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
2.51k stars 1.5k forks source link

MIMIC-IV 2.2: Question about "uo_rt_6hr" #1493

Open vasilissoti opened 1 year ago

vasilissoti commented 1 year ago

Prerequisites

Description

Description of the issue, including:

Hello my questions is based on MIMIC-IV, version 2.2:

May I ask what is the "uo_rt_6hr" column located in the module "derived" within the table "kdigo_uo"? Is this the urine output rate calculated for the urine_output_6hr based on the weight? If yes why there is a small discrepancy between the actual calculation and the rate value stored and there are rows that have both weight and a urine_output_6hr values available but not a uo_rt_6hr value?

Thank you for your time and help.

alistairewj commented 1 year ago

You can see the derivation in the source code for kdigo_uo.

Specifically, first the UO is summed over the past 6 hours (and named rate, but after this summation it is an amount): https://github.com/MIT-LCP/mimic-code/blob/87f3a027267afee82d3bb4526872fc1f48ed2161/mimic-iv/concepts/organfailure/kdigo_uo.sql#L18-L28

Then the rate per kg is calculated by dividing by the summation duration and the associated weight column: https://github.com/MIT-LCP/mimic-code/blob/87f3a027267afee82d3bb4526872fc1f48ed2161/mimic-iv/concepts/organfailure/kdigo_uo.sql#L77-L84

The rows with weight and urine_output_6hr likely do not have at least 6 hours of documentation of the UO. This was the interpretation I eventually landed on after much back and forth about how exactly to operationalize KDIGO definitions. KDIGO has a table which says:

< 0.5 ml/kg/h for 6–12 hours

... which is interpreted here to mean they must have at least 6 hours of documentation. The alternative (not requiring a minimum amount of documentation) results in many cases of "AKI" but it's not clear if that's just a lack of UO documentation. I put some numbers in PR #1470 for reference.