Chemotherapy drugs

[loberste]

Dear MIMIC community,

First of all, thanks to the publishers for creating & offering such a great dataset! As part of my master thesis I am investigating intravenous (IV) chemotherapies. I hope that I can find corresponding drugs and durations of their inpatient administration in the MIMIC data. Unfortunately, I'm a little confused and would be very pleased about any help to the following questions and tables, respectively:

• In PROCEDURES_ICD, the code 9925 is referenced by 529 admission IDs which indicates that there should be at least correspondingly many IV chemotherapy administrations present in the data
• But when searching INPUTEVENTS_MV for commonly, continously (non-bolus) administered items, general fluids (painkillers, volume expanders) were found but almost no chemo drugs:
  • (neglecting the dose change/new dosage distinction)
  • select count(a.itemid) as ct, a.itemid, b.label from inputevents_mv as a inner join D_ITEMS as b on a.itemid = b.itemid where hadm_id in (select hadm_id from procedures_icd where icd9_code = '9925') and endtime-starttime <> '1 minute' group by a.itemid, b.label order by ct desc
  • Example: Octreotide (an anti-tumor drug) amounts to only 4 times, which is surprising because in the whole inputevents_mv table, there are 2929 occurences

Excerpt:

count	itemid	label
11265	225158	NaCl 0.9%
4528	220949	Dextrose 5%
2901	225943	Solution
		...

So, ultimatively:

• Would you think that I overlooked certain chemotherapy data in the database?
  • Codes 96401-96549 in CPTEEVENTS as well as codes 6931-6934 in DRGCODES are only referenced by 100 admission IDs, so thats probably not the main part
• Is there any chance of identifying the particular drugs and start- & endtimes of their administration?

Thanks in advance for any help!

[alistairewj]

In PROCEDURES_ICD, the code 9925 is referenced by 529 admission IDs which indicates that there should be at least correspondingly many IV chemotherapy administrations present in the data

Not necessarily; the treatments may be given outside of the ICU, in which case we wouldn't have them in the ICU information system derived inputevents_* tables. But we would expect them in the prescriptions table, which covers the entire hospitalization.

[loberste]

Not necessarily; the treatments may be given outside of the ICU, in which case we wouldn't have them in the ICU information system derived inputevents_* tables. But we would expect them in the prescriptions table, which covers the entire hospitalization.

Dear Alistair, thanks for the informative and quick answer! Sorry for re-enquiring: It makes me wonder if there is another chance to infer the duration of the treatment (outside the ICU, so probably on a chemotherapy chair or in an inpatient setting)?

If the chemotherapy is administered on a chemotherapy chair, could I identify it using the TRANSFERS table? For instance, when a patient is transferred to a non-ICU ward and stays there shortly, in which period the prescription table shows an antitumor drug? But I guess a treatment room would not be listed as a ward? Aparently, there is no oncological service either.

If the chemotherapy is administered to a regular inpatient, there wouldn't be any definite start/end time of administration, would there?

Thanks a million for any thoughts!

[alistairewj]

I would see how accurate the start and end dates are in the prescriptions table. I don't think you can do much linking to transfers. Pretty challenging to study this in MIMIC unfortunately!

[loberste]

@alistairewj Thank you very much, highly appreciated! I will investigate this last option and close the "issue" :)