MIMIC-IV KDIGO AKI Stage 3 is wrongly defined

[starmpcc]

Hello.

https://github.com/MIT-LCP/mimic-code/blob/14b9f5f7c622f4535dd6da88969de3e2f2460904/mimic-iv/concepts/organfailure/kdigo_stages.sql#L22

In KDIGO AKI criteria(link, page 21), AKI Stage 3 is defined by satisfying both of

1. sCR >=4.0
2. sCR increment >=0.3 within 48h or sCR baseline * 1.5

However, the logic of above code was defined by fixed number 3.7. I think this it was defined wrongly, so should be changed onto

and (cr.creat >= (cr.creat_low_past_48hr+0.3) OR cr.creat >= (1.5*cr.creat_low_past_7day))

Thanks you

[alistairewj]

A bit more context gives a broader view of the approach: https://github.com/MIT-LCP/mimic-code/blob/14b9f5f7c622f4535dd6da88969de3e2f2460904/mimic-iv/concepts/organfailure/kdigo_stages.sql#L14-L23

With that I can see what you mean and that makes sense, this would not catch AKI for folks with an initial creatinine of 4.0 or higher. I'll look to fix it soon, probably worth doing a before/after comparison.

[starmpcc]

@alistairewj Thank you for your response! Note that there is also a same issue on MIMIC-III.

https://github.com/MIT-LCP/mimic-code/blob/2068ccad151bf492e6cc4db4bf222c82cb6b1bbe/mimic-iii/concepts/organfailure/kdigo_stages.sql#L14-L22

[arielhasidim]

I have corrected the stage 3 serum creatinine definition in this fork. Moreover, I have added staging logic for RRT initiation and also preferred the use of baseline creatinine over the last creatinine in 7 days if applicable.

I can PR this if you would like @alistairewj. If the added changes are not wanted I can also remove them and PR it afterward.