Interpretation differences between radiologists and diagnostic errors are significant issues in daily radiology practice. An awareness of errors and their underlying causes can potentially increase the diagnostic performance and reduce individual harm. The aim of this paper is to review both the classification of errors and the underlying biases. Case-based examples are presented and discussed for each type of error and bias to provide greater clarity and understanding.
Errors, discrepancies and confounding biases are inseparable parts of radiology practice with various clinical consequences.
Radiological errors can occur before, during or after the reporting periods.
Effective communication between radiologists, radiology technicians, patients and clinicians is the key to proper patient management
Being familiar with the types of errors and underlying biases is essential for radiologists to cope with them.
Radiological imaging is an essential part of patient management. Despite significant technological developments, a radiological investigation is rarely definitive on its own, leading to discrepancies between radiological impressions and the ultimate outcome. Moreover, radiology reports, like all human endeavours, may contain errors or misunderstandings. The term "error" is described as no uncertainty about the correct finding, with no possibility for dispute or disagreement, while the word "discrepancy" stands for justifiable differences of opinion between colleagues [1,2,3].
Errors and discrepancies may cause direct or indirect, permanent, or temporary harmful effects because of a false, missed, or delayed diagnosis, or may not result in any harm if clinically insignificant, or feedback is received from clinicians or other radiologists [4, 5].
In the literature, the classification of radiological errors and underlying biases has been discussed by different authors. Radiological errors can be classified according to the reporting process as pre-reporting, reporting or post-reporting errors. Pre-reporting errors consist of technical issues and procedure-related problems, whereas post-reporting errors are mainly caused by poor communication between radiologists and clinicians. Reporting errors are directly related to radiologists and can be categorized into two parts. "Perceptual errors" are more common and related to the fact that the present finding is not noticed, while "interpretative errors" are influenced by cognitive biases that can contribute to false reasoning. The classification of radiological error types according to the reporting process is shown in Table 1 [3, 4, 6].
According to the comprehensive classification system of Kim-Mansfield, there are 12 subgroups defined for radiological error types . In this paper, the classifications are reviewed, and cases are presented related to those 12 types of radiological errors and underlying bias.
Definitions of “error” and “discrepancy”
Diagnostic error is a condition that could harm the patient, with no acceptable cause and no scientific data for defense, approved by all experts in this field (Fig. 1). Discrepancy refers to a reasonable difference of opinion between radiologists about a finding or diagnosis. It differs from error because discrepancies can be justified based on a range of scientific data, such as clinical information, laboratory results or radiological patterns [3, 5] (Fig. 2).
Common radiological error types
Diagnostic errors constitute a large and complex issue that needs to be addressed, as they can prevent proper patient management, and a delayed diagnosis could lead to important consequences. Different classifications have been proposed at various times for the classification of diagnostic errors to facilitate their comprehensibility. The most broadly accepted classification was developed by Kim and Mansfield. According to this classification, diagnostic errors are examined in 12 groups based on the cause of the error [3, 4, 6, 7]:
False-positive or over-reading error In this scenario, an abnormality is noticed during the radiological examination. However, this finding is given more clinical value than it deserves and may consequently cause unnecessary diagnostic/therapeutic effort (Fig. 3) [6, 7].
Faulty-reasoning error In this type of error, detected abnormal radiological findings are thought to be associated with a false clinical entity, mostly due to cognitive biases such as hindsight bias or attribution bias (Fig. 4) [6,7,8].
Lack of knowledge error This error type occurs when a pathological finding is noticed but cannot be interpreted correctly due to lack of adequate knowledge or experience of the radiologist about the finding, despite the availability of adequate clinical information (Fig. 5) [3, 6,7,8].
Under-reading error This is the most common error type, in which an examination is reported as normal, although there is an undeniable and detectable abnormal finding. (Figs. 6, 7) [3, 6, 7].
Poor communication-related error In this scenario, the abnormal finding is recognized and accurately reported. However, the diagnostic message is not delivered to the clinician because of communication-related problems. In some circumstances, referrers may not be aware of the importance of the reported findings in the radiology report due to individual failings in communication between radiologist and clinician or systemic issues such as lack of multidisciplinary meetings and teamwork, increased workload and understaffing. Poor communication errors may also result from referrers not understanding radiology reports, lack of knowledge about the meaning of radiological findings or their own biases. Last but not least, typing errors in the report may disrupt the communication between the radiologist and the referrer, irrespective of the quality of the information provided by radiologists or the competence of referrers (Fig. 8, 9) (Table 1) [3, 6,7,8].
Technique-related error This error occurs because of low technical quality, inaccuracies during the image acquisition process or selection of the wrong technique or modality, resulting in a decreased possibility of detecting abnormal findings and sometimes makes diagnosis impossible (Fig. 9) [1, 6, 7]
Prior examination-related error The underlying cause of this 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 the likelihood of detecting pathological findings. However, when doing this, one must be careful about the ‘satisfaction of report’ error that will be explained later (Fig. 10) [6, 7].
History-related error This error includes inaccurate reporting faults when the radiologist is not provided with adequate or correct information about the patient’s clinical history (Fig. 11) [3, 6,7,8].
Location-related error Location-related error is characterized by the inability to recognize the pathological finding, which is seen within examination limits but falls outside the purposefully examined area, especially at the edges of the evaluated area (Fig. 12).
Satisfaction of search When the radiologist defines a pathological finding during the evaluation, other findings may be overlooked due to the satisfaction of the assessment and loss of motivation. Thus, accompanying findings may be under-read even if they are seen very clearly (Fig. 13) [6, 7].
Complication As a general description, complications are the conditions that occur during or after the procedures and are directly related to the nature of the procedure. They are unanticipated occurrences and may happen even under ideal conditions, making it controversial to define them as errors (Fig. 14) [6,7,8]. However, according to the Kim-Mansfield classification, the term "complication," as an error type, refers to an adverse event related to invasive radiological procedures (Fig. 15) .
Satisfaction of report This results from having undue confidence in the patient’s prior reports. As a result, if a wrong assessment has been made in the previous report, it will be repeated. This type of error is closely related to alliterative bias (Fig. 10) [6, 7].
Underlying bias types of radiological errors
Different biases are defined in the literature, which may affect the decision-making process when evaluating a radiological study. Biases may result in misinterpretation and different types of diagnostic errors, so awareness of certain types of biases can contribute to diagnostic accuracy. Biases can be classified as listed below [1, 3,4,5,6, 9]:
Attribution bias This is the tendency to attribute findings of a clinical condition by looking at specific characteristics of the patient; in other words, stereotyping (Fig. 8) [6, 7].
Alliterative bias (Satisfaction of report) Alliterative bias occurs when previous interpretations of a study influence the decision-making process of the current study (Fig. 10) [2, 4, 5].
Availability bias This type of bias is characterized by an increased tendency to make the decision under the influence of recently seen cases (Fig. 16) [1, 3,4,5,6, 9].
Regret bias This happens when radiologists worry about underdiagnosing a possibility and thereby over-report it due to fear of missing a diagnosis (Fig. 16) [7, 9].
Framing bias This bias type reflects the restriction of imaging assessment to the referral situation and clinical presentation framework (Fig. 17) [1, 4].
Premature closure This results from accepting an initial impression as a final diagnosis without any verification. Radiologists may be cognitively satisfied after discovering the first finding and end the search. Therefore, premature closure bias may overlap with the “satisfaction of search” error (Fig. 7) [3, 9].
Inattentional bias Inattentional bias, also named tunnel vision, may result in a diagnostic error due to an unusual appearance or location of findings (Fig. 6). This is also named the "Gorilla effect" after the famous study, in which a gorilla was inserted in a thorax CT by researchers and was overlooked by 83% of radiologists who were busy searching for lung nodules . Inattentinal bias may cause devastating complications, even in cases with obvious imaging findings [3, 5, 9].
Hindsight bias This type of bias is characterized by believing that it would be easy to reach a correct diagnosis by ignoring or de-emphasizing the difficulties in making the initial diagnosis and discounting the scenario under which the decision making occurred. When specialists evaluate the findings retrospectively, they may have access to contributory information not available to the original reporters, and they tend to underestimate the difficulty of making an accurate diagnosis due to hindsight bias. In contrast to other types of bias, hindsight bias is retrospective in nature (Figs. 18, 19) [3, 4, 8, 9].
Zebra retreat The radiologist retreats from an accurate but unusual diagnosis due to a lack of confidence, despite the presence of supportive evidence. In one respect, this bias is the opposite of the previously described Regret bias (Fig. 19) [3, 6].
Scout neglect The scout view is the preliminary image taken just before performing the imaging. Ignoring the scout view may result in a diagnostic error (Fig. 20) .
Anchoring bias This type of bias occurs when a radiologist becomes fixated on the first-sight diagnosis, although subsequently presented findings are incompatible with the first diagnosis. Anchoring bias is usually accompanied by confirmation bias, making it more dangerous (Fig. 4) [1, 3,4,5,6, 9].
Confirmation Bias Confirmation bias defines the active searching for more data to confirm the available hypothesis rather than seeking an alternative explanation (Fig. 4) [3, 6, 9].
The types of bias with a summary of the related thought processes of the radiologist are shown in Table 2.
Possible clinical consequences caused by a radiological error
Errors in the assessment of the imaging examinations of patients are classified by Brady et al. based on the "durations of effect" and "severity of clinical consequences." According to this classification, errors causing no harm are considered "negligible," and errors with minimal ill-effects are accepted as "minor error." "Moderate" and "major errors" are distinguished based on the duration of the negative effects. "Moderate errors" have short-term effects, while "major errors" are defined by long-term undesirable effects. Finally, errors resulting in severe long-term or fatal effects are classified as “extreme errors” . However, it should be kept in mind that this classification is only an example of a method intended to demonstrate how the effects of errors could be evaluated. It is neither a validated nor necessarily an accurate risk assessment tool, and it was not described as an absolute means of classifying the impact of radiological errors.
We can also classify errors according to their effects on patient management. Although some errors have a positive influence on patient management, these are rarely encountered and are the exception (Fig. 3). Apart from those, the most innocent errors are errors that are noticed before causing any effect on patient management. As expected, those errors cause no harm (Fig. 11).
Errors that affect patient management can be divided into two groups based on their impact on the patient. In the first group, the error reaches the patient but does not cause any harm. This error type may lead to unnecessary diagnostic effort and further examination (Fig. 21), or it may be clinically insignificant (Fig. 22). In the second group, the error reaches the patient and has a harmful effect. These errors may result in temporary or permanent clinical consequences, either directly or indirectly.
The consequences of harmful errors can be examined in three groups. The first group is related to misdiagnosis or delay in the diagnosis. Thus, the imaging findings, which are essential for the patient's clinical condition, could be missed or inadvertently interpreted as an inaccurate diagnosis (Fig. 23).
The second group is the errors causing a prolonged hospital stay, which leads to additional follow-up and sometimes the requirement for further treatment. This group of errors either results in acute injury (Fig. 15) or chronic/long-term unwanted situations (Fig. 7).
The third group is related to the overlooking or underestimating of significant and life-threatening findings. The latter can result in severe morbidity, even mortality (Fig. 24).
Errors, discrepancies and confounding biases are integral parts of the daily routine for radiologists and can cause various unexpected clinical consequences. In this study, certain types of radiological errors and biases are explained with case-based examples. By so doing, it is aimed to increase awareness about radiological diagnostic errors and related biases. However, it is obvious that radiological assessment of imaging examinations is a part of overall patient management that may be limited due to the diagnostic utility of the imaging technique and referral information. Therefore, radiological reports should not be expected always to be complete and correct or be regarded as the only tool to catch, confirm, or exclude the diagnosis [1, 3].
It should be kept in mind that effective communication between radiologists, radiology technicians, patients and clinicians is one of the key factors in reducing errors and thereby enabling proper patient management (Figs. 11, 21, 25). Finally, awareness of and familiarity with errors and underlying biases is essential for radiologists to be able to cope with them, avoid false interpretations and develop counter-measures [6, 11].
Availability of data and materials
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
Bruno MA, Walker EA, Abujudeh HH (2015) Understanding and confronting our mistakes: the epidemiology of error in radiology and strategies for error reduction. Radiographics 35(6):1668–1676. https://doi.org/10.1148/rg.2015150023
Go S, Furukawa T, Yamada K, Hiraoka T, Mochizuki S (2019) A case of supra-annular mitral valve replacement using chimney technique for severe mitral stenosis with extensive mitral annular calcification. Gen Thorac Cardiovasc Surg. https://doi.org/10.1007/s11748-019-01256-7
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Onder, O., Yarasir, Y., Azizova, A. et al. Errors, discrepancies and underlying bias in radiology with case examples: a pictorial review.
Insights Imaging12, 51 (2021). https://doi.org/10.1186/s13244-021-00986-8