By: Rhonda Butler
This blog is a summary of many of the points I have been trying to make about the GEMs over the last several months. After this blog I will lay off the subject of the GEMs for awhile, and concentrate on other aspects of ICD-10.
There are three basic ground rules every user of the GEMs should be aware of. The GEMs
- Are a code translation reference—not a medical record converter and not a crosswalk
- Assume the user has no access to the original medical record
- Are not “plug and play”—cannot be used as is to create fully automated translation
These three constraints need to be taken into account for correct, efficient use of the GEMs.
Ground Rule 1: A Starting Point, Not a Panacea
First, it must be emphasized that the GEMs are a general purpose code to code translation dictionary—not a one-to-one “crosswalk” and not a medical record converter. Because a GEMs entry is based on the complete official meaning of each code, the GEMs is an “unabridged” translation dictionary. It contains all the plausible translations of a source system code, based on the level of detail in the source system code. As a general purpose, unabridged dictionary, the GEMs cannot limit the translation alternatives based on some specific purpose like research or reimbursement.
So by definition the GEMs are not a crosswalk. If an organization needs a crosswalk for some reason (hopefully to “tide them over” for a defined ICD-10 transition need, and not as a substitute for implementing ICD-10 in their systems) than they must develop a crosswalk from the GEMs by using some reliable, transparent method (clinical rules, ICD-9 frequency data) to whittle down the GEMs based translation alternatives appropriate for the purpose of the crosswalk.
“Medical record converter” is shorthand for a context-sensitive translator that is aware of the other codes on an individual medical record, and can adjust the translation alternatives accordingly for coherent, clinically plausible results. The GEMs are plain vanilla text files, not software, and so they cannot “intelligently” convert medical records.
In the GEMs, the unit of meaning is one code, period—the code you are looking up. Put another way, the GEMs is a “transliterator” in the same way a foreign language dictionary is. For example, you are trying to translate a simple sentence from Chinese to English. If your method of translating is to look up all the Chinese words in the sentence and write down the first definition listed in the Chinese-to-English dictionary, the result may be comprehensible but it will certainly not be correct English. More likely, the result will be gibberish.
Sophisticated medical record conversion software can do better than a “transliterator.” It can look at all the ICD-9 codes on the record as a single unit, and use the information in one ICD-9 code to whittle down the ICD-10 translations alternatives for another ICD-9 code, so that it can create a plausible converted record that can be used with a bit more confidence for testing ICD-10 applications and forecasting ICD-10 impact. It can use programmed rules to make more informed choices than the code translations found in a GEM entry.
Ground Rule 2: The Medical Record is Long Gone
The GEMs assume a user has no access to the original medical record. Therefore, all a GEM entry can do is correlate the complete official meaning of a code in one system with the complete official meaning in the other coding system. And what is “the complete official meaning of a code?” It is the code title, any instructional notes such as “includes” notes that pertain to the code, and the index entries that refer a user to that specific code. Applicable reference sources like Coding Clinic and official coding guidelines are also taken into account.
If a person has access to the medical record and wants to code the record in ICD-9 or ICD-10 or both, then the GEMs are unnecessary. Just code the record directly. If you are that person—you need to code a record or look up individual codes occasionally and you are using the GEMs as a “cheat sheet,” remember that although the GEMs were developed by trained coders they lack something you have: a brain. The content is general purpose and indiscriminate—it is not capable of inferring purpose or context. Armed with information about purpose and context, a human being with common sense and a reasonable comprehension of medical terms can easily make decisions that the GEMs cannot make.
Ground Rule 3: Not Plug and Play
The third ground rule is really the conclusion reached by the first two ground rules. For the reasons above, any useful and acceptably accurate use made of the GEMs must be application specific and must involve the input of subject matter experts.
If the goal is to convert an application, policy, or other document, the conversion process must include the direct participation of experts from both the systems and HIM realms. The technical subject matter expert has a solid understanding of the application being converted, and the HIM professional is familiar with both the code sets. These subject matter experts can resolve issues exposed by a GEMs-based ICD-10 conversion in the context of a particular application, to ensure the converted ICD-10 application works as intended.
If the goal is to create a crosswalk between ICD-10 codes and ICD-9 codes, that crosswalk must be application specific. It needs clinical rules and/or rules based on the organization’s own ICD-9 legacy data to help determine which is the best choice when face with multiple translation alternatives for an ICD-10 code. These rules will help create the most accurate and useful crosswalk possible.
Rhonda Butler is a Senior Clinical Research Analyst with 3M Health Information Systems.