I am working with the Allergy dataset which has an attribute Allergy offset.

[lakshikajain]

I would like to validate if the negative values in this means that the drug allergy was already reported by the patient before he was admitted to ICU and if it was specifically after that drug administered. Can I get a little more insight on this attribute? Also, I would like to know at what frequency is the fresh patient data updated to the database? Any help would be appreciated. Thanks.

[jraffa]

Hi lakshikajain,

It would be useful to know what you're trying to do in your study, because there may be other factors you want to consider.

There are actually two offsets in this table, allergyoffset and allergyEnteredOffset. There is a little complexity to some of the offsets, and I haven't come across these two yet. From the looks of it, allergyEnteredOffset never occurs before allergyoffset, which I think makes sense. The documentation distinguishes between entry (allergyEnteredOffset) and detection (allergyoffset).

Negative offsets indicate times which occurred before the start of that specific patienitunitstayid. It may indicate that it was picked up in the emergency room or other area of the hospital, or maybe even from a previous patientunitstayid. I would not assume that all entries in this table correspond with adverse reactions to drugs in the ICU, and are probably more likely to be entered in from patient history (e.g., penicillin allergy).

I would encourage you to have a look at the documentation:

http://eicu-crd.mit.edu/eicutables/allergy/

Keep in mind there's a great deal of variation of what is entered in different hospitals, and this may even change over time, as hospitals upgrade or change EHR systems, etc.

I'm not sure what you mean by "fresh patient data updated to database". We are currently working on releasing more tables publicly for this release of the eICU-CRD. If you mean patient data at the hospital, this varies quite a bit, by type of data, hospital, year, etc.

If you give me a little more detail, and let me know what questions are still pressing, I can ask some of the people involved in the full eICU database if they know anything further.

Jesse

[lakshikajain]

Hi Jesse, Thanks for your Insight. I am working on a project where a part is to analyze the frequency of patients reporting one drug allergy and when was this allergy reported. Like, if patients suffered with a specific drug allergy after they have been administered that drug while he stayed in hospital. With the "Fresh patient data updates" I meant, as the data shows 2015 as the latest allergydate, I was curious when this gets updated to the most recent reported allergies (I.e., for allergy date being 2016 or later). Thanks for your help, I really appreciate it. Lakshika

[jraffa]

Hi Lakshika,

It may be tricky to estimate drug allergy prevalence from the allergy table alone. I will emphasize what is said in the documentation:

The absence of observation does not indicate an absence of allergy. and Allergy entry is optional.

This varies a lot by hospital even:

As for "fresh data", as you note, the current release offers admissions from 2014 and 2015. I don't anticipate updating it with more recent data (2016-) in the short term. The process to release data is quite involved, and we need to focus on improving what is currently available, or will be available shortly (e.g., we have a few more tables from 2014-15 to release).

Jesse

[lakshikajain]

Thanks for the valuable information. This helped me to a great extent.