Open kaveriC opened 2 weeks ago
Agree with the above
@cloverbunny - revisit post-ATS
@cloverbunny can we agree to drop lab_type_name
from the data dictionary and leave the other items as "to-dos" post ATS?
I think we also need to have lab_order_name
and lab_order_category
in order to distinguish blood and urine labs (e.g. creatinine
)
@cloverbunny looking through all the "Proc_names" we currently have mapped to lab_order_name
I think categorizing all the lab orders (e.g. CBC w/diff, CMP, etc.) is something we should do post-ATS.
But maybe we should replace lab_type_name
with lab_specimen_category
which will have levels c(blood/plasma/serum, urine, csf, other)
.
There will be a mapping of lab_order_name
to lab_specimen_category
. The point of this is so you can quickly filter down to just blood tests and avoid using a urinalysis creatinine or a CSF WBC in a prediction model
I know we (NU) did pre-CLIF filtering out of urine and CSF labs of things like protein, and think others probably also did? We can include a lab_specimen_category to make this more explicit, but think this can be done post-ATS? Where were the examples again of what is lab_type_name vs lab_order_name?
lab_type_name
was never well defined and lab_order_name
is the actual "procedure name", e.g. "URINALYSIS CHEMISTRY SCREEN".
Overall I think we should try to make CLIF LOINC compatible https://loinc.org/get-started/loinc-term-basics/, which is why I'm suggesting the lab_specimen_category
to correspond to the "system(specimen)" part of the LOINC code which represents "The specimen or thing upon which the observation was made"
lab_name
and lab_category
correspond to LOINC's "component (analyte)", defined as "The substance or entity being measured or observed."
So an example would be like: If we drop lab_type_name, how else would we define standard vs POC? Isn't that what it was originally intended for? lab_order_name: total_bilirubin lab_order_category: blood/plasma/serum, urine, csf, other [do we have a list of all categories we want such as tissue, synovial, etc?] I think we also need to think of some better way of naming these as it's confusing what's lab_order and what's lab_category heh.
Reference from website of current format Variable Name | Data Type | Definition | Permissible Values |
---|---|---|---|
hospitalization_id | VARCHAR | ID variable for each patient encounter. | No restriction |
lab_order_dttm | DATETIME | Date and time when the lab is ordered. | Datetime format should be %Y-%m-%d %H:%M:%S |
lab_collect_dttm | DATETIME | Date and time when the specimen is collected. | Datetime format should be %Y-%m-%d %H:%M:%S |
lab_result_dttm | DATETIME | Date and time when the lab results are available. | Datetime format should be %Y-%m-%d %H:%M:%S |
lab_order_name | VARCHAR | Procedure name for the lab. This field is not used for analysis. | No restriction |
lab_order_category | VARCHAR | Permissible CDE values for lab order names. | CDE under development |
lab_name | VARCHAR | Original lab name string recorded in the raw data. This field is not used for analysis. | No restriction |
lab_category | VARCHAR | Limited labs identified by the CLIF consortium as minimum necessary labs for the study of critical illness. | List of lab categories in CLIF |
lab_value | VARCHAR | Recorded value corresponding to a lab. Lab values are often strings that can contain non-numeric results (e.g. “> upper limit of detection”). | No restriction |
lab_value_numeric | DOUBLE | Parse out numeric part of the lab_value variable. | Numeric |
reference_unit | VARCHAR | Unit of measurement for that lab. | Permissible reference values for each lab_category listed here |
lab_type_name | VARCHAR | Type of lab. (e.g. standard, POC) | No restrictions |
lab_loinc_code | VARCHAR | LOINC code corresponding to the lab_name. | No restrictions |
Variable Name Data Type Definition Permissible Values hospitalization_id VARCHAR ID variable for each patient encounter. No restriction lab_order_dttm DATETIME Date and time when the lab is ordered. Datetime format should be %Y-%m-%d %H:%M:%S lab_collect_dttm DATETIME Date and time when the specimen is collected. Datetime format should be %Y-%m-%d %H:%M:%S lab_result_dttm DATETIME Date and time when the lab results are available. Datetime format should be %Y-%m-%d %H:%M:%S lab_order_name VARCHAR Procedure name for the lab. This field is not used for analysis. No restriction lab_order_category VARCHAR Permissible CDE values for lab order names. CDE under development lab_name VARCHAR Original lab name string recorded in the raw data. This field is not used for analysis. No restriction lab_category VARCHAR Limited labs identified by the CLIF consortium as minimum necessary labs for the study of critical illness. List of lab categories in CLIF lab_value VARCHAR Recorded value corresponding to a lab. Lab values are often strings that can contain non-numeric results (e.g. “> upper limit of detection”). No restriction lab_value_numeric DOUBLE Parse out numeric part of the lab_value variable. Numeric reference_unit VARCHAR Unit of measurement for that lab. Permissible reference values for each lab_category listed here lab_type_name VARCHAR Type of lab. (e.g. standard, POC) No restrictions lab_loinc_code VARCHAR LOINC code corresponding to the lab_name. No restrictions
Not required for ATS (only map blood to lab categories):
@08wparker Add lab_speciman_name
to the data-dictionary.
While setting up the CLIF 2.0 labs table, @08wparker and I would like to propose the following updates:
lab_category
:RDW
,Beta-hydroxybutyrate
, andLipase
.lab_order_category
and include relevant labs under lab_category, such aspH_urine
andcreatinine_urine
.lab_type_name
variable, as it has become redundant in the CLIF 2.0 version after adding lab_order and lab_order_category.