ICD-9 MCOD N / E Code Naming Convention?

Summary - We would like to agree on a naming convention for distinguishing N vs E codes in ICD-9 multiple cause of death (MCOD) analyses, and which to list first.

From Dr. Baskin at CHS - We do not have the nature-of-injury codes in our version of the ICD-9 MC. But we have these in the ICD-10 MC. In the ICD-10, these codes are codes with letters S and T (chapter XIX of the ICD-10). The nature-of-injury codes are very important for the multiple cause-of-death analysis, especially for such requests like brain (head) or spinal cord injury deaths or for another hot topic - heroin /opioids-related deaths or other specific drug poisoning mortality (ICD-10 codes T36 - T50). We have these requests almost every week and Vitalnet is very helpful to process these requests quickly (especially legislative and media requests).

The ICD-9 coding system is different and has the same numerical expression for the external (800-999) and for the nature-of-injury codes (800-999). In ICD-9 books, external cause-of-injury codes (E-codes) are designated with the capital letter "E" as "E800-E999". The nature-of-injury codes are included in a Chapter XVII and are designated by codes 800-999.9.

In Texas death database, the ICD-9 external codes do not have the letter E (unfortunately). To distinguish the external and nature-of-injury codes, there is the special Nature-Of-Injury value that accompanies the ICD-9 codes, where 1 = nature-of-injury code (from the Chapter XVII) and 0 = all other codes (natural or external). DF_RECORD field consists of 5 characters for 20 possible codes, where characters 1-4 are the ICD cause code. If the ICD code has only 3 digits, the 4th character is blank. Character 5 is the Nature-of-Injury value (index). This value is not the part of an ICD code; it is just a Flag to indicate the nature-of-injury codes and designate them from other (natural or external) codes.

In my understanding, it would be helpful to have the nature-of-injury codes in the ICD-9 MC as well, the problem would by how to code them though. Also, if you will exclude the Nature-Of-Injury value, how your programming will recognize the actual External codes (800-999.9) from the nature-of-injury codes (800-999.9)? Or you will just exclude the nature-of-injury codes from data before running the Vitalnet programming?

ICD-9 N codes vs E codes - In the ICD-9 death system, "E codes" (external cause) and "N codes" (nature of injury) share the same 800-999 numbering. For example, N800 is "Fracture Of Vault Of Skull", E800 is "Railway Accident: ...". E codes may be used as the underlying cause of death, N codes may not.

N codes not included in Vitalnet ICD-9 MCOD - N codes (N800-N999) have never been included in Vitalnet ICD-9 MCOD analyses. Nobody ever raised the issue before. In contrast, the ICD-10 MCOD data import process includes all ICD-10 codes in the data file. The S and T codes are always included in the ICD-10 MCOD.

The ICD-9 data file as currently imported - The ICD-9 MCOD 100-character field has twenty segments (for up to 20 codes). Each segment is 5 characters long. Character #5 is the NOI flag, not part of the ICD-9 code. As of now, if the NOI flag is 1, the code does NOT get imported into the MCOD module. We want to change that so all the ICD-9 codes are imported into the MCOD module.

Why import N codes into Vitalnet MCOD? - It makes perfect sense to include all the ICD-9 MCOD codes, since it would be consistent with how the ICD-10 Vitalnet MCOD works. The current results, excluding N codes, are not really correct.

Naming convention choices - We need to agree how to display N vs E codes (800-999 overlap). Looking through WHO ICD-9 books, it seems there are a few choices for N vs E codes for the MCOD system:

  #1 - N vs E = 805 vs E805
  #2 - N vs E = N805 vs E805
  #3 - N vs E = N805 vs 805

All three options clearly distinguish N vs E, because Vitalnet results always include the full ICD description. The final choice (#3) seems perhaps best, as it is consistent with the UC system (805 in UC is same as 805 in MCOD), I think whatever we decide this consistency is important.

Which comes first, N or E? - The other question would be "which comes first?" (in selection menus and output listings) (N or E?). The reference books present the N codes first, then the E codes, so that seems the best (arbitrary) choice.

Change convention for E codes in UC version? - If desired, we could switch calling the E codes in the UC module from UC "805" to "E805" version. It is feasible. If we did switch, we should then (for consistency) use E805 in the MCOD version. The disadvantage might be inconsistency with previous reports. Allowing for the possibility of using UC E805 format, here are possible combinations:

  UC - 1-799 800-999
  MC - 1-799 800-999 N800-N999

  UC - 1-799 E800-E999
  MC - 1-799 E800-E999 N800-N999

  UC - 1-799 E800-E999
  MC - 1-799 E800-E999 800-999

Feasibility / Timeline - Without going into details, the programming seems feasible. The issue of recognizing N codes from E codes is already taken care of. The programming might take a few weeks to a few months, mostly depending on my schedule. I'm interested to do it. It is a needed improvement.

N codes NOT in UC analyses - I am not suggesting to include N codes in Vitalnet ICD-9 UC analyses. The UC analyses will continue to use the E-codes for 800-999. Similarly, the ICD-10 UC analyses exclude the S and T codes.

Summary - We need to pick a naming convention for distinguishing N vs E codes in ICD-9 MCOD analyses, and which to list first (N or E). I am suggesting to use N800 to indicate an MCOD N code, to use 800 to indicate an MCOD E code, and to list N codes first. However, I am open to other suggestions, and would like to get this right.