Precedent (Australian Lawyers Alliance)
RADIOLOGICAL ERRORS: NOT ALL BLACK AND WHITE, MORE LIKE 50 SHADES OF GREY
By Stephen Littlefair
Radiological interpretation is not an exact science. Approximately 4 per cent of radiological interpretations contain errors but, fortunately, most of these errors are clinically insignificant. Where serious errors are found, most are promptly corrected, causing no harm to patients. Diagnostic errors are defined as diagnoses that are missed, delayed, or incorrect. The latter error comprises not only false negative decisions but also over-diagnosis.
Radiologists offer a plausible opinion on medical images, but neither a definitive histological nor a microbiological diagnosis. Moreover, as imaging technology advances, each new modality or procedure potentially exposes radiologists to new grounds for legal action against them.
‘Today, for a plaintiff to succeed in an action in negligence, he or she must establish (on the balance of probabilities)’:
1. That a duty of care was owed by the defendant;
2. That the defendant fell below the required standard of care;
3. That the breach of duty caused or materially contributed to the damage suffered - be it physical, mental or economic loss; and
4. That the loss or damage suffered was reasonably foreseeable.
The possibility of a radiologist being the defendant in at least one lawsuit is 50 per cent by the age of 60. Between 1985 and 2002, radiologists ranked sixth amongst specialists in the number of claims for which they were defendants, despite being only 3.6 per cent of all medical specialists.
It is estimated that 40 per cent of radiologists in the US are taken to court approximately once every five years. Tort claims for negligent diagnostic errors result in billions of dollars in compensation annually.
Radiologists can face litigation for error and misdiagnosis; irradiating a pregnant patient; adverse reaction to contrast media; interventional procedural complications; and improper consent. However, this article focuses primarily on misdiagnosis of conventional radiographs.
The diagnostic imaging process involves two distinctive parts. In breast imaging (mammography), conventional radiography and ultrasound scanning (USS), there is the technical function of image acquisition by the radiographer followed by the diagnostic task undertaken by the radiologist. A poor quality image impedes adequate reading, which may lead to misdiagnosis.
For fluoroscopic procedures such as barium enemas, cardiac imaging and the examination of blood supply (angiography), the radiologist is involved in acquiring the individual images that comprise the investigation in addition to radio-diagnosis.
However, with Computerised Tomography (CT) and Magnetic Resonance Imaging (MRI), the radiologist reads the images but does not acquire the study, apart from instructing the radiographer in the type of imaging protocol required.
Accuracy is essential, as important errors in interpreting medical images may have significant healthcare and legal consequences. Laypersons are inclined to believe that radio-diagnosis is a binary decision and that normality or abnormality can readily be established. In the early 1970s, imaging fundamentally consisted of plain film radiographs, but advances in technology have led to sophisticated tomographic technology such as CT, MRI, and USS as well as nuclear medicine and its associated imaging techniques. In addition, soft copy reporting (from computer screens as opposed to hard copy: a film) has become the norm. The interpretation of images produced by all these imaging modalities is the responsibility of the radiologist.
For example, the conventional chest radiograph is a two-dimensional projection of a three-dimensional structure. Each anatomical component (the pulmonary blood vessels, the skeleton and areas of the mediastinum) can therefore project over the lung and partly or completely mask pulmonary lesions.
Outside the profession, the thought processes in radiological diagnosis are not well understood, and little is known about how radiologists acquire their expertise.
Research into radiological error has consistently demonstrated differences in radiological performance, and studies of radiological error report significant rates of both intra- and inter-observer variability, with radiologists making both false positive and false negative decisions. Radiologists reading the same radiograph at different times disagree with each other (and with themselves) in up to 20 per cent of cases.
Why do radiologists make mistakes? There are a number of reasons why radiologists are sued: observer error (search, perception and cognition); failure to suggest further imaging; and failure to report any results in a timely manner.
Almost 40 years ago, Kundel et al described the three types of observer error. ‘Search error’ is due to a failure of a radiologist to fixate in the area of the lesion. ‘Recognition error’ involves fixating on the lesion but failing to detect the lesion. The ‘cognitive’ or decision-making error is either the misdiagnosis of an abnormal lesion as a normal structure (false negative) or reporting a normal lesion as abnormal (false positive). This error is the most common.
These observer errors are also related to several other factors, including the level of observer alertness, fatigue, the duration of the observation task, interruptions, index of suspicion (expectation), the presence of other abnormalities (satisfaction of search). ‘Satisfaction of search’ refers to ending the search for abnormalities once an abnormality has been found. It may occur when the referrer asks for assessment of a single diagnosis, instead of asking for an explanation of the symptoms.
Other influential factors include the viewing environment, the level of attention of the radiologist and any other distracting factors.
Error in diagnosis is, by far, the most common reason for malpractice suits against radiologists. These errors are threefold: perceptual, cognitive or delayed.
Breast cancer is the most common misdiagnosis, followed by non-spinal fractures, spinal fractures and chest pathology (of which almost 50 per cent is the failure to identify a lung cancer).
Much research has determined that even the most experienced radiologist often fails to recognise a lung nodule, which can be benign or malignant. A number of researchers have described the considerable difficulties of interpretation created by overlapping structures and the small size and subtlety of many lesions.
Diagnosis depends entirely on visual perception and on the identification of abnormal features on the image. Unfortunately, some conditions can have similar appearances (Table 1), while some ‘abnormal’ appearances are in fact normal variants particular to that patient as comprehensively shown in the definitive text by Keats.
Apart from misdiagnosis, the failure to pass on results also features in medical malpractice cases. Records from medical insurance companies indicate that the second most common cause of litigation is the failure to communicate the results of a radiological examination. Such communication failure may be a factor in 80 per cent of cases. Although a radiological report is generated by the imaging facility and should be sufficient, in some instances this report may not have been received or read by the treating doctor. Hence, the radiologist becomes partially culpable in any lack of intervention.
Research by Berlin also stresses the medico-legal risk of a vague radiology report. Berlin reviewed cases that involved a delayed diagnosis of malignancy that led to litigation. The reports were not inaccurate, but each of the referring physicians testified that had the radiologists emphasised their opinion more weightily, they would have acted much more readily.
Further factors responsible for misdiagnosis stem from external influences generally beyond the control of the radiologist, such as poor quality images, misleading clinical history, the presence or absence of previous images, and previously incorrect radiological reports. Any of these can serve as the basis for malpractice litigation.
While poor quality images should not be presented to a radiologist for reporting, sometimes it is impossible to avoid such an event. Sub-optimal images are often caused by patient non-compliance, rather than by poor technique by the radiographer.
Patient non-compliance may be due to misunderstanding of instructions, or the influence of drugs or alcohol. In addition, some patients are non-compliant due to their infirmity, making the subsequent image very difficult to interpret. In cases where the patient is unwell, mobile chest radiograph is often obtained, which reveals a number of tubes, lines, and monitoring devices as well as a multitude of cardio-pulmonary problems demonstrated by many different appearances on the resultant image. Adding to the diagnostic problem is the fact that a mobile image is often of inferior quality to that of a radiograph obtained in the radiology department.
Poor or incomplete information on radiology requests can have a significant effect on the radio-diagnostic process. The referrer should provide the radiologist with accurate and relevant clinical information. This allows the imaging departments to use the most appropriate imaging technique, avoid undesirable events and assist in radio-diagnosis. Referrals for imaging studies are often illegible, and relevant clinical information may be absent or misleading, leading to diagnostic error. For instance, if a radiologist believes s/he is interpreting the cervical spine radiographs of a patient with longstanding neck pain, s/he may overlook subtle evidence of acute neck trauma.
A further source of error results from the influence of a previous radiology report over another radiologist, especially if the previous radiologist is more experienced. This type of error can occur because the radiologist reads the old report before looking at the image.
Medical decision-making is prone to numerous biases. Such biases can be a factor in causing avoidable errors in diagnostic reasoning and hence subsequent litigation.
In ‘expectation bias’, a radiologist sees what s/he expects to see. For example, Jager and colleagues noted a case where a patient underwent a mammography with the request: ‘Small hard mass with some retraction of the skin: cyst or fibroadenoma?’ The senior radiology resident reported ‘a small mass with irregular margins with some spiculation to the overlying skin, which shows some thickening. Additional breast ultrasound revealed a solid mass. Conclusion: probably benign fibroadenoma.’ The final diagnosis was a malignant invasive ductal carcinoma. This patient was a healthy 23-year-old female, obviously young and healthy. The surgeon and the radiologist consequently, and wrongly, expected a benign lesion.
‘Hindsight bias’ is particularly relevant in radiological malpractice cases in the context of the expert witness who is reviewing a contentious image in a case of perceptual error. It refers to the tendency to overestimate the probability of a particular diagnosis when, in retrospect, the correct diagnosis is already known.
The failure to identify an abnormality on a chest radiograph subsequently found to be a malignancy has potentially very serious consequences in medical malpractice litigation. Accurate image-reporting remains the goal in the operation of an imaging department, but a number of factors are likely to affect the radiologist’s interpretation of a radiograph. Radiologists are often unaware of the number of biases and factors that influence their decision-making. An error is an unexpected mistake, irrespective of whether any harm ensues, and the results of radiological errors can impact upon the patient and their families, as well as members of the imaging staff involved in the case. Further consequences may negatively affect the practice, the institution and the profession itself.
Mistakes are inevitable today and will continue to occur. While various approaches have been recommended to reduce the rate of missed lung malignancies, for example, each of these exposes problems related to workflow and the limitations of human perception.
Radiologists who commit errors are not necessarily less qualified, more careless, or less well-taught than those who do not. Many believe that if a radiologist has missed a finding that is identified retrospectively on a radiograph, the miss must be negligent.
‘The answer to this question still eludes us; the distinction between radiologic errors and malpractice remains as blurry as ever.’
Stephen Littlefair is an associate lecturer at the University of Sydney. PHONE (02) 9036 7355 EMAIL firstname.lastname@example.org.
 L Berlin, ‘Radiologic errors and malpractice: A blurry distinction’, (2007) 189, American Journal of Roentgenology, pp 517-22.
 P MacFarlane, Health Law – Commentary and Materials (2000), Federation Press, Sydney, p86.
 SR Baker, JS Whang, L Luk, KS Clarkin, A 3rd Castro and R Patel, ‘The demography of medical malpractice suits against radiologists’, Radiology (2013) 266(2), 539-47.
, Physician Insurers Association of America, Claim Trend Analysis Study, Rockville MD, 2004, pp30-31 and pp97-149.
 A Fileni and N Magnavita, ‘A 12-year follow-up study of malpractice claims against radiologists in Italy’, Radiologia Medica (2006) 111(7), pp1009-22.
 AS Saber-Tehrani, HW Lee, SC Matthews et al, ‘20-year summary of US malpractice claims for diagnostic errors from 1985 – 2005’, (abstract), (2011) Proceedings of the Fourth Annual Diagnostic Error in Medicine Conference, Chicago, Illinois, Johns Hopkins University School of Medicine.
 JG Elmore, CK Wells, CH Lee, DH Howard and AR Feinstein, ‘Variability in radiologists’ interpretation of mammograms’, New England Journal of Medicine (1994) 331, pp1493-99.
 RL Eisenberg (Ed), Radiology and the Law: Malpractice and other issues, Springer, 2004, p86.
 HL Kundel, CF Nodine, D Carmody, ‘Visual scanning, pattern recognition and decision-making in pulmonary nodule detection’, Investigative Radiology (1978) 13, pp175-81.
 EA Krupinski, KS Berbaum, R Caldwell, KM Schartz, ‘Is diagnostic accuracy for detecting pulmonary nodules in chest CT reduced after a long day of reading?’, in Craig K Abbey, Claudia R Mello-Thoms (eds), Medical Imaging 2012: Image Perception, Observer Performance and Technology Assessment Proc. Of SPIE Vol 8318, Available from http://proceedings.spiedigitallibrary.org/.
 AG Pitman, ‘Perceptual error and the culture of open disclosure in Australian radiology’, Australian Radiology, (2006) 50, pp206-11.
 JS Whang, SR Baker, R Patel, L Luk, A 3rd Castro, ‘The causes of medical malpractice suits against radiologists in the United States’, Radiology (2013) 266 (2), pp548-54.
 R Fitzgerald, ‘Error in radiology’, Clinical Radiology (2001) 56, pp938-46.
 PK Shah, JH Austin, CS White et al, ‘Missed non-small cell lung cancer: Radiographic findings of potentially resectable lesions evident only in retrospect’, Radiology (2003) 226, pp235-41; PM Turkington, N Kenna, MA Greenstone, ‘Misinterpretation of the chest X-ray as a factor in the delayed diagnosis of lung cancer’, Postgraduate Medical Journal (2002) 78, pp158-60.
 TE Keats, MW Anderson, Atlas of Normal Roentgen Variants that May Simulate Disease (9th ed), 2012 Elsevier.
 L Berlin, ‘Failure of radiologic communication: An increasing cause of malpractice litigation and harm to patients’, Applied Radiology (2010) 39, pp17-23.
 W Levinson, ‘Physician-patient communication: A key to malpractice prevention’, Journal of the American Medical Association (1994) 272, p1619e20.
 L Berlin, ‘Pitfalls of the vague radiology report’, American Journal of Roentgenology (2000) 174, p1511e8.
 K Stavem, T Foss, O Botnmark, OK Andersen, J Erikssen, ‘Inter-observer agreement in audit of quality of radiology requests and reports’, Clinical Radiology (2004) 59 (11), pp1018-24.
 DM Schuster and ME Gale, ‘The malady of incomplete, inadequate, and inaccurate radiology requisition histories: A computerized treatment’, American Journal of Roentgenology (1996) 167, pp855-9.
 P Croskerry, ‘Achieving quality in clinical decision-making: Cognitive strategies and detection of bias’, Academic Emergency Medicine (2008) 9(11), pp1184-1204.
 G Jager, JJ Futterer, M Rutten, S Hertogenbosch, ‘Thinking fast and slow’, poster C-0899, ECR2014, Available from: http://posterng.netkey.at/esr/viewing/index.php?module=viewing_poster&doi=10.1594/ecr2014/C-0899.
 A speculated mass is a lump of tissue with spikes or lumps on the surface. It is suggestive but not diagnostic of malignancy.
 RB Gunderman, ‘Biases in radiologic reasoning’, American Journal of Roentgenology (2009) 192(3), pp561-4.
 Berlin, see note 1 above.