MORTALITY REVIEW AND POST-ICU RISK ============================================================================ Ideas 80 and 90. Assembling the timeline so the meeting discusses the case, and identifying who will carry harm home.
WHY THESE TWO SIT TOGETHER -------------------------- Both are retrospective reads over the same admission record, run at the two moments an ICU stay ends. One asks what happened; the other asks what follows. Sharing the extraction keeps them consistent about the same admission.
WHAT AN M&M TOOL MUST NOT DO ---------------------------- It must not adjudicate. The moment software suggests a death was avoidable, two things happen: the meeting argues with the tool instead of about the case, and people stop recording candidly.
So this engine assembles and orders. It states what happened and when, marks where the record is silent, and offers no opinion on preventability. The judgement is the meeting's work, and it is the only part that matters.
THE SILENCES ARE THE FINDING ---------------------------- In most reviews the useful discovery is not an error in the record. It is a gap — six hours with no observation, a deterioration nobody documented responding to, a result reported and never actioned. Those are visible only when the timeline is assembled, which is why assembly is worth automating even though it produces no conclusions.
PICS IS INVISIBLE IN MORTALITY STATISTICS ----------------------------------------- Post-intensive-care syndrome is cognitive, physical, and psychological harm carried home by survivors. A unit judged on mortality has no reason to look for it, which is precisely why it goes unmeasured.
Sources: standard M&M review structure; published PICS risk factors. Every factor must be checked against current evidence before use.
meta
version0.1.0
ownerUNASSIGNED — set to the responsible clinician
verified bynot set
boundary
Assembles and orders. Does not adjudicate preventability, does not assign
responsibility, and offers no opinion on whether a death was avoidable.
why no adjudication
Software that suggests a death was avoidable turns the meeting into an
argument with the tool and stops people recording candidly. Both destroy
the review.
event classes
- idadmission
labelAdmission
prominence3
- iddeterioration
labelPhysiological deterioration
prominence5
noteAnchor points. Everything else is read relative to these.
- idintervention
labelIntervention
prominence4
- idescalation
labelEscalation of care
prominence5
- iddecision
labelDocumented decision
prominence5
note
Including decisions not to escalate. A recorded decision not to act is
a decision and belongs on the timeline.
- idresult
labelResult available
prominence2
- idresult_actioned
labelResult acted upon
prominence4
- idtransfer
labelTransfer
prominence3
- idfamily
labelFamily discussion
prominence3
- iddeath
labelDeath
prominence5
gap types
- idobservation_gap
labelNo observations recorded
threshold hours4
questionWhat was happening during this period?
noteMay be entirely appropriate. The timeline cannot tell.
- idunactioned_result
labelResult available, no documented response
threshold hours6
questionWas this seen?
note
The commonest finding in any review, and usually a systems question
rather than an individual one.
- iddeterioration_no_response
labelDeterioration with no documented intervention
threshold hours2
questionWhat was the assessment at the time?
note
Absence from the record is not absence of care. Busy periods document
worst.
- idescalation_delay
labelInterval between deterioration and escalation
threshold hours3
questionWhat was the sequence here?
- idno_family_contact
labelNo family discussion recorded
threshold hours24
questionWas the family updated?
pics
domains
cognitive
labelCognitive impairment
factors
- iddelirium_days
factorDays of delirium
per unit2
unitday
cap10
note
Consistently among the strongest predictors reported, and duration
appears to matter more than occurrence. Stated as a weighting for a
panel to revise, not as a settled finding.
- idhypoglycaemia
factorAny severe hypoglycaemic episode
weight3
- idhypoxaemia
factorProlonged severe hypoxaemia
weight3
- idsedation_days
factorDays of deep sedation
per unit1
unitday
cap7
- idage_over_65
factorAge over 65
weight2
follow upCognitive screening at 3 months
physical
labelPhysical impairment
factors
- idventilation_days
factorDays of mechanical ventilation
per unit1
unitday
cap14
- idimmobility_days
factorDays without mobilisation
per unit2
unitday
cap10
- idsteroids
factorCorticosteroid exposure
weight2
- idneuromuscular_blockade
factorNeuromuscular blocking agents
weight3
- idprotein_deficit
factorSustained protein deficit
weight2
- idsepsis
factorSepsis
weight3
follow upPhysiotherapy assessment before discharge, and at 6 weeks
psychological
labelPsychological harm
factors
- iddelusional_memory
factorDelusional memories or ICU-related hallucinations
weight4
note
Repeatedly associated with later PTSD, and the one nobody asks
about because the patient rarely volunteers it. The association is
well described; calling it the strongest single predictor would
overstate what the literature settles.
- idrestraints
factorPhysical restraint
weight3
- idprior_mental_health
factorPre-existing mental health condition
weight3
- idyoung_age
factorAge under 50
weight1
noteYounger survivors report more psychological harm, not less.
- idemergency_admission
factorUnplanned admission
weight1
- idprolonged_stay
factorICU stay over 7 days
weight2
follow upScreening for anxiety, depression, and PTSD at 3 months
bands
- max4
bandlow
- max9
bandmoderate
- max999
bandhigh
note
Identifies who to follow up, not what will happen. A high score is a
reason to arrange review, not a prognosis to give the patient.
family note
Family members of ICU survivors carry measurable psychological harm of
their own. It is out of this engine's scope and is named here so its
absence is deliberate rather than an oversight.