Review the logic of mixing planned/unplanned ICU admission

[docsteveharris]

https://github.com/tharusan/snap2_final/blob/c23bbf0d922fbb363c7d1097f9cb3a430a0bf451/snap2_syntax1.Rmd#L294

This may be OK but we should just consider if we want to treat a patient who is admitted to the ICU without booking in our 'intervention/treatment' arm. Won't these be patients who have some form of intra-operative complication?

I wonder if we the target trial question might be a comparison of

direct and planned ICU admission following surgery vs others

so you would not be in the treatment group if you were admitted without a booking

I think this is important because these unexpected admissions will 'bump' planned admissions under times of strain and therefore should remain in the analysis but on the 'control' arm.

Thoughts?

[docsteveharris]

Kicking this around with Tarush

• define population at the end of surgery :

• no intraoperative disaster (i.e unexpected) and carrying a booking request and admitted or just

• booking request and admitted

in the former we exclude all intra-op deteriorations in the latter intra-op deteriorations to remain in the control group (b/c there was no choice, they had to come)

[dannyjnwong]

I think going back to the protocol paper is helpful. In that we refer to "planned postoperative critical care". I agree that if the patient met with intraoperative disaster, then unplanned ICU admission is no longer a choice. We should therefore focus on the intervention being "planned postoperative critical care".