You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am having trouble understanding how I should interpret the deathtime in the admissions table. For example, hadm_id = 28146972 shows that the patient died 2151-09-03 12:53:00. However, in the transfers table the patient is still in the MICU well after that time (until 2151-09-05 20:07:05) and is even being transferred to the PACU. I assume that the patient really died at 12:53, but why is he still shown as being in the ICU?
deathtime is from the hospital death record; transfers/outtime can lag or keep moving after that (documentation delay, organ care, disposition cleanup). treat deathtime as source of truth for death and don't expect transfers to stop exactly then.
opened #2152 — deathtime is the admission mortality stamp; later transfers are usually ADT lag / post-mortem logistics, not a join bug. prefer hospital_expire_flag/deathtime for mortality, and don't require deathtime >= last transfer outtime.
Activity
chessing234 commentedon Jul 19, 2026
deathtime is from the hospital death record; transfers/outtime can lag or keep moving after that (documentation delay, organ care, disposition cleanup). treat deathtime as source of truth for death and don't expect transfers to stop exactly then.
chessing234 commentedon Jul 31, 2026
opened #2152 — deathtime is the admission mortality stamp; later transfers are usually ADT lag / post-mortem logistics, not a join bug. prefer hospital_expire_flag/deathtime for mortality, and don't require deathtime >= last transfer outtime.
docs(mimic-iv): explain admissions.deathtime vs later transfers
docs(mimic-iv): link README(s) to deathtime interpretation notes
docs(mimic-iv): ensure documentation helper-query folder exists
docs(mimic-iv): add transfers-after-deathtime audit query
docs(mimic-iv): sample query for ICU outtime after deathtime
docs(mimic-iv): clarify we will not clip transfers to deathtime in ETL