In Part 1 of his article, Prof. Berland covered why Incidental Findings (IFs) are an ongoing issue in radiology and what solutions have been put forward to fix the problem. In Part 2, we focus on arguably the most effective solution — revised clinical and communication pathways — and where technology can assist.
Let's explore some possible solutions to managing IFs
Crable, et al, recently reviewed the literature on this topic and divided improvement strategies into four categories:
- training and distributing quick reference guides
- guideline references placed online for easy access
- structured reporting with enhanced radiology templates for improved documentation, and
- revised clinical and communication pathways.
Of course, effectively detecting incidental findings, accurately applying guidelines, and properly placing the recommendations in the report using efficient, consistent language, are important to starting the chain of follow-up. However, these are larger issues of the radiologists’ process that are not the subject of this discussion.
The first and second solutions are relatively easy to perform and have been shown to be somewhat effective, albeit usually for a relatively short duration.
The third solution, structured reporting with detailed templates, with various types of enhancements, holds promise and has been implemented successfully in several institutions. However, the topic of structured reporting is very controversial and complex, and is a very difficult solution to scale up to a broad variety of institutions. Local and individual preferences, differences in patient populations, and differing IT environments often require customization. Structure must also be developed for numerous conditions. Accomplishing this, even if using templates prepared elsewhere, requires substantial human resources to develop consensus and then ongoing work to assure compliance.
Given the limitations of these other strategies, the fourth solution of revising clinical and communication pathways may provide the best combination of effectiveness vs. cost with a more modest expense in labor and money. There are an increasing number of potential approaches to communication being discussed in the literature with commercial and homegrown solutions employed in practice, with many of these using a combination of technical and human support.
What should a practice look for as it seeks a way to help improve IF management?
A crucial component to improving efficiency is adding some form of automation in one or more of the sub-processes of communication and monitoring. With the evolution of AI and Natural Language Understanding (NLU), the extraction of information from the radiology report (whether structured or not) can be automated with clinically effective accuracy, to help effectively document and generate follow-up communications.
In addition to simply extracting and organizing that information, what is a checklist of features to look for in a solution to “close the loop” of care with the help of AI and NLU?:
- The system should identify the nature and location of the finding, the specific follow-up action recommended and when it should be performed.
- To achieve the most consistent results, extraction should be in real-time at the point of interpretation or of care within the ordinary workflow, within or alongside routinely used applications.
- The system could perform an automated search of the EHR to determine if the examination has been performed within the recommended time window.
- A computer “dashboard” may be generated, including information to be pursued by clerks or radiologists’ assistants.
- The system could have a method to communicate to the referring physician that a recommended test has gone unordered or unperformed. This can be done according to the preference of the individual physician, and may include an EHR alert or personal communication. Any method to notify referring physicians must be respectful and designed to minimize alert fatigue.
- Because the Emergency Department has unique problems and needs, the system could include a filtered ED dashboard. The ED generally encounters patients once and are focused on urgent findings in a high-pressure environment that does not lend itself to considering the implications of incidental findings.
What are other elements of implementing and sustaining a good solution?
Altering long-practiced processes requires leadership and perseverance. It has been conclusively proven in many settings that once new processes are implemented, they will rarely be sustained on their own, unless they are automated and integrated into institutional workflow and culture, and continuously supported and monitored with human and technical resources. If you are not among the leaders of your practice, you can still be a champion, potentially bringing a proposed solution to your practice leaders and helping to deploy it.
Also, in seeking to implement a novel solution, don't let perfect be the enemy of good. With an average of under 50% of IF recommendations currently undergoing appropriate follow-up, there is ample room to improve. While you want to have a highly effective platform, it does not need to be 100%. An AI/NLP solution can provide important improvement with the promise of as system that can learn from its mistakes. How effective can these systems be? One early multi-stage tracking system increased compliance with RAI from 43% to 70%. The opportunity for further improvement beyond that is significant.
What are the benefits of acting to solve this problem?
First, your husband with the hypothetical incidental finding discovered on the CT scan mentioned at the outset, will have a system in place to assure that he will get the right exam at the right time. The consequences of our recommendations represent one of our most direct connections to patient care and the perception of radiologists by patients, enhancing our relationship with our patients and our position on medical teams. The system will see more appropriate use of resources and the risks of liability to radiologists, referring physicians and your institution will be greatly lessened. So, implementing an effective solution for IF follow-through may mean that we won’t have to tolerate the results of our historically erratic performance.
About the author: Dr. Berland is retired, and Professor Emeritus from the University of Alabama at Birmingham. He founded the ACR Incidental Findings Committee in 2006, which has published about 15 white papers on this topic. He currently consults for Agamon Health.