Determining SAPS-II Admission Type from eICU data for patientunitstayids

[josephcscarpa]

I am trying to derive the SAPS-II features from the eICU dataset. Unlike the MIMIC Dataset, no official script exists for this task, so I am attempting to construct my own.

Link to SAPS-II calculator: https://www.mdcalc.com/calc/4044/simplified-acute-physiology-score-saps-ii

Definitions for the categories:

• Scheduled surgical = surgery scheduled ≥24 hours prior;
• medical = no surgery within 1 week of admission;
• unscheduled surgical = surgery scheduled ≤24 hours prior

I am aware of electivesurgery from apachepredvar (elective surgery = scheduled surgery from Googling it), so this issue is mainly to ask if there are any additional columns to distinguish between all surgeries and medical (which involves detecting a surgery for a week before admission). I acknowledge this may not be possible to do completely accurately. Here is the approach I was going to try:

1. Determine whether the icustay is related to a surgery (any of the following are true)
   a. elective surgery = 1
   b. Is the admit source:
   -. operating room
   -. recovery room
   -. PACU
   c. admissionDx WHERE admitdxpath LIKE '%All Diagnosis|Operative%’
2. Is it an elective surgery?
   1. elective surgery = 1
3. Medical is all patientunitstayids not found in 1.

Any thoughts or suggestions?

[mirkompcr]

Dear Joey,

Sounds good to see both medically as well as engineering-wise. A pull-request would be very much appreciate, this way your code would be available for everyone.
Please note: eICU is storing the same information in multiple tables (depending on the hospital) -> as such, it can be tricky to obtain the PaO₂/FiO₂ data, as you have to query multiple tables. You need to consider:
-- derived --> ventilation_events
-- original --> respiratorycharting, nursecare, note, respiratorycare, careplangeneral

Hope that helps.