Originally published December 2013
Reducing mistakes in reading and interpretation — the greatest medical-legal risk for radiologists — has become a major focus of health quality improvement initiatives for this specialty in recent years. However, the causes of radiologic discrepancies in diagnosis are complex and not always clear or easily addressed.
To determine areas of medical-legal risk for members, the CMPA analyzed 421 radiology cases that closed between 2007 and 2012. Of those cases in which care was criticized, the vast majority involved issues of misreading and misinterpretation.
The most common conditions associated with misreading or misinterpretation were cancer, mainly breast and lung, and fractures, with non-vertebral fracture greatly outnumbering vertebral. Plain X-ray, including mammography, was the most common modality followed distantly by CT, ultrasound, and MRI.
Imaging and interpretation
The radiologist's failure to request a repeat of poor quality films, or failure to order additional views to confirm or rule out a diagnosis, were recurring themes in cases involving missed diagnoses. In some cases, failing to recommend follow-up studies within a specific and appropriate time interval resulted in missing the window of opportunity to affect the course of the patient's condition. In addition, the person who was responsible for arranging subsequent studies was sometimes not made clear.
In a few cases, inadequate information from a referring physician contributed to a missed diagnosis. This was seen in cases involving unusual conditions or subtle findings. In other cases, experts were critical of radiologists who did not review previous studies for comparison.
Failure to repeat a poor quality film
An 80-year-old man undergoes a chest X-ray — postero-anterior (PA) and lateral views — at a clinic. The radiologist interpreting the films notes the PA view is overpenetrated (very dark) and uses a bright light to aid in viewing. She reports the lung fields are hyperinflated, suggestive of emphysema.
Eight months later the patient is diagnosed with lung cancer. A review of the earlier study reveals a 1 by 1.5 cm lesion in the left upper lobe. After learning of the missed finding, the radiologist adjusts her threshold for ordering repeat X-rays when studies are overpenetrated. A complaint to the medical regulatory authority (College) is dismissed after an investigation since the committee is satisfied with the radiologist's practice changes.
Diagnosis and the standard of care
Not all missed findings represent a breach of the standard of care. The Canadian Association of Radiologists has stated that radiologic discrepancy can occur in the best settings.1
Hindsight bias is recognized as a common pitfall in retrospective case analyses, especially when the reviewer is aware that an adverse outcome has occurred. That is, knowing an undesirable outcome has occurred increases the belief that it should have been evident and foreseen, and therefore preventable.
Furthermore, it has been argued that the availability of past scans and reports exposes radiologists to a level of scrutiny not experienced by other specialties. For example, it is common when reviewing past studies to be able to identify the early changes in architecture that reflect the development of a cancer. Knowing where to look, one may in hindsight identify these subtle changes which at the time were below the threshold of detection of a competent radiologist. Most often such retrospective findings reflect the limitations of the diagnostic imaging study and do not represent poor interpretation or poor clinical care. Therefore, it is especially important for physicians not to jump to conclusions that a poor outcome is the result of a diagnostic error.
Reporting and follow-up
Problems with communication were a common source of issues in the relatively small number of cases that did not involve misreading and misinterpretation. These cases often involved inadequate reporting or follow-up on the part of the radiologist or hospital.
Inadequate documentation, such as incomplete reports (e.g. no diagnostic impression or recommendation for follow-up imaging when appropriate), or insufficient detail on findings, was also identified as an issue in a number of cases.
Failure to adequately report on findings
A 35-year-old man arrives at the emergency department (ED) with a swollen, bleeding nose after being thrown from a horse. A cervical collar is applied immediately, and a CT scan of his facial bones and neck is done.
The radiologist, who is working off site, informs the ED physician that the patient might have a C1 cervical fracture, but that he is having difficulty accessing clear images from his location and is, therefore, unable to provide an opinion until the next day. Believing the C1 finding is related to an older injury, the ED physician removes the patient's collar and discharges him with a prescription for an analgesic.
The next day the radiologist posts a brief note to the hospital's PACS system indicating the patient has a comminuted nasal bone fracture and a right transverse process C1 fracture. The ED physician who treated the patient never receives the report, as she works at that hospital part time. A more detailed report is dictated and the same day the hospital mails the report to the patient's family doctor. The report is received 9 days later, and the patient is advised to go immediately to a tertiary care hospital.
The patient lodges a College complaint for the delay in diagnosing the fractures. The College counsels the radiologist that he should have informed the ED of the findings in a more direct manner in this case, for example by telephone or by other reliable means. Not doing so contributed to the patient's results "falling through the cracks" and not being acted on.
Because radiologists interact with a variety of professionals, sometimes across multiple settings, faulty systems and processes of care can contribute to adverse events.
In the cases reviewed, inadequate quality control systems led to mix-ups, most commonly interpretation of the wrong films. Similarly, vague communication protocols and inadequately aligned resources contributed to delays in the interpretation of films or the transmission of results.
Other systems issues involved transcription errors, such as referencing the wrong side of the body in a report or transcribing a report under the wrong patient's name. Other such mistakes involve confusing words that are similar, such as "renal" and "adrenal"; substituting one unit of measure for another; and confusing the word "now" for "no," or omitting the word "no" altogether.
Quality improvement in radiology
Quality improvement initiatives such as peer review programs are being implemented in organizations across the country. Peer review is a retrospective review by peers, or subject matters experts, of an individual or groups of individuals looking at specific indicators of quality of care. In The CAR Guide to Peer Review Systems, the Canadian Association of Radiologists acknowledges the benefits of such activities, but stresses that the purpose of a peer review is not to set parameters for acceptable levels of radiologic discrepancy, but to "improve overall standards by defining unperceived discrepancies and educational needs…"1
Risk management considerations
Radiologists should consider the following risk management suggestions, which are based on the experts' opinions in the cases reviewed:
- Consider the clinical information supplied, the provisional diagnosis, and whether contacting the referring physician for more information is appropriate.
- Confirm that the study being reviewed belongs to the patient listed on the requisition and that the date of the study is correct.
- If there is difficulty interpreting the study, consider if the image quality is sufficient. Consider repeating the test, or using alternative modalities, to obtain a satisfactory image.
- Perform a comprehensive evaluation of the images.
- Review any previous imaging for comparison when appropriate.
- Follow the standard protocols for imaging the anatomical structure that is being evaluated.
- Consider whether additional views are required to make a diagnosis.
- Present findings clearly, with sufficient detail, in the documentation. Important unanticipated findings should be clearly identified in the report.
- If additional studies are needed, document the recommendation and who should be responsible for arranging these.
- As stated by the Canadian Association of Radiologists: "In some circumstances, such co-ordination may require direct communication of unusual, unexpected, or urgent findings to the referring physician in advance of the formal written report."2
1. Canadian Association of Radiologists, "The CAR Guide to Peer Review Systems," 2012. Retrieved on June 13 2013 from: http://www.car.ca/uploads/standards%20guidelines/20120831_en_peer-review.pdf
2. Canadian Association of Radiologists. "The CAR Standard for Communication of Diagnostic Imaging Findings," 2010. Retrieved on June 13 2013 from: http://www.car.ca/uploads/standards%20guidelines/20101125_en_standard_communication_di_findings.pdf