Revision log

Extent of Revision/s

As well as the regular revision to published data for the previous year (2021/22), there has also been a one-off revision for data in 2021/22 due to an error identified in last year’s processing.

Regular revision

Data published for the most recent year is given a ‘p’ status (provisional), and a year later is released as final (with an ‘r’ notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%. 

The ‘finalised’ (revised) figure for RIDDOR reported non-fatal employee injuries in 2021/22 (published in November 2023), was 62,481. This is an increase of 1.2% on the previously published provisional figure (61,713).

One-off revision:

A processing error affecting the industry codes on a small number of non-fatal injury records from 2021/22 has been identified. As a result, previously published figures of number of RIDDOR reported non-fatal injuries in 2021/22 for a small number of industry groups were higher/lower than expected compared with earlier years.

A review has shown that this error was due to the omission of a stage of quality assurance that examined the industry codes for 2021/22. The main affected industry groupings were Agriculture, forestry and fishing (SIC A, SIC 01 and SIC 02), Accommodation (SIC I, SIC 55) and other personal service activities (SIC S, SIC 96), where provisional numbers for 2021/22 were higher than expected. Also affected were Human health and social work activities (SIC Q, SIC 86, SIC 87, SIC 88), as well as Security activities and Services to buildings and landscaping (SIC N, SIC 80, SIC 81), where provisional numbers in 2021/22 were lower than expected.

The table below shows the provisional and revised counts of reported non-fatal employee injuries in those industries affected most by this error.

Table: Industries where the revised count of employee non-fatal injuries in 2021/22 is +/- 5% of the provisional count published in November 2022.

(Base: Includes only those industry groups where the difference is at least +/- 100)

[Note 1]: Standard Industrial Classification (SIC): The current system used in UK official statistics for classifying businesses by type of activity they are engaged in.

Extent of Revision/s

Regular revision

Data published for the most recent year is given a ‘p’ status (provisional), and a year later is released as final (with an ‘r’ notation to depict revised). This allows for late reports and where necessary, additional quality checks. It also allows for revisions in the employment data (provided by Office for National Statistics) resulting from re-weighting to reflect the latest population estimates. These revisions affect the calculated injury rates. Additionally, this year the employment data used to calculate rates for 2020/21 has been re-aligned to reflect employment over the 2020 calendar year, rather than the 2020/21 fiscal year. This change, implemented last year to align the injury data period with the employment data period for 2020/21, (to allow for the disruption to the labour market from the coronavirus pandemic) did not materially affect the interpretation of the rate data, so has reverted to a calendar year basis for consistency with all other years. Changes from these regular revisions are usually small, numbers typically increasing by around 1-2%. These regular revisions affect published data for 2020/21.

One-off revisions:

As well as these regular revisions affecting data for 2020/21, there are two further one-off revisions to the RIDDOR non-fatal injury data published in November 2022, affecting data from 2012/13-2020/21.

1. Numbers of reported non-fatal injuries each year from 2012/13 onwards have been revised to exclude those arising on railways or offshore (around 1,125 per year on average) as these are published and maintained elsewhere. The Office for Rail and Road (ORR) is the main Enforcing Authority on railways (mainline rail, London Underground, and other non-mainline networks (trams, metros, other light rail, minor and heritage railways)). ORR publish a separate statistical release on rail safety, including statistics of non-fatal injuries to the workforce reported under RIDDOR and are the authoritative source on work-related injuries on railways - see Rail safety | ORR Data Portal. Similarly, HSE publish a separate report on offshore injuries in the oil and gas sector, including statistics on RIDDOR reported non-fatal injuries - see HSE Offshore: Statistics. This report provides the authoritative source on injuries offshore. 
   • For railways: this exclusion has affected both overall number and rate of reported non-fatal injuries as well as numbers at more detailed industry levels (Transportation and storage (SIC H) and Land transport and transport via pipelines (SIC 49)): an overall reduction of around 1,000 reports per year.
   • For offshore: this exclusion has affected both overall number and rate of reported non-fatal injuries as well as numbers at more detailed industry levels (Mining and quarrying (SIC B) and Extraction of crude petroleum and natural gas (SIC 06)): an overall reduction of around 100 reports per year.
2. A review of industry coding has resulted in changes to processing methods. These changes have been applied to data back to 2012/13, with revisions to breakdowns of numbers by industry.  
   • The RIDDOR reporting system requires notifiers to select from a drop-down list of industry classifications, and this list aligns to the UK-wide system of SIC ('Standard Industrial Classification 2007'). The information provided by notifiers allows each report to be classified at a '4-digit' SIC level and from this, higher aggregations can be produced.
   • Previously, an automated process re-coded some SIC codes as reported by the notifier, into a different code, by utilising additional information on the form such as the job title of the injured person. The basis of this re-code was to ensure consistency with the manual coding previously done in the Incident Contact Centre prior to moving to an online reporting system in September 2011.
   • A recent review of samples of records showed that in many cases it was uncertain whether the automated recoding was more reliable than the original coding, with some notable exceptions. Further, in most instances the recoding didn’t materially affect the interpretation of numbers in individual industry groups.
   • However, the review confirmed that for a few specific industries, some reports were being consistently mis-classified. This particularly applies to reports from health and social work activities (where reports are commonly mis-allocated against either accommodation or other personal services) and from equestrian activities (where reports are commonly mis-allocated against agriculture).
   • A simplified automated re-coding process has been developed that allocates a more suitable industry code to reports from these specific industries where there is evidence of consistent misclassification, utilising additional information on the form such as the job title of the injured person.  All other industry codes remain ‘as-reported’.

Table 1 below summarises the combined impact on the 2020/21 data from both the exclusion of injury reports on railways and offshore from the overall total (number and rate) and the regular revision. The net impact of these changes has had a negligible impact on the overall number and rate for 2020/21.

Table 1: Change in published number and published rate of non-fatal injuries to employees in 2020/21 arising from both the exclusion of reports on railways and offshore and the regular revision.

Table 2 summarises the change in overall number of injury reports, and the overall injury rate for years 2012/13-2019/20 arising from the exclusion of injury reports on railways and offshore from the total (number and rate).

Table 2: Change in published number and published rate of non-fatal injuries to employees arising from the exclusion of reports on railways and offshore, 2012/13-2019/20.

Figure 1 below shows the impact of the change in processing methods on the distribution of injuries by industry sector for 2019/20 (results for other years show a similar picture).

Notes

• Source: RIDDOR, industry groups based on SIC
• # Figures exclude injuries on railways and offshore
• Industry as defined at section level of the Standard Industrial Classification (SIC). SIC section code shown in brackets.
• Numbers shown on chart is the number of reports for each industry section under either the old or revised recoding system. This is for illustrative purposes, though all numbers are available to view – just select individual bars.

For the large part, the numbers by industry are similar under both the old and revised industry recoding system. The biggest difference in number is in administrative and support service activities (SIC N), arts, entertainment and recreation (SIC R) and other service activities (SIC S).

The review of the previous industry recoding algorithm showed that it was uncertain whether the automated recoding into and from administrative and support services (SIC N), arts, entertainment and recreation (SIC R) and other services (SIC S) were more reliable than the original coding.

• Within administration and support services (SIC N), the review of the previous recoding system showed that recoding of injury reports into and out of this industry covered a wide range of industry groups, with no consistent pattern for misallocation by the reporter.
• A similar pattern was found within arts, entertainment and recreation (SIC R), particularly for reports into and out of the sub-division sports, amusement and recreation activities.
• Many reports get mis-classified by notifiers to other services (SIC S) (particularly to the sub-division other personal services (SIC 96)). There are some consistent misclassifications to this industry group, particularly reports from residential and social care but also other public service activities which both the old and new automated recoding account for. However there were a wide range of other recodes in the old recoding system.

For all three of these industry sectors, the revised recoding system limits recoding into these industry groups to those instances where we are confident that the re-code is appropriate.

Extent of Revision/s

Regular revision

Data published for the most recent year is given a ‘p’ status (provisional), and a year later is released as final (with an ‘r’ notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%.

One-off revisions:

Extent.

Covers each planning year 2015/16 to 2019/20 inclusive, non-fatal injuries only relating to figures for two specific industries, see Table below. Figures for fatal injuries and other RIDDOR types are unaffected.

Context and reason for change.

The change does not affect the total numbers of reports made within each year, or to routine ‘regular revisions’; it relates specifically to the ‘industry classification’ used for some reports. Statistics previously published by HSE statisticians relating two specific industries have changed; figures for ‘SIC 47 – Retail trade, except of motor vehicles and motorcycles’ have increased; and figures for ‘SIC 81 - Services to buildings and landscape activities’ have decreased. All current RIDDOR industry statistics are published here.

Non-fatal RIDDOR injury figures published under ‘industry’ use the ONS (Office for National Statistics) system of Standard Industrial Classification (SIC 2007), and there is a process description how the industry classification is captured on each form. Most industry codes selected by the notifier are re-produced unchanged in the published statistics. In a minority of cases and where feasible, some industry codes are changed for the purposes of improving the coherence, consistency, and comparability of statistics.

An error was detected in the complex automated re-coding process relating to some reports made in SIC 47, meaning statistics for some reports were originally re-coded wrongly as SIC 81, whereas after reviewing these changes, the original code selected by the notifier was correct. Hence for the years quoted below, changes are from previously published figures, to the ‘as-reported’ industry codes. Figures below relate to number of reports of injuries to employees.

Extent of Revision/s

Regular revision

Data published for the most recent year is given a 'p' status (provisional), and a year later is released as final (with an 'r' notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%.

Extent of Revision/s

Regular revision

Data published for the most recent year is given a 'p' status (provisional), and a year later is released as final (with an 'r' notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%.

Extent of Revision/s

Regular revision

Data published for the most recent year is given a 'p' status (provisional), and a year later is released as final (with an 'r' notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%.

Regular revision

Data published for the most recent year is given a 'p' status (provisional), and a year later is released as final (with an 'r' notation to depict revised). This allows for late reports and where necessary additional quality checks, although the changes are usually small, typically increasing by around 1-2%. 

One-off revisions

1. The data field 'occupation' is no longer provided in the dataset for analysis, therefore the statistical Table 'Occupation of injured person' is no longer available. This is due to the field being provided in free-text format, and the difficulties in coding with a reasonable degree of accuracy.
2. The data field which records the local authority code on a geographical basis has changed, if an incident took place at an offshore workplace. Previously these have been coded under 'Aberdeen', on the basis a large proportion of offshore incidents take place on oil platforms serviced form Aberdeen. This has the effect of increasing the number of injuries, and corresponding injury rates, beyond those expected for conventional onshore workplaces. Therefore to maintain consistency with other GB workplace locations of the 'local authority of where the accident happened', such offshore injuries have been re-coded to 'LA unknown' from 2001/02 to date, with a corresponding reduction in accidents and rates recorded against 'Aberdeen' LA. The scale of the effect is to reduce the employee injury numbers for Aberdeen LA by around 28% for each of the most recent four years.
3. The above change to Aberdeen LA has also affected the numbers and rates for regional statistics, although in this case the 'regions' are the same geographical boundary as 'Scotland' (Region; County; and Country). Other regional figures (Scotland) reduce by a much smaller amount of around 1.5%.

Extent of Revision

RIDDOR rates from 2004/05 – 2013/14 have been revised to take account of revisions to the employment data.

The effects of these revisions on the injury rates:

• The total non-fatal injury rate for 2004/05 is about 0.1% lower using the updated employment data: for all other years the revised non-fatal rates have changed a negligible amount (within the range +/- 0.03%)
• The change in rates at more detailed levels is larger.  By broad industry group, the change is greatest in 2004/05 and 2005/06, typically in the range +/- 1% (but as high as +2.8% for the Mining and Quarrying industry in 2005/06).  For all other years the change in rate by broad industry group is generally in the range +/- 0.1%.

Reason for Revision

The Office for National Statistics has re-weighted the 2013 and 2014 Annual Population Survey (APS) data to reflect the latest population estimates.  There have also been small revisions to job estimates back to 2004 reflecting minor improvements in methodology by HSE.

Context and extent of revisions

There have been two revisions to previously-published statistics.

For the first revision. The 2004/05 -2013/14 RIDDOR injury 'rates' only (ie excluding actual numbers of injuries) have been revised. Note, this change is separate to and excludes the regular yearly changes, namely where the previous 'provisional' year is rolled-over to a 'final/revised' year to account for late reports.

RIDDOR injury rates are primarily produced in relation to employees. Given the relatively small number of reports from the self-employed, in some circumstances rates are still produced but in less detail and coverage. However where rates for the self-employed are provided, these too have been revised for the same period and the same reason. Likewise the 'worker' injury rates have been revised, which comprises employees and the self-employed combined.

For the period 2004/05-2013/14, revised employment estimates have generally increased, meaning the revised injury rates have decreased (as the number of actual injuries has remained unchanged). During the period of revision the change has been small, with reductions in the overall all-industry rates of between 0.5 and 1.0%.

At a more detailed industry level, the changes are subject to slightly more fluctuation, although again almost all industries have seen slight increases in employment (and therefore decreases in revised injury rates). Decreases in employee rates for main industries are in the range 0.1% to 1.3% since 2006/07, and slightly larger increase in some industry rates (range 0 to 3%) for the earlier years 2004/05 and 2005/06.

For the second revision. This specifically relates to the single year 2011/12, and for the two 'main industries' of Mining and Quarrying (SIC 'B'); and Electricity and Gas Supply (SIC 'D'). The change only affects statistical tables (numbers and rates), where 'main industry' is used but no lower-level industry is present, The actual data tables affected are: 'histinj'; histrate'; and 'ridlkinds 1 to 3'.

For this single year 2011/12 the numbers, and corresponding rates of injuries, have reduced for Mining and Quarrying; and increased for Electricity and Gas Supply. The changes to the number of injuries to employees are 118 major and 290 over-3-day. The all-industry numbers have not changed, they have just moved from one main industry to another.

Reasons for Revision

For the first revision. Employment estimates use the Annual Population Survey (APS) data series, as provided by the Office for National Statistics (ONS), see the employment section on the data sources page.

The data series has been re-weighted by ONS back to 2004, to reflect new population estimates based on the 2011 Census.

For the second revision. This corrects an error in data presentation. HSE previously used to publish injury data for 'Electricity and Gas Supply (SIC 'D')' alongside 'Mining and Quarrying (SIC 'B')' as a combined figure and description (SIC B,D). The descriptors were changed but the data for that year was not, such that SIC D data is now sited with Waste and Recycling (SIC E) to give SIC D,E.

Extent of Revision

The 2009/10 and 2010/11 RIDDOR rates and estimated RIDDOR reporting levels have been revised to take account of revisions to the employment data.

The effects on the employment rates:

• For 2009/10, the revised rates overall are about 0.1% higher using the updated data. Within this overall figure, the change in rates at a more detailed level of industry, occupation and region is larger, namely in the range +/- 2.5%
• For 2010/11, the revised rates overall have changed a negligible amount. Within this overall figure, the change in rates at a more detailed level of industry, occupation and region is in the range +/- 1%

The revised employment data (denominator for RIDDOR rates), has had no impact on the published overall estimates of RIDDOR reporting levels of non-fatal injuries (major and over-3-days' incapacitation injuries) by employment status. However, reporting levels by main industry have increased or decreased by one or two percentage points with the exception of public administration and defence where the reporting level of non-fatal injuries to employees increased from 66% to 69% and to workers (employees and self-employed) from 65% to 69%

Reason for Revision

The Office for National Statistics has re-weighted the 2009 and 2010 Annual Population Survey (APS) data using revised population estimates. Employment estimates from the APS are used in the calculation of RIDDOR rates.

Extent of Revision

The injury and ill health rates for all RIDDOR rate tables have been revised to 2004/5.

Reason for Revision

The Office for National Statistics has re-weighted the Annual Population Survey data back to 2004. For further information on this change please read the employment section on the data sources page.

Extent of Revision

The injury and ill health rates for all RIDDOR rate tables have been revised to 2004/5.

Reason for Revision

As of November 2011, the employment data source used by HSE's Statistics Branch is changing to the Annual Population Survey (APS). For further information on the previous sources used, and the reasons for changing to the APS, please read the employment section on the data sources page

Extent of Revision

Injury rates from 2001/02 to 2008/09 have been revised. This affects both total rates and those by industry.

The table below shows the effects of the revisions on previously used employee totals for the broad industrial sectors and overall. Overall there is little difference in the totals. There are some larger upward revisions for agriculture from 2002/3 to 2007/8, ranging from 7% to 17%. The changes for other sectors are less, the main ones being Construction (-6% in 2002/3) and Extractive (6% in 2006/7). The effect on the worker denominators will be smaller in magnitude because the self-employed component has not been subject to revision.

Reason for Revision

In July 2010, the ONS revised the employee job series which forms the denominator for all RIDDOR injury rates for employees and workers. Visit the ONS website for more details about the reason and scale of the revisions.

Extent of Revision/s

Regular revision

The 2022/23 results published in November 2023 have been revised. This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated prevalence of work-related illness in 2022/23 was revised from 1,770 thousand to 1,814 thousand, a rise of 2.5%. The impact on the overall results is small, but the percentage changes are more variable in direction and size when the data are disaggregated e.g. by industry.

Reason for Revision

The Office of National Statistics (ONS) has reweighted the 2022/23 estimates using the latest estimates of the size and composition of the UK population which incorporate data from the Census for England and Wales.

Extent of Revision/s

Regular revision

The 2019/20 and 2020/21 results published in November 2022 have been revised.  This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated prevalence of work-related illness in 2019/20 was revised from 1,635 thousand to 1,633 thousand, a fall of 0.12% and in 2020/21 was revised from 1,655 thousand to 1,663, a rise of 0.48%.  The impact on the overall results is small, but the percentage changes are more variable in direction and size when the data are disaggregated eg, by industry.

Reason for Revision

The Office for National Statistics (ONS) has reweighted the 2019/20 and 2020/21 data set estimates using updated HM Revenue and Customs (HMRC) Real Time Information (RTI) data.

Extent of Revision/s

Regular revision

The 2019/20 results published in November 2020 have been revised. This affects estimates and rates.

The table below shows the revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated prevalence of work-related illness in 2019/20 was revised from 1,627 thousand to 1,635 thousand, an increase of 0.49%. The impact on the overall results is small, but the percentage changes are more variable in direction and size when the data are disaggregated e.g., by industry, occupation, region.

Reason for Revision

The Office for National Statistics (ONS) has reweighted the 2019/20 Labour Force Survey data set using updated population estimates to better reflect changes in international migration and other impacts as a result of the coronavirus (COVID-19) pandemic. 

Extent of Revision

The 2011/12 – 2017/18 results published in October 2019 have been revised. This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated incidence of work-related illness in 2016/17 was revised from 516 thousand to 517 thousand, an increase of 0.19% and the estimated working days lost for work-related illness in 2017/18 was revised from 26,832 thousand to 26,929 thousand, an increase of 0.36%. The impact on the overall results is fairly small, but the percentage changes are more variable in direction and size when the data are disaggregated eg by industry, occupation or region.

Reason for Revision

1. The Office for National Statistics (ONS) has reweighted the Labour Force Survey data sets (2011/12 – 2017/18) to reflect the latest population estimates.

Extent of Revision

The 2013/14 - 2015/16 results published in November 2016 have been revised. This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated incidence of work-related illness in 2015/16 was revised from 529 thousand to 528 thousand, a fall of 0.19% and the estimated working days lost for work-related  illness in 2014/15 was revised from  23,314 thousand to 23,367 thousand, an increase of 0.23%. The impact on the overall results is fairly small, but the percentage changes are more variable in direction and size when the data are disaggregated eg by industry, occupation or region.

Reason for Revision

1. The Office for National Statistics (ONS) has reweighted the Labour Force Survey data sets (2013/14-2015/16) to reflect the latest population estimates.
2. Minor improvements have also been made to the methodology used to estimate work-related illness, workplace injuries and corresponding working days lost.

Extent of Revision

The 2000/01 - 2014/15 results have been revised. This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated prevalence of work-related illness in 2014/15 was revised from 1,243 thousand to 1,247 thousand, an increase of 0.32% and the estimated working days lost for workplace injury in 2013/14 was revised from 4,447 thousand to 4,436 thousand, a fall of 0.25%. The impact on the overall results is fairly small, but the percentage changes are more variable in direction and size when the data are disaggregated eg by industry, occupation or region.

Reason for Revision

1. The Office for National Statistics (ONS) has reweighted the Labour Force Survey data sets (2012/13-2014/15) to reflect the latest population estimates.
2. Minor improvements have also been made to the methodology (2000/01-2014/15) used to estimate work-related illness, workplace injuries and corresponding working days lost.

Extent of Revision

The 2000/01 -2013/14 results have been revised. This affects estimates and rates.

The table below shows the range of revisions to the overall measures of work-related illness and workplace injuries. For example, the estimated prevalence of work-related illness in 2006/07 was revised from 1,384 thousand to 1,416 thousand, an increase of 2.3% and the estimated working days lost for workplace injury in 2013/14 was revised from 4,723 thousand to 4,447 thousand, a fall of 5.9%. The impact on the overall results is fairly small, but the percentage changes are more variable in direction and size when the data are disaggregated eg by industry, occupation or region.

Reasons for Revision

1. The Office for National Statistics (ONS) has reweighted the Labour Force Survey data sets back to mid-2001 to reflect new population estimates based on the 2011 Census.
2. Implementing improvements in the days lost methodology (2000/01-2013/14), in particular the methods used to impute for missing days lost information.

Extent of Revision

The 2001/02 -2009/10 results have been revised. This affects estimates and rates.

The table below shows the range of revisions to the 2001/02- 2009/10 overall measures of work-related illness and workplace injuries. The impact on the overall results is small, but greater percentage changes occur when broken down by demographic and employment related variables. For example the estimated prevalence of work-related illness in 2005/06 for Postal workers, mail sorters, messengers and couriers (SOC code 9211) was revised from 5500 to 5600, an increase of approximately 2% with the associated rate increasing from 2570 to 2600 per 100,000 an increase of approximately 1.3%.

Reason for Revision

1. The Office for National Statistics (ONS) has reweighted the Labour Force Survey data sets back to mid-2001 to reflect new population estimates based on the 2011 Census.
2. Implementing improvements in the days lost methodology (2000/01-2013/14), in particular the methods used to impute for missing days lost information.

Extent of Revision

The 2008/09 results have been revised. This affects four industry groups and has a very small impact on the full range of results.

Reason for Revision

The Office for National Statistics (ONS) revised a small number of 1992 Standard Industrial Classification (SIC 92) codes following the identification of a few errors in the mapping from the SIC 2007 to SIC 92. This had a small impact on the weighting factors developed by HSE (explanation of weighting LFS data).

Extent of Revision

Ill health rates in the THOR industry and occupation tables, have been revised to 2004/5.

Reason for Revision

The Office for National Statistics has re-weighted the Annual Population Survey data back to 2004. For further information on this change please read the employment section on the data sources page.

Extent of Revision

Ill health rates in the THOR industry and occupation tables, have been revised to 2004/5.

Reason for Revision

As of November 2011, the employment data source used by HSE's Statistics Branch is changing to the Annual Population Survey (APS). For further information on the previous sources used, and the reasons for changing to the APS, please read the employment section on the data sources page.

Extent of revision:

The Germany percentage of workers who suffered from a work-related health problem resulting in time off work from the EU LFS (2020) (EURO_03) has been revised by Eurostat from 4.4% to 4.3%.

Reason:

This data is sourced from the Eurostat website, and the reason for the revision is due to the processing of additional survey data that had not been available before for technical reasons and because of late interviews. Consequently, some figures smoothly changed, including those of the EU-LFS module 2020 regarding work-related health problems.

Extent of revision:

The UK average rate (2006-2010) of fatal injuries has been revised from 0.97 to 0.91 (Fatal injury statistics - Table 6).

Reason:

Fatality data provided by Eurostat for 2008 and 2010 seems inconsistent with previous year's rates used by HSE. As a result, fatality rates for these two years have not been updated in the European Comparisons publication. This has subsequently affected the UK average (2006-2010) in the Fatal Injuries publication. Work is ongoing to clarify the changes made by Eurostat.

Extent of revision:

Data from the EU LFS (2007) on percentage of workers who had an accident resulting in sick leave (EU Comparisons Table 2) or ill health resulting in sick leave (EU Comparisons Table 3) has been revised by Eurostat for a number of countries. These revisions are of a small magnitude.

Reason for revision:

This data is sourced from the Eurostat website, and the reason for these revisions is being clarified by HSE.

One-off revision

A processing error led to incorrectly labelled row headings in the statistics for mesothelioma deaths for Local and Unitary Authority areas by 5-year time-period in table MESOAREA (tabs mesoarea03 and mesoarea04). Rows 100 to 218; 223 to 234 and 238 to 254 were incorrectly labelled in these two tabs. An updated table with corrected row headings was published on 18 October 2024.

Extent of Revision/s

In July 2018 provisional figures for mesothelioma deaths that occurred in 2016 were published and the figure for deaths that occurred in 2015 was subject to minor revision:

the number of deaths that occurred in 2015 was revised from 2542 to 2547

Regular revision

These regular revisions account for a small number of late death registrations and are included to ensure the data series is as complete and consistent as possible.

Extent of Revision/s

In July 2017 provisional figures for mesothelioma deaths that occurred in 2015 were published and the figures for deaths that occurred in 2013 and 2014 were subject to minor revision:

• the number of deaths that occurred in 2013 was revised from 2556 to 2560
• the number of deaths that occurred in 2014 was revised from 2515 to 2519

Regular revision

These regular revisions account for a small number of late death registrations and are included to ensure the data series is as complete and consistent as possible.

Extent of Revision

In October 2018 provisional figures for asbestosis deaths that occurred in 2016 were published and the figures for deaths occurring in 2014 and 2015 were subject to minor revision:

• the number of deaths mentioning asbestosis (excluding those also mentioning mesothelioma) that occurred in 2014 was revised from 435 to 436.
• the number of deaths mentioning asbestosis (excluding those also mentioning mesothelioma) that occurred in 2015 was revised from 467 to 468.

Reason for Revision

These regular revisions account for a small number of late death registrations and are included to ensure the data series is as complete and consistent as possible.

Extent of Revision

In November 2017 provisional figures for asbestosis deaths that occurred in 2015 were published and the figures for deaths occurring in 2014 were subject to minor revision:

the number of deaths mentioning asbestosis (excluding those also mentioning mesothelioma) that occurred in 2014 was revised from 432 to 435.

Reason for Revision

These regular revisions account for a small number of late death registrations and are included to ensure the data series is as complete and consistent as possible.

Extent of Revision

The number of deaths from asbestosis as underlying cause was revised from 104 to 123 for 2006, 93 to124 for 2007 and 107 to 148 for 2008. The total number of death certificates mentioning asbestosis remained unaffected.

Reason for Revision

Miscoding of the underlying cause of death.

Extent of revision

Revisions affect all annual cost estimates from 2004/05 to 2017/18 for workplace injury and ill health (for 2011/12, only an injury cost estimate is available). They result in an average annual increase over the period in total estimated costs of 3.4%, with a 1.9% increase in workplace injury costs and a 4.3% increase in work-related ill health costs.

There have been no substantive modelling changes to the 2017/18 model which have caused any large revisions to the cost estimates. The revisions that have been made are:

• A small revision to fatal injuries data as outlined in (1).
• Changes to indices used for inflating price data for all years outlined in (2).

Reasons for revision

1. No reweighting of the underlying injury and ill health numbers from the Labour Force Survey (2011/12-2017/18). Minor revision of fatal injuries which affects 2017/18 (three-year average period 2016/17-2018/19).
2. Use of new price data or inflation of existing sources to express each year's cost estimates in constant (2018) prices. Additionally, the calculation of long-term real earnings growth has been updated; and the model now uses NHS Cost Inflator Index (NHSCII). NHSCII was developed by the Department of Health and Social Care NHS England, the University of York, and the Office of National Statistics to replace the Hospital and Community Health Service Index.  CPI Health was produced for one year in 2017/18 Between NHSCII and HCHS. Both CPI Health and NHSCII were developed to replace the Hospital and Community Health Services index as it was discontinued.
3. The Statutory Sick Payments reimbursement scheme was removed from the model, as this ceased in 2014. This does not affect total costs to society, as it is a transfer payment from government to employers.

Extent of revision

Revisions affect all annual cost estimates from 2004/05 to 2016/17 for workplace injury and ill health (for 2011/12, only an injury cost estimate is available). They result in an average annual increase in total estimated costs of 2.5%, with a 1.2% increase in workplace injury costs and a 3.3% increase in work-related ill health costs.

Reasons for revision

1. Revisions to the underlying injury and ill health numbers from the Labour Force Survey due to (i) reweighting of the LFS data sets (2011/12-2017/18) to reflect the latest population estimates (as outlined in LFS revision log); and (ii) use of additional years' data to estimate the proportion of 'never returns' (workers who expect to permanently leave the labour market each year) due to a workplace injury relative to work-related ill health. The latter affects proportion of costs attributed to injury and ill health but does not materially affect total costs.
2. Use of new price data or inflation of existing sources to express each year's cost estimates in constant (2017) prices. Additionally, estimates of long-term real earnings growth have been updated; and the model now uses CPI Health to inflate healthcare costs as the Hospital and Community Health Services index has been discontinued.
3. Minor modelling improvements to the 2017/18 model.

Extent of revision

Revisions affect all annual cost estimates from 2004/05 to 2015/16 for workplace injury and ill health (for 2011/12, only an injury cost estimate is available). They result in an average annual increase in total estimated costs of 2.5%, with a 0.8% increase in workplace injury costs and a 3.3% increase in work-related ill health costs.

There have been no substantive modelling changes to the 2016/17 model which have caused any large revisions to the cost estimates. The revisions that have been made are:

• A small revision to 15/16 incidence data as outlined in (1i).
• Changes to indices used for inflating price data for all years outlined in (2).

These revisions result in a £5 million decrease in total costs to society for the year 2015/16 from £14,921 to £14,916.

Reasons for revision

1. Revisions to the underlying injury and ill health numbers from the Labour Force Survey due to (i) reweighting of the LFS data sets (2015/16-2017/18) to reflect the latest population estimates; and (ii) use of additional years' data to estimate the proportion of 'never returns' (workers who expect to permanently leave the labour market each year) due to a workplace injury relative to work-related ill health. The latter affects proportion of costs attributed to injury and ill health but does not materially affect total costs.
2. Use of new price data or inflation of existing sources to express each year's cost estimates in constant (2016) prices. Additionally, sources previously inflated by the Retail Price Index are now inflated by the Consumer Price Index.

Extent of revision

Revisions effect on all annual cost estimates from 2004/05 to 2014/15 for workplace injury and ill health (for 2011/12, only an injury cost estimate is available). They result in an average annual increase in total estimated costs of 2.6%, with a 2.0% increase in workplace injury costs and a 3.0% increase in work-related ill health costs. Specifically, for the year 2014/15, the revised estimate of total costs has increased by £400 million, from £14.09 billion to £14.49 billion. Of this, £200 million is due to the reweighting of LFS data described in (1) below, and £200 million due to the change in price year from 2014 to 2015 described in (2). These changes have increased estimated costs of workplace injury for the year 2014/15 by £110 million and work-related ill health by £290 million.

Reasons for revision

1. Revisions to the underlying injury and illness numbers from the Labour Force Survey due to (i) reweighting of the LFS data sets (2013/14-2015/16) to reflect the latest population estimates; and (ii) revised estimate of the number of workers who permanently leave the labour market each year due to a work-related illness or injury.  
2. Use of new price data or inflation of existing sources to express each year's cost estimates in constant (2015) prices.

Extent of revision

Revisions impact on 2006/07-2012/13 annual cost estimates for workplace injury and ill health (for 2011/12, only an injury cost estimate is available). Average annual increase of 1.7% in the overall cost of workplace injury and ill health: for ill health an average annual increase of 6.0%, for injury an annual average decrease of 5.1%. Looking in particular at 2012/13, cost of workplace injury and ill health increased slightly from £14.24 billion to £14.27 billion, cost of injury decreased from £5.64 billion to £5.12 billion and cost of ill health increased from £8.60 billion to £9.14 billion.

Reasons for revision

1. Revisions to the underlying injury and illness numbers from the Labour Force Survey due to (i) a major reweighting exercise of LFS data to reflect new population estimates produced from the 2011 Census;(ii) a new improved methodology for estimating working days lost due to work-related illness and injury from the LFS; and (iii) revised estimate of the number of workers who permanently leave the labour market each year due to a work-related illness or injury.
2. Cost estimates revised to express each years' cost estimates in current year (2013) prices.

Extent of Revision

1. Revisions impact on 2006/07-2010/11 annual cost estimates for workplace injury and ill health and 2011/12 injury cost estimate (no ill health cost estimate available for 2011/12).  Average annual increase of 1.5% in the overall cost of workplace injury and ill health: for ill health an average annual increase of 3.4%, for injury an annual average decrease of 1.1%.  Looking in particular at 2010/11, cost of workplace injury and ill health increased from £13.8 billion to £14.0 billion, cost of injury decreased from £5.4 billion to £5.3 billion and cost of ill health increased from £8.4 billion to £8.7 billion.

Reason for Revision

1. A small error in the model underestimated the time taken off work for injuries with up to 7 days off work.
2. Ongoing methodological improvements to the cost model including: improved method for accounting for lost income for those cases resulting in withdrawal from the labour market; an improved estimate of the split of cases withdrawing from the labour market as a result of workplace injury or ill health between injury and ill health; revised costing categories for workplace non-fatal injury and ill health (both injury and ill health now estimate costs separately for cases with up to 6 days off work and 7 or more days off work).
3. Cost estimates revised to express each years cost estimates in current year(2012) prices.

Extent of Revision

1. 3% increase in 2010/11 estimate (expressed in 2010 prices) from £13.4bn to £13.8bn, though the overall distribution of cost by the different cost components and cost bearers remains stable.
2. The net effect was an annual downward revision of less than 0.5% to annual total cost estimate.

Reason for Revision

1. An error was found in the source data for the 2010/11 estimate - the source data had not been updated with the time off work profile for illness and injury cases for 2010/11 and also, the prosecution costs were found to be based on 2011/12 data.
2. Number of refinements to the methodology, including (i) linking future increases in benefits to the Consumer Prices Index (rather than the Retail Price Index, in line with Government policy); (ii) excluding costs arising from cases of occupational cancer from compensation and Industrial Injuries Disablement Benefit (IIDB) costs; and (iii) changes in definition of non-fatal reportable and non-reportable injury in line with amendments to RIDDOR regulations. Also, in line with usual practice annual cost estimates revised to current year prices (2011).

Extent of Revision

Expressing previous years cost estimates in 2010 prices had an upward effect on the cost estimates. However these were countered by other changes within the model, resulting in overall negligible impact to the cost estimates.

Reason for Revision

Cost estimates for 2006/07 to 2009/10 were revised to express costs in 2010 prices. Additionally (i), the estimate for the number of workers who permanently withdraw from the labour market as a result of their illness or injury was revised across the back series, and (ii) an improved method for accounting for Industrial Injuries Disablement Benefit was introduced.