search

MIT-LCP / mimic-code

MIMIC Code Repository: Code shared by the research community for the MIMIC family of databases
https://mimic.mit.edu
MIT License
2.55k stars 1.51k forks source link

Itemids for a list of variables #540

Closed osmanii closed 5 years ago

osmanii commented 5 years ago

Prerequisites

Description

For one of my projects, I have to pull data for the following variables from the MIMIC-III database:

  1. SpO2
  2. FiO2
  3. PaO2
  4. PaCO2
  5. SaO2
  6. Hemoglibin
  7. pH
  8. Lactate
  9. SOFA
  10. COPD
  11. VT
  12. Active Smoker
  13. ARDS
  14. PEEP
  15. Peak Airway Pressure
  16. Minute Ventilation
  17. Ischemic heart disease

------Variables-------------------ItemdID--------------------------------

  1. SpO2---------------------- (CE) 646,220277
  2. FiO2---------------------- (CE) 190, 3420, 3422, 223835, (LE) 50816
  3. PaO2---------------------- (CE) 779, 50832, (LE) 50821, 490
  4. PaCO2--------------------- (CE) 778
  5. SaO2---------------------- (CE) 834, 220227, (LE) 50817
  6. Hemoglibin---------------- (CE) 814, 220228, (LE) 50811
  7. pH------------------------ (CE) 780, 223830, (LE) 50820, 50831
  8. Lactate------------------- (CE) 1531, 225668, (LE) 50813,
  9. SOFA---------------------- Used sofa.sql script
  10. COPD--------------------- Looked at icd9 codes: 490-496
  11. VT----------------------- (CE) 682, 224685
  12. Active Smoker------------ Didn't find anything
  13. ARDS--------------------- Didn't find anything
  14. PEEP--------------------- (CE) 505, 506, 220339
  15. Peak Airway Pressure----- (CE) 535, 224695
  16. Minute Ventilation------- (CE) 450, 224687
  17. Ischemic heart disease--- Looked at icd9 codes: 410-414
    • I am worried about pulling data from the right sources (itemids and icd_9 codes) which will greatly influence the experiment I will do later. Now, could you please help me to validate the itemids for these variables and also, point me to the right direction. Thanks in advance. Your help will be greatly appreciated.
alistairewj commented 5 years ago

You don't need to pull lab data from both chartevents and labevents. The data in chartevents is inserted during routine care from the underlying lab database, and labevents is the underlying lab database. The main reason you may be interested in using chartevents is that it has the actual time the data was available to the practicing clinicians (the store time).

Defining patient diagnosis using ICD-9 codes is equivalent to defining patient diagnosis using billing data. It can work but there are caveats you should be aware of and many are discussed in the issues.

You won't really find a satisfactory answer to such broad questions as "how do I define ARDS?" and "how do I define COPD?". You'll have to find local colleagues to discuss it with and come up with the best solution. If you have more specific questions about a plan then we can try to help, and of course we always welcome new contributions to the code repository for new concepts!

osmanii commented 5 years ago

Thanks for the quick and helpful reply. What do you think about the chartevents itemids for the variables? Am I pulling data from the right source?

alistairewj commented 5 years ago

It's up to you to validate that the data you are using is correct. With over 9000 users we can't provide individual study support.

osmanii commented 5 years ago

I understand. Thanks!

santinon commented 5 months ago

for the itemids and their corresponding physiological variables, I have a question : are the itemids the same in both mimic 3 and 4 ? did they change or there was a consistency ? (especially for those mentioned above, by osmanii)