Diagnostic errors are extremely common; one in every ten diagnoses is probably wrong [8]. Fortunately, the vast majority of diagnostic errors are inconsequential; the original problem resolves, the error is caught in time, the patient is resilient, or the treatment that was provided worked anyway. For some fraction of patients, however, the error results in harm and death. One estimate places the annual toll of diagnostic error in the USA at 40,000–80,000 deaths per year [9]. When timely surgical intervention is critical, misdiagnosing conditions such as spinal cord compression, necrotizing fasciitis, acute myocardial infarction, among many others, is lethal. The decision whether to operate on patients in whom these diagnoses are being considered is also a diagnostic decision, and the pressure and angst inherent in these situations is substantial and undeniable [10].

Data compiled from malpractice claims have clarified the relative incidence of surgical errors and what conditions are most commonly encountered. Diagnostic errors are the number one or two categories of claims in all of these studies. More than half of the diagnostic errors originate in ambulatory settings. In one recent study of 2531 cases of diagnostic error in ambulatory settings, 17 % were surgery related, with orthopedics, urology, and general surgery being the leading categories [11]. Most of these cases involved patients with cancer, cardiovascular conditions, and various injuries, especially orthopedic injuries. In another study, of 7438 closed claims from 2007 to 2013, 1877 were attributed to diagnostic error [12]. Of the 3963 claims involving surgeons, 524 were related to issues in diagnosis. The top five claims in each specialty are noted in Table 25.2.

While data from filed claims can help determine the relative distribution of surgery-related cases, the true incidence of diagnostic error in surgery is not known because the number of cases with good outcomes is large and not precisely known. A similar situation is encountered in internal medicine and emergency medicine—the actual incidence of diagnostic error is not known, and at the present time, no organizations report rates of diagnostic errors [13, 14]. This reflects primarily the difficulty in identifying diagnostic errors, but also the challenges physicians encounter in coming to an agreement on what comprises an error, as opposed to the normal evolution of a diagnosis over time, or the normal variability from one physician to another in diagnostic evaluation. Although healthcare has focused on patient safety for almost two decades, diagnostic errors have received relatively little attention. This has reflected cultural attitudes discouraging discussion of misdiagnosis, the challenges in finding and defining these errors, assumptions about the impracticality of potential process or outcome measures of diagnostic quality, and the belief that diagnostic errors are less amenable than other types of medical errors to systems-level solutions [15, 16].

Although real-time data is lacking, there is a wealth of research data that suggests diagnostic errors are quite common [8]. In internal medicine, one in every ten diagnoses is believed to be wrong, based primarily on studies involving standardized patients with classic presentations in real-world settings. A recent study using chart reviews found that one in every 20 ambulatory care patients will experience a diagnostic error every year [17]. Autopsy studies consistently show major discrepancy rates (discrepancies with a high likelihood to have changed management and treatment) in the range of 10–30 % [18, 19]. In programs providing second opinions, the likely diagnosis changes for one in seven patients, and treatment recommendations change for one in three patients [20]. The surgical specialties differ greatly, however, in how often the second opinion differs from the first. In terms of treatment recommendations, changes are less frequent in surgical oncology (19 %) and urology (28 %), and are more frequent in neurosurgery (42 %) and obstetrics (42 %) [20]. Finally, retrospective collections of cases are available for many conditions, and invariably report either bad or shocking statistics on diagnostic accuracy and timeliness. Several examples relevant to surgical care are listed in Table 25.3.

Diagnostic errors in surgical patients evolve from the same set of cognitive- and system-related factors as in other clinical settings. A very small fraction of errors derives from patient-related factors, for example patients with Munchausen’s syndrome who feign symptoms [35], or patients who choose not to undergo diagnostic tests that were recommended or attend follow-up appointments. Most errors, however, reflect shortcomings of the clinician’s cognitive processes, in the face of one or more breakdowns in the systems of care [2].

Cognitive errors involve one of three problems:

There is no data on the relative frequency of these error types in surgery, but in internal medicine, the vast majority of cognitive errors are in the third category, which entails synthesizing the available information [2]. The current paradigm of clinical reasoning involves the use of two very different cognitive pathways [36] (Fig. 25.2). Except for early trainees , most new problems are recognized immediately, and using a subconscious, intuitive pathway, the diagnosis is evident within milliseconds. If the problem is not recognized, we resort to deliberate, rational consideration of the situation, a process that takes longer and involves cognitive “work.” Humans and probably all animals have evolved to take advantage of the intuitive pathway whenever we can, and indeed almost all everyday actions and thoughts derive from this system. In practice, both systems may come into play in diagnosing a new patient problem, and in theory, the rational system has the opportunity and responsibility to be constantly monitoring intuitive processing. If some discrepancy is noted or something just “doesn’t fit,” the rational pathway takes over and we sense the need to slow down, or look for additional data or input to affirm our hunches or heuristics. If there are no such flags, we assume our assessment is correct and proceed. Unfortunately, the “feeling of right” in these situations is exactly the same whether our diagnosis is correct or not, until that unpleasant point that we realize that the diagnosis may be wrong [37]. Physicians, like all decision-makers, are generally not accurate in predicting which of our diagnoses are correct or not, a problem of calibration [38–40].

Both systems are error prone but for different reasons. The rational pathway for understanding a clinical dilemma in surgery can be degraded by insufficient knowledge or experience, or by flaws in logical thinking, or reasoning. Intuitive decision- making can be degraded by a large range of innate cognitive “biases,” of which over 150 have been described, (See Wikepedia’s ever expanding “List of Cognitive Bias”) and commonly encountered examples are shown in Table 25.4 [41]. The IOM report emphasized the importance of the environment and the work system in determining the quality and outcome of diagnosis and clinical decisions . The local culture of safety is critical, along with human factors that can influence the immediate situation, such as stress, distractions, fatigue, and team support.

Surgeons face a number of unique cognitive challenges that may predispose to diagnostic errors. First, patients undergoing surgery have typically been seen by a number of physicians leading up to the surgical event, creating the unavoidable assumptions that all of the requisite diagnostic thinking has already been completed, and that the diagnostic conclusions can be trusted (see Box 25.2).

Conversely, patients presenting with conditions that are considered primarily surgical, such as patients with an “acute abdomen,” may not be seen by other internists or emergency medicine staff, thus losing the opportunity to be assessed from a different perspective. Secondly, the large amount of task-oriented activities surrounding preoperative preparation and the mental rehearsal a surgeon or anesthesiologist must go through may not leave sufficient cognitive capacity to avoid diagnostic errors or prevent biases [42]. It is generally acknowledged that cognitive overload directs cognition away from the rational, deliberate pathway and toward the more error-prone intuitive approach. Finally, surgeons require a high level of confidence to lead a team through high-risk operations, raising the question of whether this may sometimes engender overconfidence and a tendency to disregard other opinions or novel information. This requires training surgeons on becoming team leaders and being aware of how these factors can shape their actions and the actions and outcomes of others [43].

System-related errors that contribute to diagnostic error include breakdowns in communication or coordinating care, reliably transmitting test results and consults, erroneous laboratory or imaging interpretations, difficulties associated with using an electronic health record, supervising trainees, and a host of other issues [2]. Cognitive load contributions are identified as the lead cause for most cases of diagnostic error [2].

As the old saying goes, the first step in addressing any problem is recognizing that you have one. This is particularly relevant in the case of diagnostic error, because physicians generally believe that they are practicing at a very high level, and tend to attribute diagnostic errors to other clinicians who aren’t as experienced or careful. All humans and, perhaps, physicians in particular are overconfident in their abilities and in the correctness of their clinical decisions [44] The cognitive errors made in clinical diagnoses are the same errors people make in their everyday lives; we jump to conclusions, we trust information given to us without verifying it, we accept an assigned diagnosis without rethinking it, and our emotions get in the way of good judgment. All physicians can improve the quality of their practice by accepting the universal predisposition to error, understanding the causes of diagnostic error, and addressing these problems transparently.

Optimal surgical diagnosis and care relies heavily on accurate cyto-pathological diagnosis, and errors deriving from the interpretation of cytology, biopsy, or surgical specimens are important to recognize and address. Errors may arise at any point in the “total testing cycle” [45], which begins with specimen acquisition, labeling, and delivery to the laboratory, where the specimen is prepared for the analytic phase. The post-analytic phase begins with a report generation and ends with delivery of the report to the clinician, and the clinician acting on it appropriately. Unlike the other phases of the total testing cycle, the analytic phase is substantially different in surgical pathology and cytology (vs. clinical pathology and automated lab testing) in that it involves visual interpretation and the judgment of the pathologist to arrive at the correct interpretation [33, 46, 47]. Compared to automated lab testing, which operates at error rates in the range of 0.01–0.001 %, the error rate in surgical pathology is orders of magnitude higher, in the range of 2–5 % [34, 48, 49]. Errors in the post-analytical phases of testing are also especially common [50, 51], and many involve failures to reliably communicate test results [52], as illustrated in the case vignette above.

There are many factors that contribute to an accurate interpretive diagnosis, including: (1) the pathologist’s knowledge and experience, (2) clinical correlation, (3) standardized diagnostic criteria and taxonomy, (4) confirmatory ancillary studies when available, and (5) secondary review of cases.

Studies have shown the additive value of clinical correlation , standardization of diagnostic criteria, and taxonomy and confirmatory ancillary testing to the accuracy of surgical pathology and cytology diagnoses [53–55]. Several of these factors contribute to establishing a precise diagnosis, but the pathologist’s knowledge and experience remain the essential factors in interpretive diagnosis such as in neuropathology tissue ambiguity. Although numerous studies have shown that second opinions help detect interpretive diagnostic errors [56], there have been only scattered efforts to formalize and adopt this practice as a clinical standard. Targeted case reviews could be an integral component of a quality assurance plan that is aimed proactively at preventing errors before they have a potential adverse impact on patients. The College of American Pathologist has issued a recent guideline on the use of second opinions in surgical pathology [49], (see Table 25.5) and a much more detailed and specific set of guidance on second opinions in cancer diagnosis is available in the 11-part series from Cancer Care Ontario [57].