- Email: [email protected]
Applying the Toyota Production System: Using a Patient Safety Alert System to Reduce Error
The Joint Commission Journal on Quality and Patient Safety
Methods, Tools, and Strategies
Applying the Toyota Production System: Using a Patient Safety Alert System to Reduce Error
Cathie Furman, R.N., M.H.A. Robert Caplan, M.D.
Recently, an ambulatory surgery center registered nurse (R.N.) called a patient safety alert (PSA) on herself. At the start of the day, she had prepared a vial of Ancef® (cefazolin for injection), noting to herself that “sometime today I am going to need it.” Sure enough, a few hours later, when readying a patient for a procedure, she hung the already prepared intravenous (IV) Ancef. She then began reviewing the patient’s medical record. When noticing that the patient was allergic to penicillin, she immediately stopped the infusion (Ancef is contraindicated in patients with known allergy to penicillin). Even though it was a near miss, with no harm done, she used the PSA system to report the event. Within hours, the senior executive accountable for the area had conducted an investigation and implemented system-based changes, such as implementation of standard work to ensure that no nurse uses a previously prepared IV without first checking the medical record. This is one of thousands of changes that we have made at Virginia Mason Medical Center (VMMC; Seattle) during the past few years to keep our patients safe. Ideally, error-prone processes should be fixed as soon as possible, but conventional tools for quality improvement can be very slow and inefficient. At VMMC, it often took weeks or months before a traditional quality incident report even received any attention. This delay made it hard to fully understand all the facts and the root causes that had led to the original problem; the trail was long since cold. Our traditional quality incident program was also characterized by a culture of 376
July 2007
Article-at-a-Glance Background: In 2002, Virginia Mason Medical Center (VMMC) adapted the Toyota Production System, also known as lean manufacturing. To translate the techniques of zero defects and stopping the line into health care, the Patient Safety Alert (PSA) system requires any employee who encounters a situation that is likely to harm a patient to make an immediate report and to cease any activity that could cause further harm (stopping the line). Implementing the PSA System—Stopping the Line:
If any VMMC employee's practice or conduct is deemed capable of causing harm to a patient, a PSA can cause that person to be stopped from working until the problem is resolved. A policy statement, senior executive commitment, dedicated resources, a 24-hour hotline, and communication were all key features of implementation. Results: As of December 2006, 6,112 PSA reports were received: 20% from managers, 8% from physicians, 44% from nurses, and 23% from nonclinical support personnel, for example. The number of reports received per month increased from an average of 3 in 2002 to 285 in 2006. Most reports were processed within 24 hours and were resolved within 2 to 3 weeks. Discussion: Implementing the PSA system has drastically increased the number of safety concerns that are resolved at VMMC, while drastically reducing the time it takes to resolve them. Transparent discussion and feedback have helped promote staff acceptance and participation. Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
secrecy that limited knowledge of the problem to those with a “need to know.” This reduced creative thinking about remedies and limited the opportunity to broadly leverage safety lessons. It also meant that the organization had no clear sense of which processes were broken or dangerous, thereby hampering engagement, collaboration, and buy-in regarding safety priorities.1 Overall, the long delays and lack of feedback contributed to the prevalent cultural belief among providers and administrators that a certain frequency of “complications” was expected and tolerable. All this changed dramatically at VMMC when we converted to the Virginia Mason Production System in 2002, our version of the Toyota Production System, for management of our health care system.2 A primary goal of the Toyota System is to pursue zero defects by finding and fixing reversible mistakes as soon as possible—at the times and the places at which these reversible mistakes occur. We believe that this approach, originally developed to improve manufacturing processes, provides an effective and efficient way to improve health care safety. This article describes the implementation of one key aspect of the VMPS—reporting safety-related problems and, if necessary, “stopping the line.”3,4 We also describe our parallel efforts to create a culture in which employees feel safe about reporting problems and in which management feels comfortable about disseminating the lessons learned from reporting and solving them.
Background Health care depends on complex human activity. Human failings in dealing with complexity are therefore bound to result in failings in the delivery of health care. For example, humans are not adept at identifying the complex coincidences that cause system failures, particularly the inefficiency and errors of health care.5–7 James Reason has cautioned that complex systems can “spring nasty surprises.”8 In health care, these “surprises” result in enormous losses, including the loss of life. According to Morath and Tunbull, preventable medical accidents could cost the United States nation $17–$29 billion a year.9 Zero defect theory, developed by Shigeo Shingo, a Japanese industrial engineer at Toyota, regards human mistakes as inevitable but also reversible if they are caught soon enough.10 If mistakes are found and fixed soon
July 2007
enough, the final product—be it cars or health care—will not have defects, defined as mistakes that have persisted so long that they have become irreversible or hard to fix. Shingo discovered that mistakes can be kept from becoming defects if processes are broken down into small, understandable steps and if inspection and correction occurs at each step. Weick and Sutcliffe characterize a high-reliability organization as one that acts “mindfully,” organizing itself in a way that makes it “better able to notice the unexpected in the making and halt it [our italics].”5 (p. 3) Certainly, the pursuit of zero defects requires such mindfulness—“attention must be paid”—everywhere and all the time. Thus, in a system that deploys zero defects, every individual in the workplace must become a safety inspector and must have the authority to call out mistakes (or “stop the line”) as soon as possible. Some manufacturing companies have pursued the mindful elimination of error in all processes so that they can provide the highest quality, the safest products and services, outstanding customer satisfaction, and healthy balance sheets. These companies use a set of concepts and methods which have become known collectively as lean production, although at Toyota, where these methods originated, they are known as the Toyota Production System.10 In 2002, VMMC adopted this system as its management method and named it the Virginia Mason Production System. The basic tenets of lean production include the standardization and detailed specification of work processes, the organization of work in such a way that unexpected and undesirable events are easy to spot, and the deployment of activities that find and fix mistakes as soon as they are made, correcting those mistakes before they become defects that are difficult to reverse or remedy.2,11 Additional assistance is provided, if needed, by supervisors and managers, who must be adept at collaborative problem-solving. If the mistake cannot be fixed on the spot, the faulty process is stopped until the mistake is corrected; this is so no other defects occur. (No other patient will have to suffer this particular harm or risk of harm.) This step is called stopping the line. At Toyota, if an employee encounters a problem that cannot be fixed within a prescribed time, he or she physically makes a signal that stops the production process. In health care, this is comparable to a well-staffed
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
377
The Joint Commission Journal on Quality and Patient Safety
and coordinated response to a “code call” for cardiac arrest, bringing needed resources to help the staff fix the problem rather than using traditional work-arounds.
to our community. We also believe that the long-term safety benefits of honesty and transparency will far outweigh any short-term liabilities.
Culture of Safety and Reporting
Implementing the PSA System
A key requirement for deploying zero defects is a workforce empowered to speak up when a mistake or error occurs. This requires overcoming employees’ natural and justified fears of retribution; it is a deliberate, slow, and difficult process. Health care has deeply ingrained cultures of blame, denial, and fear5; the culture of blame, for example, creates the presumption that errors and poor outcomes occur because of incompetence. In addition, the hierarchical culture of health care discourages subordinates from calling out the mistakes of colleagues in more powerful positions. Hence, surgery residents are reluctant to point out mistakes made by surgery attendings, and nurses are often fearful of mentioning mistakes made by physicians. To implement zero defects, these old attitudes must be eliminated, and three very different expectations must be developed5: ■ It is safe to report mistakes. ■ When mistakes are reported, they will be corrected. ■ Those who report mistakes will be praised. At VMMC, we have engaged in focused efforts to improve the culture of safety since 2002. This work has been guided in four important ways: 1. We have used culture of safety surveys since 2002. 2. We implemented executive walkrounds12 in 2003, a process that brings all 28 members of our executive team into direct dialogue with our staff about any obstacles to delivering safe health care. 3. Every month our professional staff and management team hear about errors that occurred and what actions were taken to correct them. We believe that this kind of transparency within our organization eradicates the concept that errors are something “kept secret” or something that “can’t happen here.” 4. We work very hard to “close the loop” and let our staff know that each report has been received and what has been done about it. We recognize that transparency may increase the risk for legal action and negative media coverage, but we believe it to be the right thing to do for our patients. Put another way, we believe that failing to report, recognize, or act on safety problems is an unacceptable breach of duty
VMMC’s PSA system, which went into effect in September 2002 by the chief executive officer (CEO)’s demand, is a direct application of stopping the line in health care. The goal is to detect and fix every safety hazard (mistake) as soon as it occurs, thus preventing the hazard from creating more serious harm downstream. Stopping the line can mean stopping a person as well as a process. If any VMMC employee’s practice or conduct is deemed capable of causing harm to a patient, a PSA can cause that person to be stopped from working until the problem is resolved ([Sidebar 2] see page 383). A policy statement, senior executive commitment, dedicated resources, a 24-hour hotline, and communication were all key features of implementation.
378
July 2007
POLICY STATEMENT The first step was to create a PSA policy. Any staff member who encounters a situation that he or she believes is likely to harm a patient is required to make a report to the department of patient safety immediately, using the 24-hour hotline. Furthermore, the staff member must cease any activity that he or she believes could cause further harm. Initially, in September 2002, only “significant” safety incidents—near misses and actual adverse events—were to be reported. Yet staff members had told us that they felt “ignored” or “unheard” if we made a determination that a report was not considered to be a PSA. We learned that “significant” is truly in the eye of the beholder. In addition, VMMC executives found it hard to distinguish between a safety PSA and a quality incident. Such disagreement or confusion about what was to be considered a safety incident absorbed time and resources, delaying resolution. Consequently, in February 2005, we implemented a policy that differentiated between high, moderate, and low risk of harm and high, moderate, and low likelihood of recurrence, as follows: ■ Red PSAs are events that cause serious harm or have the potential to occur frequently; sentinel events are placed in
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
this category. The accountable executive is expected to conduct an investigation, stop the process or provider if needed, and resolve the problem within 24 hours. ■ Orange PSAs are events that cause less severe harm or are less likely to recur; near misses are placed in this category. Orange PSAs require investigation and resolution within 72 hours. ■ Yellow PSAs, events that were previously called quality incidents, have the lowest potential for harm. We expect yellow PSAs to be resolved within one week. The patient safety specialist, who categorizes the PSA as red, orange, or yellow in concert with the accountable executive leader (for example, chief of surgery for an averted wrong-site surgery), is expected to conduct the entire investigative and problem-solving process within 24 hours to one week, in stark contrast to the lengthy quality incident report process that was formerly used at VMMC.
SENIOR EXECUTIVE COMMITMENT Each PSA requires that the accountable senior executive be notified and that he or she go to the “shop floor” to determine whether the process or person needs to be stopped. To remove a high-powered physician or to stop a revenue-generating process requires executive authority at VMMC, as we believe it would at most hospitals. Each PSA is also reviewed by the CEO, the president, and the senior vice president of quality. If the accountable executive does not believe that the reported incident is a patient safety concern, the CEO and the senior vice president of quality must agree with that determination. If they do not, the PSA process moves forward. The senior vice president of quality is responsible for overseeing each PSA in order to remove barriers and to aid in resolution if needed. The board-level quality oversight committee reviews the PSA data at each meeting and engages in discussion about the process and specific corrective action plans, holding the senior executive team accountable for timely, effective actions.
DEDICATED RESOURCES The PSA initiative launched a consolidation of the risk management and quality assessment functions into the newly created patient safety department. The new role of patient safety specialist was created. This role, for example, provides consultation to the executive and facilitation of
July 2007
team huddles, coordinates documentation of analytical tools (for example, Failure Mode and Effects Analysis, root cause analysis) when needed, reviews the PSA documentation for completeness, tracks and trends final reports, and generates and distributes data for the quality oversight committee. People who had been claim managers and quality resources specialists (nurse reviewers) under the old system were invited to apply for this new position. Before the merger, there had been animosity between the two groups. Nevertheless, within two months, members of both groups were pooling their skills and were cross-trained, and by a year later or so, the team worked well together.
A 24-HOUR HOT LINE The commitment to rapid resolution of PSAs required a hotline, with a patient safety specialist always on call, so that a PSA can be called at any time. In 2005, VMMC implemented a Web-based reporting system so that staff could also report PSAs on the hospital intranet. When a report is entered, the on-call patient safety specialist is alerted and the case is entered into a database that is used for analysis and aggregate reporting.
COMMUNICATION Data were collected, analyzed, and reported quarterly regarding the total percentage of staff who were aware of the PSA policy and knew what to do if they identified an error or had a patient safety concern. During each executive walkround, we distributed printed business cards that included the PSA hotline telephone number and held a question-and-discussion session. In addition, at all monthly management meetings and professional staff meetings, a case study of an actual PSA was presented.
MEASURES We use a Web-based database to collect the following information: ■ Number of PSAs reported, by type of PSA ■ Number of PSAs reported, by role function ■ Number of days elapsed from first report to resolution ■ Number of staff and processes taken off-line ■ Percentage of staff who are aware of the PSA policy ■ Percentage of staff who feel comfortable in reporting errors
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
379
The Joint Commission Journal on Quality and Patient Safety
Table 1. Categories of Patient Safety Alerts (PSAs), 2002–2006 Category*
2002 (5 mos)
Adverse drug reaction Airway management Blood/blood component Care/service coordination Diagnosis/treatment Diagnostic test Employee general incident Environment (facilities) Fall ID/documentation/consent Infection control Lab specimen/test Line/tube Medication/IV safety Restraints/supportive devices Safety/security/conduct Skin/tissue Surgery/procedure Surgical site infection Systems† Vascular access device Total declared PSAs
2003
2004
6
6
19
1
22
22
4
17
32
3
6
13
4
74
118
18
125
204
2005
2006
23 24 29 452 62 82 35 73 260 166 116 449 16 448 6 78 20 84 11
31 28 16 1103 61 100 27 76 285 170 124 405 22 580 5 119 34 99 2
16 2450
28 3315
* ID, identification; IV, intravenous. † Since 2004 “Systems”-related PSAs have been reassigned to more specific categories.
■
Percentage of staff who believe that VMMC treats patient safety as a high priority Culture of safety surveys have been conducted three times: in 2002—before the PSA system’s implementation; and after its implementation, in 2004 and 2005. The first two surveys used a commercially available tool,13 and the 2005 survey used the Agency for Healthcare Research and Quality (AHRQ) culture of safety survey tool.14
Results MEASURES Since the PSA system’s inception in September 2002, 6,112 PSAs have been reported (as of December 2006), addressing errors ranging from outdated tuna sandwiches on dietary trays to fatal medication errors. The PSA system has undergone continuous refinement since its inception. In the beginning, staff members were 380
July 2007
slow to call PSAs. As awareness of the policy grew and staff noticed that concerns were really being fixed, the number of PSAs increased dramatically (Table 1, above). By 2004, 100% of the staff were aware of the PSA system. The terms incident report or QA report have now been replaced by PSA. Table 2 (page 381) provides examples of the types of safety problems that have been reported. The goal is to resolve each PSA within 48 hours, but we have not achieved this yet. In fact, the number of reports significantly increased in 2006, resulting in an average time until resolution of 33 days (Table 3, page 382). In response, we have almost doubled the number of patient safety specialists—to five—dedicated to this work in 2007. Even these values represent a large improvement over the 3-month to 18-month turnaround time associated with our previous conventional system of incident reporting and committee review.
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Table 2. Patient Safety Alert Category Examples* Airway Management ■ Self-extubation ■ Nebulizers not available during emergent treatment
ID/Documentation/Consent ■ Patient misidentification ■ Inadvertent release of patient health information ■ Missing consents
Blood/Blood Product ■ Delayed processing of blood orders
Infection Control ■ Sharps use ■ Inadequate setup of isolation precautions ■ Personal protective equipment
Care/Service Coordination ■ Communication and handoff issues among healthcare team
Lab Specimen/Test ■ Mislabeled specimens or unlabeled specimens
Diagnosis/Treatment ■ Delayed diagnosis or treatment resulting in worsened condition ■ Incorrect diagnosis or incorrect treatment provided to patient
Line/Tube ■ Infiltration of IV line ■ Incident related to order, preparation, insertion, or use of line or tube
Diagnostic Test ■ Missed orders or delayed communication of critical results
Medication/IV Safety ■ Delay in obtaining and administering STAT medications ■ Wrong medications administered to patient ■ Use of dangerous abbreviations on medication orders ■ Near miss of wrong dose or wrong medication
Environment ■ Staffing concerns ■ Malfunctioning equipment
Safety/Security/Conduct ■ Wandering patient ■ Allegations of sexual misconduct or assault ■ Verbal or physical abuse by staff ■ Drug-seeking behavior by staff
Employee ■ Injury from assisting patient ■ Noncompliance with fitness-for-duty requirements ■ Needlestick injury
Skin/Tissue ■ Trauma of skin/tissue, such as phlebitis, rashes
Fall ■ Patient fall in room, with or without assistance
Surgery/Procedure ■ Ordering, preparation, or performance of surgical procedure or anesthesia ■ Equipment not available for scheduled surgery
* ID, identification; IV, intravenous.
SPEED OF THE PSA SYSTEM
INAPPROPRIATE PHYSICIAN BEHAVIOR AS AN ISSUE
One foundation of the PSA process is quick identification of the root cause, followed by a quick remedy. This is particularly important when providers or processes are taken off-line. The intent is to keep patients safe, but long delays in a process can lead to other harm. As an example of the speed of the PSA system, consider Case Study 1 (Sidebar 1, page 383).
At the outset of implementing the PSA policy and process, we made a deliberate and explicit decision that inappropriate physician behavior would be considered a PSA. We felt that this would be an important way to signal to the organization that safety was a responsibility that applies to everyone in the health care hierarchy. Case Study 2 (Sidebar 2, page 383) provides an example of how
July 2007
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
381
The Joint Commission Journal on Quality and Patient Safety
Table 3. Average Number of Patient Safety Alerts (PSAs) and Days to Resolution, 2002–2006 Metric
2002 (5 mos)
2003
2004
2005
2006
Average no. of PSAs/month
3.6
10.4
17
204
276
Average no. of days to resolution
18.4
13
13.7
12
33
Table 4. Number of Systems and Providers Taken Off Line, 2002–2006 2002 (5 mos)
2003
2004
2005
2006 (6 mos)
Employees Taken Off-line
6
5
14
12
19
Processes/Equipment Taken Off-line
1
4
8
8
10
this works. As shown in Table 4, above, 44 providers (physicians, nurses, and other health care workers) have been taken offline since the inception of the PSA process. Twenty-six (60%) of these persons have returned to work after appropriate remedial plans were developed, and 30% were terminated. Common problems identified include fatigue due to on-call duties, conflicts between operating room schedules and clinic schedules, and difficulties adjusting to computerized medical systems. The organization has put into place improvement efforts to address these issues. We report and discuss PSAs that involve providers, as we do any other PSAs, at our regular manager meetings and professional staff meetings. This makes it clear that safety applies to everything and everyone in health care delivery.
SOURCE OF PSAS Who reports PSAs? Initially, most reports came from nursing managers. Given the frontline position of nurses and the frequency of their contact with patients, we did not find this surprising. We were surprised that no reports came from pharmacists, given that VMMC delivers more than one million inpatient drug doses a year. We learned from the pharmacists that, in many cases, errors which we considered “significant” were taken by the pharmacists to be “normal” errors which they “corrected routinely.” This attitude changed dramatically after a patient was harmed by an illegible order that was misinterpreted by both a pharmacist and a nurse. At the time, we had already devel-
382
July 2007
oped preprinted admission orders (to reduce legibility issues at admission) and had implemented The Joint Commission’s recommendation to eliminate the use of dangerous abbreviations.15* Our policy on dangerous abbreviations requires health unit coordinators and pharmacists to stop the line if a handwritten order includes one of the specified dangerous abbreviations. However, a PSA occurred nonetheless (Sidebar 3, page 384). The investigation uncovered (1) the lack of clarity about whose job it was to stop the line and (2) the fear of retribution that the health unit coordinators and pharmacists felt. As part of the PSA’s resolution, stepby-step protocols were prepared that outline each group’s role if a dangerous abbreviation is used. Given the gravity of the error, VMMC’s top administration, for the first time, publicized the PSA incident—in an e-mail to all 5,000 staff and physicians. The e-mail stated that VMMC had harmed a patient and that everyone needed to be a safety inspector to ensure that it didn’t happen again. In addition, the hospital’s senior vice president and the director of pharmacy held focus groups with pharmacists to learn about and discuss their fears about and barriers to PSA reporting. One result of this event was that the pharmacists developed a clearer understanding of their responsibility to stop the line. As of December 2006, a total of 6,112 PSAs had been * Joint Commission National Patient Safety Goal 2B, “Standardize a list of abbreviations, acronyms, symbols, and dose designations that are not to be used throughout the organization.”
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Sidebar 1. Case Study 1—A Mix-up of Armband Colors At VMMC, we often use visual controls to enhance our understanding of patient needs. At VMMC, patients are given colored wristbands that signal special conditions such as allergies and “no code” status. One afternoon, a registered nurse (R.N.) coming on duty was rounding on her patients. She went in to do an assessment on a patient who had been admitted earlier that day with a new diagnosis of lung cancer. While doing the physical assessment, the R.N. noted that the patient had a pink wristband, which is our visual control for “do not resuscitate” (DNR) status. The R.N. thought that was odd for a newly diagnosed patient who was about to receive therapy, so she asked the patient about the wristband. The patient explained that the previous nurse had put on the wristband to indicate drug allergies. The color for drug allergies is orange, not pink. The R.N. immediately removed the pink band, gave the patient the correct orange band, and called a PSA. The chief nursing executive and the director of central ser-
vices immediately came to the floor to assess the situation. They huddled with the nurse and the manager. The investigation to find the root cause included asking “why” five times (each “why” demands an explanation at a deeper level or at a further step upstream). The root cause analysis quickly revealed that the R.N. on the previous shift was color-blind and had mixed up the colors. The team came up with the idea of adding words to the wristbands. Information systems reprogrammed the software to add allergy to the orange wristbands, DNR to the pink wristbands, and so forth. All of this was completed within hours of the initial event, ensuring that the process of banding of patients could continue—but safely. The nurse explained the error to the patient and said that a system was being put into place to fix it. When the new identification bracelets were produced the next day, the patient could see what had changed to ensure that an error would not occur again.
Sidebar 2: Case Study 2—A Noncompliant Oncologist In 2002, an oncologist ordered chemotherapy for a patient. The standard of care for this protocol required prior determination of an adequate cardiac ejection fraction and an alkaline urine. Before initiating the protocol, the oncology R.N. noted that the patient’s urine was still acidic and that the ejection fraction had not been assessed. The nurse called the physician, who replied that neither determination was necessary and ordered the nurse to start chemotherapy. The implicit message was, “Just do what I say because I am the doctor and you are just a nurse.”
dard of care. The oncologist then called the nurse back and was verbally abusive with her for calling the chief. The nurse notified the chief of cancer services that she had received an abusive phone call. The chief, determining that the oncologist’s behavior created a potential for serious and recurring harm, called a PSA, placed the oncologist on administrative leave, and temporarily suspended his privileges to admit patients to the hospital. A remedial plan for this oncologist was developed and implemented within 12 hours. Reinstatement of privileges required the following:
Instead, the nurse contacted the chief of cancer services and explained her concerns. The chief of cancer services, who agreed that the standard of care was to complete the standard testing before starting the chemotherapy protocol, called the oncologist to remind him of the applicable stan-
1. An apology to the nurse 2. A review of the literature on the standards for chemotherapy for the case in question 3. Anger management counseling 4. A follow-up plan to document improved behavior
reported, with the following sources: ■ Managers, 20% ■ Physicians, 8% ■ Nurses, 44% ■ Pharmacists, 5% ■ Nonclinical support personnel, 23% ■ Patients/families .003% In the early years of the PSA system, reports came almost exclusively from members of VMMC. However,
July 2007
since 2005 we have seen more participation by the community physicians who practice there. Furthermore, VMMC’s admission brochures tell patients about the safety programs and invite them and their families to participate in our safety efforts. Recently we had our first PSA report from a patient’s family!
CULTURE OF SAFETY AND REPORTING A variety of data and anecdotal evidence suggest that
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
383
The Joint Commission Journal on Quality and Patient Safety
Sidebar 3. A Misread Dosage An R.N. called a PSA after a patient had received three times the amount of a prescribed medication, which led to the patient’s acute deterioration and an emergency transfer to the critical care unit. The hospital administrator, the chief nursing executive, and the medical director of hospital services were called to investigate. They found that the handwritten medication order had been misinterpreted. The order was for 25 mg, but had been read as 75 mg. The medical director interviewed the hospitalist, using the technique of asking “why” five times, and found that a sequence of errors had occurred. Why had the wrong dose been given? In looking at the order, half of the team—process owners who help with the investigation and issue resolution—thought that it was 25 mg and the other half thought that it was
75 mg. It was noted that the prescribing physician had used one of the dangerous abbreviations that was not permitted. Why, then, had neither the unit clerk nor the pharmacists stopped the line, which policy dictated and which would have created an opportunity to determine the correct dose of the drug? Each thought the other role was accountable. Why did the R.N. administer a drug which didn’t make clinical sense for a frail patient? The R.N. was inexperienced and unfamiliar with the patient. In addition, the prescribing physician was so busy that he did not follow his usual practice of discussing the medication orders with the nurse, which would have created another opportunity to determine the correct dose.
Table 5. Subset of 2005 AHRQ Patient Safety Culture Survey*
* Agency for Healthcare Research and Quality: Comparing Your Results: Preliminary Benchmarks. Hospital Survey on Patient Safety Culture, Mar. 2006. http://www.ahrq.gov/qual/hospculture/prebenchmk.htm (last accessed May 9, 2007).
VMMC staff are beginning to believe that the PSA system leads to improvement in safety and quality. Although the change in the survey tool prevents direct comparison of culture-of-safety results for 2005 with those of 2002, the results for 2005 are statistically higher than the AHRQ 384
July 2007
benchmark data (Table 5, above). Anecdotally, we receive positive feedback about the PSA system during executive walkrounds, and we believe that the strong increase in reporting is another indicator that the system is accepted by our staff.
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Patient Safety Alerts (PSAs) by Level, 2005 and 2006
infrequent red and orange PSAs as a type of mistake that we want to eradicate immediately and permanently. In contrast, we look at the aggregate data for our yellow PSAs and ask which types occur most frequently. We then study those types to determine if there are broader, systembased strategies that can be used to make improvements in safety and quality. We are particularly looking for problems that might turn up throughout the hospital. Each year, as part of the goal setting process for the organization, these data are used to determine what the organizational safety initiatives will be.
Summary and Conclusions The PSA system has proven to be the single most important tool to make our care Figure 1. Red PSAs are high risk for actual harm; orange are those that are "near safer. It provides a quick, timely process for misses;" and yellow PSAs are concerns that are low risk but concerning to the improvement and feedback to our staff reporter—those that get in the way of providing safe care. that their concerns will be listened to. Lessons learned are as follows: 1. Executive leadership is a prerequisite. How many PSA reports are enough? We had initially 2. Reporting should be easy, with multiple methods availset a yearly target of 800 PSAs, based on our belief that able. formal complaints about adverse outcomes still reveal only 3. “Significant” is in the eye of the beholder: Open the a small fraction of the preventable medical errors made. floodgates for all concerns. We have well exceeded that target when red, orange, and 4. Claims management staffing will go down as patient yellow PSAs are all included. The target is now defined as safety alerts go up. Right before the start of the PSA sysa total of 800 red and orange PSAs, of which we are avertem, we had six claims managers and three R.N. reviewers; aging about 300 a year (for example, 57 and 245 in 2006, we now have three claims managers and five patient safety as shown in Figure 1, above). specialists. 5. Be prepared to change the processes of care as the orgaDiscussion nization learns from the PSAs. J Implementing the PSA system has drastically increased the number of safety concerns that are resolved at VMMC while drastically reducing the time it takes to resolve them. Transparent discussion and feedback have played an Cathie Furman, R.N., M.H.A., is Senior Vice President, important role in staff acceptance and participation. Our Quality and Compliance, Virginia Mason Medical single best indicator of the viability of this approach is that Center, Seattle, and Robert Caplan, M.D., is Medical reporting of safety and quality concerns has continued to Director, Quality, and Staff Anesthesiologist, Department of Anesthesiology. Please address requests for grow since the inception of this system in 2002. reprints to Cathie Furman, [email protected]. We are beginning to develop tactics for understanding and reacting to PSA data. We view each of our relatively
July 2007
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
385
The Joint Commission Journal on Quality and Patient Safety
References 1. Tucker A., Edmondson A.: Why hospitals don’t learn from failures: Organizational and psychological dynamics that inhibit system change. California Management Review 45:55–72, Winter 2003. 2. Ohno T.: Toyota Production System: Beyond Large Scale Production. Portland, OR: Productivity Press, 1988. 3. Davidoff F., Batalden P.: Toward stronger evidence on quality improvement. Draft publication guidelines: The beginning of a consensus project. Qual Saf Health Care 14:319–325, Oct. 2005. 4. Barley S.R.: Images of imaging: Notes on doing longitudinal field work. Organizational Science 1:220–247, Aug. 1990. 5. Weick K., Sutcliffe K.: Managing the Unexpected: Assuring High Performance in an Age of Complexity. San Francisco: Jossey-Bass, 2001. 6. Institute of Medicine: To Err Is Human: Building a Safer Health System. Washington, D.C.: National Academy Press, 2000. 7. Institute of Medicine: Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C.: National Academy Press, 2001. 8. Reason J.: Human Error. New York: Cambridge University Press, 1990.
386
July 2007
9. Morath J., Turnbull J.: To Do No Harm. San Francisco: Jossey-Bass, 2005. 10. Shingo S.: Zero Quality Control: Source Inspection and the Poka-yoke System. Portland, OR: Productivity Press, 1986. 11. Womack J., Jones D.: Lean Thinking: Banish Waste and Create Wealth in Your Corporation. New York: Free Press, 2003. 12. Frankel A., et al.: Patient Safety Leadership WalkRounds. Jt Comm J Qual Saf 29:16–26, Jan. 2003. 13. Berry D., Carlson J.: SafePlace: A Starting Point. Deerfield, IL: St. Paul Health Care, Mar. 2002. 14. Agency for Healthcare Research and Quality: Patient Safety Culture Surveys. http://www.ahrq.gov/qual/hospculture/ (last accessed May 2, 2007). 15. The Joint Commission: 2007 Hospital/Critical Access Hospital National Patient Safety Goals. http://www.jointcommission.org/ PatientSafety/NationalPatientSafetyGoals/07_hap_cah_npsgs.htm (last accessed May 2, 2007).
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
Methods, Tools, and Strategies
Applying the Toyota Production System: Using a Patient Safety Alert System to Reduce Error
Cathie Furman, R.N., M.H.A. Robert Caplan, M.D.
Recently, an ambulatory surgery center registered nurse (R.N.) called a patient safety alert (PSA) on herself. At the start of the day, she had prepared a vial of Ancef® (cefazolin for injection), noting to herself that “sometime today I am going to need it.” Sure enough, a few hours later, when readying a patient for a procedure, she hung the already prepared intravenous (IV) Ancef. She then began reviewing the patient’s medical record. When noticing that the patient was allergic to penicillin, she immediately stopped the infusion (Ancef is contraindicated in patients with known allergy to penicillin). Even though it was a near miss, with no harm done, she used the PSA system to report the event. Within hours, the senior executive accountable for the area had conducted an investigation and implemented system-based changes, such as implementation of standard work to ensure that no nurse uses a previously prepared IV without first checking the medical record. This is one of thousands of changes that we have made at Virginia Mason Medical Center (VMMC; Seattle) during the past few years to keep our patients safe. Ideally, error-prone processes should be fixed as soon as possible, but conventional tools for quality improvement can be very slow and inefficient. At VMMC, it often took weeks or months before a traditional quality incident report even received any attention. This delay made it hard to fully understand all the facts and the root causes that had led to the original problem; the trail was long since cold. Our traditional quality incident program was also characterized by a culture of 376
July 2007
Article-at-a-Glance Background: In 2002, Virginia Mason Medical Center (VMMC) adapted the Toyota Production System, also known as lean manufacturing. To translate the techniques of zero defects and stopping the line into health care, the Patient Safety Alert (PSA) system requires any employee who encounters a situation that is likely to harm a patient to make an immediate report and to cease any activity that could cause further harm (stopping the line). Implementing the PSA System—Stopping the Line:
If any VMMC employee's practice or conduct is deemed capable of causing harm to a patient, a PSA can cause that person to be stopped from working until the problem is resolved. A policy statement, senior executive commitment, dedicated resources, a 24-hour hotline, and communication were all key features of implementation. Results: As of December 2006, 6,112 PSA reports were received: 20% from managers, 8% from physicians, 44% from nurses, and 23% from nonclinical support personnel, for example. The number of reports received per month increased from an average of 3 in 2002 to 285 in 2006. Most reports were processed within 24 hours and were resolved within 2 to 3 weeks. Discussion: Implementing the PSA system has drastically increased the number of safety concerns that are resolved at VMMC, while drastically reducing the time it takes to resolve them. Transparent discussion and feedback have helped promote staff acceptance and participation. Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
secrecy that limited knowledge of the problem to those with a “need to know.” This reduced creative thinking about remedies and limited the opportunity to broadly leverage safety lessons. It also meant that the organization had no clear sense of which processes were broken or dangerous, thereby hampering engagement, collaboration, and buy-in regarding safety priorities.1 Overall, the long delays and lack of feedback contributed to the prevalent cultural belief among providers and administrators that a certain frequency of “complications” was expected and tolerable. All this changed dramatically at VMMC when we converted to the Virginia Mason Production System in 2002, our version of the Toyota Production System, for management of our health care system.2 A primary goal of the Toyota System is to pursue zero defects by finding and fixing reversible mistakes as soon as possible—at the times and the places at which these reversible mistakes occur. We believe that this approach, originally developed to improve manufacturing processes, provides an effective and efficient way to improve health care safety. This article describes the implementation of one key aspect of the VMPS—reporting safety-related problems and, if necessary, “stopping the line.”3,4 We also describe our parallel efforts to create a culture in which employees feel safe about reporting problems and in which management feels comfortable about disseminating the lessons learned from reporting and solving them.
Background Health care depends on complex human activity. Human failings in dealing with complexity are therefore bound to result in failings in the delivery of health care. For example, humans are not adept at identifying the complex coincidences that cause system failures, particularly the inefficiency and errors of health care.5–7 James Reason has cautioned that complex systems can “spring nasty surprises.”8 In health care, these “surprises” result in enormous losses, including the loss of life. According to Morath and Tunbull, preventable medical accidents could cost the United States nation $17–$29 billion a year.9 Zero defect theory, developed by Shigeo Shingo, a Japanese industrial engineer at Toyota, regards human mistakes as inevitable but also reversible if they are caught soon enough.10 If mistakes are found and fixed soon
July 2007
enough, the final product—be it cars or health care—will not have defects, defined as mistakes that have persisted so long that they have become irreversible or hard to fix. Shingo discovered that mistakes can be kept from becoming defects if processes are broken down into small, understandable steps and if inspection and correction occurs at each step. Weick and Sutcliffe characterize a high-reliability organization as one that acts “mindfully,” organizing itself in a way that makes it “better able to notice the unexpected in the making and halt it [our italics].”5 (p. 3) Certainly, the pursuit of zero defects requires such mindfulness—“attention must be paid”—everywhere and all the time. Thus, in a system that deploys zero defects, every individual in the workplace must become a safety inspector and must have the authority to call out mistakes (or “stop the line”) as soon as possible. Some manufacturing companies have pursued the mindful elimination of error in all processes so that they can provide the highest quality, the safest products and services, outstanding customer satisfaction, and healthy balance sheets. These companies use a set of concepts and methods which have become known collectively as lean production, although at Toyota, where these methods originated, they are known as the Toyota Production System.10 In 2002, VMMC adopted this system as its management method and named it the Virginia Mason Production System. The basic tenets of lean production include the standardization and detailed specification of work processes, the organization of work in such a way that unexpected and undesirable events are easy to spot, and the deployment of activities that find and fix mistakes as soon as they are made, correcting those mistakes before they become defects that are difficult to reverse or remedy.2,11 Additional assistance is provided, if needed, by supervisors and managers, who must be adept at collaborative problem-solving. If the mistake cannot be fixed on the spot, the faulty process is stopped until the mistake is corrected; this is so no other defects occur. (No other patient will have to suffer this particular harm or risk of harm.) This step is called stopping the line. At Toyota, if an employee encounters a problem that cannot be fixed within a prescribed time, he or she physically makes a signal that stops the production process. In health care, this is comparable to a well-staffed
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
377
The Joint Commission Journal on Quality and Patient Safety
and coordinated response to a “code call” for cardiac arrest, bringing needed resources to help the staff fix the problem rather than using traditional work-arounds.
to our community. We also believe that the long-term safety benefits of honesty and transparency will far outweigh any short-term liabilities.
Culture of Safety and Reporting
Implementing the PSA System
A key requirement for deploying zero defects is a workforce empowered to speak up when a mistake or error occurs. This requires overcoming employees’ natural and justified fears of retribution; it is a deliberate, slow, and difficult process. Health care has deeply ingrained cultures of blame, denial, and fear5; the culture of blame, for example, creates the presumption that errors and poor outcomes occur because of incompetence. In addition, the hierarchical culture of health care discourages subordinates from calling out the mistakes of colleagues in more powerful positions. Hence, surgery residents are reluctant to point out mistakes made by surgery attendings, and nurses are often fearful of mentioning mistakes made by physicians. To implement zero defects, these old attitudes must be eliminated, and three very different expectations must be developed5: ■ It is safe to report mistakes. ■ When mistakes are reported, they will be corrected. ■ Those who report mistakes will be praised. At VMMC, we have engaged in focused efforts to improve the culture of safety since 2002. This work has been guided in four important ways: 1. We have used culture of safety surveys since 2002. 2. We implemented executive walkrounds12 in 2003, a process that brings all 28 members of our executive team into direct dialogue with our staff about any obstacles to delivering safe health care. 3. Every month our professional staff and management team hear about errors that occurred and what actions were taken to correct them. We believe that this kind of transparency within our organization eradicates the concept that errors are something “kept secret” or something that “can’t happen here.” 4. We work very hard to “close the loop” and let our staff know that each report has been received and what has been done about it. We recognize that transparency may increase the risk for legal action and negative media coverage, but we believe it to be the right thing to do for our patients. Put another way, we believe that failing to report, recognize, or act on safety problems is an unacceptable breach of duty
VMMC’s PSA system, which went into effect in September 2002 by the chief executive officer (CEO)’s demand, is a direct application of stopping the line in health care. The goal is to detect and fix every safety hazard (mistake) as soon as it occurs, thus preventing the hazard from creating more serious harm downstream. Stopping the line can mean stopping a person as well as a process. If any VMMC employee’s practice or conduct is deemed capable of causing harm to a patient, a PSA can cause that person to be stopped from working until the problem is resolved ([Sidebar 2] see page 383). A policy statement, senior executive commitment, dedicated resources, a 24-hour hotline, and communication were all key features of implementation.
378
July 2007
POLICY STATEMENT The first step was to create a PSA policy. Any staff member who encounters a situation that he or she believes is likely to harm a patient is required to make a report to the department of patient safety immediately, using the 24-hour hotline. Furthermore, the staff member must cease any activity that he or she believes could cause further harm. Initially, in September 2002, only “significant” safety incidents—near misses and actual adverse events—were to be reported. Yet staff members had told us that they felt “ignored” or “unheard” if we made a determination that a report was not considered to be a PSA. We learned that “significant” is truly in the eye of the beholder. In addition, VMMC executives found it hard to distinguish between a safety PSA and a quality incident. Such disagreement or confusion about what was to be considered a safety incident absorbed time and resources, delaying resolution. Consequently, in February 2005, we implemented a policy that differentiated between high, moderate, and low risk of harm and high, moderate, and low likelihood of recurrence, as follows: ■ Red PSAs are events that cause serious harm or have the potential to occur frequently; sentinel events are placed in
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
this category. The accountable executive is expected to conduct an investigation, stop the process or provider if needed, and resolve the problem within 24 hours. ■ Orange PSAs are events that cause less severe harm or are less likely to recur; near misses are placed in this category. Orange PSAs require investigation and resolution within 72 hours. ■ Yellow PSAs, events that were previously called quality incidents, have the lowest potential for harm. We expect yellow PSAs to be resolved within one week. The patient safety specialist, who categorizes the PSA as red, orange, or yellow in concert with the accountable executive leader (for example, chief of surgery for an averted wrong-site surgery), is expected to conduct the entire investigative and problem-solving process within 24 hours to one week, in stark contrast to the lengthy quality incident report process that was formerly used at VMMC.
SENIOR EXECUTIVE COMMITMENT Each PSA requires that the accountable senior executive be notified and that he or she go to the “shop floor” to determine whether the process or person needs to be stopped. To remove a high-powered physician or to stop a revenue-generating process requires executive authority at VMMC, as we believe it would at most hospitals. Each PSA is also reviewed by the CEO, the president, and the senior vice president of quality. If the accountable executive does not believe that the reported incident is a patient safety concern, the CEO and the senior vice president of quality must agree with that determination. If they do not, the PSA process moves forward. The senior vice president of quality is responsible for overseeing each PSA in order to remove barriers and to aid in resolution if needed. The board-level quality oversight committee reviews the PSA data at each meeting and engages in discussion about the process and specific corrective action plans, holding the senior executive team accountable for timely, effective actions.
DEDICATED RESOURCES The PSA initiative launched a consolidation of the risk management and quality assessment functions into the newly created patient safety department. The new role of patient safety specialist was created. This role, for example, provides consultation to the executive and facilitation of
July 2007
team huddles, coordinates documentation of analytical tools (for example, Failure Mode and Effects Analysis, root cause analysis) when needed, reviews the PSA documentation for completeness, tracks and trends final reports, and generates and distributes data for the quality oversight committee. People who had been claim managers and quality resources specialists (nurse reviewers) under the old system were invited to apply for this new position. Before the merger, there had been animosity between the two groups. Nevertheless, within two months, members of both groups were pooling their skills and were cross-trained, and by a year later or so, the team worked well together.
A 24-HOUR HOT LINE The commitment to rapid resolution of PSAs required a hotline, with a patient safety specialist always on call, so that a PSA can be called at any time. In 2005, VMMC implemented a Web-based reporting system so that staff could also report PSAs on the hospital intranet. When a report is entered, the on-call patient safety specialist is alerted and the case is entered into a database that is used for analysis and aggregate reporting.
COMMUNICATION Data were collected, analyzed, and reported quarterly regarding the total percentage of staff who were aware of the PSA policy and knew what to do if they identified an error or had a patient safety concern. During each executive walkround, we distributed printed business cards that included the PSA hotline telephone number and held a question-and-discussion session. In addition, at all monthly management meetings and professional staff meetings, a case study of an actual PSA was presented.
MEASURES We use a Web-based database to collect the following information: ■ Number of PSAs reported, by type of PSA ■ Number of PSAs reported, by role function ■ Number of days elapsed from first report to resolution ■ Number of staff and processes taken off-line ■ Percentage of staff who are aware of the PSA policy ■ Percentage of staff who feel comfortable in reporting errors
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
379
The Joint Commission Journal on Quality and Patient Safety
Table 1. Categories of Patient Safety Alerts (PSAs), 2002–2006 Category*
2002 (5 mos)
Adverse drug reaction Airway management Blood/blood component Care/service coordination Diagnosis/treatment Diagnostic test Employee general incident Environment (facilities) Fall ID/documentation/consent Infection control Lab specimen/test Line/tube Medication/IV safety Restraints/supportive devices Safety/security/conduct Skin/tissue Surgery/procedure Surgical site infection Systems† Vascular access device Total declared PSAs
2003
2004
6
6
19
1
22
22
4
17
32
3
6
13
4
74
118
18
125
204
2005
2006
23 24 29 452 62 82 35 73 260 166 116 449 16 448 6 78 20 84 11
31 28 16 1103 61 100 27 76 285 170 124 405 22 580 5 119 34 99 2
16 2450
28 3315
* ID, identification; IV, intravenous. † Since 2004 “Systems”-related PSAs have been reassigned to more specific categories.
■
Percentage of staff who believe that VMMC treats patient safety as a high priority Culture of safety surveys have been conducted three times: in 2002—before the PSA system’s implementation; and after its implementation, in 2004 and 2005. The first two surveys used a commercially available tool,13 and the 2005 survey used the Agency for Healthcare Research and Quality (AHRQ) culture of safety survey tool.14
Results MEASURES Since the PSA system’s inception in September 2002, 6,112 PSAs have been reported (as of December 2006), addressing errors ranging from outdated tuna sandwiches on dietary trays to fatal medication errors. The PSA system has undergone continuous refinement since its inception. In the beginning, staff members were 380
July 2007
slow to call PSAs. As awareness of the policy grew and staff noticed that concerns were really being fixed, the number of PSAs increased dramatically (Table 1, above). By 2004, 100% of the staff were aware of the PSA system. The terms incident report or QA report have now been replaced by PSA. Table 2 (page 381) provides examples of the types of safety problems that have been reported. The goal is to resolve each PSA within 48 hours, but we have not achieved this yet. In fact, the number of reports significantly increased in 2006, resulting in an average time until resolution of 33 days (Table 3, page 382). In response, we have almost doubled the number of patient safety specialists—to five—dedicated to this work in 2007. Even these values represent a large improvement over the 3-month to 18-month turnaround time associated with our previous conventional system of incident reporting and committee review.
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Table 2. Patient Safety Alert Category Examples* Airway Management ■ Self-extubation ■ Nebulizers not available during emergent treatment
ID/Documentation/Consent ■ Patient misidentification ■ Inadvertent release of patient health information ■ Missing consents
Blood/Blood Product ■ Delayed processing of blood orders
Infection Control ■ Sharps use ■ Inadequate setup of isolation precautions ■ Personal protective equipment
Care/Service Coordination ■ Communication and handoff issues among healthcare team
Lab Specimen/Test ■ Mislabeled specimens or unlabeled specimens
Diagnosis/Treatment ■ Delayed diagnosis or treatment resulting in worsened condition ■ Incorrect diagnosis or incorrect treatment provided to patient
Line/Tube ■ Infiltration of IV line ■ Incident related to order, preparation, insertion, or use of line or tube
Diagnostic Test ■ Missed orders or delayed communication of critical results
Medication/IV Safety ■ Delay in obtaining and administering STAT medications ■ Wrong medications administered to patient ■ Use of dangerous abbreviations on medication orders ■ Near miss of wrong dose or wrong medication
Environment ■ Staffing concerns ■ Malfunctioning equipment
Safety/Security/Conduct ■ Wandering patient ■ Allegations of sexual misconduct or assault ■ Verbal or physical abuse by staff ■ Drug-seeking behavior by staff
Employee ■ Injury from assisting patient ■ Noncompliance with fitness-for-duty requirements ■ Needlestick injury
Skin/Tissue ■ Trauma of skin/tissue, such as phlebitis, rashes
Fall ■ Patient fall in room, with or without assistance
Surgery/Procedure ■ Ordering, preparation, or performance of surgical procedure or anesthesia ■ Equipment not available for scheduled surgery
* ID, identification; IV, intravenous.
SPEED OF THE PSA SYSTEM
INAPPROPRIATE PHYSICIAN BEHAVIOR AS AN ISSUE
One foundation of the PSA process is quick identification of the root cause, followed by a quick remedy. This is particularly important when providers or processes are taken off-line. The intent is to keep patients safe, but long delays in a process can lead to other harm. As an example of the speed of the PSA system, consider Case Study 1 (Sidebar 1, page 383).
At the outset of implementing the PSA policy and process, we made a deliberate and explicit decision that inappropriate physician behavior would be considered a PSA. We felt that this would be an important way to signal to the organization that safety was a responsibility that applies to everyone in the health care hierarchy. Case Study 2 (Sidebar 2, page 383) provides an example of how
July 2007
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
381
The Joint Commission Journal on Quality and Patient Safety
Table 3. Average Number of Patient Safety Alerts (PSAs) and Days to Resolution, 2002–2006 Metric
2002 (5 mos)
2003
2004
2005
2006
Average no. of PSAs/month
3.6
10.4
17
204
276
Average no. of days to resolution
18.4
13
13.7
12
33
Table 4. Number of Systems and Providers Taken Off Line, 2002–2006 2002 (5 mos)
2003
2004
2005
2006 (6 mos)
Employees Taken Off-line
6
5
14
12
19
Processes/Equipment Taken Off-line
1
4
8
8
10
this works. As shown in Table 4, above, 44 providers (physicians, nurses, and other health care workers) have been taken offline since the inception of the PSA process. Twenty-six (60%) of these persons have returned to work after appropriate remedial plans were developed, and 30% were terminated. Common problems identified include fatigue due to on-call duties, conflicts between operating room schedules and clinic schedules, and difficulties adjusting to computerized medical systems. The organization has put into place improvement efforts to address these issues. We report and discuss PSAs that involve providers, as we do any other PSAs, at our regular manager meetings and professional staff meetings. This makes it clear that safety applies to everything and everyone in health care delivery.
SOURCE OF PSAS Who reports PSAs? Initially, most reports came from nursing managers. Given the frontline position of nurses and the frequency of their contact with patients, we did not find this surprising. We were surprised that no reports came from pharmacists, given that VMMC delivers more than one million inpatient drug doses a year. We learned from the pharmacists that, in many cases, errors which we considered “significant” were taken by the pharmacists to be “normal” errors which they “corrected routinely.” This attitude changed dramatically after a patient was harmed by an illegible order that was misinterpreted by both a pharmacist and a nurse. At the time, we had already devel-
382
July 2007
oped preprinted admission orders (to reduce legibility issues at admission) and had implemented The Joint Commission’s recommendation to eliminate the use of dangerous abbreviations.15* Our policy on dangerous abbreviations requires health unit coordinators and pharmacists to stop the line if a handwritten order includes one of the specified dangerous abbreviations. However, a PSA occurred nonetheless (Sidebar 3, page 384). The investigation uncovered (1) the lack of clarity about whose job it was to stop the line and (2) the fear of retribution that the health unit coordinators and pharmacists felt. As part of the PSA’s resolution, stepby-step protocols were prepared that outline each group’s role if a dangerous abbreviation is used. Given the gravity of the error, VMMC’s top administration, for the first time, publicized the PSA incident—in an e-mail to all 5,000 staff and physicians. The e-mail stated that VMMC had harmed a patient and that everyone needed to be a safety inspector to ensure that it didn’t happen again. In addition, the hospital’s senior vice president and the director of pharmacy held focus groups with pharmacists to learn about and discuss their fears about and barriers to PSA reporting. One result of this event was that the pharmacists developed a clearer understanding of their responsibility to stop the line. As of December 2006, a total of 6,112 PSAs had been * Joint Commission National Patient Safety Goal 2B, “Standardize a list of abbreviations, acronyms, symbols, and dose designations that are not to be used throughout the organization.”
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Sidebar 1. Case Study 1—A Mix-up of Armband Colors At VMMC, we often use visual controls to enhance our understanding of patient needs. At VMMC, patients are given colored wristbands that signal special conditions such as allergies and “no code” status. One afternoon, a registered nurse (R.N.) coming on duty was rounding on her patients. She went in to do an assessment on a patient who had been admitted earlier that day with a new diagnosis of lung cancer. While doing the physical assessment, the R.N. noted that the patient had a pink wristband, which is our visual control for “do not resuscitate” (DNR) status. The R.N. thought that was odd for a newly diagnosed patient who was about to receive therapy, so she asked the patient about the wristband. The patient explained that the previous nurse had put on the wristband to indicate drug allergies. The color for drug allergies is orange, not pink. The R.N. immediately removed the pink band, gave the patient the correct orange band, and called a PSA. The chief nursing executive and the director of central ser-
vices immediately came to the floor to assess the situation. They huddled with the nurse and the manager. The investigation to find the root cause included asking “why” five times (each “why” demands an explanation at a deeper level or at a further step upstream). The root cause analysis quickly revealed that the R.N. on the previous shift was color-blind and had mixed up the colors. The team came up with the idea of adding words to the wristbands. Information systems reprogrammed the software to add allergy to the orange wristbands, DNR to the pink wristbands, and so forth. All of this was completed within hours of the initial event, ensuring that the process of banding of patients could continue—but safely. The nurse explained the error to the patient and said that a system was being put into place to fix it. When the new identification bracelets were produced the next day, the patient could see what had changed to ensure that an error would not occur again.
Sidebar 2: Case Study 2—A Noncompliant Oncologist In 2002, an oncologist ordered chemotherapy for a patient. The standard of care for this protocol required prior determination of an adequate cardiac ejection fraction and an alkaline urine. Before initiating the protocol, the oncology R.N. noted that the patient’s urine was still acidic and that the ejection fraction had not been assessed. The nurse called the physician, who replied that neither determination was necessary and ordered the nurse to start chemotherapy. The implicit message was, “Just do what I say because I am the doctor and you are just a nurse.”
dard of care. The oncologist then called the nurse back and was verbally abusive with her for calling the chief. The nurse notified the chief of cancer services that she had received an abusive phone call. The chief, determining that the oncologist’s behavior created a potential for serious and recurring harm, called a PSA, placed the oncologist on administrative leave, and temporarily suspended his privileges to admit patients to the hospital. A remedial plan for this oncologist was developed and implemented within 12 hours. Reinstatement of privileges required the following:
Instead, the nurse contacted the chief of cancer services and explained her concerns. The chief of cancer services, who agreed that the standard of care was to complete the standard testing before starting the chemotherapy protocol, called the oncologist to remind him of the applicable stan-
1. An apology to the nurse 2. A review of the literature on the standards for chemotherapy for the case in question 3. Anger management counseling 4. A follow-up plan to document improved behavior
reported, with the following sources: ■ Managers, 20% ■ Physicians, 8% ■ Nurses, 44% ■ Pharmacists, 5% ■ Nonclinical support personnel, 23% ■ Patients/families .003% In the early years of the PSA system, reports came almost exclusively from members of VMMC. However,
July 2007
since 2005 we have seen more participation by the community physicians who practice there. Furthermore, VMMC’s admission brochures tell patients about the safety programs and invite them and their families to participate in our safety efforts. Recently we had our first PSA report from a patient’s family!
CULTURE OF SAFETY AND REPORTING A variety of data and anecdotal evidence suggest that
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
383
The Joint Commission Journal on Quality and Patient Safety
Sidebar 3. A Misread Dosage An R.N. called a PSA after a patient had received three times the amount of a prescribed medication, which led to the patient’s acute deterioration and an emergency transfer to the critical care unit. The hospital administrator, the chief nursing executive, and the medical director of hospital services were called to investigate. They found that the handwritten medication order had been misinterpreted. The order was for 25 mg, but had been read as 75 mg. The medical director interviewed the hospitalist, using the technique of asking “why” five times, and found that a sequence of errors had occurred. Why had the wrong dose been given? In looking at the order, half of the team—process owners who help with the investigation and issue resolution—thought that it was 25 mg and the other half thought that it was
75 mg. It was noted that the prescribing physician had used one of the dangerous abbreviations that was not permitted. Why, then, had neither the unit clerk nor the pharmacists stopped the line, which policy dictated and which would have created an opportunity to determine the correct dose of the drug? Each thought the other role was accountable. Why did the R.N. administer a drug which didn’t make clinical sense for a frail patient? The R.N. was inexperienced and unfamiliar with the patient. In addition, the prescribing physician was so busy that he did not follow his usual practice of discussing the medication orders with the nurse, which would have created another opportunity to determine the correct dose.
Table 5. Subset of 2005 AHRQ Patient Safety Culture Survey*
* Agency for Healthcare Research and Quality: Comparing Your Results: Preliminary Benchmarks. Hospital Survey on Patient Safety Culture, Mar. 2006. http://www.ahrq.gov/qual/hospculture/prebenchmk.htm (last accessed May 9, 2007).
VMMC staff are beginning to believe that the PSA system leads to improvement in safety and quality. Although the change in the survey tool prevents direct comparison of culture-of-safety results for 2005 with those of 2002, the results for 2005 are statistically higher than the AHRQ 384
July 2007
benchmark data (Table 5, above). Anecdotally, we receive positive feedback about the PSA system during executive walkrounds, and we believe that the strong increase in reporting is another indicator that the system is accepted by our staff.
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
The Joint Commission Journal on Quality and Patient Safety
Patient Safety Alerts (PSAs) by Level, 2005 and 2006
infrequent red and orange PSAs as a type of mistake that we want to eradicate immediately and permanently. In contrast, we look at the aggregate data for our yellow PSAs and ask which types occur most frequently. We then study those types to determine if there are broader, systembased strategies that can be used to make improvements in safety and quality. We are particularly looking for problems that might turn up throughout the hospital. Each year, as part of the goal setting process for the organization, these data are used to determine what the organizational safety initiatives will be.
Summary and Conclusions The PSA system has proven to be the single most important tool to make our care Figure 1. Red PSAs are high risk for actual harm; orange are those that are "near safer. It provides a quick, timely process for misses;" and yellow PSAs are concerns that are low risk but concerning to the improvement and feedback to our staff reporter—those that get in the way of providing safe care. that their concerns will be listened to. Lessons learned are as follows: 1. Executive leadership is a prerequisite. How many PSA reports are enough? We had initially 2. Reporting should be easy, with multiple methods availset a yearly target of 800 PSAs, based on our belief that able. formal complaints about adverse outcomes still reveal only 3. “Significant” is in the eye of the beholder: Open the a small fraction of the preventable medical errors made. floodgates for all concerns. We have well exceeded that target when red, orange, and 4. Claims management staffing will go down as patient yellow PSAs are all included. The target is now defined as safety alerts go up. Right before the start of the PSA sysa total of 800 red and orange PSAs, of which we are avertem, we had six claims managers and three R.N. reviewers; aging about 300 a year (for example, 57 and 245 in 2006, we now have three claims managers and five patient safety as shown in Figure 1, above). specialists. 5. Be prepared to change the processes of care as the orgaDiscussion nization learns from the PSAs. J Implementing the PSA system has drastically increased the number of safety concerns that are resolved at VMMC while drastically reducing the time it takes to resolve them. Transparent discussion and feedback have played an Cathie Furman, R.N., M.H.A., is Senior Vice President, important role in staff acceptance and participation. Our Quality and Compliance, Virginia Mason Medical single best indicator of the viability of this approach is that Center, Seattle, and Robert Caplan, M.D., is Medical reporting of safety and quality concerns has continued to Director, Quality, and Staff Anesthesiologist, Department of Anesthesiology. Please address requests for grow since the inception of this system in 2002. reprints to Cathie Furman, [email protected]. We are beginning to develop tactics for understanding and reacting to PSA data. We view each of our relatively
July 2007
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations
385
The Joint Commission Journal on Quality and Patient Safety
References 1. Tucker A., Edmondson A.: Why hospitals don’t learn from failures: Organizational and psychological dynamics that inhibit system change. California Management Review 45:55–72, Winter 2003. 2. Ohno T.: Toyota Production System: Beyond Large Scale Production. Portland, OR: Productivity Press, 1988. 3. Davidoff F., Batalden P.: Toward stronger evidence on quality improvement. Draft publication guidelines: The beginning of a consensus project. Qual Saf Health Care 14:319–325, Oct. 2005. 4. Barley S.R.: Images of imaging: Notes on doing longitudinal field work. Organizational Science 1:220–247, Aug. 1990. 5. Weick K., Sutcliffe K.: Managing the Unexpected: Assuring High Performance in an Age of Complexity. San Francisco: Jossey-Bass, 2001. 6. Institute of Medicine: To Err Is Human: Building a Safer Health System. Washington, D.C.: National Academy Press, 2000. 7. Institute of Medicine: Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C.: National Academy Press, 2001. 8. Reason J.: Human Error. New York: Cambridge University Press, 1990.
386
July 2007
9. Morath J., Turnbull J.: To Do No Harm. San Francisco: Jossey-Bass, 2005. 10. Shingo S.: Zero Quality Control: Source Inspection and the Poka-yoke System. Portland, OR: Productivity Press, 1986. 11. Womack J., Jones D.: Lean Thinking: Banish Waste and Create Wealth in Your Corporation. New York: Free Press, 2003. 12. Frankel A., et al.: Patient Safety Leadership WalkRounds. Jt Comm J Qual Saf 29:16–26, Jan. 2003. 13. Berry D., Carlson J.: SafePlace: A Starting Point. Deerfield, IL: St. Paul Health Care, Mar. 2002. 14. Agency for Healthcare Research and Quality: Patient Safety Culture Surveys. http://www.ahrq.gov/qual/hospculture/ (last accessed May 2, 2007). 15. The Joint Commission: 2007 Hospital/Critical Access Hospital National Patient Safety Goals. http://www.jointcommission.org/ PatientSafety/NationalPatientSafetyGoals/07_hap_cah_npsgs.htm (last accessed May 2, 2007).
Volume 33 Number 7
Copyright 2007 Joint Commission on Accreditation of Healthcare Organizations