- Appreciate the role of Reason's Swiss Cheese Model in medical errors
- Understand the process of analyzing a single error
- Provide suggestions for remediation
A 79-year-old woman presented to an after hours clinic with a 1-week history of diarrhea and progressive weakness. Due to signs of dehydration, the patient was directly admitted to the hospital. Past medical history was notable for stroke with residual left-sided hemiparesis, hypertension, coronary artery disease with ischemic cardiomyopathy, peptic ulcer disease, asthma, and obesity. Two weeks prior to this admission, she had spontaneously developed right ankle and foot pain and had been evaluated in the emergency department (ED) of another hospital. The family was told of a possible fracture and a splint was applied. She was instructed to follow up with an orthopedist as soon as possible. Due to transportation difficulties, the patient was not seen in follow up.
On physical examination, she was afebrile and appeared weak. She had a left-sided hemiparesis. The right ankle and foot were in the same splint that had been applied 2 weeks earlier. When examined, the ankle had a normal range of motion with no localized tenderness. A stool specimen collected in the ED was subsequently positive for Clostridium difficile toxin. At the time of admission, a release of information was signed and faxed to the other hospital to obtain records of the recent ED visit for the ankle and foot injury. The family requested an orthopedic consultation to expedite work-up. Outside records of the previous ED visit did not arrive promptly, so another x-ray was taken of the right foot and ankle. This x-ray was read by the radiologist as showing a right ankle trimalleolar fracture and dislocation. The consulting orthopedist reviewed the x-ray report then briefly examined the patient. Surgery was recommended and discussed with the family, and consent was obtained.
The next morning, the patient was taken to the operating room (OR), and spinal anesthesia was administered. The orthopedist was scrubbed and was preparing to operate. The ankle x-ray was on the view box in the OR. Prior to making an incision, the orthopedist reviewed the x-ray and was shocked to notice that it was a left ankle x-ray showing a trimalleolar fracture. A prompt examination of both of the patient's ankles under anesthesia did not demonstrate any clinical evidence of fracture or dislocation. The x-ray was clearly labeled as belonging to the patient. Stat x-rays of both ankles were then done in the OR. The left ankle was intact, and the right showed an intact ankle with a healing fracture of the fifth metatarsal bone.
During the ensuing confusion, one of the OR technicians recalled that another patient had undergone an operative reduction-internal fixation (ORIF) of a left ankle trimalleolar fracture 2 days prior. It was later confirmed that the x-ray showing the left ankle trimalleolar fracture was mislabeled by date and patient and belonged to this other patient who already had surgery.
The spinal anesthesia was reversed, and the patient was returned to her room and fortunately did not have any consequences. Full disclosure and an apology were given to the family.
The patient continued to recover from the dehydration and colitis and was able to be discharged from the hospital. Treatment for the metatarsal fracture consisted of a supportive boot. By the time of discharge, a faxed copy of the ED records from the outside hospital had been received. Included in these records was an x-ray report describing a non-displaced, fifth metatarsal fracture of the right foot.
Performing an invasive procedure on the wrong patient should never happen. Neither should operating on the wrong side of the right patient. Reliable estimates of the frequency of this kind of adverse event are not available. The data that do exist suggest that events like these are underreported to the voluntary sentinel events database of the Joint Commission on Accreditation of Healthcare Organizations (JCAHO), as well as to state programs that require adverse event reporting.(1-3) Wrong-patient or wrong-site invasive procedures may be uncommon but “for patient and doctor it is the ultimate nightmare.”(4)
Broadly speaking, adverse events in hospitals occur through two different mechanisms. A single mistake, if it is serious enough, may cause harm by itself. Alternatively, and more commonly, many smaller errors may occur—not one of which alone is severe enough to cause harm. In combination, however, as seen in this case, they become toxic. James Reason has developed a powerful explanatory model to understand adverse events that occur by the second mechanism, which he has called “organizational accidents.”(5) Synthesizing the work of many psychologists, accident experts, and organizational sociologists, Reason provides a conceptual framework that delineates how errors made by individuals interact with system defects in complex organizations to cause harm.(5) Like other complex organizations, hospitals put defenses into place to prevent errors from doing harm. Examples of such defenses include training programs; safety protocols, policies, and procedures; and computerized decision support tools. Every layer of defense has weaknesses. If an error gets past one defense, another layer of defense usually prevents harm from occurring. Adverse events occur only when all the defenses around a particular patient's situation have been circumvented by many errors. Reason has referred to this framework for understanding adverse events as the “Swiss Cheese Model.”(6)
Let us review this adverse event to determine (i) which of the two causal pathways was involved (single error versus Swiss cheese); (ii) whether the Swiss cheese pathway was causative and which defenses failed to prevent harm; and (iii) what remedial action might be called for. We will start by identifying the primary errors made by individuals during the course of this patient's care that contributed to causing the event.
The first error (or series of errors) occurred 2 weeks prior to admission when the staff in the ED who diagnosed the non-displaced metatarsal fracture failed to communicate this diagnosis clearly and unambiguously to the patient and her family. If the patient and family had known the correct diagnosis of her foot pain and reported that history at the time of admission, this mishap might have been prevented, particularly if that information had been supplemented by confirmatory documentation from the ED. The next group of errors occurred when the patient was admitted. The physician in the ED who examined the patient and found a normal right ankle failed to communicate his or her findings to the physician responsible for admitting the patient or to the consulting orthopedic surgeon. The case summary does not identify the admitting physician, but since the patient was admitted primarily to treat dehydration, it is unlikely that she was admitted directly to the service of the orthopedic surgeon. It is not clear from the case narrative whether the ED physician ordered the repeat foot and ankle radiographs while the patient was in the ED or the consulting orthopedic surgeon ordered them after the patient had been admitted. If the ED physician ordered them, he or she erred further in not personally reviewing the films. If the patient was admitted to an internist or other primary care physician, that physician also erred in not examining the patient and her x-rays.
The case summary does not reveal how the patient's foot and ankle radiographs were mislabeled in radiology, but it is very likely that several errors were involved. Initially, an erroneous report on the patient was generated showing a right trimalleolar fracture with dislocation. It is highly implausible that the patient's actual films were misread by the radiologist. A non-displaced fifth metatarsal fracture can hardly be mistaken for a trimalleolar fracture with dislocation. As we learn later, it is far more likely that a different patient's films were somehow mislabeled with the current patient's identifying information and current date. Why the initial erroneous report identified the side of the trimalleolar fracture as right is also unclear. Did the radiologist misread the side of the trimalleolar fracture because the order for the current patient called for right foot and ankle films?
By any reasonable standard of judgment, the orthopedic surgeon committed by far the most serious errors in this case. The orthopedist failed to elicit or to discover the history of the patient's previous diagnosis or treatment. The history reported by the patient's family was not consistent with the x-ray findings. It is very unlikely that any hospital would let a patient with a displaced trimalleolar fracture (at risk for vascular compromise) leave the ED in a splint without an orthopedic consult. Had the orthopedist obtained this history, he or she might have been led to question the x-ray findings. Compounding this error, the orthopedist “briefly examined the patient.” It is difficult to comprehend how even a brief physical examination of this patient's foot and ankle could not have caused this physician to question the radiographic diagnosis. One is forced to suspect that the surgeon may not have removed the bed covers and visualized the right ankle. This single error is serious enough, but the surgeon made it even worse by recommending surgery on the basis of this incomplete evaluation, obtaining “consent” (which could hardly be called informed) and scheduling the operation.
The final set of errors occurred prior to the patient receiving spinal anesthesia. Surgical site verification was not conducted prior to the patient going to the OR or prior to administering spinal anesthesia. JCAHO has recommended a series of steps to prevent wrong-site, wrong-side, and wrong-person procedures.(7) The Universal Protocol clearly delineates these steps, which include (i) a preoperative verification process to ensure that all studies and records are available, have been reviewed, and are consistent; (ii) marking the operative site; and (iii) a “time out” to conduct a final verification of the correct patient, procedure, and site prior to starting the procedure.(7) There is no indication in this case that any of these procedures were undertaken. While a recent study notes the fact that the effectiveness of preoperative verification protocols has not been evaluated (8), there is every reason to believe that in this case the patient would have been spared an unnecessary spinal anesthesia if the Universal Protocol had been properly applied. The patient was spared unnecessary surgery at the last moment when the orthopedist discovered that the presumptive diagnosis was wrong. We suspect that the orthopedic surgeon was prompted to review the x-rays in the OR after finally recognizing that the ankle he or she was “preparing to operate” on looked nothing like an ankle with a trimalleolar fracture/dislocation.
Which adverse event pathway did this case follow? This is clearly an example of the Swiss cheese pathway (Figure). There was no single, critical error. Many individuals made many errors. Many defenses failed to protect this patient. Communication failed in several instances (eg, between the first ED and the patient, between the second ED physician and the orthopedist, between the admitting physician and the orthopedist). Policies and procedures were inadequate (eg, mislabeling of the x-ray, failure to implement the universal protocol). Teamwork failed in the OR when no one—nurses, OR technicians, the anesthesiologist—observed that the patient's ankle appeared normal, and no one questioned whether the procedure should continue. The failure of the orthopedist to perform an adequate history and physical examination on the patient prior to surgery stands out as the most serious individual error. But even that error combined with the second-most serious error (the mislabeling of the ankle x-ray) was not sufficient to cause this adverse event. Many other individuals, including the staff in the first ED, the second ED physician, the admitting physician, and the OR staff, had the opportunity to stop this sequence of errors before the patient was harmed. Only when all the defenses surrounding this patient failed did she experience the harm of unnecessary spinal anesthesia.
While some might call this case a “near miss” or close call—a situation that could have led to an adverse event but did not—we would call this case an adverse event. This patient was subjected to spinal anesthesia for no reason. Although the risks are low, spinal anesthesia is occasionally associated with severe risks, such as cardiac arrest and neurological complications.(9)
The institution at which this adverse event occurred should consider several remedial actions to strengthen its defenses. The process of identifying and labeling radiographs should be reviewed to discover exactly how the errors occurred in this case and improved to avoid them in the future. The Universal Protocol should be implemented in all ORs and procedure areas. The multiple communication failures suggest the need for a formal protocol to delineate precisely how responsibility for care is handed off from the ED to admitting physicians and consultants. Finally, the actions of the orthopedist should be examined by the appropriate peer review committee.
- Reason's Swiss Cheese Model, in which multiple errors combine to create major adverse events because of inadequate defenses, explains many adverse patient events in health care.
- Analysis of adverse patient events should focus on discovering which defenses failed and bolstering them.
- To strengthen defenses against wrong-site and wrong-patient invasive procedures, hospitals should implement the principles of the Universal Protocol.
Elizabeth A. Howell, MD, MPP Assistant Professor Departments of Health Policy and Obstetrics, Gynecology, and Reproductive Science Mount Sinai School of Medicine
Mark R. Chassin, MD, MPP, MPH Edmond A. Guggenheim Professor of Health Policy Chairman, Department of Health Policy Mount Sinai School of Medicine
Faculty Disclosure: Dr. Howell and Dr. Chassin have declared that neither they, nor any immediate member of their family, have a financial arrangement or other relationship with the manufacturers of any commercial products discussed in this continuing medical education activity. In addition, their commentary does not include information regarding investigational or off-label use of pharmaceutical products or medical devices.
1. Sentinel Event Alert: A follow-up review of wrong site surgery. Oakbrook Terrace, IL: Joint Commission on Accreditation of Healthcare Organizations; December 5, 2001. Available at: http://www.jointcommission.org/SentinelEvents/SentinelEventAlert/sea_24.htm. Accessed April 3, 2006.
2. Chassin MR, Becher EC. The wrong patient. Ann Intern Med. 2002;136:826-833. [go to PubMed]
3. NYPORTS—New York Patient Occurrence and Tracking System—Annual Report 1999. Albany, NY: New York State Department of Health; February 2001. Available at: http://www.health.state.ny.us/press/releases/2001/nyports/nyports.htm. Accessed April 3, 2006.
4. Bakalar N. Before first incision, checking for an X to mark the spot. The New York Times. April 18, 2006:F8.
5. Reason J. Managing the risk of organizational accidents. Aldershot, United Kingdom: Ashgate Publishing Ltd; 1997.
6. Reason J. Human error: models and management. BMJ. 2000;320:768-770. [go to PubMed]
7. Universal Protocol for Preventing Wrong Site, Wrong Procedure, Wrong Person Surgery. Oakbrook Terrace, IL: Joint Commission on Accreditation of Healthcare Organizations. Available at: http://www.jointcommission.org/NR/rdonlyres/E3C600EB-043B-4E86-B04E-CA4A89AD5433/0/universal_protocol.pdf. Accessed April 3, 2006.
8. Kwaan MR, Studdert DM, Zinner, MJ, Gawande AA. Incidence, patterns, and prevention of wrong-site surgery. Arch Surg. 2006;141:353-358. [go to PubMed]
9. Horlocker TT. Complications of spinal and epidural anesthesia. Anesthesiol Clin North America. 2000;18:461-485. [go to PubMed]
Figure. Swiss Cheese Model