Casus
title: CASUS (Cardiac Surgery Score) tags: #FFICM notebook: 🌑-FFICM
So who is the CASUS score (the cardiac surgery score, for ICU mortality prediction) suitable for? It's post-op cardiac surgery patients. It was made in 2005 What's different about cardiac ICU scoring vs other ICU patients? Cardiac patients can look better than they are because the balloon pump or other MCS can give them better numbers.
There are two models in CASUS:
- Additive
- This is an add the factors together to get a certain score.
- Then this score corresponds to a group of risk %
- Logistic (log-CASUS)
- This is a adjust the value by the coefficient for each factor to get your specific mortality prediction %
- More useful for long term patients as takes days of stay into account
In terms of validation most recently:
log-CASUS
CASUS depends on:
- Pressure Adjusted Heart Rate (PAR)
- PAR = HR * CVP / MAP
- Lactate
- IABP (Y/N)
- VAD (Y/N)
- Creatinine (micromol/L)
- Neuro State
- Adequate/Coperative/Confused/Sedated/Neuropathy
- PaO2/FiO2
- Serum Bilirubin
- Platelets
Model
℮ (β0 + ∑βi * Xi) / (1 + ℮ (β0 + ∑βi * Xi))
Variables (Coefficients)
| Descriptor | \(\beta\) coefficient |
|---|---|
| PF ratio (mmHg/%) | -0.0013498 |
| Serum Creatinine (mg/dl) | 0.3063634 |
| CVVH/Dialysis - No | 0 |
| CVVH/Dialysis - Yes | 0.4109223 |
| Serum Bilirubin (mg/dl) | 0.2214784 |
| PAR (HR*CVP/MAP) | 0.0642751 |
| Lactic acid (mmol/L) | 0.2115760 |
| IABP - No | 0 |
| IABP - Yes | 0.6770239 |
| VAD - No | 0 |
| VAD - Yes | 2.2976240 |
| Platelets *10^3/microL | -0.0042251 |
| Neurological State - Normal | 0 |
| Neurological State - Confused | 0.4736275 |
| Neurological State - Sedated | 0.7020852 |
| Neurological State - Diffuse Neuropathy | 1.4715528 |
| ICU Day - 1 | 0 |
| ICU Day - 2 | 0.0097085 |
| ICU Day - 3 | 0.8372058 |
| ICU Day - 4 | 1.0426010 |
| ICU Day - 5 | 1.2561380 |
| ICU Day - 6 | 1.4975238 |
| ICU Day - 7 | 1.6260023 |
| ICU Day - 8 | 1.9023001 |
| ICU Day - 9 | 2.1064412 |
| ICU Day - 10 | 2.2268852 |
| ICU Day - 11 | 2.2597632 |
| ICU Day - 12 | 2.3782868 |
| ICU Day - 13 | 2.5289064 |
| Constant | -5.6412079 |
What are my thoughts on this
The logistic does seem a step up from the additive, in that it gives us a way to take into account the length of stay.
What it does seem to do though, is treat each day fresh. So doesn't matter what your score was yesterday.
I wonder if adding in the previous day's score would be useful or not? There's probably some statistical badness in doing that maybe, as you're using the same factors twice in the same model. But then maybe it'd actually be fine, cos it's not really the same factors. Instead it's a different days.
What would we ideally need from CIS?
- patient ID (preferably CHI so can link w/ wardwatcher)
- post-op cardiac surgery
- I don't know if there's a value for this on CIS
- On wardwatcher you could get this list through:
- Surgery this admn, and under care of cardiac surgeons
- Or in the admission diagnosis
- date of admission
- for each date in admission:
- date
- PaO2
- FiO2
- Creatinine
- RRT (Y/N)
- Bilirubin
- Lactate (is on ABGs)
- IABP (Y/N)
- VAD (Y/N)
- Platelets
- HR
- CVP
- MAP
- GCS
- Sedative Infusions
- ?If we have a CAM-ICU/AVPU
- ICU Outcome
If we have a straightforward way of getting:
- Hospital Outcome
- 90 Day Mortality