Getting rid of the leading 0s in 'icd_code' column in 'd_icd_diagnoses' table causes multiple ICD-9 codes with different meanings

[tnamli]

Prerequisites

• [X] Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

Description

You don't use the leading 0s while converting actual ICD-9 codes to values in "icd_code". For some of the codes, this causes duplications with different meanings. For example, in the table there are two entries for ICD-9 code "1882" from which one of them is in fact "018.82" meaning "Other specified miliary tuberculosis, bacteriological or histological examination unknown (at present)" and other one "188.2" meaning "Malignant neoplasm of lateral wall of urinary bladder".

[tompollard]

Thanks @tnamli. I have double checked the data files for MIMIC-IV. Just to confirm, we do list the leading zeros for ICD-9 codes, e.g.:

icd_code	icd_version	long_title
00321	9	Salmonella meningitis

My guess is that whatever tool you are using to view the data is loading the field as an number. As you see in the building scripts, the field should be treated as a string/char:

https://github.com/MIT-LCP/mimic-code/blob/8047c9d7e6ad0d0f8549c8a116cd1799ca1c973a/mimic-iv/buildmimic/postgres/create.sql#L61-L67