ICD-10: Translating Headers

In a few days, it will have been a year since HHS made the announcement that a delay in ICD-10 was in the wind. I imagine a lot of folks heaved a big sigh of relief then and went off to work on more pressing demands. Well, that year is about used up and we’re back to where we were then—with the implementation date looming.

Here in 3M HIS Clinical Research, we’ve spent that year overhauling our primary vehicle for getting from ICD-9 to ICD-10, the Code Translation Tool (CTT 2.1). Among many new features, it includes a DRG Financial Impact Analysis module, which the people who pay for things seem to be particularly interested in. Hence the last few of these blog entries have highlighted some of its results. But my real sympathy (and the bulk of the tool) is for the people who have to slog through the process of finding everything in their institution that is “said” in ICD-9 and say it anew in ICD-10. While CTT users have been doing that, I’ve been saving up my answers to the questions they have asked.

How do I translate non-billable codes?

What do they mean by “non-billable” codes? This is an area where the terminology is murky. If you look at the coding books for ICD-9 or ICD-10 diagnoses, they are organized in outline form:  The section “Diseases of other endocrine glands” has 250, Diabetes mellitus, in it (along with ten others), which in turn has 250.0, Diabetes without mention of complication, under it (along with nine others), and that has 250.03, Juvenile type, uncontrolled, under it (along with three others). In the general sense of “code” as “a string of letters and digits that uniquely stands for something,” they are all codes. But only the last one (250.03) can you put on a HIPAA-compliant claim form and get paid for it. So when a coder says “code,” she probably means the codes at the bottom of the outline—the leaves on the tree—the entries in the book that are not further subdivided.

What do you call those other things? I prefer “headers.” Some people say “non-billable codes,” but it’s possible a payment system somewhere pays on them. Technically, the three-character ones (no decimal point) are called “categories,” and the ones under them that are further subdivided are called “subcategories.” So you could say “categories and subcategories.” But, for example, category 042, HIV, is not further subdivided and so is also a (billable) code. So “categories and subcategories” can be confusing, too.

All right, already. So, how do I translate headers?

Easy there, not all of us are coders. We don’t own coding books or coding software. ICD codes are just little bits of technical jargon that we find in our documents and computer programs. And in those places, people often get away with headers. It is quite common to see “250 Diabetes” in a list of, say, pre-existing conditions. When that happens, we assume it means any subcategory or code that begins with “250.”

Now to translate that into ICD-10. Where things go wrong is that people assume that ICD-10-CM is the same tree as ICD-9-CM, just with more branches and more leaves. If you look at the first page of the book, you would be reassured in that opinion. ICD-9 starts off with 000, Cholera, subdivided into three codes. ICD-10 starts off with A00, Cholera, subdivided into three identical codes. But go further, and the impression is quickly dispelled. Detail has been added, sure, but things that don’t make sense anymore have been removed, other things have been moved around, and in some areas, completely new approaches to classification have been taken.

So to translate a header, you have to do what CTT does. First, express it in terms of its constituent codes. 250 in ICD-9 resolves into 40 codes, from 250.00 through 250.93. Next, translate each of those codes into ICD-10. Using the GEMs, CTT tells us there are 140 ICD-10 codes that collectively carry the meaning of the 40 codes under 250. Some of them may have too much meaning – for example, E10.630, Type 1 diabetes mellitus with periodontal disease, implies the patient has diabetes, but it also implies periodontal disease, which nothing under 250 said anything explicitly about. This is where your understanding of what the list is for comes into play. If you wanted to catch all the people with Diabetes and related complications of diabetes, then E10.630 stays. If you want to catch those people with only Diabetes and no related complications, then it might not.

Finally, after throwing out any ICD-10 codes that do not suit your purpose, you collect the remaining codes into a range or set of ranges. If you keep all of the ICD-10 codes CTT identified, you get E10.10-E13.9. You can readily see this cannot be expressed as a single header—though by checking that there is no E10 code before E10.10 and no E13 code after E13.9, you see you can get away with E10-E13 as the translation of 250.

I picked a relatively well-behaved example. There are much uglier ones. Take for example 174, Malignant neoplasm of female breast. When translated you get:

C50.011-C50.019 Malignant neoplasm of nipple and areola, female

C50.111-C50.119 Malignant neoplasm of central portion of breast, female

C50.211-C50.219 Malignant neoplasm of upper-inner quadrant of breast, female

C50.311-C50.319 Malignant neoplasm of lower-inner quadrant of breast, female

C50.411-C50.419 Malignant neoplasm of upper-outer quadrant of breast, female

C50.511-C50.519 Malignant neoplasm of lower-outer quadrant of breast, female

C50.611-C50.619 Malignant neoplasm of axillary tail of breast, female

C50.811-C50.819 Malignant neoplasm of overlapping sites of breast, female

C50.911-C50.919 Malignant neoplasm of breast of unspecified site, female

Why? ICD-10 includes, intermingled throughout C50, neoplasm of male breast.

Bottom line: translation of a header can be straightforward (000 to A00), tolerable (250 to E10-E13), or a nightmare (174 as above). But with a little care, it can be done. Enjoy.

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

2 responses to “ICD-10: Translating Headers

  1. I get 128 ICD-10 codes for ICD-9 250 codes using the latest GEM 9 to 10. Any chance you could send me a list of the 140 codes you got or explain how you got them?

    Thanks
    momenee@email.unc.edu

  2. These are obtained from the 10-to-9 GEM using reverse lookup. Why? This is fully explained in the DIY ICD-10 Conversion series on this site, specifically Part 7 posted in February, 2014. The 9-to-10 GEM has very limited use, and if we had the GEMs to do all over again, it wouldn’t exist at all.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s