Meaning of SEQ_NUM in DIAGNOSES_ICD table and missing values in ICD9_CODE field

Question

I am just starting to analyze v1.3 of MIMIC-III data. I have two questions about the DIAGNOSES_ICD table.

1. According to the documentation [1] : "ICD diagnoses are ordered by priority - and the order does have an impact on the reimbursement for treatment." What does priority really mean in clinical settings? Also, does it mean the dataset does not have any information about when a diagnosis is confirmed? E.g. a patient may have some cancer on day 1 but acquire pneumonia on day 5?

2. There are 47 rows in this table where the ICD9_CODE is an empty string, what do they mean? Should I just remove them from my analysis?

[1] http://mimic.physionet.org/mimictables/diagnoses_icd/

Thank you very much.

Cheers.

Answer 1

For Q1 see diagnoses_icd.sequence in MIMIC-III: does the order matter aside from the primary diagnosis? and Priority sequence column in diagnosis_icd. (#156 - The SEQUENCE column in DIAGNOSES_ICD and PROCEDURES_ICD has been renamed SEQ_NUM in MIMIC-III v1.2)

For Q2: see https://github.com/MIT-LCP/mimic-code/issues/24:

Question:

I see 47 diagnoses_icd.hadm_id are null:

SELECT count(*) FROM  mimiciii.diagnoses_icd  
WHERE diagnoses_icd.seq_num is null; Bug or intended?

Bug or intended?

Answer:

Depends on your perspective! The null entries appear in the raw hospital data for reasons unknown to us.

We try to provide a true representation of the underlying data and generally err on the side of too little cleaning rather than too much, so the rows were not removed during the process of building MIMIC.

We could remove them in the future. I'll make a note to consider this for the next release.