Was urine output really measured hourly? [MIMIC-IV]

[Qinyu-Allen-Zhao]

Hi!

I learned the KDIGO_uo.sql and also carefully read #476, but I think there is evidence that urine output maybe isn't recorded hourly.

The kdigo-uo query is based on the assumption that data measured corresponds to urine output during the last hour. But I found some documentation was not hourly. I counted the number of those "special" records:

SELECT 
    SUM(CASE WHEN charttime != DATE_TRUNC(charttime, HOUR) THEN 1 ELSE 0 END) AS special
    , count(*) AS total_num
    , ROUND(SUM(CASE WHEN charttime = DATE_TRUNC(charttime, HOUR) THEN 1 ELSE 0 END) / count(*) *100, 2) AS ratio
FROM `physionet-data.mimic_derived.urine_output`

The result is:

Line	special	total_num	ratio
1	375347	3497267	10.73

That means about 10% of documentation is not hourly, for example, at 3:20 instead of 3:00.

Besides, I found that much missing documentation. Take stay_id 30367285 as an example

SELECT 
    *
FROM `physionet-data.mimic_derived.urine_output`
WHERE stay_id = 30367285
ORDER BY charttime

There are no records from 18:00 to 20:00, but there is a row of data at 20:31.
If the documentation is hourly, the data between 18 and 20 were missing?

Do those mean in fact, the previous charttime is a start time for the current UO measure? If so, these results can be explained I suppose.

Many thanks!