def test_clinical_event_with_category(): session = make_session() session.add_all( [ Patient(), Patient( CodedEvents=[ CodedEvent(CTV3Code="foo1", ConsultationDate="2018-01-01"), CodedEvent(CTV3Code="foo2", ConsultationDate="2020-01-01"), ] ), Patient( CodedEvents=[CodedEvent(CTV3Code="foo3", ConsultationDate="2019-01-01")] ), ] ) session.commit() codes = codelist([("foo1", "A"), ("foo2", "B"), ("foo3", "C")], "ctv3") study = StudyDefinition( population=patients.all(), code_category=patients.with_these_clinical_events( codes, returning="category", find_last_match_in_period=True, include_date_of_match=True, ), ) results = study.to_dicts() assert [x["code_category"] for x in results] == ["", "B", "C"] assert [x["code_category_date"] for x in results] == ["", "2020", "2019"]
def test_bmi_with_zero_values(): session = make_session() weight_code = "X76C7" height_code = "XM01E" patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=weight_code, NumericValue=0, ConsultationDate="2001-06-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=height_code, NumericValue=0, ConsultationDate="2001-06-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1995-01-01", on_or_before="2005-01-01", ), BMI_date_measured=patients.date_of("BMI", date_format="YYYY-MM-DD"), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.0"] assert [x["BMI_date_measured"] for x in results] == ["2001-06-01"]
def test_bmi_rounded(): session = make_session() weight_code = "X76C7" height_code = "XM01E" patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent( CTV3Code=weight_code, NumericValue=10.12345, ConsultationDate="2001-06-01" ) ) patient.CodedEvents.append( CodedEvent(CTV3Code=height_code, NumericValue=10, ConsultationDate="2000-02-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( "2005-01-01", include_measurement_date=True, include_month=True, include_day=True, ), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.1"] assert [x["BMI_date_measured"] for x in results] == ["2001-06-01"]
def test_explicit_bmi_fallback(): session = make_session() weight_code = "X76C7" bmi_code = "22K.." patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=weight_code, NumericValue=50, ConsultationDate="2001-06-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=bmi_code, NumericValue=99, ConsultationDate="2001-10-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1995-01-01", on_or_before="2005-01-01", include_measurement_date=True, include_month=True, include_day=True, ), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["99.0"] assert [x["BMI_date_measured"] for x in results] == ["2001-10-01"]
def test_bmi_when_only_some_measurements_of_child(): session = make_session() bmi_code = "22K.." weight_code = "X76C7" height_code = "XM01E" patient = Patient(DateOfBirth="1990-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=bmi_code, NumericValue=99, ConsultationDate="1995-01-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=weight_code, NumericValue=50, ConsultationDate="2010-01-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=height_code, NumericValue=10, ConsultationDate="2010-01-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="2005-01-01", on_or_before="2015-01-01", include_measurement_date=True, include_month=True, include_day=True, ), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.5"] assert [x["BMI_date_measured"] for x in results] == ["2010-01-01"]
def test_patients_categorised_as(): session = make_session() session.add_all( [ Patient( Sex="M", CodedEvents=[ CodedEvent(CTV3Code="foo1", ConsultationDate="2000-01-01") ], ), Patient( Sex="F", CodedEvents=[ CodedEvent(CTV3Code="foo2", ConsultationDate="2000-01-01"), CodedEvent(CTV3Code="bar1", ConsultationDate="2000-01-01"), ], ), Patient( Sex="M", CodedEvents=[ CodedEvent(CTV3Code="foo2", ConsultationDate="2000-01-01") ], ), Patient( Sex="F", CodedEvents=[ CodedEvent(CTV3Code="foo3", ConsultationDate="2000-01-01") ], ), ] ) session.commit() foo_codes = codelist([("foo1", "A"), ("foo2", "B"), ("foo3", "C")], "ctv3") bar_codes = codelist(["bar1"], "ctv3") study = StudyDefinition( population=patients.all(), category=patients.categorised_as( { "W": "foo_category = 'B' AND female_with_bar", "X": "sex = 'F' AND (foo_category = 'B' OR foo_category = 'C')", "Y": "sex = 'M' AND foo_category = 'A'", "Z": "DEFAULT", }, sex=patients.sex(), foo_category=patients.with_these_clinical_events( foo_codes, returning="category", find_last_match_in_period=True ), female_with_bar=patients.satisfying( "has_bar AND sex = 'F'", has_bar=patients.with_these_clinical_events(bar_codes), ), ), ) results = study.to_dicts() assert [x["category"] for x in results] == ["Y", "W", "Z", "X"] # Assert that internal columns do not appear assert "foo_category" not in results[0].keys() assert "female_with_bar" not in results[0].keys() assert "has_bar" not in results[0].keys()
def test_no_bmi_when_measurement_after_reference_date(): session = make_session() bmi_code = "22K.." patient = Patient(DateOfBirth="1900-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=bmi_code, NumericValue=99, ConsultationDate="2001-01-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1990-01-01", on_or_before="2000-01-01", include_measurement_date=True, include_month=True, include_day=True, ), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.0"] assert [x["BMI_date_measured"] for x in results] == [""]
def test_to_sql_passes(): session = make_session() patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code="XYZ", NumericValue=50, ConsultationDate="2002-06-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.with_these_clinical_events(codelist(["XYZ"], "ctv3")) ) sql = "SET NOCOUNT ON; " # don't output count after table output sql += study.to_sql() db_dict = study.get_db_dict() cmd = [ "sqlcmd", "-S", db_dict["hostname"] + "," + str(db_dict["port"]), "-d", db_dict["database"], "-U", db_dict["username"], "-P", db_dict["password"], "-Q", sql, "-W", # strip whitespace ] result = subprocess.run( cmd, capture_output=True, check=True, encoding="utf8" ).stdout patient_id = result.splitlines()[-1] assert patient_id == str(patient.Patient_ID)
def test_vaccination_events_sql(): session = make_session() session.add_all( [ # This patient is too old and should be ignored Patient( DateOfBirth="2002-05-04", Vaccinations=[ Vaccination( VaccinationName="Infanrix Hexa", VaccinationDate="2002-06-01", ) ], ), # This patient is too young and should be ignored Patient( DateOfBirth="2019-10-04", Vaccinations=[ Vaccination( VaccinationName="Infanrix Hexa", VaccinationDate="2019-11-04", ) ], ), Patient( DateOfBirth="2018-10-28", Vaccinations=[ Vaccination( VaccinationName="Infanrix Hexa", VaccinationDate="2018-11-01", ) ], MedicationIssues=[ MedicationIssue( MedicationDictionary=MedicationDictionary( DMD_ID="123", MultilexDrug_ID="123" ), ConsultationDate="2019-01-01", ), ], CodedEvents=[CodedEvent(CTV3Code="abc", ConsultationDate="2019-06-01")], ), ] ) session.commit() sql = vaccination_events_sql( date_of_birth_range=("2012-01-01", "2019-06-01"), tpp_vaccination_codelist=codelist( [("Infanrix Hexa", "dtap_hex")], system="tpp_vaccines", ), ctv3_codelist=codelist([("abc", "menb")], system="ctv3"), snomed_codelist=codelist([("123", "rotavirus")], system="snomed"), ) results = sql_to_dicts(sql) result_tuples = [(x["date_given"], x["vaccine_name"]) for x in results] # Results are ordered by patient ID but within each patient's results the # order is arbitrary. To make testing easier we sort them here. result_tuples = sorted(result_tuples) assert result_tuples == [ ("2018-11-01", "dtap_hex"), ("2019-01-01", "rotavirus"), ("2019-06-01", "menb"), ]
def test_patients_satisfying_with_hidden_columns(): condition_code = "ASTHMA" condition_code2 = "COPD" session = make_session() patient_1 = Patient(DateOfBirth="1940-01-01", Sex="M") patient_2 = Patient(DateOfBirth="1940-01-01", Sex="F") patient_3 = Patient(DateOfBirth="1990-01-01", Sex="M") patient_4 = Patient(DateOfBirth="1940-01-01", Sex="F") patient_4.CodedEvents.append( CodedEvent(CTV3Code=condition_code, ConsultationDate="2010-01-01") ) patient_5 = Patient(DateOfBirth="1940-01-01", Sex="F") patient_5.CodedEvents.append( CodedEvent(CTV3Code=condition_code, ConsultationDate="2010-01-01") ) patient_5.CodedEvents.append( CodedEvent(CTV3Code=condition_code2, ConsultationDate="2010-01-01") ) session.add_all([patient_1, patient_2, patient_3, patient_4, patient_5]) session.commit() study = StudyDefinition( population=patients.all(), sex=patients.sex(), age=patients.age_as_of("2020-01-01"), at_risk=patients.satisfying( """ (age > 70 AND sex = "M") OR (has_asthma AND NOT copd) """, has_asthma=patients.with_these_clinical_events( codelist([condition_code], "ctv3") ), copd=patients.with_these_clinical_events( codelist([condition_code2], "ctv3") ), ), ) results = study.to_dicts() assert [i["at_risk"] for i in results] == ["1", "0", "0", "1", "0"] assert "has_asthma" not in results[0].keys()
def test_simple_bmi(include_dates): session = make_session() weight_code = "X76C7" height_code = "XM01E" patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=weight_code, NumericValue=50, ConsultationDate="2002-06-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=height_code, NumericValue=10, ConsultationDate="2001-06-01") ) session.add(patient) session.commit() if include_dates == "none": bmi_date = None bmi_kwargs = {} elif include_dates == "year": bmi_date = "2002" bmi_kwargs = dict(include_measurement_date=True) elif include_dates == "month": bmi_date = "2002-06" bmi_kwargs = dict(include_measurement_date=True, include_month=True) elif include_dates == "day": bmi_date = "2002-06-01" bmi_kwargs = dict( include_measurement_date=True, include_month=True, include_day=True ) study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1995-01-01", on_or_before="2005-01-01", **bmi_kwargs ), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.5"] assert [x.get("BMI_date_measured") for x in results] == [bmi_date]
def test_simple_bmi(include_dates): session = make_session() weight_code = "X76C7" height_code = "XM01E" patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=weight_code, NumericValue=50, ConsultationDate="2002-06-01") ) patient.CodedEvents.append( CodedEvent(CTV3Code=height_code, NumericValue=10, ConsultationDate="2001-06-01") ) session.add(patient) session.commit() if include_dates == "none": bmi_date = None date_query = None elif include_dates == "year": bmi_date = "2002" date_query = patients.date_of("BMI") elif include_dates == "month": bmi_date = "2002-06" date_query = patients.date_of("BMI", date_format="YYYY-MM") elif include_dates == "day": bmi_date = "2002-06-01" date_query = patients.date_of("BMI", date_format="YYYY-MM-DD") study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1995-01-01", on_or_before="2005-01-01" ), **dict(BMI_date_measured=date_query) if date_query else {} ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.5"] assert [x.get("BMI_date_measured") for x in results] == [bmi_date]
def test_mean_recorded_value(): code = "2469." session = make_session() patient = Patient() values = [ ("2020-02-10", 90), ("2020-02-10", 100), ("2020-02-10", 98), # This day is outside period and should be ignored ("2020-04-01", 110), ] for date, value in values: patient.CodedEvents.append( CodedEvent(CTV3Code=code, NumericValue=value, ConsultationDate=date) ) patient_with_old_reading = Patient() patient_with_old_reading.CodedEvents.append( CodedEvent(CTV3Code=code, NumericValue=100, ConsultationDate="2010-01-01") ) patient_with_no_reading = Patient() session.add_all([patient, patient_with_old_reading, patient_with_no_reading]) session.commit() study = StudyDefinition( population=patients.all(), bp_systolic=patients.mean_recorded_value( codelist([code], system="ctv3"), on_most_recent_day_of_measurement=True, between=["2018-01-01", "2020-03-01"], include_measurement_date=True, include_month=True, include_day=True, ), ) results = study.to_dicts() results = [(i["bp_systolic"], i["bp_systolic_date_measured"]) for i in results] assert results == [("96.0", "2020-02-10"), ("0.0", ""), ("0.0", "")]
def test_using_expression_in_population_definition(): session = make_session() session.add_all( [ Patient( Sex="M", DateOfBirth="1970-01-01", CodedEvents=[ CodedEvent(CTV3Code="foo1", ConsultationDate="2000-01-01") ], ), Patient(Sex="M", DateOfBirth="1975-01-01"), Patient( Sex="F", DateOfBirth="1980-01-01", CodedEvents=[ CodedEvent(CTV3Code="foo1", ConsultationDate="2000-01-01") ], ), Patient(Sex="F", DateOfBirth="1985-01-01"), ] ) session.commit() study = StudyDefinition( population=patients.satisfying( "has_foo_code AND sex = 'M'", has_foo_code=patients.with_these_clinical_events( codelist(["foo1"], "ctv3") ), sex=patients.sex(), ), age=patients.age_as_of("2020-01-01"), ) results = study.to_dicts() assert results[0].keys() == {"patient_id", "age"} assert [i["age"] for i in results] == ["50"]
def test_sqlcmd_and_odbc_outputs_match(): session = make_session() patient = Patient(DateOfBirth="1950-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code="XYZ", NumericValue=50, ConsultationDate="2002-06-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.with_these_clinical_events(codelist(["XYZ"], "ctv3")) ) with tempfile.NamedTemporaryFile() as input_csv_odbc, tempfile.NamedTemporaryFile() as input_csv_sqlcmd: # windows line endings study.to_csv(input_csv_odbc.name, with_sqlcmd=False) # unix line endings study.to_csv(input_csv_sqlcmd.name, with_sqlcmd=True) assert filecmp.cmp(input_csv_odbc.name, input_csv_sqlcmd.name, shallow=False)
def test_no_bmi_when_measurements_of_child(): session = make_session() bmi_code = "22K.." patient = Patient(DateOfBirth="2000-01-01") patient.CodedEvents.append( CodedEvent(CTV3Code=bmi_code, NumericValue=99, ConsultationDate="2001-01-01") ) session.add(patient) session.commit() study = StudyDefinition( population=patients.all(), BMI=patients.most_recent_bmi( on_or_after="1995-01-01", on_or_before="2005-01-01", ), BMI_date_measured=patients.date_of("BMI", date_format="YYYY-MM-DD"), ) results = study.to_dicts() assert [x["BMI"] for x in results] == ["0.0"] assert [x["BMI_date_measured"] for x in results] == [""]
def test_patients_satisfying(): condition_code = "ASTHMA" session = make_session() patient_1 = Patient(DateOfBirth="1940-01-01", Sex="M") patient_2 = Patient(DateOfBirth="1940-01-01", Sex="F") patient_3 = Patient(DateOfBirth="1990-01-01", Sex="M") patient_4 = Patient(DateOfBirth="1940-01-01", Sex="F") patient_4.CodedEvents.append( CodedEvent(CTV3Code=condition_code, ConsultationDate="2010-01-01") ) session.add_all([patient_1, patient_2, patient_3, patient_4]) session.commit() study = StudyDefinition( population=patients.all(), sex=patients.sex(), age=patients.age_as_of("2020-01-01"), has_asthma=patients.with_these_clinical_events( codelist([condition_code], "ctv3") ), at_risk=patients.satisfying("(age > 70 AND sex = 'M') OR has_asthma"), ) results = study.to_dicts() assert [i["at_risk"] for i in results] == ["1", "0", "0", "1"]
def _make_clinical_events_selection(condition_code, patient_dates=None): # The default configuration of patients and dates which some tests assume if patient_dates is None: patient_dates = ["2001-06-01", "2002-06-01", None] session = make_session() for dates in patient_dates: patient = Patient() if dates is None: dates = [] elif isinstance(dates, str): dates = [dates] for date in dates: if isinstance(date, tuple): date, value = date else: value = 0.0 patient.CodedEvents.append( CodedEvent( CTV3Code=condition_code, ConsultationDate=date, NumericValue=value ) ) session.add(patient) session.commit()
def test_study_definition(tmp_path): session = make_session() session.add_all([ # This patient is too old and should be ignored Patient(Patient_ID=1, DateOfBirth="2002-05-04"), Patient( Patient_ID=2, DateOfBirth="2019-01-01", RegistrationHistory=[ RegistrationHistory( StartDate="2019-01-10", EndDate="9999-12-31", Organisation=Organisation(Organisation_ID=678), ), ], ), Patient( Patient_ID=3, DateOfBirth="2018-10-28", RegistrationHistory=[ RegistrationHistory( StartDate="2010-01-01", EndDate="2015-10-01", Organisation=Organisation(Organisation_ID=123), ), # Deliberately overlapping registration histories RegistrationHistory( StartDate="2015-04-01", EndDate="9999-12-31", Organisation=Organisation(Organisation_ID=345), ), ], Vaccinations=[ Vaccination( VaccinationName="Infanrix Hexa", VaccinationDate="2018-11-01", ) ], MedicationIssues=[ MedicationIssue( MedicationDictionary=MedicationDictionary( DMD_ID="123", MultilexDrug_ID="123"), ConsultationDate="2019-01-01", ), ], CodedEvents=[ CodedEvent(CTV3Code="abc", ConsultationDate="2019-06-01") ], ), ]) session.commit() study = VaccinationsStudyDefinition( start_date="2017-06-01", get_registered_practice_at_months=[12, 24, 60], tpp_vaccine_codelist=codelist( [ ("Infanrix Hexa", "dtap_hex"), ("Bexsero", "menb"), ("Rotarix", "rotavirus"), ("Prevenar", "pcv"), ("Prevenar - 13", "pcv"), ("Menitorix", "hib_menc"), ("Repevax", "dtap_ipv"), ("Boostrix-IPV", "dtap_ipv"), ("MMRvaxPRO", "mmr"), ("Priorix", "mmr"), ], system="tpp_vaccines", ), ctv3_vaccine_codelist=codelist([("abc", "menb")], system="ctv3"), snomed_vaccine_codelist=codelist([("123", "rotavirus")], system="snomed"), event_washout_period=14, vaccination_schedule=[ "dtap_hex_1", "menb_1", "rotavirus_1", "dtap_hex_2", "pcv_1", "rotavirus_2", "dtap_hex_3", "menb_2", "hib_menc_1", "pcv_2", "mmr_1", "menb_3", "dtap_ipv_1", "mmr_2", ], ) study.to_csv(tmp_path / "test.csv") with open(tmp_path / "test.csv", newline="") as f: reader = csv.DictReader(f) results = list(reader) assert results == [ { "patient_id": "2", "date_of_birth": "2019-01-01", "practice_id_at_month_12": "678", "practice_id_at_month_24": "678", "practice_id_at_month_60": "678", "dtap_hex_1": "", "menb_1": "", "rotavirus_1": "", "dtap_hex_2": "", "pcv_1": "", "rotavirus_2": "", "dtap_hex_3": "", "menb_2": "", "hib_menc_1": "", "pcv_2": "", "mmr_1": "", "menb_3": "", "dtap_ipv_1": "", "mmr_2": "", }, { "patient_id": "3", "date_of_birth": "2018-10-01", "practice_id_at_month_12": "345", "practice_id_at_month_24": "345", "practice_id_at_month_60": "345", "dtap_hex_1": "2018-11-01", "menb_1": "2019-06-01", "rotavirus_1": "2019-01-01", "dtap_hex_2": "", "pcv_1": "", "rotavirus_2": "", "dtap_hex_3": "", "menb_2": "", "hib_menc_1": "", "pcv_2": "", "mmr_1": "", "menb_3": "", "dtap_ipv_1": "", "mmr_2": "", }, ]
def test_number_of_episodes(): session = make_session() session.add_all( [ Patient( CodedEvents=[ CodedEvent(CTV3Code="foo1", ConsultationDate="2010-01-01"), # Throw in some irrelevant events CodedEvent(CTV3Code="mto1", ConsultationDate="2010-01-02"), CodedEvent(CTV3Code="mto2", ConsultationDate="2010-01-03"), # These two should be merged in to the previous event # because there's not more than 14 days between them CodedEvent(CTV3Code="foo2", ConsultationDate="2010-01-14"), CodedEvent(CTV3Code="foo3", ConsultationDate="2010-01-20"), # This is just outside the limit so should count as another event CodedEvent(CTV3Code="foo1", ConsultationDate="2010-02-04"), # This shouldn't count because there's an "ignore" event on # the same day (though at a different time) CodedEvent(CTV3Code="foo1", ConsultationDate="2012-01-01T10:45:00"), CodedEvent(CTV3Code="bar2", ConsultationDate="2012-01-01T16:10:00"), # This should be another episode CodedEvent(CTV3Code="foo1", ConsultationDate="2015-03-05"), # This "ignore" event should have no effect because it occurs # on a different day CodedEvent(CTV3Code="bar1", ConsultationDate="2015-03-06"), # This is after the time limit and so shouldn't count CodedEvent(CTV3Code="foo1", ConsultationDate="2020-02-05"), ] ), # This patient doesn't have any relevant events Patient( CodedEvents=[ CodedEvent(CTV3Code="mto1", ConsultationDate="2010-01-01"), CodedEvent(CTV3Code="mto2", ConsultationDate="2010-01-14"), CodedEvent(CTV3Code="mto3", ConsultationDate="2010-01-20"), CodedEvent(CTV3Code="mto1", ConsultationDate="2010-02-04"), CodedEvent(CTV3Code="mto1", ConsultationDate="2012-01-01T10:45:00"), CodedEvent(CTV3Code="mtr2", ConsultationDate="2012-01-01T16:10:00"), CodedEvent(CTV3Code="mto1", ConsultationDate="2015-03-05"), CodedEvent(CTV3Code="mto1", ConsultationDate="2020-02-05"), ] ), ] ) session.commit() foo_codes = codelist(["foo1", "foo2", "foo3"], "ctv3") bar_codes = codelist(["bar1", "bar2"], "ctv3") study = StudyDefinition( population=patients.all(), episode_count=patients.with_these_clinical_events( foo_codes, on_or_before="2020-01-01", ignore_days_where_these_codes_occur=bar_codes, returning="number_of_episodes", episode_defined_as="series of events each <= 14 days apart", ), ) results = study.to_dicts() assert [i["episode_count"] for i in results] == ["3", "0"]