eliminate two thirds of the chemotherapy order sets from the clinical computer system. Second, for those with computerized physician order entry systems, it greatly speeds the development of on-line order sets, as programmers can cut and paste from the standard order sets into protocol-specific order sets. Third, and most importantly, chemotherapy standards make errors more recognizable: physicians, nurse practitioners, nurses, families, and patients know that variations are likely to be errors and cannot be blithely attributed to the quirks of a particular protocol. As we have put more standard order sets on-line and continued to train our prescribers in their use, the fraction of our chemotherapy orders that are standard has steadily increased (Fig 3). This has also had an impact on errors: examining chemotherapy orders from a 10-month period, we found that although nonstandard orders were 13% of all chemotherapy orders, they were involved in 33% of all of the actual and intercepted errors.
Use Constraints and Forcing Functions Forcing functions are features of a system that only allow one to do something the right way. Common examples are the interlocks on American cars that prevent starting unless the transmission is in park or neutral, and automatic teller machines that require removal of the card before dispensing cash. Using chemotherapy standards, we were able to build several forcing functions into our clinical computing system. For example, our error trapping had revealed several episodes in which hydration or mesna was not ordered correctly with cyclophosphamide or ifosfamide. We changed the computerized orders so that when a physician orders cyclophosphamide or ifosfamide, the orders for hydration and mesna appear along with the orders for the drug, and the system will not accept the order until all components are complete. Computers are helpful, but not necessary, for forcing functions. An example of a procedural forcing function is our set of rules to prevent intrathecal vincristine administration. In protocols for the treatment of acute lymphoblastic leukemia, intravenous vincristine is often scheduled at the same time as intrathecal methotrexate or cytarabine. A possible error is the intrathecal administration of vincristine, which is a catastrophe: although a few patients have recovered with severe permanent neurologic damage, the majority experience several days of ascending paralysis, seizures, coma, and finally death. Intrathecal vincristine incidents have occurred at several large, excellent pediatric cancer centers, and 10 such incidents have occurred in the United Kingdom since 1985.7 Our rules for minimizing this risk are listed in Table 1; they are part of the orientation program for new oncology nurses and physicians, and signs are posted in the treatment rooms as reminders. The forcing function is the requirement that the empty vincristine bag be returned to the pharmacy before the intrathecal medication and sedation are dispensed.
Reduce Handoffs Each time a task is transferred between people or groups of people (handed off) there is a possibility of error from inaccurate or incomplete communication, which some have termed gaps. Many professional routines, such as handoffs between shifts of nurses and physicians signing out to each other, are designed to bridge gaps, and new gaps arise as technologies and routines change. Although handoffs or gaps are inevitable, their number can be minimized and their safety maximized with appropriate systems design.8 When our patients are being admitted for chemotherapy, they visit the oncology clinic first for laboratory work and a history and physical examination. Often, the first doses of chemotherapy are given in the outpatient clinic while the patient is awaiting a bed on the inpatient unit. Our outpatient unit still uses a paper system for chemotherapy ordering and documentation rather than the Eclipsys system used on the inpatient unit. Our error trapping indicated that this handoff between the clinic and the inpatient unit was involved in several potential and actual errors, which were compounded by handoffs between nursing shifts shortly after admission. Thus, we began several ramps to reduce handoffs by making chemotherapy admissions occur earlier in the day. This would have the added benefit of having the complex process of beginning a course of chemotherapy occur earlier, when people are most focused and more staff are available. We began by establishing the rule that chemotherapy could be given in only one setting per cycle: outpatient, inpatient, or at home. Because the outpatient clinic used a paper-based ordering and documentation system and the inpatient unit used a computer-based system, consistent and complete transfer of orders and documentation was difficult if more than one setting was involved. Meanwhile, we timed several patients as they went through our system of chemotherapy admission and took note of the steps associated with delays. With these results, we worked with the clinical laboratories to speed the processing of specimens and the reporting of results; we worked with the housekeeping department to speed the cleaning of vacated rooms and beds; we streamlined the discharge process, including the ordering of blood products before discharge, to make discharges occur earlier; and we modified the roles and schedules of physicians and nurse practitioners in the outpatient clinic to speed the performance of histories and physical examinations and the entry and activation of chemotherapy orders. These changes resulted in earlier admissions and the streamlining of some troublesome processes but did not cause the chemotherapy to begin earlier. Another problem our data revealed was that it took an average of 6 hours to hydrate patients sufficiently to receive cyclophosphamide, ifosfamide, cisplatin, or high-dose methotrexate. Our solution to this problem, the rapid hydration protocol, is the subject of a separate report.9 Briefly, when efforts to improve patients’ oral hydration before coming to the hospital failed, we developed a system of giving patients 750 mL/m2 of D5 0.4% NaCl over 1 hour. This shortened the time from beginning intravenous fluids to having a urine specific gravity of 1.010 (the criterion for adequate hydration) from a mean of 6 hours to a mean of 1.5 hours. This in turn resulted in an improvement in the time chemotherapy began, but most patients were still not receiving their first doses until about 10 PM. Limiting each cycle of chemotherapy to only one setting eliminated troublesome handoffs, but it prevented giving the first doses in the clinic while awaiting an inpatient bed. We solved this problem by creating a "virtual inpatient unit" in collaboration with the information systems department. This allowed patients to be "admitted" to nonexistent beds while they were still in the clinic, so that their chemotherapy could be ordered, activated, and documented on the inpatient computer system. We evaluated the impact of this change by randomly sampling 20 charts of patients admitted for chemotherapy before the creation of the virtual unit, during its first year, and during its second year. The proportion of sampled patients receiving their first dose of chemotherapy by 6 PM rose from approximately 10% to 100%, and the number of nurses involved in the first day’s chemotherapy dropped from 2.2 to 1.1 (Fig 4).
The accumulated effect of all of these changes has been a reduction in the average interval between registration and the first dose of chemotherapy from approximately 9 hours to approximately 4.5 hours. This has had obvious implications in reducing handoffs and allowing the complex tasks involved to occur during daylight hours, which is a "human factors" improvement. Pay Attention to Human Factors One of our first improvement ramps was concentration and communication, because physicians and nurses identified distractions and interruptions as frequent contributors to errors. For prescribers, we arranged for two call rooms adjacent to the oncology inpatient unit to be available as chemotherapy-ordering sanctuaries during the day, equipping them with protocol books, reference materials, and a computer terminal. There, physicians and nurse practitioners could order or review and activate chemotherapy orders behind a closed door, without interruption. Although its contribution to safety cannot be quantified, this change was enthusiastically received, not only by prescribers but also by house staff, who found the computer terminal and reference materials useful when on call. We also pursued a series of change cycles to reduce distractions and interruptions for nurses while they were administering chemotherapy. Before these changes, standard procedure on our unit had nurses paged for every telephone call or request from anyone. The changes consisted of putting a message board at the nurses’ station, to which notes could be taped for each nurse on duty, and instructing the unit clerks to take messages for telephone calls rather than page the nurses. We had a group of nurses log interruptions during their shifts and classify them (urgent, routine, patient/parent, physician, pharmacy, telephone, another nurse, or other). When six prechange shifts were compared with eight postchange shifts, the number of interruptions recorded dropped by approximately half, from 27.6 per shift to 13.8. The improvement applied to every category of interruption. A similar change did not work for oncology fellows and nurse practitioners. We arranged to have their pagers automatically recorded for 2 weeks before and after the installation of a different message board. The board was largely unused, and the number and origins of pager activations did not change.
The number of actual errors per 1,000 inpatient chemotherapy doses (actual error rate) has declined from 6.2 before the project to 1.0 in calendar year 2000, whereas the number of doses delivered per year has risen from 8,022 to 10,924 (Fig 5). Thus, the error rate has decreased 84% while the number of doses has increased 36%. This change between the baseline period in 1995 and the project period beginning in 1996 is statistically significant (P = .029 by exact Kruskal-Wallis test). Comparing the numbers of actual errors observed with those expected at the baseline rate of actual errors, one can calculate that the project prevented 226 actual chemotherapy errors through calendar year 2000.
The changes in our chemotherapy administration system have focused on inpatient chemotherapy, but some (especially the chemotherapy standards) have also applied to outpatient treatment. We instituted the pharmacy error tracking system in the outpatient oncology pharmacy 3 years ago, and the number of actual errors has remained low for the 3 years the system has been in operation.
We have described here a highly successful approach to reducing chemotherapy errors that is applicable to many other systems. Most departments and units at our hospital have adopted nonpunitive reporting, and other units have launched rapid cycle change improvement projects as well. The decline in actual chemotherapy errors early in 1996 began shortly before our first change cycles, and they may simply reflect the performance improvements that sometimes occur when people observe or change a system (the Hawthorn effect). However, observer effects would be impossible to sustain for several years, as our improvements have been. The system changes we have made are clearly responsible for the sustained improvement in chemotherapy safety. In our view, our success is attributable to several factors. First, the method of rapid cycle change has been indispensable. Making small changes on a small scale before generalizing them minimizes the "law of unintended consequences," quickly rejects the many changes that are not improvements, and reduces the innate human resistance to change. Furthermore, it maximizes the potential for synergy between changes in different parts of the system: for example, the virtual unit change could not have succeeded without all of the changes previously made in physician and nurse practitioner work assignments, admissions procedures, discharge procedures, handling of laboratory tests, and even ordering of inpatient blood products. Second, we have enjoyed strong support from our hospital’s administration and the leadership of the oncology division. They made possible our participation in the IHI collaborative, which was invaluable in providing us the philosophy, support, consultation, analytic tools, and peers to help our project succeed. They also made time available for the physicians and nurses working on the project during its first year and committed support from the information services department. Third, ours was (and continues to be) a "ground-up" rather than a "top-down" effort. The core team of four of us who traveled to IHI conferences and directed the project led a larger committee of 18, consisting of physicians, nurses, pharmacists, and clinical research associates (data managers). Most of the ideas for improvement cycles came from them and from our colleagues in all disciplines who constantly supply us with reports of problems and ideas to solve them. Fourth, we are blessed with a tradition of interdisciplinary collegiality that has been invaluable. In other hierarchical industries, such as aviation, training in "crew resource management" is used to encourage communication, cross-checking, and shared decision making between levels of the hierarchy.10 The hospital administration and the physician and nursing leadership of the oncology division long have actively promoted open communication between disciplines as the best means to ensure high-quality patient care. Finally, our core team’s leader for the first year of the project was a nursing administrator from outside the oncology division (J.H.B.). This gave her a perspective free of oncology traditions, habits, and biases that greatly sped our progress. Computerized physician order entry has been a mixed blessing. On the positive side, it eliminates handwriting and transcription errors, which bedevil many other hospitals’ systems, and it makes the use of standard order sets and forcing functions practical. The introduction of computerized physician order entry has reduced medication errors by 50% to 80% in some hospitals,11 with particularly large impacts on transcription and dispensing errors.12 On the negative side, changing almost any aspect of our medication system involves computer programming, in a system that is difficult to program. Changes to standard orders, which can be accomplished in a matter of hours with a word processor and photocopy machine in a paper system, can take months to encode and debug in the Eclipsys system. Our system does not yet automatically calculate doses, detect out-of-range doses, screen orders against lists of allergies, or provide decision support; we are looking forward to easier programming in addition to these advanced features in upgrades to our current computer system. Participation in national cooperative groups is also a mixed blessing; the majority of pediatric cancer patients are treated on national trials under the auspices of the COG (successor to the Pediatric Oncology Group, the CCG, the Intergroup Rhabdomyosarcoma Study, and the National Wilms Tumor Study). National protocols involve some degree of standardization within a protocol, but wide variation continues between protocols. Our substitution of institutional chemotherapy administration standards for protocol instructions has been controversial, as some worry that the results of a study could be altered. To minimize the risk of affecting the outcome of a study, we specifically exclude from the chemotherapy standards studies for which the method of administration is critical, for example, phase I trials, studies involving pharmacokinetics, and phase II or III studies that test a hypothesis involving the method of administration (such as recent trials of different irinotecan schedules). Also, our chemotherapy administration standards are designed to deviate as little as possible from the methods of administration in the protocols; probably the most radical substitution was an institutional 48-hour doxorubicin infusion for a protocol’s 24-hour and 72-hour infusions. Finally, the variations between institutions, physicians, and nurses in their interpretation and execution of protocol instructions, and the variations between patients in absorption, metabolism, and excretion of medications, are so large that the variations introduced by institutional administration standards are relatively trivial. We believe that the demonstrable importance of such standards for patient safety outweighs their possible impact on studies. The COG’s pharmacy discipline committee is developing standard chemotherapy administration guidelines for all studies, but their use is not mandatory. Uniform chemotherapy administration standards and uniform criteria for using weight-based and surface area–based dosing would promote pediatric chemotherapy safety nationwide. The national and global obstacles to error prevention have been exhaustively covered elsewhere and have certainly applied to us. In particular, Philadelphia’s highly litigious climate chills the sharing of data and experience, especially with outsiders, and we who work in this area feel under constant legal threat. We also have to be constantly on guard against our own perfectionism: the tendency of physicians and nurses to view any error as a personal failing rather than as a systems problem. In many improvement projects, holding the gains is a problem. We have held our gains of the project’s first year, and even furthered them, only with continuing effort. Many of the changes reported here were made in the past 3 years. Increasing numbers of patients, a nationwide shortage of nurses, and increasing demands on physicians’ time are continuing challenges. The systems we have put into place require continuing effort to maintain; for example, constantly changing cooperative group protocols, emerging knowledge, and changing equipment mandate frequent revisions to chemotherapy standards and order sets. Our multidisciplinary committee continues to meet weekly to review systems and data and devise new change cycles, and the chemotherapy standards committee continues to update the chemotherapy standards twice yearly. Others have published lists of steps to enhance chemotherapy safety. Particularly valuable are the recommendations of the Institute for Safe Medication Practices (which include the guidelines of the National Institutes of Health Clinical Center)13 and the guidelines in use at Yale-New Haven Hospital.1 Not all of their recommendations are suitable for pediatrics or for all institutions, but they provide a valuable guide for beginning safety improvement efforts. Establishing a culture of safety and sustaining improvements in patient safety, however, require a long-term commitment.
Anna Meadows, MD,and Jeffrey Rivest were generous in their encouragement and provision of resources to get the project underway. Lucian Leape, MD, provided invaluable guidance and encouragement in the first 2 years of this project, and helpfully reviewed the manuscript. The Children’s Hospital of Philadelphia information services department found itself at the center of many changes, and we appreciate their continuing cooperation. Avital Cnaan, PhD, provided statistical assistance. Thomas Nolan, PhD, Beverly Lange, MD, Garrett Brodeur, MD, and Angelo Giordino, MD, provided useful reviews of the manuscript. Finally, the participation, cooperation, and collegiality of the physicians, nurse practitioners, nurses, and pharmacists of the division of oncology at The Children’s Hospital of Philadelphia were, and continue to be, indispensable.
1. Fischer D, Alfano S, Knobf M, et al: Improving the cancer chemotherapy use process. J Clin Oncol 14: 3148-3155, 1996[Abstract]
2. Reason J: Human error: Models and management. BMJ 320: 768-770, 2000 3. Leape L, Kabcenell A, Gandhi T, et al: Reducing adverse drug events: Lessons from a Breakthrough Series collaborative. Jt Comm J Qual Improv 26: 321-331, 2000[Medline]
4. Berwick D, Nolan T: Physicians as leaders in improving health care: A new series in Annals of Internal Medicine. Ann Intern Med 128: 289-292, 1998
5. Nolan T: Understanding medical systems. Ann Intern Med 128: 293-298, 1998
6. Sievers T, Lagan M, Bartel S, et al: Variation in administration of cyclophosphamide and mesna in the treatment of childhood malignancies. J Pediatr Oncol Nurs 18: 37-45, 2001
7. Dyer C: Doctors suspended after injecting wrong drug into spine. BMJ 322: 257-258, 2001
8. Cook R, Render M, Woods D: Gaps in the continuity of care and progress on patient safety. BMJ 320: 791-794, 2000 9. Tracy E, DiTaranto S, Womer RB: Evolution of a rapid hydration protocol. J Pediatr Oncol Nurs (in press)
10. Helmreich R: On error management: Lessons from aviation. BMJ 320: 781-785, 2000
11. Bates D: Using information technology to reduce rates of medication errors in hospitals. BMJ 320: 788-791, 2000
12. Bates D, Leape L, Cullen D, et al: Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA 280: 1311-1316, 1998 13. Cohen M, Anderson R, Attilio R, et al: Preventing medication errors in cancer chemotherapy. Am J Health Syst Pharm 53: 737-746, 1996[Abstract] Submitted April 17, 2002; accepted August 12, 2002.
Related Correspondence
Related Reply
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
