def test_address_dtype_generation():
study = StudyDefinition(
# This line defines the study population
population=patients.all(),
rural_urban=patients.address_as_of(
"2020-02-01", returning="rural_urban_classification"
),
)
result = _converters_to_names(study.pandas_csv_args)
assert result == {
"dtype": {"rural_urban": "category"},
"parse_dates": [],
"date_col_for": {},
"converters": {},
}
"STP7": 0.1,
"STP8": 0.1,
"STP9": 0.1,
"STP10": 0.1,
}
},
},
),
imd=patients.address_as_of(
"2020-02-29",
returning="index_of_multiple_deprivation",
round_to_nearest=100,
return_expectations={
"rate": "universal",
"category": {
"ratios": {
"100": 0.1,
"200": 0.2,
"300": 0.7
}
},
},
),
ethnicity=patients.with_these_clinical_events(
ethnicity_codes,
returning="category",
find_last_match_in_period=True,
include_date_of_match=True,
return_expectations={
"category": {
"ratios": {
on_or_before="2020-06-01",
returning="date_of_death",
include_month=True,
include_day=True,
),
# The rest of the lines define the covariates with associated GitHub issues
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/33
age=patients.age_as_of("2020-02-01"),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/46
sex=patients.sex(),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/52
imd=patients.address_as_of(
"2020-02-01", returning="index_of_multiple_deprivation", round_to_nearest=100
),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/37
rural_urban=patients.address_as_of(
"2020-02-01", returning="rural_urban_classification"
),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/54
stp=patients.registered_practice_as_of("2020-02-01", returning="stp_code"),
# region - one of NHS England 9 regions
region=patients.registered_practice_as_of("2020-02-01", returning="nhse_region_name"),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/10
bmi=patients.most_recent_bmi(
"ratios": {
"MSOA1": 0.5,
"MSOA2": 0.5
}
},
},
),
# https://github.com/ebmdatalab/tpp-sql-notebook/issues/52
imd=patients.address_as_of(
"2020-02-01",
returning="index_of_multiple_deprivation",
round_to_nearest=100,
return_expectations={
"rate": "universal",
"category": {
"ratios": {
"100": 0.1,
"200": 0.2,
"300": 0.7
}
},
},
),
rural_urban=patients.address_as_of(
"2020-02-01",
returning="rural_urban_classification",
return_expectations={
"rate": "universal",
"category": {
"ratios": {
"rural": 0.1,
def test_patients_address_as_of():
session = make_session()
patient = Patient()
patient.Addresses.append(
PatientAddress(
StartDate="1990-01-01",
EndDate="2018-01-01",
ImdRankRounded=100,
RuralUrbanClassificationCode=1,
)
)
# We deliberately create overlapping address periods here to check that we
# handle these correctly
patient.Addresses.append(
PatientAddress(
StartDate="2018-01-01",
EndDate="2020-02-01",
ImdRankRounded=200,
RuralUrbanClassificationCode=1,
)
)
patient.Addresses.append(
PatientAddress(
StartDate="2019-01-01",
EndDate="2022-01-01",
ImdRankRounded=300,
RuralUrbanClassificationCode=2,
)
)
patient.Addresses.append(
PatientAddress(
StartDate="2022-01-01",
EndDate="9999-12-31",
ImdRankRounded=500,
RuralUrbanClassificationCode=3,
)
)
patient_no_address = Patient()
patient_only_old_address = Patient()
patient_only_old_address.Addresses.append(
PatientAddress(
StartDate="2010-01-01",
EndDate="2015-01-01",
ImdRankRounded=100,
RuralUrbanClassificationCode=1,
)
)
session.add_all([patient, patient_no_address, patient_only_old_address])
session.commit()
study = StudyDefinition(
population=patients.all(),
imd=patients.address_as_of(
"2020-01-01",
returning="index_of_multiple_deprivation",
round_to_nearest=100,
),
rural_urban=patients.address_as_of(
"2020-01-01", returning="rural_urban_classification"
),
)
results = study.to_dicts()
assert [i["imd"] for i in results] == ["300", "0", "0"]
assert [i["rural_urban"] for i in results] == ["2", "0", "0"]