UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An epidemiologic investigation of injury mortality among sawmill workers Barroetavena, Maria Cristina 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-610608.pdf [ 7.52MB ]
Metadata
JSON: 831-1.0090596.json
JSON-LD: 831-1.0090596-ld.json
RDF/XML (Pretty): 831-1.0090596-rdf.xml
RDF/JSON: 831-1.0090596-rdf.json
Turtle: 831-1.0090596-turtle.txt
N-Triples: 831-1.0090596-rdf-ntriples.txt
Original Record: 831-1.0090596-source.json
Full Text
831-1.0090596-fulltext.txt
Citation
831-1.0090596.ris

Full Text

A N EPIDEMIOLOGIC INVESTIGATION OF INJXJRY M O R T A L I T Y A M O N G S A W M I L L WORKERS by M A R I A CRISTINA B A R R O E T A V E N A M.D. The University of Buenos Aires (ARGENTINA), 1972 M.H.Sc., The University of British Columbia, 1990 A THESIS SUBMITTED IN P A R T I A L F U L F I L M E N T OF THE REQUIREMENTS FOR T H E D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Health Care and Epidemiology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A April, 2001 © Maria Cristina Barroetavena, 2001 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract Sawmill jobs rank among the most hazardous occupations in Canada due to the high work-related injury morbidity and fatality rates. As little is known about the circumstances of fatal injuries in this workforce, a study of all fatal injuries (on-the-job and off-the-job) among sawmill workers was undertaken. The study population comprises over 26,000 workers from the British Columbia Sawmill Workers Cohort. We combined data from this database with B.C. Coroners Service (BCCS) and Workers' Compensation Board (BCWCB) data to analyze the facts surrounding the occurrence of fatal injuries among sawmill workers during 1950-1990. This study consisted of three components: 1) an analysis of the usefulness of the data sources available for surveillance of work-related injuries, 2) an analysis of work-related fatalities among sawmill workers for the period 1950-1990, and 3) an analysis of all fatal injuries in this workforce during 1950-1990. The W C B records alone identified more work-related fatalities than the BCCS files alone (95.4% vs. 81.8%). Each data source has specific limitations precluding a complete count of fatal injuries at the workplace. Forty cohort members were fatally injured while working at a sawmill. The work-related fatality rate was 18.3 per 100,000 person-years. The risk of fatal work-related injury was lower during 1970-1990 than during 1950-1969. Crude fatality rates were higher among workers older than 35 years and in the occupational categories of "machine operators/clearers/sorters" and "mobile equipment operators". Factors related to the workplace physical and socio-cultural environment (e.g. unsafe work station, defective equipment, safety policies) were the main contributors to fatal injuries. When compared to the general male population, sawmill workers had fewer deaths than expected from every injury type except "being caught by machinery". The group of "skilled labour" workers had the lowest risk of dying from motor vehicle crashes and suicides. Fatal motor vehicle crashes and suicides were highest among workers living in isolated mill towns. These findings have implications for prevention. Effective countermeasures should not only address a combination of factors related to the environment, the injury event and the person but these interventions should be evaluated in terms of cost-effectiveness. iii TABLE OF CONTENTS Abstract ii TABLE OF CONTENTS iv List of Tables viii List of Figures xi Glossary xiii Acknowledgments xiv CHAPTER ONE 1 1.0 Introduction 1 CHAPTER TWO 5 2.0 Background 5 2.1 Injury Research: Concepts and issues 5 2.2 The development of the epidemiological model of injuries 7 2.2.1 Pioneering work into the causes of injuries 7 2.2.2 The epidemiological model 9 2.2.3 The Haddon matrix 11 2.3 Canadian injury facts 12 2.3.1 Data Sources 12 2.3.2 Injury Mortality 14 2.3.3 Injury Morbidity 17 2.3.4 The economic cost of injuries 17 2.3.5 Overview of risk patterns for fatal injuries in British Columbia 19 2.4 Scientific literature about sawmill workers 22 2.4.1 The work environment 22 2.4.2 Work-related injuries 24 2.4.3 Non work-related fatal injuries among sawmill workers 28 CHAPTER THREE 29 iv 3.0 Methodology 29 3.1 The study population and the original data base: The BC Sawmill Workers Cohort 29 3.2 Data sources 32 3.2.1 B.C. Coroners Service 32 3.2.2 Workers' Compensation Board of B.C. 33 3.2.3 B. C. Division of Vital Statistics 33 3.3 Exposure assessment 34 3.4 Analytical methodology 35 CHAPTER FOUR 39 4.0 Completeness of ascertainment of work-related injury fatalities: implications for surveillance 39 4.1 Introduction 39 4.2 Methods 42 4.2.1 Datasources 42 4.2.2 Analysis 46 4.3 Results 49 4.3.1 Completeness of ascertainment of work-related fatal injuries to B.C. sawmill workers 49 4.3.2 Accuracy of "place of injury" to identify work-related fatal injuries 54 4.4 Discussion §6 CHAPTER FIVE 60 5.0 Epidemiology of work-related fatalities among sawmill workers, 1950-1990 60 5.1 Introduction 60 5.2 Methods 63 5.2.1 Study population- The original B.C. Sawmill Workers study 63 5.2.2 Work-relatedness of a fatality 65 5.2.3 Assessment of exposure 66 5.2.4 Statistical analysis 67 5.2.5 Data preparation 70 v 5.2.5 Internal cohort analysis by job categories 71 5.3 Results 75 5.3.1 Rates and time trends in fatal injuries at work 75 5.3.2 Factors related to the worker 76 5.3.3 Factors related to the environment 79 5.3.4 Factors related to the injury event 83 5.4 Discussion 86 CHAPTER SIX '. 96 6.0 AH fatal injuries among sawmill workers, 1950-1990 96 6.1 Introduction 96 6.2 Methods 98 6.2.1 BC Sawmill Workers Cohort 98 6.2.2 Cause of death - International Classification of Diseases 100 6.2.3 British Columbia Coroner's Office files 100 6.2.4 Occupational categories 101 6.2.5 Data Analysis 101 6.3 Results 104 6.3.1 Demographic characteristics of the cohort 104 6.3.2 Overview of mortality 105 6.3.3 Overall Mortality from Injuries 107 6.3.4 Motor Vehicle Crashes 109 6.3.5 Suicide U9 6.3.6 Falls, drowning, and machinery-related fatalities 126 6.4 Discussion 128 CHAPTER SEVEN 139 7.0 Conclusions 139 7.1 Completeness of ascertainment of work-related fatal injuries 139 7.1.1 Implications for public health policy 140 7.2 Work-related fatal injuries 141 vi 7.2.1 Implications for public health policy 143 7.3 All fatal injuries 145 7.5.7 Implications for public health policy 146 1A Limitations 147 7.4.1 Implications for public health policy 151 Bibliography 154 Appendix A - Location of study sawmills 166 Appendix B - Data collection forms 168 Appendix C - Job title groupings according to task, process area, and skill level 173 Appendix D - External causes of death categories by ICD revision 183 vii List of Tables Table 2.1 Hazardous conditions associated with high injury rates in B.C. sawmills 28 Table 4.1 Layout of data by source of ascertainment 47 Table 4.2 Sawmill work-related fatal injuries in all B.C. sawmill workers, ascertained by each source, 1984-1990 50 Table 4.3 A l l B.C. sawmill fatalities, 1984-1990, identified by each source: case description 50 Table 4.4 Characteristics of fatal injuries at work in B.C. sawmill, 1984-1990, by source of ascertainment 53 Table 4.5 Sensitivity and specificity of the B.C. death certificate "place of injury" item to identify work-related fatal injuries among cohort members, occurring at a sawmill or other work site, 1984-1990 54 Table 4.6 Fatality incidence in the B.C. sawmill industry between 1984-1990 55 Table 5.1 Location of study mills and number of workers participating in the study 64 Table 5.2 Occupational categories, based on job tasks 66 Table 5.3 Major process areas in a sawmill and exposure to safety hazards.... 67 Table 5.4 Job history file with a record for a hypothetical worker who held 3 jobs between 1950 and 1967 71 Table 5.5 Job history file with artificial records created for each job 72 Table 5.6 Demographic file with record for a hypothetical worker who held 3 jobs between 1950 and 1967 73 Table 5.7 Demographic file with records for 3 hypothetical workers who held 1 job each between 1950 and 1967 73 Table 5.8 Data analytic file for job comparison analysis 74 Table 5.9 Crude rates and adjusted rate ratios of work-related fatal injuries, sawmill workers, 1950-1990 76 viii Table 5.10 Rate ratio of work-related fatal injuries by occupational categories, 1950-1990 77 Table 5.11 Fatalities by occupational categories according to type of injury and seniority in current job 78 Table 5.12 Number and crude fatality rates of work-related injuries by process area, 1950-1990 79 Table 5.13 Number of Coroner's Jury recommendations to prevent work-related deaths 80 Table 5.14 Factors contributing to work-related fatal injuries, 1950-1990 81 Table 5.15 Number and percentage of work-related fatalities by nature of injury, 1950-1990 84 Table 5.16 Objects involved in work-related fatal injuries, 1950-1990 84 Table 5.17 Machinery involved in work-related fatal injuries, 1950-1990 84 Table 5.18 Work-related fatalities by type of injury, 1950-1990 85 Table 6.1 Categories of independent factors 103 Table 6.2 Vital status of the cohort members prior to linkage to the income tax file 104 Table 6.3 Vital status of the cohort members after linkage to the income tax file 104 Table 6.4 Characteristics of the cohort of 26,411 sawmill workers 105 Table 6.5 Location of mills and number of workers in each region 105 Table 6.6 SMRs for selected causes of death, among sawmill workers (n=26,411), 1950-1990 106 Table 6.7 SMRs and crude rates for external causes of death among sawmill workers, during sawmill employment only, 1950-1990 108 Table 6.8 Motor vehicle-related deaths as percentage of all unintentional Injuries 110 ix Table 6.9 Crude rates and adjusted rate ratios of motor vehicle fatalities among sawmill workers by age, seniority in mill, calendar period, and region, 1950-1990 I l l Table 6.10 Crude rates and adjusted rate ratios of motor vehicle crashes among sawmill workers by occupational groups, 1950-1990 112 Table 6.11 Rate of fatal motor vehicle crashes by day of occurrence, 1950-1990 113 Table 6.12 Crude rates and adjusted rate ratios of motor vehicle crashes among sawmill workers by mill, 1950-1990 115 Table 6.13 Motor vehicle deaths by road user type and age group 116 Table 6.14 Deaths by nature of injury, fatal motor vehicle crashes, 1950-1990.. 116 Table 6.15 Head injuries as cause of death by decade 117 Table 6.16 Types of motor vehicle crashes, 1950-1990 118 Table 6.17 Collision with roadside objects, fatal crashes, 1950-1990 119 Table 6.18 Crude rates and adjusted rate ratios of suicide among sawmill workers, by age, seniority in mill, calendar period and region, 1950-1990 120 Table 6.19 Crude rates and adjusted rate ratios of suicide among sawmill workers, by occupational groups, 1950-1990 122 Table 6.20 Rate of suicides by day of occurrence, 1950-1990 122 Table 6.21 Crude rates and adjusted rate ratios of suicide among sawmill workers by mill, 1950-1990 123 Table 6.22 Conditions associated with suicide, 1950-1990 124 Table 6.23 Suicide rates by method, sawmill workers, 1950-1990 125 Table 6.24 SMRs for falls, drowning, and machinery related by age group and calendar period, sawmill workers, 1950-1990 126 Table 6.25 Crude mortality rates from falls, drowning, and machinery related by worksite region, sawmill workers, 1950-1990 126 x List of Figures Figure 2.1 Percentage of deaths and hospitalizations due to unintentional injuries by age groups, Canadian males, 1985 15 Figure 2.2 Percentage of deaths and hospitalizations due to unintentional injuries by age groups, Canadian females, 1985 15 Figure 2.3 Years of life lost before age 65 for selected causes of death, Canada, 1985 16 Figure 2.4 Costs of pension and benefits, and research costs share for selected illnesses, Canada, 1986 18 Figure 2.5 Percentage of leading causes of injury deaths by sex, Canada, 1985 20 Figure 4.1 Methodology for analyzing completeness of ascertainment of sawmill work-related fatal injuries 43 Figure 4.2 Methodology for assessing accuracy of "place of injury" on death certificates in identifying work-related fatalities at a sawmill or other work site 45 Figure 4.3 Number of work-related fatalities in all B.C. sawmill workers, 1984-1990, identified by each data sources... 49 Figure 5.1 Work-related injury fatality rates per 100,000 person-years, 1950-1990 76 Figure 5.2 Work-related fatality rates by time of occurrence, 1950-1990 82 Figure 5.3 Work-related fatality rates by day of occurrence, 1950-1990 82 Figure 5.4 Work-related fatality rates by month of occurrence, 1950-1990 83 Figure 5.5 Activity at time of incident : 83 Figure 6.1 Comparison of motor vehicle injury fatality rates between BC male population and sawmill workers by age group, 1950-1990 I l l Figure 6.2 Motor vehicle-related death by time and day of crash, 1950-1990... 114 Figure 6.3 Rate of M V crashes by season, 1950-1990 114 Figure 6.4 Motor vehicle crashes by road user type, 1950-1990 115 Figure 6.5 Contributing factors to motor vehicle crashes, 1950-1990 117 Figure 6.6 Comparison of suicide rates between B.C. male population and sawmill workers by age group, 1950-1990 120 Figure 6.7 Suicide rates by trimester 123 Figure 6.8 Suicide by methods 124 Figure 6.9 Suicide rates by methods and age groups, 1950-1990 125 xii Glossary BC British Columbia BCCS British Columbia Coroners Service BCDF British Columbia Death File B C D V S British Columbia Division of Vital Statistics BCSWC British Columbia Sawmill Workers Cohort C.I. Confidence Intervals C M D B Canadian Mortality Data Base CR Capture-Recapture ICD International Classification of Diseases LTAS Life Table Analysis System M V Motor Vehicle NIOSH National Institute for Occupational Safety and Health OSHA Occupational Safety and Health Administration P M R Proportionate Mortality Ratio P Y L L Person Years of Life Lost RR Rate Ratio SMR Standardized Mortality Ratio U.S. United States WHO World Health Organization xiii Acknowledgments This study could not have been possible without the help, support and encouragement of many colleagues and friends, specially the members of my committee. They know only too well that this research would never have been completed without their support and patience. To my supervisor, Kay Teschke, special thanks for her vision, guidance and encouragement during the seemingly endless succession of crises that marked this period. Clyde Hertzman provided me with valuable comments and advice to enhance my skills as researcher. Paul Demers guided me with patience and wisdom through the methodological challenges of the statistical analyses without dismay. I also value the professional guidance I have received from Sam Sheps. I am sincerely grateful to my fellow colleagues who started the U B C Sawmill Workers Cohort study. Especially valuable were the support and encouragement provided by Aleck Ostry, Hugh Davies, Shona Kelly, Ruth Hershler, and Lisa Chan. Needless to say this work would not have been possible without the support of the many people at the Department of Health Care and Epidemiology whose support and friendship are treasured. I am grateful to Virginia, Laurel, Zeba, and Stephanie for their friendship. I am also grateful to Tej Sidhu, from the Chief Coroner's Office, for her kind assistance locating files and to June Wick for her warm hospitality while I was collecting data in Victoria. Finally, I thank Roberto, Daniel, Laura, Analia and Martina whose love sustained me during the completion of this research. xiv CHAPTER ONE 1.0 Introduction The major goal of this research is to expand knowledge on the epidemiology of injuries. The ultimate goal is to help to reduce the number of unnecessary deaths due to preventable injuries. In Canada, the forestry industry contributes substantially to the wealth of the country. The value of shipments by the sawmill, planing and shingle mill products reached $12.5 billion in 1993. With a workforce of approximately 70,000 sawmill and planer mill workers, this country is among the world's leading producers of forestry products (Statistics Canada, 1993). In 1992, Canada was the world's largest producer of newsprint, the second largest producer of pulp, and the third largest manufacturer of sawn lumber (Statistics Canada, 1994; Statistics Canada, 1995a). British Columbia with 45 million hectares of productive forest, accounts for approximately 50 percent of the activity of the Canadian forest industry. In addition to being a principal component of the economy of the country, the forestry industry is among the most hazardous. Forestry based industries have the second highest fatality rate in Canada (117 fatalities per 100,000 workers in 1987) after fishing (Labour Canada, 1990). Sawmill workers have the second highest rate of work-related injury morbidity and mortality of the wood products industries. During 1987-1991 sawmill workers had an average of 9 wage loss claims per 100 person-years which exceeded the average for all industries combined of 6.5 wage loss claims per 100 person-years (WCB, 1992). l In spite of being an occupational group at high risk of work-related fatal injuries little research has been done to answer these questions: Who is fatally injured? How, where, and when do fatal injuries occur? Is this a workforce at high risk of non-work-related injuries as well? Much of the difficulty in addressing these types of questions is due to the lack of good quality data on the multiple factors and risks that contribute to injuries. An injury surveillance system that provides comprehensive and timely information on injury incidents is a basic requirement for the development of injury control strategies (Robertson, 1992). The present study, stemming from an ongoing investigation of a British Columbia cohort of over 26,000 sawmill workers (Hertzman et al., 1997), is aimed at applying the basic epidemiological framework of the Haddon matrix (Haddon, 1980) to the study of all injury fatalities that occurred to the cohort members. To our knowledge, a longitudinal epidemiologic study of both work-related and non-work related injury deaths among a cohort of blue-collar workers for a forty-year period has not been conducted in Canada. Our literature search identified only a mortality study in a cohort of fisherman from the Atlantic Provinces in Canada, which estimated the risk of dying from work-related injuries from 1975 to 1983 (Hasselback andNeutel, 1990). The results will enhance our understanding of the range of opportunities to control hazards and the risk of fatal injuries. The study will generate information that can be used to prioritize and guide future research into specific high risk groups within this industry. 2 Furthermore, the findings will provide evidence of the potential usefulness of the available data sources for surveillance. The following specific aims define the structure of the research: 1. To investigate completeness of ascertainment of work-related fatal injuries using records from the B.C. Coroner's Office and the Workers Compensation Board of B.C., and also to assess the accuracy of the item "place of injury" on death certificates in identifying work-related deaths among sawmill workers. Capture rates for each data source are calculated. Capture-recapture techniques for two independent data sources are applied to estimate the degree of underreporting. The sensitivity and specificity of the designation "place of injury" on death certificates are estimated. The implications of these findings for injury surveillance are discussed. 2. To examine all work-related fatal injuries among the sawmill workers in terms of a) characterizing the pattern and trend of fatal injuries at work b) estimating the risk of fatal injuries at work c) assessing the relationship between selected variables and the risk of fatal injuries at work d) identifying the occupational categories at highest risk of fatal injuries at work Both descriptive and analytical methods are applied. Frequency distributions are used to characterize the pattern of fatalities according to factors related to the worker, the physical environment and the socio-cultural environment. Mortality rates are estimated to illustrate trends from 1950 through 1990. 3 Poisson regression modeling is used to assess the relationship of selected variables and the risk of fatal injury at work, and also to identify the occupational categories at highest risk of fatal injury at work. 3. To examine mortality among the sawmill workers from selected external causes in terms of a) characterizing the pattern and trend of mortality from selected external causes b) estimating the risk of mortality from selected external causes c) comparing injury mortality from external causes in the cohort with that of the male population of British Columbia. d) assessing the association of selected variables with the risk of mortality from selected external causes. e) comparing injury mortality from selected external causes among subcohorts defined according to occupational categories. Both descriptive and analytical methods are applied. Frequency distributions are used to characterize the pattern of fatalities according to factors related to the worker, the physical environment and the socio-cultural environment. Mortality rates are estimated to illustrate trends from 1950 through 1990. Standardized Mortality Ratios (SMRs) are calculated to compare the injury mortality of cohort members with that of the B.C. male population. Poisson regression modeling is used to assess the association of selected variables with the risk of mortality from selected external causes, and also to compare fatality rates among occupational categories. 4 CHAPTER TWO 2.0 Background The background for this study is presented according to the following themes: • Injury research: concepts and issues • Development of the epidemiological model of injuries • Canadian injury facts • Scientific literature about injuries among sawmill workers The first theme is concerned with the use and misuse of the term "accident" as synonym for "injury". The detrimental connotations and consequences of using this term particularly hindering the advance of knowledge in the area of injury control, are explained. The second theme addresses the evolution of knowledge into the causes of injury. It covers the successive steps from the pioneering work of DeHaven, Gordon and Gibson to the adoption of an epidemiological model and the development of the Haddon matrix. The third theme consists of a summary of statistics regarding the magnitude of the injury problem in Canada. Mortality and morbidity data are presented together with the estimated costs of injury in Canada. The fourth theme summarizes epidemiological studies of injury among sawmill workers. 2.1 Injury Research: Concepts and issues According to the Webster's Dictionary, an accident is " a mishap, a chance event commonly involving catastrophe, suffering or damage". The definition implies that such events are random and can not be foreseen or expected. The word accident has also the connotation of lack of intent. Furthermore, the etymological history of the word reveals that accidents have 5 also been interpreted as "acts of God" for centuries and this fatalistic view persists to this day. Another common understanding is that accidents are a result of inattention or in some cases deliberate risk taking (Loimer and Guarnieri, 1996; Rivara, 1982; Pless, 1989; Robertson, 1992). The term accident has a number of misleading connotations that underscore the inappropriateness of its use to describe injury events. First, randomness signifies that every person has an equal chance of being affected. For many years it has been known that injuries occur more frequently among certain groups than others and that there is a differential distribution of injuries according to age, sex, time and place as well as other demographic factors (Baker, et al., 1992). In fact, insurance companies use this knowledge to estimate the likelihood of an injury event for specific populations and circumstances and set their rates accordingly (Waller, 1985). Injuries are non-random events and they do not occur by chance. Most injuries are predictable and preventable. Second, accidents are viewed as acts of God or fate, sometimes assumed to be the consequence of human error or behaviour. This misconception has led to a focus on the notion of fault; assuming that accidents are necessarily the result of carelessness, unawareness or ignorance on the part of the individual suffering the injury. This attention on the person involved has resulted in blaming the victim and has hindered the development of preventive countermeasures by ignoring the full range of factors that contribute to the 6 occurrence and severity of injuries. To believe that the reduction of the injury toll depends primarily on changing individual behaviour is far too narrow a focus considering the present knowledge about causation (Waller, 1985; Robertson, 1983; Kraus and Robertson, 1992). For these reasons the term accident will not be used here. The term injury (from the latin, injuria, harm or insult) will be used instead to denote an event in which 1) damage occurs over a short period of time (i.e. acute exposure), and 2) the damage results either from one of the forms of physical energy in the environment (kinetic, chemical, thermal, electrical, ionizing radiation) or because normal body mechanisms for using such energy are blocked by external causes (such as by drowning). The term injury is also chosen because it allows to focus on the damage or harm rather than the incident, and apply an epidemiologic approach to the study of injuries (Rivara, 1982; Waller, 1985). Both unintentional (e.g. falls, machinery-related, transportation-related) and intentional (e.g. suicide, homicide) fatal injuries are included in this study. 2.2 The development of the epidemiological model of injuries 2.2.1 Pioneering work into the causes of injuries Although injuries are as old as mankind and have always been at least endemic, it was not until the middle of the 20 t h century that researchers started to gain insight into the causes of injuries (Haddon, et al., 1964). Among the pioneers in this field is Hugh De Haven, an engineer who in 1942 published a paper that began a conceptual shift in the understanding of the determinants of injury severity. His study of people who survived falls from great heights 7 showed that the injury severity in falls is mainly determined by the characteristics of the environment, including the surface of impact and what is beneath that surface. It also illustrated that, under appropriate conditions, the normal human body can tolerate very high deceleration forces without significant injury (DeHaven, 1942; National Committee on Injury Prevention and Control, 1989). In 1949, John E. Gordon, proposed that injuries behaved like infectious diseases and were characterized by epidemic episodes, seasonal variation, long-term trends, and geographic, socioeconomic, and rural-urban distributions. He suggested that in a broad sense injuries are in many ways, a problem in medical ecology. Most important, injuries could be studied like classical infectious diseases recognizing that they are product of the interaction of factors related to the host, the agent and the environment (Gordon, 1949; Haddon, et al., 1964). It was not until 1961, that an experimental psychologist, James J. Gibson identified the causes of injuries. He wrote: "Man.. .responds to the flux of energies which surround him -gravitational and mechanical, radiant, thermal, and chemical.. .Injuries to a living organism can be produced only by some energy interchange. Consequently, a most effective way of classifying sources of energy is according to the forms of physical energy involved. The analysis can thus be exhaustive and conceptually clear. Physical energy is either mechanical, thermal, radiant, chemical, or electrical" (Gibson, 1961; Haddon, et a l , 1964). 8 At approximately the same time, William Haddon concluded as Gibson did, that the several kinds of energy are the necessary causes of injuries (Haddon, 1963). He also included the notion of "negative agents" for injuries resulting from the absence of elements such as oxygen or heat needed for normal health (i.e. drowning). Haddon expanded the model by identifying the vehicles and vectors of energy transmission and including the notion of human resistance to these energy forces (Haddon, 1970). The importance of these pioneers' work and others that followed their concepts, is the shift in injury prevention from a naive preoccupation with distributing educational pamphlets and posters towards modifying instead the environments in which injuries occur (Waller, 1987; Baker, et a l , 1992; Kraus and Robertson, 1992). 2.2.2 The epidemiological model Like any other health problem, injuries can be analyzed using the epidemiological model with the triad of components: the host, the agent (and mechanisms of transfer of the agent), and the environment. The agent: injury results from a transfer of energy to the human body at levels that exceed the body threshold. It is the nature and extent of the energy transfer which determines the type and severity of injury. Waller suggests that the occurrence of injury and its severity is determined by 1) the mass of energy liberated and the speed of that release, 2) the rate of transfer of that energy to the tissues (considering the amount of energy impacting the tissue in an specific time), and 3) the inherent characteristics of the tissue and the organism. (Waller, 1985). 9 The understanding of the mechanisms of energy transmission and the changes that this energy produces at the cellular, tissue or organ level is fundamental for the development of injury control strategies. Vehicles and vectors: energy in any of its forms can be carried by inanimate objects (vehicles) or living organisms (vectors) from its reservoir to the host. Thus, electrical wires are vehicles of electricity and moving objects, like cars, are vehicles of mechanical energy. Animals that produce an injury by tearing (i.e. dog bit) or crushing (i.e. horse kick) are vectors of mechanical energy. The identification of the vehicles of energy is useful since many preventive interventions can be directed against the vehicles and vectors rather than against the physical and chemical agents they carry. Host: the human body has certain resistance to energy impacting on it. However, there are limits beyond which the energy received by the host cannot be tolerated or neutralized. In addition, there are variations in resistance to the agents of injury which affect the occurrence and severity of injuries. The susceptibility to injury agents is greatly affected by personal factors that can be related to biological features (e.g. physiological changes due to aging or pathological changes due to diseases) or conditions due to peculiar temporary circumstances (e.g. fatigue, alcohol). The importance of injury threshold in bodily mechanical energy exchanges was demonstrated by De Haven's studies (De Haven, 1942; Waller, 1985). He realized that these injury thresholds and the conditions of impact (and not velocity of impact per se) are the determinants of the injury outcome. 2.2.3 The Haddon matrix Further to the notion of agent, host and environment, Haddon proposed a two dimensional matrix to describe all factors related to injury events and the means to control their occurrence or severity (Haddon, 1980). The first dimension analyzes the injury factors in a sequential fashion related to the timing of the event: pre-event, event and post-event. The second dimension adds the enumeration in each phase of the factors in relation to: the host (human), vehicle (equipment), and environment (physical and socio-cultural). It was developed as a tool to identify appropriate prevention strategies and analyze resource allocation. Its greatest contribution is that it shows that actions from a variety of disciplines are needed to effectively reduce the occurrence and severity of injuries. A n example of the use of this matrix to analyze work-related injuries is shown below. Phases Factors Host (Human) Vehicle (Equipment) Physical Environment Socio-cultural environment Pre-event Age. Start date at mill. Start date at job. Regular task. Alcohol/drugs tests. Training received. Proper tool for task. Design of tool. Proper maintenance of equipment. Equipment failure. Ambient conditions. Housekeeping. Supervisor present. Training required/provided. Event Activity during event. Protective equipment used. Type of energy. Source of energy. Mechanism of injury. Secondary sources of injury. Worksite location. Isolation. Housekeeping. Protective equipment provided. Supervision required. Violation of safety regulations. Post-event Body part injured. Nature of injury. Other contributory causes of death. Failure of safety devices. First Aid available. Good emergency communication. First A id service required. Accident investigation required. .11 In the pre-event phase attention is given to the characteristics of each factor that will increase the likelihood of an incident to occur. For example, a sensory impairment like hearing loss (human factor); or the operation of a tool that has not received proper maintenance (vehicle factor); or extreme temperature conditions (physical environment factor); or the lack of safety policy regarding provision of training for the worker (sociocultural environment factor), can contribute to the occurrence of an injury. During the event phase the interest is placed on those features of each factor that will increase the probability of having an injury once the incident has happened (e.g. use of protective equipment, guards of mobile parts, housekeeping conditions, regulations regarding supervision, etc.). Finally in the post-event phase it is important to list the peculiarities in each factor that will worsen the injury sequelae (e.g. age, failure of safety devices, availability of first aid, regulations requiring accident investigation, etc.). 2.3 Canadian injury facts 2.3.1 Data Sources Although injuries are a major cause of morbidity and mortality, they have not received the attention given to many other diseases of comparable or lesser public health importance (Pless,1989). One of the major challenges faced by safety professionals concerned with injury control is the lack of a comprehensive surveillance system to describe the magnitude of the injury problem in terms of who is more likely to suffer an injury, and where, when, and how injuries occur. While data on injury mortality is available from a couple of sources, morbidity data is very limited. It has been said that for every death, there are at least 45 hospitalized injuries, and approximately 1300 others that are sufficiently serious so as to 12 require treatment in emergency departments. However, these figures alone are insufficient to illustrate the real magnitude and nature of the injury problem (Saunders and Stewart, 1991; Pless, 1989). Data on fatal injuries can be obtained from a couple of sources that systematically collect information on deaths in the population: 1) death certificates, and 2) Coroner's records. Death certificates contain basic demographic information and the cause of death. While these certificates are a good source for counting fatalities resulting from broadly defined causes of injury, they do not provide information on the circumstances of the event. A l l deaths resulting from injuries are classified according to the International Classification of Diseases External cause of death codes (E800-E999). This group of causes of deaths comprises two subcategories defined as unintentional injuries (E800-E949), such as falls, drowning, motor vehicle related, etc. and intentional or violent injuries (E950-E999), such as suicide, homicide, legal execution and war. Data on the circumstances of a fatal injury can be retrieved from the Coroner's records. The Chief Coroner is responsible for the investigation of the facts related to sudden, and unexpected deaths, like injury deaths. Other causes requiring coroners investigations include deaths in a correctional institution or police custody, and deaths during pregnancy. The coroner may decide to hold an Inquiry (without a jury) or an Inquest before a jury and with presentations by witnesses, of the circumstances of the fatality. Coroners and coroners' juries may make recommendations to agencies and individuals involved, to prevent similar loss of 13 life in the future. In order to investigate the circumstances of a sudden death, the Coroner gathers information from several sources including pathologists, police, workers' compensation board inspectors, and any other professional considered relevant to clarify the sequence of events that led to a fatal injury. Although these records are very useful for epidemiological research, the lack of a standardized reporting format results in poor consistency from one report to the next and the level of detail in the narrative depends on the coroner in charge of the investigation. Data on specific injuries like work-related injury deaths are collected by the Workers' Compensation Boards. These files contain detailed descriptions of injuries that have occurred at the workplace and required compensation for medical treatment, disability or death. Severe injuries and fatalities are investigated by field officers from the WCB with the purpose of determining the cause of the incident, identifying unsafe conditions (i.e. violations of safety regulations) and developing recommendations for corrective action to prevent future injuries. The fatality investigation report has a standardized format with very comprehensive information on the pre-event and event phases of the incident. While the use of these records may result in an undercounting of fatal cases (e.g. non-universal coverage, incomplete workers report, and count of adjudicated claims only), the database is a very useful resource for surveillance of work-related injuries (Spiegel and Yassi, 1991). 2.3.2 Injury Mortality Injuries are the leading cause of both death and hospitalizations up to age 45. For the whole population, injuries are the fourth cause of death after cardiovascular diseases, cancer and 14 respiratory diseases (Riley and Paddon, 1990). Generally, injury frequency is higher for men in all age groups. In 1985, the ratio of male to female fatality rates was highest in the two age groups between 15 and 44 years. Specifically, the highest male/female ratio was observed in the age group 25-44. (Figures 2.1 and 2.2). Fig. 2.1 Percentage of deaths and hospitalizations due to unintentional injuries by age groups, Canadian males, 1985 35 30 & 2 5 | 20 a is CU o- io Under 5 5 14 15 24 25 44 45 64 65+ H Deaths I Hospitalizations Fig. 2.2 Percentage of deaths and hospitalizations due to unintentional injuries by age group, Canadian females, 1985 50 40 Ml « 30 C V u L . 20 <U 0-10 0 Under 5 5 14 15 24 25 44 45 64 65+ B Deaths U Hospitalizations Based on data from Riley and Paddon (1990) The effect of injuries in early mortality is best illustrated by a measure of the years of potential life loss (PYLL). Unintentional injuries rank first in terms of P Y L L before age 65 15 (MacWilliam and Fortier, 1987). In 1985 injuries resulted in a total of 219,641 (22.3%) P Y L L followed by cancer (19%) and cardiovascular diseases (14%). (Figure 2.3). Fig. 2.3 Years of life lost before age 65 for selected causes of death, Canada, 1985 250,000 -r 200,000 -150,000 -o XI 100,000 -s Z 50,000 -0 Unintentional Cancer Cardiovascular Diseases Suicide and Homicide Congenital anomalies Respiratory Diseases injuries Cause of death Based on data from Mac William and Fortier (1987) In terms of historical trends, during the period 1951 to 1986, the age standardized mortality rates of unintentional injuries have decreased. While no major change was observed during the 1960's, from 1971 to 1986 the age standardized mortality rate decreased by 44% for men (from 78 to 44 per 100,000 population) and by 39% for women (from 33 to 20 per 100,000 population). The rates for men during the period 1951-1986 were always at least twice the rate for women, the ratio of male to female ranging from 2.2 to 2.6 (Riley and Paddon, 1990). The decrease in injury fatality rates reflects the decrease of fatal motor vehicle crashes. Seat belt use and also greater public awareness of the importance of safety measures in general may have contributed to this decline in fatality rates (Wilkins,1989). 16 Suicide rates were lowest during the 1940's (13.8 per 100,000 in 1945) and stabilized during the period from 1946 to 1964. They increased sharply during the 1960's and 1970's reaching a maximum of 28.8 per 100,000 in 1983 (Canada Health and Welfare, 1987). 2.3.3 Injury Morbidity In 1987, approximately 4 million Canadians, aged 15 and over, reported that they had an injury. Motor vehicle injuries were the most frequent (one in three incidents), followed by sports injuries (23%), injuries at work (21%), and at home (13%). The highest injury rate among men was 107 per 100,000 population for motor vehicle related injuries (Millar and Adams, 1991). Approximately 80% of all incidents resulted in personal injury, and almost half of these resulted in medical attention in a hospital (2.3 million injury events). Overall, 79% of incidents resulted in activity-loss days, and 36% involved bed-disability days. In 1987, approximately 51 million activity-loss days, nine million disability days and two million hospital days were attributable to injuries. Apart from the hospital utilization costs, about 13 % of incidents involved out-of-pocket expenses. 2.3.4 The economic cost of injuries The economic impact of injury in Canada, both in terms of direct (e.g. hospital care, professional services, medications, research, pensions and benefits) and indirect costs (e.g. loss of future income and value of productivity lost due to chronic disability) was estimated at $11 billion in 1986. This amount was the second highest after cardiovascular disease costs 17 and was followed by cancer costs. These three conditions accounted for almost 50% of the total $97.2 billion estimated cost of illness in 1986 (Wigle, et al., 1991). Pensions and benefits (including Canada/Quebec Pension Plans, Unemployment Insurance and Workers' Compensation expenditures) made up 57% of the total direct costs of injuries which is a much greater proportion than for any other disease category. Interestingly, injuries' share of total research costs was barely 0.3% (Figure 2.4). Research in the area of endocrine and related diseases (19%), cancer (14.2%) and cardiovascular diseases(13.9), account for almost 50% of the total research costs. 60 50 „ 40 R 1 30 U L. tU ^ 20 10 0 Fig 2.4 Costs of pension and benefits, and research costs share for selected illnesses, Canada, 1986 • H Cardiovascular diseases Injuries Cancer Endocrine and related diseases • % of direct costs assigned for pension and benefits • % of research costs Based on data from Wigle, et al., (1991) The authors of the report noticed that the impact of health problems in economic terms is always an underestimate of the real societal cost because effects such as pain, grief, disfigurement and disability cannot be measured in these terms. Injury events may result in 18 economic dependence, social isolation, loss of opportunities for education as well as other unwanted changes in life plans (Rice et al., 1985). 2.3.5 Overview of risk patterns for fatal injuries in British Columbia Sex In B.C. the age-standardized injury mortality rate for men was 53.1 per 100,000 population in 1987. At all ages men are more likely to suffer fatal injuries than women. The ratio male/female varied from 1.2 times greater for adults aged 85+ to 4.1 times greater for those aged 20-24. The unintentional injury type with the greatest male/female differential was drowning with males having 5.0 times more deaths than females (B.C. Vital Statistics, 1997). The average suicide rate for B.C. males for 1984-1992 was 20.9 per 100,000 male population and 5.8 per 100,000 female population. The largest proportion of male suicides occurred between the ages of 20 and 29 (22.8%), followed by men aged 30-39 (21.6%) (Tuk and Macdonald, 1995). Age Data from B.C. (1987-1995) show that the elderly (75 and over) had the highest mortality rate from unintentional injuries, followed by young adults, aged 15-24. The lowest rates were among children under age 15 (B.C. Vital Statistics, 1997). These figures show that despite of being the 5 t h cause of death among individuals aged 85 and more, injuries occur more frequently among the elderly than among the rest of the population. During the period 1985-1993 the highest age specific suicide rates were in the 80+ age group. The lowest rates were in the 10-19 year age group (Tuk and Macdonald, 1995). Type of injury The six leading causes of death from unintentional injuries show a slightly different order among men than among women.(Figure 2.5). Drowning is the third cause of death for men but the fifth for women. Falls were the second leading cause of death for both but the proportion of fatal falls was greater among women in the age group 64 and over. Fig. 2.5 Percentage of leading causes of injury death by sex, Canada, 1985 50 T . 45 -I I 35 Motor vehicle Falls Drowning Fires Suffocation Poisoning H Males 8 Females Based on data from Riley and Paddon (1990) Suicide is the 8 most frequent cause of death among British Columbians. Its importance as cause of death varies at different stages of life. It is the second cause of death for young adults aged 15-24, and the third for the groups aged 25-44 and 45-64. The largest proportion of male suicides involved the use of guns. On the other hand, the largest proportion of female suicides was by drug overdose. 20 The highest ratio of observed to expected suicide was among the unemployed labour force. High ratios for suicide, homicide and gun deaths, were also observed among blue collar workers in the primary industry areas of forestry, fishing, hunting/trapping and construction (Tuk and Macdonald, 1995). Regional variation Compared with other provinces, British Columbia has the highest rate of unintentional injuries. (Statistics Canada, 1995b). The age standardized injury rates for men and women in the Atlantic region, Quebec and Ontario, were lower than the national average. In contrast, the rates in the Prairie Region and British Columbia were higher than the national average. For example, in the Atlantic region, the motor vehicle injury rates was 47 per 1,000 compared with 85 per 1,000 in Ontario and 130 per 1,000 in British Columbia (Millar and Adams, 1991). Within British Columbia, people living in the northern interior and rural areas of the province are at greater risk of injuries (B.C. Provincial Health Officer's Annual Report, 1996). Coincidentally, using an index of socio-economic status based on three variables (i.e. percent of population on income assistance, percent of population aged 25-34 with less than high school graduation, and unemployment rate), it was found that the regions of the province with the worst scores (i.e. lowest socioeconomic status) were the northern interior, south central, and Cariboo. These regions had the highest mortality rates from unintentional injuries. Other factors likely to affect this distribution are that more miles are driven in rural 21 areas, the roads may be in poor conditions, driving practices may differ (greater speeds, lower use of seat belts). There are regional differences regarding the occurrence of intentional deaths as well. For example, statistically significantly higher standardized mortality ratios of violent deaths were reported in the northern and rural areas of the province (observed cases higher than expected number, calculated based on B.C specific mortality rates). This pattern coincides with the highest involvement of guns as a means of death in those regions. Urban areas (except for Vancouver and New Westminster) had generally lower SMRs. 2.4 Scientific literature about sawmill workers 2.4.1 The work environment The physical environment of a sawmill In a survey of approximately 700 workers from 15 Swedish sawmills, noise was identified as the most disturbing environmental problem (Ager, 1974). The level of noise was measured at different workstations for 5 hours of daily exposure and it was concluded that 41% of the workers were exposed to noise levels between 85-95 dBA and 28 % to levels between 95-110 dBA. Therefore about 70% of the workers were exposed to noise levels that indicated the need for hearing protection. Apart from causing hearing loss, excessive noise was implicated as a factor for social isolation, increased stress and reduced performance. 22 Other environmental stressors identified by the Swedish sawmill workers were wood dust and other particulate matter, temperature extremes, and insufficient lighting. Approximately 2/3 of the workers complained about wood dust, particularly in large mills (Ager,1974). Physical workload Ager (1974) calculated the amount of energy consumed by sawmill workers in different mill jobs in order to ascertain their physical work load. Jobs with an energy consumption exceeding an average of 5.4 kcal/min during an 8 hour working day were classified as physically strenuous. Thirty one per cent of the jobs in small mills were physically strenuous compared to 4 percent in large mills. However, the proportion of workers experiencing their jobs as strenuous was about the same in large and small mills. The reason for this is that work in large mechanized mill implies more time in fixed postures because of the work station layout (e.g. position of pedals, etc.), thus producing muscle strain. Approximately 80% of the 700 workers interviewed performed their jobs standing or standing and walking. These workers found that the postures at workstations were unnatural and uncomfortable. Uncomfortable postures were reported with similar frequency by workers from large mills (highly mechanized) and small mills (more manual tasks). Psychosocial stressors Psychosocial conditions were also assessed in the Swedish survey. In terms of job satisfaction, it was found that the majority of the workers (70%) claimed that pay was the 23 only satisfaction obtained from work. A great number of workers described their work as too hectic and mentally strenuous. For most sawmill jobs, the psychological job content is determined by the features pertaining to an assembly line production: a) fast, machine-paced working rhythm, b) repetitiveness and monotony, and c) social isolation as a result of the above mentioned factors, combined with geographical distance between workers and high noise levels (Baneryd,1974; Nilsson, 1974). It was concluded that work in a sawmill is characterized by small demands for skill, short operational cycles, small task variation, restricted possibility to influence product quantity and quality, and restrained possibilities to communicate with fellow workers. The most negative occupations in terms of job content were sawyers, edgerman, sorters, stickers, trimmer-graders and packagers. 2.4.2 Work-related injuries Experience from Sweden Aminoff (1974) analyzed 330 injuries reported to the National Social Insurance Board by sawmill workers from 15 sawmills in Sweden during 1970 and 1971. The highest frequency of injuries was observed among maintenance workers followed by those working in the packaging, sawing and bundling areas. While the highest frequency of injuries was observed among young workers, the most severe injuries occurred to workers aged 50-59. According to the workers, injuries occurred because of: 1) deficiencies of technical equipment, 2) stress, and 3) hurry. 24 Larsson (1991) reviewed 352 claims filed with the Swedish insurance agency for severe injuries that occurred during 1988-1989 among sawyers and planers with the objective of identifying the most hazardous activities, exposures and equipment. The following findings were reported: 1) 30% of accidents were associated with some type of contact with machines (mostly sawing machines and conveyors); 2) four circumstances were identified as posing the greatest risks for injury: a) Getting caught in conveyor/in-feed roller, b) Hit by /against tool, c) Inadvertent action of belt or rollers, and d) Acute overexertion. Experience from Australia In Australia, Driscoll reviewed the coroner's records to obtain information on fatal injuries among forestry workers, including sawmill workers, for the period 1982-1984. Sixteen sawmill workers suffered fatal injuries at work. The activities at the time of the injury event were working with a bench saw (n=6; 38%), loading/unloading (n=5; 31%), and traffic related (n=4; 25%). A n injury fatality rate of 30 per 100,000 was reported. The denominator was estimated from the Employed Civilian Labour Force survey (Driscoll, et al., 1995). Experience from the United States O'Gara (1978) carried out a study of injuries and illnesses among workers from sawmills and planing mills in California in 1976. A total of 1,922 injuries and illnesses were sampled from all Employer's Reports of Occupational Injury completed during 1976. The results were expressed in terms of the proportion of injuries or illnesses that could have been prevented by compliance with safety and health orders, those that could have been prevented by better 25 safety and health training (but not by compliance with safety orders), and those that were not preventable by either of the two means. Interestingly, 32 percent of the injuries and illnesses were classified as preventable by compliance with safety orders. The remaining proportion of cases (66%) were considered as preventable by better safety training and only 1% were considered random occurrences, not preventable by either compliance with safety orders or training. A booklet produced by OSHA titled "Coping with Hazards in Sawmills" enumerates the workplace conditions considered to be related to the occurrence of injuries in sawmills. The authors included the four following factors: 1) Failure of foremen to properly instruct and manage the work for which they are responsible; 2) Poor training of workers in work methods and safety procedures; 3) Failure to enforce safety regulations; and 4) Excessive pressure for production (OSHA, 1976). Cooke & Blummenstock (1979) abstracted data from employers' reports of injuries filed with the Maine Workers Compensation Commission during 1972 with the purpose of identifying risk factors for work-related injuries. The results showed that injuries were more severe, as measured by days lost from work, in plants processing hardwood than those using softwood. Age was also associated with severity for the extreme groups. Younger (less than 24 years old) and older workers (more than 45 years old) had more severe injuries. The third factor associated with injury severity was "temporary assignment". 26 Punnett (1994) in a case-control study examined the relationship between environmental risk factors and work-related injury among sawmill workers in Maine. A total of 157 cases and 251 controls were interviewed to gather information on demographics, work history, and characteristics of tasks, equipment, tools, and other conditions of the job. The strongest risk factors associated with acute traumatic injuries were processing hard wood, high physical effort, one year or less on the job, and the lack of a lockout/tagout program in effect in the mill. Social support had a small protective effect. The distinctive working conditions that prevailed on injured workers were jobs in machine-paced workstations, exposure to dangerous work methods and materials, loud noise levels, fast work pace, high lifting demands, and frequent postural stress. Experience in Canada Laflamme and Cloutier (1989) investigated the typical scenarios where injuries occur most often. The study included analysis of 728 injuries reported in nine sawmills in Quebec in 1987. The authors were able to identify 10 scenarios of injury, 6 of them related to production, 3 including maintenance operations and 1 in relation to movement of workers from one place to the other. Most injuries occurred in the sawing section while workers were handling logs/planks. Regarding type of machine involved, accidents occurred more frequently at the sawing and conveying machines. An analysis of the relationship between hazardous conditions and injury rates in 61 sawmills in British Columbia carried out by Keith Mason from the Workers Compensation Board, showed that several hazards were significantly correlated with injury rates (Mason, 1973). 27 The regression model used injury rates as the dependent variable; the independent variables were 15 hazardous conditions (i.e. workplace conditions were rated by the W C B inspectors according to level of safety hazard). The relationship between hazardous conditions and injury rates is summarized in Table 2.1. Table 2.1 Hazardous conditions associated with high injury rates in B .C. sawmills Significant (*) Not significant Competence of workers Adequacy of tools Competence of supervisors Personal protective equipment Pressure of operation First aid Future prospects of business Fire protection Work practices Operating procedures Materials handling Warning signs Housekeeping Guarding Based on Mason (1973) * ordered according to relative weights in descending order 2.4.3 Non work-related fatal injuries among sawmill workers Two reports from Statistics Canada have identified sawmill workers as having an excess of mortality from injuries. Gallagher et al.(1986) in a study on occupational mortality in British Columbia during 1950-1978, reported a proportionate mortality ratio (PMR) of 119 for all unintentional injuries among sawyers. Tuk and Macdonald (1995) found that forestry workers represent the occupation with the highest suicide rate (11.9 per 10,000 population) together with mining (12 per 100,000). They found that the ratio of observed over expected for suicides was 5.8. (Observed=52 vs. expected=9). 28 CHAPTER THREE 3.0 Methodology 3.1 The study population and the original data base: The BC Sawmill Workers Cohort The study population analyzed in this investigation comprises all workers of the British Columbia Sawmill Workers Cohort (BCSWC). In 1988, researchers at the Department of Health Care & Epidemiology at the University of British Columbia, started an historical cohort study of B.C. sawmill workers, with the purpose of examining exposure to chorophenate wood preservatives and the risk of soft tissue sarcoma and non-Hodgkin's lymphoma. Fourteen sawmills from BC were included in the study. M i l l selection was based on the use of chlorophenate fungicides and completeness of job history records. Eleven mills that had used chlorophenate based fungicides and three mills where these fungicides had not been used were included in the study. Six mills were located on Vancouver Island, five on the south coast of BC mainland, and three in the interior of the province. A map of B.C showing the location of the study mills and their classification by region is provided in Appendix A . The degree of urbanization of these regions ranges from well established urban settings (e.g. Vancouver Lower Mainland) to isolated mill towns (e.g. Tahsis, Youbou, Mackenzie) where the only source of employment is the mill. The original study population consisted of 26,487 men who had been employed for at least one year in a study mill, between January 1, 1950 and December 31, 1985 ( Hertzman et al. 1997). For those employees who worked on a non-permanent basis, a total of 260 days of 29 cumulative service was required to complete a qualifying year. The original follow-up period extended to December 31, 1990. The number of workers in this study (n=26,411) differs slightly from the original report because after various data quality control checks for validation, 76 workers did not meet the requirements for cohort membership. At each mill data were abstracted from personnel records to construct two files. One with personal information included worker's name and date of birth; and another with the job history included hire date, job title, work department, and start and end dates for each new job held at the mill. Vital status was ascertained by means of probabilistic linkage to the British Columbia Death File (BCDF) and the Canadian Mortality Data Base (CMDB). Probabilistic linkage was dependent on the level of agreement on workers' full name and date of birth between the cohort demographic file and the two death datasets. The generalized iterative record linkage system was used to determine probability of agreement on each item matched (Howe and Lindsay, 1981). Secondary sources of ascertainment were Pension plans, B C motor vehicle records and the income tax files. As a result of these procedures, vital status was determined for 85.7% of the cohort members. The steps followed to ensure completeness of ascertainment were thorough. First, with the linkage to the C M D B an additional 4.4% of deaths, from workers who died outside the province, was obtained. Mortality ascertainment through the C M D B has been estimated to be as high as 97.6% for deaths in Canada (Schnatter, et al., 1990). Secondly, the information from income tax files added an extra 16% 30 of workers who were known to be alive. For these reasons, it can be stated that the probability of any of the subjects with unknown vital status being dead is very low. The BC Division of Vital Statistics and the Canadian Mortality Data Base provided information on death date and cause of death from the death certificates for the period 1950-1990. Underlying causes of death were coded according to the International Classification of Diseases (ICD) codes in effect at the year when death occurred. Therefore, codes for the 6 t h to the 9 t h revision of the ICD were utilized. For this study on injury deaths among sawmill workers, all deaths coded as due to external causes (codes E800-E999, ICD 9 t h Revision) that occurred during the period 1950-1990 identified by the provincial death file were selected. The data base constructed for the BC sawmill study constitutes a unique source of information for studying health outcomes in a cohort of blue collar workers. This is to our knowledge the largest assembled cohort of sawmill workers and has a follow-up period of over 40 years. Teschke et al.(1998 ) examined the benefits in terms of efficiency regarding savings in time and cost, that this type of data set offers for new investigations on this workforce. The paper also includes a summary of the work either completed or in progress based on this cohort database. 31 3.2 Data sources 3.2.1 B.C. Coroners Service Information on the circumstances prior to or concurrent with the fatal incident was obtained from the B.C. Coroners Service (BCCS) records. Under the Coroner's Act of British Columbia all sudden, unexpected deaths as well as those resulting from violence, negligence or during pregnancy or from any cause other than disease or in a correctional institution shall immediately be reported to the Coroner. A coroner's investigation may proceed as an inquiry or an inquest. A n inquiry comprises a full investigation and formal report of the coroner's findings. An inquest is an independent public hearing involving a jury and the examination of witnesses under oath, required when deaths occur in prison or police custody. Inquests are also held at the coroner's discretion i f it serves the public interest, when there is a need to clarify evidence, or to address family concerns. The jury may propose recommendations for preventing the occurrence of future deaths based on findings from the case analyzed, but is not allowed to assign legal responsibility. A coroner's inquiry or inquest determines the identity of the deceased and clarifies the facts about the fatal incident. The coroner's files contain the investigation reports and may include autopsy reports, police reports, Workers' Compensation Board reports and other information relevant to the fatal injury. A list with the names of the cohort members who died from external causes between 1950 and 1990 was produced. This roll was matched to the Coroner's records to retrieve the relevant investigation reports. 32 Data on factors related to the individual, the environment and the injury event abstracted from the BCCS files were recorded on forms designed specially for this study (Appendix B). 3.2.2 Workers' Compensation Board of B.C. According to the WCB of B.C. (BCWCB) act, all employers covered by this act are required to report immediately to the agency the death of a worker where the death is or is claimed to have resulted out of and in the course of employment. An officer of the board may investigate the fatality with the purpose of determining the cause of the incident, identifying unsafe conditions which contributed to the incident and developing recommendations to prevent reoccurrence of similar fatal injuries. To confirm work-relatedness of a death, a list of all accepted claims for fatalities in the sawmill industry from 1950 to 1990 was obtained from the WCB including the worker's name, the date of injury and the company name, size and location. 3.2.3 B.C. Division of Vital Statistics The Division of Vital Statistics of B.C. (BCDVS) provided the underlying and immediate causes of death codes pertaining to cohort sawmill workers who died during the period 1950-1990. Additionally, for all deaths due to external causes (E800-E999) among the cohort members from 1984 to 1990, the BCDVS cross-checked the item "place of injury" on the death certificate to assess its accuracy in identifying fatalities that occurred at a workplace. 33 3.3 Exposure assessment Exposure categories were defined based on job titles and work areas. In order to meet the objectives of this study, a new grouping of the original job titles was needed. The original job history file contained approximately 4,500 unique job titles. The first compilation was based on a thorough review of potential for exposure to chlorophenate fungicides and as a result, approximately 800 job titles were identified which were useful for the original study analyses of chlorophenol's association with cancer outcomes. Subsequently, through a series of consecutive steps, the job history file was shortened to a roster of 86 job titles. Briefly, the data reduction procedure comprised the following operations: a) Thorough spelling checking. Spelling differences on equivalent job titles, were resolved by a researcher with ample expertise in the sawmill industry. b) Job titles standardization. Based on similarity of tasks, job titles were standardized across the fourteen mills. c) Consultation with knowledgeable mill personnel. A group of retired sawmill workers with knowledge of sawmill operations and the historical changes that have occurred in the mills since the 1950's provided information to resolve unclassified jobs. In order to investigate whether certain jobs or work areas were associated with increased injury mortality, workers were classified into broad groups defined on the basis of similarity of 1) tasks, 2) process areas, and 3) skills. Details of each one of these classification schemes are provided in the methods sections of subsequent chapters. The complete list of job titles included in each of these three schemes is presented in Appendix C. 34 3.4 Analytical methodology Crude rates per 100,000 person-years were calculated to illustrate the absolute risk of fatal injuries among sawmill workers. Comparison with external reference population- SMR Observed deaths in the cohort were compared with expected numbers based on rates for the province of BC male population for 1950-1990, via the standardized mortality ratios (SMRs). The numerator of the SMR is the sum of observed deaths across all strata and the denominator is the sum of expected deaths across all strata. The expected numbers of deaths for each cause of death were calculated by multiplying the age and calendar year-specific person-years of observation by the age and calendar year-specific provincial death rate (Berry, 1983). Also two-sided 95% confidence intervals were calculated under the assumption that the observed deaths have a Poisson distribution. For each worker, total person-years at risk comprises the period between the date of entry in the cohort and the death date or last day of follow-up whichever occurred first. Person-years contribution for those workers with unknown vital status at the end of the study, can be handled in different ways (Checkoway, et al., 1989) and each approach will yield different mortality rates. For example, i f person-years are counted up until the day of last contact, the mortality rates will be spuriously inflated. Inversely, mortality rates will be underestimated if person-years are counted until the end of the study. Both SMRs counting person-years up to the last day of contact and up to the end of the study are presented to illustrate both the least and the most conservative effect estimates. 35 SMR analysis is used in cohort studies to describe the mortality experience of the workforce. However, it is susceptible to several drawbacks that may limit its validity as a measure of exposure-response associations (Gaffey, 1976; Breslow, 1990). One criticism is that an SMR provides a summary measure of effect, and as such it gives no information on exposure effects that may be different among subgroups of the study population (Chiazze, 1976). Moreover, because the external general population is composed of people too i l l to work as well as employed individuals, the estimated SMR will always underestimate any true increased or decreased risk (Hennekens and Buring, 1987). Another limitation is that SMRs for sub-cohorts are not comparable because the distribution of confounders (e.g. age or calendar year) may be different among the groups being compared. Therefore, for a valid comparison of SMRs, the assumption required is that the age/calendar year specific rates in each stratum are a constant multiple of the external standard rates (Miettinen, 1972; Breslow and Day, 1987). Taking into consideration the limitations discussed above, SMRs are presented as the first analysis of this cohort, mainly to compare the injury mortality risk among sawmill workers with another group with stable estimates and to remove the confounding effect due to structural differences between the groups being compared. 36 Comparison with internal reference population- Poisson regression modeling The use of an internal reference group removes the differences in employment and socioeconomic status associated with the use of the general population and thus reduces the bias due to the healthy worker effect (Wen and Tsai, 1982). Mortality rates were compared among subcohorts defined on the basis of occupational categories using regression analysis. Generally, mathematical modeling provides an efficient way to describe the joint effects of multiple variables on mortality rates (Breslow, 1990; Selvin, 1995; Armstrong, 1988). Particularly, in occupational cohort studies, Poisson regression is used to assess the effect of time-related variables. The death rates are modeled on a logarithmic scale as a linear function of the covariates of interest. Thus, the rate is a product of the effects related to each independent variable (Frome and Checkoway, 1985; Pearce, et al., 1988). Causes of death selected for internal comparison analysis were those with 20 or more deaths. Thus, only the following will be considered for Poisson regression analysis: all work-related injuries, motor vehicle crashes, and suicide. The statistical packages used for SMR and Poisson regression analyses were Life Table Analysis System (LTAS) version 1.0.c (Cassinelli, et al., 2000) and EGRET version 0.26.06 (SERC, 1993). Capture-Recapture To investigate the completeness of ascertainment of work-related deaths from WCB data, capture recapture methods were utilized (Hook and Regal, 1992). This technique permits 37 calculation of an unbiased estimate of the degree of underreporting but relies heavily on the assumption of independence of the sources. Cases of work-related fatal injuries obtained from WCB records were cross-checked with the number of cases identified by the Coroner's Office records. In order to determine the accuracy of the denomination "place of injury" on death certificates, sensitivity, specificity and predictive values (positive and negative) were calculated to determine the usefulness of the item in identifying fatal injuries occurring at a workplace. 38 CHAPTER FOUR 4.0 Completeness of ascertainment of work-related injury fatalities: implications for surveillance 4.1 Introduction In Canada, during the last decade there has been increasing recognition from the public health sector of the importance of injury as a major public health problem. Approximately 21% (1.1 million) of all injuries among Canadians aged 15 and over, occur at work (Millar and Adams, 1991). Moreover, almost two out of three (65%) work-related injuries result in hospital care. A worker suffers a compensable work-related injury every seven seconds and one worker is killed every two hours of each working day (Labour Canada, 1991). While more than one million work-related injuries were reported to the General Social Survey in 1987 (Millar and Adams, 1991), the total numbers of time-loss claims accepted by the Workers' Compensation Boards across Canada was 602,531. These claims represented injuries severe enough to require time off work to recover but do not count the claims for non severe injuries for which only medical expenses are paid. The number of work-related injuries reported by the National Work Injury data base is much less than reported by the General Social Survey. This was not surprising because data from the Workers' Compensation Boards represent only adjudicated claims. For example, Labour Canada reported that of 1090 fatalities reported to the WCBs across the country, only 796 received compensation (Labour Canada, 1989). Furthermore, there are sectors of the labour force which are not covered by Workers' Compensation. Millar and Adams concluded that 39 given the magnitude of the social and economic losses resulting from work-related injury, it is necessary to develop an improved system of data collection and analysis to provide the basic knowledge for intervention strategies. In 1991, a symposium was held in Alberta to define the Year 2000 Injury Control Objectives for Canada. Five areas were targeted: home and community; occupational health and safety; sport and recreation; transportation; and violent and abusive behaviour. The working group on occupational health and safety set as the first objective a reduction of 30% in the national rate of work-related deaths (from 8.2 per 100,000 to 5.7 per 100,000). This group also stressed the difficulty in setting and evaluating objectives in this area where baseline data is lacking. Currently, statistics on work-related fatalities are based solely on the reports produced by the National Work Injury Statistics Program which compiles information from all Workers' Compensation Boards (Saunders and Stewart, 1991). Similarly, in 1996 the British Columbia Minister of Health's Injury Prevention Advisory Committee developed the Provincial Injury Prevention Plan for Children and Youth which identified the objectives, targets and strategies for the province. Although the target group was children and youth, injuries occurring at workplaces were also considered for those aged 15-24. The goal proposed for 2001 was to reduce the average number of industrial fatalities for youth ages 15-24 by 14 %. The WCB of B.C. reported that from 1990-1994 an average of 14 deaths received compensation annually in this age group (BC Ministry of Health, 1997). This was, as in the previous example, the only source for ascertaining number of deaths in the workplace and underlines the importance of accurate surveillance data for public health planning. 40 Stout and Bell (1991) in a review of 10 studies from various states in the U S A aimed at determining the best method for surveillance of occupational fatalities using different data sources including death certificates, Workers' Compensation reports, Medical Examiner records, and OSHA reports, concluded that death certificates are the best source, ascertaining as a single source an average of 81% of cases, followed by medical examiners records (61%) and workers compensation reports (57%). Incomplete statistics lead to inaccurate assumptions about the magnitude of health and safety problems in the workforce, hampering the planning and implementation of prevention strategies. Data sources with high sensitivity and representativeness are needed to establish an efficient surveillance system for injury fatalities. The objectives of this study were 1) to determine the completeness of ascertainment of sawmill work-related fatal injuries among all B.C. sawmill workers between 1984 and 1990, from two sources of information, the Workers' Compensation Board of B.C.(BCWCB) and the British Columbia Coroner's Service (BCCS) and the characteristics of the cases missed by each source; and 2) to assess the usefulness of using the item "place of injury" on death certificates to identify work-related fatal injuries at a workplace (sawmill or other industrial site) among members of the B.C. sawmill cohort during 1984-1990 using the BCCS data as the gold standard. 41 4.2 Methods 4.2.1 Datasources British Columbia Coroners Service The BC Coroners Act establishes that the coroner shall be notified of any death resulting from unnatural and unexpected causes. Coroner's investigations are carried out to determine the identity of the deceased and how, when and where the person died. Depending on the circumstances of the incident, these files may include policy investigation reports, autopsy reports, toxicological reports or WCB reports. By definition, all injury deaths are Coroner's cases. In order to investigate completeness of ascertainment of sawmill-related fatal injuries, the coroner's reports of deaths that occurred to all sawmill workers in B.C. between 1984 and 1990 were sought (Figure 4.1). Since 1984 the Coroners Service has computerized records. The retrieval of these reports was carried out by using the entry code "death at industrial site". Unlike other industries which have their own specific code (e.g. logging, construction, farming, mining, etc), sawmills are included in a group comprising various industrial sites. Subsequent manual sorting of every record was required to select only those where the place of death was a sawmill. This selection was not always easy because the only means to identify the type of industry was by the firm name recorded in the report. When the name was unclear, the industry was resolved using the Madison Lumber Directory which provides a listing of sawmills by province. Additionally we obtained the coroner's reports of all cases identified through the W C B list as work-related fatal injuries in the sawmill industry, for the same period 1984-1990. 42 Fig 4.1 Methodology for analyzing completeness of ascertainment of sawmill work-related fatalities Workers Compensation Board of BC We requested the WCB to provide a listing of all accepted fatal claims in the B.C. sawmill industry for the period 1984-1990. For each claim we asked for the deceased's name, and death date as well as information about the employer (i.e. company name, size and location). For this study, a sawmill-related injury death is defined as a fatal injury suffered by a worker in the course of work and while working in the sawmill industry. Deaths from occupational diseases, as well as non-traumatic events and suicides were excluded from this study. The B.C. Sawmill Workers Cohort f B C S W Q This cohort comprises 26,411 workers from 14 sawmills in the province of British Columbia. The cohort entry requirement was defined as: every male who worked in a study mill for at least one year between January 1, 1950 and December 31, 1985. Personal identifiers and job history information were abstracted from personnel records at each mill. The personal identifier file was linked using probabilistic linkage methods to the B.C. Death file and the Canadian Mortality Data Base (CMDB). Follow-up information was supplemented through linkages with other data sources including pension plans, provincial motor vehicle records, union records and the Canadian income tax file. After these linkages a total of 689 deaths due to "external causes" were identified among the cohort members between 1950 and 1990. Of these, only deaths that occurred during 1984-1990 were selected to assess the accuracy of "place of injury" on death certificates (Figure 4.2). 44 Fig 4.2 Methodology for assessing accuracy of "place of injury" on death certificates in identifying work-related fatalities at a sawmill or other work site B C Sawmill Cohort Database 14 Sawmills Cohort member deaths (1984-1990) Codes E800-E999 B C Division Vital Statistics Death certificates 1984-1990 1 T "Place of injury" codes cohort members 1984-1990 Linkage B C Coroners Database Injury deaths 1984-1990 Injury deaths with Coroner's rerjort Injury deaths at a work site Sensitivity and specificity analysis Division of Vital Statistics of B.C. Personnel from the Division of Vital Statistics agreed to participate in this pilot study to determine the accuracy of death certificates in identifying work-related fatal injuries. They were interested in validating the completeness and accuracy of the "place of injury" field on 45 the death certificates. This item is coded according to the categories provided in the ICD 9 t h Revision Manual (E849). According to these categories, work-related fatalities among our sawmill workers would be identified if the place of injury was coded as "Industrial place and premises" (subcategory #3). A list with all death certificate numbers for deaths that occurred among workers from the BCSWC study between 1984 and 1990 was sent to Vital Statistics. This roll was cross-checked with the provincial death file to add the injury place codes. The period 1984-1990 was selected because the BCCS has computerized files starting in 1984 and the information from these files was used as the gold standard to identify work-related fatal injuries. 4.2.2 Analysis Completeness of ascertainment of sawmill-related fatal injuries The primary analysis is descriptive, presenting the number of cases obtained from each source and the frequency distribution by age groups and mill size. Each injury fatality is then briefly described to illustrate the types of cases recorded by each institution. The number of work-related fatal injuries among all B.C. sawmill workers between 1984-1990 obtained from the W C B records were compared to the number obtained from the BCCS records. Capture-recapture methods were used to estimate the degree of underreporting and the total number of fatalities (Hook and Regal, 1992). The method of capture-recapture (CR) has been used primarily by ecologists to monitor the census of bird, fish and insect populations. Chiu et al. (1993) 46 describes the methodology as conceptually simple and provides the following example: " i f we want to estimate the number of fish in a pond, we would net fish in various parts of the pond, tag these fish and then release them back into the pond. The following day, after the tagged fish had mixed with the other fish, we would net again. Some of the fish from the first net (tagged) would be recaptured in the second net. It is the degree of overlap between nettings that allows the estimation of the total population of fish in the pond, even though we did not net all the fish". These methods have been used in demography to adjust for undercounting in population censuses (Himes and Logg, 1992). Within epidemiology capture-recapture techniques are increasingly being used to estimate degree of underreporting of morbidity or mortality from various datasources (Hook and Regal, 1992; Chiu et. al., 1993; Keifer, et al., 1996; Cormack, et a l , 2000). When data from only two sources are compared, a 2x2 table is constructed with the count of cases in each source as is shown below Table 4.1 Layout of data by source of ascertainment Cases reported in source A Yes No Cases reported in source B Yes a No c d=? The formula to obtain a nearly unbiased estimator for the number of unascertained cases is d= bc/(a+l) 47 Accuracy of "place of injury" to identify work-related fatal injuries We estimated sensitivity (the percentage of work-related fatal injuries coded as occurring at industrial premises on the death certificates), specificity (the percentage of non-work related fatal injuries coded with any other code, different from industrial premises, on the death certificate), and the predictive values positive and negative (Kelsey et al., 1986) of the "industrial premises" item to properly denote work-related and non-work related fatalities. The BCCS records were used as the gold standard for determining work-related fatal injuries. In order to calculate predictive value we calculated the prevalence of work-related deaths. We used the average mortality rate from fatal injuries among all B.C sawmill workers between 1984-1990. Fatality rates per 100,000 person-years were calculated for the sawmill industry between 1984 and 1990. Numerators are the sum of cases identified in this study using information from the BCSWC, the BCCS and the WCB. Denominator estimates for the period 1985-1990 were obtained from the report titled "Fatal claims in sawmills" published by the W C B of B.C.(Howard, 1996). For 1984, it was assumed that the number of person-years was the same as that for 1985. The number of person-years was calculated by the Statistical Department of the B C W C B from company payroll records and accounted for part-time workers. Having a person-years count for the denominator, improves the accuracy of the rates calculated because the risk of injury death is expressed in relation to hours worked. This avoids the underestimates that result when rates are expressed per 100,000 workers whereby part-time workers are counted as equivalent to full time workers (Conroy, 1993). Confidence Intervals were calculated using the Poisson distribution assumption (Breslow and Day, 1987). 48 4.3 Results 4.3.1 Completeness of ascertainment of work-related fatal injuries to B.C. sawmill workers A total of 18 fatal sawmill work-related injuries were identified from the BCCS files for the period 1984-1990. Of these, one was not reported by the WCB of BC. The WCB of BC on the other hand, identified a total of 21 fatal injuries of which 4 were not reported by the BCCS (Figure 4.3). Seventy seven percent of the fatalities were reported by both sources. Figure 4.3-Number of work-related fatalities among all B.C. sawmill workers, 1984-1990, identified by each data source Adding the number of cases reported by the two sources a total of 22 fatalities were identified among all B.C. sawmill workers. BCCS records alone permitted the identification of 81.8% of this total and the W C B records 95.4%. Applying capture-recapture technique the estimate of the unascertained cases was 0 (d= 0.22), thus implying complete ascertainment of work-related fatal injuries (Table 4.2). 49 Table 4.2. Sawmill work-related fatal injuries in all B.C. sawmill workers ascertained by each source, 1984-1990 Work-related fatalities identified by the BCCS Yes No Work-related fatalities Yes 17 4 identified by the BCWCB No 1 0 The scenarios of each fatal event are described in Table 4.3 with the purpose of characterizing the type of incidents missed by each source. Table 4.3. AH B.C. sawmill fatalities, 1984-1990, identified by each source: case description Description of injury event Case# Age Death WCB BCCS Year Record Record 1 38 1984 Yes Yes While sizing a board sticking out of the planer, a piece of lumber caught his watch & pulled his body into an outfeed chute where he was struck by 2 other timbers coming from the planer at 400 ft/min. 23 1984 Yes Yes He was standing below the end squeeze gate of a packer (1711 lb. weight) when the travel chain failed, the cylinder rod broke and a steel compressor fell on him. 43 1984 Yes Yes While the worker was attempting to clear the crossed pieces of lumber from the edger transfer table, two pieces of hemlock fell off the roll case and struck him. 56 1984 Yes Yes He and a co-worker were piling loads of side cuts for temporary storage. The band of one load in the main pile broke and three loads fell and buried him beneath. 57 1984 Yes No He was to scale a boom of logs. He was attempting to cross a double set of railway tracks. He paused on 1 set to permit a train to pass and apparently did not hear a train approaching on the track he was standing. 21 1985 Yes No He took the loader to a gas station for replacement of the rear tires. He was inflating the tire on a split rim wheel when it exploded, striking him in the head. 30 1985 Yes Yes He was trying to free a log jam when he fell through the opening between the roll case and the mill wall to the basement floor (20-30 feet). 50 Description of injury event Case# Age Death WCB BCCS Year Record Record 8 55 1986 Yes Yes He was supervising the installation of a gang saw (weight 4,000-5,000 lb.). The saw was chained to a forklift mast. This chain failed and the saw fell crushing the worker. 9 29 1986 Yes Yes As he was attempting to re-adjust the log & the carriage of a portable sawmill, he stepped onto a moving slab belt and was thrown backwards onto the head rig. 10 54 1987 Yes Yes He was operating a Caterpillar-Front End loader. The motor stalled and the loader started free wheeling backwards. The loader flipped over crushing the driver. 11 19 1988 Yes Yes He was cleaning the tail spool area around a planer conveyor. His shirt was caught in a tail spool, his right arm dragged under the tail spool and the left arm under tension roller. The arms crossed caused asphyxia. 12 37 1988 Yes Yes She was operating an automatic strapping machine. She reloaded the strap reel. For unknown reason she returned to the outfeed roll case and was caught between a load and a roller. 13 56 1988 Yes No While riding a bicycle at his worksite, he collided with a forklift. The worker's front wheel struck the right hand fork and he was catapulted through the air striking his head on a large steel upright of the forklift. 14 36 1988 Yes Yes He noticed that a chain had dropped off the double idler sprocket at the bottom of the unscrambler. He went under the machine that was turned off. Another fellow turned it on and the worker was trapped there. 15 52 1988 Yes Yes He climbed the stairways to the sorter trays to clear a problem. The stacker operator started the electrical motors by switches to clear the tray and the worker was caught between moving lumber and sorter frame. 16 49 1988 Yes Yes He was trying to solve a vibration problem in a Mark II Chip n Saw. By doing so he approached the motor too much and was caught & pulled into the revolving chipping head knives of the machine. 17 60 1988 Yes Yes He was going to the lunch room to retrieve his jacket as he was ready to go off shift. He came down the blind side of the breezeway & not hearing the straddle loader coming had stepped out in front of it. 18 32 1989 Yes Yes He was trying to release a pile of boards to the stacker. A bin of lumber dropped from the sorter and pinned the worker causing fatal crushing injuries. 51 Case# Age Death WCB BCCS Year Record Record Description of injury event 19 37 1989 Yes Yes He crossed a planer infeed ramp behind a parked lumber forklift. On seeing another forklift in the area he stopped right there and the parked forklift started backing up and knocked the worker down. 20 22 1989 Yes Yes He was clearing a feeder platform of lumber when he was either knocked or fell onto the platform and was carried into a shallow ditch and then buried by the lumber being fed along the platform. 21 38 1989 No Yes He was operating a small circular saw and because he needed to cut a slab off he also operated a power saw. In doing so he fell backwards into the rotating saw blade. 221 1 9 9 0 Y e s N o No record existed in the BCCS f No BCCS record. Other cases identified only by the WCB had records in the BCCS although they were not identified as work-related, but in this case no BCCS data were available to give a description of the injury event. The information provided under the last column, characterizes the type of death identified by each source. For the cases identified by the WCB only (n=4), the corresponding BCCS files were pulled afterwards to abstract the incident information. BCCS records were found for all these deaths except case #22. The BCCS failed to identify the following as fatal injuries in sawmills: case #5 which occurred within the sawmill but on the railroad tracks and may have been considered a non-sawmill worker; case #6 which occurred off the sawmill premises; case #13 which occurred in the mill yard; and case # 22 which had no record at all in the BCCS. The case missed from the B C W C B records was a fatal injury of a sawmill owner. This person was in the process of registering the company with the B C W C B . Table 4.4 shows the distribution of injury fatalities by worker's age at death, place of injury, and mill size. Case retrieval from the BCCS files was poorest for the oldest group of workers. The youngest worker in this series was 19 years old at the time of death and the oldest was 60 years old. While ascertainment of "on premises" fatal injuries differed only by 2 cases between data sources, "off premises" cases were only captured through the WCB roll. Table 4.4. Characteristics of fatal injuries at work, in BC sawmills 1984-1990, by source of ascertainment BCWCB (n=21)* BCCS (n=18) Age groups (years) <34 7 6 35-44 6 6 45-54 3 3 55+ 5 3 Unknown 1 0 Place of injury On premises 20 18 Off premises 1 0 Unknown 1 0 M i l l size"!->100 person-years 15 14 <20 person-years 6 3 Unknown 0 1 * 1 case from the W C B list had no record at BCCS, thus no data describing age or place of injury were available. f M i l l size categories provided by the B C W C B 53 The highest number of fatalities was reported from the largest companies. Seventeen out of the 22 cases (77%) occurred in mills with more than 100 person-years of employment. 4.3.2 Accuracy of "place of injury" to identify work-related fatal injuries Based on the review of the BCCS records there were 19 work-related fatalities among workers from the BCSWC during 1984-1990. This may seems strange since 18 fatalities were identified by the BCCS as occurring at a sawmill, among workers from all B.C. sawmills for the same 1984-1990 period. The reason for counting 19 work-related deaths among the cohort members is that it includes work-related fatalities that occurred while working either at a study mill (n=5) or at another industrial site (n=14). The sensitivity of the "place of injury" notation on the death certificates was 74% (Table 4.5). Twenty six percent of the deaths were incorrectly coded as not occurring at industrial premises. Table 4.5. Sensitivity and specificity of the B.C. death certificate "place of injury" item to identify work-related fatal injuries among cohort members, occurring at a sawmill or other work site, 1984-1990 BCCS Death Certificates Work-related Not work-related At industrial premises 14 0 Not at industrial premises 5 110 Total 19 110 Sensitivity= 74% Specificity=100% The specificity of death certificate for non-work-related fatal injuries was 100%. Total 14 115 129 54 The fatality rates per 100,000 person-years by year from 1984 through 1990 are presented in Table 4.6. Table 4.6. Fatality incidence in the BC sawmill industry between 1984-1990 Year Number Person-Yearst Fatality Incidence t of Deaths* (95% CI) 1984 5 31,000 16.12 (5.22-37.55) 1985 2 31,000 6.45 (0.78-23.28) 1986 2 28,000 7.14 (0.86-25.77) 1987 1 34,400 2.90 (0.07-16.15) 1988 7 36,200 19.33 (7.75-39.81) 1989 4 34,600 11.56 (3.14-29.59) 1990 1 33,600 2.97 (0.07-16.54) * Deaths obtained from BCCS and B C W C B pertaining to the whole sawmill industry. t Figures from Howard, 1996. For 1984 the number of person-years was assumed to be the same as in 1985. J deaths per 100,000 person-years Over the seven year period, the maximum fatality rate was reported in 1988 with 19.3 deaths per 100,000 person-years (95% C.I. 7.75- 39.81). However, because the annual number of fatalities is small, the calculated rates are very unstable as shown in the wide confidence intervals. Using the average fatality rate of 10 per 100,00 person-years calculated for the whole period 1984-1990, a sensitivity of 74 % and a specificity of 100 %, the positive predictive value of the field "place of injury" on the death certificate would be 100% and the negative predictive value would be 99.9%. 55 4.4 Discussion A total of 22 fatal work related injuries in all B.C. sawmills were identified between 1984 andl990. The proportion of cases identified by the WCB was approximately 14% higher than the BCCS . The capture-recapture analysis suggests that the ascertainment of work-related fatal injuries is complete. There are however some caveats in interpreting this result. First, the capture recapture methods relies on the assumption of independence of the sources. The estimates obtained from this method are sensitive to assumptions about independence of sources and these estimates have some uncertainty because of statistical fluctuation. In our study, the sources can not be considered totally independent. While not all BCCS records for work-related fatal injuries included a B C W C B report, the fact that some records contained a B C W C B report suggests that these agencies are interdependent. If the sources are not independent then the estimated underreporting is an underestimate of the true population value (Hook and Regal, 1995). Thus, the zero underreporting estimated in our study is most likely an underestimate because of the sources' interdependence. Second, the precision of the estimate increases proportionally to the number of overlapped cases. In our study the number of cases reported is small and the estimate may be unstable. More evidence of underreporting by these data sources, is provided with the description of the cases specifically excluded by each agency. The identification of the types of fatality missed by each source clearly reveals the fact that every time there is such a case it will be missed because of the procedures for data collection applied by each source. For example, the people not covered by the B C W C B will not be found on the compensation files, neither will fatally injured workers whose claims were not accepted. 56 Likewise i f a worker dies of a work-related fatal injury outside the premises of the company but while doing work for the company, he will most likely not be found in the BCCS i f the retrieval based on the entry code "on premises". Thus, from the beginning we are undercounting the total number of work-related fatal injuries by using either data source alone. The ascertainment rate for the B C W C B (95.4%) is higher than that reported in U.S. by Stout and Bell (1991) where it ranged from 40 to 70 percent. The BCCS ascertainment rate of work-related fatal injuries (81.8%) was within the range for U.S. coroner's records reported in the same study (32 to 90 percent). The authors suggested that the gap between the proportion of cases ascertained for all industries by different data sources may be the result of differences in compensation coverage among states (Stout and Bell, 1991). In this study, because sawmills are covered by the WCB, we expected that the list of W C B claims for sawmill workers would include most cases and this was confirmed in our results. In fact our findings are comparable to those reported by Rossignol (1994) regarding capture rates for the industrial group "forestry, fishing and mining" in Quebec. He reported that 91.8% of all deaths in this industrial group were captured by the WCB. In contrast, Rossignol found that the proportion of cases ascertained was much lower for Agriculture and Transportation (45.5% and 76.6% respectively). Applying capture-recapture methods, this author calculated that the true number of fatal injuries at work was 24% higher than that obtained from Quebec WCB alone and 9% higher than that obtained using Quebec WCB and Coroners death certificates combined. The author concluded that more self-employed and business owners are likely to be found in agriculture and transportation. 5 7 With regard to the accuracy of the designation "place of injury" on the death certificates, our findings show that it is better at identifying deaths that have not occurred at industrial places than at identifying those that occurred at a workplace. The use of the notation "on industrial premises" results in failure to capture fatal injuries that occur off premises but that are work-related (e.g. motor vehicle-related). This is the same failure observed in the BCCS system. While the small number of deaths included in this study may have produced unstable estimates (most likely underestimates) regarding completeness of ascertainment of fatal work-related injuries, it has served the purpose of illustrating the weaknesses and strengths of each system. Criteria for developing, implementing and evaluating an injury surveillance system have been proposed by several authors (Rubens, et al., 1995; Waller and Clemmer, 1993; Graitcer, 1987; Klaucke et al, 1988). In agreement with them, we propose that in order to improve surveillance of work-related fatal injuries, both the BCCS and the B C W C B should use standardized industrial and occupational codes (e.g. North American Industry Classification System, NAICS, 1997). The adoption of standard industrial and occupational coding systems will facilitate comparability among data sources at the provincial, national and international level. Similarly, the use of E codes to classify causes of death would help make the data comparable with that recorded by Vital Statistics. The development by the BCCS of standardized data collection procedures similar to the B C W C B Investigation Report will eliminate the great variability in the type and amount of 58 information collected by different Coroners. Notwithstanding these limitations, most of the time the investigation reports from the Office of the Chief Coroner provided a detailed narrative of the circumstances of every injury death, making this data source a very useful resource for epidemiological research on injury fatalities. The possibility exists for building a population-based occupational injury fatality surveillance system in British Columbia. Having automated files with personal identifiers, it should be feasible to link death certificates with Coroners records and WCB files. This record linkage procedure would facilitate the use of current resources in a more efficient way and promote cooperative relationships between participating agencies. Likewise, linking records would improve the quality and completeness of information by allowing reliability and validity checks among sources (Conroy, 1993; Colorado Department of Health, 1984). The benefits from this strategy would be the availability of the best possible information to plan, implement and evaluate injury prevention programs. 59 CHAPTER FIVE 5.0 Epidemiology of work-related fatalities among sawmill workers, 1950-1990 5.1 Introduction In British Columbia (B.C.), the forest products industry constitutes a significant component of the provincial economy. In terms of production, in 1992 Canada was the world's largest producer of newsprint, the second largest producer of pulp and the third largest producer of sawn lumber. Within Canada, the British Columbia sawmill industry with a workforce of approximately 27,881 workers, produced the largest amount of lumber in 1992 (Statistics Canada, 1994). Although the type of equipment may vary from one mill to another depending on the type of wood processed and the degree of modernization of the plant, in general terms the production phases in a typical sawmill are the same around the world (Driscoll et al., 1995). A summary of the processes at a typical mill comprises the following sequence: after arrival to the mill, the logs are transported by a system of conveyors, belts and rollers from one cutting station to the next, until the desired shape and size are obtained. The next step consists of the grading and sorting of the boards according to the standards of quality of the market. Then, and depending on the subsequent industrial use, the lumber may be transported through another cutting and planing station for final shaping and smoothing of the surfaces. This finished product is then sorted, stacked, bundled, and prepared for shipping (Davies, et al., 1998). Forestry ranks among the industries with the highest rates of both lost time injury and fatalities (Labour Canada 1990; Salisbury, et al., 1991). The Workers' Compensation Board of B.C. (2000) reported that in 1999, a total of 6,574 claims were accepted from the sawmill industry, and sawmill workers stayed off work 128,453 days due to short and long term 60 disability. During the ten-year period from 1990 to 1999, 42 sawmill workers died from injuries sustained at the workplace, outnumbered only by loggers who have the highest fatality rates within the forest industry. (WCB of BC, 2000). Sawmill workers are exposed to several chemical and physical hazards including wood dust, pesticides, metal fumes, noise, temperature extremes, mould and bacteria (Davies, et. al., 1998). Similarly, exposure to safety hazards is virtually ubiquitous, and risks of falls, cuts, slips, trips, strains, and being struck by objects are present throughout the different process areas of a mill. It a workplace where strict adherence to safety regulations and procedures is indispensable (Demers 1998a). Ergonomic studies in sawmills have found that heavy physical workload, uncomfortable working postures, fast work pace, noise, dust and extreme temperatures are the most common concerns voiced by workers. The problem is compound by the psychological stress that results from the high repetitiveness of the tasks and the social isolation caused by the attention demands, noise and distance between operators (Aminoff, 1974; Baneryd,1974). A few studies have been conducted among sawmill workers to investigate the sequence of events leading to non-fatal injuries. Characteristics of injury scenarios suggest that injuries do not occur by chance, but rather in specific production areas during the performance of specific tasks. Handling logs or unblocking jammed logs (Laflamme and Cloutier, 1989); performing maintenance work (Aminoff, 1974); clearing away logs, and feeding machines (Larsson, 1991) have been identified as the most common activities preceding a work-related injury. 61 Injured workers were most frequently working in the following sections: sawing, planing, sorting and piling (Laflamme and Cloutier, 1989); packaging and sawing (Aminoff, 1974). Cooke and Blumenstock(1979) reported that among sawmill workers in Maine, injury severity was associated with worker's age, processing hardwood (as opposed to softwood) and being temporarily assigned to a job (as opposed to permanently assigned). Punnett (1994) also studied work-related injuries among sawmill workers in Maine with the purpose of identifying environmental risk factors associated with injury occurrence. The strongest risk factors associated with injury were: processing hard wood and high physical demand. Having 1 year or less of experience on the job, inability to take a break when tired, and absence of a lockout program in effect in the plant were also found significantly related to injury occurrence. Driscoll et al. (1995) analyzed 16 sawmill work-related fatalities in Australia and reported that workers most frequently died as a result of being struck by objects while handling logs or operating bench saws. Although studies on work-related injuries in the sawmill industry are limited, they provide a basis for understanding the type of factors that are part of the causal pathway of injuries for this particular workforce. However, most of these studies, were designed as case series (except Punnett's case-control study) and included almost exclusively non-fatal injuries without differentiation of severity (except Driscoll et al.). In the present study complete counting of person-years contributed by every cohort member allows the calculation of mortality rates which are not commonly reported in statistics of industrial injuries, and the focus is on the most severe outcomes where preventive strategies would have the greatest impact. 62 The specific objective of this study was to examine all work-related fatal injuries among sawmill workers during 1950-1990, in terms of a) characterizing the pattern and trend of fatal injuries at work b) estimating the risk of fatal injuries at work c) assessing the relationship between selected variables and the risk of fatal injuries at work d) identifying the occupational categories at highest risk of fatal injuries at work 5.2 Methods 5.2.1 Study population- The original B.C. Sawmill Workers study This study is part of a larger investigation of health effects of exposure to fungicides among sawmill workers (Hertzman, et al., 1997). The original cohort was enumerated from mill personnel records and included all workers who had at least 1 year of cumulative service between January 1st, 1950 and December 31 s t. 1985 at any of the study mills. Fourteen sawmills operating in different regions within British Columbia participated in the study. They were selected based on completeness of records, use of fungicide and willingness to participate. Eleven mills were located on the coast and three in the interior of B.C. Table 5.1 shows the location of the 14 mills and the number of participating workers from each one. Tahsis, Youbou and Mackenzie are mill towns where the only source of employment is the mill. 63 For each worker demographic and occupational history data were abstracted from the company personnel records. The demographic file includes birth date, date of hire and date of termination from the mill. The job history file comprises a complete history of all jobs held at a mill, including job title, process area, start date and end date for each job held. Table 5.1. Location of study mills and number of workers participating in the study Mill Region # workers 1 Vancouver Lower Mainland 2259 3 Vancouver Lower Mainland 2481 5 Vancouver Lower Mainland 2709 2 Vancouver Island (Chemainus) 2311 4 Vancouver Island (Port Alberni) 2211 6 Vancouver Island (Port Alberni) 2597 8 Vancouver Island (Youbou)* 1970 9 Vancouver Island (Tahsis)* 2336 10 Vancouver Island (Nanaimo) 1459 7 Coast Garibaldi (Squamish) 661 16 Coast Garibaldi (Powell River) 2768 11 Interior (Kelowna) 487 13 Interior (Mackenzie)* 781 15 Interior (Mackenzie)* 1381 * Mill town Twenty six thousand four hundred and eleven male workers met the entry criteria and were enrolled in the study. Females were not included because of their small numbers. The file with personal identifiers was linked by means of probabilistic linkage methods, to the British Columbia Death File and the C M D B for the period 1950 to 1990. Secondary sources used for vital status tracing included pension plan, motor vehicle, union and income tax records. As a result of these linkages, for the original study cohort, vital status as of December 31 s t 1990 was obtained for 85.7% of the cohort. 64 Cause of death was obtained from the provincial Division of Vital Statistics and was coded according to the International Classification of Diseases in effect at the time of death. A l l causes of deaths with an External cause of death code (E800-E999), were selected as cases of interest for this study. 5.2.2 Work-relatedness of a fatality The determination of whether or not a death was work-related was based on information abstracted from the B.C Coroners Service (BCCS) and the WCB records. With the list of names of sawmill workers who died from injuries in our Sawmill cohort, a search was performed in the Coroner's Office files. Under the Coroner's Act of British Columbia all sudden, unexpected deaths shall immediately be reported to the Coroner. The coroners investigate the identity of the deceased and the facts about the death. Coroners records may include autopsy reports, police reports, WCB reports and other information deemed important for the clarification of the circumstances of the event. The Coroner's Office has computerized records of fatalities from 1984 to present. For deaths that happened before 1984, hard copies were manually searched to retrieve the reports corresponding to deaths that occurred during the period 1950 to 1983 inclusive. The data describing the circumstances of a worker's death were first recorded on forms purposely designed containing the variables of interest for this research (Appendix B) and 65 later entered into a computer database. From the narrative of the investigation report it was possible to identify all deaths from injuries sustained at the workplace. To confirm work-relatedness of a death, a list of all accepted claims for fatalities in the sawmill industry from 1950 to 1990 was obtained from the WCB including the worker's name, the date of injury and the company name, size and location. A work-related injury was defined as any fatal injury that occurred at the worksite or as a result of injuries sustained while working at any of the 14 study sawmills, between January 1, 1950 and December 31 s t, 1990. 5.2.3 Assessment of exposure To summarize the data for the analysis in the current study, the approximately 800 job tittles used in the analysis of the original database were collapsed into groups based on similarity in job tasks (8 groups), and process area (13 groups). Each group includes one "unclassifiable" category (Tables 5.2 and 5.3). Table 5.2 Occupational categories, based on job tasks Foreman/Supervisor Skilled trades (Use tools)* Mobile equipment operator Material handling/unskilled* Machine operators/attendants/clearers/sorters * Inspector, grader, other skilled worker (non tool user) Non-wood industry & non-production Unclassifiable Adopted from Punnett, 1994 * jobs that have been associated with increased risk of injury in previous studies 66 Table 5.3 Major process areas in a sawmill and exposure to safety hazards Process areas Exposure Debarking Elevated walk-ways; unsecure logs/lumber Sawing/trimming/edging* conveyor belts; saws/cutting equipment; Planing flying debris; failure to lock-out machinery Chipping and related operations Clean-up Maintenance* Sorting and clearing* Elevated walk-ways; unsecure logs/lumber; conveyor belts Grading flying debris; failure to lock-out machinery Yard and pond Mobile equipment; unsecure logs/lumber; conveyor belts Stenciling, packing and shipping* Elevated walk-ways; unsecure logs/lumber; conveyor belts; mobile equipment; failure to lock-out machinery Kiln drying Mobile equipment General plant functions Occasional exposure to any of the above safety hazards Unclassifiable Adapted from Davies et al., 1998 *: Areas that have been associated with increased risk of injuries in previous studies The task-based classification was developed by Laura Punnett (Punnett, 1994). The grouping by process areas is based on the work by Davies, et al. (1998) with a slight modification to include a category for general plant functions that comprises non-production workers (i.e. clerical, first aid, watch-man). 5.2.4 Statistical analysis The first analysis was descriptive and examined the frequency distribution of fatalities by selected variables. These variables were grouped into three categories according to the 67 Haddon matrix: factors relating to the environment, factors relating to the person, and factors relating to the injury event. Crude rates of fatal injuries per 100,000 person years were calculated for work-related injuries for the 40-year period for the whole cohort, and also for every job and process area categories. For the present investigation, because the outcomes of interest are work-related injury deaths that occurred to workers while actively employed at a mill, the follow-up was truncated at the date of termination plus 30 days. This criterion was adopted to reduce the probability of counting deaths that may have occurred in other jobs, but also to avoid missing deaths that may have occurred as late effects or complications of injuries sustained at the work-site. Internal cohort analyses were conducted to compare the job categories in terms of their risk of death from work-related injuries. Rate ratios comparing mortality rates for each job category against the rest of the cohort adjusted for age and calendar time, were calculated using Poisson regression modeling (Frome and Checkoway, 1985; Breslow and Day, 1987). The decision to use the rates for the whole cohort (except the category being compared) as the reference category was based on the fact that the only group that could be treated as relatively unexposed, "non-wood products", would have very unstable rates given the small number of deaths in that category. Similarly, due to the small number of fatalities, all types of injury ( e.g. falls, struck by falling object, etc.) were combined for this analysis. 68 Poisson regression was also used to compute rate ratios and 95 % C.I. for the relationship between duration of employment in current job, age and calendar period. The covariate duration of employment in current job is defined as the cumulative length of employment in the current job (i.e. the one held at the moment of the injury event). Two duration of employment categories were defined as 1) less than 2 years in current job, and 2) equal to or greater than 2 years in current job. Age was dichotomized into 2 categories: 1) less than 35 years, and 2) equal to or greater than 35 years. The boundaries for the calendar periods of interest were "<1970", and ">= 1970". The categories "less than 2 years in current job", "<35 years", and "<1970", were used as the reference groups for the rate ratios. The selection criteria for the independent variables boundaries were based on an "a priori" concept of selecting one distinctive category against which the others will be compared; ideally with a large enough number of observations that a stable estimate will result against which relative rates will be derived. Having a small number of deaths (number of deaths = 40), further stratification of variables would result in several cells having 0 or 1 case each. One methodological and technological challenge was the reformatting of the database for input onto Life Table Analysis System (LTAS) so that the program could be used to analyze acute effects of risk factors instead of the common cumulative effects studied in most occupational cohort studies. A detailed explanation of the data preparation process for the analyses performed in this study, is provided below. 69 5.2.5 Data preparation Two statistical packages were used in the analysis of the cohort database. First, we used the Life Table Analysis System (LTAS, version 1.0.c) developed by researchers from NIOSH (Cassinelli et al., 2000), to produce a file with observed deaths, expected deaths, and person time at risk for each combination of level of risk factors for each relevant cause of death. Next, we used EGRET version 0.26.06 (SERC, 1993) for Poisson regression modeling to compare risk of fatal injuries across job categories. LTAS was developed by researchers from the National Institute for Occupational Safety and Health (NIOSH), primarily to analyze cohorts defined according to workplace exposures. The program requires input of two types of files: a demographic file and a job history file. Observed deaths are divided by person-time at risk to construct rates by age, sex, race, calendar time, and duration or level of exposure. Person time at risk of death usually begins when exposure begins and continues until death date, date of last observed, or end of follow-up period. Mortality rates can be calculated for each stratum and compared with external rates via indirect standardization to compute Standardized Mortality Ratios (SMRs). Apart from calculating rates, the program can generate stratified files with observed deaths and person-time at risk, that can be exported and used with other software for regression analysis. The database compiled for the B.C. Sawmill Workers Cohort consisted of two files: the demographic file and the job history file which were reconstructed for comparison of risk of work-related fatal injury among job categories. The reason for the reconstruction was to be able to accurately assign person-time at risk of death according to the outcome of interest of 70 this study. The interest of most occupational cohort studies is chronic diseases (e.g. cancer) that develop after long periods of exposure. For these chronic outcomes the program easily computes person-time at risk across the period of interest. In the present study, the outcome of interest is traumatic injuries (acute) that occurred to individuals while actively employed at a sawmill. Therefore, the exposure that matters is the latest (i.e. current job) and the follow-up period ends at date of death or at the date of termination from employment. 5.2.5 Internal cohort analysis by job categories Editing the job history file: In the occupational history file, each record contains the worker's identification number, start and termination date for each job held, and the job category. For workers who held more than one job the following changes were necessary. • First, convert each person into as many persons as the number of jobs held. For example, a person who worked at 3 different jobs, was converted into 3 persons, each with a single job. • Second, assign new identification numbers to account for each job as i f it corresponded to a different worker and also to avoid LTAS rejecting the record as a duplicate. An example is shown below Table 5.4. Job history file with a record for a hypothetical worker who held 3 jobs between 1950 and 1967 Worker ID Start date End date Job category 1000001 12/3/50 10/23/52 I 1000001 10/24/52 5/23/58 2 1000001 5/24/58 7/30/67 3 71 Table 5.5. Job History file with artificial records created for each job Worker ID Start date End date Job category 1000010 12/3/50 10/23/52 I 1000011 10/24/52 5/23/58 2 1000012 5/24/58 7/30/58 3 Editing the demographic file The demographic file contained 26,411 records, one for each worker who worked at least one year between January 1st, 1950 and December 31 s t, 1985 at any of the 14 study sawmills. The variables included in this file are the following: worker identification number (ID), birth date, sex, race, vital status, start date (at which person-years counting starts), date of last observed (at which person-years counting ends), date of death and cause of death. In order to have a demographic file consistent with the new job history file the following modifications were carried out. For each worker, the counting of person years (i.e. date last observed) was truncated at any of these points in time: a) The last day of work (termination date) plus 30 days: workers who left the industry before study end. b) The date of death: workers who died while actively employed at a mill or within 30 days after termination. c) December 31, 1990: workers who were actively employed at a mill at the end of the study period. The rationale for counting person-years only up to 30 days after termination was that the study aim is to characterize the pattern of mortality from injuries that occurred to individuals while actively 72 employed as sawmill workers. The extra 30 day-period allows for the accounting of deaths which occurred after the date of injury. A re-coding of vital status was necessary to be consistent with the criteria for case selection. Thus, for deceased workers who died more than 30 days after termination, the death date and the cause of death fields were not recorded and the vital status was changed to alive. A demographic file containing the same number of records as the job history file, was constructed. This process consisted of creating records with demographic information to match each artificial person added in the job history file. For workers who remained in the same job from the date of entry to the date of termination, no changes were needed in the demographic file. However, i f a worker had more than one job, an artificial demographic record was created for each artificial job record as i f each new job corresponded to a different worker. Table 5.6. Demographic file with record for a hypothetical worker who held 3 jobs between 1950 and 1967 Worker ID Vital Date of Person-year Date last Date of Underlying cause of death Status birth begin observed death 1000001 2 12/15/30 12/3/51 7/30/67 8/24/67 850 Table 5.7. Demographic file with records for 3 hypothetical workers who held 1 job each between 1950 and 1967 Worker ID Vital Status Date of birth Person-year begin Date last observed Date of death Underlying cause of death 1000010 1 12/15/30 12/3/51 10/23/52 1000011 1 12/15/30 10/24/52 5/23/58 1000012 2 12/15/30 5/24/58 8/24/67 8/24/67 850 73 These artificial records (in bold) were constructed according to the following rules: The start date in the demographic file was consistent with the start date on the job history file. The stop date (or date last observed) was the same as the end date in that job. The birth date from the original record was repeated for each of the artificial records pertaining to the same person. The death date was recorded only in the last artificial record pertaining to a deceased worker. New identification numbers were assigned to account for every record created in the new demographic file and corresponding to each record of the job history file. At the end, the original 26,411 workers were partitioned into 199,371 fictitious individuals. The new job history and demographic files were used as input to LTAS to produce one export file for each selected cause of death and each of the 8 job categories stratified by age and calendar groups. These files were then merged and indicator variables were created. Next, using Poisson regression from the statistical package EGRET, the risk of work-related fatal injury for each job category was compared to the risk for the rest of the cohort. Table 5.8 shows an example of the file generated for regression analysis. Table 5.8. Data Analytic File for job comparison analysis Age Calendar Cause of Observed Person-days Job 1 Job2 Job3 Job4 Job5 Job6 Job7 Job8 Stratum Stratum death deaths 1 1 825 0 2500 1 0 0 0 0 0 0 0 1 2 825 1 5678 0 1 0 0 0 0 0 0 74 5.3 Results 5.3.1 Rates and time trends in fatal injuries at work Eighty-eight cohort members died from work-related injuries from 1950 through 1990. Of these, forty workers died from injuries sustained at work while employed at any of the study sawmill and the rest occurred while working at a different job. The total number of person-years at risk was 220,965. The overall incidence of work-related fatal injuries was 18.10 per 100,000 person-years. Coroner's files were found for 35 of these 40 work-related fatalities (87.5%). For the remaining five deaths, all of which occurred prior to 1984, a manual search of the hard copies was conducted. One of the five cases had a record in the Coroner's files but the investigation report could not be found. None of the other four deaths had any record in the BCCS. For this reason, part of the descriptive analysis (i.e. that which is based on information from Coroner's reports) could include data for only 35 deaths. In addition, because information on some variables is not consistently collected by the coroners, the data presented may sometimes be limited to fewer than 35 deaths (see footnotes of each table and figure). Figure 5.1 shows that the fatality rates were higher during 1950-1959 than during the last three decades of the study. 75 CB <u >^ I c o (SI •— u a. o o o ©" o CO CO IX Figure 5.1. Work-related injury fatality rates per 100,000 person-years, 1950-1990 35 30 25 20 15 10 5 0 -•—Rate 1950-1959 29.58 1960-1969 15.37 1970-1979 12.23 1980-1990 18.33 5.3.2 Factors related to the worker The highest crude fatality rate was observed for workers over 35 years (Table 5.9). Table 5.9. Crude rates and adjusted rate ratios of work-related fatal injuries, sawmill workers, 1950-1990. Parameter Number Rate/100,000f Age <35 >=35 Period at risk <1970 >=1970 Seniority in current job <2 years >=2 years 11 12.92 29 21.35 24 16 17 23 21.96 14.32 54.45 93.92 | Crude rate per 100,000 person-years *RR is rate ratio; reference level indicated by RR=1.00 RR* 1.00 1.00 1.17 95% C.I 1.89 0.94-3.79 1.00 0.39 0.21 -0.74 0.61 -2.21 The youngest worker who died as a result of an injury on the job was 21 years old and the oldest was 61 years old. Over 70% of the deaths (n=29) occurred among workers over 35 years. 76 While the number and crude rate of work-related fatal injuries were higher among workers with more than 2 years of experience in current job, once adjusted for age and calendar period, seniority does not seem to affect the risk of death. The job categories with the highest crude fatality rates were: machine operators/clearers/sorters, mobile equipment operator, and skilled trades (30.34, 29.40 and 22.30 per 100,000 person-years respectively) (Table 5.10). Table 5.10. Rate ratio of work-related fatal injuries by occupational categories, 1950-1990. Occupational category Observed Rate/100,000f Rate Ratio* 95% C.I. Mobile equipment op. 4 29.40 1.70 0.60-4.79 Machine op./clearers 12 30.34 1.70 0.85-3.40 Foreman 3 20.35 1.04 0.32-3.39 Skilled trades 8 22.30 1.01 0.44 - 2.28 Material handling 11 16.87 0.95 0.47- 1.92 Non-wood industry 1 17.29 0.87 0.12-6.32 Unclassifiable 1 2.72 0.58 0.21 - 1.63 t Crude rate per 100,000 person-years * Rate ratio relative to the entire cohort (excluding the category considered) adjusted for age and period at risk These groups accounted for 60% of all work-related fatalities (n=24). When rate ratios adjusted for age and calendar period were calculated, there were no significant differences between any occupational group and the rest of the cohort. However the regression analysis showed a significantly higher risk of dying from injuries at the workplace during the first half of the study period (1950-1969) than during the last half. 77 Machine operators and material handlers had the largest numbers of deaths from being caught by machinery. Also for both groups the second cause of injury was being struck by or against objects (Table 5.11). Table 5.11. Fatalities by occupational categories according to type of injury and seniority in current job Occupation Type of injury Seniority in current job (years) Total <2 2-5 6+ Machine operators/clearers Caught by machinery 2 1 2 5 Struck against/by 3 2 5 Fall 1 1 2 Material handling Caught by machinery 1 3 4 Struck against/by 2 2 Fall 1 1 Water transportation 1 1 2 Drowning 1 1 Motor vehicle 1 1 Skilled trades Caught by machinery 2 2 Struck against/by 2 1 3 Suffocation 1 1 Late effects of injury 1 1 Electrocution 1 1 Mobile equipment operator Caught by machinery 2 2 Struck against/by 1 1 Water transportation 1 1 Foreman Fall 2 2 Drowning 1 1 Non-wood industry Railway 1 1 Unclassifiable Cutting object 1 1 Datasource: B C S W C database (N=40) Approximately 50% of the fatalities among the groups of machine operators (6 out of 12) and skilled trades (6 of 8), occurred to workers with less than 2 years of experience in the current job. 78 Information on the level of alcohol in blood was tested post-mortem in 15 cases (43%). Of these, only one case tested positive for alcohol above 0.08 mg%, which is the legal limit for driving in British Columbia. No alcohol was detected in the other 14 cases. One case tested positive for morphine, but it was a consequence of resuscitation procedures. It is a Coroner's prerogative to request a toxicological examination whenever alcohol or drugs may have contributed to the death. Therefore the lack of testing for the remaining 25 deaths may be interpreted as cases for which there was no reason for suspecting alcohol or drug involvement. 5.3.3 Factors related to the environment Table 5.12 shows the number and fatality rates by process area. The three areas or departments with the highest number of fatalities were: maintenance, clean-up and yard/pond, but the highest rates were among debarking, planing, and chipping operations. The lowest risk of fatal injuries was observed in the area of kiln drying, manual sorting and grading, and stenciling/packaging/shipping. Table 5.12. Number and crude fatality rates of work-related injuries by process area, 1950-1990 Process area Person-years No. Rate/100,000 95% CI Debarking 1,310 1 76.33 1.93 -425.15 Planing 6,105 2 32.76 3.96-118.26 Chipping and rel. op 3,098 1 32.27 0.81 - 179.74 Yard and pond 24,750 7 28.28 11.34- 58.25 Non-cutting equipment 7,588 2 26.35 3.18-95.12 Clean-up 35,852 8 22.31 9.61-43.95 Maintenance 42,304 9 21.27 '9.74-40.41 Sawing, trimming 26,802 5 18.65 6.04-43.45 Stencilling, packaging, shipping 6,895 1 14.50 0.36 - 80.76 Sorting, grading (manual) 27,910 1 3.58 0.09- 19.94 Kiln drying 465 0 0 0-859 Unclassifiable 28,808 3 10.41 2.14-30.39 Datasource: BCSWC database (N=40) 79 In 85% of the fatalities with Coroners reports (n=30), the type of investigation conducted was an inquest. A l l deaths were classified as "accidental" except one that was classified as homicide, meaning that the death resulted from an injury intentionally inflicted by the action of another person but without implying legal responsibility or fault. This worker died as a result of injuries from a fall because he was standing on insecure platform, a fact that was known by the supervisor without corrective action being taken. Recommendations to prevent reoccurrence of work-related deaths were made by the Coroner's jury in 29 fatalities. These were classified as shown in Table 5.13. Table 5.13 Number of Coroner's Jury recommendations to prevent work-related deaths Recommendation category No. of reports with No. of category recommendations Equipment/environment 11 42 Company safety policies 10 23 Communications 9 16 Safety equipment 5 5 Job process 5 8 First aid 4 7 Training 3 4 Supervision 1 1 Datasource: BCCS Reports It can be seen that the majority of the time the jury recommended actions on the physical environment or equipment, to prevent future deaths under similar circumstances. For example there were 11 inquest reports containing at least one recommendation in the physical environment and equipment category. Furthermore, it is noteworthy that in each of these 11 inquest reports, the jury made several recommendations for the same category. Thus, the 11 reports contained, in total, 42 recommendations about the need to modify the environment or equipment. Twenty-three recommendations were made to the employer with 80 regards to the company safety policies. Deficiencies in training and supervision were the least noted in the reports. In some cases, it was possible to identify contributing factors either because they were mentioned as such in the course of the narrative of the event or, i f not explicitly mentioned, were the result of our interpretation of the facts described in the Coroner report. These contributing causes were classified as environmental factors (physical environment and socio-cultural environment), personal factors, and factors related to the vehicle of injury (Table 5.14). Unsafe work-station was the most frequently mentioned contributory factor for a fatal injury. Table 5.14. Factors contributing to work-related fatal injuries, 1950-1990 Factor Condition Number Physical environment Unsafe work station 14 Socio-cultural environment Insufficient training, instructions 6 Personal Poor work practice 13 Lack of experience 8 Lack of personal protective equipment 3 Injury event Equipment failure 4 Inadvertent start of equipment 4 Datasource: BCCS Reports Workers employed at the mills in Vancouver Island and Coast Garibaldi were at higher risk of dying from work-related injuries than their fellow workers employed in Vancouver region or the interior of the province. The fatality rate for the former region was 21.23 deaths per 100,000 person-years vs. 13.07 and 12.89 for Vancouver and the interior, respectively. The frequency of fatal injuries varied with the time of the day, the day of the week and the month of the year (Figures 5.2; 5.3; 5.4). 81 Figure 5.2. Work-related fatality rates by time of occurrence, 1950-1990 <D >> i e o <D OH o O q o a. Pi 3.5 3 2.5 2 1.5 1 0.5 0 • 4:00 6:00 8:00 10:00 12:00 2:00 4:00 6:00 8:00 10:00 2:00 A M A M A M A M PM PM PM PM PM PM A M Datasource: BCCS Reports (N=31) There appears to be a peak between 6:00 am and 7:00 am. Twelve out of the forty fatalities (30%) occurred from 2:00 to 7:00 am. Fridays (35%) and Mondays (20%) were the days with the highest proportion of injuries. Fatal injuries among sawmill workers showed a peak in September. However there was no discernible seasonal pattern. Figure 5.3. Work-related fatality rates by day of occurrence, 1950-1990 5 \ <U A Z- 3 o % 2 a. ca 1 Pi Monday Tuesday Wednesday Thursday Friday Saturday Sunday Datasource: BCSWC database (N=40) 82 Figure 5.4. Work-related fatality rates by month of occurrence, 1950-1990 Datasource: BCSWC database (N=40) 5.3.4 Factors related to the injury event Regarding activity at time of the incident, most of the time workers were doing routine tasks such as handling pieces of wood/lumber or cleaning up the work area (34%). The next most common activity reported was doing maintenance of equipment (31%). Furthermore, seven workers (20%) were operating equipment at the time of the injury, 3 were on their breaks and 2 were doing other production tasks (Figure 5.5). Figure 5.5. Activity at Time of Incident On break/leaving equipment/machinery 20% Datasource: BCCS Reports (N=35) 83 The cause of death was abstracted from the death certificates filed with the Coroner's reports. Exsanguination and multiple injuries accounted for approximately 50 % of all fatal injuries (Table 5.15). Table 5.15. Number and percentage of work-related fatalities by nature of Cause of death No. % Exsanguination 9 26 Multiple injuries 7 20 Head injury 6 17 Asphyxia 6 17 Crushed chest/abdomen 4 11 Transection spinal cord 2 6 Cardiac arrythmia 1 3 Datasource: BCCS reports (N=35) The objects involved in the injury event as primary contact were most of the time machinery or parts of machinery, 15 cases (43%), followed by contact with structure (e.g. floor, crossbeam), and log/lumber (Tables 5.16 and 5.17). Table 5.16. Objects involved in work-related fatal injuries, 1950-1990 Primary contact at incident No. % Equipment/machinery 15 42 Structure 8 23 Logs/lumber 5 14 Water 3 9 Electricity 1 3 Other 3 9 Datasource: BCCS Reports (N=35) Table 5.17. Machinery involved in work-related fatal injuries, 1950-1990 Machinery No. Forklift/lumber carrier 4 Log rollers 2 Saw (gang, chain) 2 Conveyor 2 Hoist equipment 2 Chipper 1 Planer 1 Strapper machine 1 84 The most common type of fatal injury at the workplace was being "caught by machinery"(32.5%) followed by "struck by or against objects" (25%) and falls (12.5%). Altogether these three causes represent 70% of the fatalities. A list of all deaths by type of injury is given in Table 5.18. Table 5.18. Work-related fatalities by type of injury, 1950-1990 Type of injury Number % Caught by machinery 13 33 Struck against/by object 11 28 Falls 5 13 Water transportation 3 8 Drowning 2 5 Suffocation 1 2 Motor vehicle 1 2 Railway 1 2 Caused by cutting object 1 2 Electrocution 1 2 Late effect of injury 1 2 Datasource: BCSWC database (N=40) The majority of the fatalities (32 out of the 35 with Coroners reports) were one-person incidents. Reference as to whether or not the supervisor was present at the scene of the injury was found in 21 reports. Of these, in all but one case the supervisor was absent. 85 5.4 Discussion Pattern and trend of fatal injuries at work The overall fatality rate in our cohort was 18.10 per 100,000 person-years for the period 1950-1990. For comparison purposes, given the lack of data from comparable cohorts, rates by decades are selected. The fatality rate reported among Australian sawmill workers for the period 1982-1984 (Driscoll, et al., 1995) was higher than the one observed in our cohort for the period 1980-1990 (30 vs.18.33 per 100,000 person-years). In comparison with other industrial branches, our forty-year rate of 18.10 per 100,000 person-years is higher than the 13.20 per 100,000 person-years observed among U.S. electrical workers during 1950-1988 (Loomis, et al., 1999). Similarly, the fatality rate among our cohort members for the period 1980-1990 (18.33 per 100,000 person-years) was higher than the 13.8 and 11.6 per 100,000 workers in the construction industry in U.S and Australia respectively during 1988-1991 (Ore and Stout, 1996) and also higher than the B.C. fatality rate of 12 per 100,000 person-years for all industries combined during 1985-1994 (Howard, 1996) There was a 38% decline in the rate of fatalities across the 41-year study period. The fatality rates observed during 1950-1959 decreased by almost 50 % in the subsequent decade (29.58 and 15.37 deaths per 100,000 person-years respectively) and then they fluctuated only slightly during the last two decades of the study. The same pattern has been reported by other investigators (Baker, et al., 1992; Stout et al., 1996). Several reasons may account for the observed decline of fatality rates. First, it may reflect the beneficial impact of safety policies implemented at the workplaces since the late 1950's onwards. For example, in 1957 the WCB of B.C. made mandatory the use of 86 buoyancy equipment for workers exposed to risk of drowning. The regulation states : "The device provided shall not be dependent upon manual manipulation to produce the buoyant effect" (WCB, 1966). This safety measure alone saved many lives among boom men. Before 1957, an average of 5 boom men drowned every year (WCB, 1970). It is noteworthy that, in our cohort, 27 % of all deaths during 1950-1959 were due to drowning (3 coded as watercraft incidents and 1 as drowning). There were no deaths from drowning during 1960-1969, one during 1970-1979 and none during the last decade. In 1966 new safety regulations included a revised and expanded section on personal protective equipment, machine guards, and specific injury prevention strategies for sawmills. The right for a worker's representative to accompany an inspector during workplace inspections, the need for the company to have a joint management/labour safety committee and noise control regulations are a few examples of the regulations implemented by the WCB of B.C during the 1970's. (Rest and Ashford, 1992). Moreover, workers are now better informed about hazardous conditions on the job and more conscious about safety measures, and employers are being held responsible for providing safe workplaces (Wegman and Fine, 1990; Stout, et al., 1996). The decline in fatality rates may also be related to the process of modernization of the mills initiated in the 1960's and characterized by major technological changes that led to the automation of several production stages (Government of British Columbia, 1972). 87 Personal characteristics This study found that older sawmill workers were at higher risk of suffering a fatal injury at work than younger workers. Our finding is consistent with other reports from U S A and Canada, whether considering fatalities in all industries (Kisner and Pratt, 1997; Baker, et al., 1992; Sniezek and Horiagon, 1989; Rossignol and Pineaud, 1993) or in particular industries like construction or retail (Prat, et al., 1996; Peek-Asa, et al., 1999; Ore and Stout, 1996). Injuries may have more severe consequences among older workers. Pre-existing diseases, medication side effects, complications and infections could contribute to a poor prognosis in older workers who have been injured (Doering, et al., 1983; Schiller, et al., 1995; De Maria et al., 1987; Rossignol and Pineaud, 1993). Furthermore, a decline in visual acuity, hearing and alertness contribute to a diminished ability to perceive and avoid hazards (Baker, et al.,1982; Robertson, 1992; Salisbury, et al., 1991). In our cohort, one worker of 56 years died while crossing the breezeway on his way to the lunchroom; he didn't hear the straddle loader and was run over by it. Three workers over 55 years died as a result of injuries sustained from falling while doing maintenance work. The most common type of incident among older workers was being struck by or against objects. Two deaths were equipment-related, in one the contact was a forklift and in the other was a case roll. It seems that experience or seniority in the plant was not a factor for these fatalities, since two of the 13 workers over 55 years were foremen and eleven had been employed for more than 10 years in the mill. The small number of cases in this study preclude further analysis 88 for this group of workers but it concurs with other investigations in identifying older workers as a high risk group for work-related deaths. The implications for prevention are that this group needs to be given special consideration, especially since recent population projections estimate that the population of Canadians aged 55-64 years is expected to increase 73% by 2016 and 76% by 2026 (Statistics Canada' Internet Site,2001). Machine operators and mobile equipment operators were the occupational groups with the highest crude fatality rates. Material handlers and clean up workers on the other hand, even though they had the second largest number of fatalities, were at lower risk of suffering fatal injuries compared to the rest of their coworkers. These findings were expected since the physical environment in a sawmill is dominated by high-speed machinery, cutting blades, and open conveyors transporting large and heavy logs from one machine to the next. In this setting workers need to act fast and accurately to maintain a continuous production line eliminating the flow disturbances as they arise and in the shortest time possible. Maintenance work is often done during break time (e.g. lunch time) so that the production line is not slowed down (Larsson, 1991). Not surprisingly workers were most frequently clearing jammed logs or doing maintenance work when they were caught by a machinery or struck by a falling object. The same activities at time of injury and mechanism of injury have been reported by Larsson (1991) and Laflamme and Cloutier (1989) in their studies of non-fatal injuries among sawmill workers in Sweden and Canada respectively. In contrast, Driscoll (1995) reported that Australian sawmill workers were loading/unloading logs when they were fatally injured. 89 Blood alcohol levels were above 0.08 mg% in only one case out of the 15 (0.06%) for whom test results were available. Although it seems that alcohol was not a factor in the occurrence of fatal injuries, conclusions are limited due to the small number of tests reported. However this result is consistent with previous studies. Salisbury et al. (1991) reported that none of the fatalities among B.C. loggers occurred as a consequence of alcohol consumption (only two tested positive and were below 0.08 mg%, the legal limit for driving in Canada). Other studies on work-related fatalities in all industries have similarly found that alcohol does not play a role in the occurrence of fatal injuries at work. (Baker, et al., 1982; Copeland,1985; Shannon, et al.,1993; Parkinson, et al., 1986; Sniezek and Horiagon,1989; Lewis and Cooper, 1989). Therefore the lack of testing for alcohol for most of the deaths included in this study, may indicate that the Coroner in charge of the investigation did not consider it relevant to the incident. According to our results, a non significant increase in the risk of work-related injury deaths was directly observed among workers with more seniority in current job. In contrast, it has been reported that sawmill workers with less than 1 year in current job or in temporary job assignment were more likely to suffer non-fatal injuries (Punnett, 1994) or have more severe injuries (Cooke and Blumenstock, 1979). In our cohort, workers with less than 1 year of employment were excluded so it was not possible to assess risk of injury specifically during the first year of employment. In another industry, a direct relationship between seniority in current job and risk of fatal injury at work was found in a cohort of electrical workers in U.S.A. (Loomis, et al., 1999). Seniority has generally been interpreted as a measure of experience in the job and increased fatality rates in new hired workers has been interpreted as lack of experience or training. High fatality rates for workers with more than 2 years of 90 employment in their current job, can be the result of excessive confidence performing their daily routine. It may also be possible that the occupational history data from company records used to estimate seniority, did not always reflect the actual seniority in the current job. For example, according to Coroner's records, 3 workers were replacing coworkers the day of the incident but according to the company records these workers had been for more than 2 years in their current job. It is possible that replacements or temporary assignment of duties are decided on the shop floor but not recorded on personnel records because the situation is only temporary. It is likely that workers performing non-routine tasks have not received appropriate and complete training (Shannon, et al., 1993). Environmental characteristics The higher fatality rates observed among mills located in Vancouver Island and Coast Garibaldi compared to the mills in the City of Vancouver and the interior of B.C., is consistent with the pattern for injury deaths in the general population of B.C. for the period 1980-1990 (B.C. Ministry of Health, 1997). The City of Vancouver had lower rate of deaths from injuries from all causes than Vancouver Island. In the U.S. the highest fatality rates have been found in rural areas (Baker, et al., 1992). In this study the fatality rate in Vancouver Island and Coast Garibaldi mills was 38% higher than the other two locations. Two of the mills on Vancouver Island (Tahsis and Youbou) are located in small, isolated towns economically dependent on the mill. Rural/urban differences in rates may reflect unequal access to medical treatment (Baker, et al., 1992). The risk of sustaining a fatal injury seems to peak both at 6:00 am and 10:00 pm every day. The highest proportion of deaths occurred between 2:00 am and 6:00 am which corresponds 91 to graveyard shift. The peak at 6:00 am may indicate fatigue for those who are finishing the shift. More workers died on Fridays and Mondays than on the other weekdays. Again, the role of fatigue at the end of the week may explain the Friday peak. The higher frequency on Mondays, also described among loggers, has been ascribed to a hangover effect from the weekend (Salisbury, et al., 1991; Driscoll, et al., 1995). In our study, more deaths occurred on Fridays than on Mondays which appears consistent with the lack of alcohol involvement in these incidents. September was the month peak for fatalities, followed by December. However there is no apparent seasonal pattern. It may depend on monthly variations in production demands or be the result of chance (Driscoll, et al., 1995). In our study, a couple more workers died as a result of machinery-related injuries (33%) than from being struck by objects (28%). Falls were the cause of death in 13% of the fatalities. Four out of the five fatal falls occurred while the operators were working on elevated surfaces that were not properly secured. Two workers stepped on insecure planks and fell. One worker fell through a hole in the wire mesh platform where he was standing. Another worker fell from a crane used as elevator, where he was standing on loose planks. Apparently this was considered a normal procedure. o 92 Environmental factors were identified as contributory causes in the majority of the fatal incidents. The contributing factors were grouped into three categories for ease of illustration but they are interrelated. Unsafe work station (e.g. insecure planks, hole in wire mesh platform) was mentioned 14 times in the narrative of the fatalities studied. Injury event characteristics Factors related to the injury event, like equipment failure and unexpected starting of equipment, was mentioned in 8 fatalities. Eighty-three percent of the workers died within 24 hours of injury. The severity of the injuries suffered seems to indicate that faster medical treatment or the use of personal protective equipment would probably not have prevented these deaths (Driscoll, et al., 1995). Over 80% of the deaths were caused by severe trauma to the head, spinal cord, or multiple sites. Limitations of the study derive from the fact that the sawmills enrolled in the study were those willing to participate and they over-represent large plants. It has been reported that largest mills have lower injury rates than smaller ones (Punnett, 1992). Thus, our estimates of fatality rates may be conservative compared to the experience for the entire industry. Larger companies are more likely than smaller companies to have resources available for safety improvements. Therefore, while the results may apply to large companies, extrapolation to the whole industry (i.e. including small mills) may be questionable. Despite the large number of workers in our cohort, there was a small number of fatalities from work-related injuries and this precluded further analysis (e.g. by specific external cause of death) of the cases, and limited the statistical precision of the estimates. Another limitation is that a number of factors studied were not consistently recorded by the BCCS; it was not possible to assess their impact on the occurrence of fatal injuries at the workplace. Summary Our results suggest that machine operators/clearers/sorters, over 35 years with more than 2 years of experience in their current job were at highest risk of suffering a fatal injury at work. Most frequently workers handling materials, clearing jammed logs or doing maintenance work were caught by machinery or struck by falling objects. Of greatest importance in the analysis of the various factors related to the fatal incident, are the recommendations of the Coroner's jury because they unveil some contributing factors which i f corrected may prevent reoccurrence of similar deaths. The majority of the recommendations pointed at factors related to the physical environment of the mill (e.g. unsafe work station, equipment) as well as organizational factors (e.g. company safety policies, training). From a practical point of view, the findings of this study are useful because they provide further evidence that injury control strategies directed at correcting the environment could have effectively prevented unnecessary deaths among these workers. Unsafe work-stations, defective equipment, inadvertent start of equipment, are all modifiable by adherence to safety 94 regulations and most effectively by engineering control. Insufficient training or instructions are dependent on the company commitment to provide a safe work environment. Poor practices can be reduced if safety is valued more than production by both management and labour. The traditional approach to the control of injuries at the workplace has mainly focused on changing the worker's behaviour (Kraus and Robertson, 1992). However, it has been recognized that the reduction of occupational injuries has been achieved mainly by eliminating high-risk tasks, changing procedures to make them less hazardous, imposing physical barriers between workers and equipment and installing fail-safe devices and secondarily by education on safety regulations (Haddon and Baker, 1981). 95 CHAPTER SIX 6.0 All fatal injuries among sawmill workers, 1950-1990 6.1 Introduction Injuries are a major national public health problem because they are associated with a significant burden of morbidity and mortality and consume substantial proportions of health expenditures (Wigle, et al., 1991; Millar, 1995; Baker, et al., 1992). Injury is the leading cause of death for Canadians between 1 and 44 years and the fourth cause of death among all age groups, after cardiovascular, cancer and respiratory diseases (Riley and Paddon, 1990). Direct and indirect costs associated with injuries in Canada were estimated at over $11 billion in 1986 (Wigle, et al. 1991). The impact of injury mortality in society can be illustrated by the number of potential years of life lost (PYLL) by those who died before the expected age for the general population. Unintentional injuries are the leading cause of P Y L L before age 65 (MacWilliam and Fortier, 1987). In 1985 injuries caused 219,641 P Y L L in Canada (22.3% of all P Y L L ) . Cancer and cardiovascular ranked second and third in the count of P Y L L for that same year (19% and 14% respectively). Compared to other industrial workers, forestry workers are at a higher risk of injuries at the workplace. Forestry based industries have the second highest fatality rate in Canada (Labour Canada, 1990). In British Columbia (B.C.), during the ten-year period from 1990 to 1999, 42 sawmill workers died from injuries sustained at the workplace, outnumbered only by loggers (WCB of B C , 2000). This would explain why most of the limited literature on injuries among sawmill workers has focused on work-related incidents. 96 Much less is known about mortality in this workforce from injuries sustained in other settings. Few studies analyzing overall mortality among forestry workers in general, have reported a higher mortality rate from unintentional injuries and suicides compared to the general population (Notkola, et a l , 1993; Green, 1991). In these studies the investigators combined different types of injuries in two groups, all unintentional injuries (e.g. transportation-related, falls, drowning, etc.) and suicide. This crude categorization does not allow the detection of increases in specific causes. A mortality analysis by broad categories hinders the understanding of the different contributory factors related to specific types of injury that, once recognized, should help plan preventive interventions. The availability of the British Columbia Sawmill Workers Cohort (BCSWC) database (Hertzman, et al., 1997), provided us the unique opportunity to fully characterize the trend and pattern of both on-the-job and off-the-job mortality from specific injury types in this workforce, filling some of the gaps in the published literature. Work-related injuries were analyzed in chapter 5. This chapter analyses all fatal injuries that occurred in a cohort of over 26,000 sawmill workers from 1950 to 1990. The 40 work-related deaths in the cohort were included in this analysis for the following reasons: 1) the injury specific standardized mortality ratios identify the type of injuries with relative excess or deficit compared to the general population, 2) it makes the study comparable with studies of overall injury mortality in other workforces, 3) work-related injuries are easily identifiable and are not mixed with the causes analyzed in more detail. 97 The objective of this study was to examine mortality from selected external causes among sawmill workers between 1950 and 1990 in terms of: a) characterizing the pattern and trend of mortality from selected external causes (motor vehicle-related, falls, drowning, machinery-related and suicide). b) estimating the risk of mortality from external causes. c) comparing injury mortality from external causes in the sawmill cohort with that of the male population of British Columbia. d) assessing the relationship between age, seniority in mill and calendar period, and the risk of mortality from motor vehicle crashes and suicide. e) comparing injury mortality from motor vehicle crashes and suicide among subcohorts defined according to occupational categories. 6.2 Methods 6.2.1 BC Sawmill Workers Cohort The original B.C. Sawmill workers cohort included 26,487 workers who were employed for at least 1 year cumulative service in any of the 14 study sawmills between January 1st, 1950 and December 31 s t, 1985. Workers employed on a casual basis or any other non-permanent plan, were required to work a minimum of 260 days to qualify for membership. A l l cohort members are men. No women were included due to small numbers. The number of workers in this study (n=26,411) differs slightly from the number analyzed in the original study as a result of data cleaning and compilation for this study. 98 Fourteen sawmills were selected to participate in the investigation based on the use of chlorophenate fungicides, completeness of company records, and willingness to participate. Eleven mills were located in the coast and three in the interior of B.C. Most of the mills are situated in cities or towns built by mill companies for exploitation of the local forests (Demers, 1998b). Some of these towns have evolved into large urban areas where the mills are no longer the sole source of employment (e.g. Vancouver Lower Mainland), and some remain as forest-based communities (e.g. Squamish, Tahsis, Youbou). Two of the six mills located on Vancouver Island are very special from the point of view of their geographic and socio-cultural characteristics : Tahsis and Youbou. They may be described as isolated single industry-towns where the mill is the only source of employment for the community. In the interior of B.C. there are three mills, one in the city of Kelowna, at 400 kilometers east of Vancouver, and two mills in the town of Mackenzie which is at 750 kilometers east of Vancouver and 190 Km. north of Prince George, the closest urban centre. The location of the mills and their classification by region is shown in Appendix A . The cohort was folio wed-up until the end of 1990 and several sources were consulted for vital status ascertainment. The list of personal identifiers was linked by means of probabilistic record linkage methods to the death files of both the Vital Statistics Division of British Columbia and the Canadian Mortality Database. Secondary sources of ascertainment like pension plan, union, motor vehicle records and income tax records were consulted as well. As a result of these procedures, vital status was determined for 85.7% of the cohort members. Given the linkages with the two death databases and the income tax file, it seems highly unlikely that a significant proportion of the 3,791 workers for whom no vital status was confirmed were dead. 99 6.2.2 Cause of death - International Classification of Diseases The Division of Vital Statistics of British Columbia provided the underlying cause of death coded according to the International Classification of Diseases (WHO, 1977) codes in effect at the year of the death. The B.C. sawmill workers study spanned four decades and four revisions of the International Classification of Diseases. For this study, a reconciliation scheme devised by researchers from the British Columbia Cancer Agency was used (Appendix D). A total of 26 categories were used to identify deaths due to external causes (E800-E999). 6.2.3 British Columbia Coroner's Office files Information on the circumstances prior to or concurrent with the fatal incident was sought from the Coroner's Office. According to the B.C. Coroner's Act all sudden, unexpected deaths, are cases requiring investigation regarding the causes of death, the identification of the deceased and the place of death. The coroner's files contain the investigation reports and may include autopsy reports, police reports, Workers' Compensation Board reports and other information relevant to the fatal injury. A list with the names of the cohort members who died from external causes between 1950 and 1990 was produced. This roll was matched to the Coroner's records to retrieve the relevant investigation reports. 100 Data on factors related to the individual, the environment and the injury event abstracted from the BCCS files were recorded on forms designed specially for this study. 6.2.4 Occupational categories Workers were grouped into 4 broad categories considered to be relatively similar with respect to skill level required for job duties. Every worker was assigned to one occupational category according to the level of skill required in the current job (i.e. the one held at the moment of the injury event). The five categories (including an unclassifiable category) are: 1) Clerical, 2) Unskilled labour, 3) Skilled labour, 4) Maintenance, and 5) Unclassifiable. This scheme was developed by Aleck Ostry, with ample experience in this industrial sector, and has been used previously in this cohort (Ostry, et al., 2000). We adapted the original version of the classification by including the group "supervisor" in the category of "skilled labour". The purpose of using this skill-based classification was to categorize workers into groups that could reflect socio-economic differences (e.g. maintenance workers have higher salary and education than unskilled workers) that may affect the risk of fatal injuries. 6.2.5 Data Analysis Standardized Mortality Ratios (SMRs) were computed with indirect standardization (Hill, 1972). Expected numbers of deaths for the cohort members were calculated by applying age, sex, and calendar-specific cause of death rates for the British Columbia male population for the period 1950-1990, to the age, sex, and calendar-specific number of years of person-years contributed by the sawmill workers during the same period. Follow-up began January 1, 1950, for workers with one or more years of experience at that date. Follow-up began on the date of accruing 1 year of employment for all other workers. SMRs for all causes of deaths were calculated counting person-years at risk up to the date last observed and also up to the end of the study period (December 31 s t., 1990) to present both the least and the most conservative estimates. SMRs for injury mortality were calculated counting person years at risk either until the death date or up to 30 days after the date of termination from the mill whichever was earlier. This criterion was adopted to reduce the probability of counting deaths that may have occurred in other jobs, but also to avoid missing deaths that may have occurred as late effects or complications of injuries sustained while employed at a sawmill. Internal cohort analyses were carried out for two purposes: 1) to identify the job categories at highest risk of death from motor vehicle related injuries and suicide; and 2) to characterize the association between age, calendar period, region and seniority in the mill, and the risk of mortality from motor vehicle crashes and suicide. Rate ratio estimates were calculated using a multiplicative regression model ( Breslow and Day, 1987; Pearce, et al., 1988). The Poisson regression model for a rate ratio can be formulated as D/Do = exp ( b l X l + b2X2+.. .+bjXj) where D is the incidence rate for persons with specific values of X I , X2, .. . ,Xj , and Do is the baseline or reference incidence rate. The outcome variables analyzed were mortality from motor vehicle-related injuries and suicide. The independent variables were age, calendar, region of work-site, and seniority in the mill (i.e. length of employment in the mill). A l l these 102 covariates were included in the Poisson regression model as categorical variables (Table 6.1). Table 6.1. Categories of independent factors Age Calendar Region Seniority in mill 1=18-29 1=1950-1969 1= Vancouver Lower Mainland l=<7years 2=30-39 2=1970-1990 2=Vancouver Island & Coast Garibaldi 2=>= 7 years 3=40-65 3=Interior The six mills located in Vancouver Island, the one in Powell River (at 135 Kilometers northwest of Vancouver) and the one in Squamish (60 Km. north of Vancouver) were grouped in one region: Vancouver Island and Coast Garibaldi. The selection criteria for the covariate boundaries were based on an "a priori" concept of selecting one distinctive category against which the others will be referred ( e.g. young workers vs.adult workers vs. mature adult; etc.); and ideally with a large enough number of observations so that a stable estimate will result. Therefore, the distribution of the data was evaluated and the boundaries chosen according to the above criteria. For example, calendar period at risk and seniority were categorized in two periods each with similar proportions of person-years of observation (i.e. < 1970 accounted for 49% of person-years; and >=1970 for 51%; seniority < 7years= 47% person-years and >=7 years comprised the remaining 53% person-years). The Life Table Analysis System, LTAS version 1.0.c (Casinelli, et al., 2000) and EGRET (SERC, 1993) were used for SMR and Poisson regression analyses respectively. Ninety five percent confidence intervals were calculated. 103 6.3 Results 6.3.1 Demographic characteristics of the cohort The cohort included 26,411 sawmill workers from 14 sawmills in British Columbia. There were 4,813 deaths from all causes during the entire study period, 1950-1990. Of these, 689 were due to external causes (E-codes 800-999). For this study we analyze only the 207 deaths due to injuries (both on-the-job and off-the-job) that occurred to workers while actively employed at any of the study sawmills. Vital status information for the cohort members is shown in Tables 6.2 and 6.3. The former table illustrates the number and percentage of workers for whom vital status was known after linkage with both the provincial and national death databases plus the motor vehicle and pension plan records. It can be seen that by the end of the study period, a total of 13,479 workers were known to be alive, and 8,118 workers had unknown vital status. Table 6.2. Vital Status of the cohort members prior to linkage to the income tax file Vital Status No. % Alive 13,479 51.03 Unknown 8,118 30.74 Dead 4,814 18.23 Table 6.3. Vital Status of the cohort members after linkage to the income tax file Vital Status No. % Alive 17,806 67.42 Unknown 3,791 14.35 Dead 4,814 18.23 After the linkage with the Canadian income tax file, the number of workers with unknown vital status was reduced from 8,118 to 3,791 (30.74% vs. 14.35%) and the final number of workers 104 known to be alive by the end of follow-up, was 17,806 (Table 6.3). Thus, vital status was determined for 85.6% of the cohort. Given that the data were linked with the two mortality data sets (provincial and national) and the income tax files, it can be assumed that the likelihood of any of the workers with unknown vital status being dead is very low, provided that we had accurate personal identifying information. Demographic characteristics of the sawmill workers are presented in Table 6.4. The average worker was 27 years old when he started working at a mill. The mean length of employment was 10.4 years and 50% of the workers died at or before age 68. Among the 207 workers who died from injuries, the average age at death was 38 years. Table 6.4. Characteristics of the cohort of 26,411 sawmill workers 207 workers dead from injuries Mean Median Range Mean Median Range Age at hire 27 24 10 75 27 23 15 55 Year of birth 1935 1938 1877 1968 1933 1934 1890 1964 Length of employment 10.4 5.8 1 52 8.7 6.8 1 29.2 Age at death 66 68 18 102 38 37 18 65 Over 60% of the workers in this cohort worked and lived in the Vancouver Island and Coast Garibaldi region. (Table 6.5). Table 6.5. Location of mills and number of workers in each region Region No. mills No. workers % Vancouver 3 7,449 28.20 Vancouver Island and Coast 8 16,313 61.77 Interior 3 2,649 10.03 6.3.2 Overview of mortality Table 6.6 shows standardized mortality ratios (SMRs) for selected causes and all causes combined calculated for the original cohort study. 105 Table 6.6. SMRs for selected causes of death, among sawmill workers (n= 26,411), 1950-1990 Person-years to last known year Person-years to 1990 Cause of death Observed SMR 95% CI SMR 95% CI All causes 4814 0.99 0.96-1.01 0.81** 0.78 - 0.83 All cancers 1197 1.08** 1.01 - 1.14 0.89** 0.83 - 0.93 Specific cancer sites - Lung cancer 379 1.09 0.98-1.20 0.91 0.81 - 1.01 - Cancer of the lip 1 1.26 0.03 - 7.00 1.05 0.02- 5.81 - Cancer of oropharynx 0 0.00 — 0 — -- Cancer of larynx 16 1.36 0.77-2.20 1.11 0.63 - 1.81 - Cancer of esophagus 31 1.13 0.76- 1.59 0.92 0.62- 1.30 - Cancer of pancreas 63 0.96 0.73 - 1.23 0.8 0.61 - 1.02 All circulatory system diseases 2102 0.99 0.95 - 1.03 0.81** 0.77- 0.84 Specific circulatory diseases - Hypertension 34 0.78 0.53 - 1.08 0.66** 0.45 - 0.91 - Arteriosclerotic 1460 1.00 0.95 - 1.05 0.83** 0.78 - 0.86 - Cerebro-vascular 308 0.99 0.88- 1.11 0.78** 0.69- 0.87 Non malignant respiratory diseases 361 0.81 0.72 - 0.89 0.64** 0.57- 0.71 Specific non-malignant resp. diseases - Bronchitis & emphysema 85 0.78 0.62 - 0.96 0.64** 0.51 - 0.79 Cirrhosis of liver 79 0.71 0.55-0.87 0.59** 0.46- 0.73 All unintentional injuries 519 0.97 0.88- 1.05 0.82** 0.74- 0.88 Specific unintentional injury type - Motor vehicle-related 234 1.17* 1.02- 1.32 0.99 0.86- 1.12 - Water transportation 27 0.87 . 0.57-1.26 0.74 0.48 - 1.07 -Aircraft 23 1.01 0.63 - 1.50 0.84 0.53 - 1.26 - By machinery 23 1.48 0.93 - 2.22 1.26 0.80- 1.89 - Falls 53 0.90 0.67-1.17 0.70** 0.52- 0.91 - Drowning 28 0.89 0.59- 1.28 0.77 0.50- 1.10 - Poisoning 30 0.51 0.34-0.73 0.43 0.29- 0.61 - Caught between objects, unspecified 25 1.09 0.70-1.60 0.92 0.59- 1.35 - Struck by 23 0.81 0.51 - 1.21 0.69 0.43 - 1.03 - Burns 19 0.83 0.49-1.29 0.70 0.42- 1.09 Suicide 148 0.96 0.80-1.12 0.80** 0.67- 0.94 Homicide 21 0.82 0.50- 1.25 0.68 0.42- 1.04 War 1 3.12 0.08-17.31 2.18 0.05 - 12.1 *p<0.05;**p<0.01 Columns 2, 3, and 4, present the observed deaths, the SMR and the 95% confidence intervals respectively, based on counting person-years until the last known year alive. Columns 5, 6 106 and 7 show the results obtained when person-years were counted until the end of follow-up period (1990). Mortality from all causes combined was lower than that for men of corresponding age in the general population of British Columbia (SMR = 0.99 and SMR=0.81 depending on number of person-years counted). This result reflects the lower than expected mortality from diseases of the circulatory system, diseases of the respiratory, digestive and genitourinary system. A few rates were marginally above one and not statistically significant. SMRs for all cancers and motor vehicle-related deaths were above one if person-years were counted until date of last known alive but when person-years were added until 1990, the rates decreased. There is suggestive evidence that there were more machinery-related deaths in the cohort than in the general population (23 observed vs. 18 expected), with an SMR of 1.48 (C.I= 0.93 - 2.22) or 1.26 (C.I= 0.80-1.89) when person-years were counted until 1990. 6.3.3 Overall Mortality from Injuries Of the 689 deaths from external causes that were observed for the whole study period, only 207 occurred while the worker was actively employed at a sawmill. Standardized mortality ratios for the person-years of sawmill employment only are shown in Table 6.7. A total of 157 sawmill workers died from unintentional injuries and 50 died from intentional injuries. Overall, the sawmill workers experienced fewer deaths from injuries than the general male population of B.C. 107 Table 6.7. SMRs and crude rates for external causes of deaths among sawmill workers during sawmill employment only, 1950-1990+ Injury typef Observed SMRs 95% C.I. Rate/100,000 95% C.I All unintentional injuries 157 0.62** 0.53 - 0.73 71.05 60.68 - 83.20 Motor Vehicle 76 0.81 0.63 - 1.01 34.39 27.17-43.33 (8100-8299) By machinery 15 1.76* 1.00-2.90 6.79 3.80-11.20 (9270-9289) Drowning 11 0.65 0.32- 1.15 4.98 2.49-8.91 (9100-9109) Falls 10 0.55 0.26 - 100 4.53 2.17-8.34 (8800-8889) Water Transportation 9 0.55 0.25 - 1.05 4.07 1.86-7.73 (8300-8389) Poisoning 7 0.29* 0.07-0.71 3.16 1.27-6.51 (8500-8580,8600-8699) Caught between objects 6 0.57 0.20- 1.23 2.72 1.00-5.93 (9030-9039,9170-9189,9270-9289,9800-9899) Aircraft 5 0.43 0.14 - 1.01 2.26 0.73 - 5.27 (8400-8459) Burns 5 0.48 0.15 - 1.12 2.26 0.73-5.27 (8900-8999, 9230,9232,9239) Struck by 4 0.24** 0.07 - 0.62 1.81 0.49-4.63 (9160-9169) Late effects, misadventure 3 1.5 0.31 -4.38 1.36 0.28-3.97 (8700-8799,9290-9490) Suffocation 3 0.83 0.17-2.41 1.36 0.28-3.97 (9110-9139) Railway 1 0.2 0.01 - 1.10 0.45 0.01 -2.51 (8800-8079) Electrocution 1 0.57 0.02-3.17 0.45 0.01-2.51 (9250-9259) Caused by cutting object 1 2.76 0.07-15.36 0.45 0.01 -2.51 (9200-9209) All intentional injuries 50 0.63** 0.47 - 0.83 22.63 16.79-29.87 Suicide 44 0.65** 0.47 - 0.87 19.91 14.51-26.68 (9500-9599) Homicide 5 0.46 0.15 - 1.07 2.26 0.73 -5.27 (9600-9699) War 1 13.65 0.35 - 75.83 0.45 0.01-2.51 (9900-9998) + person-years were truncated at date of termination plus 30 days or death date * p O.05; ** p <0.01; f cause of deaths (ICD 9th Rev.) 108 There have been fewer than expected deaths from motor vehicle-related injuries, falls, drowning and poisoning which are the major causes of fatal injury. However, a statistically significant increased number of machinery-related deaths was observed in this cohort. (15 observed versus 8 expected). There were a couple of other SMRs above one but they were based on small number of cases (Late effects of injury, 3 deaths and SMR=1.5; and caused by cutting object, 1 death and SMR=2.76). The second last column of Table 6.7 presents sawmill workers' absolute risk of dying from injuries. The highest mortality rate from unintentional injuries was observed for motor vehicle-related injuries followed by machinery related and drowning. Motor vehicle injuries accounted for almost 50 % of all unintentional fatal injuries. The second largest proportion of deaths (10%), resulted from machinery-related injuries. However, when all deaths due to external causes (unintentional and intentional) are considered, the first cause of death was motor vehicle crashes and the second, suicide. 6.3.4 Motor Vehicle Crashes Seventy-six workers lost their lives in motor vehicle crashes. Records for 61 such deaths (80%) were found at the Coroner's Office. The degree of detail in the narrative of the circumstances of the fatal event varied from record to record, and this resulted in different counts for the variables described here. Motor vehicle crashes account for most of the deaths from external causes. The relative impact of this type of event varies across the life span. Table 6.8 shows that among the 109 youngest workers, 15-19 years old, motor vehicle related deaths accounted for 100 % of all unintentional injury fatalities. For the next age group (20-24 years old), the percentage diminished to 69% and remained about 50% until approximately the age of 40. At later ages the percentage diminishes except for the age group 50-54. Table 6.8. Motor vehicle related deaths as percentage of all unintentional injuries Age Groups MV No. % of all unint. 15-19 6 100.00 20-24 27 69.23 25-29 7 46.67 30-34 7 46.67 35-39 6 50.00 40-44 5 29.41 45-49 4 25.00 50-54 8 53.33 55-59 3 27.27 60-69 3 27.27 Datasource: BCSWC database (n=76) Factors related to the worker Figure 6.1 shows the distribution of fatalities for the sawmill workers and the B.C. male population by five-year age groups. In general, males in the general population were at higher risk of dying in fatal crashes than sawmill workers. However sawmill workers between 20 and 24 years old had higher rates than the general population. The age group 50-54 also showed a slightly higher rate. 110 Fig 6.1 Comparison of motor vehicle injury fatality rates between BC male population and sawmill workers by age group, 1950-1990 120.00 1 0.00 -I 1 1 1 1 , 1 , , 1 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 Age groups — • — B C Rate/100,000 pyears —•—Sawmill workers I Differences in crude fatality rates were found among the various age groups (Table 6.9) with young sawmill workers having the highest rates (68 deaths per 100,000 person-years). Table 6.9. Crude rates and adjusted rate ratios of motor vehicle fatalities among sawmill workers by age, seniority in mill, calendar period, and region, 1950-1990 Age groups Observed Rate/100,000f RR* 95% C.I. 15-29 40 68.23 1.00 — 30-49 22 21.57 0.38 0.21-0.67 50-66 14 23.21 0.48 0.23 - 0.98 Seniority in mill < 7 years 39 37.17 1.00 — >=7 years 37 31.87 0.69 0.38-1.24 Calendar period 1950-1969 30 27.46 1.00 — 1970-1990 46 41.17 1.27 0.78 - 2.07 Region Vancouver Lower 16 23.24 1.00 — Mainland Vancouver Island 52 38.07 1.60 0.88-2.91 and Coast Garibaldi Interior 8 51.58 1.85 0.75-4.55 •fCrude rates per 100,000 person-years * RR rate ratio adjusted for age and calendar period, referent level indicated by RR=1.00 111 This rate showed a reduction of over 60% when the drivers were older (30-49 and 50-65 years). The rate ratio analysis showed that the fatality rates for motor vehicle crashes were inversely related with age. Older workers ( 30-66 years) were significantly at lower risk of dying from motor vehicle crashes compared to the youngest workers (15-29 years). Seniority in the mill did not make a difference in the risk of dying from motor vehicle injuries. During the last two decades of the study, the crude motor vehicle fatality rates were higher than from 1950 to 1969. Poisson regression analysis did not show a significant difference in risk between the two time periods (Table 6.9). In terms of specific geographic area, the highest crude fatality rate was found among workers living in the interior of B.C. However, no significant risk differences were detected among the three regions when rate ratios were calculated. The highest number of motor vehicle crashes was observed in the group of unskilled labour workers and the lowest in the maintenance workers group (Table 6.10). Table 6.10. Crude rates and adjusted rate ratios of motor vehicle crashes among sawmill workers by occupational groups, 1950-1990 Observed Rate/100,000f RR* 95% C.I. Occupational group Skilled labour 18 28.49 0.45 0.23-0.88 Unskilled labour 33 38.11 1.12 0.70- 1.79 Maintenance 13 36.77 1.53 0.85-2.73 Clerical 0 0 — Unclassifiable 12 31.73 1.29 0.74 - 2.22 t Crude rate per 100,000 person-years * RR rate ratio relative to the entire cohort (excluding the category considered) adjusted for age and calendar period 112 Rate ratio estimates showed that the group of skilled labour workers was significantly at lower risk of fatal motor vehicle crashes than the rest of the cohort. The analysis did not detect risk differences between the other occupational groups and the rest of the cohort. Information on blood alcohol level was available for 36 fatalities (47%). Of these, 21 (58%) had blood alcohol level above 0.08% (legal limit for driving), 5 had less than 0.08% and in 10 cases no alcohol was detected. Factors related to the environment Motor vehicle fatalities were more frequent during nighttime. The time of crash incident was recorded for 48 motor vehicle fatalities. Of these, 33 crashes (68%) occurred between 6:00 pm and 5:59 am. with a peak from 2:00 am to 2:59 am. Fatal crashes were more likely to occur on the weekends (63%), particularly on Saturdays (Table 6.11). Table 6.11. Rate of fatal motor vehicle crashes by day of occurrence, 1950-1990 Day Observed Rate/100,000 Sunday 15 6.79 Monday 6 2.72 Tuesday 5 2.26 Wednesday 7 3.17 Thursday 10 4.53 Friday 11 4.98 Saturday 22 9.96 Datasource: BCSWC database (n=76) The highest rates of motor vehicle crashes were observed between 6:00 pm and 5:59 am on Friday and Saturday nights. (Figure 6.2). 113 F i g . 6 . 2 M o t o r V e h i c l e R e l a t e d D e a t h b y T i m e a n d D a y o f C r a s h -•— 12:00AM-5:59AM —•—6:AM-11:59 AM —A— 12:00PM-5:59PM —K—6:00PM-11:59PM The season with the highest rate of fatal crashes was summer (Fig. 6.3 ) Fifty-five percent of the fatalities occurred from July to September. F i g . 6 . 3 Rate of M V crashes b y season .-ears 16 14 erson-j 12 10 8 a. o 6 o 4 o 2 0 £ J a n u a r y - April-June July- October-March September December The analysis by mill identified workers from mills # 4, 9 and 15, all in isolated rural communities, as having the highest frequency of fatal motor vehicle crashes. Workers from mill # 4 appeared to have a significantly higher risk of dying from motor vehicle injuries than the rest of the cohort (Table 6.12). There were no deaths from motor vehicle crashes among workers from mill #13. 114 Table 6.12 Crude rates and adjusted rate ratios of motor vehicle crashes among sawmill workers by mill, 1950-1990 Observed Rate/100,000f RR* 95% C.I. Mill 1 7 36.38 1.24 0.56-2.70 2 10 51.79 1.27 0.91 - 1.77 3 5 23.88 0.91 0.67- 1.24 4 13 60.88 1.17 1.01 - 1.36 5 4 13.86 0.84 0.69- 1.08 6 6 21.94 0.92 0.80-1.05 7 1 24.55 0.92 0.69-1.22 8 4 22.25 0.94 0.83 - 1.06 9 5 51.30 1.03 0.93 - 1.14 10 4 32.29 0.98 0.88 - 1.08 11 2 40.74 1.02 0.90- 1.16 15 6 94.90 1.06 0.99- 1.13 16 9 37.33 1.01 0.96- 1.06 f Crude rates per 100,000 person years * RR rate ratio relative to the entire cohort (excluding the mill considered) adjusted for age and calendar period Factors related to the injury event Fifty-five % of all motor vehicle related fatalities occurred when the worker was an occupant of a car, either as a driver (32 incidents) or as a passenger (10 incidents) (Figure 6.4). Fig. 6.4 Motor Vehicle Crashes by Road Users Bicyclist 1% Unknow n 20% Motorcyclist r x 14% Ftedestr ian^l— mm w W MV occupant 56% 115 Table 6.13 shows the distribution of fatalities by type of road user and age at the moment of the incident. Table 6.13 Motor vehicle deaths by road user type and age group Age group Road User 18-24 25-44 45-64 Bicyclist 1 Motorcyclist 5 6 MV occupant 21 10 11 Pedestrian 1 3 3 Unknown 6 6 3 Datasource: BCCS records The least common events involved a worker riding a bicycle (n=l) or walking (n=7). Motorcycle crashes occurred to workers up to 44 years old and not at later ages. As pedestrians, workers were at risk of fatal injuries at all ages. Workers involved in motor vehicle crashes were most likely to die from severe injuries to the head. Table 6.14 shows the frequency of fatalities according to the nature of injury. Table 6.14 Deaths by nature of injury, fatal motor vehicles crashes, 1950-1990 Nature of injury No. % Skull fracture 26 34 Internal injury of chest, abdomen, 23 30 and pelvis Intracranial injury 13 17 Neck and trunk 6 8 Blood vessels injury 3 4 Drowning 2 3 Unknown 3 4 Datasource: BCCS records 116 Skull fracture and intracraneal injuries were reported as the main causes of death in 51% of the incidents. The highest rate of deaths from severe injury to the head was noted during the last decade (1980-1990). (Table 6.15). Table 6.15 Head injuries as cause of death by decades No. Rate/100,000 1950-1959 5 9.86 1960-1969 10 17.08 1970-1979 14 19.04 1980-1990 10 26.19 Forty-five workers, out of the 61 for whom data were available, were dead at the scene of the vehicle crash when aid was available and only 11 were alive. In five cases this information was not recorded. Data on contributing factors are presented in (Fig.6.5). Alcohol was mentioned as a contributing factor to the death of the worker, in 38 investigation reports. Excessive speed was noted in 17 reports and environmental factors (e.g. road or weather conditions) in 14 reports. Scarce data was available on the use of seat belts (n=2). Fig.6.5 Contributing factors to motor vehicle crashes, 1950-1990 Fatigue 117 Table 6.16 shows the frequency of fatalities by type of crash. Table 6.16 Types of motor vehicles crashes, 1950-1990 Type of impact car crashes No. % of car crashes % of all M V crashes One car Collision with object 13 31.1 17.11 Roll over 11 26.2 14.47 Two cars Head-on 14 33.3 18.42 Angle 2 4.8 2.63 Three or more cars 1 2.3 1.32 Unknown 1 2.3 1.32 Subtotal car crashes 42 55.26 Type of impact motorcycle crashes No. % o f motorcycle crashes Collision with car/truck 7 63.6 9.21 Collision with object 3 27.3 3.95 Roll over 1 9.1 1.32 Subtotal motorcycle crashes 11 14.47 Bicycle collision with car 1 1.32 Pedestrian hit by car 7 9.21 Unknown 15 19.74 Total motor vehicle-related fatalities 76 Datasource: BCCS records Twenty four deaths occurred from single-vehicle crashes and 16 from crashes involving two motor vehicles. For single-vehicle events, the most frequent collision was with a fixed object. Roll over caused 11 deaths. Head-on collision was the predominant type of crash involving 2 motor vehicles. There was only one incident involving more than 3 motor vehicles. Light 118 pole was the object most frequently hit by motor vehicles (Table 6.17). Other structures like post (n=2) or bridge (n=2), were involved less frequently. T a b l e 6 . 1 7 C o l l i s i o n w i t h r o a d s i d e o b j e c t s , f a t a l c r a s h e s , 1 9 5 0 - 1 9 9 0 Light pole 6 Post 2 Bridge 2 Ditch 2 Embankment 2 Trees 2 Rock 1 Lake 1 Datasource: BCCS records 6.3.5 Suicide A total of 44 cases of suicide were identified among sawmill workers between 1950 and 1990, making this the second leading cause of death from external causes. Coroner's records were found for 38 of those deaths (86%). Factors related to the worker Overall, for all ages combined, fewer suicides were reported among sawmill workers than in the general male population of B.C. However there were two age groups (20-24 and 25-29 years) that had more deaths than expected (Figure 6.6). Young adults (18-29 years old) and old workers (50-65 years old) were more likely to commit suicide than those in the middle age group. The highest crude rate, 29 deaths per 119 100,000 person-years, was recorded among workers between 18 and 29 years old (Table 6.18). Fig. 6.6 Comparison of suicide rates between BC male population and sawmill workers by age group, 1950-1990 45 40 35 H 30 25 20 15 10 5 0 15-19 20-24 25-29 30-34 35-39 4044 Age groups 45-49 50-54 55-59 60-69 • B.C Rates — -H— Sawmill workers Table 6.18 Crude rates and adjusted rate ratios of suicide among sawmill workers, by age, seniority in mill, calendar period, and region, 1950-1990 Age group 15-29 30-49 50-65 Seniority in mill <7 years >= 7 years Calendar period 1950-1969 1970-1990 Region Vancouver Lower Mainland Vancouver Island and Coast Garibaldi Interior Observed Rate/100,000f RR* 17 14 13 18 26 14 30 31 5 29.00 13.72 21.55 17.15 22.4 12.81 26.85 11.3 23.36 28.61 1.00 0.38 0.55 1.00 1.98 1.00 1.57 1.00 1.8 2.41 95% C.I. 0.17-0.83 0.23 - 1.31 0.95-4.14 0.80-3.04 0.82 - 3.96 0.76 - 7.64 t Crude rates per 100,000 person-years * RR rate ratio adjusted for age and calendar period, referent level indicated by R R : 1.00 120 Comparison of suicide risk among age groups showed that the youngest workers are at significantly higher risk than the group of workers aged 30-49. Suicide was more frequent among workers with more seniority in the mill. The crude suicide rate was 30% higher in workers with 7 or more years of seniority. The rate ratio showed a doubling of risk in the more senior workers, but this result was not statistically significant. The number and crude rate of suicides from 1950 to 1969 doubled during the period 1970-1990. Rate ratio estimates showed no difference between the two periods. Crude suicide rates were highest among workers residing in the interior and Vancouver Island and Coast Garibaldi regions. The results of comparing suicide risk among regions do not indicate difference ( wide confidence intervals including one). Maintenance workers had the highest crude rate of suicides (34 deaths per 100,000 person-years). The highest proportion of suicide (36%) occurred among unskilled workers although their death rate was about half the rate of the former group. Compared to the rest of the cohort, the groups of skilled labour workers were significantly at lower risk of committing suicide. No differences were apparent between the other occupational groups and the rest of the cohort (Table 6.19). 121 Table 6.19 Crude rates and adjusted rate ratios of suicide among sawmill workers, by occupational groups, 1950-1990 Observed Rate/100,000f RR* 95% C.I. Occupational groups Skilled labour 8 12.91 0.38 0.17-0.83 Unskilled labour 16 18.88 0.77 0.44- 1.33 Maintenance 13 34.89 1.25 0.65 - 2.43 Clerical 0 0 — Unclassifiable 7 27.6 2.03 1.24-3.32 t Crude rates per 100,000 person-years * RR rate ratio relative to the entire cohort (excluding the category considered) adjusted for age and calendar period Blood level of alcohol was reported for 24 suicides. Of these, on five occasions the level of alcohol was below 0.08%, 6 were above that limit and no alcohol was detected in 13 cases. Factors related to the environment Information on the time of the occurrence of the incident was available for 24 suicides. Of these, the majority (66%) occurred between 6:00 am and 6:00 pm. More suicides were recorded on Mondays than in other day of the week (Table 6.20). Table 6.20 Rate of suicides by day of occurrence, 1950-1990 Day Observed Rate/100,000 Sunday 7 3.17 Monday 11 4.98 Tuesday 6 2.72 Wednesday 5 2.26 Thursday 8 3.62 Friday 6 2.72 Saturday 1 0.45 Datasource: BCSWC database 122 The trimester with the highest percentage of suicides was April-June (Figure 6.7). Fig. 6.7 Suicide rates by trimester 10 8 -6 -4 -2 -0 January-March April-June July-September October-December — • — Series 1 3.17 9.05 2.72 4.98 Analysis by mill indicates that the highest suicide rates occurred among workers from mills 8, 9, 15 and 16. Interestingly, the rate ratio estimates indicate that workers from mills 9 and 15 were significantly at higher risk compared to the rest of the cohort members (Table 6.21). These mills are located in rural towns. There were no suicides among workers from mills 5, 7, 11 and 13. Table 6.21 Crude rates and adjusted rate ratios of suicide among sawmill workers by mill, 1950-1990 Observed Rate/100,000f RR* 95% C.I. Mill 1 5 25.99 1.90 0.80-4.53 2 4 20.71 1.04 062- 1.74 3 3. 14.33 0.79 0.48-1.26 4 3 14.05 0.91 0.68-1.22 6 4 14.63 0.94 0.79-1.12 8 5 27.81 1.05 0.93 - 1.18 9 5 51.30 1.11 1.00-1.22 10 2 16.14 0.96 0.83 - 1.11 15 5 79.08 1.10 1.02- 1.18 16 8 33.18 1.05 0.99-1.11 fCrude rate per 100,000 person years * RR rate ratio relative to the entire cohort (excluding the mill considered) adjusted for age and calendar period Almost 50% of the workers who took their lives had a history of mental health problems. Family problems (family or relationship breakup in the last year) were the second most frequently reported 123 condition (36%) coincident with the suicide. Twenty two percent had a history of substance abuse (Table 6.22). Table 6.22 Conditions associated with Suicide Conditions No. Mental Health Disorders 21 Family problems 16 History of substance abuse 10 Economic problems 7 Terminal Disease 3 Legal problems 2 Other health problems 2 Unknown 6 Economic problems were identified as relevant to the fatal outcome in 15% of the cases. Among these were 5 workers who had been laid off. Four of the deceased had talked before of suicide and 3 left suicide notes. Factors related to the injury event Shooting was the most common method of committing suicide (48%), followed by poisoning (36%) and hanging (14%) (Figure 6.8). Fig. 6.8 Suicide by Methods Jumping from elevation 2% Hanging 14% 124 The different age distribution of suicide methods is shown in Figure 6.9. Fig. 6.9 Suicide rates by methods and age groups,1950-1990 18-29 30-39 40-49 50-59 60-65 Age group • Gunshot • Hanging • Poisoning Table 6.23 shows the distribution of suicides according to method and decade. Table 6.23 Suicide rates by method, sawmill workers, 1950-1990 Method Decade No. Rate/100000 Gunshot 1950-1959 1 1.97 1960-1969 4 6.83 1970-1979 10 13.60 1980-1990 6 15.72 Hanging 1970-1979 2 2.72 1980-1990 4 10.48 Jump window 1980-1990 1 2.61 Poisoning 1950-1959 4 7.89 1960-1969 5 8.54 1970-1979 3 4.08 1980-1990 4 10.48 Datasource: BCCS records From 1950 to 1969 the suicide rates from poisoning were higher than those from firearms. This pattern was reversed during the last two decades. Data on the type of firearm used was available for 13 fatalities out of the 21 reported. Rifles were used in 12 incidents and shotgun in 1. 125 6.3.6 Falls, drowning, and machinery-related fatalities Sixty percent of mortality from falls occurred among workers between 50 and 65 years old. Both the crude mortality rates and the SMRs were highest among the oldest workers, aged 50-65 years. Increased crude rates and SMRs for fatal falls were observed during the second half of the study period compared to the first half (Table 6.24). Table 6.24 SMRs for falls, drowning, and machinery related by age group and calendar period, sawmill workers, 1950-1990 Falls Drowning Machinery-related Age groups Observed Rate* SMR Observed Rate* SMR Observed Rate* SMR 15-29 0 0 0 3 5.12 0.53 3 5.12 1.26 30-49 4 3.92 0.61 7 6.86 1.06 9 8.82 2.39 50-66 6 9.95 0.66 1 1.66 0.21 3 4.97 1.27 Calendar perio d 1950-1969 3 2.75 0.32 4 3.66 0.37 10 9.15 1.65 1970-1990 7 6.27 0.79 7 6.27 1.11 5 4.48 2.04 *Rate per 100,000 person-years Fatal falls occurred with a similar frequency among workers living in Vancouver (5.65 per 100,000 person-years) and those living in Vancouver Island and Coast Garibaldi (4.52 per 100,000 person-years). No deaths from falls were reported for workers living in the interior of the province (Table 6.25). Table 6.25 Crude mortality rates from falls, drowning and machinery-related by worksite region, sawmill workers, 1950-1990 Falls Drowning Machinery-related Region Observed Rate* Observed Rate* Observed Rate* (95% C.I.) (95% C.I.) (95% C.I.) Vancouver 4 5.65 2 2.82 4 5.65 Vancouver Island (1.54- 14.46) (0.34- 10.18) (1.54- 14.46) and Coast 6 4.52 (1.66 - 9.85) 8 6.03 (2.60 - 11.88) 11 8.29 (4.14-14.84) Interior 0 0 (0-26.18) 1 5.72 (0.14-31.86) 0 0 (0-26.18) *Rate per 100,000 person-years 126 Regarding place of occurrence, 5 fatal falls happened at home and 5 at work. Mortality rate from drowning was highest for workers aged 30-49. Drowning showed a calendar trend similar to falls. There was approximately a twofold increase in drowning death rates from 1950-1969, to 1970-1990 (3.66 and 6.27 deaths per 100,000 person-years respectively). The SMRs indicate the same calendar trend. Workers living in Vancouver Island and coast Garibaldi were at higher risk for drowning than workers residing in the Lower Mainland and interior regions of B.C.. Nine of the eleven drowning deaths occurred in recreational settings and two at the workplace. The highest rate of deaths from machinery-related injuries was for the group of workers aged 30-49. The SMRs calculated for various age group indicate that this workforce had more machinery-related deaths than expected based on general male population rates. Sawmill workers between 30 and 49 years had a risk of dying from machinery-related injuries 2.4 times that of the males of the same age group from the general B.C. population (SMR= 2.39; 95% C.I.= 1.09 - 4.54). There were more machinery-related deaths from 1950 through 1969 than during 1970-1990; yet the highest SMR was observed during 1970-1990. Fatal injuries from contact with machinery were more frequent among workers from Vancouver Island and coast Garibaldi than for workers living in other regions. There were no deaths from this cause among workers residing in the interior of the province. 127 6.4 Discussion Overall unintentional injury mortality We examined the mortality from injuries in a cohort of 26,411 sawmill workers from 1950 to 1990. The overall injury mortality rate for this period was 71.05 per 100,000 person-years. Compared to the mortality rate in the male general population for the same period, the 157 deaths observed among the sawmill workers were 37 percent fewer than expected. This finding is in line with the lower overall mortality observed in the analysis of the original study. The all cause standardized mortality ratio was below unity (SMR for all causes of death = 0.81) and illustrates the healthy worker effect. The healthy worker effect was expected because this cohort represented physically fit individuals selected because the nature of most mill tasks is characterized by high physical demands. Although the healthy worker effect is strongest during the first years of employment and then decreases, it is also more pronounced in workers who stay longer in their jobs (survival of healthier workers). The effect is known to disappear rapidly after termination of employment, especially if the worker retires before the usual age for retirement (Checkoway, et al., 1989; Fox and Collier, 1976). It has been indicated that part of the healthy worker effect may be due to the use of an inappropriate comparison population (Wen and Tsai, 1982). The general population includes those who are not fit to work due to illness as well as those who are employed. Therefore, the expected number of deaths is calculated based on the rates of a population that combines exposed and unexposed individuals. Internal cohort analyses were conducted to remove the 128 differences in employment and socioeconomic status associated with the use of the general population and reduce the bias due to the healthy worker effect (Wen and Tsai, 1982). Regarding mortality from external causes, the healthy worker effect is not expected to affect the injury mortality rates. Contrary to other reports of injury mortality among forestry workers, our cohort had fewer deaths from injuries (even when work-related injuries were included) than the B.C. male population. Previous reports (Millar and Adams, 1991) have indicated that active workers are at greater risk of motor vehicle injuries than unemployed people. Notkola et al. (1993) in a mortality study in Finnish forestry and construction workers from 1970-1985, reported a significant increased mortality rate from all injuries among forestry workers compared to the rest of the working population (R.R 2.17; 95% C.I. 1.98 - 2.39). The authors could not determine whether or not the excess of injury deaths was driven by work-related incidents because death data were obtained from death certificates. Similarly, Asp (1994) and Green (1991) reported non significant increased standardized mortality ratios (SMR 1.13 and 1.09 respectively) for all unintentional injuries among workers exposed to chlorophenoxy herbicides. In the Poisson regression analyses the injury mortality rate for each occupational group was compared to the mortality rates of the rest of the cohort members. This showed that except for the category "skilled labour" there were no significant differences among the occupational groups regarding risk of fatal injuries for the two selected outcomes (i.e. motor vehicle crashes and suicide). 129 Mortality from motor vehicle crashes The pattern of motor vehicle crashes found in this study is consistent with that reported for Canada and U.S population (Waller, 1985, Riley and Paddon, 1990). Young drivers, aged 15-24, had the highest fatality rates. The peak of fatal crashes during the summer may be the result of increased volume of traffic and length of time driving during this season, as well as increased numbers of young drivers on the road at that time. The high frequency of fatalities at nighttime during the weekends may reflect the problem of driving while impaired (Wilson, 1984). The trend in mortality across the forty year period according to the nature of injury did not show decreases in skull fractures, intracraneal injuries or internal injuries to the chest, abdomen and pelvis. In contrast Millar and Last (1988) in an analysis of motor vehicle fatalities in Canada during 1921-1984, concluded that the shift from skull fractures to intracraneal and internal injuries of the chest, abdomen and pelvis reported during the period 1970-1980, could be consistent with an increased use of seat belts among the drivers. In this study, the mortality rates from skull fractures and intracraneal injuries increased from 1950 to 1990. Internal injuries of the chest, abdomen and pelvis increased from 1950 to 1980 and then decreased during the last decade. It is known that compliance with seat belt use laws is far from complete. In Canada there are notable differences in seatbelt use between provinces with and provinces without legislation mandating seatbelt use. The average percent of users of seat belt in provinces with relevant legislation was 61.4 (B.C. 67.4) and for provinces without, was 15.0 (Millar and Last, 1988; Last, 1984; Minister of Supply and Services and Minister of National Health and Welfare, 1981). 130 The results of the internal cohort analyses have confirmed the relationship between age and town of residence and mortality from motor vehicle crashes. The youngest workers (15-29 years) were significantly at higher risk of fatal motor vehicle crashes (68.23 per 100,000 person-years) than the older workers, consistent with the pattern observed at both the provincial and national level. The overrepresentation of young drivers in fatal crashes has been attributed to several reasons. One is that young drivers are new and inexperienced drivers. Another reason is that young drivers are more frequently exposed at times and locations posing greater safety hazards. In addition, young people are more sensitive to small amounts of alcohol than older, experienced drinkers (Waller, 1985). It has been reported that the risk of crashing increases as blood alcohol level exceeds 0.05 percent by weight (U.S. Department of Transportation, 1968; Jones and Joscelyn, 1978). For young people the risk of motor vehicle crash increases with even one drink and thereafter it rises at a faster rate than for drivers over 25 or 30 years (Hyman, 1968). In terms of behavioral factors, the risk of fatal crashes seems to be closely related to feelings of hostility and rebellion, especially among male drivers (Pelz and Shuman, 1971). It has also been suggested that the excess of car crashes among young males may be attributed to men's tendency to drive faster, violate motor vehicle traffic regulations, drive while impaired and without seat belts (Baker, et al., 1992; Waters, et al., 1993; Millar and Last, 1988). In this study, 21 workers out of the 36 tested had blood alcohol levels above the legal limit for driving (0.08 %) and 5 below. No alcohol was detected in 10 fatalities. Since blood alcohol tests are done when it is considered relevant to the event, it may be that the cases 131 tested were the more likely to give a positive result. In 1993, it was reported that alcohol was a contributing factor in approximately 31 % of all motor vehicle-related fatalities in B.C. (BC Ministry of Transportation and Highway, 1993). Because of the link between motor vehicle crashes and alcohol, several causes of death known to be associated with alcohol were examined to ascertain whether or not there was evidence of excessive alcohol consumption among the sawmill workers. SMRs for oral and laryngeal cancers and cirrhosis were below unity, suggesting that sawmill workers consume less alcohol than the general population. The association of fatal motor vehicle crashes and residence in several of the rural isolated mill towns also agrees with the provincial pattern of injury mortality. Most of the mills in our study are located in rural areas or newly developed urban centers rather than in well established cities. Fatal motor vehicle crashes occur at higher frequency in rural areas where more miles are driven and at greater speed, the roads are in poor conditions, the distance to medical facilities is greater, and seat belt use is lower (Waller, 1985; Baker, et al.,1992). Freeways have fewer crash fatality rates than primary and secondary roads, mainly because their sides have been generally cleared of trees, rocks or other hazards but also because there are no intersections. Suicide Forestry workers have been identified as a workforce at high risk of suicide in several reports (Tuk and Macdonald, 1995; Notkola, et a l , 1993; Asp, et al., 1994; Green, 1991). However, both the study by Asp et al. and the one by Green included exclusively chlorophenoxy acid herbicides applicators. Green suggested that the increased suicide rates could be related to 132 neurological toxicity resulting from exposure to phenoxy acid herbicides. Notkola, et al., studied mortality in forestry workers but the authors did not provide details of the occupational groups included in the broad category of forestry workers and thus is not possible to know whether sawmill workers were part of the study. Tuk and Macdonald also refer to "forestry and logging" as the industrial group with the highest suicide rate among all industries in B.C. It implies that the category may include sawmill workers as well as loggers and workers in other forestry-related activities. In this cohort, sawmill workers had higher suicide rates than the B.C population only at ages 20-29. Similar to the pattern of the general population of B.C. suicide ranked second as cause of death among persons 15 to 24 years of age, following motor vehicle crashes. Factors that have been found to contribute to the risk of suicide include psychiatric conditions, substance abuse, genetic and family background, sociological and biological factors, and stressful life events (Health Canada, 1995). Some common stressful situations arise from breaking up with a girl friend, problems at school, arguments with parents. This is the age of entry into the job market as well. The level of stress triggered by these circumstances varies with the individual personality as well. Social factors such as alcohol consumption seem to play the role of facilitator for the act. Besides, alcohol consumption is likely to be associated to other risk factors such as depression or other mental illness (National Committee for Injury Prevention and Control, 1989). In this cohort, the leading conditions noted as relevant to the suicides were mental health disorder, history of substance 133 abuse and family and economic problems. This cohort experienced massive layoffs during the recession of the early 1980's when the labour force was reduced by 50 percent. Geographic differences in suicide rates have also been reported in U.S, with rural areas having the highest rates (Baker, et al., 1992). One possible explanation for this is that in rural areas the availability of firearms may be greater for hunting purposes. Another possible explanation may be related to some qualities of rural life, particularly loneliness (National Committee for Injury Prevention and Control, 1989). In this cohort the highest suicide rates (79.08 and 51.30 per 100,000 person-years) were observed in the two most isolated mill communities. These rates are far greater than the ones reported during the periods of great economic depression of the late 1920s and early 1930s (17 per 100,000 person-years) (National Committee for Injury Prevention and Control, 1989; Health Canada, 1995). Three mill communities shared special socio-demographic features that make them different from the other mill locations; these may be related to the results. These three are mill towns that are totally dependent on the labor force needs of the company that built the town (Marchak, 1990). Their survival is subordinate to the economic success of one particular company. Adverse market conditions, technology changes, depletion of the natural resources are the major threats for these communities. Marchak (1990) provides an example of forest-based towns describing the socio-demographic characteristics of Mackenzie. This town, situated in the northeast of British Columbia, was constructed in the mid 1960's by a large company mill. Although the average income of the workers is high, the population seems to have no strong ties with the town. There are few long-term residents. This instability has 134 been ascribed to the following circumstances: 1) mill workers are known to experience frequent lay-offs; 2) geographical isolation of the town (i.e. Mackenzie is 190 K m from the nearest urban center); and 3) lack of opportunities for women and children (Marchak, 1990). Possible limitations of this study include the number of workers for whom vital status was not known. In this cohort, after the linkage with both the provincial and national mortality databases and the income tax records, vital status was ascertained for 85.6 % of the cohort. Given that death ascertainment through Statistics Canada's mortality database is estimated to be 97.6%, it can be stated that the likelihood that workers with unknown vital status are dead is very low. In addition, injury mortality must be reported to the Coroner. One possible explanation for the cases with unknown vital status is that they may have moved out of the country. Another limitation relates to the number of workers who could not be assigned to specific job categories. Instead of deleting these workers, we created a category "unclassifiable" so that all person-years were used in the internal cohort analysis (Checkoway, et al., 1989). The intent of this skill-based classification was to categorize workers into groups that could somehow reflect socio-economic differences (e.g. maintenance workers have higher salary and education than unskilled workers). The results seem to indicate that there are no significant differences among occupational groups regarding the risk of fatal injuries either from motor vehicle crashes or suicide. The exception was the group "skilled labor" that had significantly fewer deaths from both motor vehicle crashes and suicide. Such outcomes may be related to particular characteristics of the skilled labour group. The fact that this group of 135 workers lies between the unskilled labour group and the maintenance group (in terns of skills and income), suggests that the "skilled labour" workers may be at a comfortable work stage. This group may experience lower stress level than unskilled workers. Karasek and Theorell (1990) have found that if the worker's skill is developed and used, the worker is more likely to feel in control of the situations that may arise. Job induced skill development results in improved confidence and self-esteem thus reducing stress response. In contrast, unskilled workers having unstable jobs (i.e. great turnover), with little or no control over tasks are more susceptible to experience more dissatisfaction and stress. Maintenance workers, on the other hand, having a formal education and being the best paid in the industry, have more responsibility. While their work is not repetitive and monotonous, they are pressured to fix problems as quickly as possible without interrupting the production flow. Maintenance work is often done during break times or night shifts. The degree of stress thus can be high and exceed the person's resilience. It has been suggested that the risk of personal injury is affected by socioeconomic factors, but this association seems to be complex. The following findings from the General Social Survey illustrate this relationship. Motor vehicle injuries were highest among working people. Also, for men, the highest risk of motor vehicle crashes was observed among individuals with some postsecondary education. In general, income level and injury events appear to be directly associated. People with the highest income had the highest motor vehicle-related injury rates. Additionally, married people tended to have the lowest injury rates (Millar and Adams, 1991). 136 The understanding of the causes of the different risk of injuries among an otherwise homogeneous industrial group would require knowledge of other important determinants of health such as housing, social support, or life-style (Evans, et al., 1994). Finally, the study could not analyze the contribution of other factors (e.g. from the physical and social environment). The coroner's files contain the investigating coroner's report of the events surrounding each event. These were based on the accounts of individuals who knew the deceased, relatives and friends or coworkers. Although the documentation provided useful information describing the scenario of the injury event, the thoroughness of each investigation and or completeness of the coroner's report is not consistent. Therefore many conditions relating to personal and environmental factors associated with the occurrence of fatalities are likely to have been underreported. Also it is not possible to develop rates for these factors since no denominator data are available. The strengths of the study are the length of the follow-up, the large number of workers included, and the availability of information from the Coroner's records that are the best source of information on pre-event and event phases of injury fatalities. The most important feature of this study is that it provides mortality rates for cause-specific injury mortality. This has allowed us to discriminate the risk of fatal injuries in different settings. Relationship of this study with the chapter on work-related injuries This study completes the characterization of the pattern of mortality from injuries among sawmill workers. It allowed us to quantify the magnitude of both the absolute and relative risk of fatal injuries in different settings in a cohort of blue collar workers. These analyses have provided some interesting results. First, the overall mortality rate from unintentional injuries is unacceptably high (71.05 per 100,000 person-years) given that most injuries are preventable. Comparatively, the rate is lower than the one observed in the general male population of B.C.( SMR= 0.62). This deficit in injury mortality is mainly due to the reduced mortality from motor vehicle crashes, falls, and drowning that represent the three leading causes of death in the general population. Second, machinery-related injuries with a mortality rate of 6.79 per 100,000 person-years was the second leading cause of unintentional injury deaths in this cohort, and the only cause with significantly more deaths than expected. According to this analysis, sawmill workers experienced 76% more machinery-related deaths than the general population. This was expected given the nature of the production processes typical of a mill including exposure to various safety hazards including different types of saws, conveyor belts, chippers, planers, and mobile equipment such as forklift, and lumber carriers. Third, surprisingly, the only setting where this workforce had more fatal injuries than the general population was in the workplace. While the increased mortality (compared to the general population) from work-related injuries was expected, the lower mortality from other types of injury was not. If the occurrence of a work-related fatal injury were attributable solely to personal factors such as risk-taking behavior, carelessness, or lifestyle characteristics, we would expect greater mortality rates for other injury types as well. These results seem to indicate that work-related, not personal, factors play the decisive role in the occurrence of injuries at work. 138 CHAPTER SEVEN 7.0 Conclusions The primary purpose of this research was to advance our knowledge of the epidemiology of injuries. The injury mortality among a cohort of blue collar workers was analyzed in terms of to whom, how, when and where injuries occur with the purpose of understanding why injuries happened and what can be done to prevent future deaths. In addition, the usefulness of the data sources available for injury surveillance was explored. In addition to the limitations, strengths, and conclusions that were discussed with respect to each research objective in the preceding chapters, the following discussion addresses key issues influencing the results and interpretation of this study. 7.1 Completeness of ascertainment of work-related fatal injuries Epidemiologic surveillance is defined as the ongoing and systematic collection, analysis, and interpretation of health data in the process of describing and monitoring a health event. Surveillance data are essential both to determine the need for public health action and to assess the effectiveness of programs (Klaucke, et al, 1988). In this study, although the evaluation of completeness of the data sources using capture-recapture methods was limited due to the small number of fatalities and the interdependence of the data sources, it was possible to identify the types of cases that will be systematically missed by each source due to the data collection procedures. The specific limitations of using BCCS or B C W C B records to account for all fatal injuries occurring at a workplace have been documented. 139 The use of death certificates to identify work-related fatal injuries was assessed. The selection of the item "place of injury" on the death certificate was suggested by an officer from the Division of Vital Statistics of B.C. because Vital Statistics was interested in validating the accuracy of this item. The results indicated better specificity than sensitivity when using this source of ascertainment. Given that there is no universally accepted gold standard against which to compare each source, it is difficult to assess the degree of underreporting of fatal injuries occurring at workplaces. Therefore any overall estimate of the number of occupational fatalities is likely to be uncertain and surely an underestimate no matter what data source is used. The evaluation of the completeness of reporting of fatal injuries occurring at workplaces using data from the BC Coroners Service, the Workers Compensation Board and the BC Vital Statistics Division has not been documented before. 7.1.1 Implications for public health policy The lack of an accurate number of fatalities should not divert our attention from the seriousness of the social and economic losses due to work-related deaths. A major issue is how people are being killed and what can be done to prevent future deaths. Causal factor information on occupational fatalities can be obtained from Coroners and W C B investigation reports. Standardization of the present data collection system used by both agencies will yield data of greater value for injury control countermeasure. We need to develop a method to 140 abstract the important information from narrative data in a concise way. The current system of presenting injury statistics focuses on the outcome (i.e. injury) rather than the preceding events or circumstances that led to it. Thus it is not prevention oriented. For example, the death of a worker (i.e. case # 7, Table 4.3 Chapter 4) can be recorded in different ways 1) narrative form: John Doe, 30 years old, was trying to free a jammed log. He stepped on an insecure plank (57 inches overhang) which up-ended dropping him through an opening to the basement floor, 20 feet below. He hit his head on a steel cross beam 2) Current system of WCB report/ Work Injury Program: Type of injury: fall from elevation Source of injury: steel cross beam Nature of injury: fracture Part of body: head Unlike the narrative form, the current system of reporting work-related injuries does not allow the identification of the obvious immediate cause of the incident: "the worker stepped on an insecure plank". Likewise other violations of safety regulations could be flagged and be the basis for developing special programs within the mill. 7.2 Work-related fatal injuries Given that injuries are preventable, the injury fatality rate of 18.10 per 100,000 person-years indicates that injury prevention strategies tailored to this workforce are needed. The downward trend of injury fatalities at the workplace observed from the 1960's onward may be ascribed partly to the implementation of more stringent safety regulations in the 141 workplace as well as the automation of production processes experienced by all industrial sectors. The impact of a single protective measure in reducing the number of deaths, such as the one observed by the mandatory use of buoyancy equipment proves that passive strategies (i.e. not requiring the active participation of the worker) are effective for injury control. Environmental factors were identified as contributory causes in the majority of the fatal injuries occurring at the workplace. Similarly, the most frequent Coroner's jury recommendations were related to the need for improving the workplace environment. The majority of factors thus corresponded to the pre-event phase of the incident according to the Haddon matrix: e.g. unsafe work stations, inappropriate maintenance of equipment/machinery, job process (physical environment) and inappropriate safety policies, communications, training, supervision (socio-cultural environment). Equipment failure and inadvertent starting of equipment were identified factors in the pre-event phase in relation to the equipment (carriers of energy) producing the injury. Most of these conditions are amenable to correction by engineering control and enforcement of safety regulations. Another interesting finding is the absence of positive alcohol results in the deaths analyzed. This indicates that, in this workforce, alcohol was not a factor for being fatally injured at the workplace. The data analyzed also pointed to a few characteristics common to fatal injuries that deserve further research. Older workers were at higher risk of injuries than their younger fellow workers and 50 % of the workers were alone at the moment of the incident. It has been 142 suggested that one of the characteristics that distinguish fatal from non-fatal injuries is working alone, without supervision. Fatigue may also have an effect on the risk of injury. In our study, 38% of the deaths with reports occurred between 2:00 am and 7:00 am and on Mondays. Future research should include the study of psychosocial factors that reflect the workplace organization and affect the well-being of the workforce. Low lost-time frequency rates have been associated with demonstration by management of its concern for the employees, greater involvement of workers in general decision making, greater willingness of the health and safety committee to solve problems internally as well as greater experience of the workforce (Shannon, et al., 1994). 7.2.1 Implications for public health policy Attention should be focused on work-related fatalities occurring among workers in occupations at highest risk of injury (i.e. machine operators/clearers; mobile equipment operators, and skilled trades). Another way of establishing priorities for prevention countermeasures is to focus on groups of fatalities that share common circumstances or patterns. For example, we found that most fatal injuries occurred while workers were performing routine tasks (i.e. clearing jammed logs, repairing equipment) which suggests that redesign of job procedures or equipment may be necessary. Similar patterns regarding injury event types such as "inadvertent start of equipment" (i.e. failure to lock out equipment) point to situations that may require consideration of the multiple factors that may be involved when there is no compliance with safety regulations. One could label this as an unsafe act on the 143 part of the worker even when the unsafe act may be a standard operating procedure of the mill. The findings from this study, in agreement with other reports of non-fatal injuries, suggest that sawmill workers are routinely exposed to "normal" hazardous situations when they get injured (Larsson, 1991). Therefore, countermeasures should focus on safer ways of doing routine tasks such as redesigning the job or developing special tools. From the point of view of planning injury control strategies, our findings suggest that prevention measures directed at modifying the workplace physical environment should be prioritized to reduce the unnecessary loss of lives from preventable causes. However, to be maximally effective in preventing injury occurrence and severity it is necessary to develop countermeasures that take into consideration the full range of factors (human, equipment and environment) interacting in each of the three phases: pre-event, event and post-event. Passive countermeasures are preferable to active measures that require voluntary action from the worker. Finally, cost and effectiveness are important considerations since frequently prevention measures are chosen without previous evaluation of their efficacy (Haddon and Baker, 1981). It has been indicated that fatal injuries are different from non-fatal injuries in terms of the type of injury and the circumstances surrounding the incident (Salminen, et al., 1992). Fatal work-related injuries are primarily the result of being struck by a falling object, or being run over by a moving vehicle. Also, injuries to the head or trunk are more likely to be fatal than injuries to other parts of the body. Such fatal injuries are more likely to occur when the worker is alone, without supervision, and in situations where injuries have previously 144 occurred. Our results suggest that sawmill workers were fatally injured in circumstances similar to the ones reported for non-fatal injuries in previous studies (Larsson, 1991; Laflamme and Cloutier, 1989). This is encouraging because it implies that by focusing on the circumstances identified, not only are we addressing the most severe injuries (and less frequent), but also the most frequent, less severe injuries. Thus the greatest benefit can be expected. 7.3 All fatal injuries One interesting finding of the study is the significantly higher risk of fatal motor vehicle crashes and suicide among workers living in isolated towns. The socio-economic characteristics of these mill towns are 1) the economic dependence of the community on a company and 2) the social isolation of the residents and 3) the lack of opportunities for children and women. Furthermore, residents of rural areas drive more miles, usually on poorer roads and at greater speed than residents of urban settings. The contribution of this study is that it completed the characterization of the broad pattern of mortality from injuries among sawmill workers from 1950 through 1990. It was possible to quantify the magnitude of both the absolute and relative risk of fatal injuries in different settings. Interestingly machinery-related injuries were the second leading cause of unintentional injury deaths in this cohort, and the only cause with significantly more deaths than expected when compared to the general BC population. While the increased risk of mortality from machinery-related injuries was expected given the physical environment of a sawmill, what was not expected is the fact that there were many fewer deaths from non-145 work-related injuries among these workers than in the general population. The results of previous mortality studies in forestry workers indicated that the opposite result was more likely. In addition it has been suggested that blue collar workers operating machinery tend to experience more depression and worse global health than those employed in managerial or professional jobs (Grosch and Murphy, 1998). Furthermore sawmill jobs may comprise working rotating shift schedules, and overtime. These factors may result in fatigue and affect the risk of injuries outside the workplace. Our results indicate that sawmill workers are at greater risk of fatal injuries at work than at other settings. The causes of this different risk can not be explained with the data available in this study. However, our findings suggest that the workplace environment plays a critical role in the occurrence of fatal injuries at work. If these fatalities were solely related to personal factors such as risk-taking behaviour or carelessness or fatigue, then the same factors would likely effect the risk of fatal injuries away from work. 7.3.1 Implications for public health policy The highest fatality rates of motor vehicle crashes and suicides observed in the isolated mill towns of B.C. indicate where countermeasures should be first implemented. Consideration should be given to the particular environmental and socioeconomic characteristics of the region that may be contributing to this increased risk of fatal injuries. The significantly lower risk of fatal motor vehicle crashes and suicides in the occupational group of "skilled labour" prompts a question regarding the relative impact of income and education on health outcomes. Why were workers who are between the occupational 146 categories with the highest and lowest level of income and education (tradesmen and unskilled labour respectively) at the lowest risk? Studies addressing other important determinants of health such as housing, social support, self-esteem, and lifestyle are warranted to understand better what makes some people healthier than others (Evans, et al., 1994). More funding should be made available for research into the risk factors for severe non-fatal as well as fatal injuries in order to plan preventive strategies. The implementation of cost-effective countermeasures will result in a significant reduction of the social and economic costs of preventable injury which is the second most expensive disease category for Canadians (Wigle, et a l , 1991). 7.4 Limitations There are some limitations of the study that mostly derive from the research design. One limitation is related to the small number of deaths included in the analyses. This was the result of the criteria used to define the follow-up period in this investigation. Since the aim of our study was to characterize the mortality from injuries among sawmill workers, only fatal injuries that occurred to our cohort members while actively employed at any of the study sawmills were analyzed. This criterion was applied to the study of both on-the-job as well as off-the-job fatal injuries (i.e. chapters 5 and 6). Consequently, the small number of cases precluded more detailed analysis of the cohort mortality and limited the statistical precision of the estimates. For example, it was not possible to model the effect of age, seniority and calendar period on mortality by specific injury type. Another limitation derives from the quality of information from the Chief Coroner's Office records and the lack of consistency in recording data on factors relevant to the fatal incidents. The breadth of information of the investigation reports varies from coroner to coroner due to the absence of a standardized format for data collection. The lack of standardized procedures for the Coroner to investigate injury deaths has also been recognized in other countries like U.S and England (Hanzlick and Parrish, 1996; Gaffney, 1993; Runyan et al., 1994). Several reasons, including differences in statutes, funding, staffing, training, automation and methods of managing records have also been cited as important determinants of the quality of Coroners data (Hanzlick and Parrish, 1996). In spite of the limitations, Coroners records are the best available source for epidemiologic research of fatal injuries. Furthermore, in our study, for most of the factors included in the descriptive analysis of the fatal injuries, the lack of proper denominator precluded any definitive conclusion about the role of these factors on the injury event. For example, the finding that the highest proportion of work-related deaths occurred between 6:00 and 7:00 am and also at 10:00 pm would suggest the effect of fatigue but this can not be definitively concluded with the data available. The generalizability of our results may be limited by the over-representation of large mills in our study. Punnett (1992) has reported that largest mills have lower injury rates than smaller companies. Large companies are likely to have more resources available to implement occupational health and safety programs. If that is the case, then the mortality rates estimated in this study are likely to be conservative compared to the experience of the entire 148 industry. Therefore, while the results may apply to large mills, generalizability beyond them is questionable. With respect to the analyses of the cohort mortality for all types of injuries, reported in chapter 6, two different approaches were used. First, SMRs were computed using the B.C. general male population as the standard for calculating expected deaths. The use of an external population as a reference has been criticized because it provides a summary measure of effect but it doesn't give information on exposure effects that may be different among subgroups of the study population (Chiazze, 1976). In addition, because the external general population is composed of people too i l l to work as well as employed individuals, the estimated SMRs will underestimate any true decreased or increased risk (Hennekens and Buring, 1987). Those in favour of using SMRs as the primary analytic method for occupational cohort data argue that one of the advantages of using national or provincial rates to compare mortality experience of exposed cohorts would be that rates for the general population are readily available. Secondly, national or provincial rates generally provide stable stratum specific rates which improve the precision of the estimates for the study population (Checkoway et al. 1986). Breslow (1990) discusses the problem of SMRs in terms of its susceptibility to selection and confounding biases and proposes alternative analytic methods to control for confounding, mainly through the use of Poisson regression analysis. The reason for using such 149 mathematical modeling is that there may be several factors that could influence the outcome and should also be controlled to the greatest possible extent. For example, differences in injury rates between the working population and those in the provincial comparison group may reflect the different composition of the groups in terms of ethnicity, education, socioeconomic status and lifestyle habits. Breslow concludes that despite its limitations, a list of SMRs by cause of death provides a useful summary of the relative effects of cohort membership on the risk of cause specific mortality. Given the limitations of the SMRs, internal analyses were conducted comparing injury mortality rates among subcohorts defined according to 1) occupational categories; and 2) age groups, calendar period at risk, seniority in mill and region. Poisson regression techniques were used to eliminate confounding within the limits of the data available. This approach removed the differences in employment and socio-economic status noted with the use of an external comparison group. Mortality rates were compared among subgroups controlling for the confounding effects of different risk factors which were included in the regression equation. For example, in chapter 6, the use of Poisson regression allowed us to identify the group of youngest workers (i.e. 15 to 29 years) as the age category at the highest risk of suicide. This increased risk was statistically significant after adjustment for the effect of calendar period, seniority in the mill and region. 150 7.4.1 Implications for public health policy The discussion of the limitations of this study prompts the following query: How can we improve further research on the risk factors for injuries and identify the most effective interventions to reduce injury morbidity and mortality? According to experts in this field, the main components of any comprehensive plan to reduce the toll of injuries should be surveillance and research (CDC, 1992; Saunders and Stewart, 1991; Pless, 1989; Waller, 1993). An injury surveillance system is important to describe and monitor the injury problem. Surveillance data not only assist to determine the need for public health action but to assess the effectiveness of preventive strategies. Criteria for developing, implementing and evaluating an injury surveillance system have been proposed by several authors( Rubens et al., 1995, Waller and Clemmer, 1993; Graitcer, 1987; Klaucke et al., 1988). In agreement with them, we propose that in order to improve surveillance of work-related fatalities, the following actions be considered: 1) To use standardized industrial and occupational codes to record injury events. This will facilitate the comparison of statistics at the provincial, national and international level. It would be helpful i f both the B.C.Coroners Service and the Workers Compensation Board of B.C. adopted these coding systems. 2) To use the E-codes from the most recent International Classification of Diseases to classify fatal injuries. This will facilitate the identification of the various types of injury when abstracting information from different sources. 3) To develop guideliness to standardize the data collection and recording processes used by the Coroners Service to ensure comprehensiveness and consistency of the investigation reports. In this regard, the format of the WCB Investigation Report provides an example of the type of data relevant to the analysis of work-related fatal injuries. 4) To develop a system that takes into account the information contained in the narrative of the Coroners Investigation reports. This will add important information on the pre-event and event phases of the injury event. For example: what was the worker doing? what went wrong? what was the contact? what equipment was involved? how did the incident occurred? where did the event occur? As a general rule it is necessary to develop a data collection system that describes in a more comprehensive way the components of the preinjury and injury phases of injury event. Research is the other priority area requiring attention because funds for injury investigation and training programs for professionals are scarce. Research could be used to improve injury surveillance systems and to develop and test hypothesis aimed at identifying specific risk factors in special populations. For example, the present investigation of injuries among sawmill workers not only has identified job categories at higher risk of suffering fatal injuries on-the-job and off-the-job, but has also generated hypotheses about other potential risk factors that hopefully could be addressed in future studies. Being a workforce at high risk of injury disability future studies could aim at identifying organizational risk factors for non-fatal injuries. We propose a 152 population-based case-control study where injured workers will be identified from W C B claims and their injury experience compared to fellow workers (i.e. controls) from the same company who had not suffered a work-related injury on that day, matched to the case on age. A questionnaire would then be administered to both cases and controls to identify organizational (e.g. work satisfaction, support from supervisor and management, working hours), work environment (e.g., machinery, heights of work platforms), and personal (e.g., family support, age, experience) characteristics that may have been relevant to the injury event. The case-control design would ensure that both numerator and denominator information was available. Examining morbidity instead of mortality outcomes would allow a large number of injured sawmill workers to be included so that differences in risk can be estimated with precision. In addition, mills of various sizes could be included in the study so that the sample will be representative of the entire industry and minimize the problem of generalizability of the results. 153 Bibliography Ager, B. " A Review of the Ergonomic Problems in Sawmills and Wood-Working Industries." In Proceedings of IUFRO Joint Meeting Divisions 3 and 5. Ergonomics in Sawmills and Woodworking Industries. Sweden, 1974. Aminoff, S. "Accidents and Accident Risks Within the Sawmill Industry." In Proceedings of IUFRO Joint Meeting Divisions 3 and 5. Ergonomics in Sawmills and Woodworking Industries. Sweden, 1974. Armstrong, B. "Choosing between SMR and Poisson Regression Analyses of Cohort Data." Progress in occupational epidemiology ed. Hogsted C. and C. Reuterwall 45-48. Elsevier Science Publisher B.V.,1988. Asp, S., V . Riihimaki, S. Hernberg, and E. Pukkala. "Mortality and Cancer Morbidity of Finnish Chlorophenoxy Herbicide Applicators: an 18-Year Prospective Follow-up." American Journal of Industrial Medicine 26 (1994):243-253. Baker, S., J.S. Samkoff, R. Fisher, and C.B. Van Buren. "Fatal occupational injuries." Journal of the American Medical Association 248 (1982):692-697. Baker, S.P., B. O'Neill , M.J. Ginsburg, and G. L i . The Injury Fact Book. 2 n d . Edition. New York: Oxford University Press, 1992. Baneryd, K. "Analysis of the Psychological Job Content in Sawmill Work." In Proceedings of IUFRO Joint Meeting Divisions 3 and 5. Ergonomics in Sawmills and Woodworking Industries. Sweden, 1974. Berry, G. "The analysis of mortality by the subject-years method." Biometrics 39 (1983):173-184. Breslow, N . "Statistical Issues in the Analysis of Data from Occupational Cohort Studies." Recent Results in Cancer Research 120 (1990):78-93. Breslow, N.E. and N.E. Day. Statistical Methods in Cancer Research. Vol II.: The Design and Analysis of Cohort Studies. IARC Scientific Publications No.82. Lyon: International Agency for Research on Cancer, 1987. British Columbia Ministry of Transportation and Highways. 1993 Traffic Accident Statistics. B.C. : Ministry of Transportation and Highways, 1993. British Columbia Ministry of Health and Ministry Responsible for Seniors. BC-Injury Free-A Framework for Action. The Provincial Injury Prevention Plan for Children and Youth (Ages Birth to 24). Report of the Minister's Injury Prevention Advisory Committee. Victoria. B.C.: Ministry of Health and Ministry Responsible for Seniors, 1997. 154 British Columbia Provincial Health Officer. A report on the health of British Columbians: Provincial Health Officer's Annual Report 1995. Victoria: B.C. Ministry of Health and Ministry Responsible for Seniors, 1996. British Columbia Provincial Health Officer. A report on the health of British Columbians: Provincial Health Officer's Annual Report 1996. Victoria: B.C. Ministry of Health and Ministry Responsible for Seniors, 1997. British Columbia Vital Statistics Agency. Vital Statistics Quarterly Digest. Ministry of Health. Victoria, B.C., 6 (1997). Canada Health and Welfare. Suicide in Canada. Report of the National Task Force on Suicide in Canada. Ottawa: Minister of National Health and Welfare, 1987. Cassinelli, R., K.J . Kock, K. Steenland, S. Spaeth, and P. Laber. Life Table Analysis System for Use on the PC. User documentation PC LTAS. Version 1.0.c. NIOSH. Cincinnati: Division of Surveillance, Hazards Evaluation, and Field Studies, 2000. CDC. Centers for Disease Control. Position papers from the third national injury control conference: setting the national agenda for injury control in the 1990s. Atlanta, Ga 1992. M M W R Vol . 41(1992):No. RR-6. Checkoway, H. , N . Pearce, and D.J. Crawford-Brown. Research Methods in Occupational Epidemiology. New York: Oxford University Press, 1989. Chiazze, L. "Problems of study design and interpretation of industrial mortality experience." Journal of Occupational Medicine 18 (1976): 169-170. Chiu, W., S.R. Dearwater, D.J. McCarty, T.J. Songer, and R.E. LaPorte. "Establishment of accurate incidence rates for head and spinal cord injuries in developing and developed countries: A Capture-Recapture approach." The Journal of Trauma 35 (1993):206-21. Colorado Department of Health. Colorado Population-based Occupational Injury and Fatality Surveillance System. Report 1982-1984. Health Statistics Section. Denver: Colorado Department of Health, 1984. Conroy, C. "The Methodology of Fatal Occupational Injury Surveillance." American Journal of Public Health 83 (1993): 1491. Cooke, W.N., and M.W.Blumenstock. "The Determinants of Occupational Injury Severity: The Case of Maine Sawmills" Journal of Safety Research 11 (1979):115-120. Copeland, A.R. "Fatal Occupational Accidents- the Five Year Metro Dade County Experience, 1979-1983." Journal of Forensic Science 30 (1985):494-503. 155 Cormack, R. M . , Y . Chang, G.S. Smith. "Estimating deaths from industrial injury by captrure-recapture: a cautionary tale." InternationalJournal of Epidemiology 29 (2000):1053-1059. Davies, H. , P. Demers, T. Kauppinen, and K. Teschke. "Industries Based on Biological Resources. Major Sectors and Processes: occupational hazards and controls." In Encyclopedia of Occupational Health and Safety, 4th Edition. International Labour Office. Editor: Mager Stellman J. Geneva, 1998. De Haven, H. "Mechanical analysis of survival in falls from heights of fifty to one hundred and fifty feet." War Medicine 2 (1942):586-596. DeMaria, E.J., P.R. Kenney, M . A . Merriam, L .A. Casanova, and D.S. Gann. "Survival after Trauma in Geriatric Patients." Annals of Surgery 206 (1987):738-743. Demers, P. "Industries Based on Biological Resources. Disease and Injury Patterns." In Encyclopedia of Occupational Health and Safety, 4th Edition. International Labour Office. Editor: Mager Stellman J. Geneva, 1998a. Demers, P. "Industries Based on Biological Resources. General Profile." In Encyclopedia of Occupational Health and Safety, 4th Edition. International Labour Office. Editor: Mager Stellman J. Geneva, 1998b. Doering, M . , S.R. Rhodes, and M . Schuster. The aging Worker. Beverly Hi l l , C A : S A G E Publications, Inc., 1983:61-82. Driscoll, T.R., G. Ansari, J.E. Harrison, M.S. Frommer, and E.A. Ruck. "Traumatic Work-Related Fatalities in Forestry and Sawmill Workers in Australia." Journal of Safety Research 26 (1995):221-233. Evans, R.G., M . L . Barer, and T.R. Marmor.Ed. Why are some people healthy and others not? The determinants of health populations. New York: Aldine de Gruyter, 1994. Fox, A.J . and F.F. Collier. "Low mortality rates in industrial cohort studies due to selection for work and survival in the industry." British Journal of Preventive and Social Medicine 30 (1976):225-230. Frome, E.L. and H. Checkoway. "Epidemiologic Programs for Computers and Calculators Use of Poisson Regression Models in Estimating Incidence rates and Ratios." American Journal of Epidemiology 121 (1985):309-323. Gaffey, W.R. " A Critique of the Standardized Mortality Ratio." Journal of Occupational Medicine 18 (1976): 157-160. Gaffney B.P. "Use of Coroner's reports for surveillance of accidental death." Journal of Public Health Medicine 15 (1993):272-276. 156 Gallagher, R.P, WJ. Threlfall, P.R. Band, J.J. Spinelly, and A.J . Coldman. Occupational Mortality in British Columbia 1950-1978. Catalogue No. 84-544.0ttawa: Minister of Supply and Services Canada, 1986. Gibson, J.J. "The contribution of experimental psychology to the formulation of the problem of safety- a brief for basic research." In Behavioural approaches to accident research. Association for the Aid of Crippled Children. New York, 1961. Gordon, J.E. "The epidemiology of accidents." American Journal of Public Health 39 (1949):504-515. Government of British Columbia. The Sawmilling Industry of British Columbia. Department of Industrial Development, Trade and Commerce. Victoria, B.C.: 1972. Graitcer, P. "The Development of State and Local Injury Surveillance Systems." Journal of Safety Research 18 (1987): 191-197. Green, L . M . " A cohort mortality study of forestry workers exposed to phenoxy acid herbicides." British Journal of Industrial Medicine 48 (1991):234-238. Grosch, J.W. and L.R. Murphy. "Occupational Differences in Depression and Global Health: Results from a National Sample of U.S. Workers." Journal of Occupational and Environmental Medicine 40 (1998): 153-164. Haddon, W. "Advances in the Epidemiology of Injuries as a Basis for Public Policy." Landmarks in American Epidemiology. Public Health Reports 95 (1980):411-421. Haddon, W.Jr. " A Note Concerning Accident Theory and Research with Special Reference to Motor Vehicle Accidents." Annals of New York Academy of Sciences 107 (1963):635-646. Haddon, W.Jr. "On the Escape of Tigers: An Ecologic Note." American Journal of Public Health 60 (1970):2229-2234. Haddon, W.and S.P. Baker. Injury control. In: Clark C. MacMachon B, eds. Preventive and Community Medicine. Boston: Little and Brown Co, 1981.109-40. Haddon, W.Jr., E. Schuman, and D. Klein. Accident Research, Methods and Approaches. New York: Harper & Row, 1964. Hanzlick R. and R.G. Parrish. "The Role of Medical Examiners and Coroners in Public Health Surveillance and Epidemiologic Research." Annual Review of Public Health 17 (1996):383-409. 157 Hasselback, P. and C.I. Neutel. "Risk for commercial fishing deaths in Canadian Atlantic provinces." British Journal of Industrial Medicine Al (1990):498-501. Health Canada. Suicide in Canada: Update of the report of the Task Force on Suicide in Canada. Cat No. H39-107/1995E. Ottawa: Mental Health Division, Health Services Directorate, Health Programs and Services Branch, 1995. Hennekens, C. H. and J.E. Buring. Epidemiology in Medicine. Boston: Little Brown and Company, 1987. Hertzman, C , K . Teschke, A. Ostry, R. Hershler, H. Dimich-Ward, D. Kelly, J.J. Spindly, R.P. Gallagher, M . McBride, and S. Marion. "Mortality and Cancer Incidence among Sawmill Workers exposed to Chlorophenate Wood Preservatives." American Journal of Public Health 87 (1997): 71-79. Hi l l , I.D. "Computing man-years at risk." British Journal of Preventive and Social Medicine 26 (1972):132-134. Himes, C.L., C C . Clogg. "An overview of demographic analysis as a method for evaluating census coverage in the U.S." Population Index 58 (1992):587-607. Hook, E.B. and R.R. Regal. "The value of capture-recapture methods even for apparent exhaustive surveys." American Journal of Epidemiology 135 (1992):1060-1067. Hook, E.B. and R.R. Regal. "Capture-recapture methods in epidemiology: methods and limitations." Epidemiologic Reviews 17 (1995):243-264. Howard, G. Fatal Claims in Sawmills (classification 010501 to 010509) Analysis of Causes and Costs from 1985 to 1994. Final Report. Engineering Department. Vancouver, B.C.: Workers' Compensation Board of B.C., 1996. Howe, G.R. and J.A. Lindsay. " A generalized record linkage computer system for use in medical follow-up studies." Computers and Biomedical Research 14 (1981):327-340. Hyman, M . M . "Accident Vulnerability and Blood Alcohol Concentrations of Drivers by Demographic Characteristics." Quarterly Journal of Studies on Alcohol 29 (1968):34-57. Jones, R.K. and K . B . Joscelyn. Alcohol and Highway safety 1978: A review of the State of Knowledge. AnnArbor, Mich.: Highway Safety Research Institute, University of Michigan, Report DOT HS-803 764, 1978. Karasek, R., and T. Theorell. Healthy Work. Stress, productivity, and the reconstruction of working life. New York: Basic Books, Inc., 1990. 158 Keifer, M . , R. McConnell, A.F. Daniel, L. Rosenstock. "Estimating underreported pesticide poisoning in Nicaragua." American Journal of Industrial Medicine 30 (1996): 195-201. Kelsey, J., W.D. Thompson, and A.S. Evans. Methods in Occupational Epidemiology. New York: Oxford University Press, 1986. Kisner, S.M. and S.G. Pratt. "Occupational Fatalities Among Older Workers in the United States: 1980-1991." Journal of Occupational and Environmental Medicine 39 (1997):715-721. Klaucke, D.N. , J.W. Buehler, S.B. Thacker, R. Gibson Parrish, F.L. Trowbridge, R.L. Berkelman, and the Surveillance Coordination Group. "Guideliness for Evaluating Surveillance Systems." MMWR. 37 (1988):1-18. Kraus, J.F. and L.S. Robertson. "Injuries and the Public Health" In: Maxcy-Rosenau-Last. Public Health & Preventive Medicine. 13 th Edition. 1021-1034. Connecticut: Appleton and Lange, 1992. Labour Canada. Bulletin No. 14. Ottawa: Labour Canada, 1991. Labour Canada. Employment Injuries and Occupational Illnesses 1985-1987. Catalogue No. L151-2068/90B. Ottawa: Labour Canada, 1990. Labour Canada. Occupational Injuries and their cost in Canada 1987-1989.Ottawa: Labour Canada, 1989. Laflamme, L. and E. Cloutier. "Typical Accident Scenarios in Sawmills: Their Distribution among Production Stages and Types of Machines." International Union of Forestry Research Organizations. Conference on Working Conditions in Woodworking Industry, Conference of Division 3 P3.03-04 Ergonomics in the Timber Industry. Lappeenranta, Finland, 1989. Larsson, T.J. Safety in Sawmills- three Swedish studies-1988-1990. Stockholm:Institute for Human Safety and Accident Research (IPSO), 1991. Last, J .M. "Health -whose responsibility is it anyway.? Canadian Journal of Public Health 75 (1984):410-41. Lewis, R.J., and S.P. Cooper. "Alcohol, Other Drugs, and Fatal Work-related Injuries" Journal of Occupational Medicine 31 (1989):23-28. Loimer, H. and M . Guarnieri. "Accidents and Acts of God: A History of the Terms." American Journal of Public Health 86 (1996): 101-107. 159 Loomis, D., V . Dufort, R.C. Kleckner, and D.A. Savitz. "Fatal Occupational Injuries Among Electric Power Company Workers." American Journal of Industrial Medicine 35 (1999):302-309. Mac William, L. and L. Fortier. "Early Mortality due to Unintentional Injury in Canada, 1985." Chronic Diseases in Canada 8 (1987): 11 Marchack P. "Forestry Industry Towns in British Columbia." In Community and Forestry: Continuities in the sociology of natural resources. Ed. Lee, R.G. et al. Boulder: Westview Press, 1990. Mason, K. A Correlation Between Types of Hazardous Conditions and Accident Rates. Vancouver: Workers" Compensation Board of British Columbia, 1973. Mietinnen, O.S. "Standardization of risk ratios." American Journal of Epidemiology 96 (1972):383-388. Millar, W. "Accidents in Canada, 1988 and 1993." Health Reports 7 (1995):7-16. Millar, W. and O. Adams. Accidents in Canada. General Social Survey. Analysis Series. Housing, Family and Social Statistics Division. Catalogue No. l 1-612E No.3. Ottawa: Statistics Canada, 1991. Millar, W.J., and J .M. Last. "Motor Vehicle Traffic Accident Mortality in Canada, 1921-1984." American Journal of Preventive Medicine 4 (1988):220-230. Minister of Supply and Services and Minister of National Health and Welfare. The Health of Canadians. Report of the Canada Health Survey. Catalogue No. 82-538E. Ottawa: Minister of Supply and Services and Minister of National Health and Welfare, 1981. NAICS -North American Industry Classification System, United States, 1997-. Virginia, U.S.: Department of Commerce. Technology Administration, 1997. National Committee for Injury Prevention and Control "Injury Prevention: Meeting the Challenge." American Journal ofPreventive Medicine 5 (1989): 1-303 Supplement. Nilsson, C. "Working Conditions in the Sawmill Industry: A sociopsychological approach based on subjective data." In Proceedings of IUFRO Joint Meeting Divisions 3 and 5. Ergonomics in Sawmills and Woodworking Industries. Sweden, 1974. Notkola, V.J . , P. Martikainen, and P.I. Leino. "Time trends in mortality in forestry and construction workers in Finland 1970-195 and impact of adjustment for socioeconomic variables." Journal of Epidemiology and Community Health 47 (1993): 186-191. 160 O'Gara, K. California Sawmills and Planing Mills Industry: Analysis of Work Injuries and Illnesses. Division of Labor Statistics and Research. Department of Industrial Relations. San Francisco, CA, 1978. Ore, T. and N . Stout. "Traumatic Occupational Fatalities in the U.S. and Australian Construction Industries." American Journal of Industrial Medicine 30 (1996):202-206. OSHA. Occupational Safety and Health Administration. Coping with hazards in Sawmills. Ocupational Safety Training. University of Wisconsin- School for Workers. Wisconsin, June 1976. Ostry, A. , S.A. Marion, L. Green, P.A. Demers, R. Hershler, S. Kelly, K. Teschke, and C. Hertzman. "Downsizing and Industrial Restructuring in Relation to Changes in Psychosocial Conditions of Work in British Columbia Sawmills." Scandinavian Journal of Work Health and Environment 26 (2000):273-277. Parkinson, D.K, W.F. Gauss, J.A. Perper, and S.A. Elliott. "Traumatic Workplace Deaths in Allegheny County, Pennsylvania, 1983 and 1984." Journal of Occupational Medicine 28 (1986):100-102. Pearce, N . , H . Checkoway, and J. Dement J. "Exponential Models for analyses of Time-related Factors, Illustrated with Asbestos Textile Worker Mortality Data." Journal of Occupational Medicine 30 (1988):517-522. Peek-Asa, C , R. Erickson, and J.F. Kraus. "Traumatic Occupational Fatalities in the Retail Industry, United States 1992-1996." American Journal of Industrial Medicine 35 (1999):186-191. Pelz, D.C. and S.H. Schuman. Motivational Factors in Crashes and Violations of Young Drivers. Ann Arbor, Mich.: University of Michigan, paper presented at the Annual Meeting of American Public Health Association, 13 October 1971. Pless, B.I. "Challenges for Injury Prevention: Two Neglected Aspects." Chronic Diseases in Canada 10 (1989):2. Pratt, S.G., S.M. Kisner, and J.C. Helmkamp. "Machinery-Related Occupational Fatalities in the United States, 1980 to 1989." Journal of Occupational and Environmental Medicine 38 (1996):70-76. Punnett, L. Case-Control Study of Sawmill Injuries in Maine. Final Report to the National Institute of Occupational Safety and Health. University of Massachusetts, Lowell. Lowell, M A . , 1994. Punnett, L. Epidemiologic study of sawmill injuries in Maine. Final Report to The Centers for Disease Control. University of Massachusetts, Lowell. Lowell, M A . , 1992. 161 Rest, K . M and N . A . Ashford. Occupational Safety and Health in British Columbia: An Administrative Inventory. Richmond: Workers' Compensation Board of British Columbia, 1992. Rice, D.P., T.A. Hodgson, and A . N . Kopstein. "The economic cost of illness: a replication and update." Health Care Financing Review 7 (1985):61-80. Riley, R. and P. Paddon. "Accidents in Canada: mortality and hospitalization." Health Reports 1 (1990): 23-50. Rivara, F.P. "Epidemiology of Childhood Injuries. Review of current Research and Presentation of Conceptual Framework." American Journal of Diseases of Children 136 (1982):399-405. Robertson, L.S. Injuries: Causes, Control Strategies, and Public Policy. Lexington, Mass., DC: Heath, 1983. Robertson, L.S. Injury Epidemiology. New York: Oxford University Press, 1992. Rossignol, M . "Completeness of Provincial Workers' compensation Files to Identify Fatal Occupational Injuries." Canadian Journal of Public Health 85 (1994):244-247. Rossignol, M . , and M . Pineault. "Fatal Occupational Injury Rates: Quebec, 1981 through 1988." American Journal of Public Health 83 (1993):1563-1566. Rubens, A.J . , W.A. Oleckno, and L. Papaeliou. "Establishing Guidelines for the Identification of Occupational Injuries: A Systematic Appraisal." Journal of Occupational and Environmental Medicine 37 (1995): 151-159. Runyan, C.W., D.Loomis, and J. Butts. "Practices of County Medical Examiners in Classifying Deaths as On the Job." Journal of Occupational Medicine 36 (1994):36-41. Salisbury, D.A., R. Brubaker, C. Hertzman, and G.R. Loeb. "Fatalities among British Columbia Fallers and Buckers 1981-7." Canadian Journal of Public Health 82 (1991):32-37. Salminen, S., J. Saari, K . L . Saarela, and T. Rasanen. "Fatal and non-fatal occupational accidents: identical versus differential causation." Safety Science 15 (1992): 109-118. Saunders, L.D. and L . M . Stewart. " A Safer Canada Year 2000 Injury Control Objectives for Canada" In Proceedings of a symposium held in Edmonton, Alberta, Canada. May 21-22, 1991. Schnatter, R.A., J.F. Acquavella, F.S. Thompson, D. Donaleski, and G. Theriault. "An Analysis of Death Ascertainment and Follow-up through Statistics Canada's Mortality Data Base System." Canadian Journal of Public Health 81 (1990):60-65. 162 Selvin, S. Practical Bio statistical Methods. California: Wadsworth Publishing Company, 1995. Shannon, H.S., L. Hope, L. Griffith, and D. Stieb. "Fatal Occupational Accidents in Ontario, 1986-1989." American Journal of Industrial Medicine 23 (1993):253-264. Shannon, H.S., V . Walters, W. Lewchuk, J. Richardson, L .A. Moran, T. Haines, and D. Verma. "Workplace Organizational Correlates of Lost-time Accident Rates in Manufacturing." American Journal of Industrial Medicine 29 (1996):258-268. Sniezek, J.E., T .M. Horiagon. "Medical-Examiner-Reported Fatal Occupational Injuries, North Carolina, 1978-1984." American Journal ojIndustrial Medicine 15 (1989):669-678. Spiegel, J. and A . Yassi. "Occupational Disease Surveillance in Canada: A Framework for Considering Options and Opportunities." Canadian Journal of Public Health (1991):294-299. SERC Statistical Epidemiological Research Corporation. EGRET Reference Manual. Seattle (WA): Statistical Epidemiological Research Corporation, 1993. Statistics Canada. Canadian Forestry Statistics 1992. Industry Division. Annual Survey of Manufactures Section. Catalogue 25-202. Ottawa: Statistics Canada, 1994. Statistics Canada. Industry and the Class of Worker. The Nation. Ottawa: Statistics Canada, 1993. Statistics Canada. National Population Health Survey. Overview. 1994-1995. Catalogue No. 82-567. Ottawa: Statistics Canada, 1995b. Statistics Canada. Wood Industries. SIC 25,1993. Industry Division Annual Survey of Manufactures Section. Catalogue 35-250. Ottawa: Statistics Canada, 1995a. Statistics Canada's Internet Site. Population Projections, http://www.statcan.ca. 23 Jan 2001. Stout, N . , and C. Bell. "Effectiveness of source Documents for Identifying Fatal Occupational Injuries: A Synthesis of Studies. "American Journal of Public Health 81 (1991):725-728. Stout, N .A. , E.L. Jenkins, and T.J. Pizatella. "Occupational Injury Mortality Rates in the United States: Changes from 1980 to \9%9." American Journal of Public Health 86 (1996):73-77. Teschke, K. , A . Ostry, C. Hertzman, P. Demers, M.C. Barroetavena, H. Davies, H . Dimich-Ward, H. Heacock, and S. Marion. "Opportunities for a broader Understanding of Work and Health: Multiple uses of an Occupational Cohort Database." Canadian Journal of Public Health S9 (\998):\32-\36. 163 Tuk, T.A. and J. Macdonald. Suicide, Homicide, and Gun Deaths British Columbia: 1985 to 1993. Victoria, B.C.: Division of Vital Statistics, Ministry of Health and Ministry Responsible for Seniors, 1995. U.S. Department of Transportation. Alcohol and Highway Safety. Report to the U.S. Congress. Washington, D C : U.S. Department of Transportation, 1968. Waller, J.A. "Injury: conceptual shifts and preventive implications." Annual Review of Public Health 8 (1987):21-49. Waller, J.A. Injury Control: A guide to the causes and prevention of trauma. Lexington, Mass., D . C : Health and Company, 1985. Waller, J.A., Clemer D.I. A Scheme for Describing Injury Events. The Journal of Trauma. 1993; 35(6):909-919. Waters, C , L. Gibbons, R. Semenciu, and Y. Mao. "Motor Vehicle Traffic Accidents in Canada, 1978-87 by Time of Occurrence." Canadian Journal of Public Health 84 (1993):58-59. Wegman, D.H., and L.J. Fine. "Occupational Health in the 1990s." Annul Review of Public Health 11 (1990): 89-103. Wen, C P . and S.P. Tsai. "Anatomy of the healthy worker effect- a critique of summary statistics employed in occupational epidemiology." Scandinavian Journal of Work, Environment and Health 8 (1982):48-52. Wigle, D.T., Y . Mas, T. Wong, and R. Lane. "The economic burden of illness in Canada, 1986." Chronic Diseases in Canada 12 (1991): 1-37. Wilkins, K . "The major causes of death among young adults: trends from 1926 to 1985." Chronic Diseases in Canada 10 (1989):3-7. Wilson, R.J. A national household survey on drinking and driving. Knowledge, attitudes and behaviours of Canadian drivers. A report prepared for the Road Safety and Motor Vehicle Directorate. Catalogue No. TP5865-E62,68. Ottawa: Transport Canada, 1984. Workers' Compensation Board of British Columbia. Accident Prevention Regulations. B.C.Reg.43/66. Vancouver, B.C.: Workers' Compensation Board of British Columbia, 1966. Workers' Compensation Board of British Columbia. Annual Report 1969. Richmond, B.C.: Workers' Compensation Board of British Columbia, 1970. Workers' Compensation Board of British Columbia. Annual Report 1999. Richmond, B.C.: Workers' Compensation Board of British Columbia, 2000. 164 Workers' Compensation Board of British Columbia. News. Volume 34, No. 2. Richmond, B.C.: Workers' Compensation Board of British Columbia, 1992. World Health Organization: Manual of the International Classification of Diseases, Ninth Revision. Geneva: WHO, 1977. 165 Appendix A Location of study sawmills Location of study sawmills Mill - Region Mill - Region M i l l 1 - Vancouver Lower Mainland M i l l 3 - Vancouver Lower Mainland M i l l 5 - Vancouver Lower Mainland M i l l 2 - Vancouver Island (Chemainus) M i l l 4 - Vancouver Island (Port Alberni) M i l l 6 - Vancouver Island (Port Alberni) M i l l 8 - Vancouver Island (Youbou) M i l l 9 - Vancouver Island (Tahsis) M i l l 10 - Vancouver Island (Nanaimo) M i l l 7 - Coast Garibaldi (Squamish) M i l l 16 - Coast Garibaldi (Powell River) M i l l 11 - Interior (Kelowna) M i l l 13 - Interior (Mackenzie) M i l l 15 - Interior (Mackenzie) 167 Appendix B - Data collection forms HUMAN FACTORS STUDY N° SEX M f ~ i F n BIRTH DATE ~ I M r n . Y AGE BIRTH PLACE: CANADA | | OTHER COUNTRY:. MARITAL STATUS: S • M • W Q D | | START DATE INDUSTRY D START DATE JOB D OCCUPATION: 1 1 i M l • l Y l | | N |Y | | | C H I L D R E N : Y • N • # C H I L D R E N : S T A R T D A T E M I L L D 1 1 IMl | Y | | | 1 M | | Y | | | INJURY DATE AND TIME D 1 1 | M | 1 | Y | I I IAM I I I I I P M 1 1 :l 1 ACTIVITY AT TIME OF INJURY: EMPLOYMENT STATUS: EMPLOYED | | UNEMPLOYED | | RETIRED | | TRAINING RECEIVED: Y u N L PROTECTIVE EQUIPMENT USED: Y • ALONE AT SCENE OF INJURY Y n PART OF BODY: CHEST ABDOMEN MULTIPLE SITES HEAD/FACE UPPER EXTREMITIES LOWER EXTREMITIES INTERNAL ORGANS . TYPE OF EVENT: FALL-FROM ELEVATION FALL-SAME LEVEL STRUCK AGAINST STRUCK BY RUBBED-ABRADED CAUGHT IN/UNDER/BETWEEN . SOVEREXERTION VEHICLE ACCIDENT OTHER CAUSE OF DEATH: #1: #2: #3: #4: AUTOPSY Y | | N | | SUBSTANCE CONCENT. SPECIMEN ALCOHOL BENZODIAZEPINE MEDICATION BASIC DRUGS I [OTHER ACID DRUGS OPIATE . HEALTH PREVIOUS TO ACCIDENT: RESPIRATORY DISEASE ALCOHOL/DRUG ABUSE HEALTHY PSYQUIATRIC DIS. MEDICATION USE CARDIOVASCULAR/HYPERTENSION OTHER CHRONIC DISEASE V E H I C L E FACTORS STUDY N° Y N TYPE O F ENERGY • PROPER TOOL FOR TASK • • MECHANICAL EQUIPMENT FAILURE • • ELECTRICAL • FAILURE SAFETY DEVICES • • CHEMICAL • PROPER FRCCSDURES • • THERMAL • PROPER MAINTENANCE O F EQUIPMENT • • RADIANT SOURCE O F INJURY 1 • 2 • 1 2 HAND T O O L S MATERIAL HANDLING EQUIPMENT • • SAW • • : LADDER • • CONVEYOR • • MACHINERY • • W O O D / P L A N K S / B O A R D S / L O G S • • HOISTING APPARATUS • • FIREARM • • PERSON • • MOTOR VEHICLE (INDUSTRIAL) • • WORKING SURFACE • • MOTOR VEHICLE (ROAD/HIGHWAY) • • OTHER • • NATURE O F INJURY A S P H Y X I A / S T R A N G U L O R O W N I N G • A M P O T A T I O N • F O E C N I N G • B U R N S • C O N C U S S I O N • M U L T I P L E INJURIES • B R U I S E ' C O N T U S S y C R U S H I N G fj F R A C T U R E Q C U T / L A C E R A T I O N / P U N C T U R E Q OTHER Q ADDITIONAL INFO: 170 E N V I R O N M E N T A L - P H Y S I C A L F A C T O R S STUDY N ° MILL : L O C A T I O N : # E M P L O Y E E S AMBIENT CONDITIONS EXTREME TEMPERATURE: ABOVENORMAL • BELOW NORMAL |~] RAIN: Y • N • VISIBD_rTY: G O O D | | B A D | | HOUSEKEEPING: GOQp| | BAD f J J OTHER ADVERSE CONDITIONS: INJURY PLACE WORKPLACE URBAN • RURAL • OUTDOOR • INDOOR • AREA/REGION HOME/RESIDENCE • • • • HIGHWAY/ROAD • • OTHER: FIRST AID AT SCENE: Y Q N Q TRANSPORTATION TO HOSPITAL: AMBULANCE • OTHER Q CONCURRENT INCIDENT OTHER/INJURED Y • N Q DEAD A l SCENE • HOSPITAL:.. .ADDITIONAL INFO-CONTRIBUTORY FACTORS 171 E N V I R O N M E N T A L SOCIO C U L T U R A L F A C T O R S STUDY NP INQUIRY Q INQUEST • Y N SUPERVISOR P R E S E N T • j | WRITTEN S A F E T Y P R O C E D U R E S REQUIRED • | | S A F E T Y COMMfTTES • O Y N • • W C B R E P O R T TRAJNING/INSTRUCTIONS PROVIDED • • UMON • • VIOLATIONS OF OHS REGULATIONS WCB FINDINGS-INCIDENT ATTRIBUTED TO: WORKPLACE/ORGANIZATIONAL [ | HUMAN FACTORS [_| ACTION TAKEN AS A RESULT OF THE ACCIDENT: Y N FAMILY PROBLEMS • f £ ] WORKPLACE PROBLEMS • • EMOTIONALLY STABLE Q CORONER'S CLASSIFICATION OF DEATH SUICIDE • ' HOMICIDE • ACCIDENTAL Q NATURAL UNDETERMINED | | ADDITIONAL INFO - CORONER'S RECOMMENDATIONS: Y/N Appendix C - Job title groupings according to task, process area, and skill level Similar task Job Title Foreman/Supervisor Dept Foreman Logs Foreman Manager/Superintendent M i l l Foreman Yard Foreman Inspector, grader, other skilled worker (non tool user) Grader Ki ln Operator Marker Scaler Scowman Stenciller Tallyman Machine operators/attendants/clearers/sorters Barker Operator Boomman Breakdown Chipnsaw Operator Chipper Feeder Chipper Operator Cut-Off Sawyer Dogger Dropsorter Edger Operator End Loader Operator End Sealer Gang Sawyer Head Sawyer Hog Operator Hoist Operator J-Bar Sorter Operator J-Bar Sorter Patrolman Jump Rolls Operator Leverman Log Deck Operator Package Press Operator Planer Feeder Planerman Pony Edger Operator Resawyer Similar task Job Title Machine operators/attendants/clearers/sorters Sawyer Setter Stacker Operator Tailsawyer Timber Deckman Trimmer Operator Unscrambler Operator Material handling/unskilled Blockman Casual Labour Clean Up Helper Labour Lumber Straightener Offbearer Slab Picker Slipman Spotter Swamper Wire Tie Woodpicker Mobile equipment operator Boat Operator Carrier Driver Cat Operator Crane Operator Forklift Log Loader Operator Truck Driver Non-wood industry & non-production Clerk Fire Protection First A i d Shortage Clerk Watchman Yard Locomotive Skilled trades (Use tools) Benchman Carpenter Chipper Maintenance Electrician Similar task Skilled trades (Use tools) Job Title Machinist Mechanic Millwright Oiler Painter Pipefitter Power House Maintenance Sawfiler Sawfiter Steam Engineer Utility Welder Process Area Job Title Chipping and related op. Chipper Feeder Chipper Operator Scowman Clean-up Debarking General Plant functions Casual Labour Clean Up Helper Labour Barker Operator Clerk Fire Protection First Aid Manager/Superintendent Shortage Clerk Watchman Ki ln drying Ki ln Operator Maintenance Benchman Carpenter Chipper Maintenance Electrician Machinist Mechanic Millwright Oiler Painter Pipefitter Power House Maintenance Sawfiler Sawfiter Steam Engineer Welder Non-cutting equipment- (Sorting and clearing equipment) Breakdown Dropsorter J-Bar Sorter Operator J-Bar Sorter Patrolman Jump Rolls Operator Process Area Job Title Non-cutting equipment- (Sorting and clearing equipment) Leverman Log Deck Operator Tailsawyer Timber Deckman Planing Planer Feeder Planerman Sawing, trimming and edging Sorting and grading Chipnsaw Operator Cut-Off Sawyer Dogger Edger Operator Gang Sawyer Head Sawyer Hog Operator Pony Edger Operator Resawyer Sawyer Setter Trimmer Operator Unscrambler Operator Utility Grader Lumber Straightener Offbearer Slab Picker Spotter Woodpicker Stenciling, packaging, shipping End Loader Operator End Sealer Marker Package Press Operator Stacker Operator Stenciller Tallyman Wire Tie Yard and pond Blockman Boat Operator Process Area Job Title Yard and pond Boomman Carrier Driver Cat Operator Crane Operator Forklift Hoist Operator Log Loader Operator Logs Foreman Scaler Slipman Swamper Truck Driver Yard Foreman Yard Locomotive Skill level Job Title Clerical Clerk Shortage Clerk Maintenance Benchman Carpenter Chipper Maintenance Electrician Machinist Mechanic Millwright Oiler Painter Pipefitter Planerman Power House Maintenance Sawfiler Sawfiter Steam Engineer Welder Yard Locomotive Boat Operator Carrier Driver Cat Operator Chipnsaw Operator Crane Operator Cut-Off Sawyer Edger Operator First A i d Forklift Gang Sawyer Grader Head Sawyer Hoist Operator J-Bar Sorter Operator Ki ln Operator Log Deck Operator Log Loader Operator Marker Package Press Operator Skilled labour Barker Operator Skill level Job Title Skilled labour Unskilled Labour Pony Edger Operator Resawyer Sawyer Scaler Stacker Operator Tallyman Timber Deckman Trimmer Operator Utility Dept Foreman Logs Foreman Manager/Superintendent M i l l Foreman Yard Foreman Blockman Boomman Breakdown Casual Labour Chipper Feeder Chipper Operator Clean Up Dogger Dropsorter End Loader Operator End Sealer Fire Protection Helper Hog Operator J-Bar Sorter Patrolman Jump Rolls Operator Labour Leverman Lumber Straightener Offbearer Planer Feeder Scowman Setter Slab Picker Slipman Spotter Skill level Job Title Unskilled Labour Stenciller Swamper Tailsawyer Truck Driver Unscrambler Operator Watchman Wire Tie Woodpicker Appendix D - External causes of death categories by ICD revision > OS ON Q U o 00 I o o o ON o so 00 I o o NO oo O OS o «-> KO so OO 00 OS SO so 00 o so SO OS so so oo OS r-o oo o o o oo o o 00 o oo so °°» Os" o SO oo I o o 00 OS OS so oo o OS so oo Os" r-SO OS 00 , 9230,9232,923 co CN OS , 9230,9232,923 OS Os co OS , 9230,9232,923 OS Os co Os , 9230,9232,923 o CN , 9230,9232,923 OS OS os OS , 9230,9232,923 o o o , 9230,9232,923 t^- OS co OS , 9230,9232,923 o CN CN , 9230,9232,923 Os OSK CTs^ OS OS OS OS Os" OS OS Os OS OS os" os" OS OS OS Os" OS OS OS OS 00 OS <N o co «o SO o oo CN "3- >o so OS OS Os oo SO 00 OS o CN 1^- CN <N CN ( N r~ </"> SO 00 00 00 OS OS OS OS OS OS oo OS OS OS OS oo OS OS OS o o o o o o o o o o o o o O o o o o o o o o SO o so •* in so o o o o SO oo OS o <N "d- CN rs| (N <N r- <n so r-oo oo oo OS OS OS OS OS OS oo OS OS OS OS oo OS OS OS OS oo co co CM CM OS Os CN OS co OS co OS co OS CM o CN -d-OS OS °\ os o" o o co co o <N o CN OS Os OS OS OS OS- os" OS OS OS OS OS OS os" OS OS OS Os" OS OS OS r- OS CN o co ' O so o oo CN «-> SO so OS OS 00 oo OS o <N <N (N CN CN CN m </-> so r-oo oo 00 OS OS Os OS OS OS OS OS OS OS Os Os OS OS OS o o o o o o o o o o o o o o o o o o o o o o o SO o r- •q- <n SO o o o o r- 00 OS o O J CN CN CN CN CM >o SO l> 00 00 00 OS OS Os OS OS OS OS OS OS OS OS Os Os OS OS OS OS Os OS OS OS OS «o «-l m OS CN OS CO co CN <N SO r- oo Os OS OS OS Os OS OS o o o o o o o o co Os o o o co <N CN 1/1 00 OS °\ °\ °\ OS OS OS OS Os Os Os" OS os" os" OS OS os" OS OS OS Os" Os" Os" OS «n SO oo OS CM o o CO OS «/-> r- oo SO m OS o CN CN CN ( N <N 3^- so so oo 00 OS OS OS OS OS OS OS OS Os • OS OS OS OS OS OS OS OS o o o O o o o o o o o o o O o o o o o SO SO OS o o co o «/-> r-- 00 o m OS o CN CN CN <N 1—1 »—< f—1 .—i «—i — so so oo oo Os Os OS OS OS OS OS OS OS OS OS OS OS OS OS Os OS <u o ffects I w x 1 in O J3 ing1 3 m ors ing1 jec 00 <u o apour IFact o Pierc :hPrp TOSJVi ment > cd m X? gor & o O en c o gor ne u ID c Gases i vironm c Foreig Falling Cuttin Machi or Sho Hot or ion c o .S3 idical T rventio by En 60 c ffocatio by >> by by on by 3 o Rai u S de u UOSI in rns o owni ffocatio used uck 1 used used plosi used lectro^ o u o 1) icide imici gal Ii o 3 3 cd S3 cd X lectro^ 3 3 3 <x m Q Q C/3 U </} U U w U w Q Q VI X 1—1 OS O OS r- os —< oo Os Os OS o m o O OO OS OS o os" CO oo O CN OS OS o OS CN OS Os" OS OS o OS t-- OS oo Os OS Os" o oo CO o OS o oo OS OS °\ OS o Os" o ro OS o O CN OS OS OS OS OS Q u o o o oo o o ^r oo o o CO 00„ Os" >n CN oo OS oo O OS O «o — 00 oo OS so SO oo I o o so OS 00 oo u o en •a 4> l> 1-O O 'I 60 l l O i l X! 60 3 cd U 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0090596/manifest

Comment

Related Items