In an article published in the online Journal of the American College of Radiology1, authors from Duke University Medical Center Department of Radiology present a study conducted to demonstrate the variability and complexity of radiologists’ dictated notes. The authors chose to analyze the language used to describe normal thyroid glands in chest CT reports as a “surrogate for the broader readability of radiology reports”. In a sample of nearly seven thousand non-contrast chest CT reports, the researchers found 342 unique sentences or phrases describing a normal thyroid gland. Furthermore, linguistic analysis suggested that descriptors for a normal thyroid gland require an advanced college-level education for comprehension. This text is well above the national average health literacy level and results in reports that are difficult for patients to understand.2
Establishing clarity and structure in radiologic documentation has been a topic of academic discussion for decades.3 Referring physicians rely on unambiguous communication of imaging findings for accurate interpretation. As the authors note, the growing adoption of online portals for sharing test results directly with patients has increased the demand for simplicity and readability. Likewise, recent developments in natural language processing (NLP) applications used for the interpretation of findings create new imperatives for improved uniformity of radiology reports.
Gregory Nicola, MD, serves as executive leadership of Hackensack Radiology Group and is also a thought leader in the economics of artificial intelligence (AI) in radiology. In this article from Health Data Management4, he states that, “Machine learning algorithms have a much easier time if the exact same words are used by everyone. Some people think that natural language processing can be used to address this problem. But being very consistent in the language you use is a more effective strategy. We’re laying the groundwork first by standardizing and structuring so that, when machine learning is more universally applicable, we’re already there to use it.”
At HAP, our team has developed an analytics process that helps physicians identify such variations in dictated language so that they can improve documentation procedures now, and be better prepared for the implementation of applications that use machine learning.
In the same article, Arun Krishnaraj, MD, associate professor of radiology and medical imaging at University of Virginia, states that radiologists “are sitting in the middle of an information storm…Our job as radiologists is to synthesize all this information into something actionable.” As radiology revenue cycle experts, we at HAP agree with this perspective and are fully focused on enabling our clients to address such challenges. Our goal is to use our analytics to realize efficiencies for healthcare providers already overwhelmed with data.
Case in point: HAP’s Deep Dive Analytics application is accounting for the predictable differences in human language, addressing the limits of NLP applications, and helping to identify the intended content of physician dictation. By working in collaboration with physicians, we empower next-steps across many crucial actions impacting patient care and practice productivity. Results-to-date have demonstrated significant efficiencies. These include identifying common documentation issues to help educate physicians and increased patient compliance with incidental findings due for follow-up by as much as 20 percent in one year.
Standardizing radiology documentation remains a prime area of focus for practices that need to maximize their revenue cycle. Today, radiology practices have even more incentive to do so in order to benefit from the efficiencies of the many technological advancements available right now, as well as those right around the corner. HAP will continue to help modern practices make sense of these changes. Subscribe to our radiology RCM blog now to stay current on our latest advice.
- Short, Ryan G. et al., “A Normal Thyroid by Any Other Name: Linguistic Analysis of Statements Describing a Normal Thyroid Gland from Non-contrast Chest CT Reports”. Journal of the American College of Radiology, published online. Accessed 5/27/18 at https://www.jacr.org/article/S1546-1440(18)30503-9/fulltext
- National Center for Education Statistics, “The Health Literacy of America’s Adults: Results from the 2003 National Assessment of Adult Literacy”. Accessed 5/27/18 at https://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2006483
- Friedman, P.J., “Radiologic reporting: structure”. AJR Am J Roentgenol. 1983;140:171–172. Accessed 5/27/18 at https://www.ajronline.org/doi/pdf/10.2214/ajr.140.1.171
- Van Dyke, Maggie, “A New Focus for Imaging”. Health Data Management. Accessed 5/27/18 at https://assets.sourcemedia.com/68/2c/ac7437974a168bb3be12494c37d3/hdm-091017.pdf