Skip To Content

The Role of Workers Compensation-based Data in the Development of Effective Occupational Health and Safety Interventions

Written and prepared by: Greg Foley

Statistics Unit, Worksafe Australia - June 1996

© Commonwealth of Australia 1996

ISBN 0 644 45922 0

This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by any process without prior written permission from the Australian Government Publishing Service. Requests and inquiries concerning reproduction and rights should be addressed to the Manager, Commonwealth Information Services, Australia Government Publishing Service, GPO Box 84, Canberra ACT 2601.

Acknowledgements

I would like to thank Neill Stacey, Tim Driscoll, Petra Macaskill, Claire Mayhew, Tim Williams and Brad Cole of Worksafe Australia for their helpful comments and suggestions in the preparation of this publication. Their assistance in this regard has been most useful and greatly appreciated. However, any errors in content or faults in the logic or presentation of the publication remain soly mine.

Greg Foley

Contents

Abbreviations

Executive Summary

Introduction

Workers' Compensation Systems in Australia

The National Data Set for Compensation-based Statistics

The Objectives of Occupational health and Safety Surveillance

Some Criticisms of Workers' Compensation Data and Some Thoughts in Response

Understatement of the `True Extent' of the Problem

Bias in Reporting/Nonreporting

Coverage of Diseases is Poor

Exclusion of Self-employed Workers

Lack of `Near Miss' Data

Data Inconsistency Over Time

Untimely Data Release

Advantages of Workers' Compensation Data in Comparison to `Alternative' Data Sources

How Compensation Data can be Used

Conclusions

Appendix 1 - Extracts from Occupational Health and Safety Performance Overviews, Selected Industries

Appendix 2 - Key Aspects of Selected Industry Occupational Health and Safety Performances as Revealed by Analysis of Workers' Compensation Data

References

Further Reading

Abbreviations

ABS - Australian Bureau of Statistics

NDS - National Data Set for Compensation-based Statistics

NESB - non-English speaking background [workers of]

OHS - occupational health and safety

Executive Summary

Work-related injury and disease in Australia is associated with literally thousands of deaths, hundreds of thousands of injuries/illnesses and tens of billions of dollars of expenses each year. In addition, there is a high human cost which can be thought of in terms of physical pain, loss of future prospects and reduction in the quality of life. To confront this problem and develop viable preventive measures for the reduction of the burden it imposes on the Australian community, reliable data on the occupational health and safety (OHS) experience of the workforce are an essential prerequisite.

A major source of data on work-related injury and disease in Australia has been workers' compensation data. These data have been collected, since 1993, in accordance with standards detailed in the National Data Set for Compensation-based Statistics (NDS)[1] and its associated classificatory systems. However, there has been considerable criticism of workers' compensation data, with varying degrees of severity, as a useful source of information in the task of improving OHS performance. The purpose of this publication is to address some of these criticisms, assess their validity and importance, and present some ideas on the practical role of workers' compensation data in the development of effective interventions to improve the OHS experience of Australian workers.

In the process of assessing purported limitations of workers' compensation data, the availability of reliable OHS data is considered to serve six objectives (in no particular order of priority):

  1. Hard data on the prevalence of occupational injury/disease highlights its importance relative to other public health problems and thereby establishes its claim for resources and legislative support.
  2. Analysis of objective data provides a basis for determining research and control priorities.
  3. It supports action strategies, through data-driven targeting of problem industries and occupations, to minimise the impact of risk factors associated with occupational injury and disease. (Concomitant with data-driven targeting is the opportunity to identify better performers and, therefore, best practice approaches.)
  4. Following implementation of intervention strategies, analysis of data allows evaluation of the effectiveness of those interventions and monitoring of overall progress with OHS performance improvements.
  5. Organised, pertinent data can be disseminated to all stakeholders, including workers, employers, unions, government agencies and the general public, to raise awareness of the problem and reduce the general cooperation and commitment essential for effective action to diminish the problems of poor OHS performance.
  6. To assist in the early identification of new workplace hazards.

Several criticisms of workers' compensation data are identified and are discussed against the background of the above objectives and in terms of NDS-based data.

One of the most frequently cited limitations of workers' compensation data is that they greatly understate the `true' extent of the OHS problem in Australia. To some extent, this type of criticism appears to stem from misapprehensions regarding the scope (what is intended to be covered) and coverage (what is actually covered) of workers' compensation data.

The scope of NDS data is the more serious work-related injuries and diseases, which are validated by the various compensation systems as work-related. While there are a large number of less severe cases, it is argued that a focus on more severe cases facilitates the speedier achievement of the maximum reduction in human suffering and economic costs within a given set of budget constraints. It is also argued that the fact that only compensated cases are covered can be considered a strength of the data because cases have been confirmed as work-related. Furthermore, the most meaningful measure of performance is rates of occurrences per number of workers (incidence rates) or hours worked (frequency rates) and, as is pointed out, the numerator data and denominator data used in calculation of NDS rates are consistent in scope. Therefore, it is argued, there would have to be quite significant differences between the injury/disease experience of workers covered by compensation provisions, vis a vis those who are not, for these rates to be invalid measures of the OHS experience of specific groups of workers (it is suggested that it is unlikely that any differences would prove sufficiently large to be considered truly significant).

Coverage concerns which are centred on the degree to which workers are aware of their rights under workers' compensation do not appear to be supported by available data showing that a high proportion of workers are, in fact, aware of their rights. Claims suppression and contestation have more important implications for coverage, and it is agreed that there is some degree of coverage deficiency as a result. However, it is suggested, on an intuitive basis, that undercoverage resulting from these causes is more likely to impact on cases involving less than five days lost time from work, which are outside the scope of the NDS in any event.

The problem of bias in reporting/nonreporting is recognised and addressed. Broadly speaking, it is argued that industry and occupation under-reporting bias does not greatly undermine the utility of NDS data. Data available from the Australian Bureau of Statistics (ABS) imply that underreporting by workers from a non-English speaking background (NESB) is not a highly significant problem. There is insufficient quantitative data available to ascertain whether or not there is reporting bias in relation to certain types of injury/disease.

It is recognised that there are problems with NDS data relating to disease. However, it is pointed out that any attempt to measure work-related disease is fraught with difficulty, simply because of the innate difficulty in unequivocally determining that a disease is actually `work-related'. It is considered that, treated with caution, workers' compensation data can be used as a guide (particularly if complemented by the knowledge and experience of OHS practitioners) in making decisions regarding the need for more in-depth study of particular diseases.

The exclusion of self-employed workers from NDS data is highly unlikely to make decisions regarding OHS interventions, based on these data, entirely irrelevant for the self-employed. For this to be the case, the OHS experience of the 83% of workers covered by compensation would have to be significantly different from the 16% of workers who are self-employed. There is little convincing evidence that this is the case (once collection methodologies are taken into account), although longer hours worked by the self-employed and, therefore, longer exposure to risk factors, might result in higher numbers of cases. In these circumstances, it might be expected that the frequency rates of occurrence would be much the same as for employees.

It is argued that for most purposes it would be inappropriate, and in many instances counterproductive, to associate `near miss' data with data obtained for the objectives of OHS surveillance listed above as numbers 1 to 5. It is suggested that, in the main, `near miss' data can best support trouble shooting systems, which must be highly responsive to specific hazardous situations as they emerge in order that occurrences can be either entirely avoided or minimised.

The question of consistency in workers' compensation data over time is briefly considered. Data on fatalities and diseases is likely to be vulnerable to inconsistency, to some extent, as jurisdictions change policies regarding claims contestation and the types of cause-effect relationships readily recognised in these types of claims. Data on injuries are considered to be more stable, as long as jurisdictions adhere to NDS thresholds. While there are some short term concerns in this regard, at this stage it is considered that any resultant inconsistencies will be restricted to the 1995-96 and 1996-97 periods.

A number of extracts from Worksafe Australia's publication series Occupational Health and Safety Performance Overviews, Selected Industries are presented as illustrations of how compensation data might best be used. These extracts show that compensation data can identify problem areas for an industry and/or occupation. Subsequently, more in-depth research can focus on those problem areas and isolate important causal factors. Ultimately, this should lead to development of specific preventive measures. It is suggested that without the starting point provided by NDS-based compensation data, the improvement of OHS performance in Australia would largely become a process of disjointed incrementalism.

It is noted that workers' compensation data are capable of being used, to varying degrees, to meet the requirements of all OHS surveillance objectives (with the exception of `near miss' situations). As administrative by-product data, they are quite low cost, particularly in comparison to the amount of detail obtained. Furthermore, there are few cost-effective and reliable data alternatives.

It is concluded that many of the perceptions of weaknesses in compensation data are significantly overstated. In fact, it is considered that many of the problems with compensation data might lie more with a need to ensure users are fully alert to the implications of its source, scope, coverage and utility, rather than with any intrinsic deficiencies in the data themselves.

Finally, it has to be recognised that there is unlikely to be one data source which will, on its own, fully satisfy all the requirements of OHS surveillance. NDS-based compensation data are a sound starting point for further OHS data-driven research. They represent the `top level' picture which highlights aspects of performance where an investment of resources is likely to be most productive.

Introduction

Work-related injury and disease has a highly damaging impact on the general Australian community. It causes literally thousands of deaths, hundreds of thousands of injuries/illnesses and tens of billions of dollars of expenses each year[2]. Moreover, there is the social (or human) cost dimension of this problem which can be thought of in terms of the physical pain, loss of prospects for further development, general decline in quality of life, etc, which impact on individuals who suffer work-related injury or disease.

The need for reliable data on the OHS experience of Australia's workforce is accepted by OHS practitioners as an essential prerequisite to the development of viable preventive measures for the reduction of this burden and for monitoring progress in doing so. The practical uses of these data which have been put forward to support the need for them include the need to objectively determine the size of the problem, to develop interventions to reduce the extent of the problem through comparisons between industries and across jurisdictions, and to monitor progress in dealing with the problem by comparison of performance measures over time.

One of the major sources of data on work-related injury and disease in Australia has been workers' compensation data. These data are produced largely as a by-product of the operation of the various Australian workers' compensation schemes. Steps have been taken in Australia to standardise the recording of these data by the various jurisdictions and thereby enable their use as national statistics. Nevertheless, there has been continual criticism, with varying degrees of severity, of workers' compensation data as a useful source of information in the task of reducing the burden upon the community of poor OHS performance. This criticism of compensation data is most frequently predicated on their inherent scope and coverage limitations and their inconsistency over time, due, it is argued, primarily to the ease with which they can be affected by government/bureaucratic interference.

The purpose of this publication is to address some of the criticisms raised, comment on their validity and importance, briefly consider whether potential alternative sources might obviate perceived problems and, finally, to present some ideas on how workers' compensation data might be used in practice to assist in improving the OHS experience of Australian workers. With regard to this latter purpose, some examples are provided at Appendix 1 of findings from analyses undertaken during preparation of various issues of Worksafe Australia's publication series Occupational Health and Safety Performance Overviews J Selected Industries. These examples hopefully illustrate how compensation data might initially be used most effectively.

Workers' Compensation Systems in Australia

Each State and Territory in Australia has its own compulsory workers' compensation system, with Commonwealth Government employees being covered by a separate system. All systems are based on the premise that employers are liable for work-related injuries and illnesses suffered by their employees. However, there are differences between the systems, primarily in the areas of administration, insurance arrangements, benefits payable, threshold limits, dispute resolution procedures, and rehabilitation and return-to-work strategies and programs.

All of the above-mentioned system differences will have some effect, among other things, on the proportions of employees entering the compensation system in one State or Territory vis a vis another, and the length of time for which they will be compensated (or, put another way, the number of working days which will be lost for each occurrence). Perhaps the most significant factor causing inconsistency between workers' compensation systems across Australia (particularly in terms of statistical output) is the variation in threshold. This is tied to the employer excess operating in each workers' compensation jurisdiction. Western Australia, Tasmania, the Australian Capital Territory and the Commonwealth have no excess applicable, with the result that notionally all cases reported should enter the system. In Queensland and the Northern Territory, the excess effectively means that cases (other than those involving a fatality or permanent disability) enter the compensation system if they entail more than one day's time lost from work. In New South Wales, the excess is the first $500 of wages paid. In South Australia, the excess is in the process of changing from the first week of incapacity to the first ten days. In Victoria, it is the first ten days.

Obviously, any statistical data produced as a direct by-product of these differing systems, with differing in-built classification standards, etc, would be of highly dubious integrity and utility. In response to this problem, the National Occupational Health and Safety Commission developed the NDS.

The National Data Set for Compensation-based Statistics

The workers' compensation data which are used as the basis for evaluation in this publication are those collected in Australia in accordance with standards detailed in the NDS and its associated classification systems. These data are provided to Worksafe Australia by the various workers' compensation jurisdictions for creation of a national data set.

As noted above, workers' compensation systems and, therefore, statistics available from them, vary across workers' compensation jurisdictions in Australia. The NDS was created to enable data to be collected from these systems which would be largely unaffected by the differences between the systems. Therefore, it would probably be worthwhile at this point to briefly provide some background on the NDS and its implications for data quality.

The NDS was approved by the National Occupational Health and Safety Commission in February 1987. It recommends a standard set of data items, concepts and definitions for inclusion in workers' compensation collections. It is supported, in practice, by several classification systems including the Australian Standard Industrial Classification, the Australian Standard Classification of Occupations and the Type of Occurrence Classification System [NOHSC:12006(1990)] which is consistent with the International Classification of Diseases (Ninth Revision) and covers nature of injury/disease, bodily location of injury/disease, mechanism of injury/disease and breakdown agency.

Its primary purpose is to enable the production of national and nationally comparable workers' compensation-based data. Since the release of the NDS, it has been progressively implemented in the State, Territory and Commonwealth workers' compensation systems in Australia. The first data produced under this system, relating to the 1991-92 reference period, were provided to Worksafe Australia in late 1993.

Development of the NDS largely depended on gaining consensus agreement from the various workers' compensation jurisdictions as to what common data elements could be provided and the type and extent of coding which could be applied to them. The outcome might be described as a `lowest common denominator' data set. For example, one of the most frequently commented upon features of the data set is that its scope is limited to cases resulting in a fatality, permanent disability or five days or more time lost from work. The five-day threshold for inclusion was reached as a compromise based on what jurisdictions could provide as a group (see above regarding threshold differences and note that, until the 1993-94 reference period, Victoria was able to provide information relating to cases involving 6 days or more lost time). Likewise, some potential data items were excluded from the data set because one or more jurisdictions considered that they could not provide them, for example, breakdown mechanism, agency of injury/disease, type of shift, etc.

Compliance with the NDS and provision of data has generally been reasonable, with two exceptions. As yet, the Australian Capital Territory has not been able to provide any data at all, and Victoria has not yet been able to provide data coded in accordance with ASCO and the standardised type of occurrence classification. In addition, Victoria has not been able to provide data in accordance with the agreed threshold cut-off. However, it is expected that full compliance will be achieved over the next two to three years.

With the introduction of the NDS, Australia now has available to it a workers' compensation database covering the more serious compensated work-related injury and disease cases. With the exceptions mentioned above, the data have been consistently coded to agreed standard classifications with respect to industry of employer, occupation of employee, and the type and circumstances of the injury/disease (that is, nature, bodily location, mechanism and breakdown agency of injury/disease). Information is also available on the age, gender, employment status (full or part-time) of the injured worker, the day and time of injury, and costs and working days lost in respect of the new cases reported each year.

The database is quite large, containing information on approximately 170,000 cases for 1991-92, 160,000 cases for 1992-93 and 160,000 cases for 1993-94 - nearly half a million unit records in total (with the inclusion of 1994-95 data expected to occur by the end of June 1996, this will rise to approximately 650,000 unit records). Moreover, as the data are essentially an administrative by-product, the database has been established at relatively low cost.

The Objectives of Occupational Health & Safety Surveillance

Broadly speaking, the availability of reliable OHS data can be thought of as serving six vital objectives (in no particular order of priority):

  1. Hard data on the prevalence of occupational injury/disease highlights its importance relative to other public health problems and thereby establishes its claim for resources and legislative support.
  2. Analysis of objective data provides a basis for determining research and control priorities.
  3. It supports action strategies, through data-driven targeting of problem industries and occupations, to minimise the impact of risk factors associated with occupational injury and disease. (Concomitant with data-driven targeting is the opportunity to identify better performers and, therefore, best practice approaches.)
  4. Following implementation of intervention strategies, analysis of data allows evaluation of the effectiveness of those interventions and monitoring of overall progress with OHS performance improvements.
  5. Organised, pertinent data can be disseminated to all stakeholders, including workers, employers, unions, government agencies and the general public, to raise awareness of the problem and educe the general cooperation and commitment essential for effective action to diminish the problems of poor OHS performance.
  6. To assist in the early identification of new workplace hazards.

In this publication, criticisms of the usefulness of workers' compensation data will be evaluated against the background of the above objectives.

Some Criticisms of Workers' Compensation Data & Some Thoughts in Response

The most common criticisms of workers' compensation data as tools for OHS surveillance focus on understatement of the `real' level of work-related injury/disease, reporting bias, inadequate disease coverage, noncoverage of self-employed workers, lack of `near miss' information, inconsistencies in data reporting from year to year due to jurisdictional changes to compensation systems and their administration, and lack of timeliness in data release. In the following chapter, these criticisms are not considered in full. Rather, they are treated more in the manner of discussion points.

Understatement of the `True Extent' of the Problem

A frequent criticisms of workers' compensation data is that they greatly understate the `real' level of work-related injury/disease experienced by the Australian workforce and (implicitly) thereby understate the seriousness of the problem in Australia. The Industry Commission[3] in its inquiry into occupational health and safety considered workers' compensation data unsuitable as a `true measure' of the `level' of workplace injury and disease. Queensland's Department of Employment, Vocational Education, Training and Industrial Relations[4] concluded that workers' compensation statistics `... give seriously misleading information on the ... extent of workplace injury and disease'.

It appears that this type of criticism sometimes stems from misapprehension concerning the scope and coverage of workers' compensation data. Scope, in this context, refers, among other things, to the fact that not all workers can claim workers' compensation, for example, the self-employed are usually not entitled to lodge claims. Coverage relates to the proportion of workers eligible to make a claim who, in fact, do make a claim.

Scope

With regard to scope, workers' compensation data do not purport to represent all injury/disease occurrences irrespective of the degree of severity. Under the NDS, the scope of the statistical collection is restricted to the more serious compensated cases (involving a fatality, permanent disability or five days or more time lost from work). Therefore, the NDS only attempts to be a `true measure' of the extent of the more serious work-related injuries and diseases.

It should be noted that the pattern of injury/disease revealed for these more serious cases may not entirely reflect the experience in less serious cases. If so, this means that the data do not provide representative information on less severe cases which comprise a large proportion of the total number of cases. (To put the number of less severe cases into perspective, analysis of data available from a small number of compensation jurisdictions indicates that occurrences of less than five days duration, on average, represent approximately 46% of all new claims lodged annually, but account for only 8% of total working days lost).

However, it might be argued that despite the high number of less severe cases, if the objective of OHS interventions is to reduce the human and economic costs of poor OHS as quickly and cost-effectively as possible, a focus on severity is probably an advantage in prioritising action strategies. In this respect, it should be kept in mind that budgets available (from all sources) to enable improvements in OHS performance are limited. Therefore, to maximise reduction in human suffering and economic costs, expenditures on OHS interventions must be made in such a way as to equalise, at the highest possible level, the marginal benefit achieved per dollar outlaid across all investments in OHS improvements.

While it cannot be measured objectively, it is probably fair to say that, in general, an injury/disease occurrence causing five days lost time from work involves considerably more human suffering than one which involves only one day. Similarly, the average dollar cost per occurrence rises significantly with the duration of the injury/disease[5]. Consequently, on the assumption that more dollars would be required to achieve a quantum improvement in cases with a low average cost than would be required to achieve a commensurate improvement in those with a high average cost, there is more scope to maximise the marginal benefit per dollar expended by focussing on the more serious cases (as NDS-based compensation data do).

The second feature of the collection's scope, which is often portrayed as a weakness, is that the statistics only relate to cases which have been compensated. Taken from a different perspective, this might actually be considered a strength of the statistics, in so far as the validation processes associated with compensation systems help ensure the objectivity of the data. (Most other potential sources of national data are vulnerable to the subjective opinions of data providers in recording cases and in coding industry, occupation, type of occurrence, etc.)

Moreover, in measuring the `true extent' of the problem, rates of occurrence per number of workers (incidence rates) or hours worked (frequency rates) are the most meaningful measures. The denominator data used in conjunction with NDS data to calculate these rates exclude virtually all workers not covered by compensation (such as the self-employed). For these rates to be invalid measures, it would have to be conclusively proven that there were significant differences between the injury/disease experience of workers covered by compensation provisions vis a vis those who are not.

Several studies have been undertaken which imply that, for example, self-employed workers experience injury/disease rates substantially higher than employees in the same industry. The weakness in many of these studies is that they rely on self-reported injury/disease information, frequently obtained by approaching the self-employed worker. This raises two fundamental questions:

  1. How many of these self-reported occurrences would survive the scrutiny of the normal compensation process and be verified as work-related?; and
  2. In cases where the self-employed worker is actively approached for data, do the apparently higher number of occurrences obtained simply reflect the greater collection effectiveness of this methodology? (To clarify, it is known that employees will not bother to report less severe injuries for compensation purposes[6] but, if they were approached for this information, it is likely that they would report higher rates of injury/disease, probably fully commensurate with those reported by the self-employed.)

As yet, there is little convincing evidence of significant differences in injury/disease rates and risk factors for those covered by workers' compensation compared to those who are not. However, it is possible that longer hours worked and, therefore, longer exposure to risk factors, might give the impression of higher injury/disease rates for the latter group (in other words, the incidence rate, that is, occurrences per number of workers, might be high for the self-employed, but the frequency rate, that is, occurrences per number of hours worked, might be much the same as for employees).

Coverage

Concerns regarding coverage centre on the degree to which workers are aware of their rights under workers' compensation and on what Hopkins refers to as the `... twin phenomena of claims suppression and claims contestation'[7]. Hopkins describes claims suppression in terms of employers discouraging workers from submitting claims, lest it affect their premium costs, and employees' reluctance to make claims that might affect their standing in the eyes of their co-workers and/or their employment prospects. Claims unnecessarily contested by insurers are a concern because they are not included in workers' compensation statistics.

Coverage does represent an aspect of workers' compensation data about which there is not a great deal of hard data. Nevertheless, an ABS survey[6] suggests that undercoverage of occurrences due to lack of awareness is not necessarily a significant problem. It found that 82% of employed persons were already aware of workers' compensation rights. It might be expected that if, and when, any of the residual 18% of workers experienced a relatively serious injury, most would take positive steps to ascertain their rights, or be given the necessary information by their work colleagues or employers.

The effect of claims suppression and claims contestation on coverage is harder to quantify and cannot be lightly dismissed. The ABS survey mentioned above found that only 47% of injured workers (subjectively defined) in New South Wales applied for workers' compensation. Taking into account the threshold differences mentioned earlier (New South Wales, first $500 of wages, Queensland, more than one working day lost), this is not inconsistent with James'[8] finding that Queensland workers' compensation data omitted 27% of injury/disease cases.

The above findings are, of course, not an ideal basis upon which to gain an insight into the undercoverage of workers' compensation data in terms of the stated scope of the NDS. Considering this undercoverage in terms of NDS scope, the question arises, would proportionately more of these non-reported events relate to cases involving less than five days lost time from work and, therefore, be outside the scope of the NDS collection? Intuitively, because it would be expected that workers are less likely to claim in respect of less serious occurrences, the extent of any undercoverage in the NDS collection would probably be the relatively smaller proportion. Nevertheless, on balance, it must be considered that there is some degree, possibly significant, of coverage deficiency in workers' compensation data. Furthermore, because compensation claims for fatalities are not always lodged for a variety of reasons including, for example, situations where there are no heirs to claim on behalf of the deceased worker, it is likely that coverage deficiencies for cases resulting in a fatality would be greater than for other types of occurrences.

Importance of Any Deficiencies Due to `Understatement'

Clearly, there are scope and coverage considerations which mean that the number of occurrences captured in NDS-based workers' compensation data do not represent the total number of work-related injury and illness cases experienced annually. This begs the question, what is the practical effect of this fact in light of the broad objectives of OHS surveillance outlined above?

The objective listed earlier as 1 deals with the need for the dimensions of the OHS problem to be known for the purposes of funding and legislative support. NDS-based workers' compensation data only deal with the more serious cases and, as alluded to above, have at least some degree of uncertainty regarding coverage. Even so, enough data are available from this source, and others, to unequivocally establish occupational injury and illness as a substantial public health problem in Australia worthy of high priority funding and urgent action.

The available evidence reveals the following:

  • It has been estimated that there are at least 2,900 deaths related to work annually[2] - one and a quarter times the annual suicide number and nearly one and a half times the number of annual motor vehicle traffic accident deaths[9].
  • One in 12, or over 650,000[3], workers suffer a work-related injury or illness each year, approximately 170,000[2] of whom require at least five days absence from work as a result.
  • In August 1994 Worksafe Australia's estimated the annual cost of work-related injury and disease as being between $15 billion and $37 billion, in 1992-93 dollar terms. Assuming a 6% earnings potential, these costs become between $196 billion and $490 billion over 10 years, in 1992-93 dollar terms.
  • The Industry Commission[3] subsequently estimated the yearly cost at $20 billion, with approximately 30% being borne by employers, 30% by employees and 40% by government.

These objective data outline the case for treating OHS as a priority issue and substantiate its claims for resources. They come from workers' compensation data and other sources.

In summary, as far as Objective 1 is concerned, workers' compensation data do not, on their own, show the full dimensions of the problem, in terms of total numbers and costs, for any given year. However, it is this author's contention that they clearly can and do show the full extent of the problem in terms of rates of occurrence for the more serious cases. It is also possible that compensation data can be a reliable proxy for movement in the `level' of all occurrences from year to year.

It should also be noted that the broad picture types of data, which show the full extent of the problem in terms of total numbers and costs, are not the types of data which would require frequent revision to support claims for funding and priority. They are highly unlikely to change with sufficient significance or rapidity to justify more than, say, triennial collection. Consequently, it is probably quite appropriate that data to fully meet the needs of Objective 1 are collected through a number of other means rather than the highly detailed workers' compensation data collections.

The way in which workers' compensation data meet the requirements of Objectives 2 through 5 will depend, to a certain extent, on whether non-randomness in reporting/nonreporting exists, and the degree of consequent bias. It is contended that workers' compensation data, as reliable measures of the more serious injuries and quick onset illnesses, do have a significant role to play in all of these objectives, and their utility in doing so is not seriously affected by NDS scope and coverage limitations. This is particularly evident when it is considered that NDS-based data relate to over half a million unit records, consistently coded in accordance with NDS recommendations—in effect, half a million `mini case studies'. These issues will be further addressed under criticisms considered below. (It is recognised that workers' compensation data will not meet the requirements of Objective 6, except, perhaps, in indirect fashion. Some thoughts on this are raised under the heading Lack of `Near Miss' Data.)

Bias in Reporting/Nonreporting

A common criticism of the usefulness of workers' compensation data in OHS surveillance is that they understate the problem due to under-reporting by those entitled to claim compensation. For many purposes, under-reporting does not necessarily represent a problem, as long as it is consistent across the whole spectrum of occurrences. If this is the case, the researcher will still know, for example, the highest risk industries, occupations and age groups; the most frequently occurring nature of injury; the bodily location most at risk; and the injury mechanism and breakdown agency most often involved in occurrences. Thus, as long as relativities revealed by the data are reliable reflections of the `real' situation, research and control priorities can still be appropriately set, data-driven targeting is still valid, etc.

However, if under-reporting is not the result of some randomly distributed tendency, workers' compensation data might provide a seriously distorted picture of the nature of work-related injury and disease and those most affected by it. In these circumstances, it would be risky to take some of the decisions currently predicated on workers' compensation data. The degree of risk would, of course, be dependent upon how much was known about the way and extent to which under-reporting was skewed away from the normal.

Industry Under-reporting Bias

Much of the evidence supporting bias in industry under-reporting is anecdotal. It has been claimed that there is a concentration, in certain industry segments, of incentive payment schemes based on the length of time without a work-related injury being reported. It has also been suggested that casual workers in some industries, with relatively high proportions of casual employment and high staff turnover rates, are reluctant to exercise their rights to claim compensation for fear of losing employment (and there is evidence which seems to support these suggestions by showing apparent under-representation of casual workers in claims data, given their relative exposure to risk factors and the experience of full-time workers as reported in compensation data for that industry, for example, Restaurants, Hotels and Clubs[10] and Cleaning Services[11]). In other cases, when `contracting-out' work, some industry managers appear to have taken the opportunity to attempt to avoid compensation responsibilities/claims, for example, pieceworkers in the Textiles, Clothing and Footwear industry[12,13] who may actually be covered by workers' compensation but are unaware of the fact, or are unwilling to submit claims.

If the above practices are widespread within the industries concerned, they militate towards under-reporting, biased according to industry. This has implications primarily for the usefulness of interindustry OHS performance comparisons (data-driven targeting) based on compensation data, unless allowances are made for the existence of these practices.

The industries concerned can, to a certain extent, be identified. In the case of incentive payments schemes, it is possible analysts would have to rely on industry knowledge. In instances where casual workers are loathe, for various reasons, to lodge compensation claims, statistical significance tests on compensation data can be used to identify industries where relatively small proportions of occurrences involving part-time workers are unlikely to be the result of chance.

Awareness of the existence of under-reporting bias for a particular industry provides analysts with the opportunity to make allowances for it in setting priorities and in targeting. Where incentive payment schemes are perceived to be a problem, in a worst case scenario, in the absence of better information, the analyst would be obliged to factor-up on the basis of an educated guess. However, the NDS focuses on the more serious cases (five days or more lost time) and it is likely that there will be a practical limit to the effect of incentive schemes—a point at which the trade-off between the incentive payment and the compensation payment for working days lost will no longer be attractive. Therefore, despite (perhaps apocryphal) claims regarding the synergistic effect of some rehabilitation schemes on this practice, it seems unlikely, given the trade-off referred to above, that it could have a highly significant impact on compensation data employing the NDS threshold (in any case, it appears that there are only a small number of industries where this practice has any prevalence and compensation data already show them as having significantly higher injury/disease rates than other industries). In the case of under-reporting bias for an industry as a result of low part-time employee claims, analysts have the experience of full-time workers upon which to base adjustments (again, there are considered to be only a small number of industries where this type of under-reporting represents a problem).

Occupation Under-reporting Bias

Occupation under-reporting bias presents a different set of problems. There is not a great deal of hard evidence to support any assumption that this is a serious problem at this stage. James[8] finds, from a sample which she advises is small and not completely representative, that under-reporting in workers' compensation data for Queensland tends to be more concentrated among lesser skilled (`blue collar'), occupationally mobile and geographically isolated workers. It is also suggested that the propensity to under-report is exacerbated in an adverse economic environment. (Interestingly, the ABS study on work-related injuries and illnesses[6] tends to show that the more skilled occupations lodge less applications for compensation, but the study is not clear on relative entitlements to claim.)

Intuitively, James' basic findings are appealing and consistent with anecdotal evidence, particularly in regard to the development of working cultures which influence workers not to lodge claims. However, the sample size (309 interviewees) is too small to reach a firm conclusion on national occupational under-reporting bias, let alone attempt to quantify the extent of it. (It should also be noted that this study is vulnerable to self-reporting problems mentioned earlier.)

Perhaps it is pertinent to note that, even though under-reporting bias might possibly be lowering the number of cases reported for `blue collar' workers, these workers still accounted for approximately two-thirds of all NDS-based compensation cases reported in 1992-93, despite making-up only one-third of total employment[2]. It would seem that occupational under-reporting bias, if it exists, is by no means sufficiently widespread to prevent compensation data from conveying the right signals in regard to OHS targeting and prioritisation.

Other Under-reporting Bias

It has been suggested that under-reporting bias is also associated with specific natures of injury/disease and injuries/diseases suffered by NESB workers.

It is probable that some proportion of workers suffering certain types of diseases might have a tendency not to lodge compensation claims, for example, those suffering mental stress. It is also possible that certain types of injuries, for example, those resulting from workplace violence, might not always be reported. However, there is no conclusive, quantitative evidence to support the existence of significant under-reporting connected with specific injuries/diseases. Much of the opinion on this matter comes down to conjecture and it would be a useful contribution to the debate if sound data on reporting propensities in this regard could be collected.

The ABS survey on work-related injuries and illnesses produced some data on the extent of NESB under-reporting (as cited by the Industry Commission[3]). This showed that under-reporting for NESB workers was 5% higher than for English speaking background workers. This is by no means a significant difference in terms of overall bias. Further, a superficial examination of general relative standard errors associated with the survey suggests that this difference might simply result from random chance associated with sample selection procedures.

Thus, there is no convincing evidence that significant under-reporting bias in compensation data results from the above sources.

Coverage of Diseases is Poor

Firstly, it should be recognised that anyattempt to collect national data on work-related disease will be confronted with a number of serious difficulties. As Driscoll[14] points out `... diseases are often characterised by uncertain aetiology, potential non-occupational causes that are difficult to exclude, long latency periods, and/or uncertain diagnostic criteria'. In practice, the crux of the problem lies in determining whether the disease is actually `work-related'.

The onset of most diseases stem from exposures which can occur in the normal living environment, as distinct from a genesis which is unequivocally connected solely to the occupational environment. Frequently, it can be argued that a disease results from a combination of an individual's susceptibility, work exposures and non-work exposures. To illustrate, consider a worker who develops cancer and claims it is the result of passive smoking in the workplace. Assuming he/she is the only worker in the workplace to lodge such a claim and that he/she would be exposed to some degree of passive smoking outside the workplace, the following questions arise:

  1. Would he/she have developed the cancer in any case, irrespective of the passive smoking?; and
  2. To what extent would non-workplace exposures have contributed to development of the cancer?

To answer these types of questions, it is impossible to avoid dealing with probabilities and, consequently, very difficult to produce a definitive response.

To further compound the problem, there is often a lengthy latency period between the exposure, the emergence of symptoms and the diagnosis of the disease. This tends to further obscure the nexus between exposure and disease, for example, there may have been several different exposures which could have potentially caused the disease. The longer the latency period, the less likely it is that there will be a trail of convincing evidence connecting the exposure to the resultant disease.

All of the above factors, in addition to others not mentioned, place severe constraints on options for collecting data on work-related disease. Certainly, self-reported data collections on work-related disease can produce results, but these are frequently complicated to interpret and of highly uncertain reliability (although the latter point is not necessarily true of all diseases reported upon in this manner, for example, mesothelioma, pneumoconiosis, vibration white finger and deafness can be self-reported as work-related with a fairly high degree of accuracy—Hodgson et al[15]). In addition, epidemiologists have developed some probabilistic techniques which can produce reliable estimates of the level of disease in the community which is work-related (not detailed here), registers of certain disease sufferers have been created, etc. Nevertheless, it is clear that it is virtually impossible for any one data source to give a comprehensive picture of the situation with work-related disease, and the full picture can only be pieced together from a number of different data sources. So, the issue distils to what information does workers' compensation data provide on the level of occupational disease, and how can it help in meeting OHS surveillance objectives?

Workers' compensation data relate to successful claims. The success of claims in Australian compensation systems, which were initially designed to handle occupational injuries, depends on the clarity of the connection between the disease and its cause. In instances where exposure is ill-defined or its consequences poorly understood, it is unlikely claims will be successful. Essentially, this creates a pattern of compensation claim success heavily tilted towards those diseases which have a well recognised cause-effect relationship to exposures at the workplace and/or are basically of quick onset.

Consider, for example, chemical exposures. Chemical exposures can result in relatively minor health effects, for example, dermatitis or folliculitis, serious acute toxicity or chronic systemic poisoning. In simplistic terms, it is most likely that a minor health effect such as dermatitis would result in a successful claim because of long established recognition of its connections with chemicals at certain workplaces, for example, cutting fluids. Diseases resulting from serious acute toxicity, especially in the case of single exposure events, will also most likely produce successful claims because of the clear connection between the occurrence and the outcome and the short time lag between exposure and symptoms. It is relatively less likely that diseases manifested as a result of chronic systemic poisoning will result in a successful claim. The symptoms of these diseases frequently require some type of medical screening for detection and, as a result, might be diagnosed a relatively long time after the exposure.

Deafness is another illustration of variability in workers' compensation coverage of disease. The cause-effect relationship between the workplace and deafness was once strong enough for some to argue that nearly all deafness cases (except congenital and old-age) were related to occupation[16]. Despite increasing recognition of a number of non-workplace causes of noise-induced deafness, the connection is still considered extremely strong. (Note though that legislation in some jurisdictions weakens recognition of this connection, to some extent, through measures such as minimum hearing loss thresholds and specifically nominated starting dates for coverage of deafness claims, effectively precluding pre-existing conditions.) A weakness in workers' compensation data on deafness stems from the insidious nature of occupational hearing loss. Most workers do not know when they are suffering from it until testing reveals significant hearing impairment. So, while claims tend to be successful because of the strength of the cause-effect nexus, there appears to be some fluctuation in the number of occurrences reported due to factors unrelated to the prevalence of the disease. These influences include random pressure for the conduct of hearing tests across industries and a type of `bandwagon' effect when successful claims are made in an industry/workplace. This could lead to incorrect trend interpretation, unless, for example, occurrences are averaged (smoothed) over a reasonable period of time.

Thus it can be seen that work-related diseases are patchily covered by workers' compensation data. As illustrated above, generalisations about the reliability of this coverage on the basis of nature of disease are not dependable. However, coverage is undoubtedly more reliable for quick onset disease occurrences and diseases where there is a strong and well-recognised cause and effect relationship linking them to the occupational environment (with certain, specific exceptions, for example, mesothelioma which can frequently have a very long latency period).

The advantage of having workers' compensation data on diseases lies in having a measure of the number of cases of verified, work-related disease which are of quick onset and/or readily identifiable as work-related. The remainder of reported disease cases represent, in most cases, at least the minimum level of occurrences diagnosed and confirmed as work-related in a period. Treated with caution, these data can be used as a guide (particularly if complemented by the knowledge and experience of OHS practitioners) in making decisions regarding the need for more in-depth study of particular diseases. In view of the expense and difficulty frequently involved in collecting this type of information, this is a worthwhile aid in deciding data collection and/or research priorities.

It should be noted that for effective prioritisation of research and control and data-driven targeting of OHS problem areas, allowances have to be made for the difference in workers' compensation coverage of injuries compared to their coverage of diseases.

Exclusion of Self-employed Workers

At June 1993 approximately 16% of workers in Australia were self-employed and 1% were unpaid family helpers (ABS[17]). A frequently expressed concern in relation to workers' compensation data is that, given these data almost totally exclude this segment of the working population, decisions based on them may be totally inappropriate to the OHS problems and needs of self-employed workers.

Some reservations regarding the validity of these concerns were raised under the earlier heading Understatement of the `True Extent' of the Problem. To reiterate, the most meaningful measures of OHS performance ate rates of occurrences per number of workers or hours worked. Both the denominator data and the NDS numerator data used in the calculation of these rates exclude self-employed workers. For these rates to be non-representative of the self-employed, the OHS experience of the 83% of workers covered by workers' compensation would have to be significantly different from the 16% of workers who are self-employed. Given that they are usually confronting the same risk factors, this would be surprising.

Assessment of the difference in injury/disease patterns between the self-employed compared to employees is hindered by lack of a substantial body of reliable data and the difficulties involved in collecting such data. Studies undertaken to clearly establish whether or not significant differences in experience do exist will have to confront the problem of how to collect information on injury/disease occurrences from the self-employed in a way which mimics the same conditions of verifying work-relatedness that apply to processing a successful workers' compensation claim. Further, given that employees frequently do not bother to report less severe injuries for compensation[1], a technique would have to be developed to correctly `discount' the number of occurrences reported by the self-employed to reflect this propensity on the part of employees not to report.

To date, there is little convincing evidence of a substantial disparity between the work-related injury/disease experience of the self-employed vis a vis employees (although incidence rates—not frequency rates—might be higher for the self-employed if longer hours are worked). Further, the question should be asked, how substantial would such a difference have to be for it to make decisions regarding OHS interventions, based on workers' compensation data, entirely irrelevant for the self-employed?

Lack of `Near Miss' Data

The thrust of another criticism levelled at workers' compensation data is that they do, for example, include a case where someone has slipped on a wet set of stairs, injured his foot and been absent from work for five days, but they do not include a case where a crane cable has snapped, dropping a cargo container which narrowly misses killing six workers (a `near miss'). The key issue here is that the objectives of a system which would count `near misses' are different from those of a statistical data collection such as the NDS-based workers' compensation system.

The former should be thought of as an early hazard identification system (see Objective 6 in the chapter The Objectives of Occupational Health and Safety Surveillance) aimed at speedy information collection and dissemination to prevent injury/disease before it has a chance to occur. Its focus would be on new types of hazards, specific equipment malfunctions, problems with specific work conditions, etc, and it would probably be most effective operating at the workplace or industry level. The latter is a system aimed at meeting the requirements of Objectives 2 through 5, as outlined above. If an early hazard identification system was tied to this latter system, with its greater degree of detailed data content and its sheer size and time-consuming analysis requirements, most of the potential hazards about which the system was attempting to provide early warning would already have become serious incidents before they could be notified.

The concept of collecting `near miss' data as some type of measurement of OHS performance is intrinsically less than persuasive. The definition of what would constitute a `near miss', constructed so that it could be consistently applied and capable of producing meaningful statistics, would be a problematic task in itself. If this central problem of a useful definition could be overcome, in what way could `near miss' data provide a better measure of national/industry OHS performance?

The circumstances (risks) surrounding an actual occurrence, as recorded in compensation statistics, in all probability, would be basically the same as those which result in a `near miss'. In most cases, whether a `near miss' or an injury/disease results is largely a matter of chance. In other words, if near misses were included in performance measurements, the numbers would be larger, in absolute terms, but the underlying pattern of work-related injury and disease would be much the same. Further, in general management practice, it is considered that performance measurement indicators which are outcome related are more productive in facilitating achievement of goals than indicators which are vague or predicated on superficial features of program activity, for example, process measures. It is likely that the same holds for OHS interventions.

It seems that `near miss' data systems (or early hazard identification systems) form one, albeit crucial, aspect of OHS surveillance. They represent that part of surveillance which allows identification of potential, very specific, work-related injury/disease problems at a level where they can be acted upon and occurrences can, consequently, be either avoided entirely or minimised. They are often systems based on the concept of information and solution sharing through networking.

Early hazard warning systems have no real role in broad level prioritising of research and control objectives, nor in data-driven targeting of OHS problem areas, nor in the monitoring of general progress with OHS performance improvements. They are of the nature of trouble-shooting systems, which must be highly responsive to hazardous situations as they emerge, and it would be counter-productive if they were entangled with policy level data systems such as workers' compensation data.

Data Inconsistency Over Time

As mentioned earlier in this publication, the first set of NDS-based workers' compensation data (relating to 1991-92) was made available to Worksafe Australia in late 1993. In the initial stages of data provision, there were some problems relating to coding quality (and there still are with respect to some data items received from Victoria). In addition, there have been problems with variation in threshold. Initially, Victoria could only provide data relating to occurrences resulting in six days or more lost time from work, and for 1993-94 have only been able to provide data with a lost-time threshold of more than ten days. Other administrative actions will also have some effect on data consistency from time to time, for example, changes to rehabilitation systems, periodic case reviews, etc.

Of these issues, the threshold variations represent a fairly significant problem. However, at this stage it is only considered to be a short term concern. Currently, all of the jurisdictions are expected to be able to provide data with a threshold of five days lost time from the 1996-97 reference period onwards. In the interim, the comparability gap over 1993-94 to 1994-95 might, to some extent, be bridged by linking the trends of all jurisdictions using a threshold of more than ten days to trends displayed by all jurisdictions excluding Victoria.

It is considered by those with access to NDS source data that inconsistency in fatalities data over time will continue to be evident as jurisdictions change policies with regard to claims contestation and which types of cause-effect relationships are readily recognised as valid in fatalities claims. All other potential inconsistencies over time will have only a relatively marginal effect on the validity of time series comparisons (it is expected that the major effect will impact on cases below the five day threshold, which are outside the NDS scope, or upon cases of very long duration, which are relatively few in number).

Untimely Data Release

A consistent criticism of NDS-based workers' compensation data is that there is considerable delay in production of data following the reference period concerned. The observation made is that untimely data detracts from its utility, at least to some extent. This is a valid point only in regard to some data uses, although not, in practice, as far as the major applications of the data are concerned.

Teething problems experienced during introduction of the NDS undoubtedly contributed to delays in data release. The greatest potential for improvement in the timeliness of data release is now in the hands of workers' compensation jurisdictions which provide the data to Worksafe Australia. The earlier that they can provide data, the earlier that national data can be produced and released.

Advantages of Workers' Compensation Data in Comparison to `Alternative' Data Sources

With regard to workers' compensation data, Macaskill et al[18] find that `Although this data set suffers from a number of limitations, it does provide one of the best available data sets on: occupational injury; factors associated with the injury; details of the occupation and industry of the injured person; and the cost of compensation.' They go on to say `Rather than replace the NDS with something which is certain to be flawed in other respects, the most useful strategy would be to identify opportunities for providing a context in which to interpret these data and to integrate the information with that obtained from other sources.'

The foregoing discussion of common criticisms of compensation data tends to show that many of the perceptions of weaknesses in compensation data are significantly overstated. Indeed, many of the ways in which aspects of compensation data are portrayed as limitations suggest that the problem with compensation data lies more with a need to ensure users are fully advised regarding its source, scope, coverage and utility, rather than with any intrinsic deficiencies in the data themselves. To illustrate, the Industry Commission expressed doubt regarding the cost-effectiveness of maintaining the NDS because of its `deficiencies', then, curiously, went on to suggest `Equally meaningful national estimates could be made by aggregating workers' compensation agency data ...' [3](which is precisely the substance of NDS-based data).

In comparison to alternative (potential) data sources, the strengths of the compensation data, as they are currently available through the NDS, are that they represent around half a million unit records on the more serious, verified, work-related injuries/diseases, all (with the exception of occupation and type of occurrence in Victoria) consistently coded for industry, occupation and type of occurrence, and including information on age, gender, employment status, costs, day and time of injury and working days lost. As such, they are capable of being used, to varying degrees, to meet the requirements of all OHS surveillance objectives (with the exception of those associated with `near miss' situations). Furthermore, as administrative by-product data, they are quite low cost, particularly in comparison with the amount of detail obtained.

Perhaps one of the implicit strengths of workers' compensation data is that there are few extant, cost-effective, alternative data sources. For example, surveys are dubious alternatives due to unavailability of adequate sample frames, the excessively high sample numbers required to produce necessary cross-classificatory information, their frequent reliance on self-reporting of injury/disease and the burden that they impose on informants.

There is never likely to be one data source which, on its own, fully satisfies all the requirements of OHS surveillance. There will always be the need for information from a variety of other sources to complement and flesh-out compensation data. In the Australian context, it would seem that NDS-based workers' compensation data are a sound starting point for OHS data research, and there are a number of other data sources which could be put in place to build on that base (rather than replace it). Macaskill et al[18] provide a comprehensive assessment of options in this respect.

How Compensation Data can be Used

Hopkins[7] notes that `... workers' compensation data ... have led to the realisation that relatively mundane back sprains are far more widespread than the more dramatic and traumatic `blood on the floor' injuries.' He sees this as a `... progressive development from the point of view of reducing the total amount of pain and suffering which workers experience.'

While Hopkins' observations in this respect are clearly correct, taken alone they run the risk of trivialising the usefulness of workers' compensation data in the development of effective OHS interventions. These data can be used to fine-tune data-driven targeting of OHS problems a fair deal better than Hopkins' observation might be construed as suggesting. They are also capable of playing a significant role in research and control prioritisation, and their analysis provides an ideal means of raising awareness of OHS problems in a manner which should help to get a more positive and effective response from all those involved.

A number of analyses of the OHS performance of various industries have been undertaken using workers' compensation data, and the results disseminated, primarily through Worksafe Australia's publication series Occupational Health and Safety Performance Overviews, Selected Industries. This series is not meant to provide definitive statements on the OHS performance of the industries concerned. Rather, it is intended to provide an indication of what appear to be potential problem areas for those industries and, thereby, initiate further discussion, investigation and research.

Set out in Appendix 1, on an industry by industry basis, are some findings and other salient points from a selection of these analyses. A table at Appendix 2 provides some summary information from the analyses. Basically, they illustrate the way in which compensation data can be used as a starting point for the development of effective OHS interventions.

Compensation data are the broad level picture which highlight aspects of performance where it might be productive to initially invest resources. As progressively more is found out about the circumstances surrounding injuries/diseases during the undertaking and completion of this `top-down' process, specific, effective interventions should be developed.

Thus, as can be seen from the extracts provided in Appendix 1, compensation data can identify problem areas for an industry and/or occupation. Ideally, the subsequent course of events should commence with more in-depth research to focus on those problem areas and isolate important causal factors. This should be followed by development of specific preventive measures, for example, modification of certain types of plant/equipment, wider use (or development) of relevant protective clothing, improvement of workforce skills in certain areas of activity, etc. Clearly, without the starting point and framework provided by compensation data, the reduction of the OHS burden on the Australian community would largely become a process of disjointed incrementalism.

It will be noted that many of the points raised in the extracts might have been expected intuitively, or might be expected to have been known by persons with good industry knowledge. It is important to realise that by confirming some of these points by analysis of objective data, it prevents them from being dismissed as dubious (based on `folklore'), and more precisely establishes their importance in the order of OHS problems for an industry/occupation. In some cases, of course, the conventional wisdom regarding OHS performance in an industry is not supported by the findings of an objective analysis.

The selected extracts at Appendix 1 are provided in point form, without any of the supporting details or graphs that appeared in the original publications. The discussion on criticism of workers' compensation data earlier in this publication should put these extracts into better perspective. It is suggested that readers who would like more information on any of the industries analysed should refer to the relevant issue in the series (see the chapter References).

Conclusions

Workers' compensation data are one of many potential sources of information on the nature and extent of work-related injury and disease in Australia. The data are a good source of information on the more serious injuries, and a less reliable indicator of the extent of disease. With respect to disease, they are inclined to best cover those diseases which have a well recognised cause-effect relationship to exposures at the workplace and/or are basically of quick onset. With regard to other diseases, they can generally be expected to reveal problems in a manner only suited to `tip of the iceberg' assessments. However, perhaps surprisingly, their coverage of deafness, a long onset disease, can be useful in targeting some industries/occupations as having a particular problem.

A number of criticism of NDS-based data have been addressed in this publication to ascertain whether there is a compelling argument against the utility of workers' compensation-based data in the development of OHS interventions. These criticisms are not really considered fully. Rather, they are treated more in the manner of discussion points. These points are set against the background of what are seen as the major objectives of OHS data.

In summary, in terms of the six vital objectives of OHS surveillance recognised in this publication, workers' compensation data would generally appear to be:

  1. Useful in providing a general picture of the level of occupational injury/disease, although when more comprehensive data are required, it is appropriate that they be supplemented by information from other sources. It is considered that, for most practical purposes, these more comprehensive data would not be required annually.
  2. Quite useful as a basis for determining research and control priorities, although users have to remain mindful of the need for `tip of the iceberg' assessments as far as disease data are concerned.
  3. Quite useful in the data-driven targeting of problem industries/occupations, although the same provisos exist for disease as mentioned above.
  4. Able to be used effectively in the monitoring of progress in OHS performance.
  5. Very useful in raising awareness of the problem, and, thereby, eliciting cooperation and commitment from those decision-takers who can play a role in improving OHS performance.
  6. Not useful in workplace-based early hazard identification systems.

In respect to the specific criticism considered, the following points are made:

  • Workers' compensation data do not fully represent the OHS situation in Australia in terms of total numbers and costs, and to this extent they could be said to understate the dimension of the problem. However, it can be argued that they do show the full extent and severity of the problem by expressing it in terms of rates of occurrence (per number of workers and hours worked) for the more serious cases.
  • The threshold cut-off adopted for the NDS probably means that, in those few industries where incentive payment schemes exist to discourage reporting, under-reporting bias is probably relatively small. Other industries affected by the reluctance of part-time workers to lodge claims can be identified by statistical significance testing. Subsequently, notional adjustments might be possible, based on rates reported by full-time workers, to counteract under-reporting bias. There is no convincing evidence, as yet, to indicate that significant occupational reporting bias, or other types of reporting bias, exist in workers' compensation data.
  • Coverage of diseases is only acceptably reliable in respect of cases where there is a strong and well-recognised cause and effect relationship linking them to the occupational environment and/or where cases are of quick onset disease. (An exception to this general observation is mesothelioma where, despite its well-recognised link to occupational exposure, less than one-tenth of cases appear in compensation data due to the sometimes extremely long latency periods involved.) Other disease coverage through compensation data only provides a `tip of the iceberg' type of assessment. The point is made though that any attempt to collect national data on work-related disease will face serious difficulties. The crux of the problem lies in determining whether disease is actually work-related. Consequently, other systems for collecting this type of information must also be flawed to some degree. There is, simply, no single means of collecting completely comprehensive and precise national data on work-related disease. Nevertheless, steps should and could be taken to adequately support compensation data with data from a range of other sources.
  • The fact that self-employed workers are excluded from workers' compensation data is not a problem unless the injury/disease experience of the self-employed is significantly different to employees. This is so because occurrence rates are calculated using both numerator and denominator data which exclude the self-employed. In practice, the disparity between the experience of the self-employed vis a vis employees would have to be very substantial to make OHS interventions based on workers' compensation data entirely irrelevant for the self-employed. As yet, there is no conclusive evidence that there is a difference of any great magnitude between the two groups and the difficulties in collecting such data are noted.
  • Workers' compensation data are not suitable as the basis for early hazard warning systems. These systems should be workplace and/or industry-based. In fact, it would be counterproductive if they were entangled with policy level data systems such as workers' compensation data.
  • Data inconsistency over time is essentially seen as only a short term problem.
  • Untimely data release is a valid problem to some extent. Perhaps the greatest potential for improvement in this regard rests predominantly with the workers' compensation jurisdictions that provide the data to Worksafe Australia.

It is concluded that there is unlikely to be one data source which will, on its own, fully satisfy all the requirements of OHS surveillance. NDS-based compensation data are a sound starting point for further OHS data-driven research. They represent the `top level' picture which highlights aspects of performance where it might be productive to invest resources. From this starting point, progressively more can be found out about the circumstances surrounding injuries/diseases and, at the completion of this `top-down' process, it is possible for specific, effective interventions to be developed.

Appendix 1

Extracts from Occupational Health and Safety Performance Overviews, Selected Industries

Fire Brigades Industry [19]

  • Noise —Deafness represents over 8% of compensable disease experienced by firefighters. Further, it is a gradual onset disease with a variable latency period and it might be expected that workers' compensation data could understate risk in this respect. Therefore, the data probably flags the area as one which requires further attention and additional `hard' data.
  • Falls and Work Environment —Over 40% of injuries are attributed to falls from heights and on the same level. Of these, nearly 45% had the breakdown agency as steps and stairways, suggesting there may be a potential problem with access, maintenance or positioning; further data would be useful in this respect. The fairly high number of injuries with the mechanism of muscular stress with no objects involved might also be partly attributable to confined conditions in the working environment.

(Feedback received on this report indicated that by far the majority of cases occurred as a result of training activity. It would seem that an appropriate preventive strategy would have to have a focus on safer training techniques, routines, etc.)

Forestry Logging & Log Sawmilling Industries[20]

  • The number of instances involving power tools in Logging suggests that further investigation of problems in this area might also be productive.
  • Plant —In both Logging and Log Sawmilling, being hit by moving objects, trapped between stationary and moving objects, trapped by machinery, and falls from the same level constitute significant proportions of the mechanisms of injury. Despite the fact that Logging, and to a relatively lesser extent Log Sawmilling, have natural risks which are somewhat random in nature, this suggests that plant and general site safety is an area of potential concern and worth further attention.
  • Manual Handling Practices —Muscular stress while lifting, carrying, putting down or handling objects causes more than a quarter of Log Sawmilling injuries and 13% of Logging injuries. A large proportion of the resultant injuries are lower back injuries indicating that manual handling practices are an area of potential concern.

Paper, Printing & Related Industries[19]

  • Across most ASIC classes there appears to be a relatively higher incidence of fingers/hands being trapped by moving printing machinery; lower back strains from lifting/carrying or handling crates/cartons of print output; falls from steps and stairways; and falls from the same level due to outdoor/indoor environment conditions.
  • Plant —Mechanism of injury data show that being trapped by moving machinery, hit by a moving object, and vehicle accidents account for 21% of injury occurrences. This suggests that plant is an area of potential concern.
  • Manual Handling Practices —Muscular stress while lifting, carrying or putting down objects plus muscular stress while handling objects accounts for over 27% of injuries, indicating that manual handling practices are also an area of concern.
  • Falls —Falls from heights and falls on the same level account for nearly 18% of occurrences. Further, in 10.7% of cases, various aspects of indoor environment (eg, steps, stairways, wet, oily surfaces) have been identified as the breakdown agency. This suggests further research to identify causes associated with falls would be productive.

Construction Industry[21]

  • There were 65 compensated fatalities in the Construction industry reported during 1992-93. This represents approximately one-eighth of all compensated work-related fatalities in Australia over this period. By way of contrast, the Construction industry represented only a little more than one-twentieth of total employment numbers.
  • Overall, the Construction industry was a significantly poorer performer than the average Australian industry with an incidence rate which was 90% higher than the All Industries rate. Also, there was a wide variation between performance at the more detailed level of industry classification. Road and Bridge Construction had a rate more than three and a half times the All Industries rate, while both Residential Building Construction, n.e.c., and Roof Tiling had rates around two and a half times the All Industries rate.
  • It is interesting to note that, while most injury cases occurred on Tuesday, the largest number of cases of longer duration occurred on Monday, followed by Wednesday. Looking at day of occurrence by time of occurrence, the working hour of the week in which employees were most likely to be injured was 9.00am to 10.00am on a Monday. It might be productive to undertake further investigations aimed at identifying any relationship between day, time of day, injury occurrences and the factors/risks involved in this industry.
  • Most falls on the same level were associated with the outdoor environment, specifically traffic and ground surfaces. In this respect it might be useful to analyse specific workplace data to see, for example, if there is scope for more effective preventive action through improving general site tidiness and/or removing traffic obstructions.
  • The objects by which most workers are struck, whether those objects are moving, falling or stationary, are bars, rods, pipes, rails, girders, roofing iron, tinplate, sheet metal, aluminium roofing and cladding. Among other things, this raises the question of whether there might be any better means of on-site storage and/or restraint for these types of items.
  • Plant —Machinery and Plant was involved in over 10% of occurrences. Nearly a third of these entailed self-propelled plant such as graders, dozers, excavators, backhoes, front-end loaders, road rollers, etc. Over a fifth were associated with semi-portable plant such as pneumatic tools, compressors, pumps, etc. Nearly a fifth resulted from the use of cutting, slicing, and sawing machinery such as powered saws, lathes and grinders. Perhaps there would be some benefit in further investigation of the use of plant in this industry, particularly in certain segments of it where plant played a greater part, proportionately, in the number of cases experienced, eg, the Electrical Work class, the Earthmoving and Dredging class and the Non-Building Construction group. In these industry sectors, falls from or on plant, muscular stress from relocating plant and long term exposure to plant noise appeared to play a significant part in injury/disease occurrences. A closer look at causal relationships in these selected industry sectors might prove productive in establishing more effective preventive practices.
  • Handtools, Appliances and Equipment —Powered handtools, appliances and equipment accounted for nearly 5% of cases while non-powered handtools, appliances and equipment accounted for a further 6%. The number of muscular stress cases involving this equipment raises the question of whether there is room for improving adherence to those manual lifting/handling techniques recognised as reducing risk in this regard.
  • Manual Handling Practices —Body stressing was associated with nearly a third of all cases and manual handling activities were involved in five-sixths of these cases. A frequent outcome of this type of body stressing was sprains and strains of joints and adjacent muscles (this was the nature of injury in over 40% of cases, accounting for half the compensable working days lost at an average 48 days lost per occurrence). Large numbers of these occurrences involved timber, rods, pipes, girders, roofing material, sheet metal cladding, etc. This suggests it might be worthwhile to attempt to establish whether handling practices involving heavy material currently employed within the industry are optimal.
  • Noise —Deafness represents over 11% of all injury/disease cases reported in this industry, despite the fact that the workers' compensation data upon which this analysis is based have coverage limitations in respect of this type of gradual onset disease. The majority of the reported cases were associated with long term sound exposure, as distinct from single, sudden sounds. Therefore, it would seem advisable to consider ways in which long term exposure to sound might be reduced in this industry, particularly in the Road and Bridge Construction class where 15% of cases reported were deafness and in the Non-Building Construction n.e.c. class where 29% of reported cases were deafness.
  • Work Environment —Falls, slips and trips, and being hit by, or hitting, objects accounted for a high proportion of injury/disease occurrences. The general work environment appears to have played a part in most of these instances. This tends to suggest that any measures which can be taken to improve general site tidiness and/or remove traffic obstructions, and provide better protection/prevention against falling/moving objects, might be effective in significantly reducing injury/disease risk. The significant number of cases involving ladders, mobile ramps, external and internal stairways and indoor steps suggests that further investigation of this type of injury occurrence might also be productive.

Mining Industry[10]

  • The injury/disease incidence rate for the Mining Industry as a whole is almost three times the All Industries rate. The Silver-Lead-Zinc Ores industry Class has a rate which is nearly six times the All Industries rate, while the largest employing industry in the sector, Black Coal, has a rate which is more than four times the All Industries rate.
  • The industry class with more than half the recorded injuries (and fatalities) in the Mining Industry is Black Coal. There is a discernible difference in the occupational injury pattern vis a vis the industry as a whole. The most striking difference in this respect is the far higher proportion of injuries affecting Excavating and Earthmoving Plant Operators. To some extent, this occupation group probably features more prominently due to a different employment profile within the industry class. Nevertheless, the data suggest that further investigation of causal factors in plant operator injuries might be productive and also flag plant operators as a potential target group for effective OHS action in this industry.
  • The major proportion of falls on the same level ie, slips, trips, stumbles, etc) were associated with conditions on the mine floor, eg, wet, oily or icy surfaces, hazardous object, etc. However, there also appears to be a significant number of occurrences resulting from the condition of external traffic and ground surfaces which are at a fair distance from mining activity. It seems that analysis of more detailed data on this type of occurrence might be productive.
  • For the most part, being hit by falling objects involved factors which would intuitively be expected to be the more hazardous aspects of mining, eg, falling rocks, stones and boulders, falling ore, roof, face and walls movement. However, there was also a slightly higher incidence of plant and machinery involvement than might be expected, eg, integrated mining plant and roof bolting machines. (With respect to roof bolting machines, they appeared as a significant factor in a fairly broad spectrum of incidents, eg, muscular stress while lifting or handling objects, being trapped by moving/falling objects, etc.)
  • Plant —Machinery, fixed plant, mobile plant and transport figure prominently in analyses of the breakdown agency of injury. It appears that there is a higher incidence of falls from mobile plant and transport equipment than might be expected. It also appears that use and relocation of plant (eg, integrated mining plant, conveyors, roof bolting machines) is involved in a significant number of injury occurrences. Consequently, Plant is an area of potential concern and worthy of further attention.
  • Manual Handling Practices —Apart from injuries associated with lifting, carrying or putting down plant and equipment, most muscular stress occurs in connection with use of non-powered equipment and the lifting/handling of materials (rocks, stones, boulders, timber and ore) which would normally be expected to pose a problem at mine sites. As muscular stress accounts for over 22% of injury mechanisms, it would appear productive to monitor problems more closely, perhaps with a focus on methods of moving plant/equipment and of using non-powered tools/equipment.
  • Noise —Long term exposure to sound presents as the major mechanism of injury. This is the case despite the fact that workers' compensation data are not an entirely reliable measure of this type of injury/disease. The data also reveal an unusually high concentration of deafness in two industries, viz, Silver-Lead-Zinc Ores and Black Coal. This might be more a reflection on data limitations than an indication that the problem is restricted to these industry classes. However, given 18% of injuries reported in these data are deafness, it seems advisable to consider additional ways of reducing exposure to excessive noise.
  • Work Environment —People familiar with the industry might possibly contend that there is little that can be done to change mining work environments so as to significantly improve OHS performance. Nevertheless, the breakdown agency is attributed to environmental factors in over 34% of cases and while it might prove more difficult, particularly in the short term, to reduce problems which result from `internal conditions' (mine floor, wall and face, etc), it might be worthwhile to explore strategies aimed at reducing risk related to `external' traffic and ground surfaces.

Restaurants, Hotels and Clubs Industry[10]

  • Falls (from a height or the same level) accounted for more than a quarter of all occurrences. In fact, at 26%, this industry had 10 percentage points more occurrences in this mechanism category than did the All Industries group. It is also worth noting that females in this industry appear to be disproportionately affected by falls, with over 30% of occurrences involving falls, compared to the All Industries average, for females, of around 18%. Even though considerably less males are employed in the industry, they still had disproportionately less falls than females, in terms of number of occurrences. It may well be worthwhile undertaking further investigation into falls, slips and trips involving females to establish whether there are any specific work practices or conditions which might be contributing to this disparity. The use of appropriate footwear might be a specific consideration in this regard.
  • Manual Handling Practices —Nearly 30% of cases were associated with muscular stress, and a significant proportion of these involved lifting, carrying or handling crates, cartons, kegs or other types of containers. As sprains and strains of joints and adjacent muscles comprised nearly 45% of injuries, it may well be worth addressing the issue of improving practices in this regard.
  • Work Environment —This appears to be an area where there might be scope to make a significant impact on reducing injury occurrences. Falls, slips and trips (especially by female workers), hitting or being hit by objects, and contact with hot objects account for more than half of all occurrences. The general work environment, wet/oily traffic areas, steps and stairways, and hazardous objects played a part in many of these cases. This tends to suggest that any measures which can be taken to improve conditions in `wet' areas of the work environment, or to ensure clear and easy access to, and egress from, steps and stairways should be worth consideration.

Meat Products Industry[22]

  • The OHS performance of the Meat Products Industry in 1992-93 was extremely poor. For the Group as a whole, the incidence rate was nearly five times the All Industry rate. Frequency rates (number of occurrences per million hours worked by employees) show a level nearly four times the All Industries rate. Incidence rates for all industry Classes within the Group are several times the All Industries rate, but the Meat (except Smallgoods or Poultry) Class is by far the worst, at slightly over six times the All Industries average. In fact, it appears to be the second worst performing industry Class in Australia, behind Brown Coal mining, which has one-twentieth the employment and for which, as a result, incidence rate calculations are not entirely reliable.
  • The injury/disease pattern suggests that consideration of the adequacy of training provided to younger workers might be productive in the prevention of open wounds and infectious diseases.
  • Data tend to support the notion that an effective range of preventive strategies might ideally be initially targeted on 20-29 year olds where the numbers of occurrences are high and the frequency and incidence rates are also quite high.
  • Non-powered Handtools and Equipment —Non-powered handtools, edged, accounted for nearly 30% of all injuries. This suggests that a review of practices in the usage of this equipment might be productive in identifying appropriate preventive procedures. Certainly, a focus on this area will be necessary to make any significant impact on reducing the total number of injury/disease occurrences.
  • Manual Handling Practices —About one-third of cases were associated with muscular stress, and more than half of these occurrences were associated with lifting, carrying, putting down or handling dead animals. As sprains and strains of joints and adjacent muscles comprised over 40% of injuries, it would appear well worthwhile to address the issue of improving practices in this regard. It is also worth noting, that a large proportion of muscular stress cases are associated with handling non-powered, edged handtools. In assessing any possible action as a result of the above, it might be worthwhile to consider those practices causing open wounds and those causing muscular stress separately.

Textiles Industry[12]

  • The under 20 year age group experienced the highest proportion of Fractures and Open Wounds. The prevalence of Open Wounds decreases with increasing age suggesting that consideration of the adequacy of training provided to young workers might be worthwhile.
  • Overall, the data indicate that, for the Textiles Industry, females make less workers' compensation claims in proportion to their numbers than do males. The reasons for this may be that males work in more hazardous occupations than females or engage in more hazardous work-practices or that females under-report work-related injury and disease. Perhaps closer examination of these issues would prove productive.
  • It is worth examining the difference between males and females in terms of duration of absence from work after an injury/disease occurrence. For all claimants from the Textiles Industry, the average duration of absence was 56 working days compared to an All Industries average of 41 working days. When broken down by gender, women working in the Textiles Industry were found to have an average duration of absence of 75 working days while men had a significantly lower average time lost of 49 working days. These values compare with All industry average values of 52 working days lost for females and 37working days lost for males. The average time lost for the Textile Fibres, Yarns and Woven Fabrics industry group was 53 working days (75 working days for females, 45 working days for males) while that for the Other Textile Products industry group was 62 working days (74 working days for females, 55 working days for males). Clearly, the overall performance of the Textiles Industry is poor in this regard and particularly so for females.
  • On the basis of the above data, an examination of Textiles Industry work practices, and compensation claims management involving female employees may be warranted.
  • Occupation —Four out of ten of the injury/disease occurrences reported in the Textiles Industry during 1992-93 were experienced by Machine Operators, an occupational group which includes the three highest risk occupations. The workers at the highest risk of injury/disease were the Fabric Production Machine Operators, accounting for over one-fifth of the reported occurrences. An analysis of work-practices for this occupation may result in a reduction in work-related injury and disease.
  • Plant —Machinery, fixed plant, mobile plant and transport accounted for one in three of the newly compensated injury/disease occurrences reported in the Textiles Industry during 1992-93. In particular, textile, clothing and footwear machinery was the largest specific breakdown agency, accounting for over one in eight occurrences. The injury/disease risk factors encountered while using this type of machinery are likely to include: being hit or trapped by the machinery; suffering muscular stressing; or deafness as a result of long term exposure to sound. An examination of the way in which plant is utilised could make a significant impact in reducing the incidence of work-related injury and disease.
  • Work Environment —Falls, slips and trips accounted for nearly one-eighth of the injury/disease occurrences suggesting that, in particular, an examination of the layout and conditions in the indoor working environment may be worthwhile.

Clothing and Footwear Industry[13]

  • Occupation —Over one-half of the injury/disease occurrences reported in the Clothing and Footwear Industry during 1992-93 were experienced by Machine Operators, an occupational group which includes three out of four of the highest risk specific occupations. The workers at by far the highest risk of injury/disease were the Textile Sewing Machinists, accounting for nearly two-fifths of the reported occurrences. An analysis of work practices for this occupation may result in a reduction in work-related injury and disease.
  • Plant —Machinery, fixed plant, and mobile plant accounted for over one-third of the newly compensated injury/disease occurrences reported in the Clothing and Footwear Industry during 1992-93. In particular, textile, clothing and footwear machinery was the largest specific breakdown agency, accounting for over one-fifth of occurrences. Over seven in ten of the claims involving textile, clothing and footwear machinery resulted from muscular stress and, specifically, five in ten claims involving this type of machinery were from repetitive movement, low muscle loading, a mechanism which includes OOS claimants. An examination of the way in which this machinery is utilised could make a significant impact in reducing the incidence of work-related disease.
  • Work Environment —Falls, slips and trips accounted for over one-eighth of occurrences and more than one-half of these were associated with hazardous indoor traffic and ground areas and steps and stairways. An examination of the layout and conditions in the indoor working environment together with the usage of appropriate footwear may be worthwhile.

Cleaning Services Industry[11]

  • It would appear worthwhile, on several counts, to undertake further investigation to establish whether part-timers represent a particular OHS problem in this industry.
  • Overall, given the high proportion of injury/disease cases attributable to them and the high frequency rates they experience, strategies for improving OHS performance might initially be targeted on the 35-54 year old age group to achieve best short term results.
  • Handtools, Appliances and Equipment —Powered cleaning equipment was involved in nearly 16% of occurrences, while brooms and mops accounted for a further 4% of cases. The majority of cases involved lifting/handling of this equipment. While this has obvious implications for manual handling practices, perhaps it would also be worthwhile considering more detailed investigation to ascertain whether the designs, weights and other practical features of powered equipment are optimal and whether the number of occurrences could be reduced by wider usage of appropriate powered equipment. As one in five cases were associated with this breakdown agency, it appears to be an area of potential concern and worthy of further attention.
  • Manual Handling Practices —Two in every three injury cases are sprains and strains of joints and adjacent muscles and half of all injury/disease occurrences result from some form of body stressing. A large proportion of these affect the lower back and are associated with lifting/handling of cleaning equipment and movement of furniture and fittings. Clearly, a focus on improving manual handling practices will be necessary to make any significant impact on reducing the total number of injury/disease occurrences.
  • Work Environment —About one fifth of injury/disease cases resulted from slips and trips and over half of these were connected with wet, oily, icy or otherwise slippery surfaces. Given the nature of cleaning activity, this type of surface condition is largely unavoidable. However, it may be productive to investigate, among other things, the possibility of reducing risk by wider utilisation of more appropriate footwear.

Hospitals, Nursing Homes and Related Industries[23]

  • Hospitals, Nursing Homes and Related Industries overall, showed a rate 30% above the all industries rate. The ASIC Class with more than two-thirds of the total employment, Hospitals (except Psychiatric Hospitals), was nearly 20% higher than the All Industries rate, Psychiatric Hospitals was nearly 30% higher, Nursing Homes was over one and three-quarters the All Industries rate and Community Health Centres (Medical) was nearly twice that rate.
  • At around 78% of the workforce, the Health subdivision has a far higher proportion of female workers than the general workforce where, on average, around 42% of workers are female. Consistent with their employment proportion, females accounted for about three-quarters of compensation claims lodged in the Health subdivision.
  • Of the nearly 13,000 compensated injury/disease occurrences reported during 1992-93 in the Hospitals, Nursing Homes and Related Industries sector, nearly a quarter affected Registered Nurses, over a fifth affected Ward Helpers and more than an eighth affected Enrolled Nurses.
  • The Hospitals (except Psychiatric Hospitals) class had by far the highest number of cases involving Registered Nurses. Enrolled Nurses also had their highest number of cases in the hospitals class, but also had a relatively large number of cases in Nursing Homes. Ward Helpers had most of their occurrences in Nursing Homes, but with a large proportion in the hospitals class. Of course, these differences across classes might simply reflect the different occupational employment profiles in these classes. Alternatively, it could be that a worker in a particular occupational classification is affected by the fact that he/she undertakes a different mix of functions and/or works with different technology/equipment, depending on the industry class in which he/she is employed. A comparison of relevant occupational incidence and frequency rates across industry classes would be an ideal starting point in deciding whether there was room for improved practices in an industry class, based on better performances in other classes. Unfortunately, there are currently no adequate denominator data available for this purpose. However, it might prove productive to undertake this type of analysis, perhaps through appropriately designed case studies.
  • Similarly, differences across jurisdictions in incidence and frequency rates for a particular occupation might also flag opportunities to identify and implement best practice approaches. A comparison of frequency rates across jurisdictions (excluding Victoria) suggests that the poorer performing jurisdictions for Registered Nurses were Tasmania, Northern Territory and South Australia, while the best performance occurred in New South Wales. For Enrolled Nurses the poorer performing jurisdictions appeared to be Northern Territory, South Australia and Tasmania while the best performer was Queensland. For Ward Helpers the poorer performers appeared to be Tasmania, South Australia and New South Wales while the best performer was Western Australia. Further investigation to establish the precise extent of these apparent differences and any underlying factors influencing them might prove worthwhile.
  • Overall, given that nearly 45% of injury/disease cases are attributable to them and they experienced high frequency rates relative to other age groups, with longer average time lost per occurrence, strategies for improving OHS performance might initially be targeted on the 35-49 year old age group to achieve best short term results. Perhaps a particular focus on Ward Helpers and Enrolled Nurses in these age groups would also be productive.
  • Over a third of the occurrences shown as Hit by Moving Objects were instances where workers were hit by another person, either accidentally or deliberately (anecdotal evidence and other studies undertaken suggest that violence to health workers might be an area of significant under reporting due to factors such as peer pressure, a tendency to avoid lengthy paperwork and concern regarding accusations of patient abuse).
  • It is possible that the peak number of occurrences appears between 9am and 10am as a result of more workers being on hand at this time of day for this industry. However, it might be useful to establish whether there are any other contributory factors affecting the increased number of occurrences at this time of day.
  • Manual Handling Practices —Body stressing accounted for an inordinately high proportion (57%) of occurrences. By far the majority of these were the result of muscular stress while lifting, carrying, putting down or handling patients. Most of these cases appear to have affected the lower back, indicating that manual handling practices are a primary area of concern and, probably, the greatest area of potential savings from improved OHS performance.
  • Work Environment —Over an eighth of injury/disease cases resulted from falls, slips and trips and a third of these were connected with wet, oily or otherwise slippery surfaces. This tends to suggest that any measures which can be taken to improve conditions in the `wet' areas of the work environment, or to ensure clear and easy access to, and egress from internal trafficways might be worth consideration. At over 4% of cases, Mental Stress presented a greater than usual problem for Hospitals (except Psychiatric Hospitals), and most of these occurrences appeared to impact on Registered Nurses. While these cases constituted a relatively small proportion of the total number of compensated cases, anecdotal evidence suggests that cases which reach this stage represent only a part of the overall effects of stress on work-related injury/illness and general productivity in this industry. Some of the factors identified by other studies as contributing in this regard include burnout, substance abuse, shiftwork, and mental illness. Further investigation to establish the extent of this problem and the stress/stressors involved might be productive.
  • Specific Occupation Groups —Ward Helpers had an injury/disease incidence rate more than 3 times the All Occupation rate and a frequency rate nearly 4 times the all Occupation rate. The incidence rate for Enrolled Nurses was twice the All Occupations rate and the frequency rate nearly two and a half times the All Occupations rates. The high rates being experienced by these workers suggests that it might be productive to undertake a review of the functions of these occupations and the techniques employed to accomplish tasks with a view to improving procedures. In this respect, the degree of variation in rates between jurisdictions indicates that there might be significant scope for an exchange of information about operational procedures and equipment with a view to the implementation of best practice approaches to reduce specific types of problems.
  • Other —Being hit by patients accounts for relatively low but still significant proportion of injuries, particularly in the Nursing Homes industry class. It may be worth considering ways in which risk can be reduced in this area.

Agriculture and Services to Agriculture Industries[24]

  • The fatality incidence rate for the Agriculture sector was 26 fatalities per 100,000 wage and salary earners. This rate was over three and one-half times the Australian All Industries rate of 7 fatalities per 100,000 wage and salary earners.
  • The Agriculture sector was a significantly poorer performer, in terms of occupational health and safety, than the average Australian industry, with an injury/disease incidence rate which was over 90% higher than the All Industries rate.
  • The injury/disease frequency rates are consistent with incidence rate information in showing the Agriculture sector as a poor performer relative to other industries. The overall Agriculture sector rate was two-thirds higher than the All Industries rate. Within the sector, the Services to Agriculture group experienced the highest frequency rate at more than two and one-half times the All Industries rate.
  • Over one-half of the occurrences were experienced by Agricultural Labourers and Related Workers (ASCO 82) while more than one-seventh were experienced by Sheep Shearers (ASCO 4929), and one-tenth by Farmers and Farm Managers (ASCO 1401).
  • Most occurrences involved 20-24 year olds. The highest incidence rate was experienced by the 25-29 year olds followed by 20-24 year olds and 30-34 year olds. The highest frequency rate was experienced by 2S-29 year olds, followed by 15-19 year olds and 20-24 year olds.
  • Sprains and strains of joints and adjacent muscles was the most frequently occurring injury in the industry, accounting for 36% of cases and three out of ten compensable working days lost, at an average 49 days lost per occurrence. The second most frequently occurring injury was fractures. They represented just over one-sixth of all injury/disease cases and were responsible for 20% of all compensated days lost, at an average 57 days lost per occurrence. The third most common occurrence was open wounds not involving traumatic amputation which constituted 17% of cases, accounting for 6% of compensated days lost at an average 21 days per occurrence.
  • Plant —Mobile Plant and transport was involved in 14.7% of occurrences. Over one-quarter of these involved motorbikes/trailbikes, while one-sixth involved tractors. Almost half the occurrences involving tractors were as a result of falls. Machinery and (mainly) fixed Plant was involved in 9.4% of occurrences. Over one-half of these involved the use of sheep shearing plant, while one-fifth were associated with conveyors and lifting plants.
  • Handtools, Appliance and Equipment —Non-powered handtools appliances and equipment accounted for over 17% of occurrences. Almost one-third of these involved edged handtools, in particular, knives, scissors, shovels, scythes and axes. A large majority of the accidents involving edged handtools resulted in open wounds. Powered equipment, tools and appliances accounted for nearly 3% of cases. Over one-fifth of these involved chainsaws, while a further one-fifth involved abrasive, planing or cutting powered tools.
  • Manual Handling Practices —Body stressing was associated with over one-quarter of all cases and manual handling activities were involved in four-fifths of these. A frequent outcome of this type of body stressing was sprains and strains of joints and adjacent muscles. Large numbers of these occurrences involved handling livestock (in particular sheep). This suggests it might be worthwhile to attempt to establish whether handling practices involving livestock currently employed within the industry are optimal. Sprains and strains connected with the manual handling of fastening, packing and packaging equipment also presented a problem for the industry.
  • Work Environment —Falls slips and trips accounted for almost one-quarter of occurrences. The general condition of the work environment accounted for the majority of falls on the same level. This tends to suggest that any measures which can be taken to improve general site tidiness and/or remove traffic obstructions might be useful in reducing injury risk.

Meat Products Industry[25]

(Note that this industry analysis was the second undertaken using workers' compensation data. The first, based on 1992-93 data, was used by various industry representatives to assist in development of a set of occupational health and safety guidelines for the industry, focused on the more prevalent problems in the industry. Consequently, the second analysis, based on 1993-94 data, was able to identify the continuing, specific injury/disease problems in the industry and provide information on practical ways of initially addressing those problems. To date, this has been the only issue of this series where this approach has been possible.)

  • The OHS performance of the Meat Products Industry in 1993-94 remained extremely poor. For the Group as a whole, the incidence rate (number of occurrences per 1,000 wage and salary earners) was more than five times the all industry rate. This poor performance is also reflected in frequency rate information (number of occurrences per million hours worked by employees) which is available at the ASIC Group level and shows a rate four times the all industries rate for the Meat Products Industry Group. Incidence rates for all industry Classes within the Group were several times the All Industries rate, but the Meat (except Smallgoods or Poultry) Class was by far the worst, at slightly over six times the All Industries average. In fact, it appears to be the second worst performing industry Class in Australia, behind Brown Coal mining, which has one-twentieth the employment and for which, as a result, incidence rate calculations are not entirely reliable.
  • The data indicate that the various occupations were confronted with different mixes of injury/disease problems. Trades Assistants and Factory Hands, who experienced more than half the injury/diseases recorded in the Meat (except Smallgoods or Poultry) industry class, were most frequently affected by Sprains and Strains of joints and adjacent muscles, but also had significant problems with Open Wounds and the highest proportion of Zoonoses and Burns. In contrast, Meat Tradespersons were most frequently affected by Open Wounds, followed by Sprains and Strains. The most serious problem for Meat Packers was clearly Sprains and Strains.
  • Data provide some insight into the types of injury /disease which are of concern to particular age groups within the Meat (except Smallgoods or Poultry) Industry Class. It is noteworthy that the proportion of Open Wound cases tended to decline with age, raising the possibility that improvements in training for younger workers in this area might be productive in reducing the number of occurrences. The proportion of Sprains and Strains increased steadily with age before plateauing between ages 35 to 50, suggesting that it might be beneficial for workers in this age range to pay close attention to appropriate manual handling practices. It also appears that Zoonoses was a more frequent occurrence for younger, rather than older, workers.
  • The injury/disease pattern suggests that consideration of the adequacy of training provided to younger workers might be productive in the prevention of open wounds and infectious diseases.
  • Looking at the large number of cases, the types of injury/disease experienced and the high incidence and frequency rates, it would seem that strategies for improving OHS performance might initially be targeted on the ` to 29 year old age group to achieve best short term results.

Type of Occurrence Cross-classifications

(Mechanism of injury cross-classified by breakdown agency provides a focus on connections which figure in a large proportion of injury/disease occurrences in the Meat (except Smallgoods or Poultry) Industry Class.)

  • A third of all reported cases resulted from muscular stress of some type. Almost a third of these cases were associated with the lifting, carrying, putting down or handling of offal and waste products (noticeably more with handling than with lifting, etc). Over 15% of muscular stress cases resulted from handling knives, and about the same percentage resulted from lifting, carrying, putting down or handling carcases.
  • Hitting objects with the body accounted for nearly one-quarter of all cases. Nearly seven-eighths of these cases were associated with knives.
  • Nearly one-tenth of cases resulted from falls, slips and trips. The agencies most commonly associated with this mechanism were steps and stairways and wet, oily surfaces.

(Taking this type of analysis from a different perspective, by considering data which show nature of injury/disease cross-classified by mechanism and breakdown agency, allows some insight into the circumstances surrounding the most frequent type of injury/disease occurrences.)

  • Sprains and strains of joints and adjacent muscles was the most frequently occurring injury in the Meat (except Smallgoods or Poultry) Industry Class accounting for over a third of cases and almost a half of the total compensable working days lost, at an average 34 days lost per occurrence. This type of injury also entailed direct costs which were about 25% higher than the average cost per occurrence for the industry.
  • Nearly one-quarter of these cases resulted from the lifting, carrying, putting down or handling of offal and waste products. The National Guidelines for Health and Safety in the Meat Industry (produced by the Australasian Meat Industry Employees Union and the Meat and Allied Trades Federation of Australia) provide some useful ideas on how risk in this area might be minimised, including:
    1. reorganising layout of trimmers' and sorters' workplaces for easy transfer of product between tables
    2. provide height-adjustable tables to suit the height of the worker and the task at hand
    3. position tubs for trimmings and inedible products within easy reach so that workers do not have to throw or twist
    4. use of mechanical aids to reduce risk, for example, conveyor belts to transport containers of product, screws (with appropriate guards), self-tipping trolleys, etc.

  • About an eighth of sprains and strains cases derived from trips, slips, stumbles and falls on the same level. Over two-thirds of these cases were associated with wet or oily, or otherwise hazardous, traffic and ground surfaces, steps and stairways. The National Guidelines suggestions include:

    1. treating the surface of existing floors to improve slip-resistance, for example, acid etching, sand grinding, grooving,
    2. maintenance procedures that facilitate prompt repair of leaks from equipment or fittings
    3. designing appropriate routine cleaning procedures
    4. provision of suitable footwear with slip-resistant soles should be considered
    5. stairs, steps and ramps could be fitted with slip-resistant tread or surface.

  • Nearly one tenth of sprains and strains resulted from handling knives. The National Guidelines suggestions include:

    1. consideration of alternative knife designs to reduce wrist and arm strain, for example, designs that bend the knife handle rather than the wrist
    2. have a supply of knives with different handle sizes
    3. ensure knives are correctly maintained to have the sharpest possible cutting edge at all times.

  • Open wounds, not involving traumatic amputation, accounted for over one-quarter of all cases in the industry and over an eighth of the total working days lost at an average time lost of 14 days. On average the direct cost of this type of injury was half the cost of an average cost of a case within this industry. Not surprisingly, given the nature of the industry, two-thirds of these injuries resulted from being cut by a knife. The National Guidelines suggestions include:

    1. provide sufficient work space for each employee to reduce the risk of employees cutting or stabbing one another
    2. ensure that knives are pouched when not in use in pouches which are properly designed so that the blade is not exposed, too much handle does not protrude and only one knife is stored per compartment
    3. ensure knife handles are cleaned regularly during the day
    4. where there is a risk of a knife cut to a particular bodily location, provide and ensure proper use of personal protective equipment such as mesh gloves and arm or abdominal guards.

  • Disorders of muscles, tendons and other soft tissue accounted for just under 5% of cases, but almost 10% of total days lost at an average time lost of 54 working days. The average direct cost per occurrence for this type of disease was nearly one-third higher than the average cost of all types of injury/disease in the industry. Nearly a third of cases with this type of occurrence were connected to handling knives and about 30% were connected to lifting, carrying, putting down and handling offal and waste products. Points from the National Guidelines mentioned above with regard to sprains and strains are also relevant in this respect.

  • Specified Zoonoses accounted for 6% of cases and about 4% of working days lost at an average time lost of 19 days. The average direct cost for this type of disease was about 40% lower than the average cost of all types of injury/disease. Naturally enough, these disorders result from contact with, or exposure to, germs, bacteria and other micro-organisms. Two-fifths of these exposures were connected with offal and waste products, while just under this proportion was connected to carcases and less than 5% were related to live animals. The National Guidelines suggestions include:
    1. ensure work practices minimise the risk of contamination and infection
    2. provide adequate personal hygiene facilities
    3. laundering of all work clothing on site or by a professional off-site laundry
    4. develop an occupational health program which includes relevant vaccination and first aid facilities
    5. provide and ensure use of appropriate equipment including personal protective equipment
    6. training for all employees on the risks of zoonotic infection and possible control measures and, for employees undertaking high risk tasks, skills training to assist them in identifying and controlling the risks.

  • Looking at day of occurrence by time of occurrence, the working hour of the week in which employees were most likely to be injured was 10am to 11am on a Thursday, marginally ahead of Monday between 8am and 9am.

Summary Points

  • Non-powered Handtools and Equipment —Non-powered knives were associated with nearly 30% of all injuries. They featured significantly in open wound injuries, sprains and strains and disorders of muscles and other soft tissue. This suggests that a review of practices in the maintenance and usage of this equipment might be productive in identifying appropriate preventive procedures and the analysis sets out a number of suggestions as starting points for discussion. Certainly, a focus on this area will be necessary to make any significant impact on reducing the total number of injury/disease occurrences in this industry. (Following release of the first analysis of this industry, it was suggested that the high proportion of open wounds (nearly 40% of cases) was the result of workers standing too close to each other in the processing line when using knives, etc. It was also suggested that many workers in the process-line were not wearing all the necessary protective gear as they considered that the clothing, etc, currently available reduces the speed at which they can work. These issues may be worth further investigation.)
  • Plant —Only about 4% of injury/disease occurrences were attributed to a breakdown agency of machinery and (mainly) fixed plant. This tends to suggest that existing Plant did not present as a major area of concern for the industry. However, the analysis does beg the question of whether the absence or underutilisation of more technologically advanced types of Plant might be a contributory factor to high injury rates in this industry.
  • Noise —On the basis of the available data, deafness represents under 3% of cases reported. This might be interpreted as meaning that deafness does not present as a major area of immediate concern for the industry. However, it should be borne in mind that deafness is a gradual onset disease which is not always easily detected at the time at which it occurs.
  • Manual Handling Practices —Nearly one-third of cases were associated with muscular stress of some type. Almost a half of these cases were associated with lifting, carrying, putting down or handling carcasses, offal and waste products. As sprains and strains of joints and adjacent muscles comprised over a third of injuries, a half of total compensable days lost, and had an average cost 25% higher than the average cost for all injuries, it would appear well worthwhile to address options for improving practices and introducing more mechanical aids to reduce risks in this regard. (Following release of the first analysis of this industry, it was suggested that a large proportion of muscular stress/Sprains and Strains were associated with movement of animal innards/offal. In a relatively small number of works this problem was apparently obviated through mechanisation of the process. The effectiveness of this approach might be worth further investigation). It is also worth noting that a large proportion of muscular stress cases are associated with handling non-powered, edged handtools (knives) and the analysis makes a number of suggestions as starting points for improving performance in this regard.
  • Work Environment —Falls, slips and trips appear to account for a little under 10% of occurrences. Therefore, it would appear worthwhile to address this issue.

Appendix 2 - Key Aspects of Selected Industry Occupational Health and Safety Performances as Revealed by Analysis of Workers' Compensation Data

References

  1. Worksafe Australia, National Data Set for Compensation-based Statistics, Australian Government Publishing Service, Canberra, 1987.
  2. Williams T, Foley G, Cole B, Silman G, Davis M, Gale J, Hickling C, Elliot J, Compendium of Workers' Compensation Statistics, Australia, 1992-93,Australian Government Publishing Service, Canberra, 1995.
  3. Industry Commission, Work, Health and Safety—Inquiry into Occupational Health and Safety, Report No. 47, Australian Government Publishing Service, Canberra, 1995.
  4. Division of Workplace Health and Safety, Workers' Compensation Data—A Poor Indicator of Workplace Injury and Disease, Division of Workplace Health and Safety, Sydney, 1994.
  5. Foley G, Gale J, Gavenlock L, `The Cost of Work-related Injury and Disease, J Occup Health Safety—Aust NZ, 11(2):171-194, 1995.
  6. Australian Bureau of Statistics, Work-related Injuries and Illnesses, New South Wales, Australian Bureau of Statistics, Sydney, 1994.
  7. Hopkins A, `Are workers' Compensation Statistics a Health and Safety Hazard?', Aust J Public Admin, 53(1):78-86, 1994.
  8. James C, Work and Health: The Origins, Management and Regulation of Occupational Illness, Macmillan Education, Melbourne, vol 33-56, 1993.
  9. Australian Bureau of Statistics, Australian Bureau of Statistics Yearbook 1995,Australian Bureau of Statistics, Sydney, 1995.
  10. Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 2 [NOHSC:11016(1994)], Worksafe Australia, Sydney, 1994.
  11. Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 5—Cleaning Services Industry[NOHSC:11024(1995)], Worksafe Australia, Sydney, 1995.
  12. Cole B, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 4—Textiles Industry[NOHSC:11023(1995)], Worksafe Australia, Sydney, 1995.
  13. Cole B, Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 6—Clothing and Footwear Industry[NOHSC:11025(1995)], Worksafe Australia, Sydney, 1995.
  14. Driscoll TR, `Are Work-related Injuries More Common than Disease in the Workplace?', Occup Med, 43:164-166, 1993.
  15. Hodgson JT, Jones JR, Elliott RC, Osman J, Self-reported Work-related Illness, Health and Safety Executive, London, 1993.
  16. Getty L, Hetu R, Noble W, Waugh R, Occupational Noise-induced Hearing Loss, Worksafe Australia/The University of New England, Sydney, 1991.
  17. Castles I, Labour Statistics Australia, Australian Bureau of Statistics, Canberra,1994.
  18. Macaskill P, Mandryk J, Leigh J, Improved Coverage and Relevance of National Occupational Health and Safety Surveillance in Australia, Worksafe Australia, Sydney, 1995.
  19. Foley G, Occupation Health and Safety Performance Overviews, Selected Industries, Issue No. 1 [NOHSC:11013(1994)], Worksafe Australia, Sydney, 1994.
  20. Foley G, `Forestry Logging and Log Sawmilling Industries: Occupational Health and Safety Performance Overview, Australia 1991-92', J Occup Health Safety— Aust NZ, 10(5):467-474, 1994.
  21. Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 8—Construction Industry[NOHSC:11030(1995)], Worksafe Australia, Sydney, 1995.
  22. Foley G, `Meat Products Industry: Occupational Health and Safety Performance Overview, Australia 1992-93', J Occup Health Safety—Aust NZ,11(6):629-644, 1995.
  23. Foley G, Cole B, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 7—Hospitals, Nursing Homes and Related Industries [NOHSC:11028(1995)], Worksafe Australia, Sydney, 1995.
  24. Cole B, Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 9—Agriculture and Services to Agriculture Industries, Worksafe Australia, Sydney, 1995.
  25. Foley G, Davis M, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 10—Meat Products Industry, Worksafe Australia, Sydney, 1996.

Further Reading

Baker EL, Melius JM, Millar JD, `Surveillance of Occupational Illness and Injury in the United States: Current perspectives and future directions', J Public Health Policy, 9(2):198-221, 1988.

Cole B, Foley G, Occupational Health and Safety Performance Overviews, Selected Industries, Issue No. 9—Agriculture and Services to Agriculture Industries[NOHSC:11034(1995)], Worksafe Australia, Sydney, 199S.

Coleman PJ, `Injury Surveillance', Scand J Work Environ Health,9:128-135, 1983.

Foley G, `Construction Industry: Occupational Health and Safety Performance Overview, Australia 1991-92', J Occup Health Safety—Aust NZ,10(6):566-578, 1994.

Frumkin H, Camara V, `Occupational Health and Safety in Brazil', Am J Public Health, 81(12):1619-1624, 1991.

Hales T, Seligman PJ, Newman SC, Timbrook CL, `Occupational Injuries due to Violence', J Occup Med, 30(6):483-487, 1988.

Harrison JE, Frommer MS, Ruck EA, Blyth FM, `Deaths as a Result of Work-related Injury in Australia, 1982-1984', Med J Aust, 150:118-125, 1989.

Hilaski HJ, `Where Occupational Health Reporting Systems go Wrong', Am J Industrial Med, 8:435-439, 1985.

Hopkins A, `Injury Statistics—Where the Figures Fall Down', Aust Safety News, 66(1):42-51, 1995.

Hopkins RS, Nelson M, Lalonde C, `The Incidence of Work-related Injury: Comparison of Emergency Department and Workers' Compensation Data', Am J Epidemiol, 132(4):762, 1990.

James C, Papaicsik I, Wyatt T, `Work-related Injuries: A Comparative Study of Self-employed and Employee Transport Workers in Brisbane', J Occup Health Safety— Aust NZ, 9(3):245-253, 1993.

Jensen RC, `Workers' Compensation Claims Relating to Heat and Cold Exposure', Professional Safety, 28(9):19-24, 1983.

Korrick SA, Rest KM, Davis LK, Christiani DC, `Use of State Workers' Compensation Data for Occupational Carpal Tunnel Syndrome Surveillance: A Feasibility Study in Massachusetts', Am J Industrial Med, 25:837-850, 1994.

Larsson TJ, `We Need Applied Prevention—Not Statistics', J Occup Health Safety—Aust NZ, 7(4):287-294, 1991.

Markowitz S, Environmental and Occupational Medicine,2nd Edition, Brown and Company, Boston, Chapter 3:19-28, 1992.

Mathias CGT, Sinks TH, Seligman PJ, Halperin WE, `Surveillance of Occupational Skin Disease: A Method Utilizing Workers' Compensation Claims', Am J Industrial Med, 17:363-370, 1990.

Solving the Workers' Compensation Puzzle, MMI, Sydney, 1994.

Oleinick A, Gluck JV, Guire KE, `Establishment Size and Risk of Occupational Injury', Am J Industrial Med, 28:1-21, 1995.

Ore T, `Inter-industry Variations in Occupational Injuries in Australia', J Occup Health Safety—Aust NZ, 8(1): 41-45, 1992.

Rossignol M, `Planning Preventive Occupational Health Services at the Community Level', Canadian J Public Health, 82:115-119, 1991.

Seligman PJ, Halperin WE, Mullan RJ, Frazier TM, `Occupational Lead Poisoning in Ohio: Surveillance Using Workers' Compensation Data', Am J Public Health, 76:12991302, 1986.

Sorock GS, Smith E, Hall N, `An Evaluation of New Jersey's Hospital Discharge Database for Surveillance of Severe Occupational Injuries', AmJ Industrial Med, 23:427-437, 1993.

Spiegel J, Yassi A, `Occupational Disease Surveillance in Canada: A Framework for Considering Options and Opportunities', Canadian J Public Health,82:294-299, 1991.

Tanaka S, Seligman P, Halperin W, Thun M, Timbrook CL, Wasil J, `Use of Workers' Compensation Claims Data for Surveillance of Cumulative Trauma Disorders', J Occ Med, 30(6):488-492, 1988.

Tengs TO, Adams ME, Pliskin JS, Safran DG, Siegel JE, Weinstein MC, Graham JD, `Fivehundred Life-saving Interventions and their Cost-effectiveness', Risk Analysis, 15(3):36390, 1995.

Vandenheuvel A, Wooden M, `Self-employed Contractors in Australia: How Many and Who are They?', J Industrial Relations, 37:263-279, 1995.

Wigglesworth EC, `Serious Occupational Injuries in Australia: Some Deficiencies of the Existing Data Collections', Community Health StudiesJ14(3):279-287, 1990.

Yassi A, `Health and Socioeconomic Consequences of Occupational Respiratory Allergies: A Pilot Study Using Workers' Compensation Data', Am J Industrial Med,14:291-298, 1988.

Page last updated: 03/04/2009