Merge pull request #71 from opensafely/dataset_definition_revision

Dataset definition revision

analysis/active_analyses.R

@@ -65,14 +65,14 @@ core_covars <- c(
 
 ## Define project-specific covariates (specific to respiratory project) ----
 project_covars <- c(
-  "cov_bin_history_pneumonia_snomed", "cov_bin_history_asthma_snomed", 
-  "cov_bin_history_pulmonary_fibrosis_snomed", "cov_bin_all_stroke"
+  "cov_bin_history_pneumonia", "cov_bin_history_asthma", 
+  "cov_bin_history_pulmonary_fibrosis", "cov_bin_stroke_isch"
 )
 # Combine covariates into a single vector ----
 all_covars <- c(core_covars, project_covars)
 
 ## Combine covariates into a single string for analysis ----
-preex_FALSE_covars <- paste0(all_covars[!all_covars %in% c("cov_bin_history_asthma_snomed", "cov_bin_history_copd")], collapse = ";")
+preex_FALSE_covars <- paste0(all_covars[!all_covars %in% c("cov_bin_history_asthma", "cov_bin_history_copd")], collapse = ";")
 all_covars <- paste0(c(core_covars, project_covars), collapse = ";")
 
 # Specify cohorts ----

analysis/dataset_definition/codelists.py

@@ -423,25 +423,46 @@
 
 # Preexisting respiratory condition
 
-copd_ctv3 = codelist_from_csv(
+copd_ctv3_clinical = codelist_from_csv(
     "codelists/opensafely-current-copd.csv",
     column="CTV3ID"
 )
-asthma_snomed = codelist_from_csv(
+
+copd_icd10 = codelist_from_csv(
+    "codelists/opensafely-copd-secondary-care.csv",
+    column="code"
+)
+
+asthma_snomed_clinical = codelist_from_csv(
     "codelists/opensafely-asthma-diagnosis-snomed.csv",
     column="id"
 )
 
+asthma_icd10 = codelist_from_csv(
+    "codelists/opensafely-asthma-exacerbation-secondary-care.csv",
+    column="code"
+)
+
 # Respiratory outcome(s)
 
-pneumonia_snomed = codelist_from_csv(
+pneumonia_snomed_clinical = codelist_from_csv(
     "codelists/bristol-pneumonia-snomed.csv",
     column="code"
 )
 
-pulmonary_fibrosis_snomed = codelist_from_csv(
+pneumonia_icd10 = codelist_from_csv(
+    "codelists/opensafely-pneumonia-secondary-care.csv",
+    column="ICD code"
+)
+
+pulmonary_fibrosis_snomed_clinical = codelist_from_csv(
     "codelists/bristol-ild-snomed.csv",
     column="code"
 )
 
+pulmonary_fibrosis_icd10 = codelist_from_csv(
+    "codelists/bristol-pulmonary-fibrosis-interstitial-lung-disease.csv",
+    column="code"
+)
+
 # asthma and copd snomed codes are above under 'Preexisting respiratory condition' - same codes and variable names to be used for outcomes

analysis/dataset_definition/variable_helper_functions.py

@@ -10,8 +10,6 @@
     sgss_covid_all_tests,
     apcs, 
     ec, 
-    opa, 
-    opa_diag, 
     clinical_events, 
     medications, 
     ons_deaths,
@@ -52,23 +50,15 @@ def last_matching_med_dmd_before(codelist, start_date, where=True):
         .last_for_patient()
     )
 
-def last_matching_event_apc_before(codelist, start_date, where=True):
-    return(
-        apcs.where(where)
-        .where(apcs.primary_diagnosis.is_in(codelist) | apcs.secondary_diagnosis.is_in(codelist))
-        .where(apcs.admission_date.is_before(start_date))
-        .sort_by(apcs.admission_date)
-        .last_for_patient()
-    )
-
-def last_matching_event_opa_before(codelist, start_date, where=True):
-    return(
-        opa_diag.where(where)
-        .where(opa_diag.primary_diagnosis_code.is_in(codelist) | opa_diag.secondary_diagnosis_code_1.is_in(codelist))
-        .where(opa_diag.appointment_date.is_before(start_date))
-        .sort_by(opa_diag.appointment_date)
-        .last_for_patient()
-    )
+def last_matching_event_apc_before(codelist, start_date, only_prim_diagnoses=False, where=True):
+    query = apcs.where(where).where(apcs.admission_date.is_before(start_date))
+    if only_prim_diagnoses:
+        query = query.where(
+            apcs.primary_diagnosis.is_in(codelist)
+        )
+    else:
+        query = query.where(apcs.all_diagnoses.contains_any_of(codelist))
+    return query.sort_by(apcs.admission_date).last_for_patient()
 
 # helper function
 def any_of(conditions):
@@ -139,23 +129,15 @@ def first_matching_med_dmd_between(codelist, start_date, end_date, where=True):
         .first_for_patient()
     )
 
-def first_matching_event_apc_between(codelist, start_date, end_date, where=True):
-    return(
-        apcs.where(where)
-        .where(apcs.primary_diagnosis.is_in(codelist) | apcs.secondary_diagnosis.is_in(codelist))
-        .where(apcs.admission_date.is_on_or_between(start_date, end_date))
-        .sort_by(apcs.admission_date)
-        .first_for_patient()
-    )
-
-def first_matching_event_opa_between(codelist, start_date, end_date, where=True):
-    return(
-        opa_diag.where(where)
-        .where(opa_diag.primary_diagnosis_code.is_in(codelist) | opa_diag.secondary_diagnosis_code_1.is_in(codelist))
-        .where(opa_diag.appointment_date.is_on_or_between(start_date, end_date))
-        .sort_by(opa_diag.appointment_date)
-        .first_for_patient()
-    )
+def first_matching_event_apc_between(codelist, start_date, end_date, only_prim_diagnoses=False, where=True):
+    query = apcs.where(where).where(apcs.admission_date.is_on_or_between(start_date, end_date))
+    if only_prim_diagnoses:
+        query = query.where(
+            apcs.primary_diagnosis.is_in(codelist)
+        )
+    else:
+        query = query.where(apcs.all_diagnoses.contains_any_of(codelist))
+    return query.sort_by(apcs.admission_date).first_for_patient()
 
 def first_matching_event_ec_snomed_between(codelist, start_date, end_date, where=True):
     conditions = [