Looking for format for ICD-9 mortality codes

[lstockman]

Hi all,

Does anyone have a proc format statement that assigns the cause of death description the three-digit ICD-9 code? I have this for ICD-9-CM (hospitalization data) but looking specifically for ICD-9 for Mortality.

Thank you!

[PeterClemmensen]

i doubt that anyone has a format stored here with ICD-9 codes. Do you have the codes now? And how are they stored?

[Kurt_Bremser]

If you have a reference list for those codes, it's very easy to create a format from that.

[ballardw]

ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD-9

has codes "modified for use by the United States" as a text document. There are notes about codes groups that have been excluded so you may be able to recreate some of those if you need them. It sounds like you may only be looking for the no decimal levels which reduces the number codes but are a tad harder to read from that (not well structured for data set use) text file

Read the text document ucod.txt into a data set and use the information as a CNTLIN data set for proc format.

I'm not aware of a nicer source with relatively complete codes.

---

@lstockman wrote:

Hi all,

Does anyone have a proc format statement that assigns the cause of death description the three-digit ICD-9 code? I have this for ICD-9-CM (hospitalization data) but looking specifically for ICD-9 for Mortality.

Thank you!

---

If you only use 3 digits your codes are incomplete. You will likely need 4 characters to accommodate codes such as

E800    Railway accident involving collision with rolling stock

[lstockman]

Thanks ballardw! I have the ucod.txt doc, and am trying to work with it. NCHS has stripped out the E and the decimal in the data I have so external codes are reported beginning with 800. I will consider the 4th digit. Was hoping to find the same info in the txt doc in a more usable format. 

[ballardw · Accepted solution]

---

@lstockman wrote:

Thanks ballardw! I have the ucod.txt doc, and am trying to work with it. NCHS has stripped out the E and the decimal in the data I have so external codes are reported beginning with 800. I will consider the 4th digit. Was hoping to find the same info in the txt doc in a more usable format. 

---

I've made a few searches over the past 10 years or so but have never found a "clean" data file of any sort for either ICD-9 or ICD-10 except for smallish subsets of topics.