Skip to content

Hospital admissions

Admitted Patient Care (APC) is the national data set for hospital admissions. Admitted Patient Care Spells (APCS) is part of Hospital Episode Statistics (HES) and is provided to OpenSAFELY via NHS Digital's Secondary Use Service (SUS).

Metadata🔗

  • Data provider NHS England.
  • Participation / Coverage In-hospital patient admissions in NHS Trusts in England.
  • Update frequency in OpenSAFELY Approximately monthly.
  • Delay between event occurring and event appearing in OpenSAFELY Approximately 1-2 months.
  • Collected information Admission and discharge dates; routes of admission; reason for admission; diagnoses; treatments; discharge destination.

Basic Background🔗

Each row is a in-hospital spell. In-hospital episodes cannot be queried.

  • An episode is a period of continuous care under a single responsible consultant.
  • A spell is a period of continuous care within a single trust.

Essentially, a spell is a collection of episodes (often just one episode) denoting the entirety of a patient's stay in hospital from admission to discharge or death. An episode denotes time within a spell spent under the responsibility of a single consultant. So for example if within your hospital stay you are transferred from a medical to surgical ward, or transferred to another site (within the same trust), there will be at least two episodes within your spell.

The dataset does not include any patients who have not been discharged (i.e., patients who have an incomplete spell) at the data extraction date. In other words, the data will be missing admissions of the patients with the longest stay for the most recent admissions.

* Notes on transfers:🔗

  • Sometimes it is desirable to exclude patients admitted via transfer from another provider, e.g. if we are looking for readmissions we probably want to exclude them because it was not the original admission date of the readmission.
  • The post-transfer spell may have additional diagnoses or procedures of interest so it is often useful to include them.
  • Transfers may occasionally come from private hospitals where a patient's condition worsened to require e.g. intensive care; if the initial spell was not NHS funded we would probably not have the original spell in our data.

Available filters and returning fields🔗

admission_date🔗

Filtering🔗

  • Use on_or_before and/or on_or_after, or between.
  • Optionally specify find_first_match_in_period (or last match). NB any ties (multiple same day admissions) are resolved using APCS_Ident which does not predictably give the earliest or latest admission on a given day.

Returning🔗

returning="admission_date"

number_of_matches_in_period🔗

Filtering🔗

  • Not currently available

Returning🔗

  • returning="number_of_matches_in_period"

discharge_date🔗

Filtering🔗

  • Not currently available

Returning🔗

  • returning="discharge_date"

Caution🔗

  • discharge date is present for patients who were transferred, or died, at the end of their spell
  • patients who are still in hospital are not included in the dataset so this field is always complete but long-stay patients may be missing.

with_these_diagnoses / with_these_primary_diagnoses/ primary_diagnosis🔗

  • Coding system: ICD-10.

Filtering🔗

  • Filter using with_these_diagnoses / with_these_primary_diagnoses

Returning🔗

  • returning="primary_diagnosis"

Caution🔗

  • Primary diagnosis is not necessarily the primary reason for admission, and could represent an escalation/complication of initial reason for admission.

with_these_procedures🔗

Coding system: OPCS-4.

Filtering🔗

  • Filter using with_these_proecedures

Returning🔗

  • Not currently available

admission_method🔗

Filtering🔗

  • Filter using with_admission_method= (supply either an individual code or a list)

Returning🔗

  • returning="admission_method"

Most useful for:🔗

  • identifying planned vs emergency admissions
  • identifying/excluding transfers* ('28', '2B', '81')
  • excluding birth and maternity spells ('2C', '82', '83', '31', '32')

Examples🔗

# all normal admissions and transfers, excluding births and maternity admissions
with_admission_method = ['11', '12', '13', '21', '2A', '22', '23', '24', '25', '2D', '28', '2B', '81']

# all normal admissions excluding transfers, births and maternity admissions
with_admission_method = ['11', '12', '13', '21', '2A', '22', '23', '24', '25', '2D']

Brief Reference table🔗

Codes Admission method
11, 12, 13 Elective (planned/booked/waiting list)
21 Emergency Admission: from A&E
2A Emergency Admission: from A&E of another provider
22, 23, 24, 25, 2D Emergency Admission: other sources (e.g. GP, MH crisis team)
28 Emergency Admission: Other - NB can include transfers and births
2B Transfer (emergency)
81 Transfer (non-emergency)
2C, 82, 83 Birth of baby (in hospital or e.g. at home)
31, 32 Maternity Admission: Mother admitted ante or post partum

source_of_admission🔗

Filtering🔗

  • Filter using with_source_of_admission= (supply either an individual code or a list)

Returning🔗

  • returning="source_of_admission"

Most useful for:🔗

  • identifying patients admitted from care homes ('54', '65', '85', '88')
  • excluding birth spells ('79')
  • identifying transfers* - but this cannot distinguish transfers from wards vs patients arriving from the A&E dept of another provider (e.g. an MIU or smaller hospital which doesn't have a suitable ward) so admission_method is better for this purpose.

Example🔗

with_source_of_admission = ['19','29','54', '65', '85', '88', '39', '40', '41', '42', '66', '49', '52', '53', '87', '98', '99']
# all admissions except births (includes unknowns (98, 99))

Brief Reference table🔗

code Source of admission
19 Usual place of residence or no fixed abode (but should exclude care homes)
29 Temporary place of residence (e.g. hotel, school)
54, 65, 85, 88 Care Home / hospice
51 NHS other Hospital Provider - ordinary (physical care) ward or A&E
49, 52, 53 NHS other Hospital Provider - psychiatric accommodation, maternity ward/neonates or MH/LD ward
39, 40, 41, 42, 66 Other (Penal establishment, Court, or Police Station / Police Custody Suite, foster care)
79 Babies born in or on the way to hospital
87 Non NHS run hospital
98, 99 Default/unknown

patient_classification🔗

Most useful for:🔗

  • excluding regular admissions (i.e. where patient is admitted e.g. every week for chemotherapy or dialysis)
  • excluding mother and baby admissions (where using delivery facilities only)
  • distinguishing elective day case admissions from ordinary admissions (day case admissions are for planned procedures not requiring overnight stays)

An potential advantage of this field is that it reflects what actually happened during the spell rather than what was planned. If a patient's treatment was not limited to the planned course, e.g. patient was expected to have their routine chemotherapy but illness arose which required further treatment and an overnight stay, this would be coded as 1 (ordinary admission) rather than as 3 (regular admission).

Brief Reference table🔗

code patient classification
1 Ordinary admission (all emergency admissions and elective admissions expected to last at least one night)
2 Day case admission (patient admitted electively, not expected to stay overnight and who returns home on the same day as scheduled. If discharge does not occur same day as planned, this should be counted as an ordinary admission)
3, 4 Regular day or night admission (part of a planned series of regular admissions for an on-going regime of broadly similar treatment. If the admission lasts 24 hours or more, such an admission should be classified as an ordinary admission)
5 Mother and baby using delivery facilities only

discharge_destination🔗

Most useful for:🔗

  • Identifying/excluding patients who died at the end of their spell
  • Identifying/excluding patients who were transferred at the end of their spell
  • Identifying patients who were discharged to a care home (whether or not they were previously resident in one)

Example:🔗

with_discharge_destination = ['19', '29', '51', '52', '30', '49', '50', '53', '48', '84', '87', '37', '38', '66', '54', '65', '85', '88', '98','99']
# everyone discharged alive or transferred to another provider (includes those with an unknown destination '98','99')

with_discharge_destination = ['19', '29', '30', '49', '50', '53', '48', '84', '87', '37', '38', '66', '54', '65', '85', '88']
# everyone discharged to the community, alive (excludes those with an unknown destination '98','99')
# excludes people transferred to other NHS providers of physical health care (51, 52)
# but includes patients transferred to psychiatric / secure units and non-NHS hospitals, because
# these post-transfer spells are most likely not present in SUS and any subsequent exacerbation
# of physical symptoms would result in a new spell in SUS.

Note - include '79' for patients who died (or were stillborn) Include 98 and 99 for unknowns. There will also be some null values.

Brief Reference table🔗

Code Description
19 Usual place of residence unless listed below. It includes 'no fixed abode' but excludes care homes.
29 Temporary place of residence (includes hotel, residential Educational Establishment)
51, 52 Transfer: NHS other Hospital Provider - general ward or maternity/ neonates
30, 49, 50, 53 Transfer (other NHS):, mentally ill / Learning Disabilities, psychiatric units
48, 84, 87 Transfer: Non-NHS run hospital or medium secure unit, Scotland high-secure unit
37, 38, 66 Other: Court, Penal establishment, police station, foster care
54, 65, 85, 88 Hospice / Care Home
79 Not applicable - PATIENT died or stillbirth

Reference🔗

Link to lookup table

administrative_category🔗

Most useful for:🔗

  • Identifying/excluding patients who were not undergoing NHS treatment

admission_treatment_function_code🔗

Most useful for:🔗

  • Identifying which specialty patients were admitted under (use with caution)

days_in_critical_care🔗

Most useful for:🔗

  • Enumerating how many days were spent in critical care

Caution🔗

  • Definition of critical care may vary between trusts
  • This is counted in number of days (or part-days) not the number of nights as per normal "length of stay" calculations. E.g. a patient who was admitted on day 1 and discharged on day 2 and spent at least part of both days in critical care will have cc_days=2 but LOS=1.

duration_of_elective_wait🔗

Most useful for:🔗

  • assessing how long patient was waiting for elective treatment (use with caution)

More information🔗