move to dedicated folder and tidy

analysis/codelists.py → analysis/PRIMIS/codelists.py

@@ -1,33 +1,13 @@
-# Purpose:
-# define codelist objects from codelist files imported by codelist.txt spec
-
-# Import code building blocks from cohort extractor package
+# Import code building blocks from ehrql package
 from ehrql import codelist_from_csv
 
-
-## --VARIABLES--
-# if the variable uses a codelist then it should be added below
-# after updating the codelist.txt configuration and importing the codelist
-
-# Ethnicity
-
-ethnicity_codelist5 = codelist_from_csv(
-  "codelists/opensafely-ethnicity-snomed-0removed.csv",
-  column="code",
-  category_column="Label_6", # it's 6 because there is an additional "6 - Not stated" but this is not represented in SNOMED, instead corresponding to no ethnicity code
-)
-
-ethnicity_codelist16 = codelist_from_csv(
-  "codelists/opensafely-ethnicity-snomed-0removed.csv",
-  column="code",
-  category_column="Label_16",
-)
-
 #######################################################
 # PRIMIS
 #######################################################
 
-#Asthma
+
+# Asthma
+
 ## Asthma Diagnosis code
 ast = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-ast.csv",
@@ -65,6 +45,7 @@
 )
 
 # CKD
+
 ## Chronic kidney disease diagnostic codes
 ckd_cov = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-ckd_cov.csv",
@@ -89,7 +70,8 @@
   column="code",
 )
 
-# DB
+# Diabetes
+
 ## Diabetes diagnosis codes
 diab = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-diab.csv",
@@ -180,19 +162,21 @@
   column="code",
 )
 
-# BMI
+# Severe Obesity
+
+## BMI
 bmi = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-bmi.csv",
   column="code",
 )
 
-# All BMI coded terms
+## All BMI coded terms
 bmi_stage = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-bmi_stage.csv",
   column="code",
 )
 
-# Severe Obesity code recorded
+## Severe Obesity code recorded
 sev_obesity = codelist_from_csv(
   "codelists/primis-covid19-vacc-uptake-sev_obesity.csv",
   column="code",
@@ -203,44 +187,3 @@
   "codelists/primis-covid19-vacc-uptake-learndis.csv",
   column="code",
 )
-
-
-# Cancer
-
-cancer_haem_snomed=codelist_from_csv(
-    "codelists/opensafely-haematological-cancer-snomed.csv",
-    column="id",
-)
-
-cancer_nonhaem_nonlung_snomed=codelist_from_csv(
-    "codelists/opensafely-cancer-excluding-lung-and-haematological-snomed.csv",
-    column="id",
-)
-
-cancer_lung_snomed=codelist_from_csv(
-    "codelists/opensafely-lung-cancer-snomed.csv",
-    column="id",
-)
-
-chemotherapy_radiotherapy_snomed = codelist_from_csv(
-  "codelists/opensafely-chemotherapy-or-radiotherapy-snomed.csv", 
-  column = "id"
-)
-
-cancer_nonhaem_snomed = (
-  cancer_nonhaem_nonlung_snomed + 
-  cancer_lung_snomed + 
-  chemotherapy_radiotherapy_snomed
-)
-
-# solid organ transplant
-solid_organ_transplant=codelist_from_csv(
-    "codelists/opensafely-solid-organ-transplantation-snomed.csv",
-    column="id",
-)
-
-# HIV/AIDS
-hiv_aids=codelist_from_csv(
-    "codelists/nhsd-hiv-aids-snomed.csv",
-    column="code",
-)

analysis/PRIMIS/dataset_definition.py

@@ -0,0 +1,48 @@
+from ehrql import create_dataset
+from ehrql.tables.tpp import patients, practice_registrations
+
+# import variable definitions
+from variables_function import *
+
+#Import codelists
+from codelists import *
+
+# initialise dataset
+dataset = create_dataset()
+
+# Choose an index date
+index_date = "2020-12-08"
+
+#Dummy data
+dataset.configure_dummy_data(population_size=1000)
+
+# define dataset population
+dataset.define_population(
+  practice_registrations.for_patient_on(index_date).exists_for_patient() &
+  ((patients.date_of_death> index_date) | patients.date_of_death.is_null())
+)
+
+# Example 1: Add specific PRIMIS variables
+
+dataset.immunosuppressed = is_immunosuppressed(index_date) #immunosuppress grouped
+dataset.ckd = has_ckd(index_date) #chronic kidney disease
+dataset.crd = has_crd(index_date) # chronis respratory disease
+dataset.diabetes = has_diabetes(index_date) #diabetes
+dataset.cld = has_prior_event(cld, index_date) # chronic liver disease
+dataset.chd = has_prior_event(chd_cov, index_date) #chronic heart disease
+dataset.cns = has_prior_event(cns_cov, index_date) # chronic neurological disease
+dataset.asplenia = has_prior_event(spln_cov, index_date) # asplenia or dysfunction of the Spleen
+dataset.learndis = has_prior_event(learndis, index_date) # learning Disability
+dataset.smi = has_smi(index_date) #severe mental illness
+dataset.severe_obesity = has_severe_obesity(index_date) #immunosuppress grouped
+
+# Example 2: add the single PRIMIS "at risk" variable
+
+dataset.primis_atrisk = primis_atrisk(index_date) # at risk (at least one of the conditions above)
+
+# EXAMPLE 3: alternatively, use the `primis_variables` function to add variables programmatically:
+
+for i in range(0, 2):
+  suffix = f"_{i}"
+  primis_variables(dataset = dataset, index_date = index_date+years(i), var_name_suffix = suffix)
+