ehrQL language reference🔗
Frames🔗
Frames are the starting point for building any query in ehrQL. You can think of a Frame as being like a table in a database, in that it contains multiple rows and multiple columns. But a Frame can have operations applied to it like filtering or sorting to produce a new Frame.
You don't need to define any Frames yourself. Instead you import them
from the various schemas available in ehrql.tables e.g.
from ehrql.tables.beta.core import patients
Frames have columns which you can access as attributes on the Frame e.g.
dob = patients.date_of_birth
The schema documentation contains the full list of
available columns for each Frame. For example, see
ehrql.tables.beta.core.patients.
Accessing a column attribute on a Frame produces a Series, which are documented elsewhere below.
Some Frames contain at most one row per patient, we call these PatientFrames; others can contain multiple rows per patient, we call these EventFrames.
class PatientFrame() 🔗
Frame containing at most one row per patient.
Return a boolean patient series which is True for each patient that has a row in this frame and False otherwise.
Return an integer patient series giving the number of rows each patient has in this frame.
Note this will be 0 rather than NULL if the patient has no rows at all in the frame.
class EventFrame() 🔗
Frame which may contain multiple rows per patient.
Return a new frame containing only the rows in this frame for which condition
evaluates True.
Note that this excludes any rows for which condition is NULL.
Return a new frame containing only the rows in this frame for which condition
evaluates False or NULL i.e. the exact inverse of the rows included by
where().
Sort the rows for each patient by each of the supplied sort_values.
Where more than one sort value is supplied then the first (i.e. left-most) value has highest priority and each subsequent sort value will only be used as a tie-breaker in case of an exact match among previous values.
Note that NULL is considered smaller than any other value, so you may wish to filter out NULL values before sorting.
Return a boolean patient series which is True for each patient that has a row in this frame and False otherwise.
Return an integer patient series giving the number of rows each patient has in this frame.
Note this will be 0 rather than NULL if the patient has no rows at all in the frame.
class SortedEventFrame() 🔗
Frame which contains multiple rows per patient and has had one or more sort operations applied.
Return a new frame containing only the rows in this frame for which condition
evaluates True.
Note that this excludes any rows for which condition is NULL.
Return a new frame containing only the rows in this frame for which condition
evaluates False or NULL i.e. the exact inverse of the rows included by
where().
Sort the rows for each patient by each of the supplied sort_values.
Where more than one sort value is supplied then the first (i.e. left-most) value has highest priority and each subsequent sort value will only be used as a tie-breaker in case of an exact match among previous values.
Note that NULL is considered smaller than any other value, so you may wish to filter out NULL values before sorting.
Return a boolean patient series which is True for each patient that has a row in this frame and False otherwise.
Return an integer patient series giving the number of rows each patient has in this frame.
Note this will be 0 rather than NULL if the patient has no rows at all in the frame.
Return a PatientFrame containing, for each patient, the first matching row according to whatever sort order has been applied.
Note that where there are multiple rows tied for first place then the specific row returned is picked arbitrarily but consistently i.e. you shouldn't depend on getting any particular result, but the result you do get shouldn't change unless the data changes.
Return a PatientFrame containing, for each patient, the last matching row according to whatever sort order has been applied.
Note that where there are multiple rows tied for last place then the specific row returned is picked arbitrarily but consistently i.e. you shouldn't depend on getting any particular result, but the result you do get shouldn't change unless the data changes.
Series🔗
A Series represents a column of values of a certain type. Some Series contain at most one value per patient, we call these PatientSeries; others can contain multiple values per patient, we call these EventSeries. Values can be NULL (i.e. missing) but a Series can never mix values of different types.
class BoolPatientSeries() 🔗
One row per patient series of type boolean
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Logical AND
Return a boolean series which is True where both this series and other are
True, False where either are False, and NULL otherwise.
Logical OR
Return a boolean series which is True where either this series or other is
True, False where both are False, and NULL otherwise.
Logical NOT
Return a boolean series which is the inverse of this series i.e. where True becomes False, False becomes True, and NULL stays as NULL.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
class BoolEventSeries() 🔗
Multiple rows per patient series of type boolean
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Logical AND
Return a boolean series which is True where both this series and other are
True, False where either are False, and NULL otherwise.
Logical OR
Return a boolean series which is True where either this series or other is
True, False where both are False, and NULL otherwise.
Logical NOT
Return a boolean series which is the inverse of this series i.e. where True becomes False, False becomes True, and NULL stays as NULL.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
class StrPatientSeries() 🔗
One row per patient series of type string
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return a boolean series which is True for each string in this series which
contains the corresponding value in other as a sub-string and False otherwise (or
NULL if either value is NULL).
class StrEventSeries() 🔗
Multiple rows per patient series of type string
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return a boolean series which is True for each string in this series which
contains the corresponding value in other as a sub-string and False otherwise (or
NULL if either value is NULL).
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
Return the minimum value in the series for each patient (or NULL if the patient has no values).
Return the maximum value in the series for each patient (or NULL if the patient has no values).
class IntPatientSeries() 🔗
One row per patient series of type integer
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return the sum of each corresponding value in this series and other (or NULL
if either is NULL).
Return each value in this series with its corresponding value in other
subtracted (or NULL if either is NULL).
Return the product of each corresponding value in this series and other (or
NULL if either is NULL).
Return a series with each value in this series divided by its correponding value
in other (or NULL if either is NULL).
Note that the result is always if a float even if the inputs are integers.
Return a series with each value in this series divided by its correponding value
in other and then rounded down to the nearest integer value (or NULL if either
is NULL).
Note that the result is always if an integer even if the inputs are floats.
Return the negation of each value in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return each value in this series rounded down to the nearest integer.
Return each value in this series as a float e.g 10 becomes 10.0
class IntEventSeries() 🔗
Multiple rows per patient series of type integer
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return the sum of each corresponding value in this series and other (or NULL
if either is NULL).
Return each value in this series with its corresponding value in other
subtracted (or NULL if either is NULL).
Return the product of each corresponding value in this series and other (or
NULL if either is NULL).
Return a series with each value in this series divided by its correponding value
in other (or NULL if either is NULL).
Note that the result is always if a float even if the inputs are integers.
Return a series with each value in this series divided by its correponding value
in other and then rounded down to the nearest integer value (or NULL if either
is NULL).
Note that the result is always if an integer even if the inputs are floats.
Return the negation of each value in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return each value in this series rounded down to the nearest integer.
Return each value in this series as a float e.g 10 becomes 10.0
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
Return the minimum value in the series for each patient (or NULL if the patient has no values).
Return the maximum value in the series for each patient (or NULL if the patient has no values).
Return the sum of all values in the series for each patient.
Return the arithmetic mean of any non-NULL values in the series for each patient.
class FloatPatientSeries() 🔗
One row per patient series of type float
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return the sum of each corresponding value in this series and other (or NULL
if either is NULL).
Return each value in this series with its corresponding value in other
subtracted (or NULL if either is NULL).
Return the product of each corresponding value in this series and other (or
NULL if either is NULL).
Return a series with each value in this series divided by its correponding value
in other (or NULL if either is NULL).
Note that the result is always if a float even if the inputs are integers.
Return a series with each value in this series divided by its correponding value
in other and then rounded down to the nearest integer value (or NULL if either
is NULL).
Note that the result is always if an integer even if the inputs are floats.
Return the negation of each value in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return each value in this series rounded down to the nearest integer.
Return each value in this series as a float e.g 10 becomes 10.0
class FloatEventSeries() 🔗
Multiple rows per patient series of type float
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return the sum of each corresponding value in this series and other (or NULL
if either is NULL).
Return each value in this series with its corresponding value in other
subtracted (or NULL if either is NULL).
Return the product of each corresponding value in this series and other (or
NULL if either is NULL).
Return a series with each value in this series divided by its correponding value
in other (or NULL if either is NULL).
Note that the result is always if a float even if the inputs are integers.
Return a series with each value in this series divided by its correponding value
in other and then rounded down to the nearest integer value (or NULL if either
is NULL).
Note that the result is always if an integer even if the inputs are floats.
Return the negation of each value in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return each value in this series rounded down to the nearest integer.
Return each value in this series as a float e.g 10 becomes 10.0
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
Return the minimum value in the series for each patient (or NULL if the patient has no values).
Return the maximum value in the series for each patient (or NULL if the patient has no values).
Return the sum of all values in the series for each patient.
Return the arithmetic mean of any non-NULL values in the series for each patient.
class DatePatientSeries() 🔗
One row per patient series of type date
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a series giving the difference between each date in this series and
other (see DateDifference).
Return an integer series giving the year of each date in this series.
Return an integer series giving the month (1-12) of each date in this series.
Return an integer series giving the day of the month (1-31) of each date in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return a date series with each date in this series replaced by the date of the first day in its corresponding calendar year.
Return a date series with each date in this series replaced by the date of the first day in its corresponding calendar month.
Return a boolean series which is True for each date in this series that is
earlier than its corresponding date in other and False otherwise (or NULL if
either value is NULL).
Return a boolean series which is True for each date in this series that is
earlier than or the same as its corresponding value in other and False
otherwise (or NULL if either value is NULL).
Return a boolean series which is True for each date in this series that is later
than its corresponding date in other and False otherwise (or NULL if either value
is NULL).
Return a boolean series which is True for each date in this series that is later
than or the same as its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a boolean series which is True for each date in this series which is
strictly between (i.e. not equal to) the corresponding dates in start and end.
Return a boolean series which is True for each date in this series which is
between or the same as the corresponding dates in start and end.
The same as is_on_or_between() above, but allows supplying a start/end date
pair as single argument.
class DateEventSeries() 🔗
Multiple rows per patient series of type date
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series that is
strictly less than its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is less
than or equal to its corresponding value in other and False otherwise (or NULL
if either value is NULL).
Return a boolean series which is True for each value in this series that is
greater than or equal to its corresponding value in other and False otherwise
(or NULL if either value is NULL).
Return a boolean series which is True for each value in this series that is
strictly greater than its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a series giving the difference between each date in this series and
other (see DateDifference).
Return an integer series giving the year of each date in this series.
Return an integer series giving the month (1-12) of each date in this series.
Return an integer series giving the day of the month (1-31) of each date in this series.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
Return a date series with each date in this series replaced by the date of the first day in its corresponding calendar year.
Return a date series with each date in this series replaced by the date of the first day in its corresponding calendar month.
Return a boolean series which is True for each date in this series that is
earlier than its corresponding date in other and False otherwise (or NULL if
either value is NULL).
Return a boolean series which is True for each date in this series that is
earlier than or the same as its corresponding value in other and False
otherwise (or NULL if either value is NULL).
Return a boolean series which is True for each date in this series that is later
than its corresponding date in other and False otherwise (or NULL if either value
is NULL).
Return a boolean series which is True for each date in this series that is later
than or the same as its corresponding value in other and False otherwise (or
NULL if either value is NULL).
Return a boolean series which is True for each date in this series which is
strictly between (i.e. not equal to) the corresponding dates in start and end.
Return a boolean series which is True for each date in this series which is
between or the same as the corresponding dates in start and end.
The same as is_on_or_between() above, but allows supplying a start/end date
pair as single argument.
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
Return the minimum value in the series for each patient (or NULL if the patient has no values).
Return the maximum value in the series for each patient (or NULL if the patient has no values).
class DateDifference(lhs, rhs) 🔗
Represents the difference between two date series (i.e. it is what you get when you subtract one date series from another)
The value of the date difference in days (can be positive or negative)
The value of the date difference in whole weeks (can be positive or negative)
The value of the date difference in whole calendar months (can be positive or negative)
The value of the date difference in whole calendar years (can be positive or negative)
class CodePatientSeries() 🔗
One row per patient series of type code
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
An alias for map_values which makes the intention clearer when working with
codelists. See codelist_from_csv().
class CodeEventSeries() 🔗
Multiple rows per patient series of type code
Return a boolean series comparing each value in this series with its
corresponding value in other.
Note that the result of comparing anything with NULL (including NULL itself) is NULL.
Return the inverse of == above.
Note that the same point regarding NULL applies here.
Return a boolean series which is True for each value in this series which is NULL, and False otherwise.
Return the inverse of is_null() above.
Replace any NULL value in this series with the corresponding value in other.
Note that other must be of the same type as this series.
Return a boolean series which is True for each value in this series which is
contained in other, where other can be any of the standard "container"
types: tuple, list, set, frozenset, or dict.
Return the inverse of is_in() above.
Accepts a dictionary mapping one set of values to another and applies that mapping to the series e.g.
status = status_code.map_values(
{1: "pending", 2: "accepted", 3: "completed"},
default="unknown"
)
An alias for map_values which makes the intention clearer when working with
codelists. See codelist_from_csv().
Return a integer patient series counting the number of distinct values for each patient in the series (ignoring any NULL values). Not that if a patient has no values at all in the series the result will be zero rather than NULL.
General🔗
class Dataset() 🔗
Defines the patients you want to include in your dataset and the variables you want to extract for each patient.
Every dataset definition file must define a Dataset() instance called dataset
like so:
dataset = Dataset()
Variables are added to the dataset as attributes, for example:
dataset.age = patients.age_on("2020-01-01")
Define the condition that patients must meet to be included in the Dataset, in the form of a boolean patient series e.g.
dataset.set_population(patients.date_of_birth < "1990-01-01")
case(*when_thens, default=None) 🔗
Take a sequence of condition-values of the form:
when(condition).then(value)
And evaluate them in order, returning the value of the first condition which
evaluates True. If no condition matches and a default is specified then return
that, otherwise return NULL.
For example:
category = case(
when(size < 10).then("small"),
when(size < 20).then("medium"),
when(size >= 20).then("large"),
default="unknown",
)
Note that because the conditions are evaluated in order we don't need the condition
for "medium" to specify (size >= 10) & (size < 20) because by the time the
condition for "medium" is being evaluated we already know the condition for "small"
is False.
codelist_from_csv(filename, column, category_column=None) 🔗
Read a codelist from a CSV file as either a list or a dictionary (for categorised codelists).
filename
Path to the file on disk, relative to the root of your repository. (Remember to use
UNIX/style/forward-slashes not Windows\style\backslashes.)
column
Name of the column in the CSV file which contains the codes.
category_column
Optional name of a column in the CSV file which contains categories to which each
code should be mapped. If this argument is passed then the resulting codelist will
be a dictionary mapping each code to its corresponding category. This can be passed
to the to_category() method to map a series of
codes to a series of categories.
class days(value) 🔗
Represents a duration of time specified in days
Return a boolean indicating whether the two durations have the same value and units.
Return a boolean indicating whether the two durations do not have the same value and units.
Add this duration to a date to produce a new date.
Alternatively two durations with the same units may be added to produce a new duration.
Subtract another duration of the same units from this duration.
Invert this duration so that rather that representing a movement, say, four weeks forwards in time it now represents a movement four weeks backwards.
Return a list of time intervals covering the duration starting on the supplied date. For example:
weeks(3).starting_on("2000-01-01")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
Return a list of time intervals covering the duration ending on the supplied date. For example:
weeks(3).ending_on("2000-01-15")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
class months(value) 🔗
Represents a duration of time specified in calendar months
Return a boolean indicating whether the two durations have the same value and units.
Return a boolean indicating whether the two durations do not have the same value and units.
Add this duration to a date to produce a new date.
Alternatively two durations with the same units may be added to produce a new duration.
Subtract another duration of the same units from this duration.
Invert this duration so that rather that representing a movement, say, four weeks forwards in time it now represents a movement four weeks backwards.
Return a list of time intervals covering the duration starting on the supplied date. For example:
weeks(3).starting_on("2000-01-01")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
Return a list of time intervals covering the duration ending on the supplied date. For example:
weeks(3).ending_on("2000-01-15")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
class weeks(value) 🔗
Represents a duration of time specified in weeks
Return a boolean indicating whether the two durations have the same value and units.
Return a boolean indicating whether the two durations do not have the same value and units.
Add this duration to a date to produce a new date.
Alternatively two durations with the same units may be added to produce a new duration.
Subtract another duration of the same units from this duration.
Invert this duration so that rather that representing a movement, say, four weeks forwards in time it now represents a movement four weeks backwards.
Return a list of time intervals covering the duration starting on the supplied date. For example:
weeks(3).starting_on("2000-01-01")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
Return a list of time intervals covering the duration ending on the supplied date. For example:
weeks(3).ending_on("2000-01-15")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
class years(value) 🔗
Represents a duration of time specified in calendar years
Return a boolean indicating whether the two durations have the same value and units.
Return a boolean indicating whether the two durations do not have the same value and units.
Add this duration to a date to produce a new date.
Alternatively two durations with the same units may be added to produce a new duration.
Subtract another duration of the same units from this duration.
Invert this duration so that rather that representing a movement, say, four weeks forwards in time it now represents a movement four weeks backwards.
Return a list of time intervals covering the duration starting on the supplied date. For example:
weeks(3).starting_on("2000-01-01")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
Return a list of time intervals covering the duration ending on the supplied date. For example:
weeks(3).ending_on("2000-01-15")
[
(date(2000, 1, 1), date(2000, 1, 7)),
(date(2000, 1, 8), date(2000, 1, 14)),
(date(2000, 1, 15), date(2000, 1, 21)),
]
Useful for generating the intervals arguments to Measures.
Measures🔗
Measures are used for calculating the ratio of one quantity to another as it varies over time, and broken down by different demographic (or other) groupings.
class Measures() 🔗
Define a collection of measures to be generated. Each measure definition file must
define a single Measures() instance called measures like so:
measures = Measures()
Add a measure to the list of measures to be generated.
name
The name of the measure, as a string.
numerator
The numerator definition, which must be a patient series but can be either
boolean or integer.
denominator
The denominator definition, which must be a patient series but can be either
boolean or integer.
group_by
Optional groupings to break down the results by. Must be supplied as a
dictionary of the form:
{
"group_name": group_definition,
...
}
intervals
A list of start/end date pairs over which to evaluate the measures. These can be
most conveniently generated using the starting_at()/ending_at() methods on
years, months, and weeks e.g.
intervals = months(12).starting_at("2020-01-01")
The numerator, denominator and intervals arguments can be omitted if
default values for them have been set using
define_defaults().
When defining several measures which share common arguments you can reduce repetition by defining default values for the measures.
Note that you can only define a single set of defaults and attempting to call this method more than once is an error.
INTERVAL 🔗
This is a placeholder value to be used when defining numerator, denominator and group_by columns in a measure. This allows these definitions to be written once and then be automatically evaluated over multiple different intervals. It can be used just like any pair of dates in ehrQL e.g.
clincial_events.date.is_during(INTERVAL)
Placeholder for the start date (inclusive) of the interval. Can be used like any other date e.g.
clinical_events.date.is_on_or_after(INTERVAL.start_date)
Placeholder for the end date (inclusive) of the interval. Can be used like any other date e.g.
clinical_events.date.is_on_or_before(INTERVAL.end_date)