MIT-LCP / eicu-code

Code and website related to the eICU Collaborative Research Database
https://eicu-crd.mit.edu
MIT License
307 stars 212 forks source link

Fluid Intake Total #159

Open daisy647lsq opened 3 years ago

daisy647lsq commented 3 years ago

Hi, May I know what is the meaning of variable 'intaketotal in the eICU database under table intakeOutput? Is it the accumulative value of fluid intake since the admission time (regardless of output)? If so, I would expect this value to be monotonic increasing for a patient's ICU stay. However, I found that this value can also be decreased for a patient. (i.e. patientunitstayid = 141194)

nickfong commented 3 years ago

Hi @daisy647lsq,

I’m going to preface this by saying that I’m in a similar boat to you and I’ve spent quite a bit of time trying to figure out the nuances of the data, so this is all based on what I’ve been able to glean from looking at the data for the past couple months.

The intaketotal column in the intakeoutput table is the total of fluid intake for a given charting event/offset. Stepping back and looking at the clinical flow at play here, the bedside nurse records ins and outs for the patient at some time. Each fluid, drain, catheter, etc. is charted separately, but the data is saved in the EMR all at once. For example, our patient might have infusions of NS and epi running at the same time and a urinary catheter and bilateral chest tubes. The bedside nurse enters the volume of NS infused, the volume of epi infused, and the volume drained from the catheter and each of the respective chest tubes in the EMR, then clicks save. Consequently, all the I/O events charted at this time is given the same timestamp.

Let’s look at how this looks like for a patient. Using your example (patientunitstayid == 141194), at intakeoutputoffset == 282 the pt.’s weight was recorded along with some volume of NS, dexmedetomidine, d5 0.45 NaCl, and Zosyn. All these rows of the table have the same timestamp, and the same intaketotal. If we add up the volume of all the infusions charted at time 282 (looking at the cellvaluenumeric column), we find that we get 1000+32+42+100 = 1174, which is the intaketotal we see for each of the rows. In short, intaketotal is the total number of CCs the patient gained at the given charting time.

We can sort of sanity check ourselves here by looking at another offset for the same patient. Some time later, at intakeoutputoffset == 462, we show the pt. received 500 CCs of NS. At the earlier tilmestep, we saw the pt. got 1000 CCs of NS. But we note that the only input at intakeoutputoffset == 462 is the NS infusion, so intaketotal at 462 corresponds solely to the 500 CCs that the patient just received; the number in this column at this offset doesn’t tell us anything about the 1000 CCs the pt. got earlier. Also note that this value is reflected in the nephrostomy outputs from the same timestamp, and that the sum of the outputs from that timestamp are reflected in outputtotal for our NS input data.

If you have any further questions or if something I said doesn't make sense, please let me know and I'd be happy to try to clarify.