HOSPITAL EPISODE STATISTICS (HES)

Introduction

The data for the Department of Health Hospital Episode Statistics data warehouse are extracted from the Commissioning Data Set at the NHS-Wide Clearing Service (NWCS).

Hospital Episode Statistics and what they are used for

The Hospital Episode Statistics required by the Department of Health cover every finished CONSULTANT, NURSE and MIDWIFE episode in England (including regular day and night admissions) within the financial year, from 1 April to 31 March. Finished episodes must contain all the relevant clinical data. Hospital Episode Statistics also includes an Annual Census of episodes unfinished at midnight on 31 March, and the Psychiatric Census, a subset of the Annual Census, which contains additional data items. Hospital Episode Statistics will also cover out-patient and accident and emergency data backdated to 1 April 2003, extracted from the Out-Patent Attendance and Accident and Emergency Attendance Commissioning Data Set Types. The data warehouse represents an invaluable national source of information about patterns of treatment in hospitals throughout England as well as providing epidemiological data about diseases and operative procedures.

Hospital Episode Statistics data are published annually and are also used to feed into other published statistics including the Compendium of Clinical and Health Indicators (formerly the Public Health Common Dataset) and the Performance Indicators. Within the Department of Health, the main uses of the data include policy development, resource allocation, performance management, accountability to public and parliament and monitoring of health and healthcare variations. In addition, Hospital Episode Statistics data are widely used by clinical and other researchers, both within and outside the Department of Health. There is pressure to increase both the timeliness and completeness of the Hospital Episode Statistics data set.

Further information can be obtained from the Hospital Episode Statistics website: http://www.dh.gov.uk/PublicationsAndStatistics/Statistics/HospitalEpisodeStatistics/fs/en

How Hospital Episode Statistics data are processed

Data records must be lodged with the NHS-Wide Clearing Service regularly and routinely, preferably on a monthly basis.

Extracts for the Hospital Episode Statistics data warehouse are taken at prearranged times each quarter and these dates are published on the ClearNET website and elsewhere. http://clearnet.mhapp.nhs.uk/iclearnet.html

These quarterly extracts are generally taken one month apart in order to reduce the burden on the NHS. A&E data will be extracted to a similar timetable, initially six months in arrears but moving to quarterly extracts.

Data is extracted as cumulative quarters throughout the period 1 April - 31 March and the entire year's data is taken again - as an 'annual refresh' - approximately eight weeks after the end of the year for admitted patient data and 12 weeks after the end of the year for out-patient data.

It is expected that at each of the extract dates the records are as complete as possible both in terms of the overall activity and the completion of the relevant data items. Unfinished and psychiatric census episodes for the whole year, which form part of the admitted patient data, are taken at the same time as the annual refresh.

Before being incorporated into the main Hospital Episode Statistics data warehouse, all data are subject to a complex sequence of checks, as follows:

Verification
For finished admitted patient episodes, the NWCS service provider selects episodes that contain an end date within the data year. To be accepted for Hospital Episode Statistics, a record must contain an appropriate Hospital Provider Code relating to that data year. A record which fails this check will be rejected. Similar criteria apply for the extraction of out-patient and A&E Commissioning Data Set data in that the event must have occurred within the extract period.

Derivation
The data extracted is used singly or in combination to derive additional information, such as the PATIENT's age group or the Strategic Health Authority of treatment to facilitate the interrogation and analysis of the data warehouse.

Manual Cleaning (Annual Data Only)
Under exceptional circumstances, the processing of the annual data can be halted to make manual changes to the data. This is carried out on a Trust specific basis and requires the agreement of the Trust(s) concerned and the Hospital Episode Statistics section. Manual cleaning is extremely resource and time intensive and is only undertaken when there would otherwise be a significant impact on the integrity of the dataset. The details of the manual cleaning process are on the Hospital Episode Statistics website (see below for the website address).

Autocleaning
Various fields within the record are analysed to see whether the entries make sense on their own, and with reference to related fields. In some cases it is possible to overwrite incorrect entries by deriving the data from other fields within the record. If this cannot be done, the incorrect entry may be overwritten with the accepted code for `not known/not applicable'.

Further information

If you want to find out more on Hospital Episode Statistics processing, the Hospital Episode Statistics team at the Department of Health publish a number of documents, one of which details the autocleaning and derivation routines outlined above.

The Hospital Episode Statistics team can be contacted at:

Department of Health Hospital Episode Statistics Service
Skipton House - Room 430B
80, London Road
London
SE1 6LH
 
Tel: 020 7972 5529
Fax: 020 7972 5662
Department of Health Website - Hospital Episode Statistics.