Effect on RIDDOR statistics following recent legal and system changes
In the three years from 2011, the following three changes took place to the notification of RIDDOR incidents. Some of these changes affected RIDDOR statistics, so care is needed when comparing one year with another. Below is a summary of the changes and, where known, the effects on the statistics. Covering the most recent change first:
- In October 2013, RIDDOR was substantively changed following formal review, consequently affecting many of the RIDDOR statistics for 2013/14 and onwards.
- In April 2012, RIDDOR underwent just one specific change. The reporting threshold for those non-fatal injuries to workers, that were not ‘major’ or ‘specified’, from over-3-days’ incapacitation to over-7-days. This affects over-7-day statistics for 2012/13 and onwards.
- In September 2011, although RIDDOR did not change, the reporting mechanism did. In nearly all cases reporting is now made by employers directly online. This potentially affects RIDDOR statistics for 2011/12 and onwards.
The following describes the above changes in more detail:
Changes to RIDDOR: October 2013
These changes took effect mid-way through the reporting year 2013/14, so any effect of the change will be roughly half that for the full year, although noting there can be seasonal effects, for example bad weather, which could skew the results for the ‘winter’ half of the year since the change. In addition to following the changes, there may be subtle effects on reporting behaviour.
One of the main changes in terms of statistics, is the former ‘major’ injury to workers definition changing to ‘specified’. Some of the old major categories were removed, other new ones included. The net result is a reduction of reports of around 10% (full-year equivalent) when comparing the new specified with the old major injuries.
The over-7-day reporting threshold did NOT change on the above date – it changed the previous year. However, some of the former major injuries that are no longer reportable as specified (as described in the above 10% reduction), may still fall within the over-7-day reporting threshold and reported as such. Typically, many of these are dislocations or overnight hospitalisation, previously classified as major, now possibly reportable as over-7-day injuries.
Therefore, numbers of over-7-day reports from October 2013 are slightly higher than they otherwise would have been.
Separate to the above changes, many defined dangerous occurrence categories changed ‘type number’ as well as description.
The reporting of domestic gas incidents resulting in injury also changed. Previously the non-fatal injury was defined as ‘major’ to qualify; now it requires loss of consciousness or hospitalisation.
For the above reasons, direct comparison of statistics that span the affected years should be treated with caution.
Change to over-7-day injury reporting: April 2012
From April 2012 a legal change to RIDDOR affected the reporting threshold, for those non-fatal injuries to workers that are not ’major’ or ‘specified’. Previously the reporting threshold was over-3-days, now it is over-7-days off work (or unable to do normal duties). The requirement for employers to retain records of over-3-day injuries remains.
The net effect is to reduce the previous rate of over-3-day reports by around 29%. No discernable change in major injury reporting was evident following the change. Because of this change it is not possible to make direct comparisons between over-3-day and over-7-day statistics. More detailed analysis of the long-term trend in over-3-day injuries, and an estimate of the effects of the change to over-7-day, has been undertaken . This includes estimated effects on major injuries, by all industry and some main industries.
For illustrative purposes, a description of the relationship between all workplace injuries, and the amount of time off, can be seen through the complementary data source, the Labour Force Survey, showing accidents with: less than 4 days off; 4 to 7 days; and over-7-days.
Change in reporting system: September 2011
From September 2011 the RIDDOR notification system used by employers changed, with reporting now being online-only. In completing the online form, employers provide information about the incident using simplified drop-down lists. Less detailed information is available compared to the previous reporting arrangements, although most top-level RIDDOR statistics are still comparable. To note, although the actual reporting mechanism changed, RIDDOR itself did not at that time.
Analysis of data from the new system to date suggests that, whilst overall numbers are consistent with the previous system, in practice some of the coding and definitions may not be. This problem is frequently observed when administrative systems are changed and not unique to RIDDOR. To enable RIDDOR data to be published for the year 2011/12, a new dataset was created for the full-year using data from both old and new systems. This enabled us to publish top-level RIDDOR figures for the 2011/12 by:
- injury severity, e.g. fatal, major, over-3-day (over-7-day from April 2012)
- top-level industry, e.g. construction, manufacturing
- top-level 'kind' of accident, e.g. fall from height, handling injuries
- dangerous occurrence types and flammable gas incidents
Notes about published RIDDOR data
For statistical purposes, where feasible HSE statisticians make the following adjustments to data extracted from the RIDDOR reporting system. This statistical dataset ensures the underlying 'raw' data, as required under RIDDOR provided by the notifier, remains unchanged.
Industry classification; and occupation
The 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 are produced. Together with estimates of employment by SIC this allows reported injury rates to be calculated, thereby helping identify higher and lower risk industries.
Fatal injury statistics are subject to detailed record-level scrutiny for accuracy of industry classification before publishing, although given the volume of non-fatal reports it is not possible to manually review individual reports for industry selections made by the notifier. Hence samples of records are reviewed across various industry classifications, and where specific misalignment with standard SIC classification seems evident then deeper evaluation is undertaken, to decide whether a more relevant industry code is available.
This approach confirms the large majority of SIC selections made by notifiers is appropriate and changes are not necessary. However, reports relating to a few industries do require change to an SIC code more aligned to SIC. A manageable automated process is used that re-codes some SIC’s as reported by the notifier, into a more suitable code, by utilising additional information on the form such as the job title of the injured person. Those main industries subject to some re-coding into other industries cover: agriculture and forestry; accommodation; public administration; other service activities.
This re-coding enables statistics to be published by industry at a 2-digit level and higher, covering main industries like: construction; broad manufacturing types; agriculture; and top-level service-based industries such as retail and types of public administration. More complex systems of re-coding have been evaluated to see if more detailed re-coding might be possible, however those systems are too manually intensive or complex to maintain, and invariably only offer marginal improvements to the original SIC codes.
Occupation (job title) of the injured person. The reporting system requires notifiers to enter the job title of the injured person, in a 'free-text' format. To enable analysis, this information requires classification according to the UK-wide 'SOC 2010' ('Standard Occupational Classification 2010') or more recently ‘SOC 2020’. However, many of the job titles provided are unclear to enable coding directly, e.g. production/process/factory worker. Others use job titles that are unknown outside their industry or company. Consequently it’s not feasible to code a considerable proportion of records with sufficient accuracy and automation, therefore the coding of occupation is not available for statistical analysis.
'Local authority' classifications
When a report is made online, the notifier is required to select from drop-down options within which local authority the accident happened. In a small number of cases and for various reasons, the selected LA may not be correct in relation to the incident postcode provided. Where feasible this LA code is automatically re-assigned to a more suitable one in our statistical dataset. It also happens that LA boundaries or names change most years, usually resulting in a few small LAs replaced by a single larger and newly formed LA. Where LA boundaries or names change, any relevant existing LA codes in the statistical dataset are re-coded to the new values to create a consistent and simplified time series. In each of these scenarios where an LA code requires change, the higher statistical aggregations like region are largely unaffected.
Duplicate and 'non-reportable' notifications
The RIDDOR reporting system enables simple and efficient notifications to be made by employers, with guidance available to reporters on RIDDOR definitions. However, sometimes incidents which are not legally reportable are entered on the system; conversely for a variety of reasons, the same incident is sometimes reported more than once. Where feasible non-reportable and duplicate reports are excluded from statistical analyses.