1. Definitions and differences of “error”, “discrepancy” and “complication”.
Diagnostic error is the condition that could inflict harm on patient, with no acceptable cause and no scientific data for defense, approved by all experts in this field (Figure-1).
The discrepancy means reasonable differences of opinions between radiologists about finding or diagnosis. It differs from error because discrepancies can be justified based on a range of scientific data, such as similar clinical, laboratory, or radiological patterns (Figure-2).
Complications are the conditions that occur during or after radiological procedures and are directly related to nature of procedure. Complications are unanticipated occurrences and they may happen even under ideal conditions (Figure-3).
2. Examples of common radiological error types.
Diagnostic errors are a large and complex issue that needs to be addressed, which prevents proper patient management, and sometimes could lead to vital consequences due to delay in diagnosis. Different classifications have been proposed at different times to classify diagnostic errors and thereby facilitate their intelligibility. The most broadly accepted classification is developed by Kim and Mansfield. According to this classification, diagnostic errors are examined in 12 groups based on the cause of error.
1. False-positive error or overreading: In this scenario, abnormality is noticed during radiological examination, however, this finding is given too much clinical value than it deserves. Consequently, it may cause unnecessary diagnostic/therapeutic effort (Figures-4,6).
2. Faulty-reasoning: In this type of error, detected abnormal radiological findings are thought to be associated with wrong clinical entity mostly due to cognitive biases such as hindsight or attributable bias (Figures-1,5,17,18).
3. Lack of knowledge: It can be mentioned when a pathological finding is noticed, but could not be interpreted correctly due to lack of adequate information or experience (Figures-6,23).
4. Underreading: It is the most common error type. In this type of error, an examination is reported as normal although there is undeniable detectable abnormal finding (Figures-1,19).
5. Poor communication: In this scenario, abnormal finding is recognized and accurately reported. However, diagnostic messages cannot be reached to clinician because of communication-related problems (Figure-7).
6. Technique related: This error occurs because of choosing the wrong technique or low technical quality which decreases the possibility of detecting abnormal findings , and sometimes, even, makes diagnosis impossible (Figures-8,25).
7. Prior examination: The main reason for this type of error is skipping the "comparison with previous exams" step which is indispensable during radiological evaluation. Each evaluation must be compared with previous examinations to increase likelihood of detecting pathological findings. But when doing this, one must be careful about ‘satisfaction of report’ error that will be explained later (Figure-9).
8. History: This error includes inaccurate reporting faults which occur when the responsible radiologist is not provided with adequate or correct information about the patient’s clinical history (Figure-10).
9. Location: Location related error defines the inability to recognize the pathological finding which is seen within examination limits (especially in the edges of evaluated area) but falls outside the purposefully examined area (Figures-11,27).
10. Satisfaction of search: When the radiologist detects a pathological finding during the evaluation, (s)he subconsciously lost motivation for seeking other findings. Thus, accompanying findings may be underread even if they are seen very clearly (Figures-12,13,14).
11. Complication: It describes undesirable situations that occur during radiological procedures and problems/misdiagnosis due to these complications. They may be caused by either incorrect selection of the patient or of the radiologic procedure (Figure-15).
12. Satisfaction of report: It results from having undue confidence in the patient’s prior reports. As a result, if there is wrong assessment made in previous report, it will be repeated. There is a close relationship between this type of error and alliterative bias (Figure-9).
3. Common bias types underlying radiological errors.
Biases may result in misinterpretation and diagnostic errors, so being aware of certain types of biases can contribute to diagnostic accuracy. Types of biases can be classified as listed below;
1. Attribution bias: It defines the tendency to attribute findings of a clinical condition by looking at certain characteristics of the patient, in other words, stereotypes (Figure-7).
2. Alliterative bias (satisfaction of report): Alliterative bias occurs when previous interpretations about a study influence the decision-making process of the current study (Figure-9).
3. Availability bias: This type of bias is characterized by an increased tendency to diagnose recently seen or catchy entities (Figure-16).
4. Regret bias: It happens when radiologist worries about underdiagnosing a possibility so brings it forward (Figure-16).
5. Framing bias: This bias type reflects automatically restricting the differential diagnosis due to the referral situation and clinical framework (Figures-14,17,20).
6. Premature closure: It results from accepting an initial diagnosis as final diagnose without any verification. This is the most common type of bias in clinical medicine (Figure-18).
7. Inattentional bias: Inattentional bias, also named as tunnel vision, may result in diagnostic error due to unusual appearance or location of findings (Figure-19).
8. Hindsight bias: It defines when a specialist evaluates the findings retrospectively, they tend to underestimate the difficulty of making an accurate diagnosis. By contrast with other types of bias, hindsight bias is retrospective in nature (Figures-20,21).
9.Zebra retreat: The radiologist retreats an accurate but unusual diagnosis due to the lack of confidence, despite the presence of supportive evidence (Figure-21).
10. Scout neglect: Scout view is a preliminary image that is taken just before performing the imaging. Ignoring the scout view may result in a diagnostic error (Figure-22).
11. Anchoring bias: This type of bias occurs when radiologist fixes first-sight diagnosis, although subsequently presented findings are incompatible with the first diagnosis. Anchoring bias usually accompanied by confirmation bias, making it more dangerous (Figure-23).
12. Confirmation bias: Confirmation bias defines actively searching for more data to confirm the available hypothesis rather than seeking for the alternative hypothesis (Figure-23).
4. Possible clinical consequences caused by radiological error.
Errors in diagnosis could lead to some clinical consequences. We can also classify errors based on these clinical consequences.
The simplest and most innocent of those are errors that are noticed before affecting the patient. In this condition, the error causes no harm.
We can divide errors that reach patient into two types based on concerning the patient.
In the first type, the error reaches the patient but does not cause any harm. This error type may be clinically insignificant or may cause unnecessary further examination (Figure-24).
The second type is errors that harm the patient. These can lead either directly/indirectly, temporary/permanent clinical consequences. The consequences of harmful errors can be examined in 3 groups. The first group is related to delay in diagnosis. Thus, imaging findings that are important for patient may be missed or inadvertently interpreted as an irrelevant diagnosis (Figures-25,26). The second group is errors causing prolonged hospital-stay which leads to additional follow-up and sometimes additional treatment (Figure-15). The third group is related to the overlooking or underestimating of vital and life-threatening findings. The last one can result in serious morbidity, even mortality (Figure-27).