ICD-10 Scare Tactics

October 25th, 2011 / By Ron Mills, PhD

At AHIMA earlier this month, I kept seeing and hearing assertions that went something like this:

Because ICD-10 is so much more complex and demanding, your coder productivity is going to drop dramatically (requiring you to add coders or lose revenue) unless you buy [the speaker’s] new coding technology.

I’m all in favor of technology that improves the accuracy, reduces the cost, or diminishes the tedium of coding. But the statement that ICD-10 is so much more complex and demanding is just not true.

Worse, it is becoming the dominant meme. People who have never opened a code book hear it, believe it, and pass it on. Pretty soon it is echoing down the corridors of power and we start getting rumbles again about putting off the conversion. Puh-leeze. This has got to stop.

Let’s examine the facts. Start with diagnoses. Open an ICD-9-CM code book and an ICD-10-CM code book and put them side by side.

You see right away that the structure of the two coding systems is exactly the same – a hierarchy, an outline, a tree. Broad clinical concepts are successively refined until a code with adequate detail is defined. Hints and recommendations are given at each level (the “instructional notes”) to guide coders to the right place. Coders and software encoders are already skilled at navigating such a tree structure – no new skills or technology is needed to code ICD-10-CM diagnoses – just the new book. It is true that there are about five times as many ICD-10-CM diagnosis codes, but that just comes from there being, usually, one more level to the hierarchy – one more question to be answered. And if you look closely at what that question is, you will find it is almost always something trivial. Affected body part on the left or right? If your doctors aren’t already writing that down, you have bigger problems than loss of coder productivity with ICD-10.

A huge chunk of those new codes are in Chapter 19, Injury and Poisoning. Take a look. They’ve made that whole area much more regular. ICD-9-CM was a hodge-podge requiring the coder to go find out what detail was wanted and what not. In ICD-10-CM, these codes are almost a matrix – the same axes of classification are used consistently, over and over, throughout the chapter. Less technology, not more, suffices here. Auto-coder developers will find this chapter a joy to work with.

Okay, there are some areas where ICD-10-CM emphasizes a different initial axis of classification than ICD-9-CM did. But look closely at those and you won’t see any head-scratchers. OB is everybody’s favorite poster child. In ICD-9-CM the first consideration is “did the patient deliver?” In ICD-10-CM it is “what trimester was she in?” Now any woman can tell you, what is the first question she is asked when she shows up anywhere for obstetrical care?

Now turn to the back of the books. Both have an Index which relates common medical terms to the code hierarchy. Coders use this all the time. Same structure, same organization, same processes apply.

If you’re dealing with physician or hospital outpatient claims – the vast majority of claims – you can stop right here. CPT/HCPCS will continue to be your procedure coding system – that is not going to change in 2013. Go out to www.cdc.gov/nchs/icd/icd10cm.htm and take a look at the diagnosis code book and fortify yourself against the doomsayers.

Hospital inpatient coders read on. Instead of the 3,838 ICD-9-CM procedure codes, nominally in a hierarchy but actually stuck every which way (since space in the coding system effectively ran out years ago) you now have about 70,000 codes in a new thing called PCS. Sounds daunting, doesn’t it? But it turns out PCS is organized around about 3,000 concepts, which are fit together in an extremely regular way, where the axes of classification – the questions you have to answer – are always the same. The whole thing is elegantly expressed in a series of tables. PCS was designed to make coding easier. Once you “get” the PCS way, you’ll find that true. Even I can code in PCS. (Rhonda is rolling her eyes here.)

Okay, the hard part of inpatient procedure coding is reading the op report and understanding what went on – picking the codeable events out of the narrative detail. That doesn’t change. It doesn’t get worse with ICD-10, for a person or an NLP. But once you’ve figured out what you want to code, generating it in PCS is actually easier than searching around in the ICD-9 procedure book, trying to find something that fits and isn’t embarrassingly vague. Software that helps you do that in PCS exists – I know, I’ve written some of it.

I admit that providers who have not invested a little in coder training or installed required upgrades to their coding software will experience delays getting paid in the last quarter of 2013. But even that avoidable hiccough should be short-term, as coders discover that ICD-10 isn’t the big deal the fear-mongers want them to believe it is. I’ve heard estimates of upwards of 30% permanent reduction in coder productivity with ICD-10. I don’t believe them.

Like I said in the beginning, if you find new technology that will increase the cost-effectiveness of your coding processes, by all means go for it. Replace your coders with a bank of servers up in the cloud if you want. But please justify the move on its own merits. Don’t blame poor new ICD-10.

Ron Mills is a Software Architect for the Clinical & Economic Research department of 3M Health Information Systems.