To detect whether vasopressor was used for a patient in MIMIC, is looking at the presence of itemids whose labels are dobutamine, dopamine, epinephrine, epinephrine-k, levophed, levophed-k, vasopressin, milrinone, neosynephrine, neosynephrine-k in d_items enough? (case insensitive, as done by https://github.com/MIT-LCP/hdlvef/blob/master/sql/echo_vasopressors.sql)
or shall one be more exhaustive like:
SELECT d_items.label, MIN(d_items.itemid) item_id, count(distinct(hadm_id)) hadm_id_count
FROM mimiciii.d_items, mimiciii.chartevents
WHERE chartevents.itemid = d_items.itemid
AND (
upper(label) like '%DOBUTAMINE%'
or upper(label) like '%DOPAMINE%'
or upper(label) like '%EPINEPHRINE%'
or upper(label) like '%LEVOPHED%'
or upper(label) like '%VASOPRESSIN%'
or upper(label) like '%MILRINONE%'
or upper(label) like '%NEOSYNEPHRINE%')
group by d_items.label
ORDER BY label ASC
;
and
SELECT d_items.label, MIN(d_items.itemid) item_id, count(distinct(hadm_id)) hadm_id_count
FROM mimiciii.d_items, mimiciii.ioevents
WHERE ioevents.itemid = d_items.itemid
AND (
upper(label) like '%DOBUTAMINE%'
or upper(label) like '%DOPAMINE%'
or upper(label) like '%EPINEPHRINE%'
or upper(label) like '%LEVOPHED%'
or upper(label) like '%VASOPRESSIN%'
or upper(label) like '%MILRINONE%'
or upper(label) like '%NEOSYNEPHRINE%')
group by d_items.label
ORDER BY label ASC
;
which returns many more though not widely used itemids: