Follow Up Project

Many patients are lost to follow up as they transition through the healthcare system. Often the patient or primary provider might overlook a recommendation until the underlying pathology has progressed. In this project, we examined Radiology patients over a decade to understand which patients were lost to follow up. An automated system was then constructed to recognize Radiologists’ recommendations, and to automate flagging at risk patient populations.

In this study, our client (a leading Radiology practice) was interested in measuring the impact that they had on the clinical management of their patients. As a consultative practice, Radiology plays a key component in clinical management, but often does not have metrics to demonstrate its value.

One of the clear areas that a specialty consultation can show its value is in the recommendations made to the referring physician. In this project, the challenge was that the recommendations were made in the radiology report in a variety of phrasings dependent on the physicians. Technology was developed to extract the recommendations, classify those recommendations in categories (e.g. further imaging studies, additional procedures). The clients’ historical practice record over the last decade was then analyzed to obtain metrics of follow up compliance.

For simplicity, the client was interested in the patient follow up compliance for recommended imaging exams (roughly 15% of all patients). The metrics showed that the patients were compliant roughly 60% of the time. The 40% non-compliant group consisted of patients who went elsewhere for imaging exams, overlooked the recommendation, or had conditions that resolved in the meantime.

A simple ROI calculation was possible: assuming that of the 40% of non-compliant patients, roughly half should have clinically had an additional imaging exam. In this practice, this accounted for roughly $20M in yearly missed income. An additional risk-based ROI was also possible based on the likelihood that a non-compliant patient might have a critical finding that would later lead to risk of a lawsuit.

Our compliance algorithm was considered for use as a prospective system to flag patients that needed further communication to ensure timely follow up.

In this project, we creatively designed and implemented an automated method to address the client’s specific issue.

Snowstone can assist with designing, implementing, and deploying the right metrics for your institution.