Health and Safety Executive
(Who submit the data on behalf of the National Statistics Institute)

Website:
http://www.hse.gov.uk/

Statistics and Epidemiology Unit

Sub-website:
http://www.hse.gov.uk/statistics/index.htm

Health and Safety Executive,
Redgrave Court,
Merton Road,
Bootle. UK.
L20 7HS

The defintions, scope and key indicators are the same as for EU-level, although for a longer time period (in many cases it is possible to go-back to 1974, representing the commencement of key UK legislation - the Health and Safety at Work etalso include estimates of self-reported injuries from the UK Labour Force Survey (LFS), hence providing an overall scale for a given situation. LFS estimates have been available since around 1990.

A summary of these two sc, Act 1974).

Data is primarily collected thorugh employer reports to the enforcing authorities, of fatal and non-fatal injuries in the workplace. Reporting is required under UK legislation, primrily for the purposes of health and safety enforcement by professional inspectors. Use of these reports for statistical purposes is therefore secondary.

Phase 1 variables - There is no 'size of enterprise' or 'nationality of victim' (both are optional). Weights are calculated for non-fatal injuries, to allow for under-reporting.
Phase 2 variables: Of the nine possible variables, three provided within the scope of the Regulation are: Deviation; Contact mode of Injury; and Working Process.

Non-fatal injuries are under-reported. Statistics are normally presented 'as reported' with no adjustment, although reference is made in outputs. To counteract the effect on these statistics of under-reporting, where possible complementary outputs sources is published at
http://www.hse.gov.uk/statistics/sources.htm

As described at 'EU Level', the UK use classification systems that largely mirror these EU and other international systems.

Equivalent UK systems are
SIC 2007 (maps to NACE Rev2);
SOC 2010 (maps to ISCO 08);
Local authority, and derived regional aggregates (maps to NUTS 2013);
Employment status (maps to ISCE-93).

See:
http://www.hse.gov.uk/statistics/industry/sic2007.htm
http://www.hse.gov.uk/statistics/soc2010.htm

From 2016 (2014 data year), coverage is all economic sectors (NACE A to U) and occupations (ISCO).

Prior to this date, coverage excluded aircrew and sailors.

COVERAGE OF ECONOMIC SECTORS (NACE), (Annual update expected)

Additional comments on coverage of economic sectors

Codes:

y Sector fully covered
n Sector not covered at all
p Secor partially covered
n.a. Not applicable, i.e. sector does not exist in country

The definition of an accident, and the term 'work-related', is the same as for EU.
Deaths are reportable if the person dies up to one year after the accident.
Exclusions are the same as EU-level.

Variables covered meet the Regulation in scope and definitions, and are the same as EU-level, except:
Phase 1 variables - There is no 'size of enterprise' or 'nationality of victim' (both are optional). Weights are calculated for non-fatal injuries.
Phase 2 variables: Of the nine possible variables, three provided within the scope of the Regulation are: Deviation; Contact mode of Injury; and Working Process. No 'cause and circumstance' weights are calculated, so a default of 1 is used.

COVERAGE OF ESAW VARIABLES (Annual update expected)

Additional comments on coverage of ESAW variables

Codes:

Coverage
y Variable fully covered
n Variable not covered at all, i.e. variable is so far not collected in country
p Variable partially covered

Days lost
y num Days lost are covered and in data file listed by numbers (004 - 182)
y cat Days lost are covered and in data file listed by categories (A01 - A06)
p num, p cat Days lost are partly covered and listed by numbers (categories)

Type of weight
U weight to correct under-reporting
S weight to account for sampling
Sp weight to correct special sampling

Data are collected for each accident.

If a person is a victim of more than one accident during the reference year, several cases are reported (one for each accident).

If there are several victims in the same event of accident, also several cases are reported (one for each victim).

For employment types covered: Full coverage of employees (this includes family members, students, work placement and others if they are 'at work').

For the self-employed, fatalities are fully covered, with non-fatalities being significantly under-reported.

From 2014 data year, coverage is all industries (NACE) and occupations (ISCO). Prior to this date, coverage excluded aircrew and sailors.

COVERAGE OF PROFESSIONAL STATUS (Annual update expected)

Additional comments on coverage of professionnal status

Codes:

y Professional status fully covered
n Professional status not covered at all
p Professional status partially covered
n.a. Not applicable, i.e. professional status does not exist in country

COVERAGE OF ACCIDENTS OUTSIDE PREMISES (Annual update expected)

Additional comments on coverage of accidents outside premises

Codes:

y Accidents fully covered
n Accidents not covered at all
p Accidents partially covered
F Data of fatal road traffic accidents and fatal accidents on board of any means of transport provided in the main CSV data file
A Data of fatal and non-fatal road traffic accidents as well as fatal and non-fatal accidents on board of any means of transport provided in the main CSV data file

Since 2013 (reference year 2011), coverage is at UK level, meaning Great Britain as well as Northern Ireland.

Prior to this date, data coverage is GB only.

At a detailed level, outputs are available by local authority; and from which higher aggregates are calculated, for example NUTS 3.

Data are available since reference year 1994 for all EU-15 Member States, from 1995 also for Norway.

In addition, Croatian data are included since 2010.

Data for Switzerland was added from 2004 and Icelandic data from 2012.

Not applicable.

The following measurement units are used in ESAW data:

- Numbers of accidents

- Percentages of accidents (in relation to different totals and breakdowns)

- Incidence rates of accidents: number of accidents per 100,000 workers

- Standardised incidence rates: number of accidents per 100,000 workers adjusted for the relative sizes of economic sectors at EU level (see section 18.6 Adjustment for more details)

The calendar year during which the accidents were reported to have taken place (reference year).

Relevant reporting legislation in the UK is RIDDOR - The Reporting of Injuries, Diseases and Dangerous Occurrences Regulations.

This places a requirement on employers to report workplace deaths and non-fatal injuries to the enforcing authorities.

RIDDOR is supplemented by data arrangements with the transport authorities responsible for road, air and sea.

For legal interpretation and reporting, see:
http://www.hse.gov.uk/riddor/index.htm

Not applicable.

As at 'EU' level. In addition:

There is a Confidentiality Policy (which also covers data security) available on the website. In summary, this says:
All of our data is handled, stored and accessed in a manner which complies with Government and Departmental standards regarding security and confidentiality, and fully meets the requirements of the Data Protection Act. Access to this data is controlled by a system of passwords and strict business need access control.

See:
http://www.hse.gov.uk/statistics/about/confidentiality.htm

To avoid the disclosure of personal information through statistical outputs, disclosure control is implemented where deemed necessary, especially where small counts are involved.

In certain circumstances, the main treatment is suppression of very small counts (usually <=3) of non-fatal injuries.

This is to avoid identifying an individual.

Data released in the UK is done in accordance with the UK Statistics Authority (UKSA). Regular formal assessment by the UKSA confirms the statistics as 'National Statistics', meaning they are of high quality and integrity, by meeting the needs of the user, and released in an orderly, pre-announced and impartial manner. Statistics comply with the UKSA Code of Practice.

See
http://www.hse.gov.uk/statistics/about/index.htm

https://www.gov.uk/government/statistics/

Data released in the UK is done in accordance with the UK Statistics Authority (UKSA). Regular formal assessment by the UKSA confirms the statistics as 'National Statistics', meaning they meet the needs of the user, and are released in an orderly, pre-announced and impartial manner.

See
http://www.hse.gov.uk/statistics/about/index.htm

Annual

In accordance with the UK Code of Pactrice for Official Statistics, the statistics are released by formal advanced notice on the HSE and UK Stats Authority websites.

See:
http://www.hse.gov.uk/statistics/index.htm

Given the wide range and depth of these statistics, a variety of web-based publication formats are used (no charges are made to users).

These are normally accessed through the homepage of our statistics website, see:

http://www.hse.gov.uk/statistics/index.htm
Within this homepage, are several user-friendly Booklets giving top-level statistics and interpretation, and directing users if they require more detailed further reading.

Top-level results and commentary are presented as conventional web pages, with further analysis contained in stand-alone PDF documents. For more experienced users and those requiring specific data, supporting Excel files containing aggregated tables are also provided. PDF and Excel files are viewable by anyone, using freely-available software.

For confidentiality reasons, suppression rules are employed to avoid inappropriate disclosure. A HSE statistical confidentiality policy is publicly available. See:
http://www.hse.gov.uk/statistics/about/confidentiality.htm

Not currently available.

Until 2015 there was a very comprehensive searchable online database of workplace fatal and non-fatal injuries, at a record-level.

This allowed users to make an infinite choice of selections and tabulations to suite their specific needs. For cost and technical reasons this was withdrawn, however we are actively looking at cost-effective replacements.

'4500 monthly

ESAW data is not available as microdata.

Not applicable

Not available

Documentation on methodologies, metadata and quality are made available via the website.

As above

As above

Quality is at the heart of UK statistics and professional competence. Statisticians are members of the UK Government Statistical Service, and attend and fully contribute to the annual ESAW Working Group meetings. The independent UK Statistics Authority require the statistics comply with their Code of Practice for Official Statistics.

A statement of Quality Guidelines is publicly available on the website.

It describes at a top-level the six European measures (relevance; accuracy; timeliness; comparbility; accessibility; coherence).

See: 
http://www.hse.gov.uk/statistics/about/quality-guidelines.htm

There is also a detailed Quality Report published, describing the key features of the Quality, along very similar lines as this Quality Report.

See :
http://www.hse.gov.uk/statistics/pdf/riddor-background-quality-report.pdf

As with the EU, continuous improvement is a default position.

These statistics comply with the UK Statistics Authority Code of Practice, and are classified as ;'National Statistics' - this is a robust and highly-regarded independent measure that the statistics meet the needs of users.

The statistics conform to the ESAW Regulation and definitions, so are EU-harmonised.

The main weakness is under-reporting of non-fatalities. However, apart from differential reporting by type of industry, there is no evidence the data is otherwise biased.

One key trade-off is timeliness of release. Normal release is about six months after the end of the year of collection; any sooner would result in under-copunts due to late reports being missed; also it would not allow sufficient time for data processing and interpretation. Any later would cause frustration with users and the data becoming outdated sooner.

These statistics in the UK comply with the UK Statistics Authority 'Code of Practice (CoP) for Official Statistics'.

The CoP requires us to periodically review user needs for health and safety statistics and the relevance and use of the data we provide. We do consult with both internal and external users on a regular basis.

Methods of user engagement include several user conferences. We also make available on the website a feedback form, to allow users to comment on any aspect of the statistical outputs. In addition, we have a popular electronic email bulletin service, providing regular information to users and reminding them of forthcoming releases and other relevant key events.

Typical users are similar as at EU level. The UK user base has built-up over many years, for example direct contact with regular users, both internally and externally.

Any fundamental changes to statistics are pre-announced where possible, and user feedback encouraged.

The is a User Engagement policy, at
http://www.hse.gov.uk/statistics/about/engagement/index.htm

Users are encouraged to provide feedback via the channels described above.

Formal consultations are summarised on the website, as well as follow-up action where appropriate.

User satisfaction surveys are carried-out periodically, and results published on the website.

See metadata file in annex which lists for all countries in detail the available variables, economic sectors, employment status and information on traffic accidents.

All mandatory Phase 1 variables are provided. Currently no data is provided on Nationality or Size.
To meet the ESAW requirement, three Phase 3 variables out of nine are provided - Contact Mode of Injury; Deviation; and Working Process.

REPORTING LEVELS OF ECONOMIC SECTORS (NACE), (Annual update expected)

Global reporting level

Reporting levels by sector

Additional comments on global reporting level

Codes:

N sector not covered by data collection
UNK sector covered but unknown reporting level
(value %) reporting level
(e) rough estimate (instead of "medium" or "high")

Explanation: Please fill in every cell with information, either "N" for sector not covered by the data collection, "UNK" for sectors covered but unknown reporting level, or a percentage value for sectors covered for which the reporting level is known (such as "52%"). Instead of "medium" or "high" please fill in a approximate value and flag it with "(e)" at the end, such as "50% (e)". Please fill in both, the global reporting cells and the detailed rows for individual sectors. If for example no sector is covered for your country, then fill into every cell "0", if all sectors are covered but all reporting levels are unknown fill in "UNK" in every cell. "N/A" should be replaced by either "N" or "UNK".

All data are published ‘as-is’, with no adjustment to counts of accidents (i.e. no weightings applied, no estimations). Non-fatal injuries are under-reported, although there is no known under-coverage. Self-employed are generally covered by teh same reporting legislation as employees, however under-reporting by the self-employed is more pronounced. For this reason no self-employed non-fatality data is supplied to ESAW, although it is disseminated within the UK.

Quality assurance is carried out on all data. The initial quality focus is to make sure that the top-level number of reports is in line with expectations (severity by employment status). Subsequent quality assurance examines individual fields in detail to ensure that any unexpected changes or data values are understood and represent genuine changes.

Fatalities are not subject to under-reporting, as all are investigated by the authorities, and the data scrutinised in detail.

For non-fatalities, outputs clearly refer to the issue of under-reporting (as described elsewhere in this file). As the majority of non-fatalities are not investigated, the data is presented 'as reported by employers', subject to some logical validation checks described under 'statistical processing'.

Weighting procedures are used to make inferences from the sample to the entire eligible population, taking account of differential non-response among different sub-groups in the population.

Estimates based on the LFS sample will be subject to sampling error.

As non-fatal injuries are subject to under-reporting, the weightings are based on a survey (LFS), hence subject to sampling error. Differetial reporting by is apparent by industry, with some industries having better reporting than others. In this case reporting levels are calculated by industry section level. Where weightings are applied, estimates of the sampling error of the weighting are identified by the calculation and use of confidence intervals.

Estimates of the non-fatal 'days lost' variable are based on LFS estimates, hence subject to random variation and calculation of confidence intervals based on the LFS sampling framework.

Where sampling error is relevant, confidence intervals are calculated and presented alongside the statistics. 

The UK uses employer reports, although they are subject to under-reporting of non-fatalities.

General information about under-reporting errors can be found in other sections of this file.

Weights are applied to correct for under-reporting.

UK reports are not considered to suffer from non-sampling error.

Information about the coverage of variables, economic sectors, employment types and occupations can be found in the metadata annex to this file.

No over-coverage.

Not applicable

Not applicable

Not applicable

Not applicable

Not applicable

No known processing errors.

Not applicable.

Statistics are based on the date the accident took place, and the year of release is normally 1 April to 31 March. Fatality data is released roughly three months after the year-end; and non-fatality data about six months after.

This process is well-established and understood by producers and users.

Data provided to ESAW is based on the calendar year, and complies with the schedule as described at 'EU Level'.

As above

As above

All UK-released data has been published on-time in accordance with pre-annouced release dates.

Data is supplied to Eurostat by the legal deadline, usually in the month of June N+2.

Delivery as above.

Data on fatal and non-fatal accidents have a high level of comparability between all countries within the UK, as well as across UK regions and local authorities.

Incidence rates (standardised in some cases) enable comparisons to be made between countries and regions.

Not applicable

There is a long time-series of data covering at least 30 individual years.

In earlier years of this time series, some data was not collected for specific industries (although numerically has little effect on the statistics), however for at least the last 15 years there has been full consistency in coverage.

There was a change in UK reporting legislation in 2012, which resulted in the data being modelled to mitigate the effect on the statistics.

This modelling quantified the scale of change, and by complementing with the UK LFS, enables trends to be maintained.

Where formal classification changes have taken place, for example industry, data has been computer re-coded from 'old to new codes', to enable a back-series to be maintained.

As above

Data is also coherent across domains, such that comparisons can be made between this data and the complementary UK LFS. 

Not applicable

Not applicable

Internal domain comparisons are possible: for example between Regions; Industry; Occupation; Age; Gender.

Comparisons are possibble due to use of denominator data (employment) to calculate incident rates.

RIDDOR data in this context is referred to as ‘administrative data’. That is, reporting under RIDDOR is a legal requirement, so the data is already available for statistical use. Consequently, there is minimal cost and burden to the respondent (that is, the duty-holder), in providing the data for statistical purposes.

As described in the quality measures elswhere in this file, there is ongoing work carried out by statistical staff in HSE in processing, analysing and presenting the data. The staff costs associated with this work is approximately £150k each year. Other costs are not available, as often they are absorbed in wider work (for example formal assessment of compliance with the UKSA Code of Practice).

There is ongoing work to ensure processes are automated where possible, to minimise human input (which is often the main cost). For example input validation of online reporting forms are routinely updated, ensuring maximum levels of validation are captured at source. Automation is also maximised where possible, in reducing human error. For example the recent introduction of an SQL warehouse, to enable more efficient handling of larger datasets and automating existing manual routines.

There is a Data Revision Policy at
http://www.hse.gov.uk/statistics/about/revisions/index.htm

This allows for revisions to be made based on one of four gemeral themes:
Changes in methodology; Changes in employment estimes and/or re-weightings; Changes due to late reports; Changes to reflect processing errors.

The two main reasons for changes are:
Late reports, usually resulting in a slight increase on previously published figures; and re-weigting of employment estimaters, often due to re-alignment with updated Census estimate.

Revisions due to late reports are typically no more than about 1% increase; and no more than about 1% decrease in rates because emplyoment estimates get revised upwards by this amount.

Accident data collected is based on the EU-level description (item 3, legal reports made by employers to the enforcing authorities). The reporting requirement is RIDDOR - the Reporting of Injuries, Diseases and Dangerous Occurrences Regulations. It is therefore administrative data, the primary purpose being to alert the authorities to an incident, with a view to further operational activity.

As RIDDOR legislation does not cover most road traffic acccidents, such record-level data is sourced from the UK Department for Transport (DfT) for those persons injured whilst travelling as part of work. The primary purpose of the DfT data is statistical, which in turn is sourced from police administrative accident reports and investigations. DfT data is added to RIDDOR data in compiling the dataset.

Similarly administrative data for air and sea accidents is sourced from air and marine accident authorities. Administrative data for Northern Ireland is sourced from HSENI.

The above compilation ensures all workers are covered (all NACE 2-digits and ISCO 08), and all geographical component parts of the UK (NUTS 3).

Non-fatal injuries are known to be under-reported, so the data is weighted accordingly using the UK LFS injury estimates as a base of full reporting.

The numbers employed are provided by the UK Office for National Statistics to Eurostat, and based on the UK LFS Annual Population Survey.

Collected daily; fatalities released quarterly; non-fatals annually.

Provided to Eurostat annually.

Data is administrative.

Employers are required by law to report workplace deaths and injuries.

The source data is validated using all of the technical methods descibed by ESAW, manual and automated. The data is also sense-checked with intelligence from key operational staff; trends are monitored at a detailed level for changes over time; expected changes are used to help validate (eg increases due to weather; decreases due to holiday seasons; changes due to changes in the economy). Outliers, missing values and invalid cross-combinations are investigated.

‘Hard’ checks stop notifiers from entering incorrect data, with a warning message to explain what is wrong so that they can correct it, for example an incident date in the future. For the statistical dataset, analysts perform data cleaning checks on data items that are not covered adequately by the database internal validation checks. All analysts have considerable expertise of RIDDOR and IT skills, for example whether certain types of incident are technically reportable or not (if not, the records are removed).

Basic checks are performed on file size and characteristics to make sure they are in line with expectation. Detailed quality assurance checks are run on the raw data such as analysis of the new dataset against previous datasets, looking for changes in means; levels of change in each data item; the incidence of missing values; changes in categorical variables; extreme values.

Where possible records that require re-coding due to inaccuracies are done en-block by automation. It also often happens that batches of records require case-by-case examination of certain data fields. In both scenarios a judgement is made by analysts as to the correct codes to use. Where necessary additional expert opinion on a case is obtained from operational and policy colleagues.

Tabulations are produced and compared with the outputs generated during the same period the previous year. We look at the statistics from an operational and policy point of view to ensure that any unexpected changes or data values are understood and represent genuine changes. Where uncertainties exist in data outputs, these are highlighted and if possible an explanation given.

The source data is administrative. As non-fatal injuiries are under-reported, weights are calculated and these weights supplied to Eurostat alongside the dataset.

Some non-fatal data is imputed, scaling-up to account for accidents resulting in 4 to 7 days' off work, hence aligning with the ESAW requirement of over-3-day injuries. Imputation is done on the basis of a proportion of reported incidents, validated against the LFS estimates of over-3-day injuries. Modelling shows this imputation has no effect on the final estimates (that is, the risk profiles of over-7-day accidents is no different to those of over-3-day accidents).

Data is normally presented un-standardisesd, but is standardised by NACE in some circumstances.

To calculate the weights, the UK Labour Force Survey (LFS) is used as the reference of over-3-day injuries. Each weight is calculated according to ESAW methodolgy, such that diffewrnt weights are applied to different NACE sectors, on the basis that under-reporting differs depending on the industry sector. Fatal injuries are full-reported so not subject to uprating.

Not applicable.

No seasonal adjustments are made.

Not applicable.

Data on 'size of enterprise' and 'nationality of victim' are not currently available.

For Phase 3 variables, of the maximum of nine possible ones, three are provided: Contact Mode of Injury; Deviation; and Working Process. The Regulation requires at least three, and the choice of variables at the discretion of the Member State. Technical research is underway to see if additional variables can be provided.

UK Derogations:
(i) Road traffic accidents: from 2015 (2013 data year), the UK has provided data on persons injured or killed whilst working on the road.
(ii) Accidents to aircrew and sailors whilst working in the air or at sea: This data will be provided from 2016 (in relation to 2014 data).
(iii) 'Days Lost' variable - details on non-fatal time off work was provided as from 2015 (2013 data year).

Analysis of trends, including the effect of the recent recession, and subsequent recovery:
Detailed analysis is always ongoing, especially around the changes in the UK economy and effect on accident rates.

Data is assessed from the two main sources of injuries, namely accidents reported under legislation (and used for submission to ESAW), and self-reports from the UK LFS. This generally shows a steady reduction from around 2000 to 2010, although possibly starting to level-off since.

In terms of the economic effects, there was a discrete and expected reduction in accidents around 2008 and 2009, however the scale of reduction was probably not as noticable compared to some other Member States.