Computer Assisted Coding – Potential and Problems

Computer-assisted coding depends on the accuracy of the input.

There is considerable interest in computer-assisted coding or “CAC.” The proponents say it will reduce costs, coding backlogs and discharged not final billed claims.  There are several companies in this space, the largest being 3M with more than $30 billion in revenue every year. 

CAC is part of a larger move into artificial intelligence or AI.  Many companies, including those in the healthcare sector, would like to replace costly humans with computers.  In the 1960s, the television program, “The Twilight Zone,” had an episode in which a man running a manufacturing plant turns to robots to reduce costs and increase efficiency.  In the episode, Wallace V. Whipple replaces people with robots, despite the stern warning of the employees. The show ends with Mr. Whipple, drinking in a bar, angrily complaining as his robot doppelganger sits in his chair at the office smoking a cigar. All the humans are gone, replaced by robots that look like people.

With CAC we do not have human-looking robots sitting at desks coding accounts.  We have computer software matching data from electronic medical records into diagnosis and procedure codes.

The benefits of CAC include the fact that it is fast, scalable, consistent and accurate based on the data in the medical record.  There are no coding backlogs as CAC software can run through unlimited claims in a fraction of a second.  It never codes for things not in the medical record.

The first problem facing CAC is “garbage in garbage out,” GIGO. CAC depends on the assumption that the medical record is complete and accurate.  Let’s take evaluation and management (E&M) services as an example.  The computer can look at the electronic medical record to find documentation of the patient’s history and physical.  It can look for the documentation of the review of systems and it can look at the amount of time from the time the patient is admitted until the patient is discharged.  The software can then give a code for the service.  What is needed is to know if all the data is being correctly being loaded into the system and if the data is accurate?  Did the physician really have the “face time” the medical record states?

The second issue is oversight of the coding process.  When healthcare providers are doing manual coding, there is at least some communication between health information management (HIM) and billing.  All the stakeholders in the revenue cycle usually meet regularly to discuss challenges.  Overdependence on CAC can break down this communication.  Compounding this is the fact that computers can make mistakes faster than humans.

In conclusion, CAC can increase consistency and accuracy to the coding process, but like any other revenue cycle process, will require daily oversight and making sure the documentation supports the coding. 

Comment on this article