Radiology Report Recommendations—
Issues and Solutions
Communication breakdowns play a significant role in mal- practice litigation against radiologists [ 1]. The principal mode of communication from radiologists to clinicians has
become the written radiology report, which may be faxed, mailed,
or otherwise transmitted to an ordering provider. Radiology information systems are starting to allow patients access to radiology
reports through so-called portals. These portals are relatively
new and their use is still being optimized. Unfortunately, while
systems for conveying radiology findings to clinicians are well
established, in some cases, the lines of communication can nevertheless break down and adversely affect care.
James Moses, chief quality and safety officer, Boston Medical
Center, in Massachusetts, and Alex Norbash, of the department of
radiology at the University of California, San Diego, were among
individuals who conducted a study to better understand how often
clinicians acknowledged the radiology report as informing patient
care. The study assessed the proportion of reports that contained
recommendations that required further actions and whether these
recommendations were acted on. The study was conducted in response to a request from our institution’s quality committee in order to provide data that would justify investment into result
communication support systems in radiology.
Our retrospective study evaluated 6851 routine communications in radiology reports issued during a 15-day period in
January 2014. These were only those reports for which we had
medical record access to search for follow-up actions. We found
978 different recommendations and, for each report that contained a recommendation, we reviewed the medical record to determine whether the recommendation had been acted on. The
recommendations for which we found no evidence that they were
acted upon were further classified into four groups:
1. The ordering physician acknowledged reading the report.
2. The ordering physician acknowledged reading the report
but the patient was a “no show.”
3. Further workup became unnecessary.
4. The ordering physician determined and documented that
the radiology recommendation was not clinically warranted.
Radiology Report Recommendations continues on p. 16
Department of Radiology and
Although the majority (67%; n = 655) of radiology recommendations were followed, we found that 33% (n = 323) were not followed. When considering the recommendations that were not
followed, we found three explanations: no show in 36% (n = 115),
recommendation became clinically unnecessary in 19% (n = 64),
and clinician documented a rationale for not following the recommendation in 5% (n = 18).
In the remaining 40% (n = 126) where recommendations were
not followed, we conducted further analysis to determine which
recommendations may adversely impact patient outcomes. We
found that 68% (n = 86/126) would likely not affect patient outcomes. Examples included “clinical correlation is recommended” or “may be better characterized with MRI.” Of the remaining
32% (n = 40/126) where recommendations were not followed, our
assessment showed that 50% (n = 20/40) were issued for important findings that did not necessarily contribute to significant
morbidity and therefore would be considered appropriate for routine communications. Examples included a small labral tear with
a 3.3-cm hemorrhagic cyst of the right ovary, twin pregnancy, or
7-mm cystic lesion in the head of the pancreas. In another 45%
(n = 18/40), radiology recommendations were issued because of
mass lesions raising concerns for cancer, and a lack of documentation for a closed loop verbal notification. Examples included an
incompletely characterized mass within segment 4 of the liver or
6-cm segment of irregular bowel wall thickening of the ascending colon. In 5% (n = 2/40) there was a fetal demise and the reports did not have documentation of a verbal notification.
Our study had several limitations, including our inability to
assess lack of follow-up actions for referred patients whose medical records we could not access. It is also possible that reports
with concerning findings that we studied were verbally communicated with clinicians, although the communication itself may
not have been documented in the report, potentially leading to
our overestimating the number of concerning findings lacking
documented clinician deliberation.
After we shared our findings with the quality committee at
our institution, we received approval for a full-time reading room
assistant and software that is capable of tracking result notifications. This study provided data documenting concrete patient
safety risks that were very powerful in helping our administration’s decision to allocate funds toward our request.
With regard to follow-up in our patient population, the number of patients who were “no shows” consisted of 8% (n = 9/115)
of patients who died in the interval, and the remaining 92% (n =
106) were lost to follow-up. This high rate of follow-up loss is best
explained by the fact that our academic institution is a regional
safety-net hospital. Many of our patients are homeless, uninsured, unable to access a computer, unable to read, or not proficient in English and unable to afford follow-up care or may not