Hospital Admitted Patient Care Activity 2020-21

The HES publications focus on headline information about hospital activity. Each annual publication includes a series of national tables and also provider-level breakdowns for some main areas. Most data included in the published tables are aggregate counts of hospital activity. Where averages are published, e.g. average length of stay for inpatients or caesarean rates for maternity statistics, these data are clearly labelled stating how the data has been calculated.

The accuracy of HES data is the responsibility of the NHS providers who submit the data to SUS. These data are required to be accurate to enable them to be correctly paid for the activity they undertake. NHS Digital has a well-developed data quality assurance process for the SUS and HES data. It uses an xml schema to ensure some standardisation of the data received. The use of the schema means that the data set has to meet certain validation rules before it can be submitted to SUS. NHS Digital leads on the schema changes and consults the data suppliers about proposed changes.

Each month NHS Digital makes data quality dashboards available to NHS providers to show the completeness and validity of their data submissions to SUS. This helps to highlight any issues present in the provisional data allowing time for corrections to be made before the annual data are submitted.

An external auditor, acting on behalf of the Department of Health and Social Care (DHSC), audits the data submitted to SUS to ensure NHS providers are being correctly paid by PbR for the care they provide.

NHS Digital validates and cleans the HES extract and derives new items. The team discusses data quality issues with the information leads in hospital trusts who are responsible for submitting data. The roles and responsibilities within NHS Digital are clear for the purposes of data quality assurance, i.e. to assess the quality of data received against published standards and report the results.

Data quality information for each year to date HES data set is published alongside the provisional year to date HES data, and also alongside annual publications. These specify known data quality issues each year, e.g. if a trust has a known shortfall of secondary diagnoses. The statisticians can only check the validity and format of the data and not whether it is accurate, as accuracy checking requires a level of audit capacity and capability which NHS Digital does not currently possess. There is also further information about HES data quality published.

NHS Digital also publishes a regular Data Quality Maturity Index for providers across several datasets including HES.The UK Statistics Authority conducted case studies of quality assurance and audit arrangements of administrative data sources. HES was used as a case study and further information can be found in the published report (Annex C, case study 3).

The HES APC 2020-21 data set includes records of admitted patient episodes collected from 491 providers in England.

Table 3 provides a count and percentage of FCE records that have valid data in specific key fields.

Table 3: Count of FCEs with a valid entry in fields

Collection of HES data is carried out on a monthly basis throughout the financial year, with a final annual refresh (AR) once the year end has passed. Each monthly collection refreshes data back to the start of the financial year.

Providers are no longer offered the opportunity to re-submit data for HES after the submission deadline (top-up files), which was approved by partners on the Data Quality Steering Group (DQSG).  This has enabled us to commit to publishing Annual data earlier than for previous years.  Providers were informed about this decision emphasising the importance to submit data correctly before the inclusion date of .

Table 4 shows the change from the Month 13 provisional data and the final annual refresh data. 

Table 4: Comparing month 13 and annual refresh data

Table 5 shows the number of FCEs occurring in each month in the last three submission periods of 2020-21, including annual refresh. The number of records per month of activity generally increases as more submissions are made; the completeness of the data increases over time. 

Table 5: Monthly variation in submitted FCE records

Months 1 to 13 are published in the 'Provisional Monthly HES for Admitted Patient Care, Outpatient and Accident and Emergency Data - April 2020 to March 2021 - Data completeness' document.  

Of the 34 critical care fields collected in HES, only 14 are mandatory. A summary of completeness of some key fields is shown in the table below, also available as Table 16 in the accompanying Data Tables file (note that, of those fields shown, only ‘unit function’ is mandatory and the others are optional). The table also shows completeness of three key fields in the associated APC episode records. Completion rates are broadly similar to those for the previous year (see Table 16 in the accompanying Data Tables file).

Most raw critical care records in the data set are associated with only one record of an APC episode, but some are associated with multiple episodes as an APC spell; to avoid multi-counting these critical care records, an algorithm has been applied to link each critical care record to the APC episode record against which it is the ‘best match’. Because of limitations in the algorithm, a small number of raw critical care records are identified as a ‘best match’ with more than one APC episode and the final presented numbers of critical care records slightly over-count the true number of critical care periods.

A more detailed discussion of the algorithm and its impact is presented in the accompanying Technical Guide, and Table 17 in the accompanying data file shows a breakdown of application of the first step of the best match algorithm.

The table below illustrates the impact of applying the best match algorithm against the raw critical care / APC episode records to remove duplicates from the analysis:

Table 6: Number of Critical Care records ending in 2020-21

Note that the count of number of patients and distinct critical care periods before and after the application of the best match algorithm is different prior to 2013-14. The reason for this is that a change in the data structure of HES has taken place which means that records with no support day data recorded are excluded from the best match setting process.

Much data associated with HES APC records, including associated critical care records, is not reliably complete until the episode or critical care period is finished. Critical care periods associated with hospital APC episodes which are still ongoing at the end of the period of coverage (i.e. at the end of March 2021) will not be reported into the system until the following financial year, when they have finished, and are therefore not included for analysis. As a result, reported figures are lower than the true figures for the full financial year, and, in particular, analysis by month (including the Table 1 in the ACC tables) will present artificially depressed figures for March.

Figures for March each year are artificially depressed as any critical care period associated with a hospital APC episode which is still on-going at the end of March will not be reported into the system until the following financial year, when it has finished.

The critical care database includes fields recording the time of day at which the critical care period started and ended, albeit that the start time field has been completed in only 215,425 of the 235,262 critical care records used for analysis.

Although the times submitted are not specified to be rounded, it is clear from analysis of the data that there are rounding peaks at each multiple of 5 minutes past the hour, larger peaks at multiples of 10 minutes (and, for end times, at 59 minutes past), still larger peaks at multiples of 15 minutes, a yet larger peak at 30 minutes past the hour and the largest peak at 0 minutes past the hour; this suggests that in many cases the times are being rounded to 5, 10, 15, 30 or 60 minute boundaries.

HES data are published as early as possible. The production of the underlying annual HES data sets takes several months after the reference period. The final submission deadline for NHS providers to send annual data to SUS is normally at the end of May, almost two months after that year has finished. It then takes approximately two months to produce the HES APC data set and a further two months to complete publication production and data investigation.

In addition to annual data NHS Digital also publish provisional monthly HES data approximately two months after the reference period.

The final annual data includes additional data cleaning, validation and processing than the provisional monthly data.

Since August 2010, first the Department of Health (DH) and now NHS England has conducted censuses of the numbers of available and occupied critical care beds on the last Thursday of each month; prior to August 2010, similar censuses were conducted by DH twice a year.  These reports can be found at: https://www.england.nhs.uk/statistics/statistical-work-areas/bed-availability-and-occupancy/

Although the number of critical care beds on a given day is not directly comparable with the number of critical care periods which covered that day, a comparison between the two reveals that there are substantial differences between reporting by NHS providers to NHS England’s censuses and to the Critical Care Minimum Dataset and thus in the HES data warehouse from which this publication’s data is drawn.

In particular, there were a number of hospital providers which submitted critical care bed data to NHS England’s census but not to HES; figures reported in this publication are therefore thought to represent an undercount of the true critical care situation; conversely there were a far smaller number of providers for which there is a critical care data in HES but which did not submit any data to NHS England’s census. 

During the COVID-19 pandemic NHS England's census was suspended. As a result Table 18, which provided a comparison of NHS Census numbers and HES Critical Care Numbers has been omitted from this years report.

Hospital data for the other administrations can be found at: 

• Northern Ireland – Hospital Statistics
• Scotland –Hospital Care 
• Wales – Health and social care statistics

NHS England also publish hospital activity data.

HES and similar statistics from the devolved administrations are used to contribute to World Health Organisation (WHO), Organisation for Economic Co-operation and Development (OECD) and Eurostat compendiums on health statistics.

HES data are available from 1989-90 onwards. Changes to the figures over time need to be interpreted in the context of improvements in data quality and coverage (particularly in earlier years), improvements in coverage of independent sector activity (particularly from 2006-07) and changes in NHS practice. 

Payment by Results (PbR) is a system whereby hospitals are paid for the number of patient treatments, known as activity, they perform and the complexity of these treatments. It was introduced in a phased way from 2003-04 onwards. In order to get paid correctly, hospitals need to record the activity they perform and the clinical codes that outline the patients’ conditions and treatment.

The introduction of Payment by Results (PbR), increased private sector involvement in the delivery of secondary care and brought about some changes in clinical practice (including some procedures occurring as outpatient appointments instead of hospital admissions). It is likely that these changes will have affected trends.

This has provided a major financial incentive for hospitals to ensure all of the activity they perform and the clinical coding is fully recorded. This improved recording of information captured by HES could be one of the factors leading to the reported activity increases.

In order to manage patients’ waiting times there has been the need for additional elective operations to be performed as well as a requirement for more capacity in NHS funded care to perform this activity. In the middle of the last decade, additional capacity was brought in from the private sector via treatment centres, with the NHS funding some patients to be treated there for routine operations.

Improvements in technology and the need to increase efficiency to allow more patients to be treated have led to a reduction in the length of time patients need to stay in hospital for certain planned operations. In particular, many of those operations that would have involved an overnight stay at the start of the period are now routinely performed as day cases. In addition, many operations where a patient would have been admitted to hospital at the start of the period are now routinely performed in outpatients. This has led to increases in day case rates and outpatient attendances over the period.

The recent period has also seen a rise in the number of emergency admissions. One factor contributing to this is likely to be the increased demand on health services from an ageing population. Alongside this there has been the introduction of observation or medical assessment units at many hospitals to which patients arriving in A&E departments are admitted, often for around a day, to enable observation and tests to be performed on them.

Care should be taken when interpreting these changes, as improvements in coverage in HES will contribute alongside growth from increased activity.