ehrQL tutorial: Working with electronic health record codes and codelists🔗

Danger

This page discusses the new OpenSAFELY Data Builder for accessing OpenSAFELY data sources.

Use OpenSAFELY cohort-extractor, unless you are specifically involved in the development or testing of Data Builder.

OpenSAFELY Data Builder and its documentation are still undergoing extensive development. We will announce when Data Builder is ready for general use on the Platform News page.

Electronic health record codes🔗

Electronic health record (EHR) codes are used when recording patient events. For research, codes are often collated into codelists to classify patients. Data Builder can use codelists to help extract information about patient groups of interest.

Info

There is more information about codelists elsewhere in OpenSAFELY's documentation. We also have a codelist builder tool, OpenCodelists, that you may find useful when creating codelists.

Example dataset definition 7: Codes and codelists🔗

Warning

This is a rough draft. It is intended as a placeholder so that the tutorial flows from start to finish.

Warning

Codelists are not yet fully implemented. See https://github.com/opensafely-core/databuilder/issues/31

Todo

Complete this tutorial page. Write an example that suitably explains enough for the user testing scenario. This is pending the use of codelists there being finalised.

Loading a codelist🔗

Todo

Remove the placeholder example and replace with a real dataset definition.

You can load codelists in a comma-separated values (CSV) format with Data Builder's codelist_from_csv() function. The CSV must have named columns.

You need to specify, as strings:

the CSV filename
the name of the CSV column containing the codes

codelist_from_csv()

from databuilder.codes import codelist_from_csv

codelist = codelist_from_csv(filename="my_codelist.csv", column="code")

Checking if a code is in a codelist🔗

Todo

Remove the placeholder example and replace with a real dataset definition.

Warning

This example may not be correct.

You can use isin and isnotin to check for specific codes.

Dataset definition

from databuilder.codes import SNOMEDCTCode
from databuilder.ehrql import Dataset, codelist_from_csv
from databuilder.tables.example.tutorial import clinical_events

codelist = codelist_from_csv(filename="my_codelist.csv", column="code")
a1_code = SNOMEDCTCode("a1")
a2_code = SNOMEDCTCode("a2")

codelist_filtered_events = clinical_events.where(clinical_events.system == "snomed").code.is_in(codelist))
code_filtered_events = clinical_events.where(clinical.events.system == "snomed").code.is_in([a1_code, a2_code])

dataset = Dataset()
dataset.define_population(codelist_filtered_events.exists_for_patient() | code_filtered_events.exists_for_patient())

Todo

Explain that […] creates a sequence of ordered items termed a list in Python, if not mentioned elsewhere.

We also instantiate code objects the same way we do for Dataset().

Tutorial questions🔗

Question

…