ICD-9 and ICD-10 code regex

icd_regex.md

ICD code matching regex

The regex patterns below only help validate when something is not valid ICD-10 or ICD-9. They do not ensure that the code exists. You should consult a list of valid ICD codes for that level of verification.

ICD-10-CM list: https://gist.github.com/jakebathman/063c50cb9772e4bfc864a9e1ff4ccc8d

ICD-9 list: https://gist.github.com/jakebathman/f1ed0d473f12091a708243b0ddf03d82

ICD-10/ICD-9 combined regex

Useful for validating field values that could contain both, and may or may not use decimals

([V\d]\d{2}(\.?\d{0,2})?|E\d{3}(\.?\d)?|\d{2}(\.?\d{0,2})?)|([A-TV-Z][0-9][A-Z0-9](\.?[A-Z0-9]{0,4})?)

ICD-10-CM regex

([A-TV-Z][0-9][A-Z0-9](\.?[A-Z0-9]{0,4})?)

ICD-9 regex

([V\d]\d{2}(\.?\d{0,2})?|E\d{3}(\.?\d)?|\d{2}(\.?\d{0,2})?)

Do you have a CSV or text file with just the ICD 10 CM codes and descriptions?

@mpjohnston sure, you can find that from CMS here: ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD10CM/2017/

Specifically, you're probably looking for this file (~6 MB): icd10cm_codes_2017.txt

I think the ICD10 one can be simplified a bit: removed the ? after the . and changed the latter section to {1,4} (exactly 1 to 4 characters) because of the trailing ?.

([A-TV-Z][0-9][A-Z0-9](\.[A-Z0-9]{1,4})?)

Just about everything about these regexes is wrong.

• Purely numeric ICD-9 codes have three digits before the period, not two, so \d{2} should be\d{3}.
• The periods aren't optional if they're followed by digits, so \.? should be \. everywhere.
• A period followed by no digits is not valid either, so {0,2} and {0,4} should be {1,2} and {1,4}.
• Arguably, excluding U from the ICD-10 prefixes is incorrect too (c.f. https://en.wikipedia.org/wiki/ICD-10_Chapter_XXII:_Codes_for_special_purposes)

See also, slide 7 of https://www.cms.gov/Medicare/Coding/ICD10/downloads/032310_ICD10_Slides.pdf

@ewanmellor Slide 7 does not seem consistent with slide 9. The former says that character 2 is always numeric and the latter says alphanumeric. Do you know which it is?