MIT-LCP / eicu-code

Code and website related to the eICU Collaborative Research Database
https://eicu-crd.mit.edu
MIT License
302 stars 209 forks source link

Interpreting intakeOutputOffset #244

Open brian-mclaverty opened 3 months ago

brian-mclaverty commented 3 months ago

Hi @jraffa , @alistairewj , @obadawi, @tompollard and all,

I am trying to interpret the meaning of intakeOutputOffset from the intakeOutput table in eICU. Per table column description in https://eicu-crd.mit.edu/eicutables/intakeoutput/, intakeOutputOffset is 'the number of minutes from unit admit time that the I and O value was observed'. An example extracted from eICU intakeOutput table is as follows: at intakeOutputOffset -7 for patientUnitStayID 1021624, 1460 mL of NS and 450mL amount of urine is recorded. Can we interpret this charting as the clinician administered 1460mL of NS to the patient, measured 450mL of urine, and then afterwards these volumes were recorded by the clinician at intakeOutputOffset -7? Or is the interpretation that 1460mL of NS is recorded by the clinician at intakeOutputOffset -7 and afterwards administered to the patient?

Screen Shot 2024-04-15 at 4 04 43 PM

Thanks for your help, Brian

alistairewj commented 3 months ago

It is the time of documentation for the most part. So here you would presume they documented that fluid intake/output roughly on ICU admission (-7, i.e. 7 minutes before ICU admit). There is no guarantee that the amount actually occurred at -7, unfortunately. Usually there are patterns to documentation: some units will have documentation hourly (so you can assume that the volume corresponds for the past hour), some have it per shift (every 12 hours), and some every day. On admission, I would assume they are documenting some recent and relevant amount, e.g. that could be intake while the patient was in ED amount or the floor.

brian-mclaverty commented 3 months ago

This makes sense. Thanks for your help! @alistairewj