ICD-9 to ICD-10 Medical Coding Crosswalk

The United States healthcare system will discontinue its use of ICD-9-CM diagnostic codes and upgrade its entire system to the next revision of the code: ICD-10-CM. Because ICD codes are integral to the medical billing process, the changes are taking place at every level of the patient-provider-payer relationship. As the person who interacts with diagnostic codes on a daily basis, the medical coder must be prepared for this transition. A medical coder must be fluent in both ICD-9-CM and ICD-10-CM in order to make the switch as seamlessly as possible.

This change was originally planned for October 1st, 2014 , however, the U.S Senate  introduced a bill on March, 26th, 2014 that will delay the change of ICD-9-CM to ICD-10-CM until October 1st, 2015.

A process called crosswalking is used to translate from one code to another. Crosswalking means mapping or translating a code from one set to another. Use the AAPC tool below to crosswalk between the different coding systems.

Learn Why the Change Will Happen

In order to understand this important shift in health informatics, it’s important to first look at why the change is occurring. ICD-9-CM is being phased out because it is outdated (published in 1978) and not flexible enough to deal with changes and expansions to medical procedures and diagnoses. To put it another way, ICD-9-CM is out of room. As new methods, diseases, and diagnoses are discovered, ICD-9-CM has been unable to find space in its code to accurately report these advances in medicine. ICD-9-CM caps out at around 13,000 codes, while ICD-10-CM has roughly 68,000.

Explore How the Change Will Happen

Being able to perform translations between ICD-9-CM and ICD-10-CM codes is a vital skill for a professional coder. One way in which translation is useful is that you can crosswalk codes back to ICD-9-CM if you are tracking or analyzing data. For example, if you are creating a report on the 2014 calendar year (which, again, will see the use of both ICD-9-CM and ICD-10-CM codes), it may be easier to crosswalk ICD-10-CM codes back to ICD-9-CM to create a standard set of data. Crosswalking will also help you update your records and programs.

There are, however, a number of difficulties in this process. Because ICD-10-CM expands so significantly on the body of codes in ICD-9-CM, there are a number of discrepancies and inaccurate or incomplete translations between the two code sets. It’s the medical coder’s job to watch out for these discrepancies and become familiar with crosswalking procedure as the deadline to switch from ICD-9-CM to ICD-10-CM approaches. Note that it is impossible in most cases to perform a 100 percent accurate translation from one code set to the other, as ICD-10-CM is significantly different from ICD-9-CM in terms of the format, concept, and structure.

One resource that will help you learn to crosswalk between the two code sets is the National Center for Health Statistics’ General Equivalence Mappings, or GEMs. These GEMs find and list equivalencies between the code sets, and they are considered the authoritative source for mapping between both sets. The GEMs allow you to map forward and backward between ICD-9-CM and ICD-10-CM. Coders should be familiar with both processes.

Understanding Different Code Matches

There are different types of matches that occur between the ICD-9-CM and ICD-10-CM code sets. The AMA identifies four types of matches between the two code sets, and an additional type that is reserved for “no match.”

One-to-one exact matches

In one-to-one matches, a coder is able to identify an exact match between the two code sets. These are relatively rare; only 5% of codes translate exactly from ICD-10-CM to ICD-9-CM, and just over 24 percent map directly in the opposite direction. One example is the ICD-9-CM code 416.0 (primary pulmonary hypertension) and the ICD-10-CM code I270 (primary pulmonary hypertension).

One-to-one approximate matches with one choice

A significantly more common occurrence in ICD code crosswalking is a one-to-one approximate match. In fact, 82.6 percent of ICD-10-CM codes can be backward-mapped this way, while 49.1 percent of ICD-9-CM codes can be forward-mapped to a similar degree of accuracy. It should be noted that this is not a direct translation, but more of a “close enough” approximation. For example, the ICD-9-CM code 422.91 (idiopathic myocarditis) is an approximate match for ICD-10-CM code I401 (isolated myocarditis).

One-to-one approximate matches with multiple choices

Exact and approximate matches with one choice make up the majority of ICD code crosswalking, but there are still a large number of codes that do not translate with the same level of accuracy. Approximate matches with multiple choices put much more responsibility on the coders, as they must pick the best fit from a number of similar choices. In certain cases, different diagnoses may fall under one code in the other set. For example, ICD-10-CM codes C220 (liver cell carcinoma) and C22 (hepatoblastoma) both correspond to ICD-9-CM code 155.0 (malignant neoplasm of the liver, primary). Each code set has instances of different diagnoses corresponding to only one code in the other. Coders must pay attention to these areas, as a miscoded diagnosis could affect the status of a claim.

One-to-many matches

By far the most difficult instance of crosswalking, one-to-many matches involve one code in a code set corresponding to several codes in the other. For instance, the ICD-9-CM code 80010 (closed fracture of vault of skull with cerebral laceration and confusion, state of consciousness unspecified) corresponds to two different codes in ICD-10-CM: S02.0xxA (fracture of the vault of the skull, initial encounter for closed fracture) and S06.339A (contusion and laceration of the cerebrum, unspecified, with loss of consciousness of unspecified duration, initial encounter). Essentially, the diagnosis code in one set must be created out of multiple codes in another set. The groups of codes that translate to a single code in another set are called “clusters.” Clusters are always between two and four codes in size. Some single codes may correspond to multiple clusters. All codes in a cluster need to be listed in order to fully represent its corresponding translation. This process of translating one-to-many matches requires diligence and constant review, as a code missing from a cluster creates fundamental inaccuracies in your report.