2

I would like to get some external verification to see if I am correct.

I originally used the MIMIC II (2v26) database and I looked for fluids based on category in the d_ioitems table. For example:

MIMIC2=# select * from mimic2v26.d_ioitems where lower(category) like 'colloid%';
 itemid |         label         | category
--------+-----------------------+----------
   2036 | Sterile Water 150.0ml | Colloids
   1894 | Dextran 70            | Colloids
   1438 | Hespan 1000.0ml       | Colloids
   2688 | Albumin 25% 50.0ml    | Colloids
(4 rows)

But it seems that categories such as 'Colloids', 'Fluid Bolus', 'IV Infusions','IV Drips' are no longer present in MIMICIII (v1.4):

MariaDB [mimiciii]> select distinct category from mimiciii.D_ITEMS where 
category in ('IV Infusions','IV Drips','Colloids','Free Form Intake','Fluid Bolus');
+------------------+
| category         |
+------------------+
| Free Form Intake |
+------------------+
1 row in set (0.00 sec)

And that 'Free Form Intake' is what replaces the categories that I'm looking for.

Is there something I am missing? I haven't been able to find any documentation on this change.

If these category names no longer exist, then I will likely need to filter for specific 'label' names that are indicative of the categories I'm looking for.

1

At least for colloids, the category became null when going from MIMICII to MIMICIII.

In addition to some colloids increasing in their itemid (for example Hespan, itemid 6313 to 46313), there was also merging of itemids. Using hespan as an example, MIMICII had multiple labels (itemids) for hespan in different categories:

MIMIC2=# select category, label, d.itemid, count(d.itemid)
MIMIC2-# from mimic2v26.ioevents io
MIMIC2-# join mimic2v26.d_ioitems d on io.itemid=d.itemid
MIMIC2-# where lower(label) like '%hespan%'
MIMIC2-# group by d.itemid, label, category
MIMIC2-# order by itemid;
     category     |      label      | itemid | count
------------------+-----------------+--------+-------
 IV Infusions     | Hespan          |    246 |  2019
 Free Form Intake | hespan          |    623 |    11
 IV Infusions     | Hespan 500.0ml  |    662 |    61
 Colloids         | Hespan 1000.0ml |   1438 |    15
 Free Form Intake | 6% Hespan       |   6313 |     5
(5 rows)

But upon further investigation with specific patients, itemids 246, 623, 662, 1438 were all combined into 6313, and in MIMICIII, all became 30012, with 6313 becoming 46313 as expected:

MariaDB [mimiciii]> select category, label, d.itemid, count(d.itemid)
    -> from mimiciii.INPUTEVENTS_CV io
    -> join mimiciii.D_ITEMS d on d.itemid = io.itemid
    -> where lower(label) like '%hespan%'
    -> group by d.itemid, label, category;
+------------------+-----------+--------+-----------------+
| category         | label     | itemid | count(d.itemid) |
+------------------+-----------+--------+-----------------+
| NULL             | Hespan    |  30012 |            2150 |
| Free Form Intake | 6% Hespan |  46313 |               5 |
+------------------+-----------+--------+-----------------+
2 rows in set (0.02 sec)
| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.