ICD-10-PCS: Making It Real

Sept. 2, 2015 / By Rhonda Butler

A system for classifying knowledge is a framework for organizing information. It is usually a vastly simplified model of some aspect of reality as we understand it, like the periodic table, is a simple model for representing our understanding of chemistry. In that sense, ICD-10-PCS is one among several systems that attempt to construct a systematic way of describing the things done to the human body that we collectively call “procedures.”

When I remind coders who are grappling with some aspect of ICD-10-PCS, that PCS is a model of reality and not reality itself, it gets a wry laugh and a sort of “no kidding it’s not reality” look. Of course my saying that PCS is a model doesn’t help at all, because coders are caught up in their frustration at not being able to do the one specific thing they want to do—use ICD-10-PCS to accurately code a specific procedure.

Coders are end users of ICD-10-PCS, and end users are typically very different in cognitive style and temperament from the designers of systems like PCS. Designers look for patterns in the specific in order to build good general models. End users are interested in accomplishing a specific task, and they aren’t interested in the model, just in knowing what they need to know to get their work done. We need both designers and end users in the world, of course. They perform essential but very different functions. You don’t want the nerds who crack the unified theory making up the shopping list for your spaceship’s kitchen. Same deal here. Designers design, and end users make it work.

The group who designed ICD-10-PCS had the job of thinking globally about the essential physical actions that comprise a procedure, and asking the specific question, “How would procedure X be coded in ICD-10-PCS?” only as a way of testing the basic design. This is called “constructive testing.” You come up with specific examples that test the existing model. Under ideal conditions, this project would have also had a dedicated bunch of coders whose job is to take the existing model and try to break it. This is called “destructive testing.” They test the initial design by attempting to code from the whole spectrum of real medical records using the system. Then they report back—this thing works, that thing doesn’t work. And then designers take that very welcome information about the real world back to the drawing board.

As you might expect, ICD-10-PCS was designed under less than ideal conditions. PCS was developed by a small group whose chief function was to design the initial model. After the initial design was released, the system was subjected to a series of “test drives” over the years—by the Central Data Abstraction Centers (in 1998!), by the ICD-10 MS-DRG conversion project, by the General Equivalence Mappings project. But none of these things represent the final, actual intended use of ICD-10-PCS: for inpatient hospital medical record coding, day in and day out, by tertiary care hospitals, small hospitals, children’s hospitals, and so on.

For a system to be mature—meaning that it is both intellectually rigorous and practically useful— it needs the contributions of both designers and end users. Under ideal conditions, designers and end users collaborate closely all the way. We didn’t get to have the ideal development environment for ICD-10-PCS—the work of the designers and the ultimate end users was separated by about 15 years. The PCS design nerds are all retired (with the possible exception of me, the last nerd standing), but the people who make up the shopping list are showing up in droves. PCS has been thoroughly tested by end users in the last two or three years. They are coding the whole spectrum of real world medical records in PCS, and they are discovering things about the system that they want added or changed.

So we should all expect and, hopefully welcome, this break-in period when PCS gets thoroughly worked over by end users. When they find something they want added or changed to PCS, they send CMS their shopping list. That’s how it works.

Rhonda Butler is a senior clinical research analyst with 3M Health Information Systems.

ICD-10 is fast approaching.  Is your practice ready for the transition?  Find the tools you need to succeed.