Mapping ICD code to corresponding text

[NarenInD]

@alistairewj

Below is an example:

---

**HADM_ID**: 100001
**SUBJECT_ID**: 58526
**ICD9_CODE**: ['25013', '3371', '5849', '5780', 'V5867', '25063', '5363', '4580', '25043', '40390', '5853', '25053', '36201', '25083', '7078', 'V1351']
**TEXT**: "

... <snip> ...

Discharge Condition:
Mental Status: Clear and coherent.
Level of Consciousness: Alert and interactive.
Activity Status: Ambulatory - Independent.

...

---

If I check on Discharge Diagnosis there are only 4 diagnosis: Diabetic keotacidosis Hematemesis (blood in your vomit) Hypertension Chronic renal insufficiency

Total ICD codes are 16. But the Discharge Diagnosis is 4

From where are the other 12 ICD codes drawn. Exactly where are the sentences in discharge summary corresponding to the ICD code

Other than Discharge Diagnosis what are the other headings taken into account for ICD codes

[alistairewj]

Please do not post an entire discharge summary on a public forum like GitHub! This is a violation of the data use agreement. I understand the desire to include context but if you are doing so, please only include the bare minimum required. I have edited your post. You could include the row_id in the future if you want to reference a note in noteevents.

To answer your question: the discharge diagnosis section is written by a provider and represents what they believe to be the discharge diagnoses. After writing and signing this note, hospital staff (trained "coders") review all patient notes for the stay, and code discharge diagnoses. If necessary, they will confirm diagnoses are valid with the care provider. As such, you tend to get more billed diagnoses than those listed in the discharge diagnosis section.