Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Certification in hazardous industries : an evaluation of the British Columbia Faller Training Standard Sarkany, Daniel 2011

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

Item Metadata

Download

Media
24-ubc_2011_fall_sarkany_daniel.pdf [ 1.44MB ]
Metadata
JSON: 24-1.0072114.json
JSON-LD: 24-1.0072114-ld.json
RDF/XML (Pretty): 24-1.0072114-rdf.xml
RDF/JSON: 24-1.0072114-rdf.json
Turtle: 24-1.0072114-turtle.txt
N-Triples: 24-1.0072114-rdf-ntriples.txt
Original Record: 24-1.0072114-source.json
Full Text
24-1.0072114-fulltext.txt
Citation
24-1.0072114.ris

Full Text

Certification in Hazardous Industries: An Evaluation of the British Columbia Faller Training Standard  by  Daniel Sarkany  B.H.K., The University of British Columbia, 2006  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE  in  THE FACULTY OF GRADUATE STUDIES (Occupational and Environmental Hygiene)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August, 2011  © Daniel Sarkany, 2011  Abstract Manual tree falling is historically one of the most hazardous occupations in British Columbia (BC). In an effort to reduce injury rates in this trade, WorkSafeBC implemented regulations requiring mandatory safe work practice training and certification known as the BC Faller Training Standard (BCFTS). The goal of this thesis was to assess if mandatory certification is associated with changes in injury rates and risk of injury for manual tree fallers. Two studies were conducted to determine effects of certification: 1) a comparison of manual tree faller specific injury rates for each forest region of BC during pre and post certification periods and 2) a comparison of the individual level risk of injury associated with certification status for a cohort of manual tree fallers. Annual injury rates representing the number of injuries/10,000,000m3 wood for each BC forest region were constructed using workers‟ compensation data obtained from WorkSafeBC and wood harvest volume data from the Ministry of Forests and Range. The individual level relative risk (RR) of injury associated with certification status was determined using WorkSafeBC workers compensation data linked to a cohort listed in the BC Forest Safety Council faller registry. The injury rate study indicated that following implementation of mandatory safe work practice certification, acute injury rates in the Coast Forest Region dropped from 33.5 injuries/10,000,000m3 in 2006 to 25.8 injuries/10,000,000m3 in 2008. The cohort study based on experienced fallers who obtained grandfathered certification found that individual level risk of injury did not change significantly (RR=1.11, 95%CI 0.58,2.14) for the year immediately following certification compared to the year before certification. Combining these two study outcomes suggests the BCFTS may have an effect in the region of the province with the highest level of manual tree falling, although the individual level risk analysis suggests that the effects may take longer to be realized in the industry as a whole. ii  Preface Both studies conducted for this thesis involved the use of administrative data that contained personally identifying information for linkage purposes only. Personally identifiable fields were removed and replaced with anonymous study identifiers in the linked database provided to the researchers. The use of this data required separate research ethics approvals for each study. All study procedures were approved by the University of British Columbia Behavioural Research Ethics Board. The certificate number for the ecologic manual tree faller injury rate study was H09-01090. The certificate number for the individual level risk of injury study was H09-02725. Both ethics approval certificates are available through the University of British Columbia‟s Office of Research Services. The research also involved research agreements covering the conditions for the use of administrative data with WorkSafeBC and the British Columbia Forest Safety Council.  iii  Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ......................................................................................................................... iv List of Tables .............................................................................................................................. viii List of Figures ............................................................................................................................... ix List of Abbreviations .................................................................................................................... x Acknowledgements ...................................................................................................................... xi 1.  INTRODUCTION ................................................................................................................. 1  1.1. The BC Faller Training Standard .......................................................................................... 1 1.2. Description of Forestry Tasks and Working Environments - The Forest Harvesting Process and Role of Manual Tree Fallers........................................................................................... 3 1.3. Geographic Variation in Forestry Environments .................................................................. 5 1.4. Manual Tree Falling in British Columbia ............................................................................. 6 1.4.1. Demographics of the Manual Tree Faller Population ..................................................... 8 1.5. WorkSafeBC and the Forestry Subsector ............................................................................. 8 1.6. Injury Rates in the Forestry Subsector and Manual Tree Falling ....................................... 10 1.7. Mandatory Faller Certification as a Safety Intervention..................................................... 12 1.8. Study Rationale ................................................................................................................... 15 1.9. Objectives and Research Questions .................................................................................... 18 2.  METHODS ........................................................................................................................... 19  2.1. Study 1: Comparison of Workers‟ Compensation Claim Rates for Manual Tree Fallers. . 19 2.1.1. Data Sources ................................................................................................................. 20 2.1.2. Study Variables ............................................................................................................. 20 2.1.2.1.  Injury Type .......................................................................................................... 20  2.1.2.2.  Claim Region....................................................................................................... 23  2.1.2.3.  Wood Harvest Volume ........................................................................................ 23 iv  2.1.3. Analysis ........................................................................................................................ 24 2.1.3.1.  Data Cleaning ...................................................................................................... 25  2.1.3.2.  Descriptive Statistics ........................................................................................... 25  2.2. Study 2: Cohort Study of Risk of Injury Associated with Faller Certification Status……………………………………………………………………………………... 26 2.2.1. Data Sources ................................................................................................................. 26 2.2.2. Study Cohort ................................................................................................................. 26 2.2.3. Study Follow-up Period ................................................................................................ 28 2.2.4. Study Variables ............................................................................................................. 28 2.2.4.1.  Injury Outcome Variables ................................................................................... 28  2.2.4.2.  Fixed Covariates .................................................................................................. 30  2.2.4.2.1.  Age Group ....................................................................................................... 30  2.2.4.2.2.  Work Experience Group .................................................................................. 31  2.2.4.2.3.  Faller Type ....................................................................................................... 31  2.2.4.2.4.  Region .............................................................................................................. 31  2.2.4.2.5.  Grade Level of Certification ............................................................................ 32  2.2.4.3.  Time Varying Covariates .................................................................................... 32  2.2.4.3.1  Previous Claims ............................................................................................... 32  2.2.4.3.2  Unemployment Rate ........................................................................................ 32  2.2.5. Analysis ........................................................................................................................ 33  3.  2.2.5.1.  Survival Analysis and Censoring ........................................................................ 33  2.2.5.2.  Multivariable Model Construction and Model Fitting ........................................ 36  RESULTS ............................................................................................................................. 39  3.1. Study 1: Comparison of Workers‟ Compensation Claim Rates for Manual Tree Fallers . 39 3.1.1. Description of Workers‟ Compensation Claims among Fallers in British Columbia .. 39 3.1.2. Description of Regional Harvest Volumes (Denominator Variable) ........................... 40 3.1.3. Regional Injury Rates ................................................................................................... 41 v  3.2. Study 2: Cohort Study of Risk of Injury Associated with Faller Certification Status…. .. 43 3.2.1. Demographic and Occupational Cohort Characteristics ............................................... 43 3.2.2. Cohort Certification Characteristics ............................................................................. 43 3.2.3. Injury Rates Among the Faller Cohort ......................................................................... 46 3.2.4. Multivariable Model Fitting ......................................................................................... 49 3.2.5. Adjusted Survival Analysis Results.............................................................................. 50 3.2.6. Risk of Injury - Sensitivity Analysis ............................................................................ 54 4.  DISCUSSION ....................................................................................................................... 55  4.1. Summary of Key Findings .................................................................................................. 55 4.2. Outcomes of Certification Through a “Grandfathering” Process ....................................... 56 4.3. Synthesis of Covariate Findings ......................................................................................... 60 4.4. Policy Implications ............................................................................................................. 63 4.5. Study Limitations ................................................................................................................ 65 4.5.1. Selection Bias ............................................................................................................... 65 4.5.2. Information Bias ........................................................................................................... 67 4.5.3. Reporting Bias .............................................................................................................. 68 4.5.4. Unmeasured Confounders ............................................................................................ 68 4.6. Study Strengths ................................................................................................................... 69 4.6.1. Study Design ................................................................................................................. 69 4.6.2. Applicability of the Study Results ................................................................................ 71 4.7. Future Studies ..................................................................................................................... 73 4.7.1. New Faller Trainees ...................................................................................................... 73 4.7.2. Data Collection ............................................................................................................. 74 5.  CONCLUSIONS .................................................................................................................. 76  BIBLIOGRAPHY ....................................................................................................................... 77 APPENDIX A: WORKERS’ COMPENSATION CLAIM RATE INCIDENT LOCATION ASSIGNMENT ............................................................................................................................ 81 vi  APPENDIX B: WORKERS’ COMPENSATION CLAIM RATE CLAIM INCLUSION AND EXCLUSION CRITERIA BASED ON NATURE OF INJURY (NOI), SOURCE OF INJURY (SOI) AND ACCIDENT TYPE (ACC). .................................................................... 83 APPENDIX C: DISCRETE SURVIVAL ANALYSIS CLAIM INDICATOR INCLUSION AND EXCLUSION CRITERIA BASED ON NATURE OF INJURY (NOI), SOURCE OF INJURY (SOI) AND ACCIDENT TYPE (ACC). .................................................................... 84 APPENDIX D: DISTRIBUTION OF MANUAL TREE FALLER CLAIMS BY CLASSIFICATION UNIT (CU) INCLUDED IN THE WORKERS’ COMPENSATION CLAIM RATE STUDY COHORT. .......................................................................................... 85 APPENDIX E: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE COAST FOREST REGION OF BC ........................................................................................................ 86 APPENDIX F: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE SOUTHERN INTERIOR FOREST REGION OF BC ........................................................... 87 APPENDIX G: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE NORTHERN INTERIOR FOREST REGION OF BC ........................................................... 88 APPENDIX H: RESULTS OF MULTIVARIABLE MODEL FITTING ............................. 89 APPENDIX I: REGRESSION COEFFICIENTS FOR NON-HCO, ACUTE, STRAIN AND SERIOUS INJURIES MEETING THE SPECIFIC FALLING DEFINITION…………………………………………………………………………………...92 APPENDIX J: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE BROAD MANUAL TREE FALLING CLAIM DEFINITION (2 YEAR FOLLOWUP). ............................................................................................................................................... 93 APPENDIX K: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE SPECIFIC MANUAL TREE FALLING CLAIM DEFINITION (4 YEAR FOLLOWUP). ............................................................................................................................................... 95 APPENDIX L: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE BROAD MANUAL TREE FALLING CLAIM DEFINITION (4 YEAR FOLLOWUP). ............................................................................................................................................... 97  vii  List of Tables Table 1.1: WorkSafeBC forestry subsector classification units (CU‟s) ......................................... 9 Table 1.2: Average injury rates for the WorkSafeBC forestry subsector by CU for the years 2003-2007 ..................................................................................................................................... 12 Table 1.3: Example of Haddon's Matrix as applied to manual falling injuries............................. 17 Table 2.1: Nature of Injury (NOI) claim criteria for acute and strain injury types. ...................... 21 Table 2.2: Source of Injury (SOI) and Accident Type (ACC) claim criteria for inclusion as acute, strain or serious injuries. ............................................................................................................... 23 Table 2.3: Ministry of Forests and Range harvest billing system wood volume report configuration parameters. ............................................................................................................. 24 Table 2.4: Claim inclusion criteria based on Statistics Canada SOC and WorkSafeBC occupation code combinations for specific/ broad faller claims. .................................................................... 29 Table 3.1: Distribution of acute and strain claims by forest region for the years 1997-2008. ..... 39 Table 3.2: Distribution of claims by age group and injury type for the years 1997-2008. ........... 40 Table 3.3: Annual injury rates of the Coast, Southern Interior and Northern Interior Forest Regions indicating acute, strain and serious injury types. ............................................................ 42 Table 3.4: Demographic, occupational and certification characteristics of the cohort. ................ 45 Table 3.5: Specific manual tree falling claim injury rates. ........................................................... 48 Table 3.6: Results of adjusted, discrete survival analysis. ............................................................ 53  viii  List of Figures Figure 1.1: Map of British Columbia indicating the Coast, Northern Interior and Southern Interior Forest Regions ................................................................................................................... 6 Figure 1.2: Annual provincial forestry subsector non-HCO injury rates for the years 1997-2009 ....................................................................................................................................................... 11 Figure 2.1: Classification hierarchy for injury outcome variables. .............................................. 30 Figure 2.2: Survival analysis structure, discrete quarters, and individual date of certification. ... 33 Figure 3.1: Annual forest harvest volumes for the Coast, Northern Interior and Southern Interior forest regions of BC for the years 1997 – 2008 ............................................................................ 41  ix  List of Abbreviations ACC – Accident Type AIC – Akaike Information Criterion BC – British Columbia BCFTS – British Columbia Faller Training Standard BIC – Bayesian Information Criterion CU – Classification Unit FTE – Full Time Equivalent FTL - Fatal HBS – Harvest Billing System LTD – Long Term Disability NOI – Nature of Injury non-HCO – non Health Care Only PopDataBC – Population Data British Columbia SOC – Standard Occupational Classification (1991) SOI – Source of Injury STD – Short Term Disability RR – Relative Risk  x  Acknowledgements I would like to thank my thesis committee members, Dr. Mieke Koehoorn, Dr. Chris McLeod, Dr. Hugh Davies and Dr. Kevin Lyons, for their insights, guidance and contributions to this thesis. Every time they challenged me, not only did my skills in occupational health research improve, but my perspective of the world around me expanded. Their efforts ensured the success of this study and the development of a graduate student who each have made a great impression upon. Their encouragement and mentorship led to achievements and opportunities that I would not have achieved alone. I am truly grateful for the time spent with each one these people and the input they have had in the development of my academic and professional life. I am very thankful for the funding that WorkSafeBC provided for this project in the form of an operating grant and for the personal support through a research training award. Without such support this project would not have been possible. I greatly appreciated the collegial and supportive staff of the Center for Health Services and Policy Research/WorkSafeBC partnership, Dawn Mooney, Fan Xu, Lillian Tamburic, Jonathan Fan and Suhail Marino, whose assistance was more than helpful throughout all stages of this thesis. Their technical knowledge of how to work with administrative data was invaluable and daily friendship made the office a great place to be. The motivation to complete this thesis and pursue opportunities along the way is significantly attributable to my wife Kelly. Her consistent support played a huge role in my success. Throughout our years in university she always told me that it‟s not just about reaching goals, but also enjoying the journey along the way. Looking back over the years, I have enjoyed every moment of the journey thanks to Kelly.  xi  1.  INTRODUCTION  Forestry and logging has traditionally had the largest employment within the primary resource sector in the province of British Columbia (BC) (1), but given the occupational and environmental hazards associated with this industry, it has some of the highest occupational injury rates (2). Within forestry, manual tree falling is the most hazardous occupation and has been recognized by the WorkSafeBC (the provincial workers‟ compensation system in British Columbia) Forestry Task Force as the job in which a serious injury is most likely to occur (2). For example, between 1998 and 2001, there were 26 falling-related fatalities and over 1,400 faller injury claims in the province (3) out of an estimated population of 4710 chainsaw and skidder operators (4). This occupation and the hazards associated with it are not unique to BC; manual tree falling has been identified as the most dangerous occupation in forestry for other Canadian provinces such as Ontario (5) and countries around the world including Australia (6), China (7), New Zealand (8) and the United States (9). In an effort to reduce the injury rates of manual tree fallers, WorkSafeBC implemented the BC Faller Training Standard (BCFTS) to ensure that all tree fallers in the BC forest industry have the skills to meet industry best practices. This thesis evaluates what outcomes the BCFTS had as a safety intervention on the incidence of workplace injuries for the occupation of manual tree falling. 1.1. The BC Faller Training Standard WorkSafeBC implemented sections 26.21 (Faller Qualifications) and 26.22 (Forestry Operation Faller Training) of the Occupational Health and Safety Regulation (10) on August 1st 2006 as a safety intervention to reduce the incidence of tree falling injury by standardizing manual tree falling best practices. Section 26.22 stipulates that a worker may not fall trees1 in a forestry  1  Fallers are not required to meet the criteria of section 26.22 if a tree has a trunk diameter less than 15cm at 30cm off the ground.  1  operation unless a worker is certified in writing as a “competent faller” (10). According to the WorkSafeBC Forestry Task Force a competent2 faller was considered to be a worker that had the “knowledge, skills, abilities, work practices and attitude that enables a worker to function as a safe and productive faller” (2). Thus, the BCFTS was developed to standardize training and industry best practices for tree fallers in the forestry sub-sector. Upon obtaining BCFTS certified status, fallers were deemed to be competent and to have satisfied the requirements for sections 26.21 and 26.22 of the Occupational Health and Safety Regulation. BC Faller Training Standard certification was first introduced in 2003 as an optional requirement to work as a tree faller in BC, but became a mandatory requirement in 2006. Two methods are currently available to obtain a competent faller status and are contingent on the amount of work experience a faller has at the time of evaluation. Fallers with less than 2 years of tree falling work experience are required to complete a 30 day course that is a combination of classroom and field based components. Upon completion of the 30 days of coursework “new faller trainees” must have up to 180 days3 of on-the-job training with a qualified supervisor/trainer before becoming eligible to challenge the tree falling competence evaluation. The evaluation to become a BCFTS certified faller consists of a written/oral examination and a practical skills evaluation in the field to test safe falling procedures and falling knowledge. Faller trainees with a grade of 75% or greater on each portion of the exam are deemed to be a competent faller and designated with a BCFTS certified status. Fallers with more than 2 years4 of tree falling work experience are  2  Regulation 26.22 of the WorkSafeBC Occupational Health and Safety Regulation (10) uses the term “competent” in the context of sub-section 1 where “a worker may not work as a faller in a forestry operation unless the worker receives training for falling that is acceptable to the Board and is certified in writing as a competent faller”. 3 “A shorter period as determined by a qualified supervisor or trainer”… can be less than 180 days …”if the supervisor or trainer is satisfied that the worker is competent to perform the tasks of a faller” (10). 4 Some fallers with less than 2 calendar years of experience were permitted to attempt the faller training standard course challenge because they were deemed to have completed an equivalent amount of work experience in a shorter time period.  2  eligible for the BCFTS Course Challenge by completing the same evaluation as the new faller trainees without taking the 30 day course or the on-the-job training. Meeting the same passing criteria as new faller trainees, experienced fallers are “grandfathered” into a BCFTS certified status and deemed competent fallers. The two methods to obtain BCFTS status results in two groups of fallers: new fallers with standardized training and minimal prior work experience and „grandfathered‟ fallers with variation in job training and prior work experience. 1.2. Description of Forestry Tasks and Working Environments - The Forest Harvesting Process and Role of Manual Tree Fallers Most individuals who work in the forest industry can be classified as having an occupation in one of the sequential stages of planning, harvesting or silviculture. These occupations are, for the most part, performed by skilled laborers who typically work in one of the preceding stages. The initial stage of planning involves the survey of an area by a “timber cruiser” who assesses what types of trees are present, what wood volume may be yielded by an area and what environmental limitations a site may have. If it is deemed that an area is feasible and economical to harvest then a timber cruiser will lay-out5 a cut block6 by specifically distinguishing which areas should be harvested. Based on the location and characteristics of a cut block the next phase, operations planning, determines the method of harvesting which subsequently dictates the location of access roads and landings7. The second stage of harvesting begins with the clearing of the rights of way for roads and landings by tree fallers. Road construction is then accomplished using heavy machinery to excavate road beds and prepare a road surface that can be used in all weather conditions. Blasting is often utilized not only to construct roads through rocky terrain, but also to generate 5  The process of marking harvest area boundaries and which trees should be cut down or left standing. An area to be logged that is distinguished by specific boundaries. 7 An area where logs are de-limbed, sorted and loaded onto trucks for transportation. 6  3  road bed materials that can support heavy loads. Upon completion of road and landing construction, tree fallers begin to cut down all selected trees within a cut block. This can be accomplished either manually with a chainsaw or mechanically with heavy equipment specialized for this purpose. After a cut block has had all harvestable trees cut down, logs are extracted utilizing one of four methods: ground skidding, forwarding, cable yarding or helicopter logging. Ground skidding involves the use of a machine or horse to drag logs on the ground out to the roadside or landing. Forwarding involves the use of heavy machinery to pick up logs in a cut block and carry them to a loading location without a log touching the ground. Cable yarding involves a system of winch driven cables suspended over a cut block that have points of attachment for grasping logs. Logs are grasped by attaching a grapple or by having cables wrapped around them by a “choker-man”, and then a “yarder-man” operates winches that pulls the logs out to a landing. Heli-logging (as its name implies) involves the use of a helicopter aerially lifting logs out of a cut block via a cable wrapped around logs by a choker-man or grasping logs with a grapple. Wood extracted from a cut block accumulates at a roadside or landing where logs are loaded onto trucks and hauled to a scaling site where the volume of wood in each load is estimated. In the case of remote coastal or lakeside logging operations, logs are transported to a shoreline where they are transferred into the water, assembled into “booms” or loaded onto barges and towed by a tug boat to an end use location. Silviculture is the third stage of forestry that involves the re-establishment of a harvested area back to a state that can be harvested again. Tree planters walk into harvested cut blocks and place seedlings into the ground. After a few years of growth “spacing” occurs to cut down any competing species of trees that are inhibiting the growth of seedlings and to provide space for the growth of juvenile trees. With ongoing silvicultural management a harvested cut block can ideally be logged again when the forest has re-established itself and has a merchantable value. 4  1.3. Geographic Variation in Forestry Environments Geographically, BC is divided into three forest regions; the Coast Forest Region, Northern Interior Forest Region and Southern Interior Forest Region (see Figure 1.1). These regions are characterized by their terrain, tree species, seasons and the predominant type of harvesting methods. Although all three regions have mountainous terrain, the extent of this type of terrain varies across regions. Mountainous terrain is the predominant geography in the Coast Forest Region, and is associated most commonly with steep slope harvesting methods such as manual tree falling, cable yarding, and helicopter logging. The Northern and Southern Interior Regions have a mix of low angle and steep terrain, but are most commonly associated with low angle terrain harvesting methods, such as mechanized tree falling, forwarding, and ground skidding. Seasonally, the Coast and Interior Forest Regions have different forestry cycles. The Coast Forest Region has its peak harvesting season typically during the summer months and its low season during the winter. Conversely, the Northern and Southern Interior Forest Regions have their peak harvesting season in the winter and their low season in the spring. The differences in harvesting cycles are attributed to freezing conditions that facilitate winter access to environmentally sensitive areas in the interior area of the province.  5  Northern Interior Forest Region  Coast Forest Region  Southern Interior Forest Region  Figure 1.1: Map of British Columbia indicating the Coast, Northern Interior and Southern Interior Forest Regions (11). 1.4. Manual Tree Falling in British Columbia Although some tree fallers are engaged in activities to clear land, such as right of way clearing8, tree spacing9, seismic line cutting10, forest fire suppression and arborist occupations, the predominant type of employment for manual tree fallers is working as “production” fallers in the forestry subsector where the objective is to cut down trees to harvest wood. The primary role of those who work as production fallers in forestry is to cut down trees for harvesting and if fallen trees are too large to extract from a cut block in one piece then they must cut logs into smaller sections (“bucking”). A typical working day begins with transportation to an area where active logging is occurring, arriving to a work site by sunrise and working at least  8  Clearing a strip of trees road or utility line construction. Cutting down trees that inhibit the growth of others. 10 Clearing a narrow strip of trees for the purpose of oil and gas exploration. 9  6  a 6.5 hour day. Methods of transportation to active logging sites can vary and includes using such modes as four wheel drive vehicles, boats and helicopters. Upon arrival to a work site, manual tree faller‟s hike to the face11 of a cut-block carrying their chainsaw, axe, wedges, tools, gasoline, chain oil, lunch and water which in total can weigh upwards of 90lbs (12). Manual tree fallers are assigned to work in a cut block where trees have been selected for harvest by designated boundaries and markings. Two or more fallers are permitted to work within the same cut block as long as they are a minimum distance of two tree lengths apart and not working uphill from each other. Upon approaching a tree to be cut down, a faller surveys the immediate area to identify hazards such as surrounding trees that may drop overhead debris, the distribution of branches in a tree to be felled, the direction of lean for a tree, and the condition of the trunk. Part of the preparation to cut down a tree also includes using a chain saw to clear an area around the tree base to reduce hazards in the immediate workspace and to provide escape routes. Trees in a cut block are typically felled in a controlled manner by the use of directional control. The purpose of using directional control to cut down trees includes safety but also minimizing trunk damage and facilitating the extraction of logs with specific placement. To direct the fall of a tree, a notch is cut out (“undercut”) from the side of a tree trunk oriented in the intended landing location. An undercut no greater than 1/3 of trunk diameter (depending on diameter and falling method) is removed to leave enough holding wood so that a tree will not topple in an uncontrolled manner. Then the holding wood is carefully removed by cutting a swath (“backcut”) into the trunk on the opposite side from the undercut. A wedge is hammered into a back-cut as soon as possible to inhibit a tree from leaning backwards and to provide leverage to topple a tree in the intended direction. As soon as there are indications that a tree is falling, a faller moves away from the tree along an escape route and takes cover in a safe location from any potential 11  The leading edge of a cut block where trees are being cut down.  7  overhead hazards. After a tree has been cut down and the surrounding area is deemed to be safe, (if required) a faller removes branches from the fallen tree (“limbing”) and bucks the trunk into specific lengths before moving on to another tree selected for harvesting. 1.4.1. Demographics of the Manual Tree Faller Population Publicly available information about the population characteristics of workers who perform manual tree falling is limited and is inferred from the occupational group “Chainsaw and Skidder Operators” that Statistics Canada collects through census and Labour Force Survey‟s. The chainsaw operating occupation is almost exclusively male, with only 2% of this occupation being female (13). Areas of residence for the chainsaw operating population reflect the geographic areas in where tree falling work is typically available. The majority (46%) of chainsaw operators live in the Coast Forest Region while 38% live in the Southern Interior and 16% live in the Northern Interior Forest Regions (13). The mean age of chainsaw operators is ageing. In 2001, the most numerous age group in this occupation was 35 to 44 year olds but by 2006 had shifted to the 45 to 54 year olds (13). During this same time period the proportion of self-employed chainsaw operators versus those employed by a company has decreased in the Coastal and Northern Interior and increased in the Southern Interior (13). Province wide, the proportion of self-employed chainsaw operators was 16% in 2006 (13). 1.5. WorkSafeBC and the Forestry Subsector WorkSafeBC is the regulatory agency responsible for workplace safety on provincial worksites in BC12. Its role is to ensure employer compliance with current safety regulations, promote the prevention of workplace injury, provide compensation/rehabilitation for injured workers and manage the finances for the workers compensation system. All phases of the forest harvesting 12  Some workplaces in BC such as railways and ports are under the exclusive jurisdiction of the federal government for occupational health and safety regulation.  8  processes must be carried out in accordance with WorkSafeBC safety regulations and guidelines stated in section 26 of the Occupational Health and Safety Regulation (10). In order to manage all aspects of workplace safety and compensation for the array of companies that conduct forestry work, WorkSafeBC has classified forestry related businesses into the forestry subsector13. The forestry subsector is composed of 15 different classification units (CUs) that represent the different types of primary industrial activities that are involved in the forest harvesting process (see Table 1.1). The purpose of classifying companies into CU‟s is to group businesses that perform similar primary industrial work activities for assigning insurance premiums based on similar risks. Not only does WorkSafeBC utilize the classification unit system to determine yearly insurance premiums to each employer, but it also uses this system as a basis for descriptive statistics to indicate where workplace accident prevention may be required. Table 1.1: WorkSafeBC forestry subsector classification units (CU’s) (14). Classification Unit Number 703002 703003 703004 703005 703006 703008 703009 703011 703012 703013 703014 703015 703016 703018 703019  13  Classification Unit Description Brushing and Weeding or Tree Thinning or Spacing (not elsewhere specified) Cable or Hi-Lead Logging Dry Land Sort Forest Fire Fighting Ground Skidding, Horse Logging, or Log Loading Integrated Forest Management Log Booming Log Processing Logging Road Construction or Maintenance Manual Tree Falling and Bucking Mechanized Tree Falling Shake Block Cutting Tree Planting or Cone Picking Marine Log Salvage Helicopter Logging  The forestry subsector is nested within the primary resources sector.  9  WorkSafeBC determines the safety performance of each subsector and CU by calculating annual injury claim rates. To calculate the annual claim rate for each CU, WorkSafeBC determines the ratio of injury claims to the estimated number of workers classified in a CU defined as full-time equivalents (FTE)14:  Where the numerator is the total annual number of CU specific accepted15 non-health care only (non-HCO) claims defined as short term disability (STD), long term disability (LTD) and fatal (FTL) claims; and the denominator is the estimated FTE calculated as the amount of time employees have worked based on total payroll for all of the employers in a CU and an average CU specific annual income. To facilitate comparison of claim rates, WorkSafeBC standardizes claim rates as the number of injuries per 100 FTE person years. 1.6. Injury Rates in the Forestry Subsector and Manual Tree Falling The annual injury claim rate16 for the forestry subsector in British Columbia has decreased since 1997 with the lowest injury rate of 5 claims/100 person years in 2009 (see Figure 1.2). The decrease in claim rates suggests that the forestry subsector as a whole has become a safer industry since 1997, similar to the trend of the overall provincial claim rate (see Figure 1.2). However, the forestry subsector has experience a much greater decline in injury rates when comparing 1997 to 2009. Despite the fact that the gap between the forestry sub-sector claim rate and the overall provincial claim rate has decreased, the forestry subsector has consistently had approximately twice the claim rate of the provincial average, indicating that even though the 14  One FTE person year represents one person working full time for one year. The annual amount of FTE person years for a CU is determined by dividing the total annual reported payroll of a CU by the average annual wage paid for that CU. 15 Claims that have been accepted by WorkSafeBC to receive compensation for lost wages and health care costs. 16 Number of non-HCO claims per 100 person-years of employment.  10  incidence of injuries has dropped, the forestry subsector is still one of the more dangerous sectors to work in.  Injury Rate (number of non-HCO claims/100 person years)  Annual Provincial Subsector Injury Rates 12 11 10 9 8 7 6 5 4 3 2 1 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Year Forestry Subsector  All Subsectors  Figure 1.2: Annual provincial forestry subsector non-HCO injury rates for the years 19972009 (15-17). There is a wide range in injury claim rates within the forestry subsector CU‟s as evidenced by the low injury rate of 2 claims/100 person years in the Log Processing CU and the very high injury rate of 33 claims/100 person years in the Manual Tree Falling and Bucking CU (see Table 1.2). Based on the CU specific injury rates in the forestry subsector, the Manual Tree Falling and Bucking CU has the most dangerous primary industrial activity – manual tree falling.  11  Table 1.2: Average injury rates for the WorkSafeBC forestry subsector by CU for the years 2003-2007 (18). Classification Unit Number 703013 703015 703003 703019 703002 703005 703016 703009 703004 703006 703008 703012 703014 703011 703018  Classification Unit Description Manual Tree Falling and Bucking Shake Block Cutting Cable or Hi-Lead Logging Helicopter Logging Brushing and Weeding or Tree Thinning or Spacing (not elsewhere specified) Forest Fire Fighting Tree Planting or Cone Picking Log Booming Dry Land Sort Ground Skidding, Horse Logging, or Log Loading Integrated Forest Management Logging Road Construction or Maintenance Mechanized Tree Falling Log Processing Marine Log Salvage  Average Injury Rate (non-HCO injuries/100 person years) 33 22 18 14 13 11 10 9 6 6 5 4 3 2 2  1.7. Mandatory Faller Certification as a Safety Intervention The BCFTS is a safety intervention that has three hierarchical components to ensure that fallers employed in the BC Forestry sector have the abilities to work using industry best practices. At the individual level, the first component is educating new faller trainees with safe work practices through structured training and apprenticeship. Across the tree falling trade, the next component involves endorsement via standardized examination and certification to ensure that manual tree fallers can meet standardized criteria for industry best practices. At the forestry industry level, the final component is the requirement for industry wide adherence to regulations that require both the training and certification components. 12  Training is a means to teach the components of a safety intervention and, in some trades, is followed-up by apprenticeship to reinforce the knowledge gained in a supervised environment. Similar to the new faller training process, training that involves behavioral modification has been shown to be among the most effective means to influence safe work practices (19-21). Forestry workers who have been trained in safe work practices have been shown to demonstrate better adherence to occupational health and safety regulations than those who have not received such training (22). Improved adherence to safety regulations might improve health outcomes, but such outcomes may be dependent on the variability of the environment in which work is conducted and the amount of control regulations have to reduce exposure to hazards. Forestry work, especially manual tree falling, occurs in a highly variable environment with many natural hazards that may not be mitigated via strict adherence to safety regulations alone (23). Safe work practice training that teaches techniques to minimize the risks associated with hazardous situations has been shown to positively influence the actions of loggers when faced with situations that do not have specific safety regulations prescribed to them (24). Yet, evidence indicating the effects of safe work practice training with respect to health outcomes varies across occupations, as indicated by a decrease in time loss injuries for health care workers (25) while there were no change in injury rates for loggers (26,27). Although training has been shown to improve demonstrated safe work practices, the prior-mentioned contradictory health outcomes associated with training suggest that more than just safe work practices are required to improve the health and safety of workers (23). In the case of the BCFTS, new faller trainees are required to complete training and a 6 month apprenticeship. This combination of standardized training and supervised apprenticeship is suggested to contribute to better health outcomes for workers that perform manual labour when compared to workers that have not had both modalities of training combined (28). 13  Certification in general involves demonstrating a minimum level of competence. Methods to obtain a certified status range from less demanding requirements such as writing an open book exam to more demanding demonstrations of capabilities that involve performing practical skills in a dynamic environment. There is a distinction between methods of certification; certification by demonstrating competence alone and certification by demonstrating competence in combination with prerequisite training. Studies focused on logging safety and task specific certification in combination with prerequisite training have shown no effect on injury rates at an ecologic level (26,27). Likewise, the requirement for youth aged less than 16 years old to be trained and certified in the operation of tractors on public roads was shown to have no change on the rate of tractor crashes in the agricultural industry (29). Among the existing research on certification without prerequisite training there is conflicting evidence for the effects that certification has on health and safety outcomes. In the occupational health and safety realm a commercial driver licensing program was found to have no effect on traffic accidents (30). Yet, broader public health certification programs beyond work environments have shown some benefits. These types of certification programs also aim to reduce the incidence of injury and can be considered indicative of the health outcomes to be expected from requiring task specific certification. For example, in the United States mandatory certification for motorcycle operators requires demonstrating task specific competence and this has been associated with a decrease in crash related mortality rates (31). Regulation provides a means to state and enforce rules that govern workplace safety. The BCFTS applies to all fallers in BC and holds employers accountable to ensuring that they only employ fallers that have met the criteria to be deemed a competent faller. Ever since it has been a requirement in regulation to hold a BCFTS certificate to work as a manual tree faller in BC, it is illegal to work as an un-certified faller. Employers found to be out of compliance with 14  WorkSafeBC regulations that stipulate faller training and certification are cited or penalized with fines. Safety interventions similar to the BCFTS that are mandated through regulation and thus enforceable have shown to be an effective means to improve workplace safety (32). There is also strong evidence indicating that citations or penalties for being out of compliance with specific occupational safety regulations are associated with a reduction in injuries (33). 1.8. Study Rationale No scientific evaluations have been conducted on the effect of the BCFTS intervention on the health and safety of manual tree fallers. The need for an assessment of the BCFTS has not gone unnoticed and the Auditor General of BC has identified a need to assess the effect of the BCFTS on manual tree faller injury rates in order to determine whether the program has been achieving its mandate (34). The results of this thesis will help determine if the BCFTS reduced injuries among manual tree fallers and may guide the implementation of similar programs in other jurisdictions and hazardous industries. Research involving occupational health and safety interventions is particularly susceptible to uncontrolled sources of bias (35). Contributing to such biases is that this type of research is often conducted retrospectively and assumptions must be made with inherent biases to control confounders. The suggested method to overcome these biases is to prospectively design safety intervention evaluations (36). No prospective plans were established by WorkSafeBC to evaluate the BCFTS. Nevertheless, a retrospective study on the BCFTS can assist with the design of future prospective studies by indicating confounders that can possibly be mitigated. An evaluation of the BCFTS can be based on workers‟ compensation claims since the BCFTS and associated regulations were created for the purpose of preventing workplace injuries. At an ecologic level, claim rates during the time periods before and after the requirement of mandatory 15  certification can be compared for targeted CU‟s within the province of British Columbia. At the individual level, the BC Forest Safety Council maintains a registry of fallers in the province by certification date providing an opportunity to evaluate the effect of certification status on injury risk among a cohort of fallers. Support for conducting a quantitative assessment of the BCFTS is based on the standardized introduction of the intervention and a sound theoretical rationale for injury reduction; safe work practice training with supervised apprenticeship and changes in the social work environment (requirement of a certificate to work). Haddon‟s Matrix (37) provides a framework to characterize the multiple factors contributing to the chronologic events of a workplace injury and has been suggested as one method to identify where to target safety interventions in the forest industry (38) (see Table 1.3). An advantage of using Haddon‟s multifactorial approach to characterizing workplace injury is that it can assist with identifying variables that contribute to workplace accidents with respect to chronology before, during or after incidents (38). The benefit of identifying categorical and continuous variables to include in retrospective evaluation is that they provide a means to account for their influence upon the key outcome measures of a safety intervention. For example, evidence suggests that having a previous claim predisposes workers for future injuries (25) and it has also been shown that during periods of high unemployment there is a decrease in injury reporting when workers are concerned about security of employment (leading to artificially lower injury rates) (39,40).  16  Table 1.3: Example of Haddon's Matrix as applied to manual falling injuries. Physical Environment  Social Environment  Phase  Host (Worker)  Agent  Pre-event (primary prevention – stop the injury event)  1. Faller training/ certification, faller experience, previous injury, age, work experience, certificate level, occupation type  2*. Type and maintenance of faller equipment (e.g. chainsaws, wedges)  Event (secondary prevention – reduce the injury severity)  5*. Worker physical fitness, fatigue, choice of felling technique, placement of cuts  6*. Use of 3*. Terrain, slope, 8*. Worker, personal protective tree characteristics, supervisor and equipment tree species, employer weather conditions adherence to safety policy  3. Production 4. Safety policy volume, geographic (mandatory region certification), unemployment rate  9*. Injured 10*. Automatic 11*. Availability of Post-event* workers general shutoffs on and timely (tertiary health status chainsaws emergency prevention – response (radio treatment and communication in rehabilitation to remote regions) reduce disability) * shaded areas indicated variables not included or investigated in the current study  12*. Compensation benefits for care, rehabilitation and wage-loss  The known dates for faller certification, existing health outcome data, and available information for covariates identified using Haddon‟s Matrix present a basis for retrospective study on mandatory faller certification to indicate whether this safety intervention has had an effect at both the industry wide and individual level. Variables are available to indicate worker, physical environment and social environment characteristics supporting a study focused on prevention at the pre-event phase. However, data is not available to provide indicators of agents at the preevent phase or any variables at the event and post-event phases. The BCFTS does not require all fallers to complete standardized training since experienced fallers are grandfathered into certified status upon successful completion of a challenge exam. Grandfathering into a certified status may not be as effective for a safety intervention compared to the full certification program. Nevertheless, this study is primarily an evaluation of 17  grandfathering for the existing faller workforce as there were very few new fallers during the study period. As a result, the analyses covered in the ecologic and cohort studies are largely of the effect of experienced fallers (>2 years of work experience) challenging the certification exam, and not the effect of the full certification process. Regardless, the effect of demonstrating trade specific competencies (knowledge and skills) among fallers grandfathered through the BCFTS Course Challenge is worthy of evaluation. At the ecologic level, the requirement for proof of tree falling competence to work as a manual tree faller in BC has (in theory) excluded unskilled tree fallers from this line of work and this may have improved health and safety at an industry wide level. At the individual level, the introduction of an industry wide standard may have improved work practices and reduced the risk of injury for fallers. 1.9. Objectives and Research Questions The objective of this thesis was to evaluate the effect of the BCFTS on the risk of injury among manual tree fallers in British Columbia. The specific research questions were as follows: 1. Using aggregate provincial claim data, did the annual claim rate for manual tree fallers in British Columbia decrease after the introduction of optional certification in 2003 and mandatory certification in 2006? 2. Using an individual-level model of risk, did the risk of an injury claim decrease during the year post certification compared to the year pre certification among the cohort of registered manual tree fallers in British Columbia? Did the risk change by type of injury (any injury versus acute injuries versus serious injuries)?  18  2.  METHODS  Study 1 was a descriptive analysis of the annual claim rate among manual tree fallers in British Columbia from 1997 to 2008, a period that included the introduction of the faller certification program. Study 2 was a cohort study using survival analysis to estimate and compare the risk of injury pre and post certification among manual tree fallers in British Columbia. 2.1. Study 1: Comparison of Workers’ Compensation Claim Rates for Manual Tree Fallers. Annual injury rates were calculated as accepted workers‟ compensation claims per 10,000,000m3 of forest harvest volume for manual tree faller claims in the forestry subsector of British Columbia. Typically injury rates are calculated using Full Time Equivalence (FTE) to indicate the number of injuries per 100 full time equivalent workers. However, FTE is based on the provincial average wage in a WorkSafeBC Classification Unit (CU) and does not allow for comparisons within sub-groups defined by geographic region, for example. The goal of this study was to determine injury rates specific to manual tree falling and to be able to compare the rates by forestry region within the province of BC. Regional forest harvest volumes were utilized as a denominator since they are indicative of work performed by manual tree fallers and were available by forest region. The numerator was defined as accepted WorkSafeBC injury claims (to be called “claims” herein) for short term disability (STD), long term disability (LTD) and fatalities (FTL) among manual tree fallers with an injury date between 1997 and 2008. Claims due to occupational diseases such as red cedar asthma or non-tree felling incidents like motor vehicle accidents were excluded since they were not directly targeted for prevention by the faller certification program.  19  The WorkSafeBC CU for Manual Tree Falling and Bucking includes fallers engaged in tree falling activities as well as occupations engaged in the manual tree falling industry, but not specifically performing tree felling activities (e.g. support/administrative staff). As well, many manual tree fallers engaged in logging are captured in other industrial CUs used by WorkSafeBC (e.g. Integrated Forest Management). In order to „refine‟ the numerator (i.e. to those specifically targeted by the forestry faller certification program) the combination of the Statistics Canada Standard Occupational Classification (SOC) code for “Chainsaw and Skidder operators” (I161) with the WorkSafeBC occupation code for “Fallers, Fellers – Logging” (0003) was used to identify manual tree fallers engaged in logging across the forestry subsector and three additional industry CUs (Log Hauling, Log Towing and Forest Management Services) (41). 2.1.1. Data Sources Numerator data for claims with STD, LTD and FTL injury outcome indicators were obtained from the WorkSafeBC data warehouse. The claim records included data on the operating location of the employer associated with an injury and this was used to aggregate claims to forest regions. Denominator data was comprised of annual forest harvest volumes obtained from the Ministry of Forests and Range Harvest Billing System (HBS) (42). The HBS is an Internet based service where forestry companies report the volume of wood they harvest. The forest harvest reports indicate the volume of “scaled” wood, which was an estimate of the volume of wood extracted and transported out of a logged area. 2.1.2. Study Variables 2.1.2.1. Injury Type Claims were categorized as an acute, strain or serious injury. This was done to investigate if the impact of the forestry certification program on the claim rate may have differed by type of 20  injury. Additionally, claims were excluded for injuries that were not specific to manual tree falling related activities or the BCFTS or that were attributed to diseases with latency periods longer than the study period. The acute and strain injuries were identified using the nature of injury (NOI) code on the workers‟ compensation claim. A NOI classifies an injury based on its principle physical characteristics, for example, amputations, fractures, cuts or lacerations (43). The list of NOIs used to define acute injuries and strain injuries for the purposes of this study are provided in Table 2.1. Given the small size of the manual tree faller population, fatality rates were not determined since there were insufficient frequencies of this type of claim to determine reliable rate estimates. Table 2.1: Nature of Injury (NOI) claim criteria for acute and strain injury types. Acute Injuries Abrasion Amputation Concussion Contusion Dislocation Exposure to cold Exposure to heat Fractures Heat burns Hernia Laceration Multiple injuries Other injuries  Strain Injuries Back strain17 Bursitis & related Carpal tunnel syndrome Other strains Tendinitis, tenosynovitis  In order to determine if trends for injuries with the greatest severity had changed over time, the rate for serious injuries was also determined. The purpose of identifying serious injuries was to attain an indicator of claims that resulted in the most disability (indicated by time loss and health care costs). Serious injuries were identified from claims that (1) met the acute and strain injury criteria (see Table 2.1), and (2) by using the WorkSafeBC serious injury indicator included in the 17  Back strain was classified as a strain injury rather than an acute injury because there is evidence suggesting that back strains are the outcome of the culmination of multiple previous strain events (63). However, back strains can also occur due to a one time (acute) injury event.  21  claims data. WorkSafeBC defines serious injuries as claims that meet one or more of the following four criteria (17): 1) Long duration (28 or more days STD days lost in a period); or 2) High healthcare costs (costs greater than or equal to 28 days of wage loss); or 3) Fatality; or 4) One of 275 ICD9 medical diagnoses falling under: a) Infectious and parasitic diseases b) Neoplasms c) Mental Disorders d) Diseases of the Nervous System & Sense Organs e) Diseases of the Circulatory System f) Diseases of the Respiratory System g) Diseases of the Digestive System h) Diseases of the Musculoskeletal System and Connective Tissue i) Symptoms, Signs, and Ill-defined Conditions j) Injury and Poisoning Claims that were classified with ICD9 medical diagnoses falling under infectious and parasitic diseases, neoplasms, mental disorders, diseases of the nervous system and sense organs, diseases of the circulatory system, diseases of the respiratory system or diseases of the digestive system were not included since they did not meet the acute or strain injury criteria. Only claims that were relevant to the physical tasks required when working in a cut block felling trees and were likely targeted for prevention by faller certification were included for analyses. Thus, the serious injury rate criteria for this thesis was different from that used by WorkSafeBC described above. Each of the claims was also limited to those with specific sources of injury (SOIs) and accident types (ACCs) relevant to the tasks required when working in a cut block felling trees and were likely targeted for prevention by faller certification (see Table 2.2).  22  Table 2.2: Source of Injury (SOI) and Accident Type (ACC) claim criteria for inclusion as acute, strain or serious injuries. Source of Injury (SOI) Bodily motion Clothing, textiles Environmental cold Flame, fire, smoke & other heat Hand tools Hoisting apparatus Ladders Logs, tree products Machines Metal items Mineral items Miscellaneous Power tools Trees, plants Working surfaces  Accident Type (ACC) Caught in Exposure to heat, cold Fall from elevation Fall on same level Matter in eye Other accidents Other bodily motion Overexertion Repetitive motion Rubbed, abraded Struck against Struck by  2.1.2.2. Claim Region Each claim was geographically assigned to the Coast, Northern Interior or Southern Interior Forest Region based on the employer operating location recorded on the injury claim (Appendix A). Each operating location (recorded by geographic sub-location, eg. community, harbor or valley) was assigned a postal code to indicate the geographic location in BC. Using the postal code associated with each claim, the geospatial information software (GIS) ArcGIS (44) aggregated claims into forest regions using the 2003 Ministry of Forest and Range boundary parameters (11). 2.1.2.3. Wood Harvest Volume Customized harvest volume reports were obtained from the Ministry of Forests and Range harvest billing system using the parameters outlined in Table 2.3. These reports were obtained for each month of the study period. Due to potentially long lags between the time when trees were cut down and scaled for wood volume, monthly harvest volumes were aggregated into annual volumes to reduce the misclassification inherent for shorter time periods. The only type 23  of wood harvest end use product excluded from the harvest volumes was Christmas trees since the narrow diameter of the trunk does not require a faller to have the BCFTS certificate to cut down. The wood harvest volumes obtained were not exclusively representative of the amount of wood cut down by manual tree fallers since it also included the amount of wood cut down by mechanized harvesting equipment. Regionally, the proportion of wood harvested by manual tree fallers and mechanized equipment varies, however, no data was available to indicate these proportions. Table 2.3: Ministry of Forests and Range harvest billing system wood volume report configuration parameters. Report parameter Date Range Region Harvested Region Scaled Client Location File Type Species Product Grade Billing Type  Selection From: 1997-01-01 To: 2008-12-31, by year Coast, Northern Interior or Southern Interior All regions All locations All land types All tree species All except Christmas trees All grades All types  2.1.3. Analysis Annual acute, strain and serious injury rates for included claims were calculated per 10,000,000m3 of forest harvest volume. The value of 10,000,000m3 represents approximately half the average annual harvest volume in the Coast Forest Region (1997-2008) as indicated in Figure 3.1. Descriptive analyses were conducted on the annual injury rates for the years 19972008 with a particular focus on changes in the rates starting in 2003 with the introduction of the optional faller safety certification and 2006 with the mandatory faller certification program. All data analyses were conducted using the statistics software Stata 10 SE (45).  24  2.1.3.1. Data Cleaning The initial phase of claims data cleaning was focused on determining to which forest region a claim was attributed and was based on a total number of 3520 claims. Claims were matched with their representative employer operating location using employer identifier numbers and employer operating location numbers assigned by WorkSafeBC. Claims without employer identifying numbers (6.3%) were geographically ambiguous and were excluded from the descriptive analysis of claim rates. Following this exclusion from the cohort, 3298 claims remained in the dataset. The final phase of claims data cleaning involved excluding claims from the dataset that were not attributed to the tasks required to manually fall trees and were not considered relevant to the safe work practice objectives of the BCFTS. Claims not meeting the injury type criteria in 2.1.2.1 were excluded from the dataset based on specific NOI, ACC and SOI (Appendix B). Claims were also excluded from the dataset if they were due to occupational diseases such as hearing loss or occupational asthma since many occupational diseases have latency periods that exceed the study time period. Consequently 198 claims were excluded from the dataset with a NOI, ACC or SOI that met the exclusion criteria and 3100 claims remained in the final dataset for descriptive analyses. 2.1.3.2. Descriptive Statistics Three types of annual injury rates were calculated for each forest region using the frequency of claims in the numerator and harvest volume in the denominator. Annual frequencies of acute, strain or serious injuries for each forest region were derived from the final dataset. Each numerator was paired with the annual harvest volume for the corresponding forest region as a denominator. To obtain the key outcome measures of annual injuries per 10,000,000m3 wood the following formulas were used: 25  Each injury rate was plotted as a times series to see if there were changes in rates with respect to the implementation of the optional certification period (2004) and mandatory certification period (2006). Study 2: Cohort Study of Risk of Injury Associated With Faller Certification Status The purpose of Study 2 was to investigate the individual-level risk of injury among a cohort of BC fallers before and after BCFTS certification using a cohort study with survival analysis. 2.2.1 Data Sources All BCFTS certified fallers were identified using the BC Forest Safety Council faller registry. This database contained information about registered fallers such as name, address, age, certification date, work experience and information indicating the type of falling work typically performed. Fallers in the registry were linked to WorkSafeBC data for STD, LTD and FTL claims in the WorkSafeBC data warehouse by Population Data BC (PopDataBC) following their established linkage protocols. Personally identifiable information was removed from the linked database provided to researchers and replaced with an anonymous study identifier. 2.2.2 Study Cohort BC Faller Training Standard certified fallers who were listed in the BC Forest Safety Council faller registry and certified between March 1st, 2003 and December 31st, 2008 were considered 26  for inclusion in the study cohort. During this time period 3722 fallers were certified and recorded in the faller registry by their certification date. The majority of these fallers (81%) reported at the time of their certification exam that they were “production” fallers, indicating that their primary work was to fall trees for the purpose of harvesting wood. Among the other 19% of fallers who indicated they were “non-production”, 8% performed falling activities as a “spacer” (4%), “fire suppresser” (2%), “ arborist” (1%), “seismic faller” (1%) or “other” (0.3%). An additional 9% of non-production fallers did not report their type of falling activities. One percent of fallers in the registry were classified as new faller trainees by the BC Forest Safety Council. The cohort excluded fallers who could not be linked to the provincial population directory for identification of workers‟ compensation claim outcomes via Population Data BC data services (n=399 grandfathered fallers, n=5 new faller trainees). The population directory is a centralized register that is primarily based on registration into the provincial health care plan. Thus, it is plausible that out of province workers came to British Columbia to fell trees for short-term contracts and were not covered under the BC Provincial medical health plan at the time of their employment. While some of these individuals may have incurred a work injury covered by WorkSafeBC, it was not possible to capture them in this study due to our data access and linkage procedures governing access to data for research purposes in British Columbia. The cohort also excluded new faller trainees (n=36) as a different at-risk and exposure group. New faller trainees underwent a different type of certification intervention (i.e. full certification with classroom/field training and field mentoring versus challenging the exam) and we had insufficient numbers to analyze them separately. An additional 34 eligible fallers were excluded from the cohort because they were 65 years of age or older at the time of certification and two were excluded for an error in the injury date on their claim records. Fallers aged 65 years or older in the cohort were excluded as it was unknown 27  if these individuals were still working as fallers or if they had obtained a certified status for other reasons. The final cohort of fallers was 3251. 2.2.3 Study Follow-up Period The risk of injury associated with certification was investigated by quarter (three months) one year pre certification date and one year post. The faller registry only records the date of certification, so assumptions were made that all fallers were employed full time in the industry one year prior to certification, a logical assumption as the cohort was limited to fallers with a minimum of two years of experience18 prior19 to challenging the certification exams. It was also assumed that fallers were employed full time one year post certification; also a logical assumption as challenging the exam is in preparation for employment in the occupation in BC. 2.2.4 Study Variables 2.2.4.1 Injury Outcome Variables Dichotomous injury outcome variables were constructed to identify whether a faller was injured while working (yes=; presence of a WorkSafeBC claim during follow-up) or not injured at work (no= no claims during follow-up) during each of the 8 discrete 3 month time periods. Claims were restricted to STD, LTD or fatal claims. WorkSafeBC does not code HCO claims by occupation or injury characteristics and these claims were excluded from the analysis. Only one claim per quarter was counted towards identifying whether a faller was injured while working during each discrete time period (i.e. subsequent claims in a quarter did not contribute to the risk of injury metric). Claims were restricted initially to those where the occupation and industry of  18  Some fallers with less than 2 calendar years of experience were permitted to attempt the faller training standard course challenge because they were deemed to have completed an equivalent amount of work experience in a shorter time period. 19  Previous work experience as a faller that met the grandfathering criteria could have been achieved at any time prior to the certification exam. 28  employment coded at the time of injury was specifically among production fallers. In other words, those targeted by the faller certification program. We also investigated a broader definition of claims as part of an adjusted survival analysis sensitivity analysis. The occupations included in the broader definition were for those that involve falling work but were not coded specifically for a faller. For example, many fallers are independent owner-operators and even though they work full time falling trees as a production faller, WorkSafeBC may code these claims occupationally as “Logging Contractor”. See Table 2.4 for a list of the Statistics Canada SOC (19) and WorkSafeBC occupation codes (20) used to identify faller injuries for both the narrow and broad definitions. Table 2.4: Claim inclusion criteria based on Statistics Canada SOC (46) and WorkSafeBC (41) occupation code combinations for specific/ broad faller claims. Type of Outcome Variable Specific faller claims  Occupation Code Combination Statistics Canada WorkSafeBC Chainsaw & Skidder Operators Chainsaw & Skidder Operators Primary Production Managers (except Agriculture)  Broad faller claims  Silviculture and Forestry Workers Supervisors, Logging and Forestry  Faller, Feller – Logging Bucker – Logging Faller, Feller – Logging Logging Contractors Logging Owner/Operator Forest Fire Fighter Tree Spacer Supervisors – Forestry & Logging Foremen/women – Forestry & Logging  Each of the injury outcome variable classes were further categorized into four sub-types based on the type of injury: non-HCO injuries, acute injuries, strain injuries and serious injuries (see Figure 2.1). The acute injury and strain injury sub-types were subsets of the non-HCO injuries20 and had the same NOI definitions used in Study 1 as shown in Table 2.1. Serious injuries were defined using the WorkSafeBC serious injury criteria (17) described in Study 1 (see 2.1.2.1).  20  All acute and strain injuries are included in the non-HCO injury type.  29  Specific faller claims  All injury claims  Broad faller claims  non-HCO injuries Acute injuries Strain injuries Serious injuries non-HCO injuries Acute injuries Strain injuries Serious injuries  Figure 2.1: Classification hierarchy for injury outcome variables. The same SOI and ACC criteria specified in Study 1 were used to identify manual tree falling related claims. Claims that were not attributed to falling were identified by specific NOI, SOI, and ACC exclusion criteria listed in Appendix C. 2.2.4.2 Fixed Covariates All fixed covariates were assumed to be constant throughout the survival analysis follow-up and based on self-reported information at the time of certification, as described below. 2.2.4.2.1 Age Group Fallers were grouped into age categories using the WorkSafeBC classifications: 15 – 24, 25 – 34, 35 – 44, 45 – 54 and 55 – 64 years. Age was not considered to be a time varying covariate because the survival analysis period spanned two years. A two year change in age was thought to have a negligible effect on safe work practices or propensity for injury.  30  2.2.4.2.2 Work Experience Group Work experience groups were constructed by classifying fallers as having 1 – 2, 3 – 5, 6 – 10, 11 – 20 or 21+ years of falling specific work experience at the time of certification, consistent with previous occupational health studies (47). 2.2.4.2.3 Faller Type Two faller classifications were constructed that grouped fallers as “production” or “notproduction”. The fallers grouped into the “not-production” classification consisted of the spacers, fire suppressors, arborists, seismic fallers, and others. Fallers that did not have their faller type recorded were distinguished by a “missing” label and included in the “not production” classification. The rational for creating the two groups of faller types was to investigate the risk for participants whose primary work task is falling full time (production) versus people who have falling tasks required in their line of work, but it is not the task that is performed all day (notproduction). 2.2.4.2.4 Region The Thompson and Kootenay regions were amalgamated together to form the Southern Interior Region. The outcome of this amalgamation resulted in three regions that fallers stated was their regular place of work: Coast, Northern Interior and Southern Interior. The grouping of fallers into these three regions also helped to investigate the effect of conditions such as terrain, environment, and harvest volumes with reference to data collected by the Ministry of Forests and Range by these forest regions.  31  2.2.4.2.5 Grade Level of Certification Fallers are certified according to their ability to work with different tree diameters (5 levels, 1 – 5) and in different terrain (3 levels, A – C) for a possible 15 different certification grades. The numeric component of the grade refers to the tree diameter where 1 is the smallest (up to 20”) and 5 is the greatest (over 60”). The alphabetic component of the grade refers to the slope of the terrain where A is the least (less than 30%) and C is the steepest (greater than 60%). Due to small numbers of individuals certified to work in grades B and C (e.g. 1C accounted for 0.5% of the cohort), we relied on the tree diameter levels to differentiate fallers by grade level in the regression modeling. 2.2.4.3 Time Varying Covariates Individual previous claims status and industry-level unemployment rate were identified as two time varying variables that could bias the relative risk of injury estimates as outlined below. 2.2.4.3.1 Previous Claims Previous faller claims that met the study definition were identified from the claim database using all available records for two years prior to the start of follow-up. Individuals who did not have previous faller claims prior to the start of follow-up and had an injury claim during follow-up were subsequently coded as having had a prior injury in the next quarter and all successive quarters of follow-up. 2.2.4.3.2 Unemployment Rate To account for the effect that high rates of unemployment may have had upon the estimated risk of injury, the unemployment rate was calculated for each quarter during the study follow-up period and applied to individuals based on their quarter of employment. The quarterly 32  unemployment rates were based on the mean of the monthly provincial unemployment rates for the primary resources sector (forestry, fishing, mining, oil and gas) which were obtained from Statistics Canada Labour force surveys (2002 – 2008) (1). 2.2.5 Analysis To perform a survival analysis the dataset was structured so that each faller had a 1 year period before and after their individual certification date. Relative risk of injury for the cohort of fallers was then modeled using complementary log-log regression. All data analyses were conducted using the statistics software Stata 10 SE (45). 2.2.5.1 Survival Analysis and Censoring Survival analysis was selected as the optimal method to assess the risk of manual tree falling injury given the outcome (injuries) and follow-up time (48). The analysis was based on all fallers sharing a common survival analysis point designated as the day of evaluation for BCFTS certification and spanned a time period of 1 year before this event to 1 year after, resulting in a 2 year study period. The 2 year study periods for each faller were divided into 8 quarterly periods so that the risk of injury was determined for different time periods surrounding the certification event (Figure 2.2).  Figure 2.2: Survival analysis structure, discrete quarters, and individual date of certification. Using each faller‟s individual certification event as a common point in time, the 2 year window of survival analysis was fixed for all fallers by making the last day of the 4th quarter the 33  evaluation date. Individual fixed covariates were constant across all 8 observations for each faller. If a faller had stated at the time of certification that he/she was a “production” faller, he/she was assumed to be a “production” faller in all 8 quarters. Time varying covariates were updated over all 8 time periods. For example, previous claims status changed based on the occurrence of an injury during follow-up and unemployment rates changed based on the unemployment rates present at an actual point in time for an individual faller. Descriptive results were determined by calculating the injury rates of each injury outcome variable for quarters of survival analysis and covariate type. The numerator of the injury rate was the number of injuries during the survival analysis period and the denominator was the amount of time fallers were working full time (i.e. were not censored). To standardize the injury rates for comparison purposes, the injury rates were scaled to the number of injuries per 100 person years21 of tree falling work. The quarterly injury rates were based on a relative time scale and were the outcome of pooling injuries from many different years (2002 – 2008). Since these injury rates are based on a relative time scale no comparisons could be made to the annual injury rates determined in Study 1. Individuals who were off work as indicated by dates on a time-loss workers‟ compensation claim or fatality were censored from the relevant quarters. Starting in December 2007, the BC Forest Safety Council implemented an annual fee to maintain certified faller status. Fallers who indicated to the BCFSC that they were no longer working in the profession or had retired were also censored from this date forward. The discrete survival analysis with censoring resulted in left censoring, gaps and right censoring. Left censoring occurred when an injury prevented  21  Person years were determined by using the number of fallers in a given quarter who were uncensored. Uncensored fallers in a quarter were assumed to be working full time for 3 months and this time quantity provided the basis to determine an injury rate with a person years denominator. 34  inclusion at the start of survival analysis until becoming healthy enough to return to work. Gaps occurred when fallers were included in the survival analysis, but sustained an injury, were censored for at least a quarter and then returned to work before the end of the survival analysis period. Right censoring occurred when fallers either: 1) were injured during the survival analysis and disabled through till the end of survival analysis or 2) were removed from the study due to a fatality, retirement or withdrawal from the profession. A discrete-time survival analysis was utilized to assess the risk of manual tree falling injury in each period of time and risk of injury comparisons could be made between quarters with respect to the certification date. The outcome of an event (injury) was discrete in that you could only be one of two states during each of the eight discrete time periods, injured or not injured (49). This resulted in interval-censored data for which complementary log-log (clog-log) regression is the recommended method for statistical modeling (48). This approach models outcomes using a proportional hazards assumption, providing a relative risk (48). The quarterly relative risk of injury was determined using a multivariate model based on the probability of fallers sustaining an injury in one quarter compared to the probability of sustaining injury in a reference quarter (quarter 1). To determine the intra-covariate relative risk of injury the probability of fallers sustaining an injury in one classification (eg. age 15-24 years) was compared to the probability of fallers sustaining an injury in a reference classification (eg. age 45-54).(48) Post estimation chi2 tests were conducted to determine the significant contrasts between quarterly time periods and covariates that were not specified as reference categories. Further support for using a discrete-time approach in comparison to other survival analysis methods, such as continuous-time survival analysis, is that it is considered simpler to fit time varying covariates to discrete time periods (49,50).  35  2.2.5.2 Multivariable Model Construction and Model Fitting Complementary log-log regression model construction was performed by using time (quarters) as the independent variable and injury outcome type as the dependent variable. Initially a fully specified model was used incorporating all of the covariates stated in Error! Reference source ot found. and Error! Reference source not found.. Fixed covariates were included in the full model based on information provided by fallers at the time of the examination. Age was incorporated in to the full model since evidence indicates that young workers are more likely to be injured and that the severity of serious injury tends to increase as workers age (51-53). Work experience was included since workers with less experience have been shown to be at a higher risk for injury (54). Faller type (production vs. non-production) was included since it was hypothesized that there should be a greater risk of injury for fallers who fall more frequently than those who do not. Region was included since it was hypothesized that environmental conditions (ie. tree size and slope) might affect the difficulty of falling trees and the consequent risk of injury. Certificate grade level was included since it was thought to be indicative of skill level. Since Grade level was dependent on the tree size and slope on which a faller was examined, it was not assumed to be an accurate indicator of skill for very experienced fallers that work with large trees. However, grade level does hold some validity for indicating fallers who may not have a higher skill level since it is unlikely that an unskilled faller would be working in grade level 3 – 5 environments. Pairs of fixed covariates suspect for correlation were age/work experience, region/certificate level, & work experience/certificate level. Each of these pairs of covariates were determined to have correlations that ranged from low to moderate/strong. The work experience/certificate level pairing was shown to have a low correlation of 0.16. Region/certificate level and age/work  36  experience were determined to have correlations of 0.52 and 0.57 respectively, but were included in the full model together. Time varying covariates were incorporated into the full model based on information that was specific to each quarter of follow-up for each individual faller. Previous claims status was included in the full model since it has been shown that workers who have previous claims are more likely to sustain future injuries (25). Primary resource sector unemployment rate was included in the full model since unemployment rate has been shown to be associated with injury rates and could be used as an indicator of current economic conditions (39,40). The regression equation of the fully specified model used was:  37  Where:  = constant,  = time,  = fixed variable, = time varying variable, = x variable,  =  error term. Reference categories were quarter 1, age 45 – 54, non-production faller type, Northern Forest Region, Grade Level 1, no previous claims and 0 – 6% unemployment. Model fit was assessed using the Akaike Information Criterion (AIC) (55) and Bayesian Information Criterion (BIC) (56). Deviance statistics such as the Chi2 distribution were not applicable to assessing model fit since the models in consideration were not nested due to having different sets of predictors (48). As long as models were fit to identical sets of data, AIC‟s and BIC‟s could be used to determine the model of best fit. Using the AIC and BIC fit statistics, the model of best fit was chosen by selecting the one that yielded the smallest AIC or BIC values. The full model AIC and BIC values were compared to reduced models that differed by the removal of one covariate from the full model. Five reduced models were assessed for AIC and BIC values by the separate removal of age, certificate level, region, unemployment rate, and work experience. Reduced models without faller type or previous claims status were not assessed since these covariates were deemed to be vital to the final model. The approach used in this study to model is consistent with those used in epidemiology for testing hypotheses.  38  3.  RESULTS  3.1. Study 1: Comparison of Workers’ Compensation Claim Rates for Manual Tress Fallers 3.1.1. Description of Workers’ Compensation Claims among Fallers in British Columbia During the years 1997 to 2008, a total of 3,100 compensation claims for workplace injuries were recorded for manual tree fallers in British Columbia (Table 3.1). Among these claims, acute injuries were the most frequent (56.6%) compared to strain injuries (43.4%). Of these 3,100 claims, half met the serious injury definition (50.1%). The distribution of acute and strain injuries among the serious injuries was similar to the overall distribution of injuries in the study population at 57.5% and 42.3%. The distribution of acute, strain and serious claims was similar across the three forest regions of the province (Table 3.1). Table 3.1: Distribution of acute and strain claims by forest region for the years 1997-2008. Forest Region  Acute Injuries  Strain Injuries  Serious Injuries22  Frequency Percent Frequency Percent Frequency Percent (n) (%) (n) (%) (n) (%)  Total Injuries23 Frequency (n)  Coast  1210  55.0  992  45.0  1076  48.9  2202  Southern Interior  385  61.1  245  38.9  323  51.3  630  Northern Interior  159  59.3  109  40.7  155  57.8  268 3100  Very few women work as manual tree fallers with only 4 claims (0.1%) recorded for women during this study period (data not shown). The age range of injured workers was 18 to 77 years. 22  Serious injuries are not mutually exclusive and can include both acute or strain injuries. All injuries included in the cohort are either acute or strain and of these injuries some meet the serious injury criteria. 23 The total number of injuries is all non-HCO injuries that met the study inclusion criteria and is the sum of acute and strain injuries.  39  The distribution of claim types by age group were different for acute, strain and serious injuries (Table 3.2). Acute injuries were the most prevalent (74.4%) in the 15 – 24 year old age group. The fewest acute injuries were in the 65+ year old age group (44.4% of all claims). For serious injuries, young workers in the 15-24 year old age group had the smallest proportion (32.2%) of this type of injury and the 65+ year old age group had the greatest proportion of 88.9%. Both the 15 – 24 and 65+ year old age groups represented the two smallest proportions of injuries out of all the age groups with the 15 – 24 year olds having 2.9% and the 65+ year olds having 0.3% of all claims, thus these rates may be susceptible to high variability. Table 3.2: Distribution of claims by age group and injury type for the years 1997-2008. Acute Injuries Strain Injuries Serious Injuries Total Injuries Age Group Frequency Percent Frequency Percent Frequency Percent Frequency (n) (%) (n) (%) (n) (%) (n) 67 74.4 23 25.6 29 32.2 90 15 – 24 415 61.0 265 39.0 300 44.1 680 25 – 34 618 55.4 498 44.6 558 50.0 1116 35 – 44 467 52.8 417 47.2 486 55.0 884 45 – 54 183 57.0 138 43.0 173 53.9 321 55 – 64 4 44.4 5 55.6 8 88.9 9 65 + 1754 56.6 1346 43.5 1554 50.1 3100 All  In terms of the distribution of injury claims within the WorkSafeBC forestry subsector (i.e. workers‟ compensation classification units), the majority of all claims were incurred by manual tree fallers in the Manual Tree Falling and Bucking CU (57.1%) and the Integrated Forest Management CU (32.7%) (Appendix D). Approximately 10% of all claims were incurred by manual tree fallers in other Forestry subsector CU‟s. 3.1.2. Description of Regional Harvest Volumes (Denominator Variable) The Southern and Northern Interior Forest Regions had the greatest annual average forest harvest volumes over the study period with average yearly harvests of 2.57x107m3/year and 2.53x107m3/year respectively or approximately 35.9% and 34.9% of the provincial harvest (see 40  Figure 3.1), and the Coast Forest Region the least with an annual average harvest of 2.12x107m3/year or 29.2% of the provincial harvest. Peak forest product volumes were harvested in 2005 for the Southern and Northern Forest Regions, and in 2004 for the Coast Forest Region. Leading up to 2005, the trends in annual harvest volumes for each region fluctuated whereas from 2005 till the end of the study period all forest regions had a continuous decrease in the annual harvest volumes (Figure 3.1).  Annual Forest Harvest Volumes of British Columbia by Forest Region  Harvesrt Volume (m3)  3.50E+07 3.00E+07 2.50E+07 2.00E+07 1.50E+07 1.00E+07 1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  Year Coast  Northern Interior  Southern Interior  Figure 3.1: Annual forest harvest volumes for the Coast, Northern Interior and Southern Interior forest regions of BC for the years 1997 – 2008 (42). 3.1.3. Regional Injury Rates Throughout the entire study period the Coast Forest Region had significantly higher annual injury rates compared to the Southern and Northern Interior Forest Regions (see Table 3.3), with as much as a five-fold and ten-fold difference respectively. Some of this difference may be attributable to issues related to the denominator that is a wood volume attributed to manually and mechanically cut down trees (see 4.5.2). Overall, each of the three forest regions had lower 41  annual injury rates in 2008 compared to the injury rates at the start of the study period in 1997. It must be noted that the acute injury rate of the Coast Forest Region did not exhibit a steady decline since 2003, rather, there were years such as 2005 and 2007 where the injury rate increased in comparison to the previous year. Generally, the acute injury rate was higher than the strain injury rate, and the serious injury similar or higher than the strain injury rate for the Coast and Southern Interior Forest Regions, with a few exceptions in a couple of years. In the Northern Interior Region, the same pattern was observed although the serious injury rate exceeded the acute injury rate in several years. Table 3.3: Annual injury rates of the Coast, Southern Interior and Northern Interior Forest Regions indicating acute, strain and serious injury types. Injury Rate (injuries/10,000,000m3) Year Coast Southern Interior Northern Interior Acute Strain Serious Acute Strain Serious Acute Strain Serious No Faller Certification in Province 55.1 53.8 25.5 12.1 16.7 9.7 5.9 7.6 1997 83.5 54.7 54.7 17.9 10.9 15.3 8.2 5.4 6.7 1998 61.3 48.1 38.1 38.5 14.8 10.5 11.3 7.4 4.8 6.7 1999 40.7 45.8 19.2 8.8 13.1 4.6 5.8 7.3 2000 41.5 44.1 46.3 10.7 7.8 9.0 6.1 4.9 7.7 2001 50.7 32.1 38.4 14.6 11.9 14.6 4.8 4.4 7.0 2002 47.2 Start of Optional Faller Certification in the Province 42.8 35.2 12.6 6.1 11.4 6.0 1.3 5.3 2003 37.3 21.9 27.4 8.6 6.2 9.6 2.9 4.3 2.2 2004 33.2 35.1 28.4 38.0 8.7 6.1 9.6 1.9 2.5 3.1 2005 Start of Mandatory Faller Certification in the Province 30.3 28.5 5.9 3.9 4.2 2.6 1.0 1.3 2006 33.5 28.4 37.1 5.5 4.5 4.5 3.0 0.8 3.0 2007 37.1 25.8 35.2 4.5 4.9 4.2 3.8 0.5 1.4 2008 25.8  Although there were decreases in injury rates prior to the introduction of the BCFTS, these decreases appear to be consistent with the overall provincial injury rate trend during the same follow-up period. This is in keeping with a steady drop in provincial workers‟ compensation claim rates (17). The descriptive nature of the claim rates did not support the ability to  42  independently assess the effect of faller certification or any other influences on a declining injury rate. 3.2. Study 2: Cohort Study of Risk of Injury Associated With Faller Certification Status 3.2.1. Demographic and Occupational Cohort Characteristics Among the 3,251 manual tree fallers included in the study cohort, the age at time of certification (i.e. time of registration with the BC Forest Safety Council Registry) ranged from 19 to 64 years of age. Age at the time of certification was normally distributed with a mean of 42.2 years (sd = 9.7). Fallers aged 35 to 44 years and 45 to 54 years of age represented the majority of the workforce at the time of certification at 33.1% and 32.2% respectively (Table 3.4). Years of tree falling work experience at the time of certification ranged from one to 50 years. Workers with 11 to 20 years of work experience (25.9%) and those with 21+ years of work experience (22.2%) represented the majority of the workforce at the time of certification (see Table 3.4). A total of 8.0% of the cohort was missing information on work experience. Workers in the Coast Forest Region represented 44.9% of the cohort followed by workers in the Southern Interior Forest Region at 35.0% and the Northern Forest Region at 20.0%. Based on occupation reported at the time of certification the greatest proportion of the cohort was working as production fallers (83.8%). Non-production fallers represented 7.9% of the cohort and within this category reported occupations of spacers (47.3%), forest fire fighters (22.7%), arborists (16.4%), seismic line cutters (10.9%) and other fallers (2.8%). 3.2.2. Cohort Certification Characteristics During the optional certification period defined by the dates March 1st, 2003 to July 31st 2006, 91.5% of registered fallers obtained their faller certification. More specifically, 2.2% of the 43  registered fallers were certified in 2003, 32.4% in 2004, 41.5% in 2005 and 14.3% were certified in 2006 before the July 31st mandatory deadline. The remaining 9.6% of registered fallers obtained their certified status between the start of the mandatory certification period (August 1st, 2006) and the study end date (December 31st, 2008). The overall number of fallers who attempted the certification process is unknown since the faller registry only contains certified fallers and does not include fallers who were unable to meet the BCFTS competency criteria. Most fallers included in the cohort (44.0%) were certified in tree falling conditions at a Grade Level 1 (see Table 3.4). The proportion of fallers certified at more difficult grade levels decreased steadily to 3.4% at Grade Level 5.  44  Table 3.4: Demographic, occupational and certification characteristics of the cohort. Frequency Percent (n) (%) Total Manual Tree Fallers  3,251  100.0  Age Group (years)  15 - 24 25 – 34 35 – 44 45 – 54 55 – 64  101 669 1,075 1,048 358  3.1 20.6 33.1 32.2 11.0  Forest region where working  Coast Southern Interior Northern Interior Missing  1461 1138 651 1  44.9 35.0 20.0 0.1  Tree falling occupation type  Production Non- Production24 Missing  2725 256 270  83.8 7.9 8.3  224 563 641 841 721 261  6.9 17.3 19.7 25.9 22.2 8.0  Tree falling work experience group 1 – 2 years 3 – 5 years 6 – 10 years 11 – 20 years 21+ years missing Certificate grade level  1 2 3 4 5 missing  1431 891 404 413 109 3  44.0 27.4 12.4 12.7 3.4 0.09  Year certified  2003 2004 2005 200625 2007 2008  71 1054 1349 571 153 53  2.2 32.4 41.5 17.6 4.7 1.6  Certification period when certified  Optional Mandatory  2939 312  90.4 9.6  24  Non-production fallers in the final cohort consisted of 121 Spacers, 58 Forest fire fighters, 42 Arborists, 28 Seismic line cutters and 7 “other” fallers. 25 In 2006 465 fallers were certified before the July 31st optional certification deadline and 106 fallers were certified after the deadline when it was mandatory to be BCFTS certified to work as a faller in BC.  45  3.2.3. Injury Rates Among the Faller Cohort The average injury rate for all included claims across the 8 follow-up quarters during the study was 5.0 injuries per 100 person years of follow-up. By injury type, the average acute injury rate was slightly higher (2.9 injuries per 100 person years) than the average serious injury rate (2.4 injuries) and the average strain injury rate (2.1 injuries) (Table 3.5). It was hypothesized that the Faller Certification Standard would have the biggest impact on acute injuries given the focus of the certification and results presented below focus on the acute injury rate, although results for the other injury outcomes are included in tables and differences in results compared to acute injuries noted below. No distinct injury rate trends were observed across Quarters 1 through 8 over the follow-up period, although the lowest injury rate by quarter was recorded during Quarter 4 (1.6 acute injuries per 100 person years) representing the period immediately before certification. The highest injury rate by quarter was recorded during Quarter 6 (4.4 acute injuries per 100 person years) representing 4 to 7 months after certification. Among the age groups, fallers aged 15 – 24 years had the greatest injury rate at 6.0 acute injuries per 100 person years. The rate ranged from 2.4 to 3.0 acute injuries per 100 person years for the older age groups. Conversely, the higher rate of serious injuries occurred among the oldest age group (Table 3.5). Production fallers had a greater injury rate than non-production fallers with rates of 3.0 acute injuries per 100 person years and 1.2 acute injuries per 100 person years respectively. Fallers working in the Coast Forest Region had a high rate of acute injuries (5.0 per 100 person years) compared to those working in the Interior regions (1.4 to 0.5 per 100 person years). The rate of injury across work experience groups fluctuated between 2 to 3 acute injuries per 100 person years with no discernable trend for more or less work experience. Fallers with a 46  grade level 1 certificate had the lowest injury rate at 1.1 acute injuries per 100 person years, whereas, fallers with a grade 4 certificate level had the highest injury rate at 6.0 acute injuries per 100 person years. There was a slight drop in the acute injury rate for fallers with grade level 5 certificates to 4.6 acute injuries per 100 person years. Findings for occupation type, forest region, work experience and grade level were consistent between acute, strain and serious injury types. There were no distinct trends indicated with the acute injury rate as unemployment rates increased, however, the acute injury rate which ranged from a low of 1.9 acute injuries per 100 person years when the unemployment rate was 0 – 6% to a high of 4.6 acute injuries per 100 person years when the unemployment rate was 12 – 32%.  47  Table 3.5: Specific manual tree falling claim injury rates. Study26 Injury Rate (injuries/100 person years)  Acute Injury Rate (injuries/100 person years)  Strain Injury Rate (injuries/100 person years)  Serious Injury Rate (injuries/100 person years)  Average  5.0  2.9  2.1  2.4  Age group 15-24 (n=101) Age group 25-34 (n=669) Age group 35-44 (n=1075) Age group 45-54 (n=1048) Age group 55-64 (n=358)  8.9 4.9 5.2 4.4 4.8  6.0 2.9 3.0 2.4 2.6  2.9 2.0 2.2 2.0 2.2  2.0 1.7 2.0 2.4 2.9  5.4  3.0  2.4  2.6  1.2  1.2  0.0  0.2  8.8  5.0  3.8  4.2  2.3  1.4  0.9  1.0  0.8  0.5  0.3  0.6  6.0  3.1  2.9  2.2  3.4  2.2  1.4  1.3  5.0  3.3  1.7  2.4  5.3  2.6  2.7  2.0  5.4  3.0  2.4  3.5  1.8  1.1  0.7  1.0  4.2  3.0  1.2  2.1  8.9  5.2  3.7  4.4  12.2  6.0  6.2  5.6  11.5  4.6  6.9  3.2  Fixed variables  Production faller type (n=2725) Non-production faller type (n=256) Coast Region (n=1461) Southern Interior Region (n=1138) Northern Interior Region (n=651) Work experience 1-2 years (n=224) Work experience 3-5 years (n=563) Work experience 6-10 years (n=641) Work experience 11-20 years (n=841) Work experience 21+ years (n=721) Certificate grade level 1 (n=1431) Certificate grade level 2 (n=891) Certificate grade level 3 (n=404) Certificate grade level 4 (n=413) Certificate grade level 5 (n=109)  26  All included claims that met the study inclusion criteria. Acute, strain and serious injury claims were identified from these claims.  48  Table 3.5: Continued  Time varying variables Average  Study23 Injury Rate (injuries/100 person years exposure) 5.0  Acute Injury Rate (injuries/100 person years exposure) 2.9  Strain Injury Rate (injuries/100 person years exposure) 2.1  Serious Injury Rate (injuries/100 person years exposure) 2.4  Unemployment 0-6% Unemployment 6-9% Unemployment 9-12% Unemployment 12-32%  3.9 4.8 5.2 7.2  1.9 3.0 2.7 4.6  2.0 1.8 2.5 2.6  1.8 2.6 2.5 2.6  One Year Before Certification 4.3 2.6 1.7 1.7 Quarter 1 of Follow-up 4.3 3.0 1.3 2.1 Quarter 2 of Follow-up 4.7 3.0 1.7 1.6 Quarter 3 of Follow-up 2.7 1.6 1.1 0.9 Quarter 4 of Follow-up One Year After Certification 6.0 3.1 2.9 2.7 Quarter 5 of Follow-up 6.6 4.4 2.2 4.0 Quarter 6 of Follow-up 6.9 2.9 4.0 3.5 Quarter 7 of Follow-up 4.7 2.9 1.8 2.8 Quarter 8 of Follow-up NTotal=3251 a The number of fallers is not shown for time varying variables since the injury rates are based on time in a given state  3.2.4. Multivariable Model Fitting The full model including the covariates age group, certificate grade level, faller type, previous claims status, region, work experience, and unemployment rate was determined to have the lowest AIC value (2802.009) in comparison to the reduced models (Appendix H). However, the reduced model without the age was determined to have the lowest BIC value (2989.274) (Appendix H). The outcome of the reduced model without the age covariate suggests that this was a preferred model since BIC penalizes for degrees of freedom. However, age was deemed to be an important confounder to control for given the outcome measure of this study. Thus, the full model was specified as the final model to determine relative risk since it had the lowest AIC.  49  The constants and co-efficients determined for each of the final models used in the multivariable regression, modeling based on the fully specified model shown in Error! Reference source not ound. are listed in Appendix I. 3.2.5. Adjusted Survival Analysis Results Two changes in the RR of injury were evident in the results by quarter of follow-up defined by pre and post certification status. First, in Quarter 4, the period immediately preceding a faller‟s certification date, there was a non-statistically significant decrease in the RR of injury compared to the first quarter of follow-up. This drop in the RR was observed for all four injury outcomes (Table 3.6). Second, in Quarter 5, the period immediately following a faller‟s certification date, there was an increase in the RR of injury compared to the first quarter of follow-up and the risk remained elevated post certification particularly for strain and acute injuries. Post estimation contrast between the acute RR of quarters 4 and 5 indicated that quarter 5 had a significantly higher RR (Chi2=3.22, p=0.072). Relative risks were significantly higher for strain injuries in Quarter 7 (RR=2.77, 95% CI 1.40, 5.48) and for serious injuries in Quarter 6 of follow-up (RR=2.04, 95% CI 1.02, 4.05) in comparison to the first quarter. The RR of acute injuries was closer to „1‟ by the end of the year post certification (RR=1.11, 95% CI 0.58, 2.14). The age group with the greatest relative risk (RR) of acute injury were the youngest fallers aged 15 to 24 years with a significant three-fold increased risk of acute injuries compared to fallers aged 45 to 54 years (RR=3.44, 95% CI 1.54,7.68). Post estimation contrasts between the RR‟s of fallers aged 15 to 24 indicated significant differences from the RR‟s of those aged 25 to 34 (Chi2=5.32, p=0.0211), 35 to 44 (Chi2=5.31, p=0.0212) and 55 to 64 (Chi2=5.05, p=0.0246). The relative risk of injury among the oldest age group was closer to 1.0 (RR=1.18, 95% CI 0.65, 2.14). This same pattern was observed for strain injuries but for serious injuries the higher RRs  50  were recorded for those aged 25 to 34 years of age and those over 55 years of age (95% CIs included „1‟) (Table 3.6). Among tree faller types, production fallers had a greater RR of injury than non-production fallers for each claim type. This was particularly true for serious injuries (RR=6.02, 95% CI 0.83, 43.58). Across all injury types, fallers who worked in the Coast Forest Region had a significantly greater RR of injury than those who worked in the Northern Forest Region (RR=5.40 95% CI 2.28, 12.80 for acute injuries). Fallers who worked in the Southern Forest Region also had a greater RR of injury than those working in the Northern Forest Region, but this was typically about half of the RR of injury for fallers working in the Coast Forest Region (RR=2.64 95% CI 1.09, 6.35). Post estimation comparisons of the Coast and Southern Interior Forest Region RR‟s indicated significant difference (Chi2=9.56, p=0.0020). Based on work experience, no distinct trends in the RR of injury were observed across experience groups or by injury type. Compared to those with 3 to 5 years of experience, high relative risks of injury were observed among fallers with 6 to 10 years of experience for acute injuries (RR=1.46, 95% CI 0.86, 2.49), among fallers with 1 to 2 years of experience for strain injuries (RR=2.19 95% CI 1.00, 4.79), and among fallers with 21 plus years of experience for serious injuries (RR=2.15, 95% CI 1.03, 4.50). Compared to those with a Grade Level 1 certification, the relative risk of acute injuries was elevated across higher certification grade levels, although the risk was only significantly higher among those at Grade Level 3 (RR=2.14, 95% CI 1.25, 3.68). For strain injuries, in comparison to Grade Level 1 the relative risk increased steadily and significantly by Grade Level with the highest relative risk among those with Grade Level 5 (RR=3.14, 95% CI 1.45, 6.80). Elevated 51  relative risks compared to Grade Level 1 were observed for serious injuries among those at Grade Levels 3 and 4, although these estimates were not statistically significant. Compared to fallers who had no previous claims, fallers who had claims within 2 years prior to entering survival analysis were at a significantly greater RR of injury. Across all injury types, fallers with a previous injury had an approximate three-fold increased risk of acute and serious injury and an approximate two-fold increased risk of strain injury during the study period compared to those with no prior injury. In comparison to periods of low unemployment (0 – 6%), across all injury types the highest relative risks associated with unemployment levels were observed for periods of the highest unemployment (12-32%). In particular, there was a two-fold increased risk of acute injury during high unemployment periods (RR=2.25, 95% CI 1.19, 4.24). Post estimation tests indicated that the acute RR of injury during periods of 12 – 32% unemployment was significantly different from 6 – 9% unemployment (Chi2=2.68, p=0.1019) and 9 – 12% unemployment (Chi2=4.97, p=0.0258).  52  Table 3.6: Results of adjusted, discrete survival analysis. non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  One Year Before Certification Quarter 1! Quarter 2! Quarter 3! Quarter 4! One Year After Certification Quarter 5! Quarter 6! Quarter 7! Quarter 8!  ref. 0.94[0.58,1.55] 1.01[0.62,1.64] 0.67[0.38,1.18]  ref. 0.98[0.53,1.82] 1.06[0.58,1.96] 0.59[0.28,1.24]  ref. 0.82[0.37,1.82] 0.92[0.42,2.02] 0.75[0.32,1.77]  ref. 1.14[0.54,2.41] 0.77[0.34,1.76] 0.47[0.18,1.24]  1.45[0.91,2.32] 1.56+[0.97,2.49] 1.79*[1.13,2.85] 1.25[0.76,2.07]  1.15[0.62,2.15] 1.62[0.89,2.92] 1.11[0.58,2.13] 1.11[0.58,2.14]  1.91+[0.95,3.84] 1.38[0.65,2.93] 2.77**[1.40,5.48] 1.41[0.65,3.05]  1.58[0.78,3.22] 2.04*[1.02,4.05] 1.88+[0.93,3.80] 1.64[0.80,3.39]  Age group 15-24 Age group 25-34 Age group 35-44 Age group 45-54 Age group 55-64  2.93***[1.56,5.49] 1.48+[0.99,2.23] 1.34+[0.96,1.87] ref. 1.16[0.75,1.81]  3.44**[1.54,7.68] 1.47[0.85,2.54] 1.44[0.92,2.26] ref. 1.18[0.65,2.14]  2.25[0.82,6.21] 1.47[0.81,2.67] 1.20[0.74,1.94] ref. 1.10[0.57,2.12]  1.16[0.32,4.23] 1.73+[0.96,3.11] 1.13[0.69,1.86] ref. 1.23[0.71,2.15]  Production faller type Not-production faller type  2.09+[0.92,4.76]  1.31[0.57,3.04]  a  -  6.02+[0.83,43.58]  ref.  ref.  a  -  ref.  Coast Region Southern Region Northern Region  5.48***[2.80,10.73] 2.43*[1.22,4.84] ref.  5.40***[2.28,12.80] 2.64*[1.09,6.35] ref.  5.82**[2.00,16.88] 2.20[0.73,6.62] ref.  4.21***[1.84,9.64] 1.46[0.61,3.50] ref.  1.61+[0.97,2.67]  1.27[0.64,2.50]  2.19*[1.00,4.79]  1.76[0.77,4.01]  ref.  ref.  ref.  ref.  1.43+[0.94,2.19]  1.46[0.86,2.49]  1.39[0.69,2.83]  1.78+[0.93,3.42]  1.33[0.87,2.05]  1.12[0.64,1.96]  1.83+[0.93,3.61]  1.32[0.67,2.62]  1.35[0.82,2.20]  1.24[0.65,2.34]  1.55[0.71,3.38]  2.15*[1.03,4.50]  ref.  ref.  ref.  ref.  Work experience 12 years Work experience 35 years Work experience 610 years Work experience 11-20 years Work experience 21+ years Certificate grade level 1 Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5  1.24[0.83,1.84]  1.60+[0.98,2.63]  0.86[0.44,1.69]  1.04[0.59,1.83]  2.02***[1.33,3.07]  2.14**[1.25,3.68]  2.02*[1.05,3.88]  1.71+[0.95,3.06]  2.18***[1.44,3.31]  1.62+[0.92,2.86]  3.04***[1.63,5.67]  1.51[0.83,2.75]  2.15**[1.24,3.71]  1.48[0.66,3.32]  3.14**[1.45,6.80]  1.07[0.44,2.62]  Previous claims No previous claims  2.87***[2.18,3.76] ref.  3.30***[2.29,4.75] ref.  2.43***[1.61,3.67] ref.  2.99***[2.01,4.44] ref.  53  Table 3.6: Continued  Unemployment 06% Unemployment 69% Unemployment 912% Unemployment 1232%  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  ref.  ref.  ref.  ref.  1.13[0.80,1.61]  1.48[0.90,2.46]  0.81[0.49,1.35]  1.26[0.76,2.10]  1.16[0.78,1.71]  1.23[0.70,2.16]  1.11[0.65,1.90]  1.11[0.62,1.97]  1.77*[1.11,2.83]  2.25*[1.19,4.24]  1.47[0.74,2.94]  1.42[0.67,2.99]  3251 3251 3251 3251 N 95% confidence intervals in brackets. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 ref. = Reference value ! Study variables of interest a Insufficient number of injuries for the “not-production” faller type to include production type as a covariate  3.2.6. Risk of Injury - Sensitivity Analysis Applying the broad manual tree falling claim definition in the adjusted survival analysis did not result in different outcomes compared to those obtained using the specific manual tree falling claims (Appendix J). An additional sensitivity analysis was conducted by extending the survival analysis period with 2 years of follow-up prior to certification and 2 years of follow-up post certification for the specific and broad falling definitions (Appendix K and L). For comparison purposes between models with different survival analysis lengths (2 vs. 4 years), the same quarter was chosen as a reference period (9-12 months before certification). The quarterly RRs of injury during the time periods of 1 year before and after certification were nearly identical in both the 2 year and 4 year models. In addition, there were no significant changes in the RRs of injury in the second year of follow-up after fallers obtained certification.  54  4.  DISCUSSION  4.1. Summary of Key Findings The main objective of this thesis was to determine if there were changes in the injury claim rate for manual tree fallers at the ecologic level (industry wide) and at the individual level prior to and following mandatory safe work practice certification. In the initial phases of creating the proposal for this thesis it was thought that there were sufficient numbers of new faller trainees and grandfathered fallers to conduct a comparison of the health outcomes between the two cohorts. However, it was discovered that over 99% of the certified faller cohort was grandfathered; consequently injury rate and risk of injury results in this thesis can only be said to be representative of the occupational health outcomes for manual tree fallers that have been certified through a grandfathering process. The results from the assessment for the RR of injury show that the BCFTS grandfathering process did not change the health outcomes of manual tree fallers for all injury types assessed when comparing time points one year before certification to one year after. However, at the ecologic level acute injury rates in the Coast Forest Region were shown to be at the lowest rates ever in 2008 following the mandatory requirement of BCFTS certification in 2006. These different health outcomes suggest that the injury rates following the mandatory requirement of BCFTS certification have decreased to the lowest levels yet; and that this may be due to regulatory exclusion of uncertified manual tree fallers from the forestry industry. It may represent an industry-wide effect of reducing injury rates by restricting new and/or unqualified fallers from working in the field, but resulted in little change in the individual level risk of injury among the current experienced cohort of fallers (i.e. those grandfathered in the certification program). However, the annual forest region injury rates were not adjusted for confounders and, as such, must be interpreted with caution. 55  4.2. Outcomes of Certification Through a “Grandfathering” Process Grandfathering manual tree fallers through the BCFTS certification process may have been a contributing reason why there was an overall decrease in the acute injury rates for the Coast Forest Region when comparing 1997 to 2008. This may seem somewhat paradoxical given that it has been shown that there is no change in the long term risk of injury. However, where grandfathering may be effective is at the ecologic level since manual tree fallers who failed or did not attempt the certifying exam challenge have been excluded from working in this occupation (or new and inexperienced fallers may have been deterred from entering the workforce due to the certification commitment, both financial and time). These fallers were unable to meet the minimum standard for industry best practices and have consequently been withdrawn from the trade. Thus, the lowest acute injury rates of the entire follow-up period for the Coast Forest Region are in 2008 when only certified fallers were permitted to work in BC and are likely to be representative of the injury rates for grandfathered fallers. The overall decrease of the Coast Forest Region acute injury rates shown in Table 3.3 is similar to the overall forestry subsector (14). A similar downward trend has been observed for the overall provincial injury claim rates as well (17). Both of these injury rates do not account for the effect of confounders and are subject to the influence of other factors. Contributing factors to the decrease in the overall provincial injury rates from 1997-2008 may include such influences as regulatory changes, progressive development of safety culture or technological advancements in the workplace (53). As shown in Table 3.3, the acute injury rate for the Coast Forest Region remained relatively stable through the optional certification period (2003 – 2006). It was not until after it became mandatory to be certified that the acute injury rate for the Coast Forest Region dropped to its  56  lowest level of follow-up. As mentioned above, the drop of this injury rate may be indicative of the exclusion of un-qualified fallers from the trade who did not have the mandatory certification. Since the injury rates in Table 3.3 are not adjusted for confounders such as economic conditions, interpretation of the rates must be done with recognition that they may be influenced by uncontrolled confounders. Nevertheless, based on the comparison of injury rates during the optional and mandatory certification periods, the outcomes of this thesis suggest that at an ecologic level optional certification is not an effective safety intervention since no regulation of the intervention occurs. However, when a safety intervention such as certification is mandatory and is enforced through regulation it might result in decreased injury rates. If a safety intervention involving safe work practice skills certification is stated in occupational regulations and enforced using citations or penalties even greater compliance may be observed (33), leading to a further decrease in injury rates. The results of this thesis indicate that there was no significant change in individual-level risk of injury for certified fallers (Table 3.6) and this is consistent with the findings of other logging safe work practice certification programs (26,27), a tractor operating certification program for adolescents (29) and a commercial vehicle driver licensing program (30). This result, however, is contradictory to other certification program health outcomes, such as the decrease in mortality rates determined for mandatory licensing to operate motorcycles (31). The disagreement between the outcomes of faller certification and motorcycle operation may be due to the motorcycle licensing health outcome assessment only involving mortality as a key outcome measure and not including non-fatal injuries. All of the above mentioned studies were based on a combination of both experienced and inexperienced cohorts, thus outcome comparisons should be interpreted with caution since the certified faller cohort in this thesis only consisted of experienced fallers. No published research was found that assessed occupational certification via a grandfathering 57  process that involved a skills examination solely for experienced workers, thus this research presents an original contribution to the body of occupational and public health research. When observing the RR of injury across the eight distinct follow-up periods two trends were observed for all injury types: in the quarter immediately before the certification exam there was a non-significant decrease in the RR of injury and in the three quarters following the certification exam there were both significant and non-significant increases in the risk of injury. A plausible explanation of why there was a decrease in the RR of injury in the quarter immediately before the certifying exam was that during this point in the follow-up fallers might have been preparing for the exam by studying exam preparation materials and utilizing the safe work practices required to pass the forthcoming exam. The increases in RR of injury immediately following the certifying exam may have been due to an increase in the reporting of injuries which has been found to follow safety interventions (25). However, these trends may have been attributable to a methodological weakness involving the assumption that all fallers were working full time felling trees before and after certification (see 4.5.1). It is possible that the certification exam itself was a key point in time that indicated the transition from a period of manual tree falling unemployment to a period of full time employment cutting down trees. A hypothesis was suggested that it could be possible that manual tree fallers obtained certification during periods without employment. However, many of the fallers obtained certification while working since they were examined in a cut block where they were employed (57). An alternate hypothesis was that fallers sought the certification exam upon returning from a seasonal period of layoff from employment, however, upon determining the frequency of faller certification by date, there was no indication of months where more certification exams occurred compared to other months. In fact, certification exam dates appeared to be equally distributed throughout a calendar year.  58  As mentioned above, following the certification exam, the increased risk of injury may be attributable to an increase in the amount of injury reporting which has been found to occur with other occupational safety interventions (25). A greater propensity to report injuries may have led to an increase in the apparent incidence of injury occurrence in comparison to the precertification time period, thus leading to the determination of an increased risk. However, the increase in risk for serious injuries during the same quarters suggests that an increase in reporting may not be the only contributing factor. Serious injuries are severe enough to require reporting for workers‟ compensation and thus may not be an optional choice for injured workers to report. A potential reason why no change in the RR of injury was observed when comparing one year before to one year after obtaining certification is that no behavioral interventions effective at modifying behavior such as structured training or feedback on workplace skills (19-21) were provided. As long as manual tree fallers demonstrated industry best practices to a standard acceptable to pass the certifying exam, they were designated a certified status. The determination that there was no change in the risk of injury for grandfathered fallers in the long term could be attributable to the safe work practices the BCFTS tested for and the environment in which manual tree falling occurs. Qualitative evidence indicates that manual tree fallers in BC acknowledge that safe work practices are essential skills; however, these skills provide techniques that still must be applied in a manner that can adjust to a variable environment (23). Competence in safe work practices alone may not be a determinant of whether a faller is at risk for injury or not, rather, how a faller utilizes safe work practices and how they incorporate them into the process of felling trees may be more relevant to the risk of injury. Given that no change in the risk of injury was shown to occur (in the long term) for grandfathered fallers, alternative safety interventions may help improve health outcomes. 59  Engineered safety controls could help to remove fallers from exposure to the hazards associated with when a tree is falling over. For example, the use of hydraulic wedges that can be remotely operated removes fallers from the immediate area of a falling tree. Administrative controls could help to alleviate the fatigue that fallers may have to deal with. For example, if a workday was limited to 8 hours in total, including travel time to a worksite, increased awareness may help fallers identify potentially dangerous situations and improve reaction time in the event of an exposure to a hazard. 4.3. Synthesis of Covariate Findings The finding that injury rates in the Coast Forest Region are higher than both the Southern Interior and Northern Interior Forest Regions (Table 3.3 & Table 3.5) is in agreement with similar trends indicated by WorkSafeBC (14). A contributing reason to higher injury rates in the Coast Forest Region is that this region is where the most wood is harvested using manual tree falling. The Coast Forest Region predominantly has more steep terrain and larger trees than the interior regions; factors that inhibit the use of mechanized tree falling equipment. The youngest certified faller group (15 – 24 years old) was found to have the highest non-HCO injury rate and risk of injury (comparing risk of injury to 45 – 54 year olds). This finding is consistent with existing literature that indicates young workers have higher claim rates than older workers and elevated risk of injury (51). The acute injury rate was three times higher than the serious injury rate, suggesting that the greatest proportion of injuries sustained by young fallers was non-serious acute injuries. This finding is also consistent with existing evidence that indicates young workers to have a smaller prevalence of serious injuries in comparison to older workers (51).  60  The finding that certified fallers with the least amount of work experience had the highest nonHCO claim rates is in agreement with another study that has shown an increase in claim rates with increased work experience (54). This evidence provides support for targeting the BCFTS at new entrants into the manual tree falling occupation to complete mandatory training since fallers with the least amount of work experience have been shown to have the greatest injury rate. BCFTS certified manual tree fallers with previous claims history were shown in Table 3.6 to have a significantly higher RR for injury than those who did not have previous claims 2 years prior to and during follow-up. This finding aligns with those found by Craib et al. (2007) that indicate that workers who have a previous claim are more likely to sustain future injuries. The dangerous nature of the work that manual tree fallers perform and hazardous environment in which they work certainly may be contributing to this finding. The evidence of manual tree fallers having one of the highest injury rates in BC (2) is indicative that the odds of being injured are great and may be associated with the likelihood of sustaining more than one injury. A possible contributing factor to why workers with previous claims are more likely to sustain multiple injuries is that this may be a surrogate measure of environmental, work task, and physical characteristics. Employer encouragement for employees to submit claims may also contribute to the propensity for apparent re-injury since employers that condone submitting workers compensation claims may deter injured employees from claiming injuries (thus these employees appear to have no claims). Certified fallers that had follow-up during periods of high unemployment were shown to have higher injury rates than those who had follow-up during periods of low unemployment (Table 3.5). Several studies have shown the opposite trend, indicating that as unemployment rates increase, injury rates decrease (39,40)(39,40,58). It has been noted in these studies that injury rates are artificially low during periods of high unemployment since workers are less likely to 61  report occupational injuries due to concerns about security of employment (39,40). Reporting of injuries may not be inhibited for manual tree fallers due to the severity and disability incurred from incidents that typically occur in this dangerous occupation, necessitating workers compensation. An additional factor that may have contributed to these outcomes is that there may have been some misclassification of the unemployment rate for the forest industry included in the RR modeling. The unemployment rates were based on the entire primary resource sector which includes fishing, mining, and oil and gas in addition to forestry. It may be possible that the non-forestry industry sectors skewed the unemployment rates and interfered with obtaining accurate RR outcomes indicating the effect of unemployment. Working as a production faller was shown to have a six fold increased risk of serious injury (RR=6.02, 95% CI 0.83, 43.58) compared to non-production fallers. This result is indicative of the increased hazard of manually falling trees full time. The difference between production and non-production fallers was that production fallers were assumed to be working full time manually falling trees, whereas non-production fallers were assumed to perform other duties as a part of their work tasks. The finding that manual tree fallers who have a certificate grade level greater than one are at greater risk of injury (compared to certificate grade level of one) is suggestive of the hazards associated with cutting down trees with trunk diameters larger than 21”. Large trees are more challenging to cut down and have additional inherent risks in comparison to smaller trees. Higher certificate grade levels were based on the size of trees that fallers were tested on at the time of certification and are indicative of the ability of fallers to cut down larger tree sizes. However, they do not necessarily indicate the predominant tree size that they cut down on a daily basis. Certificate grade level 5 fallers who were certified on the largest trees (those over 60” in diameter) are distinguished from less skilled fallers since it is unlikely that a faller with little 62  work experience and resultant skills would be hired to cut down these types of trees at the time of certification. The demographics of the certified faller cohort (Table 3.4) were similar to the regional and age characteristics of 2001 and 2006 Statistics Canada census data. Comparison of the forest region distribution of manual tree fallers by typical area of work at the time of certification (based on the BC Forest Safety Council faller registry) and place of residence (based on 2001 and 2006 Statistics Canada Census‟s) indicated nearly exact results (4,13). The match in the distribution of typical work area and place of residence is suggestive that manual tree fallers, for the most part, live and work in the same forest region. We also found that from 2003 to 2008 the largest proportion of fallers (33.1%) were 35 – 44 years old at the time of certification and this corresponds with the 2001 census indicating that 34.8% of chainsaw and skidder operators in the same age group account for the majority of this workforce (4). However, the 2006 census determined that the greatest proportion of chainsaw and skidder operators (29.5%) shifted to the 45 – 54 year old age group. This shift in the most predominant age group indicates an ageing workforce. 4.4. Policy Implications The outcomes of the ecologic injury rate study suggest that when an occupational health and safety regulatory body is considering implementing safe work practice skills certification, they should make the certification a mandatory requirement by writing it into regulation. Writing mandatory safe work practice skills certification into regulation provides a means to enforce compliance and excludes unskilled workers from performing work that they are not qualified to do.  63  Most regulations come into force with a minimum notice period beforehand so that stakeholders can prepare to be compliant. The results from the ecologic injury rate study suggest that during the notice period prior to safe work practices skills certification becoming mandatory (the optional period) most workers will seek to obtain a certified status, but industry wide injury rate decreases may not be observed until after the regulation is in effect. The results from the individual level risk of injury study suggest that the safety of workers who are grandfathered in the BCFTS will not likely improve in the long term. When making a policy involving safe work practice skills certification it is understandable that experienced workers may have resistance to the requirement to complete training for skills they have already learnt. Hence, grandfathering clauses are utilized to introduce industry wide qualification standards that acknowledge past on-the-job training experienced workers have received to differentiate them from those with insufficient previous experience. If improvements for the occupational health outcomes for grandfathered workers are considered to be a part of the goal of implementing mandatory certification, an additional safety intervention will be required. Based on the outcomes of this thesis and the dynamic environment that manual tree fallers work in, it may be likely that even if experienced fallers went through the training process no change in individual risk for injury would have been observed since the safe work practices taught through training may not provide skills above and beyond what they already possess to deal with such a high risk environment. The retrospective nature of this study led to many assumptions with inherent biases. This highlights the need to prospectively collect data that enables the most accurate assessment of future safety interventions.  64  4.5. Study Limitations 4.5.1. Selection Bias Each certified faller in the BC Forest Safety Council registry was assigned a “linked” (89%) or “unlinked” (11%) status depending on unique identification in the PopDataBC population directory (based on BC Medical Services Plan billing and records). The success rate of this linkage was lower than the 97% achieved in a different study utilizing WorkSafeBC claims data to assess work related musculoskeletal disorders among health care workers in BC (59). A requirement for PopDataBC to confidently state that a faller did or did not have WorkSafeBC claims was the need for a unique identity (using name, date of birth & postal code) in their population medical registry for the province. Two factors may have contributed to why the certified faller cohort had a lower linkage success rate than in the healthcare worker study; out of the 422 unlinked certified fallers, 65.4% had no date of birth and 1.7% had no postal code recorded in the registry. An additional reason why 2.4% of the unlinked fallers were not found in the population directory was because their postal code indicated that they are not residents of BC and therefore they did not appear in the BC population medical registry. Had these fallers been included in the individual level risk of injury study they would have influenced the RR values towards the null since they do not have any identifiable injury claims. Excluding the unlinked fallers from the cohort may have biased the RR outcomes in either a negative or positive manner. The influence of this bias is dependent on the amount of claims that the excluded unlinked fallers may have had. However, whether the unlinked fallers had claims or not is unknown since they could not be identified in the BC population directory which was a requirement to confidently determine individual workers‟ compensation claims status. If the unlinked fallers had many injury claims, then the RR estimates obtained are negatively biased since these additional claims are missing from the overall risk estimates. If the unlinked fallers 65  had few claims, then the RR estimates are positively biased since the claims free contribution of these fallers to the overall risk is missing. The influence of these biases may be minor given the small number of unlinked fallers (11.3% of the original cohort). All fallers in the assessed cohort were selected for inclusion with the assumption that they were working as a manual tree faller full time throughout the follow-up. We did not have data from the faller registry on employment status or on gaps in employment during the two year follow-up period. The inability to censor workers in the cohort from such time periods when they were not employed as a manual tree faller negatively biased the injury rates and RR‟s since they appeared to be a manual tree faller who was not injured while working full time. It is possible that the decrease in RRs for injury in the quarter immediately preceding the certification exam was due to fallers not having any claims for injuries that would have resulted in a disability interfering with the ability to perform the tasks required in the practical skills portion of the certifying exam. If manual tree fallers were to have been disabled due to workplace injuries they would not have been able to complete the certifying exam and thus would have delayed the examination until reaching a healthier state. Consequently, most fallers were likely to have arranged to challenge the certifying exam when they were not injured and thus in the quarter before the exam the RR‟s for injury were at the lowest values of the study. Inclusion of fallers who were aged 65 years or older in the individual level risk of injury cohort was questioned since it was unknown if these individuals were still working as fallers or if they had obtained a certified status for other reasons. Exploratory analysis of the initial dataset indicated that during follow-up no faller aged 65 years or older had sustained any manual tree falling specific injuries. The non-existence of any claims for the 65 years and older age group was considered to be indicative that this was a group of certified fallers who were not working as 66  fallers or were not working as fallers on a full time basis. Excluding certified fallers who were 65 years old or older (0.9% of the original cohort) may have slightly biased the observed RR upward if our assumptions about them are incorrect and they were truly working as fallers but not sustaining injuries. 4.5.2. Information Bias The outcomes of this thesis are fully based on injury claims that have been occupationally coded by administrative staff at WorkSafeBC. It is possible that some injury claims may have not been identified for inclusion into the ecologic and individual level studies due to occupational misclassification of the claims themselves. If this happened, then the injury rates and RR of injury estimates are negatively biased. Conversely, some of the claims that were included in the studies may have been misclassified to appear to meet the inclusion criteria, but actually should have been excluded. If this occurred, then the injury rates and RR estimates were positively biased. In the descriptive analysis of injury rates, wood harvest volume data used for the denominator in the ecologic level injury rates includes both the wood that was harvested manually and mechanically. The proportion of wood harvested by either method was not known and consequently the volume of wood harvested by manual tree falling is inflated by assuming that all of the harvest volume is attributable to manual tree fallers. As a result the injury rates obtained using the harvest volumes as a denominator are negatively biased and appear to be lower than they actually are, in particular for two regions in BC. The Northern and Southern Interior Forest Regions have extensive mechanized harvesting due to the favorable terrain and this is very likely why the injury rates were much lower than in the Coast Forest Region. To minimize the interference that wood harvested by mechanized equipment introduced into the  67  injury rates, key findings were based on the Coast Forest Region. The finding that the greatest proportion of manual tree fallers (44.9%) worked in the Coast Forest Region at the time of certification is indicative that this region had the most manual tree falling activity during the follow-up and supports the basis of key findings on this region. 4.5.3. Reporting Bias The work experience variables included in the individual level study were based on information reported by manual tree fallers at the time of certification. Fallers who reported 10 or more years of work experience exhibited digit preference. The outcome of this was that there were particularly high frequencies of fallers who reported having 10, 15, 20, 25 or 30 years of working experience. In these upper ranges of work experience few fallers were observed to have years of work experience that was not a multiple of 5. This could have led to a misclassification of fallers into the 6 – 10, 11 – 20, and 21+ year work experience groups since there actual work experience may have classified them into a different group. However, the process of grouping fallers into categories, rather than individual years, likely mitigated the effects that the digit preference would have had on the RR estimates. Strain injuries were consistently shown to have lower injury rates in both studies. This may have been a negatively biased result due to a lack of reporting of strain injuries by manual tree fallers. It was reported by an industry expert that it is known within the industry that there are manual tree fallers who are working while suffering from strain injuries. 4.5.4. Unmeasured Confounders A major limitation to this thesis is that in both studies there is a lack of accounting for the change in the business structure of the forest industry during the follow-up periods of both studies. This change has seen shift in the industry away from large companies that perform many integrated 68  aspects of the forest harvesting process to independent contracting. These changes in the industry structure have impacted the environmental working conditions of manual tree fallers and the safety culture of the industry itself (12). It is possible that the effect of the changes in the forest industry that occurred throughout the follow-up periods has interfered with obtaining representative outcomes. Furthermore, variations in income earning by either hourly or piece rate work may have influenced the outcomes of both studies. It has been shown that in most occupations piece work has had negative effects on the health and safety of workers compared to those that are salaried (60) and this has been shown to be the case for both forestry workers (61) and manual tree fallers (62). Considering that the ecologic study was solely based on descriptive analysis, it was not possible to account for confounding variables such as the effect of independent contracting mentioned above or other economic influences on the injury rates. Furthermore, statements made about changes in injury rates from one time period to the next are based on visual observation and not based on statistical significance. 4.6. Study Strengths 4.6.1. Study Design Both studies had a claim inclusion criteria based on an occupational code specific to manual tree falling. The advantage of this is that all manual tree falling occupation specific claims were identified across the forestry subsector. Manual tree falling claims are not exclusive to the Manual Tree Falling and Bucking CU and therefore claims attributed to tree falling are found in many forestry subsector CU‟s (Appendix D). For example, claims due to manual tree falling can be found to be attributed to all forestry subsector CU‟s except for Log Booming and Marine Log Salvage. 69  The numerator and denominator used in the ecologic study provided an indication of occupation and region specific injury rates. This provided the ability to remove the interference of other occupations that are found in CU based injury rates and to focus on the Coast Forest Region where most manual tree falling occurs. Classification unit based injury rates can only be used as an indicator of occupation specific injury rates that match the primary industrial activity of a CU. The reason why CU based injury rates can only be used as general indicators of occupation specific injury rates is that a firm classified into a specific CU can have employees working in jobs that are not the primary industrial activity. For example, a firm classified into the Manual Tree Falling and Bucking CU may have a crew of manual tree fallers, supervisory staff and administrative staff. If one of the administrative staff has an accepted injury claim for an office related injury, the injury would count towards the overall Manual Tree Falling and Bucking CU injury rate even though it had nothing to do with the tasks required to fall and buck a tree. Furthermore, CU based injury rates are based on the entire province, so region specific rates cannot be determined. The individual level risk of injury study was based on known dates of certification and injury providing a framework to structure survival analysis. This provided a metric of time to indicate a meaningful scale in which events such as injuries and certification occurred. Using the date of certification as a common point in time permitted all fallers included in the cohort to be framed in a common relative time scale even though fallers were certified throughout a 6 year period. Constructing the survival analysis in this manner facilitated follow-up periods for all fallers so that they shared states of un-certified or certified status during the same quarters. During periods of disability due to any type of accepted injury claim or withdrawal from the occupation fallers were censored and did not contribute to the estimated RR of injury for a given quarter. Censoring  70  fallers from the survival analysis prevented negatively biasing RR estimates since they were assumed to not be working while disabled. Constructing the survival analysis using quarterly time periods facilitated the assessment of RR during the follow-up while mitigating the effects of variability due to low injury counts. Even though the date of injury was recorded by day and year in the claims data (essentially providing a continuous measurement of time), support for using quarterly periods as opposed to days, weeks, or months was based on the issue of obtaining reliable estimates of injury risk for a relatively small cohort. If, for example, weeks had been used for time periods there would be few total injuries per week and a large amount of variability would be introduced into the weekly risk of injury outcomes. The outcome of aggregating injuries into discrete quarters by injury date was that the total amount of injuries in a given time period was considerable enough to reduce the variability in risk of injury to a point where trends could be observed and 95% confidence intervals could be calculated. 4.6.2. Applicability of the Study Results The greatest strength of this thesis is that it determined the occupational health outcomes of manual tree fallers using injury criteria specific to manual tree falling activity. Outcomes of both studies were based on injury indicators that were specific to the BCFTS safety intervention. This provided a means to assess the people who the intervention was aimed at, the effects of the intervention at the industry and individual level, and determine differences among the types of injuries. Injury claims were classified as acute, strain or serious injury types to obtain different classifications of key outcome measures. The motivation to specifically identify acute injuries was based on the aim of the BCFTS to reduce the incidence of injuries due to one-time events 71  such as struck by, struck in or caught in accidents. However, the BCFTS information booklets also mentioned tips on how to avoid strain injuries, so a criterion was created to investigate this injury type, even though they did not appear to be the main purpose of the intervention. The serious injury criterion was able to assess the effect of the BCFTS upon the injuries that cause considerable disability, a relevant outcome for this industry and a focus of WorkSafeBC for prevention. The sensitivity analyses involving the extension of the individual level of risk follow-up period from 2 to 4 years and expanding the claim occupational coding definition supported the validity of the results obtained. For both types of sensitivity analysis the RR estimates during the followup periods one year before and after certification were nearly identical to the results obtained using the specific falling definition and a two year follow-up period. The outcomes indicating no significant changes in the RRs of injury in the second year of follow-up after obtaining certification (quarters 13 – 16 in the 4 year model) support the conclusion that BCFTS certification does not change the level of risk for injury at the individual level. Considering the specific cohort criteria and key outcome measures used in both studies, the occupational health outcomes of these studies are applicable to other occupations involving manual labour and hazardous work tasks. The outcomes determined for both of these studies are specific to certification by a grandfathering process that involves passing an exam with both a written and practical skills component. Generalizability of these outcomes to other occupations should be only considered for those that involve hazardous manual labour and a similar grandfathering process. These study outcomes are not suggested to be applied for policy decisions involving a certification process that has a training component. Training involves standardized education for workers who are 72  seeking certification and the BCFTS safety intervention involving grandfathering in this study did not involve such a component. 4.7. Future Studies 4.7.1. New Faller Trainees It is recommended that a future study on the BCFTS be conducted at a later date once more new faller trainees have obtained a certified status. The outcomes of such a study would yield indications if mandatory training and mentoring improves the occupational health outcomes of manual tree fallers. Since new faller trainees comprised only 1 % of the initial cohort in this thesis it was decided that this study is primarily focused on the issue of grandfathering into a certified status. Inclusion of new faller trainees in the study would only have led to risk of injury outcomes that would be highly variable due to the small number of individuals and the even smaller number of claims. The first group of new faller trainees was certified in 2005 and by the end of 2009 a total of 110 students completed the new faller training course; however, a condition of eligibility to attempt the certifying exam was the requirement of up to 180 days of workplace training with a qualified faller supervisor/trainer. The new faller trainees were responsible to obtain a work placement with a qualified supervisor/trainer to complete the prerequisite workplace training. Out of the 110 new faller trainees, 57 successfully achieved a certified status by the end of 2009. Two factors may explain why only 33% of the new faller trainees obtained their certification: some new faller trainees did not pass the certifying exam or some could not find a work placement to obtain the prerequisite supervised work experience due to unfavorable economic conditions. Over time the new faller trainee cohort will continue to grow as there will be a need to replace ageing workers. 73  4.7.2. Data Collection Data for fallers who were not certified would aid in making comparisons between fallers who sought certification and those that did not. This information could yield how many people may have been excluded from the occupation and what their risk of tree falling injury was in comparison to those fallers that obtained a certified status. Furthermore, data on fallers that failed the certifying exam could help determine different health outcomes between those that attempted the exam and passed or failed, as well as those that did not attempt to obtain certification. More accurate follow-up data could improve the accuracy of the censoring process in future cohort studies of this nature. The censoring in the individual level risk study of this thesis was partly based on the estimated days of disability from an accepted WorkSafeBC claim (based the amount of disability compensation) and could have been under or over estimated. Actual return to work dates would rectify the inaccuracy of using estimated days of disability and yield reliable censoring. Prospectively designing a database to assess manual tree faller risk of injury could ensure the collection of more accurate data indicating parameters such as work history to facilitate censoring without assumptions. Explicit requirements for data entry could ensure that all individuals have information to control for important confounders such as age and sex. Mechanisms could be put in place to obtain missing data in the event of a data entry error or omission such as access to documents that indicate the required data. Denominator criteria could be determined during the prospective database design in order to ensure that it is representative of the desired incident rate and is unbiased, for example, harvest volumes that are representative of the amount of wood cut down by manual tree fallers. In order to achieve the optimal  74  prospective database and design for safety intervention evaluation, an interdisciplinary team of subject matter experts has been suggested to be the ideal method (36).  75  5.  CONCLUSIONS  Based on using the Coast Forest Region as an indicator of industry wide manual tree faller injury rates, the acute injury rate dropped to the lowest level in 2008 following the regulatory requirement to be a certified manual tree faller in 2006. At the individual level, the effect of mandatory certification through grandfathering status did not change the risk of injury among an experience cohort of tree fallers. While safe work practices are shown to reduce injury rates, the grandfathering process may not have realized the full potential of the certification requirement. Further, the implementation of standard practices, while a good foundation, may not be sufficient to reduce the risk of injury in a complex work environment such as the one that manual tree felling occurs in. The outcomes of this thesis can be used to help form policy making decisions about certification as an occupational health and safety intervention. Based on this thesis, policy makers should consider the limited effect of grandfathering as a condition for certification to reduce injuries and the need to build upon certification as a foundation to continue to reduce risks in this high risk environment.  76  BIBLIOGRAPHY 1. Statistics Canada. Labour Force Historical Review, 1976-2009. 2010; Report No.;71F0004XVB. 2. WorkSafeBC. A Report and Action Plan to Eliminate Deaths and Serious Injuries in British Columbia‟s Forests - Final Report of the Forest Safety Task Force; 2004. 3. WorkSafeBC. Occupational Health and Safety Faller Serious Injury and Fatal Review; 2009. 4. Statistics Canada. British Columbia Forest Region Custom Tabulation (based on the 2001 Census); 2010. 5. West R, Shkrum MJ, Young JG. Commercial logging fatalities in Ontario, 1986-1991. American Journal of Forensic Medicine and Pathology. 1996; 17(4):299-304. 6. Brodie LR, Ibrahim JE. Fatal injury in tree felling and related activities, Victoria, Australia 1992–2007. Injury Prevention. 2010; 16(1):53-56. 7. Wang J, Bell JL, Grushecky ST. Logging injuries for a 10-year period in Jilin Province of the People's Republic of China. Journal of Safety Research. 2003; 34(3):273-279. 8. Bentley TA, Parker RJ, Ashby L. Understanding felling safety in the New Zealand forest industry. Applied Ergonomics 2005; 36(2):165-175. 9. Scott DF. A study of logger fatalities from 1992-2000. Injury Prevention. 2004; 10(4):239243. 10. WorkSafeBC. Occupational Health and Safety Regulation. 2009; Available at: http://www2.worksafebc.com/publications/OHSRegulation/Home.asp. Accessed 2/24, 2011. 11. British Columbia Ministry of Forests Resource Tenures and Engineering Branch. Forest Administrative Areas; 2004. 12. Western Fallers' Association. Western fallers' association report - A view from the field: Report on contributing factors to faller accidents in British Columbia; 2005. 13. Statistics Canada. British Columbia Forest Region Custom Tabulation (based on the 2006 Census); 2010. 14. WorkSafeBC Business Information & Analysis. Forestry-Related CUs Summary Information Including Segmentations by Regional District Groups; 2010. 15. WorkSafeBC Statistical Services Department. WorkSafeBC Statistics 2000; 2001. 16. WorkSafeBC Statistical Services Department. WorkSafeBC Statistics 2004; 2005. 17. WorkSafeBC Statistical Services Department. WorkSafeBC Statistics 2009; 2010. 18. WorkSafeBC Risk Management. Statistics for Forestry; 2008; Available at: http://www2.worksafebc.com/Portals/Forestry/Statistics.asp. Accessed 5/25, 2009.  77  19. Burke MJ, Sarpy SA, Smith-Crowe K, Chan-Serafin S, Salvador RO, Islam G. Relative Effectiveness of Worker Safety and Health Training Methods. American Journal of Public Health. 2006; 96(2):315-324. 20. Rasmussen K, Carstensen O, Lauritsen JM, Glasscock DJ, Hansen ON, Jensen UF. Prevention of farm injuries in Denmark. Scandinavian Journal of Work, Environment & Health. 2003; 29(4):288-296. 21. Robson L, Stephenson C, Schulte P, Amick B, Chan S, Bielecky A, Wang A, Heidotting T, Irvin E, Eggerth D, Peters R, Clarke J, Cullen K, Boldt L, Rotunda C, Grubb P. A systematic review of the effectiveness of training & education for the protection of workers. 2010. 22. Montorselli NB, Lombardini C, Magagnotti N, Marchi E, Neri F, Picchi G, Spinelli R. Relating safety, productivity and company type for motor-manual logging operations in the Italian Alps. Accident Analysis & Prevention. 2010; 42(6):2013-2017. 23. J. M. Colman, H. K. Kahle. Is performance variability necessary or counterproductive for safely felling trees in the coastal region of British Columbia? (MSc thesis) Sweden: Lund University; 2010. 24. Helmkamp JC, Bell JL, Lundstrom WJ, Ramprasad J, Haque A. Assessing safety awareness and knowledge and behavioral change among West Virginia loggers. Injury Prevention. 2004; 10(4):233-238. 25. Craib KJP, Hackett G, Back C, Cvitkovich Y, Yassi A. Injury Rates, Predictors of Workplace Injuries, and Results of an Intervention Program Among Community Health Workers. Public Health Nursing. 2007; 24(2):121. 26. Bell JL, Grushecky ST. Evaluating the effectiveness of a logger safety training program. Journal of Safety Research. 2006; 37(1):53-61. 27. Mujuru P, Helmkamp JC, Mutambudzi M, Hu M, Bell JL. Evaluating the impact of an intervention to reduce injuries among loggers in West Virginia, 1999-2007. Journal of Agricultural Safety and Health. 2009; 15(1):75-88. 28. Spangenberg S, Baarts C, Dyreborg J, Jensen L, Kines P, Mikkelsen KL. Factors contributing to the differences in work related injury rates between Danish and Swedish construction workers. Safety Science. 2003; 41(6):517-530. 29. Marlenga B, Doty BC, Berg RL, Linneman JG. Evaluation of a policy to reduce youth tractor crashes on public roads. Injury Prevention. 2006; 12(1):46-51. 30. Hagge RA, Romanowicz PA. Evaluation of California's commercial driver license program. Accident Analysis & Prevention. 1996; 28(5):547-559. 31. McGwin G, Whatley J, Metzger J, Valent F, Barbone F, Rue LW. The effect of state motorcycle licensing laws on motorcycle driver mortality rates. The Journal of Trauma, Injury, Infection, and Critical Care. 2005; 56:415-419. 32. Smitha MW, Kirk KA, Oestenstad KR, Brown KC, Lee S. Effect of state workplace safety laws on occupational injury rates. Journal of Occupational and Environmental Medicine. 2001; 43(12):1001-1010. 78  33. Tompa E, Trevithick S, McLeod C. Systematic review of the prevention incentives of insurance and regulatory mechanisms for occupational health and safety. Scandinavian Journal of Work Environment and Health. 2007; 33(2):85-95. 34. Office of the Auditor General of British Columbia. Preventing fatalities and serious injuries in BC Forests: Progress Needed; 2008. 35. Goldenhar L, Schulte P. Intervention research in occupational health and safety. Journal of Occupational Medicine 1994; 36(7):763-778. 36. Goldenhar LM, Schulte PA. Methodological issues for intervention research in occupational health and safety. American Journal of Industrial Medicine. 1996; 29(4):289-294. 37. Haddon W, Jr. The changing approach to the epidemiology, prevention, and amelioration of trauma: the transition to approaches etiologically rather than descriptively based. American Journal of Public Health and the Nations Health. 1968; 58(8):1431-1438. 38. Rummer B. Models for analyzing logging safety. International Journal of Forest Engineering. 1995; 7(1):63-71. 39. Boone J, van Ours JC. Are recessions good for workplace safety? Journal of Health Economics. 2006; 25(6):1069-1093. 40. Davies R, Jones P, Nuñez I. The impact of the business cycle on occupational injuries in the UK. Social Science and Medicine. 2009; 69(2):178-182. 41. WorkSafeBC Statistical Services Department. Dataman Dictionary. 2008. 42. Ministry of Forests. Mark Monthly Scaling History Selection. 2011; Available at: https://www15.for.gov.bc.ca/hbs/opq/ftas/invoiceSummary.do?actionType=P400&radSelecte dReport=scaling. Accessed 02/14/2011, 2011. 43. Centre for Health Services and Policy Research. Workers Compensation Board Data Fields Manual (Modified from original WCB metadata document); 2009. 44. Esri. ArcGIS. 2010;10. 45. StataCorp. Stata. 2007;10. 46. Statistics Canada. Standard Occupational Classification (SOC) 1991. 1993; Available at: http://stds.statcan.gc.ca/soc-cnp/1991/ts-rt-eng.asp?cretaria=i. Accessed 02/11, 2011. 47. Koehoorn M, Demers PA, Hertzman C, Village J, Kennedy SM. Work organization and musculoskeletal injuries among a cohort of health care workers. Scandinavian Journal of Work, Environment & Health. 2006; 32(4):285-293. 48. Singer JD, Willet JB. Applied Longitudinal Data Analysis. First ed. New York, NY: Oxford University Press; 2003. 49. Krickeberg K. Survival Analysis: A Self-Learning Text. New York, NY: Springer New York; 2005. 79  50. Richardson DB. Discrete time hazards models for occupational and environmental cohort analyses. Occupational and Environmental Medicine. 2010; 67(1):67-71. 51. Breslin F, Koehoorn M, Smith P, Manno M. Age-related differences in work injuries and permanent impairment: A comparison of workers' compensation claims among adolescents, young adults, and adults. Occupational and Environmental Medicine. 2003; 60(9):10e-16e. 52. Breslin FC, Smith P. Age-related differences in work injuries: A multivariate, populationbased study. American Journal of Industrial Medicine. 2005; 48:50-56. 53. Breslin FC, Smith P, Koehoorn M, Hyunmi L. Is the workplace becoming safer? Statistics Canada Perspectives. 2006; Report No.:75-001-XIE:18-23. 54. Breslin C, Smith P. Trial by fire: a multivariate examination of the relation between job tenure and work injuries. Occupational and Environmental Medicine. 2006. 63:27-32. 55. Akaike, H. Information theory as an extension of the maximum likelihood principle. Petrov BN, Csaki F, editors. International Symposium on Information Theory; 1973; Budapest, Hungary. 56. Schwarz G. Estimating the Dimension of a Model. The Annals of Statistics. 1978; 6(2):461464. 57. Mueller S. Personal communication with Steve Mueller (BC Forest Safety Council) about how an examination location was determined. Phone conference 2010 Jun 8. 58. Barmby TA, Orme CD, Treble JG. Worker absenteeism: an analysis using microdata. The Economic Journal. 1991; 101(405):214-229. 59. Koehoorn M, Cole DC, Hertzman C, Lee H. Health care use associated with work-related musculoskeletal disorders among hospital workers. Journal of Occupational Rehabilitation. 2006; 16(3):411-424. 60. Johansson B, Rask K, Stenberg M. Piece rates and their effects on health and safety – A literature review. Applied Ergonomics. 2010; 41(4):607-614. 61. Sundström-Frisk C. Behavioural control through piece-rate wages. Journal of Occupation Accidents. 1984; 6(1-3):49-59. 62. Mason K. The effect of piecework on accident rates in the logging industry (incorporating a different approach to the exposure problem). Journal of Occupational Accidents. 1977; 1(3):281-294. 63. Von Korff M. Studying the Natural History of Back Pain. Spine. 1994; 19(18S):2041S.  80  APPENDIX A: WORKERS’ COMPENSATION CLAIM RATE INCIDENT LOCATION ASSIGNMENT Claims were assigned an operating location number by WorkSafeBC to indicate the corresponding employer operating location. For 49.6% of the claims in the dataset there were employer operating location numbers that identified a corresponding operating location and indicated which forest region the claim was associated with. For 49.5% of claims in the dataset that had no employer operating location number, but had only one operating location listed, the operating location was assumed to be a one to one claim to employer operating location match. However, for the remaining 1% of claims in the dataset there were no employer operating location numbers and the operating locations associated with these claims had to be manually selected. It is worth noting that 8 claims were excluded from the dataset because the employer operating locations were not in the province of BC. The remaining claims with no employer operating location numbers had multiple operating locations that were possibly associated with the claims. In order to identify one employer operating location that was representative of the region each claim was attributed to the following decision making hierarchy was employed: 1) Only employer operating locations that had a start date and end date that encompassed an injury date were considered; 2) If all employer operating locations were in one forest region, then one employer operating location was randomly chosen to represent the claim forest region; 3) If employer operating locations were in multiple forest regions, then the operating location in the same forest region as the WorkSafeBC claim area office was chosen to represent the claim forest region; 81  4) If employer operating locations were in multiple forest regions and no WorkSafeBC claim area office information was available, then the operating location in the same forest region as the injured worker home address was chosen to represent the claim forest region. For the circumstances when a WorkSafeBC claim area office or employee home address was used to verify a forest region that a claim was associated with, an operating location in the same forest region was randomly chosen to represent the claim forest region. Using this decision making criteria 0.46% of claims were associated with a forest region based on all operating locations in one forest region, 0.09% of claims were associated with a forest region based on verification with a WorkSafeBC claim area office and 0.3% of claims were associated with a forest region based on verification with an injured worker home address.  82  APPENDIX B: WORKERS’ COMPENSATION CLAIM RATE CLAIM INCLUSION AND EXCLUSION CRITERIA BASED ON NATURE OF INJURY (NOI), SOURCE OF INJURY (SOI) AND ACCIDENT TYPE (ACC) Inclusion/Exclusion Criteria  Nature of Injury (NOI) Abrasion Amputation Back strain Bursitis & related  Included  Carpal tunnel syndrome Concussion Contusion Dislocation Exposure to cold Exposure to heat Fractures Heat burns Hernia Laceration Multiple injuries Other injuries Other strains Tendinitis, tenosynovitis  Accident Type (ACC) Caught in Exposure to heat, cold Fall from elevation Fall on same level Matter in eye Other accidents Other bodily motion Overexertion Repetitive motion Rubbed, abraded Struck against Struck by  Dermatitis  Animal products, food products Animals (live)  Hearing loss  Boilers, pressure vessels  Infected blisters Other occupational diseases Poisoning Raynaud‟s phenomenon Respiratory inflammation Stress  Boxes containers  Exposure to noise Exposure to toxic substances Motor vehicle accidents  Buildings & structures  Public transportation  Conjunctivitis  Excluded  Source of Injury (SOI) Bodily motion Clothing, textiles Environmental cold Flame, fire, smoke & other heat Hand tools Hoisting apparatus Ladders Logs, tree products Machines Metal items Mineral items Miscellaneous Power tools Trees, plants Working surfaces  Acts of violence, force  Electrical apparatus Furniture, fixtures Glass items Noise Parasitic infection People Vehicles  83  APPENDIX C: DISCRETE SURVIVAL ANALYSIS CLAIM INDICATOR INCLUSION AND EXCLUSION CRITERIA BASED ON NATURE OF INJURY (NOI), SOURCE OF INJURY (SOI) AND ACCIDENT TYPE (ACC) Inclusion/Exclusion Criteria  Nature of Injury (NOI) Abrasion Amputation Back strain Bursitis & related  Included  Carpal tunnel syndrome Concussion Contusion Dislocation Exposure to cold Exposure to heat Fractures Heat burns Hernia Laceration Multiple injuries Other injuries Other strains Tendinitis, tenosynovitis Allergic reaction Chemical burns  Excluded27  Source of Injury (SOI) Bodily motion Clothing, textiles Environmental cold Flame, fire, smoke & other heat Hand tools Hoisting apparatus Ladders Logs, tree products Machines Metal items Mineral items Miscellaneous Power tools Trees, plants Working surfaces  Accident Type (ACC) Caught in Exposure to heat, cold Fall from elevation  All conveyors Animal products, food products  Acts of violence, force  Conjunctivitis  Animals (live)  Dermatitis  Boilers, pressure vessels  Hearing loss Heart attack Infected blisters Other occupational diseases Poisoning Radiation Raynaud‟s phenomenon Respiratory inflammation Stress  Boxes containers Buildings & structures Ceramic items  Fall on same level Matter in eye Other accidents Other bodily motion Overexertion Repetitive motion Rubbed, abraded Struck against Struck by  Exposure to noise Exposure to toxic substances Motor vehicle accidents Public transportation  Chemicals Electrical apparatus Furniture, fixtures Glass items Liquids Noise Parasitic infection People Sun, other radiation Vehicles  More NOI‟s and SOI‟s were present in the Study 2 dataset than Study 1 because injury types resulting from all occupations were required for survival analysis dataset structuring. The Study 1 dataset was based only on injuries due to the occupation “Faller,Feller – logging” and therefore had fewer types of NOI‟s and SOI‟s. 27  84  APPENDIX D: DISTRIBUTION OF MANUAL TREE FALLER CLAIMS BY CLASSIFICATION UNIT (CU) INCLUDED IN THE WORKERS’ COMPENSATION CLAIM RATE STUDY COHORT Classification Unit Number  Classification Unit Description  Claim Frequency (n )  Percent (%)  703013  Manual Tree Falling and Bucking  1771  57.1  703008  Integrated Forest Management  1012  32.7  113  3.7  703006  Ground Skidding, Horse Logging, or Log Loading  703003  Cable or Hi-Lead Logging  44  1.4  703002  Brushing and Weeding or Tree Thinning or Spacing (not elsewhere specified)  38  1.2  703015  Shake Block Cutting  30  1.0  703012  Logging Road Construction or Maintenance  20  0.7  732044  Log Hauling  16  0.5  703019  Helicopter Logging  15  0.1  763015  Forest Management Services  11  0.4  703005  Forest Fire Fighting  6  0.2  703014  Mechanized Tree Falling  6  0.2  703016  Tree Planting or Cone Picking  6  0.2  703004  Dry Land Sort  5  0.2  703011  Log Processing  4  0.1  703001  Chemical Brushing and Weeding or Chemical Tree Thinning or Spacing  2  0.06  703009  Log Booming  1  0.03  703018  Marine Log Salvage  -  -  732024  Log Towing  -  -  3100  100  TOTAL  85  APPENDIX E: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE COAST FOREST REGION OF BC  86  APPENDIX F: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE SOUTHERN INTERIOR FOREST REGION OF BC  87  APPENDIX G: ANNUAL MANUAL TREE FALLER INJURY RATES IN THE NORTHERN INTERIOR FOREST REGION OF BC  88  APPENDIX H: RESULTS OF MULTIVARIABLE MODEL FITTING  One Year Before Certification Quarter 1 Quarter 2 Quarter 3 Quarter 4  Reduced model with no work experience  Reduced model with no region  Reduced model with no certificate level  Reduced model with no unemployment rate  Baseline hazard  Full Model  Reduced model with no age  ref. 1.00 [0.63,1.60] 1.08 [0.69,1.72] 0.63+ [0.37,1.07]  ref. 0.94 [0.58,1.55] 1.01 [0.62,1.64] 0.67 [0.38,1.18]  ref. 0.95 [0.58,1.55] 1.01 [0.62,1.65] 0.67 [0.38,1.18]  ref. 0.94 [0.58,1.53] 1.09 [0.68,1.76] 0.69 [0.40,1.19]  ref. 0.95 [0.58,1.55] 1.01 [0.62,1.66] 0.67 [0.38,1.18]  ref. 0.94 [0.57,1.54] 1.01 [0.62,1.65] 0.68 [0.39,1.19]  ref. 0.96 [0.59,1.58] 1.00[ 0.61,1.63] 0.62+ [0.35,1.08]  1.38 [0.89,2.13] 1.53+ [1.00,2.34] 1.60* [1.05,2.45] 1.08 [0.68,1.72]  1.45 [0.91,2.32] 1.56+ [0.97,2.49] 1.79* [1.13,2.85] 1.25 [0.76,2.07]  1.45 [0.91,2.32] 1.56+ [0.98,2.50] 1.80* [1.13,2.86] 1.25 [0.76,2.07]  1.46 [0.92,2.32] 1.70* [1.08,2.69] 1.88** [1.19,2.96] 1.24 [0.75,2.03]  1.45 [0.91,2.32] 1.55+ [0.97,2.48] 1.82* [1.14,2.89] 1.24 [0.75,2.05]  1.47 [0.92,2.36] 1.58+ [0.99,2.53] 1.84* [1.15,2.92] 1.29 [0.78,2.12]  1.31 [0.83,2.08] 1.38 [0.88,2.18] 1.56+ [1.00,2.43] 1.08 [0.67,1.76]  2.64*** [1.56,4.46] 1.43* [1.03,1.98] 1.35* [1.01,1.80] ref. 1.19 [0.79,1.80]  2.80** [1.50,5.22] 1.45+ [0.97,2.16] 1.34+ [0.97,1.87] ref. 1.16 [0.75,1.79]  3.20*** [1.71,5.98] 1.60* [1.07,2.39] 1.41* [1.01,1.95] ref. 1.13 [0.73,1.76]  2.94*** [1.57,5.51] 1.49+ [1.00,2.24] 1.34+ [0.96,1.86] ref. 1.16 [0.75,1.80]  One Year After Certification Quarter 5 Quarter 6 Quarter 7 Quarter 8 Age group 15-24 Age group 25-34 Age group 35-44 Age group 45-54 Age group 55-64  2.93*** [1.56,5.49] 1.48+ [0.99,2.23] 1.34+ [0.96,1.87] ref. 1.16 [0.75,1.81]  89  Baseline hazard  Production faller type Not-production faller type Coast Region Southern Region Northern Region Work experience 1-2 years Work experience 3-5 years Work experience 6-10 years Work experience 11-20 years Work experience 21+ years Certificate grade level 1 Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5  Full Model  Reduced model with no age  Reduced model with no work experience  Reduced model with no region  Reduced model with no certificate level  Reduced model with no unemployment rate  2.09+ [0.92,4.76] ref.  2.02+ [0.89,4.61] ref.  2.28* [1.00,5.17] ref.  2.26+ [0.99,5.15] ref.  2.50* [1.10,5.67] ref.  2.17+ [0.95,4.94] ref.  5.48*** 5.49*** [2.80,10.73] [2.81,10.74] 2.43* 2.48** [1.22,4.84] [1.25,4.92] ref. ref.  5.63*** [2.88,10.99] 2.37* [1.19,4.70] ref.  8.07*** [4.26,15.27] 2.54** [1.28,5.04] ref.  5.43*** [2.77,10.61] 2.46* [1.24,4.90] ref.  1.66* [1.00,2.76] ref. 1.52+ [1.00,2.32] 1.48+ [0.97,2.26] 1.54+ [0.95,2.50]  1.61+ [0.97,2.67] ref. 1.45+ [0.95,2.22] 1.36 [0.89,2.08] 1.37 [0.84,2.24]  1.61+ [0.97,2.67] ref. 1.43+ [0.94,2.19] 1.33 [0.87,2.05] 1.35 [0.82,2.20]  1.79* [1.09,2.95] ref. 1.27 [0.84,1.92] 1.08 [0.73,1.60] 0.97 [0.65,1.44]  ref. 1.24 [0.83,1.84] 2.02*** [1.33,3.07] 2.18*** [1.44,3.31] 2.15** [1.24,3.71]  ref. 1.24 [0.83,1.85] 2.08*** [1.37,3.16] 2.26*** [1.49,3.42] 2.24** [1.30,3.85]  1.65+ [1.00,2.74] ref. 1.38 [0.90,2.10] 1.30 [0.85,1.98] 1.35 [0.83,2.19] ref. 1.29 [0.87,1.91] 2.05*** [1.35,3.09] 2.36*** [1.57,3.55] 2.14** [1.25,3.66]  ref. 1.92*** [1.32,2.79] 3.72*** [2.55,5.44] 4.54*** [3.15,6.53] 4.37*** [2.62,7.29]  ref. 1.25 [0.84,1.86] 2.08*** [1.37,3.16] 2.26*** [1.50,3.43] 2.21** [1.28,3.82]  90  Baseline hazard Previous claims No previous claims Unemployment 0-6% Unemployment 6-9% Unemployment 9-12% Unemployment 12-32% Observations 25756 AIC 3463.982 BIC 3529.233 95% confidence intervals in brackets. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 ref. = Reference value  Full Model  Reduced model with no age  Reduced model with no work experience  Reduced model with no region  Reduced model with no certificate level  Reduced model with no unemployment rate  2.87*** [2.18,3.76] ref.  2.93*** [2.23,3.84] ref.  2.74*** [2.10,3.56] ref.  3.15*** [2.40,4.14] ref.  3.14*** [2.39,4.11] ref.  2.92*** [2.22,3.83] ref.  ref.  ref.  ref.  ref.  ref.  1.13 [0.80,1.61] 1.16 [0.78,1.71] 1.77* [1.11,2.83] 22441 2802.009 3018.512  1.14 [0.80,1.62] 1.16[ 0.79,1.71] 1.78* [1.11,2.85] 22441 2804.846 2989.274  1.11 [0.79,1.58] 1.18 [0.81,1.73] 1.86** [1.18,2.93] 23598 2942.658 3128.243  1.09 [0.77,1.54] 1.10 [0.75,1.62] 1.65* [1.03,2.62] 22449 2849.119 3049.594  1.16 [0.82,1.65] 1.22 [0.83,1.80] 1.97** [1.23,3.15] 22441 2815.052 2999.481  22441 2802.154 2994.602  91  APPENDIX I: REGRESSION COEFFICIENTS FOR NON-HCO, ACUTE, STRAIN AND SERIOUS INJURIES MEETING THE SPECIFIC FALLING DEFINITION non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  Constant Quarter 2 Quarter 3 Quarter 4 Quarter 5 Quarter 6 Quarter 7 Quarter 8  -7.81 -0.06 0.007 -0.40 0.37 0.44 0.58 0.22  -7.98 -0.23 0.06 -0.53 0.14 0.48 0.11 0.11  -7.86 -0.12 -0.09 -0.22 0.65 0.37 1.06 0.38  -9.37 0.13 -0.26 -0.76 0.46 0.71 0.63 0.50  Age group 15-24 Age group 25-34 Age group 35-44 Age group 55-64  1.07 0.40 0.29 0.15  1.24 0.39 0.37 0.17  0.83 0.41 0.20 0.13  0.15 0.55 0.12 0.21  Production faller type  0.74  0.27  0.00a  1.79  Coast Region Southern Region  1.70 0.89  1.69 0.97  1.71 0.74  1.44 0.38  0.48  0.24  0.78  0.56  0.34  0.38  0.34  0.58  0.29  0.11  0.53  0.28  0.30  0.21  0.43  0.77  0.21 0.70 0.78 0.76  0.47 0.76 0.48 0.39  -0.31 0.58 1.03 1.06  0.04 0.53 0.41 0.07  Previous claims  1.05  1.19  0.87  1.10  Unemployment 6-9% Unemployment 9-12% Unemployment 12-32%  0.12 0.15 0.57  0.40 0.21 0.81  -0.20 0.08 0.28  0.23 0.10 0.35  Work experience 1-2 years Work experience 6-10 years Work experience 11-20 years Work experience 21+ years Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5  Insufficient number of injuries for the “not-production” faller type to include production type as a covariate a  92  APPENDIX J: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE BROAD MANUAL TREE FALLING CLAIM DEFINITION (2 YEAR FOLLOWUP) non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  One Year Before Certification Quarter 1! Quarter 2! Quarter 3! Quarter 4! One Year After Certification Quarter 5! Quarter 6! Quarter 7! Quarter 8!  ref. 0.94[0.61,1.46] 1.01[0.66,1.56] 0.73[0.45,1.17]  ref. 1.07[0.62,1.87] 1.06[0.60,1.85] 0.59[0.30,1.15]  ref. 0.71[0.35,1.45] 0.95[0.49,1.84] 0.88[0.44,1.76]  ref. 1.06[0.56,2.02] 0.73[0.36,1.49] 0.55[0.25,1.19]  1.35[0.89,2.05] 1.55*[1.03,2.34] 1.46+[0.96,2.22] 1.38[0.90,2.10]  1.09[0.61,1.93] 1.52[0.89,2.61] 1.11[0.62,1.98] 1.32[0.76,2.30]  1.71+[0.93,3.12] 1.52[0.82,2.84] 1.90*[1.04,3.49] 1.47[0.77,2.79]  1.39[0.75,2.58] 1.79+[0.99,3.24] 1.46[0.78,2.71] 1.74+[0.95,3.18]  Age group 15-24 Age group 25-34 Age group 35-44 Age group 45-54 Age group 55-64  2.70***[1.56,4.69] 1.39+[0.97,1.99] 1.34*[1.01,1.79] ref. 1.16[0.79,1.69]  3.09**[1.54,6.19] 1.45[0.89,2.35] 1.45+[0.97,2.16] ref. 1.39[0.83,2.30]  2.11[0.85,5.26] 1.31[0.77,2.24] 1.21[0.80,1.82] ref. 0.96[0.55,1.69]  1.44[0.52,3.99] 1.57+[0.94,2.62] 1.08[0.70,1.65] ref. 1.33[0.82,2.15]  2.01*[1.02,3.94]  1.26[0.63,2.52]  a  -  2.75+[0.86,8.75]  ref.  ref.  a  -  ref.  3.88***[2.35,6.41] 1.85*[1.11,3.10] ref.  4.48***[2.26,8.86] 2.15*[1.07,4.32] ref.  3.34**[1.59,7.02] 1.55[0.72,3.32] ref.  2.77**[1.50,5.13] 0.99[0.51,1.93] ref.  1.38[0.88,2.16]  1.16[0.64,2.08]  1.79[0.87,3.67]  1.31[0.64,2.68]  ref.  ref.  ref.  ref.  1.27[0.87,1.83]  1.26[0.80,2.00]  1.26[0.68,2.36]  1.32[0.75,2.31]  1.27[0.88,1.84]  0.99[0.61,1.60]  1.88*[1.04,3.39]  1.33[0.76,2.33]  1.30[0.85,1.99]  1.07[0.61,1.86]  1.75[0.89,3.42]  1.73+[0.92,3.22]  ref. 1.12[0.80,1.57] 1.82**[1.27,2.60] 1.98***[1.38,2.83] 2.25***[1.42,3.58] 2.51***[1.98,3.17] ref.  ref. 1.35[0.88,2.06] 1.84*[1.15,2.93] 1.52+[0.93,2.48] 1.49[0.74,2.97] 2.80***[2.04,3.84] ref.  ref. 0.89[0.51,1.55] 1.99*[1.15,3.45] 2.70***[1.58,4.60] 3.40***[1.78,6.49] 2.19***[1.54,3.10] ref.  ref. 0.94[0.58,1.53] 1.61+[0.97,2.67] 1.48[0.88,2.47] 1.59[0.80,3.18] 2.56***[1.82,3.60] ref.  Production faller type Not-production faller type Coast Region Southern Region Northern Region Work experience 1-2 years Work experience 3-5 years Work experience 6-10 years Work experience 11-20 years Work experience 21+ years Certificate grade level 1 Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5 Previous claims No previous claims  93  Unemployment 0-6% Unemployment 6-9% Unemployment 9-12% Unemployment 12-32% N  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  ref. 1.02[0.76,1.37] 0.87[0.62,1.22] 1.33[0.88,2.00]  ref. 1.24[0.83,1.86] 0.82[0.51,1.33] 1.53[0.88,2.66]  ref. 0.80[0.52,1.22] 0.96[0.61,1.53] 1.23[0.68,2.24]  ref. 1.23[0.80,1.90] 0.92[0.56,1.52] 1.20[0.63,2.31]  3251  3251  3251  3251  95% confidence intervals in brackets. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 a Insufficient number of injuries for the “not-production” faller type to include production type as a covariate ref. = Reference value ! = Study variable of interest  94  APPENDIX K: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE SPECIFIC MANUAL TREE FALLING CLAIM DEFINITION (4 YEAR FOLLOWUP)  Two Years Before Certification Quarter 1! Quarter 2! Quarter 3! Quarter 4! Quarter 5! Quarter 6! Quarter 7! Quarter 8! Two Years After Certification Quarter 9! Quarter 10! Quarter 11! Quarter 12! Quarter 13! Quarter 14! Quarter 15! Quarter 16! Age group 15-24 Age group 25-34 Age group 35-44 Age group 45-54 Age group 55-64 Production faller type Not-production faller type Coast Region Southern Region Northern Region Work experience 1-2 years Work experience 3-5 years Work experience 6-10 years Work experience 11-20 years Work experience 21+ years  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  0.97[0.59,1.58] 0.63[0.36,1.09] 0.89[0.54,1.48] 0.90[0.54,1.49] ref. 0.95[0.58,1.56] 1.00[0.61,1.64] 0.65[0.37,1.13]  0.87[0.46,1.66] 0.53+[0.25,1.10] 0.86[0.45,1.63] 0.84[0.44,1.61] ref. 0.99[0.53,1.84] 1.06[0.57,1.95] 0.57[0.27,1.19]  1.12[0.51,2.44] 0.80[0.34,1.86] 0.96[0.43,2.15] 0.99[0.44,2.21] ref. 0.90[0.40,2.04] 0.92[0.40,2.08] 0.78[0.33,1.85]  1.24[0.59,2.59] 0.53[0.21,1.34] 1.29[0.62,2.67] 1.01[0.47,2.18] ref. 1.15[0.55,2.42] 0.77[0.34,1.76] 0.47[0.18,1.23]  1.39[0.87,2.21] 1.47[0.93,2.33] 1.68*[1.07,2.65] 1.17[0.72,1.92] 0.94[0.56,1.58] 1.13[0.68,1.86] 1.06[0.64,1.77] 0.71[0.40,1.26]  1.10[0.59,2.04] 1.51[0.84,2.69] 1.04[0.55,1.96] 1.05[0.55,1.98] 1.18[0.63,2.21] 1.03[0.53,1.97] 0.69[0.33,1.43] 0.75[0.37,1.53]  1.86+[0.91,3.80] 1.41[0.66,3.00] 2.75**[1.39,5.45] 1.38[0.64,2.99] 0.55[0.20,1.48] 1.29[0.59,2.84] 1.68[0.79,3.57] 0.66[0.25,1.70]  1.58[0.78,3.19] 2.02*[1.03,3.98] 1.86+[0.93,3.72] 1.63[0.80,3.32] 1.08[0.49,2.35] 1.75[0.86,3.56] 1.59[0.77,3.28] 1.35[0.64,2.86]  2.91***[1.77,4.77] 1.40*[1.03,1.90] 1.27+[0.99,1.62] ref. 1.02[0.74,1.39]  3.30***[1.79,6.10] 1.28[0.85,1.91] 1.18[0.86,1.63] ref. 0.94[0.61,1.43]  2.22+[0.95,5.19] 1.57+[0.98,2.53] 1.39+[0.96,2.03] ref. 1.13[0.71,1.80]  1.02[0.38,2.72] 1.23[0.80,1.90] 1.17[0.84,1.64] ref. 0.91[0.60,1.39]  1.84*[1.03,3.30]  1.57[0.79,3.09]  2.66+[0.84,8.45]  1.84[0.81,4.20]  ref.  ref.  ref.  ref.  4.45***[2.86,6.91] 1.75*[1.11,2.78] ref.  4.90***[2.76,8.70] 1.98*[1.09,3.59] ref.  3.77***[1.88,7.53] 1.43[0.69,2.98] ref.  3.38***[1.97,5.81] 1.20[0.67,2.14] ref.  1.27[0.84,1.92]  1.25[0.72,2.15]  1.27[0.67,2.42]  1.16[0.61,2.21]  ref.  ref.  ref.  ref.  1.30[0.93,1.80]  1.49+[0.97,2.27]  1.04[0.62,1.75]  1.37[0.86,2.19]  1.41*[1.02,1.95]  1.42[0.92,2.18]  1.37[0.84,2.24]  1.25[0.78,2.00]  1.61*[1.11,2.32]  1.62+[1.00,2.62]  1.57[0.90,2.76]  1.90*[1.14,3.16]  95  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  Certificate grade level 1 Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5  ref. 1.22[0.91,1.62] 1.68**[1.23,2.30] 1.89***[1.39,2.57] 1.88**[1.25,2.82]  ref. 1.28[0.89,1.82] 1.62*[1.09,2.39] 1.31[0.88,1.97] 1.11[0.61,2.04]  ref. 1.10[0.68,1.80] 1.79*[1.07,2.99] 2.92***[1.80,4.72] 3.19***[1.78,5.73]  ref. 1.15[0.77,1.71] 1.67*[1.09,2.56] 1.66*[1.08,2.54] 1.81*[1.02,3.19]  Previous claims No previous claims  2.46***[2.00,3.03] ref.  2.49***[1.88,3.31] ref.  2.39***[1.76,3.25] ref.  1.85***[1.36,2.53] ref.  ref. 1.05[0.82,1.35] 1.08[0.80,1.45] 1.36+[0.96,1.94]  ref. 1.25[0.89,1.77] 1.17[0.78,1.75] 1.57+[0.98,2.50]  ref. 0.84[0.58,1.22] 0.98[0.64,1.51] 1.15[0.68,1.96]  ref. 1.21[0.85,1.71] 1.07[0.70,1.64] 1.32[0.79,2.22]  3251  3251  3251  3251  Unemployment 0-6% Unemployment 6-9% Unemployment 9-12% Unemployment 12-32% N  95% confidence intervals in brackets. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 ref. = Reference value ! = Study variable of interest  96  APPENDIX L: RESULTS OF ADJUSTED, DISCRETE SURVIVAL ANALYSIS FOR THE BROAD MANUAL TREE FALLING CLAIM DEFINITION (4 YEAR FOLLOWUP)  Two Years Before Certification Quarter 1! Quarter 2! Quarter 3! Quarter 4! Quarter 5! Quarter 6! Quarter 7! Quarter 8! Two Years After Certification Quarter 9! Quarter 10! Quarter 11! Quarter 12! Quarter 13! Quarter 14! Quarter 15! Quarter 16! Age group 15-24 Age group 25-34 Age group 35-44 Age group 45-54 Age group 55-64 Production faller type Not-production faller type Coast Region Southern Region Northern Region Work experience 1-2 years Work experience 3-5 years Work experience 6-10 years Work experience 11-20 years Work experience 21+ years  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  0.98[0.63,1.52] 0.72[0.45,1.16] 1.01[0.65,1.56] 0.99[0.64,1.53] ref 0.95[0.61,1.47] 1.01[0.65,1.55] 0.71[0.44,1.15]  1.02[0.57,1.80] 0.62[0.32,1.19] 0.94[0.53,1.68] 0.97[0.55,1.73] ref 1.08[0.62,1.88] 1.05[0.60,1.84] 0.59[0.30,1.14]  0.93[0.47,1.85] 0.87[0.43,1.75] 1.10[0.57,2.13] 1.00[0.51,1.97] ref 0.76[0.37,1.56] 0.94[0.48,1.87] 0.89[0.45,1.80]  1.20[0.64,2.27] 0.56[0.26,1.23] 1.35[0.73,2.51] 1.08[0.56,2.05] ref 1.06[0.56,2.02] 0.72[0.36,1.48] 0.55[0.25,1.19]  1.31[0.87,1.98] 1.49+[0.99,2.23] 1.41[0.94,2.13] 1.34[0.88,2.03] 0.96[0.61,1.50] 0.99[0.63,1.55] 1.02[0.65,1.60] 0.68[0.41,1.12]  1.07[0.61,1.88] 1.46[0.86,2.48] 1.09[0.61,1.92] 1.31[0.76,2.26] 1.29[0.74,2.23] 0.98[0.54,1.76] 0.76[0.40,1.42] 0.67[0.35,1.29]  1.64[0.89,3.03] 1.52[0.82,2.84] 1.86*[1.02,3.42] 1.37[0.72,2.62] 0.50[0.21,1.16] 1.01[0.50,2.02] 1.40[0.73,2.68] 0.70[0.32,1.52]  1.38[0.75,2.54] 1.76+[0.98,3.16] 1.46[0.79,2.68] 1.75+[0.97,3.17] 1.10[0.57,2.11] 1.45[0.78,2.69] 1.52[0.82,2.81] 1.29[0.68,2.44]  2.64***[1.72,4.06] 1.30+[0.99,1.71] 1.30*[1.05,1.60] ref. 1.05[0.80,1.37]  2.78***[1.62,4.77] 1.26[0.88,1.80] 1.28+[0.96,1.69] ref. 1.05[0.73,1.52]  2.31*[1.13,4.75] 1.36[0.90,2.06] 1.33+[0.97,1.82] ref. 1.04[0.69,1.56]  1.04[0.45,2.37] 1.13[0.77,1.64] 1.10[0.83,1.47] ref. 1.01[0.71,1.44]  1.87*[1.14,3.06] ref.  1.54[0.87,2.71] ref.  2.96*[1.09,8.04] ref.  1.68[0.86,3.31] ref.  3.91***[2.74,5.59] 1.61*[1.11,2.34] ref.  4.31***[2.71,6.87] 1.66*[1.02,2.70] ref.  3.34***[1.92,5.82] 1.54[0.86,2.73] ref.  3.01***[1.94,4.67] 1.07[0.67,1.72] ref.  1.17[0.81,1.67]  1.25[0.79,1.97]  1.03[0.57,1.85]  0.94[0.53,1.67]  ref.  ref.  ref.  ref.  1.18[0.89,1.57]  1.29[0.90,1.85]  1.03[0.66,1.62]  1.15[0.77,1.71]  1.34*[1.01,1.76]  1.22[0.84,1.76]  1.47+[0.97,2.25]  1.27[0.85,1.88]  1.50*[1.09,2.05]  1.41[0.93,2.13]  1.60+[0.99,2.58]  1.62*[1.05,2.50]  97  non-HCO Injuries  Acute Injuries  Strain Injuries  Serious Injuries  Certificate grade level 1 Certificate grade level 2 Certificate grade level 3 Certificate grade level 4 Certificate grade level 5  ref. 1.10[0.86,1.41] 1.51**[1.16,1.97] 1.68***[1.29,2.19] 1.91***[1.35,2.70]  ref. 1.11[0.82,1.51] 1.43*[1.02,2.02] 1.27[0.89,1.80] 1.23[0.74,2.03]  ref. 1.08[0.73,1.61] 1.63*[1.06,2.50] 2.37***[1.58,3.56] 3.02***[1.84,4.95]  ref. 0.96[0.69,1.35] 1.43+[0.99,2.06] 1.47*[1.02,2.11] 1.97**[1.24,3.14]  Previous claims No previous claims  2.10***[1.75,2.53] ref.  2.11***[1.65,2.71] ref.  2.06***[1.57,2.70] ref.  1.73***[1.32,2.25] ref.  ref. 1.05[0.84,1.30] 0.95[0.74,1.23] 1.19[0.88,1.62]  ref. 1.15[0.86,1.54] 0.93[0.65,1.32] 1.29[0.86,1.93]  ref. 0.93[0.67,1.29] 0.98[0.67,1.43] 1.08[0.68,1.72]  ref. 1.29+[0.95,1.75] 1.08[0.75,1.57] 1.21[0.77,1.91]  3251  3251  3251  3251  Unemployment 0-6% Unemployment 6-9% Unemployment 9-12% Unemployment 12-32% N  95% confidence intervals in brackets. ref. = Reference value ! = Study variable of interest  98  

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.24.1-0072114/manifest

Comment

Related Items