“apachePredVar” dataset variable “bedcount”

[leonar994]

In the dataset called “apachePredVar” the variable called “bedcount” includes the number of ICU beds in the unit the patient was admitted or the total number of ICU beds in the hospital that the patient was admitted?

For example, if a hospital has 3 ICUs with 10, 20, and 30 beds, respectively, “bedcount” will be 10/20/30 or “bedcount” will be 60?

Thank you in advance for your time and help. Looking forward to hearing back from you.

[alistairewj]

I've never actually used the bedcount column. But this is the type of thing you can get an idea about by cross-referencing tables in the dataset. If you get the wardid from the patient table, then you can compare the bed count across wards in the same hospital:

SELECT 
pt.hospitalid
, pt.wardid
, pt.hospitaldischargeyear
, apv.bedcount
, count(*)
FROM `physionet-data.eicu_crd.apachepredvar` apv
inner join `physionet-data.eicu_crd.patient` pt
on apv.patientunitstayid = pt.patientunitstayid
group by 1, 2, 3, 4
order by 1, 2, 3, 4

hospitalid	wardid	bedcount	count
56	82	8	236
58	108	9	252
59	91	12	575

etc.

One hospital, hospitalid = 73, has quite a few rows:

hospitalid	wardid	hospitaldischargeyear	bedcount	count
73	85	2014	38	2
73	85	2015	28	206
73	85	2015	56	1
73	89	2014	17	6
73	89	2015	17	208
73	89	2015	22	4
73	89	2015	26	1
73	92	2014	38	519
73	92	2014	76	1
73	92	2015	38	644

So it looks to be ICU specific. It also changes over time but is consistent for a large number of patients, which implies it's a count of the total number of beds.

Note you can also get the number of beds in the hospital using numbedscategory in the hospital table. hospitalid = 73 has >= 500 beds in the hospital.