Merge pull request #86 from opensafely/viv3ckj/refactor-code

Viv3ckj/refactor code (part 1)

analysis/codelists.py

@@ -68,3 +68,12 @@
 # Community Pharmacist Consultation Service for minor illness - 1577041000000109
 # Pharmacy First service - 983341000000102
 pharmacy_first_consultation_codelist = ["1577041000000109", "983341000000102"]
+
+# PF codes separated for individual analysis
+pharmacy_first_event_codes = {
+    # Community Pharmacist (CP) Consultation Service for minor illness (procedure)
+    "consultation_service": ["1577041000000109"],
+    # Pharmacy First service (qualifier value)
+    "pharmacy_first_service": ["983341000000102"],
+    "combined_pf_service": ["1577041000000109", "983341000000102"],
+}

analysis/config.py

@@ -0,0 +1,15 @@
+# measures_definition_pf_breakdown.py
+start_date_measure_pf_breakdown = "2023-11-01"
+monthly_intervals_measure_pf_breakdown = 12
+
+# measures_definition_pf_condition_provider.py
+start_date_measure_condition_provider = "2023-01-01"
+monthly_intervals_measure_condition_provider = 22
+
+# measures_definition_pf_descriptive_stats.py
+start_date_measure_descriptive_stats = "2024-02-01"
+monthly_intervals_measure_descriptive_stats = 9
+
+# measures_definition_pf_medications.py
+start_date_measure_medications = "2023-11-01"
+monthly_intervals_measure_medications = 9

analysis/measures_definition_pf_breakdown.py

@@ -11,13 +11,16 @@
     ethnicity_codelist,
 )
 
-from pf_dataset import pharmacy_first_event_codes, get_latest_ethnicity
+from pf_dataset import get_latest_ethnicity
+from codelists import pharmacy_first_event_codes, pharmacy_first_consultation_codelist
+from config import start_date_measure_pf_breakdown, monthly_intervals_measure_pf_breakdown
+from pf_variables_library import select_events
 
 measures = create_measures()
 measures.configure_dummy_data(population_size=1000)
 
-start_date = "2023-11-01"
-monthly_intervals = 12
+start_date = start_date_measure_pf_breakdown
+monthly_intervals = monthly_intervals_measure_pf_breakdown
 
 registration = practice_registrations.for_patient_on(INTERVAL.end_date)
 ethnicity_combined = get_latest_ethnicity(
@@ -56,16 +59,12 @@
     otherwise="Missing",
 )
 
-pharmacy_first_ids = clinical_events.where(
-    clinical_events.snomedct_code.is_in(
-        pharmacy_first_event_codes["combined_pf_service"]
-    )
-).consultation_id
+pharmacy_first_ids = select_events(clinical_events, codelist=pharmacy_first_consultation_codelist).consultation_id
 
-# Select clinical events in interval date range
-selected_events = clinical_events.where(
-    clinical_events.date.is_on_or_between(INTERVAL.start_date, INTERVAL.end_date)
-).where(clinical_events.consultation_id.is_in(pharmacy_first_ids))
+# # Select clinical events in interval date range
+selected_events = select_events(clinical_events, start_date=INTERVAL.start_date, end_date=INTERVAL.end_date).where(
+    clinical_events.consultation_id.is_in(pharmacy_first_ids)
+)
 
 # Breakdown metrics to be produced as graphs
 breakdown_metrics = {

analysis/measures_definition_pf_condition_provider.py

@@ -8,27 +8,22 @@
     pharmacy_first_conditions_codes,
     imd_quintile,
 )
-from pf_dataset import pharmacy_first_event_codes
-
+from codelists import pharmacy_first_consultation_codelist
+from config import start_date_measure_condition_provider, monthly_intervals_measure_condition_provider
+from pf_variables_library import select_events
 
 measures = create_measures()
 measures.configure_dummy_data(population_size=1000)
 
-start_date = "2023-01-01"
-monthly_intervals = 22
+start_date = start_date_measure_condition_provider
+monthly_intervals = monthly_intervals_measure_condition_provider
 
 registration = practice_registrations.for_patient_on(INTERVAL.end_date)
 
-selected_events = clinical_events.where(
-    clinical_events.date.is_on_or_between(INTERVAL.start_date, INTERVAL. end_date)
-)
+selected_events = select_events(clinical_events, start_date=INTERVAL.start_date, end_date=INTERVAL.end_date)
 
 # Create variable which contains boolean values of whether pharmacy first event exists for patient
-has_pharmacy_first = selected_events.where(
-    selected_events.snomedct_code.is_in(
-        pharmacy_first_event_codes["combined_pf_service"]
-    )
-).exists_for_patient()
+has_pharmacy_first = select_events(selected_events, codelist=pharmacy_first_consultation_codelist).exists_for_patient()
 
 for condition_name, condition_code in pharmacy_first_conditions_codes.items():
     condition_events = selected_events.where(

analysis/measures_definition_pf_descriptive_stats.py

@@ -2,45 +2,42 @@
 from ehrql.tables.raw.tpp import medications
 from ehrql.tables.tpp import practice_registrations, patients, clinical_events
 
-from pf_variables_library import get_consultation_ids, get_consultationid_events
+from pf_variables_library import select_events_from_codelist, select_events_by_consultation_id
 from codelists import (
     pharmacy_first_med_codelist,
     pharmacy_first_consultation_codelist,
     pharmacy_first_conditions_codelist,
 )
+from config import start_date_measure_descriptive_stats, monthly_intervals_measure_descriptive_stats
 
 measures = create_measures()
 measures.configure_dummy_data(population_size=1000)
 measures.configure_disclosure_control(enabled=True)
 
-start_date = "2024-02-01"
-monthly_intervals = 9
+start_date = start_date_measure_descriptive_stats
+monthly_intervals = monthly_intervals_measure_descriptive_stats
 
 registration = practice_registrations.for_patient_on(INTERVAL.end_date)
 
 # Function to retrieve consultation ids from clinical events that are PF consultations
-pharmacy_first_ids = get_consultation_ids(
+pharmacy_first_ids = select_events_from_codelist(
     clinical_events, pharmacy_first_consultation_codelist
-)
+).consultation_id
 
 # Function to retrieve selected events using pharmacy first ids
-selected_clinical_events = get_consultationid_events(
+selected_clinical_events = select_events_by_consultation_id(
     clinical_events, pharmacy_first_ids
 ).where(clinical_events.date.is_on_or_between(INTERVAL.start_date, INTERVAL.end_date))
 
-selected_med_events = get_consultationid_events(medications, pharmacy_first_ids).where(
+selected_med_events = select_events_by_consultation_id(medications, pharmacy_first_ids).where(
     medications.date.is_on_or_between(INTERVAL.start_date, INTERVAL.end_date)
 )
 
 # Create variable which contains boolean values of whether pharmacy first event exists for patient
-has_pf_consultation = selected_clinical_events.where(
-    selected_clinical_events.snomedct_code.is_in(pharmacy_first_consultation_codelist)
-).exists_for_patient()
+has_pf_consultation = select_events_from_codelist(selected_clinical_events, pharmacy_first_consultation_codelist).exists_for_patient()
 
 # PF consultations with PF clinical condition
-has_pf_condition = selected_clinical_events.where(
-    selected_clinical_events.snomedct_code.is_in(pharmacy_first_conditions_codelist)
-).exists_for_patient()
+has_pf_condition = select_events_from_codelist(selected_clinical_events, pharmacy_first_conditions_codelist).exists_for_patient()
 
 # PF consultations with prescribed PF medication
 has_pf_medication = selected_med_events.where(

analysis/measures_definition_pf_medications.py

@@ -5,46 +5,32 @@
     practice_registrations,
 )
 from ehrql.tables.raw.tpp import medications
-
-from pf_dataset import pharmacy_first_event_codes
-from codelists import *
-
-pharmacy_first_med_codes = (
-    acute_otitis_media_tx_cod
-    + impetigo_treatment_tx_cod
-    + infected_insect_bites_tx_cod
-    + shingles_treatment_tx_cod
-    + sinusitis_tx_cod
-    + sore_throat_tx_cod
-    + urinary_tract_infection_tx_cod
-)
+from codelists import pharmacy_first_consultation_codelist, pharmacy_first_med_codelist
+from config import start_date_measure_medications, monthly_intervals_measure_medications
+from pf_variables_library import select_events
 
 measures = create_measures()
 measures.configure_dummy_data(population_size=1000)
 # Turn off during code development, but turn on before running against on the server
 measures.configure_disclosure_control(enabled=True)
 
-start_date = "2023-11-01"
-monthly_intervals = 9
+start_date = start_date_measure_medications
+monthly_intervals = monthly_intervals_measure_medications
 
 registration = practice_registrations.for_patient_on(INTERVAL.end_date)
 
 # Select Pharmacy First consultations during interval date range
-pharmacy_first_events = clinical_events.where(
-    clinical_events.date.is_on_or_between(INTERVAL.start_date, INTERVAL.end_date)
-).where(
-    clinical_events.snomedct_code.is_in(
-        pharmacy_first_event_codes["combined_pf_service"]
-    )
+pharmacy_first_events = select_events(clinical_events, start_date=INTERVAL.start_date, end_date=INTERVAL.end_date).where(
+    clinical_events.snomedct_code.is_in(pharmacy_first_consultation_codelist)
 )
 
 pharmacy_first_ids = pharmacy_first_events.consultation_id
 has_pharmacy_first_consultation = pharmacy_first_events.exists_for_patient()
 
 # Select medications prescribed with PF consultation ID
-selected_medications = medications.where(
-    medications.date.is_on_or_between(INTERVAL.start_date, INTERVAL.end_date)
-).where(medications.consultation_id.is_in(pharmacy_first_ids))
+selected_medications = select_events(medications, start_date=INTERVAL.start_date, end_date=INTERVAL.end_date).where(
+    medications.consultation_id.is_in(pharmacy_first_ids)
+)
 
 # Select first medication for group_by argument in measures
 first_selected_medication = (
@@ -53,7 +39,7 @@
 
 # Check if a medication is from our PF codelists 
 has_pharmacy_first_medication = first_selected_medication.is_in(
-    pharmacy_first_med_codes
+    pharmacy_first_med_codelist
 )
 
 # Numerator, patients with a PF medication

analysis/pf_dataset.py

@@ -2,18 +2,8 @@
 
 from pf_variables_library import check_pregnancy_status, count_past_events
 
-pharmacy_first_event_codes = {
-    # # Community Pharmacy (CP) Blood Pressure (BP) Check Service (procedure)
-    # "blood_pressure_service": ["1659111000000107"],
-    # # Community Pharmacy (CP) Contraception Service (procedure)
-    # "contraception_service": ["1659121000000101"],
-    # Community Pharmacist (CP) Consultation Service for minor illness (procedure)
-    "consultation_service": ["1577041000000109"],
-    # Pharmacy First service (qualifier value)
-    "pharmacy_first_service": ["983341000000102"],
-    "combined_pf_service": ["1577041000000109", "983341000000102"],
-}
-
+# This file contains functions for the denominators of the patient population for each clinical condition.
+# It will be used to calculate rates, and is separate from pf_variables_library
 
 # Create denominator variables for each clinical condition
 # These are based on NHS England rules using sex, age, pregnancy status and repeated diagnoses

analysis/pf_variables_library.py

@@ -1,7 +1,6 @@
 # Function to check status of a condition within a specified time window
 from ehrql import months
 
-
 def check_pregnancy_status(index_date, selected_events, codelist):
     return (
         selected_events.where(selected_events.snomedct_code.is_in(codelist))
@@ -11,7 +10,6 @@ def check_pregnancy_status(index_date, selected_events, codelist):
         .exists_for_patient()
     )
 
-
 # Function to count number of coded events within a specified time window
 def count_past_events(index_date, selected_events, codelist, num_months):
     return (
@@ -24,19 +22,40 @@ def count_past_events(index_date, selected_events, codelist, num_months):
         .count_for_patient()
     )
 
+# Function to get events linked to a specified codelist
+def select_events_from_codelist(event_frame, codelist):
+    selected_events = event_frame.where(
+        event_frame.snomedct_code.is_in(codelist)
+    )
 
-# Function to get consultation IDs linked to a specified codelist
-def get_consultation_ids(clinical_events, codelist):
-    consultation_ids = clinical_events.where(
-        clinical_events.snomedct_code.is_in(codelist)
-    ).consultation_id
-
-    return consultation_ids
-
+    return selected_events
 
 # Function to get events with specific consultation IDs
-def get_consultationid_events(event_frame, consultation_ids):
+def select_events_by_consultation_id(event_frame, consultation_ids):
     selected_events = event_frame.where(
         event_frame.consultation_id.is_in(consultation_ids)
     )
     return selected_events
+
+# Function to get events within a time frame
+def select_events_between(event_frame, start_date, end_date):
+    selected_events = event_frame.where(
+        event_frame.date.is_on_or_between(start_date, end_date)
+    )
+    return selected_events
+
+def select_events(event_frame, codelist=None, consultation_ids=None, start_date=None, end_date=None):
+    """
+    Wrapper function to select events based on codelist, consultation IDs, or a date range.
+    Allows combining multiple selection criteria.
+    """
+    selected_events = event_frame
+
+    if codelist is not None:
+        selected_events = select_events_from_codelist(selected_events, codelist)
+    if consultation_ids is not None:
+        selected_events = select_events_by_consultation_id(selected_events, consultation_ids)
+    if start_date is not None and end_date is not None:
+        selected_events = select_events_between(selected_events, start_date, end_date)
+
+    return selected_events