Errors are inevitable in the practice of medicine. The most obvious causes are failures in individual performance related to attention, memory, knowledge, judgment, skill, and motivation. However, they also result in part from the nature of medical work, such as the complexity of medical knowledge, the uncertainty of clinical predictions, and the need to make timely treatment decisions in spite of limited or uncertain knowledge. And importantly, mistakes are caused by system factors that influence working conditions. Although much attention has been focused on the harmful effects of errors on patients, it must be understood that these incidents can be correspondingly distressing for physicians, evoking shock and feelings of remorse, guilt, anger, and fear.
If dealt with effectively, errors can provide powerful learning experiences for physicians; however, difficulty in dealing with mistakes may impede both learning and efforts to prevent future errors. Professional norms that assume physician infallibility and treat mistakes as anomalies pose significant barriers to learning. Judgmental institutional responses and fear of litigation are further disincentives to the open discussion of mistakes and reduce the potential for wider learning.
It is useful to define several terms related to what are commonly referred to as errors or mistakes. The Institute of Medicine (IOM) defines an error as “the failure of a planned action to be completed as intended (i.e., error of execution), or the use of a wrong plan to achieve an aim (i.e., error of planning). An error may be an act of commission or an act of omission.” An adverse event is an injury due to health care. Errors differ from adverse events because they do not necessarily cause harm. They differ from negligence or malpractice, which require both preventable harm and violation of the standard of practice.
Most studies of medical errors have focused on the hospital setting and on adverse events. Although the overall prevalence of errors is difficult to ascertain, it appears that they are common. Studies conducted in multiple states and countries suggest that the rate of adverse events may be as high as 10%. The Canadian Adverse Events Study reported on the incidence of adverse events among hospital patients in Canada. The authors randomly selected four hospitals (one teaching, one large community, and two small community hospitals) in each of five Canadian provinces and reviewed a random sample of charts for nonpsychiatric, nonobstetric adult patients hospitalized in each hospital during 2000. Trained reviewers screened all eligible charts, and physicians reviewed the positively screened charts to identify adverse events and to determine whether they were preventable. After adjustment for sampling, the adverse event rate was 7.5 per 100 hospital admissions (7.5%). Among the adverse events experienced by patients, 36.9% were judged to have been preventable; 20.8% of these preventable events resulted in death. According to the physician reviewers, the adverse events were associated with an estimated 1521 additional hospital days.
In a study of 254 medical residents (conducted by the authors), 45% reported a significant mistake made during the prior year. In a survey of 2637 US and Canadian physicians, 55% reported involvement in a serious error.
The prevalence of errors in outpatient practice is less well studied, but one investigator has documented a prescription error rate of 7.6%, and another suggests errors in at least one-quarter of the encounters, with a rate of adverse drug events of 27 per 100 patients. A literature synthesis found that the median preventable adverse drug events prevalence rate was 16.5% in ambulatory care-based studies. A study of ambulatory oncology care found that medication reconciliation resulted in 31 changes per 100 prescriptions. A survey of patients attending primary care clinics found that nearly one in three reported experiencing a mistake or diagnostic error by a physician in the previous 10 years, with 14% of patients having changed physicians as a result.
Mistakes occur in every aspect of medical practice—diagnosis, decision making (often because of ignorance of facts), the pace of evaluation or its timing, prescribing medications, and performing tests and procedures (Table 37-1). In the Harvard Medical Practice Study, the most common adverse events involved performance of or follow-up of a procedure or operation (35%). Failure to take preventive measures (e.g., failure to guard against accidental injury) was the next most common error (22%), followed by diagnostic errors (e.g., failure to use indicated tests, act on test results, or avoid delays in response; 14%), errors involving drug treatment (9%), and system errors (2%).
Table 37-1.Types of medical mistakes. ||Download (.pdf) Table 37-1. Types of medical mistakes.
|Error ||Example |
|Diagnosis or evaluation ||Missed diagnosis |
|Medical decision making ||Inappropriate or premature discharge |
|Treatment ||Waiting when treatment is indicated |
|Medication ||Incorrect dosage |
|Procedural complications ||Faulty technique |
|Faulty communication ||Failure to convey information during sign-out |
|Inadequate supervision ||Failure to review treatment plan |
Cognitive Errors. Cognitive errors may result from failures in perception, faulty pattern recognition, and common biases in thinking. Two broad areas of cognitive functioning are relevant in these errors: the schematic and attentional control modes. In the schematic control mode of functioning, which is typical of the expert, sequences of thinking and acting are embedded in unconscious subroutines that allow more mental energy to be available for other activities. Examples would be speaking one’s native language or driving a vehicle. In medicine, some procedural skills as well as pattern recognition skills (such as an intuitive appreciation of a patient’s condition or hunch about a diagnosis) are under schematic control. When an anomaly occurs, or when the physician is fatigued or stressed, a shift occurs to the attentional control mode, which is more energy intensive and requires more deliberate thinking and action. This model is more prevalent in the novice learning any new skill (a foreign language, a new procedural skill). The attentional control mode can more quickly degrade over time. Errors can occur in each of these cognitive modes. A momentary break in a skilled routine in surgery, or a lapse in attention that leads to missing data are common examples.
Other cognitive errors result from biases inherent in human thinking. Among these are thinking tendencies such as the “availability” bias, in which the first diagnosis that comes to mind is deemed most likely; the “confirmatory” bias, in which only data supportive of the first hypothesis is entertained; the “recency” bias, in which the last serious diagnosis managed looms large in probability estimates; and the “gambler’s fallacy,” in which a series of similar diagnoses leads the clinician to expect that the next presentation will be of a different disease.
Procedural Errors. Procedural errors occur in both surgical and nonsurgical procedures. One study suggested that major complications occurred in 3.4% of surgical operations. Wrong site surgeries and retained foreign bodies are rare, but can have devastating consequences for patients. Causes include individual errors—such as surgical technique (including slips and fumbles), judgment errors, inattention to detail—and incomplete understanding of the case. Other causes occur at other levels of the system, including poor communication among members of the perioperative team, poor planning (e.g., surgical implant not available on the day of surgery), lack of teamwork, and surgical site infection. Surgical site infections are common, with overall incidence rates in the United States of about 3%, and there is great variability associated with procedure and site. Contributions to error in nonsurgical procedures (e.g., endoscopy, lumbar puncture, line placement) include fatigue and disruptions. Many might have been prevented by use of the Joint Commission Universal Protocol or by simulation training.
System Errors. Errors occur throughout the health care system, many arising from the inherent fallibility of human beings working within systems not primarily designed for safety. Errors occur at the level of patient, task, individual practitioner, team, unit, organization, and beyond. They also occur in multiple locations, such as in the clinical diagnostic laboratories, the pharmacy, and the preparation and maintenance of medical devices. Identification errors, which involve confusing one patient, test, or treatment for another, are a common element in many adverse events. Another common occasion for errors is in the handoffs between different health care providers and locations of care. Communication problems are at the root of many adverse events as are problems with team functioning.
Medication Errors. Medication errors are very common, in part because of the ubiquity of the use of medications. In the United States, more than four out of five adults take at least one medication in a given week, and almost a third of adults take at least five different medications. The goal of safe and effective medication use is referred to as “the five rights”: the right drug, right dose, right route, right time, and right patient. However, errors can occur with any drug product at any point in the medication use process. It is estimated that over 1.5 million people are injured by medication errors each year and that, on average, hospitalized patients are subjected to an average of one medication error per day.
A combination of individual human and system-related factors can lead to adverse events. Human error, which occurs at what is sometimes referred to as the “sharp end” of adverse events, happens close to the patient and includes deficits in skill (attention, memory, and execution), knowledge, decision making, and following of rules. System errors, the “blunt end” of adverse events, occur at multiple levels, including at the levels of patient, task, individual provider, team, unit environment, department, and institution. Examples of system errors include use of handwritten prescriptions, team members who are unfamiliar with one another, inadequate communication at change of shifts, understaffed units, inadequate equipment, long work hours, and production pressures. It should be kept in mind that individuals are part of and function within the system of health care delivery.
Individual physicians report a variety of reasons for their mistakes and frequently attribute the mistakes to more than one cause. In one study, house officers most often reported that mistakes were caused in part because they did not possess specific essential knowledge (e.g., being unaware of the significance of a prolonged episode of ventricular tachycardia). Almost as often, they cited “too many tasks” (one resident neglected to continue a required medication because he was “too busy with other sick patients, and supervising interns and students”). Fatigue was a significant factor (after inadvertently ordering potassium replacement as a bolus, one resident commented, “It was 3:00 a.m., and I’m not sure I was completely awake.”). Fatigue can interfere with the clinician’s ability to rely upon the “schematic control” mode of cognitive functioning, leading to more reliance on the less efficient “attentional control” mode, which is more vulnerable to decay. The common cognitive biases mentioned above also contribute to diagnostic and decision making errors by the individual clinician.
Many errors are caused by conditions in the workplace and organizational processes largely independent of clinician attributes. Studies of system factors demonstrate the predictive power of these characteristics independent of physician identity. One study of an inpatient medical service found cross-coverage to be a significant predictor of preventable adverse events. Recent studies suggest that staffing ratios tend to be related to patient outcomes. The pressures of practicing medicine in a managed care setting, especially in prospective payment systems with higher demands for productivity, may increase the risk that a hurried physician will overlook important diagnostic information or make an error in prescription. Similarly, the incentives offered by third-party payers to order fewer diagnostic tests and to limit the number of referrals to subspecialists can lead to errors of omission.
An overlooked contributor to the prevalence of medical errors is the culture of medicine itself. Historically the “hidden curriculum” in medical training has held up an ideal of perfectionism that is heroic in fantasy but unattainable in practice. This perspective views mistakes as a failure of character, often leading to a “blame and shame” response to individual physician error. This habit of over-focusing on individual perfection has led to an unrealistic reliance on error-free performance. It underplays the “blunt end” of system errors and overemphasizes the “sharp end” of clinician error. These cultural biases tend to drive discussion of mistakes underground, thus avoiding the reporting and discussion of error that can lead to meaningful change.
A systems-oriented view acknowledges that errors are expected even in the best of organizations. Lucian Leape, a leading scholar of medical error, has suggested, “Any system that relies on error-free performance is doomed to fail. We need to build systems that assume errors will occur, but that can identify and absorb them before they become catastrophes.”
Mistakes seem to occur particularly frequently during residency training, possibly because interns and residents are learning new skills, honing their clinical judgment, and accepting new responsibilities. Additionally, they are caring for patients who are acutely ill. However, although first-year residents made the highest proportion of prescription errors (4.25 per 1000 orders) according to one study, even more experienced physicians have reported making serious medical mistakes.
Many mistakes happen in the inpatient or emergency department setting. The surgical specialties and intensive care units have also been identified as areas of high risk for patient safety. Severely ill patients require rapid assessment of a complex clinical picture as well as multiple procedures, evaluations, and decisions, thus affording many opportunities for mistakes to occur. One study found that patients in intensive care units had on average 178 activities performed on them, with 1.7 errors observed per day.
Patient characteristics can also increase the risk of mistakes. The risk of adverse events increases with increased age, severity of illness, length of hospital stay, and number of drugs prescribed. Older patients, for example, are likely to have advanced disease and comorbid conditions and are more likely to be taking numerous medications. These factors increase both the risk of errors and the likelihood that these errors will cause harm.
Serious medical mistakes also occur in office practice. In the United States, the main source of medical care for patients is in ambulatory settings. Due to the decentralized structure and diverse nature of events, medical errors can be common. Three top patient safety concerns are medication adverse events, diagnostic errors, and surgical errors. A study of malpractice claims showed that the former two errors are of concern for outpatient treatment and diagnosis of cancer. Some patients may be at increased risk. For example, patients older than 64 years of age are more vulnerable to medication errors. Some practicing physicians contend that the probability of making a serious error increases with more years in practice and particularly with greater pressures to increase productivity.