Improving Trial Recruitment Processes by Moving on from Insurance Codes

The field of medicine is constantly evolving as our clinical and scientific knowledge continues to deepen on an almost daily basis. As this happens, medicine is becoming increasingly more complex.

One hundred years ago, a patient with cancer was diagnosed simply as having cancer. Today, there are multiple variations of cancers and lymphomas, each with different molecular biomarkers that can potentially be treated by any number of medications. The same can be said for any number of viruses and diseases. As we discover and learn more about each disease state, we add an increased number of lab tests and medications for consideration and change the way surgeries and procedures are done for treatment and prevention. As a matter of fact, today’s medical student needs to learn more than four times the quantity of information as was required 20 years ago due to the exponential growth of specialists and sub specialists.

For clinical researchers, the increased complexity in today’s field of medicine means that the traditional method of relying heavily on insurance diagnostic codes to identify qualified trial candidates simply does not work.

Insurance codes were initially developed to help doctors and hospitals bill insurance companies for services. Given that codes for diagnoses, medications, labs, and vaccines can be easily recorded, neatly stored, and easily searched within EMR databases, they became useful for clinical staff seeking to quickly find populations of patients with certain medical conditions.

These codes still have a role to play in the totality of the process, but they are not sufficient for obtaining the deeper level of medical information required of today’s clinical research protocols, the kind of data that can typically only be found in the unstructured notes in a patient’s medical chart. The fact is diagnosis codes will always lag behind the speed of scientific progress. They are updated slowly and infrequently. Veteran medical professionals are resistant to embrace new codes out of fear that errors or incorrect code assignments could lead to malpractice or fraud claims.

But, even if this wasn’t the case, the key to truly understanding a patient’s accurate, complete, and detailed medical history requires an analysis of both structured and unstructured data. Given the depths of detail and complexity of this data, researchers need the support of technology that can review and understand non-standardized notes that align with protocols which, by the way, continue to become more complex in their own right.

This is where BEKhealth’s AI-powered platform is at its best. Using Galileo, clinical researchers can quickly access, scan, and understand the nuanced details of a potential trial candidate’s medical history at a granular and uncoded level. With a prebuilt library of more than 24 million medical terms, Galileo helps researchers create a narrowed-down list of potential candidates earlier in the process that meet the numerous inclusion and exclusion criteria of a protocol, without engaging in the highly laborious and manual process of sifting through patient records and trying to understand handwritten notes.

Insurance codes still have value. They can serve as a foundational starting point. But with 70­–80% of all patient medical data residing in unstructured notes, they cannot be the final story of a patients medical history.