Get the order of hospital admission

[Yannis-ys]

For patients with multiple ICU admission, "unitvisitnumber" can help us identify the time order of each ICU admission. But for patients with multiple hospital admission, how can we know the order of each hospital admission? If the different hospital admission occurs in different year or have different outcome ("death" is the last admission), "age", "admission year" and "discharge status" may help. But if multiple admissions occur in the same year and all discharge alive, how can we know the order? Are the "patienthealthsystemstayid" and "patientunitstayid" associated with the admission time, which means latter admission has larger number of stayid ? Thanks !

[jraffa]

I wouldn't personally read too much into the order of the ids. Unfortunately we don't have and likely will not get this admission order information. You mention many of the strategies others have used, but you may still find these useful, https://github.com/MIT-LCP/eicu-code/issues/16 and https://github.com/nus-mornin-lab/oxygenation_kc/pull/23 and https://github.com/MIT-LCP/eicu-code/issues/123

As an aside, I know it is often preferred to have the first admission, but the fact remains that any retrospectively collected database is not able to definitively determine the precise order of ICU admissions when multiple hospital systems are involved. Even in MIMIC, where the order is thought to be known, what we really know are the order of admissions within BIDMC ICUs. If the patient is admitted to another Boston hospital it would not be apparent through MIMIC. I say this because because I think people think this problem isn't present in other databases, but I think most will suffer from it in some form.

[Yannis-ys]

Thank you for you rapid and detailed reply. Recently, I have read an article based on MIMIC and eICU, which has mentioned "Only data of the first ICU admission of the first hospitalization were used." in the Methods part.[1] But it seems hard to achieve this screen accurately in eICU. (Maybe they achieved this goal by deleting all patients with multiple hospital admissions...) By the way, I have look into the ids, it is not reliable to take the order of ids as the order of hospital admissions. Thank you Jraffa ^_^ ! [1] Serpa Neto A, Deliberato R O, Johnson A E W, et al. Mechanical Power of Ventilation Is Associated with Mortality in Critically Ill Patients: An Analysis of Patients in Two Observational Cohorts[J]. Intensive Care Medicine, 2018, 44(11): 1914–1922.