ehrQL language features

This reference is structured as a series of examples.

The intended audience is primarily:

researchers
software developers

that already have some understanding of how the ehrQL works.

Info

Please refer to the introduction and tutorial documentation sections if you need more explanation of the underlying concepts behind ehrQL.

How the examples work🔗

Each individual example demonstrates a specific ehrQL feature in isolation.

Every example here consists of:

Headings and subheadings that summarise the feature being demonstrated.
A small example data input table containing entirely fictitious variables and values.
- The table has a single-letter name referred to throughout the example
  - e for event-level table
  - p for patient-level table.
- The columns of input tables use a name constructed from a single letter with a number to create an identifier — for example, i1. The single letter in the identifier refers to the column's data type:
  - a b column contains Boolean values
  - a c column contains electronic health record codes (the codes used in this reference are fictitious, for example: abc)
  - a d column contains dates
  - an i column contains integers
  - an s column contains strings
- Both table and column names are written with code formatting throughout this reference.
An ehrQL query that extracts some data from the example table. Like the table names, ehrQL queries are displayed here with code formatting.
The resulting output from the ehrQL query, displayed as another table, to demonstrate the query's effect

Note

The examples here are automatically generated from ehrQL's specification tests.

1 Filtering an event frame🔗

1.1 Including rows🔗

1.1.1 Where with column🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	b1
1	101	T
1	102	T
1	103
2	201	T
2	202
2	203	F
3	301
3	302	F

e.where(e.b1).i1.sum_for_patient()

returns the following patient series:

patient	value
1	203
2	201
3

1.1.2 Where with expr🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	111
1	102	112
1	103	113
2	201	211
2	202	212
2	203	213
3	301

e.where((e.i1 + e.i2) < 413).i1.sum_for_patient()

returns the following patient series:

patient	value
1	306
2	201
3

1.1.3 Where with constant true🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201

e.where(True).count_for_patient()

returns the following patient series:

patient	value
1	2
2	1

1.1.4 Where with constant false🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201

e.where(False).count_for_patient()

returns the following patient series:

patient	value
1	0
2	0

1.1.5 Chain multiple wheres🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	b1
1	1	T
1	2	T
1	3	F

e.where(e.i1 >= 2).where(e.b1).i1.sum_for_patient()

returns the following patient series:

patient	value
1	2

1.2 Excluding rows🔗

1.2.1 Except where with column🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	b1
1	101	T
1	102	T
1	103
2	201	T
2	202
2	203	F
3	301	T
3	302	T

e.except_where(e.b1).i1.sum_for_patient()

returns the following patient series:

patient	value
1	103
2	405
3

1.2.2 Except where with expr🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	111
1	102	112
1	103	113
2	201	211
2	202	212
2	203	213
3	301

e.except_where((e.i1 + e.i2) < 413).i1.sum_for_patient()

returns the following patient series:

patient	value
1
2	405
3	301

1.2.3 Except where with constant true🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201

e.except_where(True).count_for_patient()

returns the following patient series:

patient	value
1	0
2	0

1.2.4 Except where with constant false🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201

e.except_where(False).count_for_patient()

returns the following patient series:

patient	value
1	2
2	1

2 Picking one row for each patient from an event frame🔗

2.1 Picking the first or last row for each patient🔗

2.1.1 Sort by column pick first🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
1	103
2	203
2	202
2	201

e.sort_by(e.i1).first_for_patient().i1

returns the following patient series:

patient	value
1	101
2	201

2.1.2 Sort by column pick last🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
1	103
2	203
2	202
2	201

e.sort_by(e.i1).last_for_patient().i1

returns the following patient series:

patient	value
1	103
2	203

2.2 Sort by more than one column and pick the first or last row for each patient🔗

2.2.1 Sort by multiple columns pick first🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	3
1	102	2
1	102	1
2	203	1
2	202	2
2	202	3

e.sort_by(e.i1, e.i2).first_for_patient().i2

returns the following patient series:

patient	value
1	3
2	2

2.2.2 Sort by multiple columns pick last🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	3
1	102	2
1	102	1
2	203	1
2	202	2
2	202	3

e.sort_by(e.i1, e.i2).last_for_patient().i2

returns the following patient series:

patient	value
1	2
2	1

2.3 Picking the first or last row for each patient where a column contains NULLs🔗

2.3.1 Sort by column with nulls and pick first🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1
1	102
1	103
2	203
2	202
2

e.sort_by(e.i1).first_for_patient().i1

returns the following patient series:

patient	value
1
2

2.3.2 Sort by column with nulls and pick last🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1
1	102
1	103
2	203
2	202
2

e.sort_by(e.i1).last_for_patient().i1

returns the following patient series:

patient	value
1	103
2	203

2.4 Mixing the order of `sort_by` and `where` operations🔗

2.4.1 Sort by before where🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	1
1	102	2
1	103	2
2	203	1
2	202	2
2	201	2

e.sort_by(e.i1).where(e.i1 > 102).first_for_patient().i1

returns the following patient series:

patient	value
1	103
2	201

2.4.2 Sort by interleaved with where🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2
1	101	1
1	102	2
1	103	2
2	203	1
2	202	2
2	201	2

e.sort_by(e.i1).where(e.i2 > 1).sort_by(e.i2).first_for_patient().i1

returns the following patient series:

patient	value
1	102
2	201

3 Aggregating event and patient frames🔗

3.1 Determining whether a row exists for each patient🔗

3.1.1 Exists for patient on event frame🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	b1
1
2
3

patient	b1
1
1
2

e.exists_for_patient()

returns the following patient series:

patient	value
1	T
2	T
3	F

3.1.2 Exists for patient on patient frame🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	b1
1
2
3

patient	b1
1
1
2

p.exists_for_patient()

returns the following patient series:

patient	value
1	T
2	T
3	T

3.2 Counting the rows for each patient🔗

3.2.1 Count for patient on event frame🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	b1
1
2
3

patient	b1
1
1
2

e.count_for_patient()

returns the following patient series:

patient	value
1	2
2	1
3	0

3.2.2 Count for patient on patient frame🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	b1
1
2
3

patient	b1
1
1
2

p.count_for_patient()

returns the following patient series:

patient	value
1	1
2	1
3	1

4 Aggregating event series🔗

4.1 Minimum and maximum aggregations🔗

4.1.1 Minimum for patient🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
1	103
2	201
2
3

e.i1.minimum_for_patient()

returns the following patient series:

patient	value
1	101
2	201
3

4.1.2 Maximum for patient🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
1	103
2	201
2
3

e.i1.maximum_for_patient()

returns the following patient series:

patient	value
1	103
2	201
3

4.2 Sum aggregation🔗

4.2.1 Sum for patient🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
1	103
2	201
2
2	203
3

e.i1.sum_for_patient()

returns the following patient series:

patient	value
1	306
2	404
3

4.3 Mean aggregation🔗

4.3.1 Mean for patient integer🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1
1	1	1.1
1	2	2.1
1	3	3.1
2
2	2	2.1
2	3	3.1
3

e.i1.mean_for_patient()

returns the following patient series:

patient	value
1	2.0
2	2.5
3

4.3.2 Mean for patient float🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1
1	1	1.1
1	2	2.1
1	3	3.1
2
2	2	2.1
2	3	3.1
3

e.f1.mean_for_patient()

returns the following patient series:

patient	value
1	2.1
2	2.6
3

4.4 Count distinct aggregation🔗

4.4.1 Count distinct for patient integer🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1	s1	d1
1	101	1.1	a	2020-01-01
1	102	1.2	b	2020-01-02
1	103	1.5	c	2020-01-03
2	201	2.1	a	2020-02-01
2	201	2.1	a	2020-02-01
2	203	2.5	b	2020-02-02
3	301	3.1	a	2020-03-01
3	301	3.1	a	2020-03-01
3
3
4

e.i1.count_distinct_for_patient()

returns the following patient series:

patient	value
1	3
2	2
3	1
4	0

4.4.2 Count distinct for patient float🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1	s1	d1
1	101	1.1	a	2020-01-01
1	102	1.2	b	2020-01-02
1	103	1.5	c	2020-01-03
2	201	2.1	a	2020-02-01
2	201	2.1	a	2020-02-01
2	203	2.5	b	2020-02-02
3	301	3.1	a	2020-03-01
3	301	3.1	a	2020-03-01
3
3
4

e.f1.count_distinct_for_patient()

returns the following patient series:

patient	value
1	3
2	2
3	1
4	0

4.4.3 Count distinct for patient string🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1	s1	d1
1	101	1.1	a	2020-01-01
1	102	1.2	b	2020-01-02
1	103	1.5	c	2020-01-03
2	201	2.1	a	2020-02-01
2	201	2.1	a	2020-02-01
2	203	2.5	b	2020-02-02
3	301	3.1	a	2020-03-01
3	301	3.1	a	2020-03-01
3
3
4

e.s1.count_distinct_for_patient()

returns the following patient series:

patient	value
1	3
2	2
3	1
4	0

4.4.4 Count distinct for patient date🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	f1	s1	d1
1	101	1.1	a	2020-01-01
1	102	1.2	b	2020-01-02
1	103	1.5	c	2020-01-03
2	201	2.1	a	2020-02-01
2	201	2.1	a	2020-02-01
2	203	2.5	b	2020-02-02
3	301	3.1	a	2020-03-01
3	301	3.1	a	2020-03-01
3
3
4

e.s1.count_distinct_for_patient()

returns the following patient series:

patient	value
1	3
2	2
3	1
4	0

5 Combining series🔗

5.1 Combining two patient series🔗

5.1.1 Patient series and patient series🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	102
2	201	202

p.i1 + p.i2

returns the following patient series:

patient	value
1	203
2	403

5.2 Combining a patient series with a value🔗

5.2.1 Patient series and value🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201

p.i1 + 1

returns the following patient series:

patient	value
1	102
2	202

5.2.2 Value and patient series🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201

1 + p.i1

returns the following patient series:

patient	value
1	102
2	202

5.3 Combining two event series🔗

5.3.1 Event series and event series🔗

This example makes use of an event-level table named e containing the following data:

patient	i1	i2	s1
1	101	111	b
1	102	112	a
2	201	211	b
2	202	212	a

(e.i1 + e.i2).sum_for_patient()

returns the following patient series:

patient	value
1	426
2	826

5.3.2 Event series and sorted event series🔗

The sort order of the underlying event series does not affect their combination.

This example makes use of an event-level table named e containing the following data:

patient	i1	i2	s1
1	101	111	b
1	102	112	a
2	201	211	b
2	202	212	a

(e.i1 + e.sort_by(e.s1).i2).minimum_for_patient()

returns the following patient series:

patient	value
1	212
2	412

5.4 Combining an event series with a patient series🔗

5.4.1 Event series and patient series🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	i1
1	101
2	201

patient	i1
1	111
1	112
2	211
2	212

(e.i1 + p.i1).sum_for_patient()

returns the following patient series:

patient	value
1	425
2	825

5.4.2 Patient series and event series🔗

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	i1
1	101
2	201

patient	i1
1	111
1	112
2	211
2	212

(p.i1 + e.i1).sum_for_patient()

returns the following patient series:

patient	value
1	425
2	825

5.5 Combining an event series with a value🔗

5.5.1 Event series and value🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201
2	202

(e.i1 + 1).sum_for_patient()

returns the following patient series:

patient	value
1	205
2	405

5.5.2 Value and event series🔗

This example makes use of an event-level table named e containing the following data:

patient	i1
1	101
1	102
2	201
2	202

(1 + e.i1).sum_for_patient()

returns the following patient series:

patient	value
1	205
2	405

6 Operations on all series🔗

6.1 Testing for equality🔗

6.1.1 Equals🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	101
2	201	202
3	301
4

p.i1 == p.i2

returns the following patient series:

patient	value
1	T
2	F
3
4

6.1.2 Not equals🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	101
2	201	202
3	301
4

p.i1 != p.i2

returns the following patient series:

patient	value
1	F
2	T
3
4

6.1.3 Is null🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	101
2	201	202
3	301
4

p.i1.is_null()

returns the following patient series:

patient	value
1	F
2	F
3	F
4	T

6.1.4 Is not null🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	101
2	201	202
3	301
4

p.i1.is_not_null()

returns the following patient series:

patient	value
1	T
2	T
3	T
4	F

6.2 Testing for containment🔗

6.2.1 Is in🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201
3	301
4

p.i1.is_in([101, 301])

returns the following patient series:

patient	value
1	T
2	F
3	T
4

6.2.2 Is not in🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201
3	301
4

p.i1.is_not_in([101, 301])

returns the following patient series:

patient	value
1	F
2	T
3	F
4

6.3 Map from one set of values to another🔗

6.3.1 Map values🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201
3	301
4

p.i1.map_values({101: "a", 201: "b", 301: "a"}, default="c")

returns the following patient series:

patient	value
1	a
2	b
3	a
4	c

6.4 Replace missing values🔗

6.4.1 If null then integer column🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201
3	301
4

p.i1.if_null_then(0)

returns the following patient series:

patient	value
1	101
2	201
3	301
4	0

6.4.2 If null then boolean column🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	101
2	201
3	301
4

p.i1.is_in([101, 201]).if_null_then(False)

returns the following patient series:

patient	value
1	T
2	T
3	F
4	F

7 Operations on boolean series🔗

7.1 Logical operations🔗

7.1.1 Not🔗

This example makes use of a patient-level table named p containing the following data:

patient	b1
1	T
2
3	F

~p.b1

returns the following patient series:

patient	value
1	F
2
3	T

7.1.2 And🔗

This example makes use of a patient-level table named p containing the following data:

patient	b1	b2
1	T	T
2	T
3	T	F
4		T
5
6		F
7	F	T
8	F
9	F	F

p.b1 & p.b2

returns the following patient series:

patient	value
1	T
2
3	F
4
5
6	F
7	F
8	F
9	F

7.1.3 Or🔗

This example makes use of a patient-level table named p containing the following data:

patient	b1	b2
1	T	T
2	T
3	T	F
4		T
5
6		F
7	F	T
8	F
9	F	F

p.b1 | p.b2

returns the following patient series:

patient	value
1	T
2	T
3	T
4	T
5
6
7	T
8
9	F

8 Operations on integer series🔗

8.1 Arithmetic operations without division🔗

8.1.1 Negate🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	111
2	201

-p.i2

returns the following patient series:

patient	value
1	-111
2

8.1.2 Add🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	111
2	201

p.i1 + p.i2

returns the following patient series:

patient	value
1	212
2

8.1.3 Subtract🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	111
2	201

p.i1 - p.i2

returns the following patient series:

patient	value
1	-10
2

8.1.4 Multiply🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	111
2	201

p.i1 * p.i2

returns the following patient series:

patient	value
1	11211
2

8.1.5 Multiply with constant🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	111
2	201

10 * p.i2

returns the following patient series:

patient	value
1	1110
2

8.2 Comparison operations🔗

8.2.1 Less than🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	201
2	201	201
3	301	201
4		201

p.i1 < p.i2

returns the following patient series:

patient	value
1	T
2	F
3	F
4

8.2.2 Less than or equal to🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	201
2	201	201
3	301	201
4		201

p.i1 <= p.i2

returns the following patient series:

patient	value
1	T
2	T
3	F
4

8.2.3 Greater than🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	201
2	201	201
3	301	201
4		201

p.i1 > p.i2

returns the following patient series:

patient	value
1	F
2	F
3	T
4

8.2.4 Greater than or equal to🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	i2
1	101	201
2	201	201
3	301	201
4		201

p.i1 >= p.i2

returns the following patient series:

patient	value
1	F
2	T
3	T
4

9 Operations on all series containing codes🔗

9.1 Testing for containment using codes🔗

9.1.1 Is in🔗

This example makes use of a patient-level table named p containing the following data:

patient	c1
1	123000
2	456000
3	789000
4

p.c1.is_in([SNOMEDCTCode("123000"), SNOMEDCTCode("789000")])

returns the following patient series:

patient	value
1	T
2	F
3	T
4

9.1.2 Is not in🔗

This example makes use of a patient-level table named p containing the following data:

patient	c1
1	123000
2	456000
3	789000
4

p.c1.is_not_in([SNOMEDCTCode("123000"), SNOMEDCTCode("789000")])

returns the following patient series:

patient	value
1	F
2	T
3	F
4

9.1.3 Is in codelist csv🔗

This example makes use of a patient-level table named p containing the following data:

patient	c1
1	123000
2	456000
3	789000
4

p.c1.is_in(codelist)

returns the following patient series:

patient	value
1	T
2	F
3	T
4

9.2 Test mapping codes to categories using a categorised codelist🔗

9.2.1 Map codes to categories🔗

This example makes use of a patient-level table named p containing the following data:

patient	c1
1	123000
2	456000
3	789000
4

p.c1.to_category(codelist)

returns the following patient series:

patient	value
1	cat1
2
3	cat2
4

10 Logical case expressions🔗

10.1 Logical case expressions🔗

10.1.1 Case with expression🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	6
2	7
3	8
4	9
5

case(
    when(p.i1 < 8).then(p.i1),
    when(p.i1 > 8).then(100),
)

returns the following patient series:

patient	value
1	6
2	7
3
4	100
5

10.1.2 Case with default🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	6
2	7
3	8
4	9
5

case(
    when(p.i1 < 8).then(p.i1),
    when(p.i1 > 8).then(100),
    default=0,
)

returns the following patient series:

patient	value
1	6
2	7
3	0
4	100
5	0

10.1.3 Case with boolean column🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1	b1
1	6	T
2	7	F
3	9	F
4

case(
    when(p.b1).then(p.i1),
    when(p.i1 > 8).then(100),
)

returns the following patient series:

patient	value
1	6
2
3	100
4

11 Operations on all series containing dates🔗

11.1 Operations which apply to all series containing dates🔗

11.1.1 Get year🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

p.d1.year

returns the following patient series:

patient	value
1	1990
2	2000
3

11.1.2 Get month🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

p.d1.month

returns the following patient series:

patient	value
1	1
2	3
3

11.1.3 Get day🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

p.d1.day

returns the following patient series:

patient	value
1	2
2	4
3

11.1.4 To first of year🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-12-15
3	2020-12-31
4

p.d1.to_first_of_year()

returns the following patient series:

patient	value
1	1990-01-01
2	2000-01-01
3	2020-01-01
4

11.1.5 To first of month🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	1990-01-31
3

p.d1.to_first_of_month()

returns the following patient series:

patient	value
1	1990-01-01
2	1990-01-01
3

11.1.6 Add days🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

p.d1 + days(p.i1)

returns the following patient series:

patient	value
1	1990-04-12
2	2000-09-20
3

11.1.7 Subtract days🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

p.d1 - days(p.i1)

returns the following patient series:

patient	value
1	1989-09-24
2	1999-08-17
3

11.1.8 Add months🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	2003-01-29	1
2	2004-01-29	1
3	2003-01-31	1
4	2004-01-31	1
5	2004-03-31	-1
6	2000-10-31	11
7	2000-10-31	-11

p.d1 + months(p.i1)

returns the following patient series:

patient	value
1	2003-03-01
2	2004-02-29
3	2003-03-01
4	2004-03-01
5	2004-03-01
6	2001-10-01
7	1999-12-01

11.1.9 Add years🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	2000-06-15	5
2	2000-06-15	-5
3	2004-02-29	1
4	2004-02-29	-1
5	2004-02-29	4
6	2004-02-29	-4
7	2003-03-01	1

p.d1 + years(p.i1)

returns the following patient series:

patient	value
1	2005-06-15
2	1995-06-15
3	2005-03-01
4	2003-03-01
5	2008-02-29
6	2000-02-29
7	2004-03-01

11.1.10 Add date to duration🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	i1
1	1990-01-02	100
2	2000-03-04	200
3

days(100) + p.d1

returns the following patient series:

patient	value
1	1990-04-12
2	2000-06-12
3

11.1.11 Difference between dates in years🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	2020-02-29
2	2020-02-28
3	2019-01-01
4	2021-03-01
5	2023-01-01
6

(date(2021, 2, 28) - p.d1).years

returns the following patient series:

patient	value
1	0
2	1
3	2
4	-1
5	-2
6

11.1.12 Difference between dates in months🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	d2
1	2000-02-28	2000-01-30
2	2000-03-01	2000-01-30
3	2000-03-28	2000-02-28
4	2000-03-30	2000-01-30
5	2000-02-27	2000-01-30
6	2000-01-27	2000-01-30
7	1999-12-26	2000-01-27
8	2005-02-28	2004-02-29
9	2010-01-01	2000-01-01
10	2000-01-01

(p.d1 - p.d2).months

returns the following patient series:

patient	value
1	0
2	1
3	1
4	2
5	0
6	-1
7	-2
8	11
9	120
10

11.1.13 Difference between dates in days🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	d2
1	2000-01-01	2000-01-01
2	2000-03-01	2000-01-01
3	2001-03-01	2001-01-01
4	1999-12-31	2001-01-01

(p.d1 - p.d2).days

returns the following patient series:

patient	value
1	0
2	60
3	59
4	-367

11.1.14 Reversed date differences🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-30
2	1970-01-15

(p.d1 - "1980-01-20").years

returns the following patient series:

patient	value
1	10
2	-11

11.1.15 Add days to static date🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	10
2	-10

date(2000, 1, 1) + days(p.i1)

returns the following patient series:

patient	value
1	2000-01-11
2	1999-12-22

11.1.16 Add months to static date🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	10
2	-10

date(2000, 1, 1) + months(p.i1)

returns the following patient series:

patient	value
1	2000-11-01
2	1999-03-01

11.1.17 Add years to static date🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	10
2	-10

date(2000, 1, 1) + years(p.i1)

returns the following patient series:

patient	value
1	2010-01-01
2	1990-01-01

11.2 Comparisons involving dates🔗

11.2.1 Is before🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_before(date(2000, 1, 1))

returns the following patient series:

patient	value
1	T
2	F
3	F
4

11.2.2 Is on or before🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_on_or_before(date(2000, 1, 1))

returns the following patient series:

patient	value
1	T
2	T
3	F
4

11.2.3 Is after🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_after(date(2000, 1, 1))

returns the following patient series:

patient	value
1	F
2	F
3	T
4

11.2.4 Is on or after🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_on_or_after(date(2000, 1, 1))

returns the following patient series:

patient	value
1	F
2	T
3	T
4

11.2.5 Is in🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_in([date(2010, 1, 1), date(1900, 1, 1)])

returns the following patient series:

patient	value
1	F
2	F
3	T
4

11.2.6 Is not in🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	1990-01-01
2	2000-01-01
3	2010-01-01
4

p.d1.is_not_in([date(2010, 1, 1), date(1900, 1, 1)])

returns the following patient series:

patient	value
1	T
2	T
3	F
4

11.2.7 Is between but not on🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	2010-01-01
2	2010-01-02
3	2010-01-03
4	2010-01-04
5	2010-01-05
6

p.d1.is_between_but_not_on(date(2010, 1, 2), date(2010, 1, 4))

returns the following patient series:

patient	value
1	F
2	F
3	T
4	F
5	F
6

11.2.8 Is on or between🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	2010-01-01
2	2010-01-02
3	2010-01-03
4	2010-01-04
5	2010-01-05
6

p.d1.is_on_or_between(date(2010, 1, 2), date(2010, 1, 4))

returns the following patient series:

patient	value
1	F
2	T
3	T
4	T
5	F
6

11.2.9 Is during🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	2010-01-01
2	2010-01-02
3	2010-01-03
4	2010-01-04
5	2010-01-05
6

p.d1.is_during(interval)

returns the following patient series:

patient	value
1	F
2	T
3	T
4	T
5	F
6

11.2.10 Is on or between backwards🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1
1	2010-01-01
2	2010-01-02
3	2010-01-03
4	2010-01-04
5	2010-01-05
6

p.d1.is_on_or_between(date(2010, 1, 4), date(2010, 1, 2))

returns the following patient series:

patient	value
1	F
2	F
3	F
4	F
5	F
6

11.3 Types usable in comparisons involving dates🔗

11.3.1 Accepts python date object🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	d2
1	1990-01-01	1980-01-01
2	2000-01-01	1980-01-01
3	2010-01-01	2020-01-01
4		2020-01-01

p.d1.is_before(datetime.date(2000, 1, 20))

returns the following patient series:

patient	value
1	T
2	T
3	F
4

11.3.2 Accepts iso formated date string🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	d2
1	1990-01-01	1980-01-01
2	2000-01-01	1980-01-01
3	2010-01-01	2020-01-01
4		2020-01-01

p.d1.is_before("2000-01-20")

returns the following patient series:

patient	value
1	T
2	T
3	F
4

11.3.3 Accepts another date series🔗

This example makes use of a patient-level table named p containing the following data:

patient	d1	d2
1	1990-01-01	1980-01-01
2	2000-01-01	1980-01-01
3	2010-01-01	2020-01-01
4		2020-01-01

p.d1.is_before(p.d2)

returns the following patient series:

patient	value
1	F
2	F
3	T
4

12 Operations on all series containing strings🔗

12.1 Testing whether one string contains another string🔗

12.1.1 Contains fixed value🔗

This example makes use of a patient-level table named p containing the following data:

patient	s1
1	ab
2	ab12
3	12ab
4	12ab45
5	a b
6	AB
7

p.s1.contains("ab")

returns the following patient series:

patient	value
1	T
2	T
3	T
4	T
5	F
6	F
7

12.1.2 Contains fixed value with special characters🔗

This example makes use of a patient-level table named p containing the following data:

patient	s1
1	/a%b_
2	/ab_
3	/a%bc
4	a%b_

p.s1.contains("/a%b_")

returns the following patient series:

patient	value
1	T
2	F
3	F
4	F

12.1.3 Contains value from column🔗

This example makes use of a patient-level table named p containing the following data:

patient	s1	s2
1	ab	ab
2	cd12	cd
3	12ef	ef
4	12gh45	gh
5	i j	ij
6	KL	kl
7		mn
8	ab

p.s1.contains(p.s2)

returns the following patient series:

patient	value
1	T
2	T
3	T
4	T
5	F
6	F
7
8

12.1.4 Contains value from column with special characters🔗

This example makes use of a patient-level table named p containing the following data:

patient	s1	s2
1	/a%b_	/a%b_
2	/ab_	/a%b_
3	/a%bc	/a%b_
4	a%b_	/a%b_

p.s1.contains(p.s2)

returns the following patient series:

patient	value
1	T
2	F
3	F
4	F

13 Defining the dataset population🔗

13.1 Defining a population🔗

define_population is used to limit the population from which data is extracted.

13.1.1 Population with single table🔗

Extract a column from a patient table after limiting the population by another column.

This example makes use of a patient-level table named p containing the following data:

patient	b1	i1
1	F	10
2	T	20
3	F	30

p.i1
define_population(~p.b1)

returns the following patient series:

patient	value
1	10
3	30

13.1.2 Population with multiple tables🔗

Limit the patient population by a column in one table, and return values from another table.

This example makes use of a patient-level table named p and an event-level table named e containing the following data:

patient	i1
1	10
2	20
3	0

patient	i1
1	101
1	102
3	301
4	401

e.exists_for_patient()
define_population(p.i1 > 0)

returns the following patient series:

patient	value
1	T
2	F

13.1.3 Case with case expression🔗

Limit the patient population by a case expression.

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	6
2	7
3	9
4

p.i1
define_population(
    case(
        when(p.i1 <= 8).then(True),
        when(p.i1 > 8).then(False),
    )
)

returns the following patient series:

patient	value
1	6
2	7

14 Defining a table using inline data🔗

14.1 Defining a table using inline data🔗

14.1.1 Table from rows🔗

This example makes use of a patient-level table named p containing the following data:

patient	i1
1	10
2	20
3	30

p.i1 + t.n

returns the following patient series:

patient	value
1	110
2
3	330

ehrQL language features

How the examples work🔗

1 Filtering an event frame🔗

1.1 Including rows🔗

1.1.1 Where with column🔗

1.1.2 Where with expr🔗

1.1.3 Where with constant true🔗

1.1.4 Where with constant false🔗

1.1.5 Chain multiple wheres🔗

1.2 Excluding rows🔗

1.2.1 Except where with column🔗

1.2.2 Except where with expr🔗

1.2.3 Except where with constant true🔗

1.2.4 Except where with constant false🔗

2 Picking one row for each patient from an event frame🔗

2.1 Picking the first or last row for each patient🔗

2.1.1 Sort by column pick first🔗

2.1.2 Sort by column pick last🔗

2.2 Sort by more than one column and pick the first or last row for each patient🔗

2.2.1 Sort by multiple columns pick first🔗

2.2.2 Sort by multiple columns pick last🔗

2.3 Picking the first or last row for each patient where a column contains NULLs🔗

2.3.1 Sort by column with nulls and pick first🔗

2.3.2 Sort by column with nulls and pick last🔗

2.4 Mixing the order of sort_by and where operations🔗

2.4.1 Sort by before where🔗

2.4.2 Sort by interleaved with where🔗

3 Aggregating event and patient frames🔗

3.1 Determining whether a row exists for each patient🔗

3.1.1 Exists for patient on event frame🔗

3.1.2 Exists for patient on patient frame🔗

3.2 Counting the rows for each patient🔗

3.2.1 Count for patient on event frame🔗

3.2.2 Count for patient on patient frame🔗

4 Aggregating event series🔗

4.1 Minimum and maximum aggregations🔗

4.1.1 Minimum for patient🔗

4.1.2 Maximum for patient🔗

4.2 Sum aggregation🔗

4.2.1 Sum for patient🔗

4.3 Mean aggregation🔗

4.3.1 Mean for patient integer🔗

4.3.2 Mean for patient float🔗

4.4 Count distinct aggregation🔗

4.4.1 Count distinct for patient integer🔗

4.4.2 Count distinct for patient float🔗

4.4.3 Count distinct for patient string🔗

4.4.4 Count distinct for patient date🔗

5 Combining series🔗

5.1 Combining two patient series🔗

5.1.1 Patient series and patient series🔗

5.2 Combining a patient series with a value🔗

5.2.1 Patient series and value🔗

5.2.2 Value and patient series🔗

5.3 Combining two event series🔗

5.3.1 Event series and event series🔗

5.3.2 Event series and sorted event series🔗

5.4 Combining an event series with a patient series🔗

5.4.1 Event series and patient series🔗

5.4.2 Patient series and event series🔗

5.5 Combining an event series with a value🔗

5.5.1 Event series and value🔗

5.5.2 Value and event series🔗

6 Operations on all series🔗

6.1 Testing for equality🔗

6.1.1 Equals🔗

6.1.2 Not equals🔗

6.1.3 Is null🔗

6.1.4 Is not null🔗

6.2 Testing for containment🔗

6.2.1 Is in🔗

6.2.2 Is not in🔗

6.3 Map from one set of values to another🔗

6.3.1 Map values🔗

6.4 Replace missing values🔗

6.4.1 If null then integer column🔗

6.4.2 If null then boolean column🔗

7 Operations on boolean series🔗

7.1 Logical operations🔗

7.1.1 Not🔗

2.4 Mixing the order of `sort_by` and `where` operations🔗