Harpur Hill, Buxton Derbyshire, SK17 9JN T: +44 (0)1298 218000 F: +44 (0)1298 218590 W: www.hsl.gov.uk HSE’s Measurement of Work-Related Ill Health: Is it fit for purpose and what are the priorities for improvement? Workshop Proceedings 19 February 2009 CWH/09/15 Project Leader: Jo Bowen Author(s): Jo Bowen, Nadine Mellor, Jo HarrisRoberts Corporate Group: Centre For Workplace Health Science Group: Human Factors DISTRIBUTION Anne Marie Grey John Hodgson Customer Project Officer Customer Authorising Officer John Ewins Karen Russ Patrick McDonald HSE HSL HSE This report and the work it describes were funded by the Health and Safety Executive (HSE). Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy. PRIVACY MARKING: Available to the public. HSL report approval: Date of issue: Job number: Registry file: Electronic file name: Andrew Curran December 2009 JN0004531 File Reference 028901 \\Vinata\hsevalue\JN0004521 Measurement of work related ill health\FINAL_REPORT\Measuring ill health workshop - final © Crown copyright (2009) ACKNOWLEDGEMENTS On behalf of the joint HSE/HSL steering team organising this event, HSL would like to thank the following workshop delegates for their valuable contribution to the workshop. David Coggon Peter Elias Ira Madan Angela Donkin Raymond Agius David Fishwick Roseanne McNamee Matthew Hotopf Keith Palmer Fintan Hurley (pre workshop contribution via email) Rob Wellens Sayeed Khan Jane Willis Simon Warne John Osman Jacky Jones Ian Spencer Maria Ottati Andy Darnton Simon Clarke Paul Buckley Special thanks to the Chair and syndicate chairs for their help with running the workshop and preparing the workshop exercises, as well as members of HSE’s Chief Scientific Adviser’s Group (CSAG): Chair: Patrick McDonald, CSAG HSE Syndicate Chair: John Ewins, CSAG HSE Syndicate Chair: Andrew Curran, HSL John Hodgson, CSAG HSE Anne Marie Grey, CSAG HSE Thanks also to Gemma McDowell (HSL) who helped to organise the workshop. iii CONTENTS 1. INTRODUCTION ......................................................................................... 7 Background ........................................................................................................ 7 Workshop Aim and objectives ............................................................................ 8 2. METHODOLOGY ........................................................................................ 9 Methodology stages ........................................................................................... 9 Figure 1. Outline workshop stages. ............................................................. 9 Discussion Paper ............................................................................................... 9 Workshop Design ............................................................................................... 9 Workshop Delivery ........................................................................................... 10 Pre workshop exercise – ‘Ideas for Improvement’ rating exercise ............ 10 Figure 2: Rating grid .................................................................................. 10 Introductory Presentations .......................................................................... 10 Syndicate Exercise 1 – ‘Ideas for Improvement’ rating exercise ............... 11 Syndicate Exercise 2 – Suggestions of How To Progress the highest Rated ‘Ideas for Improvement’ ............................................................................. 12 Conclusion of Workshop Proceedings ....................................................... 12 Collation of Workshop Outputs and Report of Proceedings ............................. 12 3. COLLATION OF WORKSHOP OUTPUTS ............................................... 13 Summary of exercise 1 – rating ‘Ideas For Improvement’ ................................ 13 Syndicate group 1 – summary of discussions ............................................. 13 Syndicate group 2 – summary of discussions ............................................. 16 Further comments in Plenary session 1 ........................................................... 20 Summary of exercise 2 - research question generation ................................... 22 Syndicate group 1 – summary of discussions ............................................. 22 Syndicate group 2 – summary of discussions ............................................. 25 Further comments in Plenary session 2 ........................................................... 27 Summary of research ideas.............................................................................. 29 Syndicate group 1 ....................................................................................... 29 Syndicate group 2 ....................................................................................... 31 Closing comments ............................................................................................ 32 Proposed next steps ......................................................................................... 33 4. REFERENCES .......................................................................................... 34 5. APPENDICES ........................................................................................... 35 Appendix 1 – Discussion Paper........................................................................ 35 Appendix 2 - Introductory presentation, Patrick McDonald and Jane Willis ...... 64 Appendix 3 - Syndicate chairs’ feedback slides................................................ 68 iv EXECUTIVE SUMMARY Background The new HSE Strategy gives prominence to “the prevention of death, injury and ill health to those at work and those affected by work activities”. HSE has therefore prioritised the need to review current statistical sources and methods of measurement of deaths, injuries and cases of ill health. The HSE’s Chief Scientific Adviser’s Group (CSAG) initiated a process of stakeholder engagement. Experts were consulted about reviewing HSE’s technical approach to measuring work-related ill health. A Discussion Paper was circulated to internal (HSE) and external (non HSE) delegates in advance of the workshop. Workshop Aim and Objectives The aim of the workshop was to review the HSE data sources and methods, looking critically at the currently used statistical sources to build up a picture of work-related ill health. The objectives of the workshop were to: Identify key ideas for improvement in data collection/analysis to better meet HSE’s strategic needs. Rate these improvements (based upon importance and practicability). Outline how the top priority improvements could be undertaken. These objectives were achieved by capturing influential stakeholders’ viewpoints regarding the identification of priority areas. The process enabled HSE to collect stakeholders’ views on the strengths and weaknesses of the respective approaches and data sources, as set out in the Discussion Paper. Workshop Outputs The workshop highlighted two broad strands of recommended improvements for inclusion in HSE’s work-related ill health measurement review. These inter-connected strands could be established as two parallel programmes, with the following aims: To clarify the leading data source for each of the different components of workrelated ill health, and (where necessary) incorporate supporting research. This would involve two main components: o Develop a systematic review identifying the best source(s) of data for each category of work-related ill health; and use this review to identify and prioritise further development work. o Test and improve the validity of self-reports in those areas where it is evident that they provide the best available estimates. To enable the exploitation of general health databases in an occupational context by attaching occupation codes at the individual record level, either via direct collection or via establishing data linkages. v o Initiate a process of engagement between HSE and other relevant organisations (including the UK Data Forum), to promote the identification and sharing of databases. Next steps The outputs of the workshop have provided CSAG with a clear picture of stakeholders’ opinions on the current array of data sources and approaches to data collection/analysis. Delegates have indicated how their highest rated ideas could be taken forward within HSE’s commissioning process. It is the intention of the HSE team to draw the outputs together into a work plan, which would be prioritised according to the temporal order of research and management of time and resources. vi 1. INTRODUCTION BACKGROUND The new HSE Strategy gives prominence to “the prevention of death, injury and ill health to those at work and those affected by work activities”. HSE has therefore viewed as a priority the need to review current statistical sources and methods of measurement of deaths, injuries and cases of ill health. The limitation of individual data sources and the need to focus on causes of work related ill health has led to the development of multiple data sources by HSE including: • Self-reported Work-related Ill health (SWI) data from the Labour Force Survey (LFS) • The Health and Occupation Reporting network (THOR) and THOR GP Survey • Industrial Injuries Disablement Benefit (IIDB) Scheme • The Reporting of Injuries, Diseases and Dangerous Occurrences Regulations (RIDDOR) • Death certificates for asbestos-related and other occupational lung diseases • Decennial supplements on occupational mortality • Attributable fraction studies of specific disease (e.g. noise exposure and hearing difficulties) These data sources are generated in three broad ways: Some rely on self-reports of work related ill health (as identified by the sufferer themselves); others on medical professionals’ diagnoses of ill health; and the third counting process is purely statistical. There is a need to review these counting methods and their advantages and disadvantages with respect to how they might independently or collectively contribute to an assessment of the nation’s state of occupational health. Several other influences call for a thorough review of existing measurement of ill health. In the last decade, the Government has increasingly sought a sound evidence base for policy-making. The advent of ‘hard’ targets in HSE’s ‘Revitalising health and safety’ strategy (RHS, 2000)1 accelerated this trend and led HSE to develop the Workplace Health & Safety and Fit3 surveys (both mainly focused on work conditions and risk control rather than health outcomes) and to bolster both the scope and frequency of its use of the LFS/SWI surveys. The trends, towards a stronger evidence base seem likely to continue. A major recent report (Hampton Report2, 2005) seeks greater co-operation, coordination and even merger between regulators, not least in the sharing of intelligence. There is also a wider move to increase data sharing between government departments. These developments should lead to new sources of data becoming available to HSE and may result in improved efficiencies in data gathering. 1 2 Revitalising health and safety: http://www.hse.gov.uk/revitalising/ Hampton Report, 2005: http://www.berr.gov.uk/files/file22988.pdf 7 In response to the RHS strategy, HSE brought together a group of experts to help with defining its technical approach to measuring progress against the RHS targets. This was set out in a statistical note published in June 20013. With the launch of the proposed new HSE strategy in 2009, a similar consultation of experts in the field was needed. The HSE’s Chief Scientific Adviser’s Group (CSAG) therefore initiated a process of stakeholder engagement. A consultation document was produced, which required further input from technical experts in the field to agree how best to take the ideas forward. This Discussion Paper can be found in Appendix 1. The next steps in this process were to capture stakeholders’ input and opinions to influence HSE’s decisionmaking process and the identification of priority areas. A workshop was chosen by HSE’s CSAG as the best method to collect stakeholder opinions on the strengths and weaknesses of the respective approaches and data sources. The workshop discussion exercises aimed to help CSAG generate a clear picture of stakeholders’ opinions on the current array of data sources and approaches to data collection/analysis. WORKSHOP AIM AND OBJECTIVES The aim of the workshop was to review the HSE data sources and methods, looking critically at the current statistical sources to build up a picture of work-related ill health and asking: What are the right components? What is the best way to use them? What are the most important ways HSE could improve its practice? The objectives were: To identify key ideas for improvement in data collection/analysis to better meet HSE’s strategic needs. To rate these improvements (on practicability and need). To outline how the top priority improvements could be undertaken. To achieve this, opinions and suggestions for improvement were elicited from subject technical experts in the field (HSE and non HSE). 3 Statistical Note: http://www.hse.gov.uk/statistics/statnote.pdf 8 2. METHODOLOGY METHODOLOGY STAGES Figure 1. Outline workshop stages. INPUT Discussion Paper Stage 1: Workshop Design Stage 2: Workshop Delivery Including Syndicate Exercises and Plenary Discussion Stage 3: Circulating Interim Outputs (Syndicate Feedback Slides) Stage 4: Collation of Workshop Outputs OUTPUT Report of Proceedings DISCUSSION PAPER Prior to the workshop, delegates received a copy of the Discussion Paper (see Appendix 1). The Paper set out the range of data sources used by HSE to indicate the level and distribution of work-related ill health in the British workforce. The discussion paper contained four broad questions that delegates were asked to consider whilst making their contributions to the workshop exercises. The paper also outlined HSE’s policy context, the intentions of the workshop and the points for delegates to consider in preparation for the workshop. WORKSHOP DESIGN The methodology included the preparation of a one-day expert workshop, which utilised the experience of the HSE project team and the HSL facilitators. The workshop design included three main stakeholder activities: an individual pre 9 workshop exercise was followed by two group exercises, which were conducted in chaired syndicate sessions. The outputs of the syndicate exercises were captured on feedback presentation slides, audio recorded and subsequently transcribed. All workshop participants were given the opportunity to discuss the outputs of syndicate group sessions in a subsequent plenary session. Further feedback on the content of the feedback presentation slides was also invited from the workshop participants via an email that was circulated following the workshop. The presentation slides and summary notes of syndicate and plenary sessions (taken from the audio transcripts) are collated in this report of proceedings. WORKSHOP DELIVERY Pre workshop exercise – ‘Ideas for Improvement’ rating exercise The Discussion Paper set out a preliminary list of ‘Ideas for Improvement’ for consideration by expert stakeholders (Appendix 1, pp.A12-A13 [paragraphs 43-44]). The pre workshop exercise encouraged the delegates to generate independent preliminary ratings, free from the influence of other delegates, in advance of the workshop discussions. Delegates were asked to rate the ‘Ideas for Improvement’ based on their perceived importance and chance of success, as indicated in the ratings grid (figure 2). Figure 2: Rating grid Rating grid for development ideas How likely is High it that this question can be successfully Moderate addressed? to low How important is progress on this question to improving the measurement of work-related ill health? High Moderate to low GREEN AMBER AMBER RED Introductory Presentations HSE's Chief Scientific Adviser and Director of CSAG, Patrick McDonald (Chair) and Jane Willis (Head of Policy Group) opened the workshop with an introductory presentation. Delegates were requested by the Chair to observe the following ground rules during the workshop discussions: Be creative Capture all points Respect one another’s contributions 10 Recognise need for shared decision making Consider issues from the perspective of HSE’s research agenda Observe the programme schedule Jane Willis highlighted the important underlying questions and also HSE’s policy context and strategy. Two guest presenters and John Hodgson (HSE CSAG) then gave presentations regarding data sources for work-related ill health: Raymond Agius – THOR Keith Palmer – Attributable Fraction Studies John Hodgson – Labour Force Survey To conclude the introductory session, the ‘Ideas for improvement’ from the Discussion Paper (Appendix 1, pp.A12-A13 [paragraphs 43-44]) were displayed on A3 sheets on the wall and delegates were invited to attach their independently derived rating for each idea using coloured stickers. This process allowed the rapid representation of all views and it was conducted without delegates being asked to discuss their rationale, at this stage. Additional ideas for improvement were also written up on the wall sheets (some from invited stakeholders who could not attend the workshop) and ratings were duly assigned. The study team divided the ideas roughly evenly between the two syndicate groups and this became the basis for syndicate exercise one. Syndicate Exercise 1 – ‘Ideas for Improvement’ rating exercise Delegates were assigned (by CSAG) to contribute to the discussion exercises in two separate syndicate groups. This aimed to provide a balanced mix of expertise and knowledge in each discussion group. The introductory slides (Appendix 2) include details of the syndicate groups and the workshop agenda. In Syndicate Discussion Exercise 1, the Syndicate Chair invited syndicate members to describe the rationale for their individual ratings assigned to each ‘Idea for Improvement’. The issues were then debated in turn, with the aim of deriving a majority viewpoint of the rating for each of the ‘Ideas for Improvement’. Majority ratings and points of consensus or contention were noted on feedback slides (Appendix 3, slides 3-20). The syndicate exercise was followed by a plenary feedback session where each Syndicate Chair presented an overview of his group’s rating decisions and discussions. There was an opportunity for delegates to discuss their viewpoints further during a subsequent plenary discussion session. The methodology for this exercise was influenced by nominal techniques, such as those described by Potter, Gordon and Hammer (2004). Nominal techniques provide a method of generating information in response to an issue(s) via group discussion involving experts. The following features of nominal techniques were harnessed by this design: Independent generation and face-to-face discussion. Enabling a large number of issues to be discussed. 11 Allowing creative thinking. Encouraging balanced input from all participants. Providing immediate feedback. Allowing the relative importance of different issues to be gauged. Syndicate Exercise 2 – Suggestions of How To Progress the highest Rated ‘Ideas for Improvement’ During Syndicate Exercise 2, each of the high rated ‘Ideas for Improvement’ (from the previous exercise) was addressed in turn. Delegates in each syndicate were asked how to develop the ideas and outline in more detail how the ideas could be taken forward, taking into account two criteria: What does HSE need to do? What is the value of the suggested approach to HSE? The immediate syndicate group outputs were once again captured on feedback slides (Appendix 3, slides 21-30), and fed back by the Syndicate Chair in a plenary discussion session. Conclusion of Workshop Proceedings The event was concluded with a feedback session where both Syndicate Chairs presented an overview of their group’s discussions. The Chair, Patrick McDonald, provided an overall summary of the workshop and explained what the next steps would be. COLLATION OF WORKSHOP OUTPUTS AND REPORT OF PROCEEDINGS Detailed summaries of the workshop proceedings, including summary points from syndicate and plenary discussions, are described in Section 3 (Collation of Workshop Outputs). The summary text is the authors’ interpretation of the workshop proceedings, based on audio recordings made at workshop and subsequently transcribed. The report contents have been reviewed and agreed by technical subject leads from HSE CSAG. 12 3. COLLATION OF WORKSHOP OUTPUTS The following commentary contains comments and viewpoints expressed by workshop delegates drawn from both within and outside HSE and HSL. Therefore, the content reflects the views expressed at the workshop by a range of stakeholders. SUMMARY OF EXERCISE 1 – RATING ‘IDEAS FOR IMPROVEMENT’ The ‘Ideas For Improvement’ discussed in the syndicate exercises are presented in the order in which they were addressed by each group (Appendix 3, slides 3-20). Syndicate group 1 – summary of discussions ‘Idea For Improvement’ #11: “Promote Recording of occupation in general health records” Agreed overall rating: Green The delegates fully accepted that accessing actual patient data (or aggregated data thereof) could potentially be a very useful source of information in the future. Medical personnel could record the occupation of any person presenting to the National Health Service (NHS) for medical treatment. Recording such data could provide the HSE and other interested parties with a broadly based, reliable source of information. However, there were outstanding questions regarding how such a change could be brought into being. One delegate asked the group to note that the response rates to surveys have been decreasing. Therefore, the validity of the information obtained from some of the other potential data sources would decline in the future. This will in turn increase the potential importance and utility of general health records. It was unanimously agreed that HSE should consider how to promote this type of data recording. ‘Idea For Improvement’ #1: “Research the meaning and validity of self reports” Agreed overall rating: Green When discussing the meaning and validity of self-reports there was a broad measure of agreement within the syndicate group that this was an idea that was clearly worth pursuing. It was noted that there is obviously some evidence in this area already and this would make a strong data source for exploring occupational ill health ‘conditions’ (i.e. such as stress and back pain), rather than ‘disease’ (i.e. medically diagnosed disease). It was said that research into the meaning and validity of self-reports should aim to better understand changes over time in reporting behaviour and also to better understand what people mean when they self report illness (i.e. both the nature of the disorder and also why and when they attribute it to their work). 13 ‘Idea For Improvement’ #2: “Research ways of improving self-reports and/or their statistical use (e.g. using adjustments or restrictions by reporter characteristics or disease type)” Agreed overall rating: Green This idea, based around improving the statistical use of self-reports, is related to the previous idea (#1). The syndicate group members strongly agreed with the notion of using adjustments or restrictions by reporter characteristics or disease type and that this should be a priority for HSE to develop. The delegates outlined some potential areas for improvement, including filtering out those conditions for which LFS statistics do not provide good data and also gaining and improved understanding of how to apply the ‘adjustors’. There is a also a need to identify which disorders the self-report data sources are most valid for and would, therefore, be best suited to describe. Furthermore, it was suggested that reporter characteristics could be adjusted to counter bias. ‘Idea For Improvement’ #5: “Research potential uses of coroners' data” Agreed overall rating: Red There was some acceptance across the syndicate group members that in certain circumstances the need may arise for HSE to undertake some ad hoc analysis in this area. However, due to some doubt about consistency of information/reporting and difficulties with obtaining meaningful data about a person’s former occupation from Coroners’ reports, the group considered that this would not be a useful data source for HSE to explore at the current time. ‘Idea For Improvement’ #7: “Systematically review (not literature review!) all disease-exposure links and determine the best estimate source for each” Agreed overall rating: Green The syndicate group delegates discussed the need for a systematic review (noting that this was not intended to be misconstrued as a literature review) of all disease exposure links to determine the best source for each. This would enable data collection to be tailored for particular disorders. This also relates to the previous point (‘Idea For Improvement’ #2) regarding the validity of the data for particular disorders within the LFS. It was stated that this review process has already been undertaken by HSE in some areas (e.g. the THOR database was considered to be the best source for occupational asthma and dermatitis) but this approach has not been applied systematically and unilaterally. There is a need for improved clarity to assist HSE in answering 14 questions about the burden of disease attributable to work and which occupations may have the highest attribution. It was proposed that a systematic review of disease exposure links and best sources could be achieved by asking, for each disease type or illness type, ‘what is our best source(s)?’ then for HSE to use this to inform improvement of these data sources. It was proposed that one of the ways in which HSE statisticians could do this would be to compare different data sources and identify where they provide different results. It was envisaged that this idea for improvement would enable HSE to focus on the best available source(s) for particular disorder and ensure it can optimally serve that purpose. ‘Idea For Improvement’ #10: “Research determinants of IIDB claim propensity and extent of under claiming in order to remove biases that may affect this source” Agreed overall rating: Red The syndicate members agreed that currently this idea was not something for HSE to pursue as a general approach. Recently, HSE has begun to shift its operational focus away from stress and MSDs towards more traditional diseases where there is more opportunity to make an impact on the causes of those diseases. Currently, IIDB claim patterns may have limited value. If however, HSE shifts its focus in the future towards more specific diseases, it was considered that the value of research into IIDB claims may be more useful. ‘Idea For Improvement’ #9: “Conduct better analyses of existing data, especially joint analyses across data sources (e.g. LFS and THOR)” Agreed overall rating: Amber This idea related particularly to building confidence in HSE’s overall estimate for ill health provided to Ministers, etc. However, there was a belief among the delegates that HSE had already conducted these analyses and could not do more in this area. The group considered that analysis across data sources may not be particularly informative with regard to the relationships across the different categories and types of condition. It was considered that HSE could use the systematic analyses described under ‘Idea For Improvement” #7 to gain relevant information on such relationships. This is an area of work that HSE could consider but the group did not view it as one of HSE’s top priorities. 15 ‘Idea For Improvement’ #12: “Research approaches to continuous measures of work related ill health (e.g. DALYs) – so that multiple causation can be coherently handled” Agreed overall rating: Amber/Green It was regarded by the syndicate group that HSE needs to understand which of its data sources are the most powerful and most useful for all the various disorders. This includes an evaluation of the LFS to identify which disorders it currently provides strong and weak data for. The syndicate chair noted that there is a clear relationship between the suggested research work for this research idea and ‘Idea for Improvement’ #2. The syndicate group discussed how continuous measures of disease severity or disease burden might be helpful in understanding both the total impact of work related disease and also, provide a way of unpacking the impacts of different sources of causation. With particular reference to DALYs, one of the advantages of this research would be to ensure that HSE captures all the downstream consequences of work related illness. This information could be applied where HSE needs to cost and justify particular interventions. Syndicate group 2 – summary of discussions ‘Idea For Improvement’ #3: “Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include National Health Service (NHS) data, large scale surveys, longitudinal study, others?” Agreed overall rating: Green The delegates suggested that there were a wide range of existing data sources that were currently ‘untapped’ by HSE. To influence the type of information collected in these larger databases, it was recommended that HSE needed to be more engaged with the research community. The ‘Understanding Society’ panel was mentioned as an example of a large-scale database where there could be opportunities for HSE to develop linkages. The emergence of new online data services was seen as an opportunity to enable linkages between data sets. Also, potential linkages could be made using NHS and Department for Work and Pensions’ (DWP) data: including the birth cohort facility; NI number and NHS numbers. If these linkages were possible, then it was considered that there could be a shift in focus from incidence data towards the antecedents and consequences of work related illness. The delegates considered that this could provide added value information to existing work-related ill health information. Furthermore, it was suggested that linkages would help to identify manifestations of occupational diseases in retired people who may have not been part of a health 16 surveillance programme or would not have details of their former occupation in their medical notes. It was considered that this would enable a better understanding of the role of work related illness in the life course of an individual. Having access to a wider range of databases was said to be useful for the generation of hypotheses, around which to base further research. However, research was deemed necessary to be more critical of data sets. Despite the green rating, some preliminary stages (as outlined above) would need to be carried out prior to this idea for improvement being addressed more fully. ‘Idea For Improvement’#4: “Explore the opportunities presented by the planned new 'Fit Note'” Agreed overall rating: Amber Delegates considered that the opportunities presented by the electronic ‘Fit Note’ should be explored. However, it was pointed out that the occupation of the person, their postcode or workplace location is currently not recorded. However, if this type of information were to be recorded in the future, it would render the data more meaningful for the study of work-related illness. It was noted that any desired future improvements to the ‘Fit Note’ should be instigated at an early stage. The future utility of the ‘Fit Note’ data was said to be highly dependent upon whether GPs would be given the right incentives to obtain and input data relating to occupation. Delegates considered that the ‘Fit Note’ could be a vehicle for changing medical recording. There was potential for an attributable fraction method to produce valuable information on trends, focusing on leading rather than lagging indicators in occupational health. However it was emphasised on several occasions that GPs were not expected to draw conclusions about the attribution of illness. It was said that the training of GPs by the Royal College makes it clear that GPs should exercise caution with respect to the attribution of illnesses such as musculoskeletal disorders (MSDs) or stress. Some illnesses (for example, work related stress, dermatitis or occupational asthma) may not be wholly attributed to work. Delegates stressed that for HSE to utilise the best available evidence, data from the ‘Fit Note’ should be utilised as part of a coherent package of data sources. This idea for improvement was rated as amber because delegates expressed concerns about the practical implications of accessing and utilising the data. It was also noted that there were possible biases in the data, as people who do not take sick leave would not be detected. Finally, there were issues about possible inconsistencies of recording occupational data by GPs. 17 ‘Idea For Improvement’#6: “Develop job exposure matrices for key risks to enhance occupational analyses” Agreed overall rating: Amber The group agreed that the development of job exposure matrices was feasible and would need to include some assumptions about the type of exposures that would be measured. The group could foresee several benefits associated with such job exposure matrices, which could be used for enforcement activities and could also be combined with other published data. However an amber rating was given because the group foresaw some limitations in this approach, such as when ill health was associated with multiple causes. Other limitations of job exposure matrices were also identified. They were: ‘not considered to be useful for all disease end points’; ‘difficult to use from a psychological perspective’; ‘focused on past events’; and were thought to be ‘condition dependent’. It was commented that if job exposure matrices were to be used in a systematic way, there was a need to take these limitations into account. ‘Idea For Improvement’#8: “Develop methods for integrating data from the different sources into a coherent overall ‘best estimate’” Agreed overall rating: Amber It was the delegates’ view, that if a system integrating several data sources could be put in place by HSE (as discussed in ‘Idea For Improvement’ #3), it would provide a means to create a coherent overall best estimate of ill health. The disability-adjusted life years (DALY)4 and quality-adjusted life years (QALY)5 approaches were mentioned as being useful metrics that could enable HSE to ‘speak the same language’ as other government departments. The group thought that this idea for improvement was potentially important for HSE, but would be difficult to put in place (hence the amber rating). Delegates wished this idea to be noted in the workshop proceedings as a first step towards making this improvement. Further steps would involve qualitative inputs, the formation of appropriate linkages and an assessment of the potential impact of these linkages This ‘Idea For Improvement’ would be temporally related to ideas #3 (rated green) and #4 (rated amber). 4 5 DALY definition: http://en.wikipedia.org/wiki/DALY QALY definition: http://en.wikipedia.org/wiki/Quality-adjusted_life_year 18 ‘Idea For Improvement’ #14: “Extending occupational health provision and collecting data from providers” Agreed overall rating: Red The delegates outlined that this idea for improvement, represented two ideas; (i) extending occupational health provision and (ii) collecting data from providers. With respect to collecting data from occupational health providers, the group members raised questions about the quality and consistency of the data that could be collected and the need for a robust quality assurance process in order to produce accurate data. It was considered that the resource implications of such a process would outweigh the potential benefits and for this reason it was rated as ‘Red’. If however, a ‘Fit for Work Service’ became operational (as proposed in the Dame Carol Black Review6) it could help to provide accurate data and the idea could be considered further. This ‘Idea For Improvement’ was linked to ‘Idea For Improvement’ #3 (rated green) regarding HSE using wider available data sources. ‘Idea For Improvement’ #15: “Establishing a cancer review panel” Agreed overall rating: Red This ‘Idea For Improvement’ was rated Red and the syndicate group questioned why a review panel should be limited to cancer only, as there were wider occupational diseases to consider. Also, the attribution of work to cancer was said to be a complex and difficult issue to clarify. Delegates rated this issue red due in part to high resource implications and costs that would be associated with taking this idea forward. ‘Idea For Improvement’ #16: “Establishing an open system for reporting suspect cases” Agreed overall rating: Amber Under the current reporting system, the reporting of cases to databases such as THOR and THOR Extra is restricted to doctors. In this context, the discussion of a more open system refers to the ‘opening up the reporting’, so that other interested parties from OH professionals through to workers themselves could report suspected cases of occupational ill health. Delegates were in favour of opening up the system to OH professionals outside the normal reporting framework (e.g. Occupational Health Nurses and factory inspectors) and considered that this would potential help to expand the dataset. However, as the data quality relies on a good understanding of case definitions, it was considered that 6 http://www.workingforhealth.gov.uk/Carol-Blacks-Review/ 19 an “open for all” system would lead to over-reporting. It was also considered that if such over-reporting occurred, it would be detrimental to data quality and could result in the database loosing its utility. As a relevant example, ‘THOR Extra’ was thought to be a very useful addition to the THOR data. It allowed reporting of additional data outside the normal reporting framework. This would not impact on formal statistical sources and was thought to be a very good source of qualitative data. Delegates regarded these data as providing leading indicators of new and emerging occupational ill health issues. ‘Idea For Improvement’ #13: “Develop and agree a definition of work related ill health (content and scope)” Agreed overall rating: Amber The delegates’ views regarding the rating of this ‘Idea For Improvement’ were divided. It was seen as important (hence rated ‘Green’ by some) to define work related ill health and to pinpoint those individuals whose health is affected by work. The group debated whom, other than directly affected person, the consequences of occupational ill health would affect (e.g. may affect other householders). Some participants viewed this as being an idea for HSE Policy team to take forward, rather than HSE statistics branch. For this reason, many delegates viewed this idea as being outside the focus of this workshop. Some delegates rated it as ‘Red’ because in their view, the content and scope of this question needed further consideration. The group concluded that an overall ‘Amber’ rating was appropriate, as there may be more than one definition of work-related ill health and HSE Policy should undertake to resolve this issue. FURTHER COMMENTS IN PLENARY SESSION 1 The comments in the morning plenary session mainly focused on the job exposure matrices, the ‘Fit Note’ and the attributable fraction studies. Developing job exposure matrices for key risks to enhance occupational analyses It was suggested that the major limitation in general population data is the lack of specificity for exposure information. If, however, the object of the exercise was to assess ‘attributable burden’, then the lack of specificity would not be so problematic. It was considered that if an analysis of general population data demonstrated a high sensitivity (but had a lack of specificity), then it would provide a non-biased estimate. Therefore, this is an area where the job exposure matrices may be useful for application in the general population. The new ‘Fit Note’ Delegates debated the reasons why the medical profession would not necessarily be in a position to cooperate and enter the required occupational data on the ‘Fit Note’. Delegates explained that currently, doctors’ training does not include this topic. Some 20 of the delegates stressed that within the medical profession, the ‘Fit Note’ data could be seen as difficult and troublesome to compile. However, it was also argued that there were reasonably accurate coding tools available which could alleviate some of the effort required for this process. Furthermore, universities were said to collect this data already, as they enquire whether university graduates are employed and their occupation. Half a million cases per year have been recorded in this way. It was concluded that compelling examples such as this could help to demonstrate to the medical profession that establishing such a recording system is possible and feasible. Data linkages Delegates in the plenary session, suggested that the medical profession may not have to input work-related information into medical records if it can be lifted via links from other databases. Scandinavian countries were said to use connected databases in this way. However, the issue of security, data protection and ethics were raised. Establishment of data linkages would require extensive sources of public information in order to initiate such an operation. The delegates acknowledged that, in relative terms, it would be much easier to establish links between existing data fields in databases (such as the patient record) than establishing new data fields. However, the issue of capturing occupation in the patient record was thought to be vitally important for HSE and it was therefore recommended that HSE should act urgently to engage with the NHS about this issue. It was stressed in the plenary discussion that the information, which will be contained in the new ‘shared patient record’ (an interface shared between primary and secondary care), is yet to be determined and this would be an ideal place to capture information about occupation. Attributable fraction studies Occupations are recorded within the LFS. The delegates discussed the scope for making attributable fraction estimates from these data, and concluded that collecting information on health outcomes and making associations between occupations and health outcomes, would be worth exploring. 21 SUMMARY OF EXERCISE 2 - RESEARCH QUESTION GENERATION The syndicate group discussed how the ‘Ideas For Improvement’ rated as Green (in the morning session) could be taken forward and also the value of such approach to HSE. Where time allowed in the sessions, some Amber rated issues (particularly those that were borderline Amber/Green) were also considered. The ideas that were discussed in the syndicate exercises are presented in the order in which the group discussed them (Appendix 3, slides 21-30). Syndicate group 1 – summary of discussions The research areas for syndicate group 1 were discussed in the temporal order in which the group considered the research might be pursued. The syndicate chair indicated that ideas #11 and #7 should run in parallel and required urgent attention. Ideas #1 and #2 were closely linked with #7 but should come later in the temporal order. Idea For Improvement’ #11: “Promote Recording of occupation in general health records” Agreed overall rating: Green The syndicate group delegates considered that it was timely for HSE to develop the business case for promoting the recording of occupation in general health records. HSE would need to identify who the key players are, how to engage with and influence them, and identify the main barriers to achieving this change. It was viewed that it would be important for HSE to simultaneously engage with GP’s and Ministers who have influence with the Department of Health (DH) at senior levels. The delegates discussed the range of ‘agents of influence’. It was considered that GP’s are more likely to adopt changed behaviour when the financial cost to the individual medical practice can be recovered. Also, the National Service Framework was mentioned in this context. It was noted that this idea also aligns with the development of the new ‘Fit Note’. Although the fit note does not include occupation, it was considered that linking these two developments together could be advantageous and beneficial to HSE, GPs and DH. It was suggested that relevant articles in journals read by GPs could be one way of promoting the recording of occupation. Other ‘agents of influence’ noted by the delegates were; the DH social marketing campaign; the Royal College of GPs; and regional GP networks. There was a shared view that this idea aligns with a variety of DH activities under the DH health and work agenda, in particular Dame Carol Black’s recent programme. It was recommended that HSE should liaise with Dame Carol Black’s team to promote this idea. Also, regarding engagement with the DH, the opportunity to use occupational data to reduce pharmaceutical budgets for GP surgeries was discussed. The point was made that the HSE could promote the ‘budgetary advantage’ for GPs recording and having access to occupational data. It was also considered that the medical profession could possibly use occupational data in the medical records to proactively identify people in certain professions who may be at risk of occupational 22 disorders. These people could be offered help and advice sufficiently early to reduce later costs of pharmaceutical treatments. The delegates debated whether it would be advantageous for an individual’s industry to be recorded in addition to their occupation. It was generally agreed that occupation would be the information of primary value, although in some circumstances (for example, back pain or use of solvents) an industry category would also be helpful. ‘Idea For Improvement’ #7: “Systematically review (not literature review!) all disease-exposure links and determine the best estimate source for each” Agreed overall rating: Green (initially Amber) The delegates were asked to consider the temporal order of HSE research for measurement of ill health. They were in agreement that the review of all data exposure links used by HSE should be carried out prior to the research for ideas #1 and #2 (which is described latterly). When considering the best source, delegates recommended that the list of disorders should be first broken down into workable categories. Although requiring further refinement, it was felt that table 2 (within the Discussion Paper), was broadly the right breakdown to work with. The delegates considered that the table provided a clear picture of the sources currently available to HSE, and the estimates they produce. This was felt to be a good starting point. It was recommended that HSE should review and develop the table further in order to evaluate the quality of the estimates. Delegates preferred solutions for determining the quality of estimates were outlined in the first session. It was therefore proposed that HSE should outline its likely course of action, (with respect to the proposed solutions) and subject this to external peer review and consultation with important stakeholders (such as the workshop attendees). It was anticipated that the goal of the review and consultation process, would be for HSE to determine the best sources. Furthermore, it was envisaged that this consultative process might reveal some further research needs regarding the degree of validation for particular sources. Delegates viewed that the desired outcomes of the proposed systematic review would be to: 1. Provide better estimates of the burden of different types of disease 2. Identify which sources are HSE’s primary data source(s) and which are supporting and background sources. The proposed research would identify which data sources should be used and the current limitations of the data sources. Much of this information is already presented alongside HSE data but it is not currently gathered together in one easily accessible place. If HSE produced this information, it could be accessible via HSE’s website and could also be presented alongside overall health statistics, whenever they are produced. 23 There was a thorough discussion about how, in some cases, attributable fraction maybe the best estimate of the burden of a particular disease. However, it was considered that studies of attributable fraction are sporadic and are conducted on an ad hoc basis. Therefore, the question was raised whether HSE can use the LFS data to derive attributable fraction measures. Further exploration of this idea may then inform future work on the LFS estimates as outlined in ideas #1 and #2 (described below). ‘Idea For Improvement’ #1: “Research the meaning and validity of self reports” Agreed overall rating: Green The delegates discussed how to increase understanding of the meaning and validity of self-reports and identify major limitations. The main approach discussed was to validate the gross estimate (of ill health) by following up the specific cases that are reported in the LFS (as was done in 1995). Although some alternative approaches were considered, the general consensus amongst the delegates was to recommend that HSE broadly follows this approach and also considers including some further validation steps. In 1995, the validation exercise involved following up with GPs, where permission had been given to do so. It was debated in the syndicate group whether the GP’s view of the individual case should be considered as ‘the gold standard’ or whether some attributable fraction routes exist that would provide the same information. A further subject of debate was whether the data could be further validated by: (1) obtaining additional qualitative information, by making enquiries directly with the affected individuals; or (2) conducting independent objective assessment of cases in some settings, e.g. to gather more data on self-reported stress. It was noted this was an expensive exercise. ‘Idea For Improvement’ #2: “Research ways of improving self-reports and/or their statistical use (e.g. using adjustments or restrictions by reporter characteristics or disease type)” Agreed overall rating: Green For this research proposition, delegates considered how HSE could improve their analysis of the cases that are obtained from the LFS and supporting data. The syndicate group strongly supported the argument that certain reports from the LFS should be adjusted, restricted or excluded. The basis for this decision was the potential for wide variation in individual perceptions of what constitutes a case of work related ill health, leading to a lack of robustness in the data reported by individuals. Furthermore, some reports are made through SWI by proxy (reported by one individual on behalf of another in the household), not by the individual themselves. The evidence suggests that varying levels of misattribution have been associated with such reports, often leading to under-reporting when compared with first-person reports. It was considered this should be adjusted for. However, it is important not to over-adjust, as some of the differences are known to be genuine. 24 There was an overriding discussion about whether adjustments or restrictions should be applied by HSE to improve the quality of self-reports or whether downstream adjustments should be made to subsequent surveys. Delegates stressed that HSE statisticians would wish to avoid making over adjustments, particularly for factors such as demographic information. For example, if a larger proportion of older people in a particular region were found to be suffering from a particular disorder, then this may be related to previous work in an industry in that region. There is a risk in making adjustments (based on flawed assumptions) that such data could become misleading. Despite the potential risks, the delegates recommended that HSE should explore the potential to make improvements related to various characteristics (including social class, age and region) in addition to the nature of the disorder being reported. The questions that HSE asks in collecting self-report data have, despite some very minor changes, been consistent since the process was introduced in 1990. However, the delegates strongly supported the view that this would be an opportune time for HSE to re-examine the collection of this data. HSE is coming up to the end of the ten-year Revitalising Health and Safety strategy period, with its associated global targets. Therefore, this would provide a timely opportunity to make changes in the question set. Syndicate group 2 – summary of discussions ‘Idea For Improvement’ #3. “Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include NHS data, large scale surveys, longitudinal study, others?” Agreed overall rating: Green The idea of tapping into a wider range of data sources will need to be taken forward at different levels. Firstly, after identifying possible linkages, HSE needs to assess whether they will add value. A strategy plan was suggested. Engagement would initially be made at higher levels (e.g. Chief Executive Officers from appropriate areas such as the Office of National Statistics [ONS], Economic and Social Research Council [ESRC]) to ensure the links were made between institutions, rather than between individuals. This was not seen without difficulties, but there were examples mentioned that suggested that this could be possible. For example, lessons were learned from the Scottish Longitudinal Study (SLS), which has achieved linkages that were initially regarded as impossible. Within the strategy plan it was suggested that HSE should be looking at working more closely with the Medical Research Council (MRC), the Wellcome Trust and the work done with the electronic patient records. HSE could undertake some immediate activities, including joining the UK Data Forum7, which was set up to allow the senior data owners from various organisations to come together and to resolve issues including working with the Office of National Statistics (ONS). It covers government departments, the MRC, the ESRC and 7 http://www.esrc.ac.uk/ESRCInfoCentre/NDS/ukdf/default.aspx 25 devolved administrations. This idea was regarded as having the potential to yield benefits very quickly. It was suggested that ‘NHS Connecting For Health8’ could also be a useful contact for HSE, particularly for patient records. A process of stakeholder engagement was suggested in order for HSE to move forward and to identify whom to engage with at the right level (key people for HSE to engage & contact). Prioritisation of available resources was also needed. The likely benefits of this approach were that access to richer data would enable HSE to measure ill health more accurately. By working in a more structured and strategic way, it would help to get to the roots of the problem. HSE was said to have been too passive in the past and the success of the data linkages that have been forged in Scotland proves how much can be achieved. An initial scoping stage was suggested that would include an inventory of current barriers to success, definition of long-range statistical objectives and an assessment of the quality of data. ‘Ideas For Improvement’ #8 and #16: “Develop methods for integrating data from the different sources into a coherent overall ‘best estimate’” “Develop and agree a definition of work related ill health (content and scope) (new question)” Agreed overall rating: Green/Amber The issue of defining work related ill health raised the question of ‘what are we trying to measure?’ and ‘how this relates to HSE’s mission?’ Participants also asked how much such a definition would need to be changed in relation to the new HSE strategy. For some, definition was a fundamental problem because it has implications on measurement issues. As the current data sources have multiple definitions, each answering a different kind of question, the ‘Idea For Improvement’ needs to be addressed in a practical perspective. A suite of measures rather than a single measure might be needed. These research activities were seen as bringing the highest value to HSE and the group agreed that these two questions if taken together, could be rated as green. 8 http://www.connectingforhealth.nhs.uk/ 26 ‘Idea For Improvement’ #6: “Develop job exposure matrices for key risks to enhance occupational analyses” Agreed overall rating: Green/Amber The development of job exposure matrices was seen as very useful. This would demand a thorough check of the quality and relevance of exposure data (time, quality). This would provide a ranking of relative risks of exposure for different jobs/activities (incorporating relevant hygiene and workplace control measure information). It might work better for more ‘tangible’ diseases (e.g. MSDs) than less tangible ones (e.g. stress). It has to be measured in the relevant context of hygiene information, which needs to use current data sets rather than reflecting the historical perspective. Triangulation across other data sets (e.g. reflecting an individual’s true exposure history) should be considered. A way forward was proposed to first explore the current evidence gaps and which conditions to consider, then, to decide on priorities within the job exposure matrices work stream. FURTHER COMMENTS IN PLENARY SESSION 2 The comments in the morning plenary session revolved mainly around case counting from health surveillance, available data sources and potential partners/collaborators. Improving case counting (case reporting) - links with health surveillance One of the underlying questions of the workshop focused on the limits of case reporting. The workshop delegates were unsure whether individuals or medical Advisers could identify cases and hence, whether these counts could be reliably used by HSE. Delegates from the HSE team highlighted the links between this question, the idea to develop a systematic review (idea #7) and HSE’s ongoing review of health surveillance. Equally, it was said that health surveillance must rely on the assumption that cases can be reliably identified and that, therefore, the limits of case counting must be more or less equivalent to the limits of surveillance. It was proposed that the answers to these questions will be considered by the review of HSE’s health surveillance guidance and that the ill health review should take note of the unfolding health surveillance debate and observe the decisions made about case counting or other indirect methods of estimation. One of the questions considered by the syndicate was whether information from Occupational Health providers can be useful and of suitable quality. From one viewpoint, an example was given of the Occupational Physicians Reporting Activity (OPRA) service. A number of the reporting physicians supply similar data to OPRA that they would also feed internally within their organisation or organisations. Therefore, the data that they provide to OPRA is of the same quality and fulfils both the aims of OPRA and the provider’s organisational purpose. There were thought to be parallels between this and the health surveillance issues (outlined above). The advantage of physicians reporting to OPRA, is that it is able to act as a ‘confidential 27 buffer’, compared with reports being made directly to HSE. Another major advantage of this system is that those who make the reports are able to make enquiries of the OPRA database and obtain anonymised information e.g. ‘What is the incidence of asthma in the chemical/pharmaceutical manufacturing industry?’ The two-way data transaction, allows physicians to receive a benefit from participating in the scheme. Delegates believed that this is an example of the possible linkages between health surveillance and reviewing sources of ill health data. A further viewpoint was put forward regarding case based reporting. It was argued that the basic underlying principle was that three conditions needed to be met: (1) ‘you can reliably diagnose the case’; (2) ‘you can reliably attribute the case’; and (3) ‘that all such cases get reported’. Based on this premise, it was proposed that the success of a case based reporting scheme would depend on the extent to which these criteria are met. For example, Mesothelioma is a case where these criteria would be met because it’s invariably fatal, almost all cases are occupational and all cases are discovered (by counting deaths from death certificates). However, occupational asthma meets the first two criteria as there are good tests for attribution and experts who can carry out those tests. A problem emerges with the third criteria because not all cases are known and only some are reported to the Surveillance Of Work-Related And Occupational Respiratory Disease (SWORD) and OPRA schemes, etc. There are several other disorders that fail to meet most of the criteria. An example such as lung cancer can be diagnosed but the attribution is particularly problematic. Therefore, it is very difficult to count the number of cases. Whereas for disorders such as upper limb disorders, back pain and stress, even the first criteria may not easily be met, depending on how those disorders might be interpreted. Engaging with relevant senior representatives in other organisations to encourage sharing of databases The workshop Chair highlighted that the workshop discussions had suggested that there were potentially useful data sources in existence that HSE have not yet mined properly. HSE would need to engage with the owners (and administrators) of the data sources in addition to other relevant interested parties, such as ‘Dame Carol Black’s team’. The Chair proposed that he was willing to campaign for the important changes that delegates had suggested, and that he would also seek the support of allies from outside HSE in this process. Delegates suggested that the Faculty of Occupational Medicine and the MRC would possibly support the idea of including occupation into health data sets if a case for the useful applications of such data was made. It was agreed that representatives of HSE would pursue this further. Based on the premise that a convincing case for changing the data recorded on health records could be prepared, it was considered that the HSE Board may wish to have a role as ambassadors in engaging Ministers and senior civil servants in the process. Delegates discussed whether these changes could be more readily introduced in Scotland and whether this would serve as a pilot for wider application. It was reasoned that in the past, Scotland has tended to lead the way in some data linkage 28 paradigms. The examples of UK Biobank9 and the SLS were given, where occupational data has been recorded. The practicalities of recording occupation in medical records Some of the delegates advised a note of caution to ensure that HSE advocated the optimal approach to collecting occupational data and applying subsequent coding to terms (used for data analysis). It was considered that various methods and systems could be applied and these would have variable ease of application and quality. It was recommended that these factors should be carefully scrutinised and lessons learned from previous examples where this data has been collected (e.g. UK Biobank). Biobank had previously provided 10 occupational categories for participants to choose from but this approach was not found to be successful and job titles are additionally now collected. When making a case for the collection of occupational data, HSE will need to investigate the feasibility, cost and resources to take the idea forward. Any data collection software used for this purpose would need to be carefully assessed to ensure it matched the data entry requirements and was reliable and easy to use. It was proposed that the occupation field could be completed when a person registers with a practice, or when the patient record is routinely updated. It was suggested that medical clerks could add or update this data or the patient themselves could check and amend the data via a suitable user-interface. Subsequent coding of the data for data analysis purposes could be automated using appropriate software. SUMMARY OF RESEARCH IDEAS The key research areas generated by the two syndicate groups are summarised below, categorised by syndicate and theme. These summaries are the author’s interpretation of the workshop consensus. Syndicate group 1 Idea For Improvement’ #11: “Promote Recording of occupation in general health records” 9 Develop the business case for promoting the recording of occupation in general health records. HSE should identify and engage with the key individuals/organisations who could influence and promote this idea. There is also a need to consider the main barriers to achieving this change. Consider what would influence the recording of occupation in medical records. o Suggested influencing agents included financial benefits to the individual medical practice, the National Service Framework and the new ‘Fit Note’. www.ukbiobank.ac.uk/about/what.php 29 ‘Idea For Improvement’ #7: “Systematically review (not literature review!) all disease-exposure links and determine the best estimate source for each” Identify the best data sources for disease-exposure links (this process would be subject to external peer review and consultation) to evaluate the quality of the estimates of work related ill health and the current limitations of the data sources. o The HSE discussion paper provides a clear overview of the data sources currently available to HSE which could be reviewed and developed in order to: Provide better estimates of the burden of different types of disease Identify which sources are HSE’s primary data source(s) and which are supporting and background sources. Explore whether the Labour Force Survey (LFS) data could be used to derive attributable fraction measures o Further exploration of this idea may inform future work on the LFS estimates as outlined in ideas #1 and #2 (described below). ‘Idea For Improvement’ #1: “Research the meaning and validity of self-reports” Increase the understanding of the meaning and validity of self-reports and identify the major limitations of this data source. o Validate the gross estimate of ill health. One way to approach this would be to follow up specific cases reported in the Labour Force Survey with GPs. Consider whether the data could be further validated by: o Obtaining additional, qualitative information through direct contact with the affected individuals. o Conducting independent objective assessment of certain disease categories, (e.g. to gather more data on self-reported stress). It was noted this would be an expensive exercise. ‘Idea For Improvement’ #2: “Research ways of improving self-reports and/or their statistical use (e.g. using adjustments or restrictions by reporter characteristics or disease type)” Certain reports from the Labour Force Survey should be adjusted/restructured (or excluded) to reduce the misattribution of work relatedness. o This includes the potential to adjust for variations in individual perceptions of what would constitute a case of work related ill health and the characteristics of self-reports (including social class, age and region) in relation to a particular disorder(s). Re-examine the collection of self-report data and identify whether any changes should be made to the question set. 30 Syndicate group 2 ‘Idea For Improvement’ #3. “Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include NHS data, large scale surveys, longitudinal study, others?” HSE should engage with individuals at the most senior level (e.g. CEO) within appropriate organisations (e.g. the Office of National Statistics, the Economic and Social Research Council, the Medical Research Council and the Wellcome Trust). o Join the UK data forum and ‘NHS Connecting For Health’ Identify possible links with other data sources and assess whether they will add value. o Examine lessons learned from similar successful ventures e.g. the Scottish Longitudinal Study. o Conduct an initial scoping study, to include an inventory of current barriers to success, definition of long-range statistical objectives and an assessment of the quality of data. ‘Ideas For Improvement’ #8 and #16: “Develop methods for integrating data from the different sources into a coherent overall ‘best estimate’” “Develop and agree a definition of work related ill health (content and scope) (new question)” Current data sources are thought to have multiple definitions of work related health. It is important to define what should be measured and how this relates to HSE’s current strategy. o A suite of measures to define work related ill health, rather than a single measure might be needed to address these Ideas for Improvement. o ‘Idea For Improvement’ #6: “Develop job exposure matrices for key risks to enhance occupational analyses” The development of job exposure matrices would require a thorough evaluation of the quality and relevance of exposure data. This would provide a ranking of relative risks of exposure for different jobs/activities (incorporating relevant hygiene and workplace control measure information). o Current data sets would need to be used rather than reflecting the historical perspective and gaps in current knowledge would need to be explored. Triangulation across other data sets (e.g. reflecting an individual’s true exposure history) should be considered. 31 CLOSING COMMENTS One important ‘take home’ message (presented by one of the delegates) was that it is never going to be possible to have perfect work-related ill health data. However, HSE’s further research will help to identify which information is ‘as good as is reasonably practicable’. For these purposes, consideration needs to be given to practically obtaining data that is not too expensive and is fit for purpose. It was stated that the current position is not, in fact, far removed from these goals and the workshop recommendations would lead to incremental rather than radical improvements. In concluding the workshop, representatives of the HSE statistics team highlighted two broad strands of recommended improvements for inclusion in the work-related ill health measurement review. It was suggested that these two areas were connected and could be established as two parallel programmes, with the following aims: To clarify the leading data source for each of the different components of work-related ill health, and (where necessary) incorporate supporting research. This would involve two main components: o Develop a systematic review identifying the best source(s) of data for each category of work-related ill health; and use this review to identify and prioritise further development work. o Test and improve the validity of self-reports in those areas where it is evident that they provide the best available estimates. To enable the exploitation of general health databases in an occupational context by attaching occupation codes at the individual record level, either via direct collection or via establishing data linkages. o Initiate a process of engagement between HSE and other relevant organisations (including the UK Data Forum), to promote the identification and sharing of databases. 32 PROPOSED NEXT STEPS The HSE team plan to draw the workshop outputs together into a work plan, which would be prioritised according to the temporal order of research and management of time and resources. The Chair advised delegates that he would make a strong case to the Board that resources are needed in this area. It was suggested that delegates would be kept engaged with this process, in part, by receiving the HSL report of workshop proceedings. Delegates also suggested that the dialogue between themselves and HSE could be maintained by circulation of HSE’s proposals in relation to the review of data sources for different occupational disorders. Delegates advised that they would be open to involvement in further consultation regarding future proposals. It was suggested that parts of the consultation process could be conducted via HSE’s website to promote wider consultation. It was proposed that an actively managed consultation process would yield the best response rate. In closing, the Chair stressed that it was important that stakeholders involved in this process are able to see the practical applications that will emerge from the workshop. Important decisions will be made and these will be influenced by the contributions made at the workshop and in the dialogue and consultation that will follow. 33 4. REFERENCES Potter, M., Gordon, S. and Hamer, P. (2004). The nominal group technique: A useful consensus methodology in physiotherapy research. New Zealand Journal of Physiotherapy, Vol. 32, No. 3, pp126-130. 34 5. APPENDICES APPENDIX 1 – DISCUSSION PAPER 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 APPENDIX 2 - INTRODUCTORY PRESENTATION, PATRICK MCDONALD AND JANE WILLIS 64 65 66 67 APPENDIX 3 - SYNDICATE CHAIRS’ FEEDBACK SLIDES Health and Safety Executive Workshop 19th February 2009 Summary Presentation Slides HSE’s Measurement of work-related ill-health – is it fit for purpose and what are the priorities for improvement? Health and Safety Executive Workshop Contents Morning Session Ratings Exercise Slides 3-11 Presented by John Ewins Slides 12- 20 Presented by Andrew Curran Afternoon Session Suggested Approaches Slides 21-25 Presented by John Ewins Slides 26-30 Presented by Andrew Curran 68 Ratings Exercise Group 1 – Presented by John Ewins Rating grid for development ideas How likely is High it that this question can be successfully Moderate addressed? to low How important is progress on this question to improving the measurement of work-related ill health? High Moderate to low GREEN AMBER AMBER RED Ratings Exercise, Group 1 Ratings Exercise 11) Promote Recording of occupation in general health records Green Key points / rationale: • Unanimous agreement for this idea for improvement to be progressed Ratings Exercise, Group 1 Ratings Exercise 1) Research the meaning and validity of self reports. Green Key points / rationale: • Research evidence exists but more in certain areas • Need to use these for ill health rather than ‘disease’ •Need to know changes in propensity to self report over time •Understand better what people mean when they self report e.g.rise in stress – what does stress mean to those people experiencing it? • Includes disorders and attribution Ratings Exercise, Group 1 69 Ratings Exercise 2) Research ways of improving self-reports and/or their statistical use (e.g. using adjustments or restrictions by reporter characteristics or disease type) Green Further explanation: some categories of self reports could be set aside to filter out some disorders eg cancer – individual does not know if they are work related eg some somatic conditions Key points / rationale: •Better understanding of how to apply adjusters is accepted fully •What disorders are these data sources most valid for and most suited to? •Could be informed by Idea 1) •Reporter characteristics: could adjust for bias Ratings Exercise, Group 1 Ratings Exercise 5) Research potential uses of coroners' data Red Key points / rationale: • Inconsistency in decisions • Difficult to get meaningful results from reports • There may be useful ad hoc studies done in the past but not for HSE to follow up Ratings Exercise, Group 1 Ratings Exercise 7) Systematically review (not literature review!) all diseaseexposure links and determine the best estimate source for each Green Originally amber Key points / rationale: • For each disorder of interest to HSE: which sources are available, which are best? Tailor data collection according to disorder •Burden of disease attributable to work is of interest to HSE and where is it •E.g occ asthma/dermatitis – THOR •Doesn’t HSE do this already? Yes, but not so systematically Ratings Exercise, Group 1 70 Ratings Exercise 10) Research determinants of IIDB claim propensity and extent of under claiming in order to remove biases that may affect this source. Red Key points / rationale: • Red, but maybe if focus is on certain diseases in future then this could be pursued •For which disorders is this going to be the primary source of information? Possibly deafness… •HSEs interest in OH is narrowing (‘back to basics’), range of conditions in IDB is closer to HSE core business •‘Take up’ is of interest to DWP – scope to combine resources on that? Could discuss with DWP representative •Claims for benefit data tend to be very unreliable, influenced by media etc, Ratings Exercise, Group 1 Ratings Exercise 9) Conduct better analyses of existing data, especially joint analyses across data sources (e.g. LFS and THOR). Amber Key points / rationale: • Better understanding of what the numbers mean •Headline figures (eg for MPs): could use data from 7) to hone in on specific diseases •Use this for overall estimate of ill health? Could use intermittently for comparative purposes. Also use other external sources of data, eg ‘one off’ surveys •Use this with caution, look back at case definition always •Related to 1) How do people progress through the steps: one person goes to GP and another does not, how does this link together, understand disparities. Ratings Exercise, Group 1 Ratings Exercise 12) Research approaches to continuous measures of work related ill health (e.g. DALYs) – so that multiple causation can be coherently handled (suggested by FH) Amber/Green Key points / rationale: • Approaching stats from case counting perspective is an oversimplification •To allow for multiple causation: instead of all or nothing – e.g. count disability life yrs lost – a proportion, a bit like attributable fraction •Could be applied to individual disorders • ?Emphasis on outcome being dimensional, rather than exposure. Yes to multiple causation but is this the right way? •‘All or none’ eg cancer or spectrum of severity eg COPD/hypertension. Attributable fraction vs distribution of severity across popn caused by work •Health of working age popn – a general measure of time working age people are well/not well is useful for making business case to industries. •DALYs (separate question) what is the overall burden on the popn? Most likely to be useful for deafness/COPD re:causation •Improves our understanding of impact of WRIH Ratings Exercise, Group 1 71 Ratings Exercise Group 2 – Presented by Andrew Curran Rating grid for development ideas How likely is High it that this question can be successfully Moderate addressed? to low How important is progress on this question to improving the measurement of work-related ill health? High Moderate to low GREEN AMBER AMBER RED Ratings Exercise, Group 2 Ratings Exercise 3) Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include NHS data, large scale surveys, longitudinal study, others? Green Key points / rationale: •Focus should not be just on analysing incidence but also on the antecedents and consequences of work related illness for which there is not enough data especially on linkages. •Linkages to other important surveys need to be made. This can be achieved if HSE is more engaged with the research community. “Understanding Society” panel (Significant number of people; Add on modules; New data services coming on line; Linkages NHS/DWP; Birth cohort facility; link between NI number and NH numbers (could be done as pilot?). •This would help with issues associated with manifestation in older people (legacy issues), work related incidents, etc. •There may be issues around the timing of doing this. •Would get a better understanding of the role of work related illness in the life course of an individual. •Could lead to some hypotheses generation. •Research needs to be done to be more critical of data sets. Ratings Exercise, Group 2 Ratings Exercise 4) Explore the opportunities presented by the planned new 'fit note' AMBER Key points / rationale: •Opportunity to create a good data source in the future and to influence at an early stage •Fundamental shift. No occupation/ No postcode at present. In the future may be paid and this information will be useful •Attributable fraction method proposed •May be useful for trends – leading vs lagging indicators •Should not be seen as the only source of information •Could be a vehicle to change the medical record •Needs to be more of a coherent package •Need to put in working hours •Rated ‘Amber’ because worries about practical side/political issues and because there might be biases in the data •Would miss people who don’t take sick leave •Issues about quality of recording occupational issues by GPs Ratings Exercise, Group 2 72 Ratings Exercise 6) Develop job exposure matrices for key risks to enhance occupational analyses AMBER Key points / rationale: •Currently associations used but no explanations about the mechanisms behind the risk are provided, this raises questions about true causality •Feasible to do (with assumptions) •Can be used for enforcement purpose •Can be combined with other published data •Links to job title – Back to fit note •Problem of multiple associations •Not the approach to be used for psychological processes •Looks back /does not take new interventions •Limitations and implications need to be understood •Depends on which conditions you are looking at •Limitations on how the information will be used Ratings Exercise, Group 2 Ratings Exercise 8) Develop methods for integrating data from the different sources into a coherent overall “best estimate” AMBER Key points / rationale: •Work to be done on linkages and impact of these linkages •Temporal relationship between this question and questions Q4 and Q3 •DALY/QUALY would carry currency with stress in Government •This event is first step towards this goal •Essential but will involve qualitative inputs Ratings Exercise, Group 2 Ratings Exercise 14) Extending occupational health provision and collecting data from providers RED Key points / rationale: •There are two questions in one •Will be as good as quality of data from providers •Resource implication •Fit for work services might help •Rated as RED but if there was a system which works properly it would then be linked to Q3. •HSE should not perhaps be focusing on this right now Ratings Exercise, Group 2 73 Ratings Exercise 15) Establishing a cancer review panel RED Key points / rationale: •Wider occupational diseases and work cancer attribution difficult •Should they be assessed in more depth •Why cancer? •Huge cost •Resource implications are huge Ratings Exercise, Group 2 Ratings Exercise 16) Establishing an open system for reporting suspect cases AMBER Key points / rationale: •Allows reports outside reporting frame/specialists •Would not impart on formal statistical sources •Very good for qualitative information •Sentinel •Leading indicators •Depends on definition of open – open to all? Ratings Exercise, Group 2 Ratings Exercise 13) Develop and agree a definition of work related ill health (content and scope) AMBER Key points / rationale: •Important for defining boundaries of HSE’s and links to others (hence rated Green) but also rated as Red because need more time to think about this question in term of content and scope, etc. •Cannot be done easily •Not a single definition •Policy question for HSE? Ratings Exercise, Group 2 74 Suggested Approaches Group 1 – Presented by John Ewins •How should HSE take the Green/Amber rated ideas forward ? •Consider what is the value of this approach to HSE Suggested Approaches, Group 1 Suggested Approaches 11) Promote Recording of occupation in general health records Green • Must have a scientific case •Identify what the barriers are and who can make it happen. Need GPs on board. Uniform system. Ministerial lever. Suggestion: get GPs interested via journals/GPs will do it if they are paid to do it. •Agents of influence? Service framework – ‘points mean prizes’, DH Health and Work agenda (HWWB analytical steering group), DH social marketing campaign, RCGP, regional GP networks, reducing pharma budgets by making better diagnoses • e.g. data collection: touch screens for patient data entry? • Base record: GP patient record (held by GP/hospital). Need to find out how HSE would set about doing this and how to influence that •Discussion point: industry vs occupation – occupation is most value but industry is also relevant. Eg back pain, solvents, occupation is more relevant than industry •Write a case for the type of analyses we want to do •Fitnote – GP does not fill in occupation but linked to pt record if occupation is on there Suggested Approaches, Group 1 Suggested Approaches 7) Systematically review (not literature review!) all diseaseexposure links and determine the best estimate source for each Green •Step 1 - List disorders (how much detail?) e.g. along the lines of breakdown level as in Table 2 discussion paper • Step 2 – Decide on best source(s) to make estimate: reach decision via ‘Delphi’ process, peer review, consultation, • Further research comparing certain sources, if req’d • Outcomes (policy etc) - would use sum of best estimates to provide a total. E.g. EPIDERM may give best exposures data but SWI gives best estimate of total – what do you use to do a regional breakdown, for example? • Prime purpose: attributable burden. Other sources more useful for breakdown, need to be considered along side • What are limitations of best possible source • Attributable fractions wouldn’t have as frequent an update cycle • Using the LFS to produce attributable fraction estimates - need a sensitive matrix (putting confounding aside) -so as not to miss exposures e.g. skin cancer through sun exposure @work 75 Suggested Approaches, Group 1 Suggested Approaches 1) Research the meaning and validity of self reports. Green •Identify what are major limitations within the data source. Interpretation and Improvement •Burden in different groups of people and disorders, how robust data are and link to validity. Don’t validate everything, start with where you want to end up • Like 1995 validation? Do we follow up specific cases on which the LFS is used to base the gross estimate? YES •LFS/SWI have taken GP opinion as being a gold standard – is that appropriate? Or compare with attributable fraction, not gold std, where do we know where the truth is? • Qualitative research, e.g. to follow up individual cases reported in LFS. Comparison with other sources, independent objective assessment of cases in some settings e.g. for stress (expensive) • Suggested Approaches, Group 1 Suggested Approaches 2) Research ways of improving self-reports and/or their statistical use (e.g. using adjustments or restrictions by reporter characteristics or disease type) Green • Restrict to conditions for which self reporting is preferred method. •Proxies record lower levels of SWI than first person – adjustments possible there. Variable amounts of misattribution, etc. Don’t want to over-adjust, some diff’s are genuine. • Where cases not validated by GPs (11%) info could be used to adjust or highlight limitations, latter preferred • Reporter characteristics: e.g. age, sex, region, social class. Applied as a test on data or standing adjustments? • Might be better ways of asking the questions – but risky to change them due to discontinuity. Less risk of that at present. E.g. stress vs mental stress. Research could assess if changes are worthwhile Suggested Approaches, Group 1 76 Suggested Approaches Group 2 – Presented by Andrew Curran •How should HSE take the Green/Amber rated ideas forward ? •Consider what is the value of this approach to HSE Suggested Approaches, Group 2 Suggested Approaches 3) Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include NHS data, large scale surveys, longitudinal study, others? (1) Green •Taken forward at different levels; Identify linkages; •Identify what is in scope for HSE •HSE policy to engage with THOR-GP more than LFS •Specific issues: Engagement needed at the right level. High level negotiation needed e.g. CEO’s from appropriate areas (National statisticians, ESRC, HSE) •HSE to join the UK data forum (2005) – SRO’ for data – help resolve issues; Covers Govt departments/MRC/ ESRC/Devolved administrations •HSE should be looking at the Scottish Longitudinal Study (SLS) which has achieved what was regarded as impossible linkages. SLS has overcome problems that ONS has not • HSE should make use of the electronic patient records, talk to the Medical Research Council; OSCHR, etc. •Some difficulty to get collaboration from the Health Protection Agency, HPA. A framework is needed •HSE needs clarity about what it needs Suggested Approaches, Group 2 Suggested Approaches 3) Tap into a wider range of data sources to analyse incidence differentials by occupation – examples potentially include NHS data, large scale surveys, longitudinal study, others? (2) Green • HSE to contact Connecting for Health • Needs a process for HSE to move forward and to identify who to engage at the right level (Suggestion to carry out a “stakeholder” analysis (key people for HSE to engage & contact)) • Prioritisation for available resources • Benefits to HSE: to get to the roots of the problem; too passive in the past; Scottish experience shows that it works • Scoping stage: Needs an inventory of current barriers to success; long range statistical objective must be clear; quality of data Suggested Approaches, Group 2 77 Suggested Approaches Discussion about both questions: Q8:Develop methods for integrating data from the different sources into a coherent overall “best estimate”) Q 16): Develop and agree a definition of work related illhealth (content and scope) (new question) Amber/green •What are we trying to measure and how does that relate to HSE’ s mission? •These two questions taken together could be rated as green •Data sources: (multiple definitions) •Might need a suite of measures •Need a clear idea of how this can be achieved practically •Value to HSE: Needs to be taken forward; Fundamental; highest value to HSE Suggested Approaches, Group 2 Suggested Approaches 6) Develop job exposure matrices for key risks to enhance occupational analyses Amber/green •Job exposure matrix: on the face of it, very useful •But must understand relevance of exposure data (time, quality) •The strength is the relative risk ranking it produces •Might work better for more ‘tangible’ diseases •Must be linked to additional hygiene and other workplace (e.g. ergonomics) measures • +ve versus -ve ; what are the gaps; which conditions to consider? 78