NOTE: This document is provided for historical purposes only.
Guiding Construction Injury Research: Data Coupled with Industry Experience—Fosbroke D, Casini V, Furrow K, Hause M, Linn H, Washenitz F
In 1996, the National Institute for Occupational Safety and Health (NIOSH), Division of Safety Research (DSR) initiated a process of examining current construction injury research, identifying research gaps, and developing a strategic research plan. Through existing injury surveillance data systems, much is known about the leading causes of fatal (falls, motor vehicles, machines, and electrocutions) and nonfatal injury (overexertion, falls, and struck by objects) in the construction industry; however, little research has focused on identifying injury problems for specific subsectors of the construction industry. Research that is focused on specific injury problems and specific types of construction work (e.g., falls during truss installation) may lead more directly to identification of effective interventions than research on general injury categories in the construction industry as a whole (e.g., falls in construction). Three high-risk construction industry sectors (highway and street construction, residential building construction, and roofing and truss installation) were selected based on a review of fatal and nonfatal injury data, the number of workers at risk, current trends in the construction industry, OSHA's regulatory agenda, an external panel evaluation of The National Program for Occupational Safety and Health in Construction, the National Occupational Research Agenda, and DSR expertise and research projects. For each industry sector, NIOSH conducted a facilitated workshop with participants from labor unions, construction companies, contractor associations, product manufacturers, insurance companies, and State and federal agencies. DSR staff identified brainstorming topics for each workshop based on the leading causes of fatal and nonfatal injury for that industry subsector. Brainstorming topics included motor vehicle incidents, falls from elevation, "struck by" incidents, and overexertion. Workshop participants were asked to suggest potential safety research ideas that could lead to a reduction of injuries from these causes in their industry sectors.
Each workshop yielded a list of more than 100 ideas about general problem areas where the participants felt further research was warranted. Following each workshop, NIOSH staff, with the assistance of federal partners, used a multivoting technique to narrow these lists to approximately 30 ideas for further study. Some suggestions were very specific (e.g., design a fall restraint system for workers riding in pickup truck beds while placing or removing traffic cones); others were broad in scope (evaluate the effectiveness of training). Other broad themes that crossed industry sector and injury categories included the following: identify high-risk construction tasks, activities, processes, and stages; improve the quality and coverage of surveillance data; develop new, or adapt existing technologies to the unique conditions of specific industries; compare the relative risks of different types of construction projects; study the relationship of training, experience, and injury; improve partnerships among research organizations, industry, and labor; and improve, simplify, and more effectively disseminate safety information. The process of combining injury surveillance data with the experience of individuals in the affected industries provided NIOSH researchers with a broader perspective on the safety research needs of the construction industry.
Work-related Fatal Injury Risk of Construction Workers by Occupation and Cause of Death—Chen GX, Fosbroke D Background.
Construction is both one of the largest and one of the most dangerous industries in the United States, but there is little literature on fatality risk by occupation within this industry. This study assessed workrelated injury fatality risk by occupation and cause of death within the construction industry, using data contained in the NIOSH National Traumatic Occupational Fatality (NTOF) surveillance system and the BLS Current Population Survey (CPS) from 1990 to 1991.
Method. Annual fatality rates were calculated for occupations with six or more work-related injury deaths during the 2-year period and presented as the number of deaths per 100,000 workers. Cause-specific annual rates were also calculated for occupations with 50 or more deaths during the 2 years. Frequencies and rates are presented for the civilian workforce only, because denominator data were not available for military personnel.
Results. A total of 1,964 work-related injury deaths occurred in the construction industry in the United States during 1990 to 1991. Construction has an average annual fatality rate of 13.2 deaths per 100,000 workers. Occupations with 100 or more deaths during the 2 years are construction laborers (463 deaths), construction supervisors (161), carpenters (153), operating engineers (146), electricians (103), and managers and administrators, n.e.c. (100). Fatalities from these six occupations accounted for 57.3% of fatalities in the construction industry. Annual fatality rates by occupation were presented and there were 22 occupations which have higher annual fatality rates than the average of 13.2 per 100,000 workers for the entire construction industry. The six occupations with the highest annual fatality rates were electrical power installers and repairers (84.6 deaths per 100,000 workers); structural metal workers (74.7); operating engineers (47.7); engineering technicians, n.e.c. (44.8); drillers, earth (40.7); and construction laborers (33.3). Operating engineers and construction laborers were the occupations with both a high annual fatality rate and a large number of injury deaths. The leading cause of death varied among occupations. Falls were the leading cause of death for construction laborers; construction supervisors; carpenters; managers and administrators, n.e.c.; roofers; structural metal workers; construction and maintenance painters; and plumbers, pipe fitters, and steamfitters. Motor vehicle crashes were the leading cause of death for truck drivers and heavy construction trades, n.e.c. Machines were the leading cause of death for operating engineers. Electrocutions were the leading cause of death for electricians. Findings from this study are consistent with previous studies and underscore the urgent need for prevention measures for work-related fatalities among construction workers. Information on fatality risk by occupation and by cause of death could be useful for planning preventive strategies in the construction industry.
Preventing Injuries and Fatalities in Highway Construction Work Zones—Melius J, Kojola W, Hoffner K, Lapping K, Blacklow B
With support from the United States Department of Transportation, the Laborers' Health and Safety Fund of North America (LHSFNA) has been conducting a project to prevent injuries and fatalities among highway construction workers. Surveillance data indicated that these workers have a fatality rate twice as high as that for the overall construction industry. The first phase of this study focused on the analysis of existing surveillance data on injuries and fatalities (including data from several highway contractors) to identify factors within the work zone contributing to fatalities and injuries among these workers. With assistance from a technical advisory group that included contractors, union members, and safety professionals, interventions to prevent these problems were identified and prioritized. These interventions ranged from better training programs to improved technology to prevent construction vehicle accidents. Currently, the LHSFNA is working with several academic institutions and other groups to evaluate or demonstrate several of these intervention approaches. These evaluation and demonstration projects should be completed by the Fall of 1997. Results from the initial surveillance study will be presented along with an overview of the intervention projects.
Machinery-related Fatalities in the U.S. Construction Industry, 1980-1992—Pratt SG, Kisner SM, Moore PH The National Traumatic Occupational Fatalities (NTOF) surveillance system identified machinery-related incidents as the fourth leading cause of traumatic occupational fatalities in the U.S. construction industry between 1980 and 1992, resulting in 1,901 deaths and 2.13 deaths per 100,000 workers. All but 22 of the victims were males. Males had nearly eight times the fatality rate observed among females (2.3 vs. 0.29). The fatality rate in the Northeast census region, 1.29, was considerably lower than the rates in other regions, which ranged from 2.05 to 2.37. Overall, fatality rates declined 50% over the study period.
Workers in three occupation divisions—precision production, craft, and repair; transportation and material moving; and handlers, equipment cleaners, helpers, and laborers—had both the highest frequency and rate of fatalities. Cranes, excavating machinery, and tractors were the machines most frequently involved. During the study period, fatality rates for tractors and cranes declined 71% and 67%, respectively, while rates for excavating machinery declined only 12%. The most common incident types were: struck by a mobile machine; overturn; and struck by a boom.
Further delineation of groups at highest risk for machinery-related injuries is complicated by a lack of data on exposure to machinery, since exposure is clearly not equivalent across all occupational groups within the construction industry. The findings suggest that injury prevention programs should focus not only on machine operators, but on those who work on foot around machines. Translation of regulations into "plain English," providing incentives for safe work practices, and addressing safety in project planning stages can also help to reduce machinery-related deaths in construction.
Manual Materials Handling Research: What We Should Know— Ayoub MM, Dempsey P
For over four decades, individuals from various disciplines have extensively researched the tasks comprising manual material handling (MMH): lifting, lowering, pushing, pulling, carrying and holding. Various criteria for defining acceptable task demands have been developed from the principles of biomechanics, physiology and psychophysics. Although significant bodies of literature exist on each class of these criteria, there are still several areas which need to be further examined. The validity of several of these criteria is unknown or in question, primarily due to a lack of epidemiological verification.
This presentation provides a critical review of MMH criteria, the conflicts between these criteria, and the shortcomings of such criteria. The presentation will also provide a set of areas needing further examination and the research needs in the areas of biomechanics, physiology and psychophysics to assist in the refinement of these criteria. Finally, the presentation addresses the validity issue for these criteria.
Risk Factors of Muscle Injury and Falls Related to Load Handling Technique—Oddsson LIE
The main goal of the postural control system is to maintain stability in all situations and during all tasks that are performed. An external perturbation of posture triggers specific automatic responses which act to restore equilibrium. Few studies have investigated the interaction between automatic postural responses and voluntary movements. It has been hypothesized that certain combinations of voluntary movement and postural perturbation, such as a slip or a trip, may cause a conflict between postural and voluntary motor commands simultaneously requiring different functions of the same muscle/muscle group. It is further suggested that this conflict increases the risk of direct tissue injury and/or a fall. The occurrence of conflicts between motor commands is probably common during different occupational activities such as walking, lifting or carrying loads as well as during reaching for or pushing and pulling objects. An unexpected slip or trip that coincides with such an ongoing voluntary activity would have the potential to cause a motor command conflict. However, most of these situations will not lead to an injury. Nevertheless, the occurrence of a motor command conflict may still represent an injury prone situation and the margin of safety, before an injury occurs, is likely decreased. The current presentation will report findings that suggest the presence of a motor command conflict in the lumbar back muscles of subjects performing a lifting motion while being perturbed on a balance platform. A series of experiments were conducted where subjects were perturbed randomly at the feet in an anteriorposterior direction during vertical lifting of a 20 kg load. Subjects stood on a specially designed moving platform (BALDER, BALance DisturbER) programmed to accelerate (9.81 m/s 2 ) and then decelerate (9.81 m/s 2 ) over a distance of 0.11 m. Kinematic information was obtained from a video based motion analysis system (ARIEL, APAS). Muscle activity was recorded from soleus, tibialis anterior, anterior deltoideus and erector spinae (ES). Posterior platform perturbation triggered a hip strategy by the subject which resulted in a sudden cessation of activity in the ES muscle. In this phase of the movement, the task of lifting the load and supporting the trunk would require the ES muscle to be continuously active. However, instead there was an interruption of the lifting motion followed by a brief sudden trunk flexion. The trunk flexion coincided with an extended period of rapid changes in activation level of the ES muscle suggesting that the muscle was activated in a lengthening contraction to brake the trunk flexion movement. This effect was not seen during anterior platform perturbations. The results support the hypothesis that certain combinations of voluntary movement and postural perturbation elicit a conflict between motor commands simultaneously trying to access the same muscle group.
Evaluation of the Effect on Spinal Curvature of Various Lifting Regimens and Recovery Techniques—Evers CT One of the most important aspects of any biomechanical research is the ability to measure the attributes of the body structure under investigation. In the field of industrial ergonomics and biomechanics, the back is a major area of concern, as back injuries account for a large percentage of both the injuries and cost of work accidents. Researchers have determined that there is a relationship between heavy work loads and back pain. Analysis methods used in many of these studies are not practical for use in most settings due to their sophisticated and invasive nature. Changes in the spine are considered a useful indicator for evaluating the effects of loading on the body. One method currently in use, the stadiometer, provides inferential data regarding spinal geometry by measuring stature. Another method directly measures the contour of the back to approximate the curvature of the spine. The proposed research will utilize the latter technique to accomplish the purpose of the research, which is to evaluate the response of spinal curvature to different combinations of workload and recovery method. This is a new area of research in industrial ergonomics. Findings of this research will expand our understanding of the spinal mechanics involved in lifting tasks, and may be used to develop recommendations for reducing the accumulated stress in the back as a result of workloads. The actual data collection will be done in June-July, 1997. In one test run using a 25 pound lift twice a minute in a task complying with the 1991 NIOSH Lifting Guidelines, an increase in the lumbar curvature in excess of 1 cm was observed after two hours. Following a 30 minute recovery period, this shift had disappeared and the lumbar arch was at its prelifting configuration. In the same trial, a decrease in spinal height of 2.2 mm was observed after the two hour lifting period. Following the recovery period, half of that decrease had been recovered. It is expected that similar results will be observed in the actual study.
EMG Analysis of Fatigue During Repetitive Work—Roy S., Bonato P, Knaflitz M
Over the past decade, the analysis of the surface electromyographic (SEMG) signal has been shown to be effective in objectively quantifying muscle impairments associated with sustained work. Application of the technique has been particularly successful for low back pain syndromes. The approach is based on indices of muscle fatigue derived by measuring the compression of the frequency content of the SEMG signal towards lower frequencies as the muscle accumulates metabolites. A primary limitation to this technique is that the quantification procedure for spectral analysis assumes signal stationarity. This precondition can only be satisfied by recording the SEMG signal during isometric, constant-force contractions. This is a serious limitation to the technique since many dynamic activities are commonly associated with work-related injuries. Fortunately, recent developments in the field of signal processing have produced methods of time-frequency (TF) analysis that are able to extract spectral information from nonstationary signals. Preliminary work has demonstrated that specific transforms belonging to the Cohen Class appear robust enough to derive spectral indices of fatigue from non-stationary SEMG signals recorded during cyclical dynamic contractions. In this paper we present the use of this procedure to measure localized paraspinal muscle fatigue during the repetitive lifting and lowering of a box. Time-dependent changes in SEMG spectral parameters, indicating the distribution of fatigue among superficial paraspinal muscles, were measured and com pared to the results obtained during a static isometric task at the same load.
The weight of the box, duration of the exercise, and frequency of the lift were specified according to safety guidelines of the National Institute for Occupational Safety and Health. Data were collected from eight subjects (five males and three females; mean age 26 ± 4 yrs) and analyzed by considering the TF representation of SEMG bursts sampled during the exercise. The contour plots of the TF distribution derived from two SEMG bursts recorded at the beginning (n = 1 s) and the end (n=5 min) of the lifting exercise clearly demonstrated the compression of the frequency components toward the lower end of the spectrum, thereby indicating the presence of fatigue. The relative amount of fatigue at the different electrode sites during the static vs. dynamic tests were qualitatively different indicating that fatigue was task-dependent and that we cannot generalize results obtained during static conditions to dynamic conditions. Furthermore, we observed that the SEMG signal recorded during dynamic contractions was also affected by factors not related to muscle fatigue. Namely, within each SEMG burst, changes in the exerted force, muscle length, and displacement between the muscle active fibers and the electrodes result in variations of the frequency content of the SEMG signal. As a result, we identified two separate nonstationary components in the SEMG signal collected during the dynamic exercise: 1) a within-burst nonstationarity related mainly to the biomechanics of the exercise (namely to variations of the exerted force as well as to the movement of the body segments which causes changes of the muscle length and a displacement between the active muscle fibers and the electrodes); and 2) a between-burst nonstationarity related to the gradual compression of the spectrum to lower frequencies as metabolites accumulate at the sarcolemma. These observations suggested that future studies focus on studying the effects of fatigue during cyclic dynamic tasks that have been implicated in repetitive use disorders.
The Human Costs of Occupational Injuries ——Strunin L
This study uses ethnographic interviews to increase understanding of the experiences of workers with back injuries, including the impact on their relationship to their pre-injury job and employer, their post-injury employment, their interactions with the workers' compensation system, medical care providers, and lawyers, and their perceptions of fairness in the process. The study is being carried out in two states, Wisconsin and Florida, among 400 workers of different racial and/or ethnic backgrounds from different industries and occupations, and different workers' compensation systems. The interviews allow us to learn how workers perceive their injuries and how they experience components of the occupational injury systems that ultimately affect both the personal and social costs of occupational injuries. The data presented will focus on respondent-identified constructs about impacts of the injury on employment, family, and related health issues.
Integrating Results from Quantitative and Qualitative Studies of Workers with Soft Tissue Injuries—Miller T, Waehrer G, Leigh P
This paper describes the use of qualitative research to understand the effect of soft tissue injury on workers. We note the importance of relations with the workplace in the experience of disability. We show how questions developed based on the qualitative results were used in a longitudinal study and proved to be important prognostic indicators for the injured workers. Qualitative research, although common in the social sciences, is rarely used in epidemiology. Fundamental differences in the underlying research paradigms, methods of inquiry, and rules of evidence in qualitative research and epidemiology pose challenges to the integration of these two genres of research. However, these same differences make the combination of methodologies a particularly powerful approach to the investigation of occupational injury problems, especially when applied to those with strong social or psychosocial components.
This is illustrated through an examination of the application of findings from a qualitative study of the experience of work-related back problems to an epidemiologic study of occupational musculoskeletal injuries. Injured workers' and key informants' verbatim interview data were analyzed using methods of grounded theory. Critical social dimensions of their experience were identified which appeared to influence recovery and return to work; in particular, problems of legitimacy and vulnerability in the workplace appeared central to the workers' experiences of injury and disability. Measurement constructs were subsequently developed and a series of closed-ended questions devised for inclusion in two prognostic cohort studies to enable the statistical testing of hypotheses which emerged from the qualitative research data. The questions were used in the Early Claimant Cohort Study, a prognostic cohort study of over 1800 injured workers. Workers were interviewed when they submitted Workers' Compensation claims, and at several other times for up to one year after injury. The time on benefits for up to one year after injury was determined from records. Using factor analysis we have confirmed the existence of two independent constructs: legitimacy and job vulnerability. The prognostic significance of these variables and their demographic correlates have also been examined. The results of these analyses indicate that the constructs are important in prognosis, but also highlight the need for more questionnaire development in this area.
Occupational Injury Costs Per Employee: Pinpointing the Risks —Shannon H
Objectives: 1. To estimate the medical and work loss costs of lost-workday occupational injuries reported to the Bureau of Labor Statistics (BLS). 2. To determine which occupations, industries, sources, events, age groups, and gender are associated with the highest costs per employee.
Methods: Work Loss Costs. BLS annual survey data show work days lost through a fixed date. Therefore, durations are censored for some cases. By major injury grouping, we built and applied non-linear regression models to estimate the full duration for censored cases in the 1993 annual survey. This was a massive modelling effort. It corrected for heterogeneity in the data and accounted for the existence of permanently and totally disabling injuries.
Once work-loss durations were available for all cases, we developed algorithms to compute lost wages. One approach used wages by occupation, industry, sex, and age group from the Current Population Survey. A problem with this approach is that an executive's injury can be weighted much more heavily than a production worker's, obscuring where the injury problems lie. A second, more egalitarian approach used an average daily wage loss. The second approach facilitates injury risk comparisons between groups, but does not accurately depict employer or societal costs.
The work loss costs were supplemented by fringe benefit costs and by household work loss costs. Household work loss was estimated from work loss duration, data showing workers typically return to housework 10% sooner than wage work (but possibly trading for less demanding tasks), and studies of the value of housework.
Medical Costs. Medical costs by diagnosis were derived separately by injury diagnosis for hospital-admitted and non-admitted cases. Diagnosis-specific costs for admitted cases came from national average lengths of stay for cases covered by Workers Compensation (from 1987-1992 National Hospital Discharge Survey data) and costs per hospital day from states where cost control regulatory agencies force hospitals to accurately report these costs. Post-discharge costs in the acute care phase came from 1987 National Medical Expenditure Survey (NMES) data. Longer-term costs came National Council on Compensation Insurance Detailed Claims Information (DCI) data.
For non-admitted cases, costs per visit came from Civilian Health and Medical Program of the Uniformed Services (CHAMPUS) data. NMES described short-term visits per case and DCI described long-term costs.
1987-1992 National Health Interview Survey data were used to compute the number of medically treated cases without work loss from the BLS lost workday case counts. Data scarcity forced us to perform these computations for broad diagnosis groupings.
Results. The analysis is in process. The data set lets us rank costs per employee by occupation, industry, source (e.g., a machine), event (e.g., a fall), body region, diagnosis, age group, and gender.
Encouraging worker participation in safety programs: A selected review of research findings—Peters RH
Various strategies have been used to encourage employee participation in safety programs. The intent of this presentation is to provide guidance to safety professionals concerning the use of three techniques for encouraging employees to participate in their safety program: (1) employee surveys, (2) incentives, and (3) fear communications. Each strategy is briefly described and a summary of the empirical evidence concerning its effectiveness and limitations is given. Some recommendations are made concerning the use of these strategies.
Management practices affecting safety: An overview of coal industry research findings—Peters RH
This presentation focuses primarily on summarizing the past 20 years of research on management practices affecting underground coal miners' safety. Three factors have consistently been found to account for differences in mine safety performance across multiple research studies involving different samples of miners, different research methods, and different researchers. The factors are: (1) the extent to which workers perceive that upper management is concerned about their welfare; (2) the extent to which management actively involves the work force in identifying safety problems and defining solutions; and (3) the favorableness of management-labor relations. The findings and implications of this research are discussed.
Reducing Accident Rates with Organizational Performance Management— Saarela KL
Safety professionals know a lot of safety and health, but do they know enough on how to achieve a change in practice. This difficult question is considered here on the basis of two intervention studies which were carried out in the shipbuilding industry and in the metal product industry. The goal of both projects was to launch an improvement process leading to a better work environment and safety, as well as to more efficient production. The cooperation partners represented industries in which the accident rate is above the national average in Finland. The action research approach combining research and practice and organizational performance management was utilized in the studies. The project at the shipyard was started in three departments at the beginning of 1995. In 1996 when the project was reported, it involved more than 10 departments (over 900 employees). The project in the metal product factory involved one department (about 200 employees) and was carried out in 1996. In the beginning of the projects, a questionnaire survey provided the personnel the opportunity to participate, to report problems they had identified, and to make suggestions for improvements. Local small groups with managers, supervisors and workers as members arranged regular meetings and followed a systematic model consisting of identification of problems, setting goals, solving problems, implementing changes, monitoring the results and providing feedback. In connection with the projects, everybody working in the intervention area participated in a two-hour training and development seminar. The questionnaire survey was repeated in order to get evaluative information. The effects of the interventions on accident rates were also investigated. The overall accident rate of the shipyard decreased by 56 % during the two year period (the goal was 25 %). According to the questionnaire survey, almost all the workers in the departments with the best results agreed that housekeeping had improved, work was progressing better, hazardous situations had decreased, cooperation had improved, new working practices had been learnt, and job satisfaction had improved. Over 70 % of the workers in these departments reported that it was easy to achieve improvements in their department; this reflected the innovative organization culture. Despite some differences between the departments, some improvements had been achieved in all departments participating in the project. In the metal product factory, the accident rate was reduced by 45 % during the one year period. In addition, improvement was noted, e.g. in delivery reliability, throughput time, productivity and profit. The department was interested in continuing with more demanding development tasks, and a new project was started this year in cooperation with the Finnish Institute of Occupational Health.
The Relationship Between Safety Climate and Injury/ Exposure in an Acute Care Hospital—Gershon RRM, Karkashian C, Martin LS, Grosch J, Murphy L
Introduction. We have previously identified organizational commitment to safety (i.e., “safety climate”) as an important correlate to safe work practices within the hospital workplace. To further advance our understanding of the complex relationship between safety climate and safe work practices and to clarify the role safety climate plays with respect to workplace injuries, such as needlestick injuries, we conducted a cross-sectional survey of employees at a large, 1000 bed, tertiary care hospital.
Methods. A questionnaire was designed to assess and characterize specific aspects of safety climate and to determine the relationship, if any, between these aspects and injury rates among hospital-based health care workers. The resulting five-page questionnaire was psychometrically analyzed and validated. Employee perception of safety climate was measured using a fifty-item safety climate scale, which factored into three distinct constructs: (1) facility-wide safety climate, (2) departmental-based safety climate, and (3) environmental safety climate. The questionnaire also contained items on injuries employees had experienced within the previous six months and these were further subdivided into categories of injuries and exposures, including exposures to bloodborne pathogens. Employees' compliance with safe work practices was measured using a 13 item compliance scale, and the questionnaires also included standard sociodemographic items. Most responses were based upon a five-point Likert scale (from strongly disagree to strongly agree). The confidential self-administered questionnaires were sent to a stratified random sample of clinical health care workers (i.e., those with direct patient or patient specimen contact) employed at a large, regional medical center.
Results. Responses were obtained from over 750 employees (60% response rate). Eighty-five percent of the respondents were female, the median age was 37 years, and the median tenure was 4.5 years. A total of 186 injuries (which occurred within the previous six months) were reported by 116 respondents. The respondents also reported 330 bloodborne pathogens exposures; 53 respondents reported a total of 74 needlesticks, 81 respondents reported 97 splashes to the mucosa, 52 respondents reported 81 cuts with sharp objects and 27 respondents reported 107 contacts with open wounds. Ninety-three (42%) of these exposures involved blood from a patient known to be infected with the human immunodeficiency virus and/or hepatitis B virus. Injuries were found to be highly correlated with two out of the three safety climate constructs. For example, low rates of injuries/exposures were correlated with strong facility-wide safety climate (p<.01) and departmental safety climate (p<.001). Employees who perceived a strong safety climate were significantly less likely to report workplace injuries. Safety climate was also significantly associated with high levels of self-reported compliance with safe work practices, and this was the case for all three safety climate constructs (p<.001).
Conclusions. Two safety climate constructs (facility-wide and departmental- based) were found to be significantly correlated with injuries and exposures; employees who perceived a strong safety climate at work were significantly less likely to report workplace injuries and exposures. Since these data are cross-sectional, we cannot determine causality—i.e., employees with fewer injuries may perceive their workplace to be safer, and this can only be determined from prospective studies. Nevertheless, these results inform us and help to identify the important determinants of safety climate. This will help us to appropriately focus our resources in our efforts to minimize the risk of injury/exposure among hospital-based health care workers.
Some Methodological Concerns in the Development of Predictive Models: Examples From Discriminant Analysis—Clancy EA
Multivariate parametric modeling is frequently used to develop predictive relationships between exposures (inputs) and the risk/ occurrence of injury and illness (outcomes). Accurate predictive models can be used to suggest interventions which can minimize illness and injury. Perhaps due to the high cost of measuring particular exposures or to the sparsity of many adverse outcomes (most adverse outcomes, including certain traumatic injuries, are relatively rare events for the individual employment establishment), modeling studies in the health and safety literature frequently lack an independent test sample for evaluating predictive model performance. Rather, these studies report only the resubstitution accuracy--the accuracy which is realized when the model is evaluated on the same sample that was used to generate the model coefficients. Unfortunately, it is well established that the resubstitution accuracy is optimistically biased, that is, it typically provides an inflated estimate of the predictive accuracy of the developed model. For small data sets (small relative to the number of recorded exposures and/or exposures included in the model), this optimistic bias can be substantial. Furthermore, small data sets may lead to selection of an entirely spurious set of exposures.
To elucidate this issue, a Monte Carlo simulation study was conducted using the classification modeling technique of discriminant analysis. Random data containing no true classification power (denoted the “Nil Model”) were generated, then analyzed using discriminant analysis. For the case of two outcome groups, the true accuracy of the Nil Model is 50% (i.e., no better than flipping a fair coin). For conditions similar to those in the literature, the random data “reported” highly accurate classification performance--results as high as 100%. These “reports” represent the bias artifact of resubstitution accuracy. Factors influencing the extent of the bias were studied. It was found that the resubstitution bias is reduced if: sample size is increased, the number of candidate exposures is decreased, the number of selected exposures is decreased, and the proportion of samples from each outcome group is equalized. These simulation studies indicate that reporting of the resubstitution accuracy alone can be problematic. The resubstitution accuracy can be made arbitrarily large, regardless of the true predictive accuracy of the model.
The most common approach to rectifying this situation is the use of a train-test methodology in which the collected subject data are separated into non-overlapping, independent training and testing sets. The model is trained (i.e., the model coefficients are computed) on the training set, then tested on the test set. The performance achieved on an adequately sized test set is considered a good estimate of the true predictive model performance. It is suggested that all research reports which develop parametric models should either (1) train the model on one data set, but report as the performance metric the accuracy achieved on an independent, adequately-sized test data set, or (2) demonstrate that the magnitude of the resubstitution bias is minimal.
Quantifying and Aiding Laymen Understanding of Risks—Cohen BL
The average layman has great difficulty in understanding the difference between a risk of 0.0001 and 0.000001, but expressing these in terms of loss-of-life-expectancy, LLE, makes it much easier. The difference between LLE = 1 day or 100 days is relevant to the layman's everyday experience and is therefore very understandable. The LLE approach has the weakness of ignoring non-fatal injuries, but it has a great advantage for quantitative treatment— dead bodies are easy to identify and count, whereas morbidity has a wide spectrum of seriousness and is subject to wide variations in perception of suffering. It may be useful to introduce a multiplier on LLE to account for morbidity, since this multiplier would normally be not much larger than unity, but it would be very difficult to reach consensus on these multipliers.
Numerous examples of LLE will be presented. Some of these LLE are many years, like alcoholism, poverty, smoking, poor social connections, and overweight, while others are less than one day, like carcinogens in peanut butter or charcoal broiling, or living near a nuclear power plant, but the public does not perceive this thousandfold difference—this is a problem in media coverage. There are also wide variations in LLE due to occupational risks between different industries, and between different subgroups of workers within industries; many examples will be given. Occupational diseases are generally much more important contributors to LLE than accidents. Some approaches to estimating LLE due to occupational diseases will be discussed.
Burden Due to Occupational Hazard—Murray CJL, Acharya AK
The Global Burden of Disease Study was initiated in 1992 at the request the World Bank and has been undertaken with full collaboration and participation of the World Health Organization. Preliminary results were used in the World Bank's World Development Report 1993: Investing in Health (1) and published by WHO in 1994 (2). The GBD was designed to address three primary goals: to infuse information on non-fatal health outcomes into debates on international health policy, which are all too often focused on mortality; to develop epidemiological assessments for major health conditions unbiased by advocacy groups; and to quantify the burden of disease using a measure that could also be used for cost-effectiveness analysis. Four specific objectives were established: (i) to develop internally consistent estimates of mortality for 107 major causes of death by age and sex for the world, divided into eight geographic regions; (ii) to develop internally consistent estimates of the incidence, prevalence, duration and case-fatality for 483 disabling sequelae of these 107 causes of disease and injury by age, sex and region; (iii) to estimate the fraction of mortality and disability attributable to ten major risk factors by age, sex and region; and (iv) to develop various projection scenarios of mortality and disability by cause, age, sex and region.
Deaths from occupational injuries and occupational diseases were evaluated as part of the risk factor assessment of the GBD. The GBD study results were based on inferences drawn from data from reporting systems in the United States, Canada, Australia, Sweden, Denmark, the United Kingdom, Switzerland, Luxembourg, Hungary, Mexico and China (selected causes only). Occupational diseases accounted for 1.0 million deaths and there were an estimated 100 thousand occupational injury deaths in 1990. The distribution of deaths by region is shown in the following table. In this study, we have also estimated the disability attributable to occupational injury. The combination of premature mortality and disability is measured in terms of Disability-Adjusted Life Years (DALYs).
| Region | Deaths | DALYs |
| EME | 14 | 540 |
| FSE | 10 | 405 |
| IND | 15 | 655 |
| CHI | 20 | 1,039 |
| OAI | 12 | 566 |
| SSA | 10 | 550 |
| LAC | 9 | 429 |
| MEC | 9 | 461 |
| World | 99 | 4,645 |
A United States burden of disease study which will include an analysis of occupational injuries and diseases is underway.
Construction Injury Rates May Exceed National Estimates: Evidence from the Construction of Denver International Airport (DIA)—Glazner JE, Borgerding JA, Lowery JT, Bondy J, Kreiss K
Aim. To describe occupational injury rates and workers' compensation (WC) payment rates on a major construction project.
Background. Most injury rate estimates rely on survey data (Bureau of Labor Statistics (BLS)), analysis of WC claims, emergency room surveillance and proportionate mortality studies. These may suffer from underreporting and/or imprecise estimates of the number of workers at risk. Construction of DIA, with 31 million person hours on 2,843 contracts held by 769 contractors, provided a unique opportunity to describe the magnitude of injury on a major construction project.
Methods. The airport owner implemented a project-wide WC insurance plan for all on-site workers involved in airport construction. All claims were recorded in a centralized database, as was payroll according to job classification. Injuries were treated at an on-site clinic operated by a designated medical provider; claims were generated whenever treatment was rendered. Using a computer file with all claims, payroll by contract (allowing calculation of hours at risk), and company characteristics, we calculated injury and payment rates by construction sector, company size, and year.
Results. DIA's total injury rates were over twice BLS's rates for the construction industry for each year of construction. Lost-work-time (LWT) injury rates were more comparable, but DIA's were slightly higher. The order of injury rates by SIC code was the same in both data sets, with SIC 17 (special trades) having the highest rates. Total injury rates for DIA's small contractors (1-19 employees) were three times BLS's reported rates and at least twice BLS's rates for all other company size categories. Injury rates declined significantly after the first year of DIA construction, co-incident with a several-fold increase in project safety personnel. DIA's total WC payment rate of $7.06 per $100 payroll was 11 percent higher than expected loss rates reported by the National Council on Compensation Insurance, in contrast to the two-fold differential between injury occurrence at DIA and BLS survey data.
Conclusion. Complete reporting, facilitated by the existence of a single WC plan, an on-site medical clinic and designated medical providers, produced injury rates significantly higher than previously reported. The relatively small difference between payment rates and expected loss rates suggests that underreporting is concentrated in the area of minor injuries, which we found to be associated at the contract level with major injuries. Our results suggest that underreporting of WC claims occurs and may result in sizable underestimates of the burden of injury in construction.
Risk Factors for Injury Among Construction Workers on the Denver International Airport (DIA) Project—Lowery JT, Borgerding JA, Zhen B, Glazner JE, Bondy J, Kreiss K
Aim. To identify risk factors for work-related injury on the Denver International Airport construction project, which generated over 31 million person-hours, employing 32,000 workers from 769 companies to complete 2,843 contracts.
Methods. We obtained workers' compensation claims data from an administrative database established under the project's owner-controlled insurance plan, which covered all on-site employees and provided on-site medical care through designated providers. We calculated injury rates by contract and over contract strata of interest by linking claims data with employee demographic information, company and contract characteristics, and contract payroll. We determined that injury rates among contracts held by the same company were independent using generalized estimating equations, enabling us to examine contract-specific factors in relation to total injuries, lost-work-time (LWT) injuries, and non-LWT injuries in Poisson regression models. To control for inherent risk of work in the model, we included expected loss rates (ELRs), which we calculated for contracts using Colorado-specific ELRs provided by the National Council on Compensation Insurance for each job classification. We used logistic regression to determine the association between LWT and non-LWT injuries on a contract level, controlling for person-hours at risk and ELRs.
Results. Injury rates were highest during the first year of construction, at the beginning of contracts and among older workers. Risk for total and non-LWT injuries was elevated for building construction contracts, contracts for special trades companies (SIC 17), contracts with payroll over $1 million and those with overtime payroll greater than 20 percent. Risk for LWT injuries, on the other hand, was increased for site development contracts and contracts starting in the first year of construction. Large companies (250+ employees) had significantly lower risk for all injuries.
Contracts experiencing one or more minor injuries were four times as likely to have at least one major injury [OR=4.0, 95%CI(2.9,5.5)].
Conclusion. Our finding of increased risk of LWT injury for contracts starting in the first year of the project suggests that enhancement of the project's safety infrastructure during the second year of construction was effective in reducing serious work-site injuries. The absence of correlation between injury rates among contracts belonging to the same company suggests that targeting of safety resources at the level of the contract may be an effective approach to injury prevention. Interventions focused on contracts with considerable overtime work, contracts of special trades companies (SIC 17), and those belonging to small and mid-sized companies may yield reductions in injury rates. Furthermore, efforts to provide adequate site-specific training to workers new to a construction site or new to a contract may also reduce injury burden on large construction sites. The joint occurrence of minor and major injuries on a contract level suggests that surveillance of minor injuries may be useful in identifying opportunities for prevention of major injuries.
Trade-specific Injury Rates at a Large Construction Project— Welch LS, Hunting K, Aleali R
Construction as an industry has a high injury rate, as documented in annual BLS surveys and other data sources. CFOI contains trade-specific data that allows a determination of rates for fatal injuries for specific trades, but trade-specific data is not readily available for non-fatal injuries. Surveillance data is collected in order to design, implement and evaluate prevention programs. Trade-specific surveillance data for construction workers is necessary if we are to understand the causes of injuries, and find ways to prevent them.
We have monitored injuries from a large construction project from 1990-1997. All injuries were treated at the George Washington University emergency department, and data on each injury was abstracted. We have obtained hours worked through the project by contractor and developed trade specific rates for this project. We will present these trade specific rates, discuss how these data differ from BLS data, and compare our rates to trade specific data collected on other large construction sites. We will also present a model for implementing trade specific surveillance on other construction sites.
Occupational Injury and Fatality Patterns of Females Employed in the Construction Industry: Kentucky, 1991-95—Mays J, Kidd P, Struttman T
Purpose. In Kentucky, almost 10% (n=7250) of the construction industry workforce is female. Recent studies have identified risks in the construction industry that are particular to females. However, epidemiologic studies of females employed in the construction industry are scarce. This study provides important epidemiologic information useful in targeting interventions for females employed in the construction industry.
Research Data. This investigation used data from the Kentucky Department of Workers Claims from 1991-95 to describe both the injury and fatality patterns of females employed in the construction industry in Kentucky. The data set contained information on sector of employment, age, gender, number of months employed, time on that particular job, body part injured, nature, extent, and cause of injury, and days of work lost.
Methods. The data were obtained electronically from the Kentucky Department of Workers Claims. Injury and fatality patterns of females are described and compared across several key stratification variables. Where appropriate, injury and fatality patterns of females are compared to males to determine differences in the characteristics of the injury or fatality. Analysis of variance and chi-square tests are used to test for statistical significance.
Results. Seven hundred and sixty-eight claims for females employed in the construction industry were filed with the Commonwealth of Kentucky's Department of Worker's Claims from 1991-1995. Less than 1 percent (n=4) of the 768 claims were fatalities, 0.1% (n=1) was a permanent total disability claim, 1.0% (n=8) was a permanent partial disability claim, 52.3% (n=402) were temporary disability claims, 43.6% (n=335) were lost time claims, and 2.3% (n=18) were no lost time claims. The average number of days lost for females was 19.8 and did not differ significantly from males (18.0 days). The body region injured differed significantly between females and males (p<0.001) with females having a higher proportion of injuries to multiple body parts than males (17.8% vs. 10.3). Regarding nature of injury, females had a significantly higher proportion than males of contusions (11.8% versus 7.4%, p<0.001), carpal tunnel syndrome (2.3% versus 0.3%, p<0.001), and “all other cumulative injuries” (12.8% versus 1.7%, p<0.001). Most of the claims filed for females were in the special trades sector (43.0%) versus heavy construction (39.8%) or general contracting (17.2%). However, while 39.8% of claims filed by females were in heavy construction, only 23.9% of claims for males were in this sector (p<0.001). Specifically, 35.6% of all claims filed for females were for those employed in highway, street, and bridge construction (SIC = 1611, 1622) compared to only 17.7% for males (p<0.001). In addition, a significantly higher proportion of claims for females versus males reported trucks, cars, highway vehicles, or earth moving vehicles involved in road construction as the source of the injury (10.0% versus 6.3%, p<0.001).
Conclusion. These findings suggest that interventions designed to reduce the number and severity of occupational injuries and fatalities of females employed in the construction industry should be focused on the heavy construction industry, particularly road, highway, and bridge construction.
Injuries Among N.C. Residential Construction Workers, 1986- 1994—Dement JM, Lipscomb H
All computerized workers' compensation claims for 7400 N.C. Homebuilders Association (NCHA) members and their subcontractors for the period 1986-1994 were obtained from their compensation insurance carrier. Average annual incidence density rates (cases/200,000 work hours) were calculated for all claims, claims involving medical costs or paid lost work time and by mechanism, nature and body part injured. Incidence density rates were calculated separately for each residential construction craft.
A total of 31,133 worker's compensation claims were filed by homebuilders during 1986-1994. Of these claims, 10,680 claims were for minor medical costs (less than $2000) without lost work time beyond the date of injury. A total of 7199 claims involved paid lost work time which begins after the seventh day away from work in North Carolina. An additional 3543 claims involving medical costs of $2000 or less and no paid lost time occurred. For the seven years studied, the following average rates of filing claims by claim type were observed.
| Overall | Paid Lost Work Day | Medical or Lost Time |
| 16.40 | 3.79 | 10.78 |
Twenty-six work related deaths occurred among residential construction workers during the seven years studied. Vehicle accidents (n=6) were observed to be the major known cause followed by falls (n=3), being struck by an object (n=3), electric shock (n=2), and contact with energy or chemicals (n=2).
Highest rates for cases involving medical costs or paid lost time by mechanism of injury were observed for being struck by an object (3.1), lifting/movement (1.97), falls from a different level (1.13), striking against an object (0.87) and falls on the same level (0.46). Rates by mechanism of injury were highest for muscle strains (2.34), wounds/punctures (2.33), bruises/contusions (1.24), fractures/dislocations (0.98) and injuries to the eyes (0.81). Among medical cost or lost work time cases, body parts with highest injury rates(cases/200,000 hours) were back/shoulders (1.99), fingers (1.31), leg/knee (1.00), hand/wrist (1.00), foot/ankle (0.86) and eyes (0.82).
Injury rates were found to vary substantially among the residential construction trades. For more serious injuries involving medical costs or lost time, rates were highest for welders and cutters (28.1), insulators (24.3), roofers (19.4), and carpenters (15.3). The same general trends by trade were observed for cases involving paid lost time except that roofers were highest with a rate of 9.1 cases/200,000 hours followed by insulators (8.5), welders and cutters (5.8) and carpenters (5.8).
Rates of falls from a different level resulting in medical costs or lost time were highest for roofers (5.54), insulators (3.53), carpenters (2.05) and drywall installers (1.99). Falls from a different level resulted in a fracture/dislocation 36.8% of the time followed by bruises/contusions (22.7%), muscle strains (21.5%) and joint sprains (10.3%). Multiple serious injuries occurred with 1.3% of the falls from a different level. Body parts most frequently injured during due to falls from a different level were back/shoulders (25.0%), foot/ankle (18.5%), leg/knee (16.4%) and arm/elbow (11.1%).
Descriptive information for falls from a different level resulting in paid lost time during 1993-94 (N=219) were reviewed in order to better determine the causes and circumstances of injuries. Falls from a roof accounted for 25.4% of the cases followed by falls involving scaffolds (23.9%), ladders (20.6%) and falls from ceiling joists, floor joists or framing (14.8%).
A Review of Methods for Estimating the Amplitude of the Surface Electromyogram (EMG): A Window into Assessing Muscular Effort in Musculoskeletal Injuries—Clancy EA
The amplitude of the surface EMG provides a measure of muscular effort and has also been investigated as an indicator of muscle force. Applications which utilize EMG amplitude in the study of traumatic musculoskeletal injuries include investigations into the mechanisms of injuries (e.g., muscular activation patterns experienced during slipping, tripping and falling), studies of muscular exposures associated with injuries (e.g., muscular efforts and tensions coincident with lifting/lowering heavy objects), and physical therapy/rehabilitation (e.g., clinical assessment of muscular function). This presentation reviews typical methods used to estimate the EMG amplitude from the EMG waveform and describes recent/ developing advances in EMG processing techniques.
Early investigators treated the EMG waveform as an amplitude modulated signal. The original amplitude estimator consisted of a full-wave rectifier (demodulator) followed by a resistor-capacitor (RC) low pass filter (smoother). Empirically, it was then found that the signal to noise ratio (SNR) of a third-order averaging filter provided a 44% improvement, and that a second-power demodulator was best. Mathematical models, representing EMG as band-limited Gaussian noise, found that second-power demodulation and averaging, i.e. RMS processing, gives optimal amplitude estimation. Typical amplitude estimators in use today utilize one of the above processors, with RMS processing preferred.
Several investigators found that inclusion of a whitening filter (a filter whose output power spectrum is constant-valued when presented with the input signal) prior to demodulation improved estimator performance. Auto-regressive modeling of the EMG power spectrum was used to form whitening filters which doubled the probability of differentiating between four contraction levels. For contractions above 10% MVC, similar whitening filters improved the SNR by 63%. Whitening can also be achieved by reducing the outer edge spacing of a pair of rectangular bipolar electrodes. A few authors have found that the shape of the whitening filter should adapt as a function of the contraction level.
Dispersing multiple electrodes about a muscle may provide a broader, more complete, measure of the underlying electrophysiologic activity. Using four electrodes, an SNR improvement of 91% has been achieved. Four electrodes have been used to improve the probability of differentiating between four contraction levels by 4070%. The combination of four electrodes and whitening via electrode geometry yielded a 176% SNR improvement. Eight whitened (auto-regressive technique), combined electrodes provided a 309% SNR improvement.
When the EMG amplitude varies throughout contraction, improved amplitude estimates can be achieved if the smoothing window length is tuned throughout the contraction. Adaptive window length processors have been implemented based on the EMG amplitude and its first derivative. When contraction levels changed rapidly or slowly, marked estimator performance improvement resulted.
Future EMG amplitude estimators should incorporate all of these improvements--adaptive whitening, multiple electrode combination and adaptive smoothing window length--into a robust processor. The improved amplitude estimator performance which results should provide more accurate assessment of muscular effort, muscular activation patterns, muscular tension and other related muscular function indicators of interest to occupational safety and health.
Assessment of a Cartographer's Workstation—Pentikis J
Office ergonomics has become an important workplace issue, however the problem of using a mouse almost exclusively to operate software has not received as much attention as other facets of office ergonomics, i.e. lighting or keyboards.
Cartographers working for the U.S. Army spend their entire workday in a seated position performing intensive input to their computers using a hand “mouse.” The nature of the work has resulted in upper extremity injuries as well as a high rate of upper extremity and back discomfort levels.
The problems faced by cartographers is threefold. First, the mouses provided to the cartographers placed their wrists in non-neutral postures and also placed mechanical stresses on the hand. Second, the tasks associated with the cartographers work activities require prolonged static postures of the upper extremity and back for the majority of the workday. Third, the workstations are not highly adjustable thereby forcing the cartographers to work in less than optimal postures.
Corrective actions included redesigning the mouse in order to place the wrists in a neutral posture, the mouse was shaped with rounded edges to reduce the mechanical stress placed on the hands, and the buttons on the mouse were rearranged and the force requirements lessened to avoid having the fingers stretch and use excessive force to operate the buttons on the mouse. Second, new chairs which are capable of being easily adjustable in height, seat pan angle, and back rest angle which also offer lumbar support are to be purchased. Third, the desks which are adjustable in height are to be replaced with desks that are adjustable in height and can also be tilted to move the desk surface closer to the cartographer. Finally, alternative tasks are being investigated to see if it is feasible to encourage the cartographers to move away from their seated workstation each hour.
Preventing Accidents Caused by Unrecognized Roof Beam Failures in Underground Stone Mines—Iannacchione AT, Prosser LJ, Oyler DC, Dolinar DR, Marshall TE, Compton CS
A safer environment for stone miners can be realized by enhancing the industry's ability to recognize and monitor hazardous ground conditions. This is necessary because 12 of the 14 underground stone miner fatalities in the last six years were caused by falls of roof or rib. Because the total underground miner population is less than 2,000, the fatality incident rate is greater than for miners working in underground coal mines. Despite society's demand for the resultant products of stone mining, i.e., highways, buildings, agriculture, water purification, etc., the public typically opposes the development of surface stone quarries. Thus underground stone mines are emerging as an alternative method for producing needed raw materials within the tolerance level of the general public.
One technique under development by NIOSH at the Pittsburgh Research Center aimed at reducing the safety risk to miners is a remote monitor to detect hazardous ground conditions. Existing mechanical roof monitoring devices installed in drill holes have experienced only limited use in underground stone mines. These monitors are generally single point extensometers which measure the overall separation of roof layers in the immediate roof. Hazardous levels of roof sag signal a high potential for roof failure. In some cases this information has been used to add roof support, remove roof rock, or danger off affected areas. Unfortunately, these instruments are difficult to read because of their location on the roof line, require a considerable number to accomplish full coverage, and require considerable experience to determine dangerous sag rates. A new generation Automated Roof Monitoring Safety System (ARMSS) developed by NIOSH improves on the existing methods for determining roof stability. This mechanical roof sag monitor includes the following features: 1) inexpensive (<$150), 2) multiple anchor points [a many as six], 3) adaptable to standard data acquisition systems, 4) easy to install, and 5) remote monitoring capabilitiy. At a recent field site test the following issues were examined: 1) what are critical sag rates?, 2) how much sag occurs prior to a roof fall?, 3) at what locations in the roof does the failure occur?, 4) how often should monitors be read?, 5) where should monitors be placed within the entry?, 6) when should monitors be used?, and 7) how should the monitor data be analyzed? This research will provide recommendations/guidelines on these issues so that this technology can be used effectively and efficiently in an effort to improve the safety conditions for underground stone miners.
SCOPE: System for the Classificaton of Operator Performance Events—Kapp EA
Human error as a cause of accidents in the workplace has been an area of concentrated interest among safety practitioners and researchers since Heinrich's first edition of Industrial Accident Prevention in 1931.
The literature is widely divergent as to what percentage of occupational accidents can be attributed to human error, with some claiming 80-90% of all accidents are due to human error (Salminen and Tallberg, 1996) while others report percentages around 50% (van der Schaaf, 1992).
Regardless of the exact figure, human error is widely accepted as a major source of or contributing factor to accidents and injuries in the workplace.
There have been numerous approaches to investigating of human error including: examination of the psychological characteristics of the employee, physical characteristics of the workplace, and management and organizational characteristics of the organization. Each approach has shed some light on the issue, but none alone has provided a complete and usable approach to analyzing human error in the workplace.
SCOPE (System for the Classification of Operator Performance Events) is a comprehensive tool for investigating workplace accidents caused by human errors. Synthesized from the research on Human Information Processing, Signal Detection Theory, Judgment and Decision Making, Organizational Design and Management, Work Physiology, and Ergonomics, SCOPE analyzes accidents to uncover the sources of the human error that caused the accident.
SCOPE uses a two phase approach for the analysis: Event Description, and Causal Factor Analysis.
Event Description involves the recording of all relevant facts concerning the accident. Causal Factor Analysis uses the event description to pinpoint the underlying workplace factors that produced the human error.
Causal Factor Analysis consists of three steps: (1) mapping of the cause and effect process that created the accident through creation of an Incident Production Tree; (2) classification of the operator errors involved in the accident through the Error Classification Flowchart; and (3) determination of the contributing factors in the workplace that produced the errors using the Workplace Factors Checklist.
Preliminary testing in the healthcare industry indicates that SCOPE has great promise for use in operational settings to determine the underlying causes of human errors in the workplace.
References: Salminen, S. and Tallberg, T. (1996). Human errors in fatal and serious occupational accidents in Finland. Ergonomics, 39(7). 980-988.
Van der Schaaf, T. (1992) Near Miss Reporting in the Chemical Process Industry. Eindhoven: Technical University of Eindhoven.
Concrete Shoring Systems Loads and Safe Design—Huston DR
This paper reports on a study that examines safety issues related to preventing the collapse of temporary support structures, called shoring, during the construction of buildings with reinforced concrete columns and floor slabs. The basic problem is that fresh concrete is quite weak and cannot support its own weight. As concrete cures, it grows stronger and is eventually able to carry its weight. It is during the pouring and curing stages that the structure is vulnerable to collapse and to cause injury and often death. Hadipriono and Wang (1986) reported on 85 such collapses that occurred during the previous 25 years. The main issue with preventing these collapses is to design and use the shoring systems to support the concrete properly. Unfortunately, there have been very few field measurements of the loads experienced by shores and the performance of shoring systems. The objective of this study is to measure the loads on shores as the buildings are being built and to use this information to evaluate the existing codes and to make recommendations for improved shoring design and use procedures. Loads have been measured in four buildings, three of which are low rise buildings of a couple stories each. The fourth building is a twenty-five story high-rise in Cambridge MA, known as the Museum Tower. The building is built with a new shoring system that allows rapid construction sequencing. Slab deflection measurements are taken on the building as well. The measured loads are compared with those predicted by computer models and by the ACI code. Preliminary results from the data indicate that there can be large variations in the loads on nominally identical shores, thermal effects are much larger than anticipated, and the average loads are close to those predicted by ACI. Certain recommendations are being formulated for the use of safer pouring sequences, the design loads for shores, and the use of electronic monitoring and hazardous condition alarm instrumentation.
Social and Economic Impact of Occupational Back and Upper Extremity Injuries: A Pilot study—Pransky G, Benjamin K, Hill-Fotouhi C
We obtained a random, stratified sample of 405 individuals who had occupational upper extremity or low back Injuries a year before a survey was mailed to them. The survey included information on demographics, employment, social and other factors, and multiple dimension of health-related quality of life, as well as economic and other consequences of Injuries. The survey was the product of extensive validation and reliability testing, and incorporated previously used questions as well as several new scales. We attained a follow-up rate of over 65% of those where addresses could be verified, with only 8% refusing to participate.
Results. Length of time out of work was directly associated with a number of adverse economic, functional, and psychological consequences. Employer accommodations were important in achieving early return to work, and were actually more strongly associated with prevention of reinjury than job modifications that were specifically designed to prevent recurrence. Although less than 10% never returned to work, persistent functional and symptomatic problems due to the injury were still present in most respondents. Time loss was also associated with negative employer responses. We conclude that immediate employer responses to work-related Injuries should include effective prevention activities, but that the nature of preventive activities should be examined more closely, as safety rules and other changes may not be effective in preventing reinjury. Work-related Injuries in a typical population are associated with significant burden in terms of lost wages, psychological and social disruption, and persistent symptoms and functional deficits, despite returns to work.
National Costs of Occupational Injuries in 1992—Leigh JP
Objective. To estimate the annual incidence, the mortality and the direct and indirect costs associated with occupational injuries in the United States in 1992.
Design. Aggregation and analysis of national and large regional data sets collected by the Bureau of Labor Statistics, the National Council on Compensation Insurance, the National Center for Health Statistics, the Health Care Financing Agency, and other governmental bureaus and private firms.
Methods. To assess incidence and mortality of occupational injuries and illnesses, we reviewed data from national surveys. To assess costs we used the human capital method that decomposes costs into direct categories such as medical and insurance administration expenses as well as indirect categories such as lost earnings, lost home production and lost fringe benefits. Some cost estimates were drawn from the literature while others were generated within this study. Total costs were calculated by multiplying average costs by the number of injuries in each diagnostic category.
Results. Approximately 6,500 job-related injury deaths, and 13.2 million non-fatal injuries are estimated to occur annually in the civilian American workforce. The total direct ($49 billion) plus indirect ($96 billion) costs were estimated to be $145 billion. Of the $49 billion in direct costs, $25.1 billion was spent on doctors, hospitals, drugs, nursing homes, and rehabilitation providers; $5.7 and $8.9 billion went to cover medical and indemnity insurance administration expenses; $8.7 billion covered property damage; and $0.8 billion paid for police and fire services. The $96.2 billion of indirect costs can also be disaggregated: $68 billion in wage losses; $14 billion in fringe benefits; $8 billion in home production losses; $5.2 billion for workplace training; and $0.3 for time delays. The estimates are likely to be low, because they ignore costs associated with pain and suffering as well as those of within-home care provided by family members, and because the numbers of occupational injuries are likely to be undercounted.
Conclusion. The costs of occupational injuries are high, in sharp contrast to the limited public attention and societal resources devoted to their prevention and amelioration. Occupational injuries are an insufficiently appreciated contributor to the total burden of health care costs in the United States.
Return to Work Following Traumatic Injury: A Prospective Study of the Effect of Trauma on Patients' Working Lives—Petersen JS, Morabito D, Boccellari A, Mackersie RC, Ochitill H
Victims of acute traumatic injuries are displaced from the labor force by an event exogenous to their usual occupational responsibilities. This displacement represents economic costs to individuals, their families, and society because of the temporary or permanent loss of productive activity. To date, little is known about the impact of non-disabling injury on an individual's economic status. The purpose of this study was to estimate the economic impact of moderate traumatic injury on labor market activity, to define additional factors that may act to facilitate or inhibit return to work, and to better assess the need for program interventions that would facilitate recovery.
Patients and Methods: Patients were recruited on the basis of: 1) pre-injury employment, 2) absence of psychiatric history, 3) absence of neurologic or other permanently disabling injury. Data were collected prior to discharge and at 6 month follow up. Data included injury severity, pre- and post-injury wages/salary, working conditions, job satisfaction, qualification for transfer payment programs, pre-injury health status, and household wealth.
Summary Statistics. The average age of the patient population was 33 yrs (s.d. 11). The patients had a mean injury severity score of 12 (s.d. 9.3) and length of stay (days) of 6.3 (s.d. 4.8). Injuries incurred were as follows: upper ext. fx. (7), lower ext. fx. (24), pelvic/spine fx. (11), chest/abd. inj. (16).
Results. Sixty-one patients were recruited for the study with occupations evenly distributed between blue (n=31) and white collar (n=30) classes as defined by the Department of Labor. Prior to injury 82% of patients found their work a rewarding experience; 74% said they would continue working even if they did not need the money; and 61% felt their injury would impair their ability to return to work. At six month follow-up interval, 67% of subjects had not returned to work. The average annual salary of the patients was $24,944 (s.d. $15,014) while average lost earnings and working hours due to injury were $10,165 and 796, respectively. Over one-third of the subjects reported that a family member took time off work to care for them. The average lost earnings of these family members was $2,760. Factors that positively influenced return to work were job satisfaction and earnings, while negative factors were stressful working conditions, mental health status, and impending lawsuits related to the injury.
Conclusions. Moderate trauma results in potentially crippling economic disruption to individuals and their families and inflicts large aggregate costs to society. Low income households may be further marginalized by the effects of trauma due to substantial loss in patient and family earnings. Trauma patients "at-risk" for not returning to the labor force may be identified on the basis of post-injury economic and vocational factors. The aggregate costs to individuals and society reflected by this data suggests a need for, and the potential feasibility of, cost-effective programs that will facilitate return to work in this patient population.
Gaining an Economic View of Construction-Related Injury: Integrating Quantitative and Qualitative Data—Kidd P, Struttman T, Mays J, Parshall M
Economic factors influence both the safety decisions and safety performance of construction workers and owners of construction companies. The framework for intervention development used in this study is an inductively derived model linking stress, economics, and injury. To develop an intervention that integrates economic factors in a meaningful fashion for these individuals, it is necessary to understand from their point of view the decision making process used in assuming greater risk. For injury prevention efforts to be successful in the construction industry, owners/operators and workers must be presented with information depicting safety as economically advantageous to their companies and careers. Because no one data source is available to understand the multifaceted, complex relationships between economics and safety, integration of multiple data bases were used. In this presentation we will discuss briefly the intervention model and the data bases used in gaining an economic view of construction-related injury. Data bases integrated include: focus group sessions, case studies of fatal and minor injuries, interviews with chiropractors, a hospital trauma registry, state worker compensation files, and insurance policies and claims. The integration process used in combining qualitative and quantitative data will be discussed and illustrations provided.
Measuring the Injury Severity of Occupational Injuries—Gillen M, Faucett J, Beaumont J, McLoughlin JJE
Finding a measure that distinguishes well between the severity levels of less serious injuries such as those found in occupational settings has been problematic. In this study of 255 construction workers who sustained nonfatal falls at work, two measures of injury severity were used--the Injury Severity Score (ISS) and the disability section of the Stanford Health Assessment Questionnaire (HAQ), a functional limitation measure. It was hypothesized that the functional limitation measure would provide a more normal distribution of scores as compared to the ISS, which is a threat-to-life scale. As predicted, the ISS scores in this study were clustered toward the lower end of the scale with 112 injuries (43%) receiving a score of 1. The mean ISS was 3.2 (SD = 3.0; median = 2.0) on an ordinal scale from 1 to 75. The highest ISS was 22.
The HAQ originally was developed to evaluate functional limitations in illness. It has primarily been used in rheumatic conditions, but has been successfully used with other chronic diseases. The HAQ was selected in this study because it measures overall functional limitations as well as limitations in upper and lower extremities. It consists of 20 questions grouped into eight components reflecting activities of daily living such as dressing, walking, reach, and grasp. Participants evaluated their ability to perform these tasks during the first week after their fall.
The mean functional limitation, as measured by the HAQ score, was 1.5 (SD = 0.7) on a scale of 0 to 3, with higher numbers representing more limited functioning. The median HAQ was 1.38. The mean scores in samples with various rheumatic diseases, participating in multi-center studies, have been reported from a low mean score of 0.55 (SD = 0.1) for systemic lupus erythematosus to a high mean score of 1.34 (SD = 0.02) for rheumatoid arthritis.
With regard to individual tasks, participants reported having the most difficulty performing heavy chores (mean = 1.89; SD = 1.02), dressing themselves (mean = 1.54; SD = 1.05), and bending and picking up clothing from the floor (mean = 1.40; SD = 1.02). Tasks which caused the least difficulty were opening car doors (mean = .62; SD = .96), turning faucets on and off (mean = .34; SD = .72), and lifting a full cup or glass (mean = .31; SD = .71). There were no statistically significant differences in means among the trades with regard to HAQ scores, but the highest scores were reported by drywallers and plasterers, and the lowest were seen in electricians.
The HAQ scores provided useful information about the degree to which workers were disabled from falls during their first week of recovery. They were moderately correlated with days lost from work (r = .52; p = <.001), hence could prove useful to occupational health practitioners. Use of the HAQ was successful with this population; however, it needs to be evaluated in other samples to more conclusively determine its utility in evaluating occupational injury disability. Sequential administration would be useful to assess whether the HAQ is sensitive in detecting minor patient status changes over time.
Formal Safety Analysis Methods as Tools for Hazard Surveillance— Kerkering JC
Traditional health surveillance methods are based on a disease-and- illness model. That is, the model is used to respond to the incidence of illness and disease by categorizing and counting such incidences and describing the circumstances surrounding the occurrence. This approach is dictated by perceptions of a cause-and-effect relationship where the cause is not readily observable and a significant time lapse lies between the initial cause (source) and its effect (illness). Hence, traditional health surveillance has, by necessity, often been reactive and descriptive rather than proactive and predictive.
Such a reactive surveillance model, as stated in the NORA research guideline document (p. 53), is deficient where the topic of concern is worker safety, and the effect of a triggered hazard is usually immediate with often fatal results. Surveillance activities, where used to prevent traumatic injury, must identify these workplace hazards and the conditions that trigger accidents. It is often possible to identify the causes of accidents in terms of hazards and triggering conditions and thus predict possible effects before an accident occurs.
The author claims that an effective hazard surveillance model for hazard identification and accident prediction requires a proactive approach, an approach that is feasible using well-established, systematic, safety-analysis procedures. This paper describes two broad safety-analysis approaches, inductive and deductive, and then suggests how these approaches can be used to anticipate accidents so that preventive measures can be taken. Preliminary hazard analysis is described as an example of an inductive approach, while fault tree and event tree analyses are described as examples of deductive approaches. The mining industry provides examples for each approach: a fault tree analysis of a blocked ore chute in a deep metal mine and a preliminary hazard analysis of a longwall escapeway in a coal mine. The author concludes that these methods could improve hazard surveillance results and provide new insights into cause-and-effect relationships related to risk and traumatic injury in mining.
Surveillance of Disaster--A View from the Denominator—Chaiken RF
Surveillance plays a very significant role in the NIOSH epidemiological model as applied to occupational health and safety. Surveillance studies are used in identifying occupational problems, evaluating the effectiveness of intervention procedures, establishing research priorities, and allocating resources for its health and safety program.
The NIOSH surveillance system has been highly successful in serving NIOSH's mission of prevention of occupational injuries as applied to relatively large number, relatively high frequency events such as falls, machinery mishaps, and environmental exposures. However in the case of relatively low number and low frequency industrial disasters such as explosions and fires, and commercial aircraft accidents, the applicability of the surveillance model is far less clear. I believe that some of the problem lies with worker exposure--its interpretation and evaluation, and some of the problem lies with assumptions (both explicit and implicit) as to weighting factors (i.e., economic, societal and hazard impacts) to be or not to be considered in the data analysis. I believe there is a definite need to broaden the data base that exists in current surveillance S&H models; particularly in the area of worker exposure.
These points will be elaborated on in terms of the `numerator' and the `denominator' of surveillance information, in which BLS and NIOSH surveillance studies revolve around a count of the injured (the numerator) normalized to some count of the workers exposed (the denominator). If one interprets the `numerator' of surveillance rate data as referring to those individuals, institutions and events which are directly affected by accidents that occurred, then weighting factors that relate to the numerator (referred to as N-factors) can be defined based on degrees of hazard or injury and the economic and societal impacts that are directly associated with the occurrence of the accident.
The `denominator' of surveillance data will likewise be affected by weighting factors, but the D-factors will refer to those individuals, institutions and events which are in harms way, i.e., those who will be affected by future accidents. As will be described, the D-factors being different from the N-factors, will lead to a ranking scheme which is different from that currently employed by NIOSH, but one that could lead to an all-inclusive model for epidemiologic analysis of accidents.
Another aspect of the model employed by NIOSH for prevention of occupational injury is what might be called the `Haddon Strategy' for reducing injuries. With this paradign, remediation efforts focus on those means available for reducing injuries associated with an accident, rather than on determining the exact cause(s) or `anatomy' of an accident. For example, water purification will control disease even in the absence of specific knowledge as to the pathogens that would cause disease. Unfortunately, low frequency disastors, such as mine explosions and fires, often involve the liberation of energies so great that prevention of occupational injury dictates that the event must be prevented from occurring. This requires a fundamental understanding of the exact causes that can lead to the event. In these cases, it is only through a detailed anatomy (research) of an accident that solutions for prevention will be found.
The Effect of Using Death Certificate Information on Industry and Occupation Specific Fatality Rates—Wolf SH, Loomis DP, Gregory E, Runyan CW, Butts JD
Death certificates are a primary source of information used to estimate industry and occupation specific fatality rates for the United States. Funeral directors completing this information are instructed to record the decedent's usual industry and occupation. However, the usual industry and occupation during a decedent's lifetime, rather than their industry and occupation at the time of the injury, may be misleading in determining industry and occupation specific rates of fatal injury at work. The purpose of this study is to compare the industry and occupation at the time of the fatal injury described by the medical examiner with the usual industry and occupation recorded on the death certificate and identify effects on computed rates.
Cases for the study were selected from records of the Office of the Chief Medical Examiner for North Carolina. All accidental deaths and homicides identified as "at-work" in the medical examiners' system between 1977 and 1991 were selected for study. Information on usual industry and occupation was abstracted from death certificates in the medical examiner's records and industry and occupation at the time of the fatal injury were determined from medical examiner investigative reports. Industry and occupation were coded to the U.S. Bureau of the Census. Tabular analyses describe the differences between usual and injury industry and occupation groups by fatality type (intentional and unintentional), age, gender and race. Case and population data were used to estimate industry and occupation specific rates of injury mortality per 100,000 worker-years.
The medical examiner's system identified 2524 injury deaths on-the- job between 1977 and 1991; 2099 from unintentional trauma, 358 from homicide and 67 from poisoning or environmental conditions. Industry at the time of injury was known for 2465 cases and the overall agreement with usual industry was 78%. Homicides had a lower percentage of agreement (73%) than did cases of unintentional traumatic injuries (79%). Occupation at the time of injury was known for 2449 cases and the overall agreement with usual occupation was 75.3%. Percentage agreement for homicides (74%) was the same as for unintentional traumatic injuries (75%). The agreement between injury and usual industry or occupation did not differ by race or gender. For workers 65 years and older, death certificate data about industry was less likely to agree with data from the medical examiner (65.9%) than for workers aged 18- 64 (80.7%). Crude mortality rates for unintentional traumatic injuries were underestimated for the ten leading industries and eight of the ten leading occupations using usual industry and occupation rather than industry and occupation at the time of the injury. For homicides, underestimates were observed for the two leading industries and occupation. Industries most likely to be affected by misclassification were: agriculture; forestry and logging; construction; truck and bus transportation; and grocery stores. Forestry and logging occupations and fishing, hunting and trapping occupations were most likely to be misclassified. While usual industry and occupation identifies the major industry and occupation groups at risk for fatal occupational injuries, its use can misstate actual fatality rates for specific industries and occupations.
Fatal Occupational Injury in North Carolina: Using Capture-Recapture Methods to Ascertain and Evaluate Annual Mortality from Multiple Data Sources—Hooten EG, Butts JD, Baker SP
Fatal injury in the context of one's employment is a problem that seems relatively easy to define and, consequently, to enumerate. Unfortunately, investigation reveals an issue plagued by imprecise definitions that vary by data source and agency mandate, further complicated by confusion about the nature of activity at the time of injury and characteristics of the decedent's employer as influences upon whether an injury death is characterized as on-the-job (OTJ) or not.
Capture-recapture methods were developed in animal ecology to improve the accuracy of estimates of wild animal populations; and, these methods compare the number of animals caught from a given population in one capture effort with another survey done at a later time. By matching animals caught in both surveys and noting the differences, an estimate of the total population can be developed. Application of capture-recapture methods to fatal occupational injury, using multiple data sources as analogous to serial capture efforts, holds promise for improving the precision of annual fatal injury incidence estimates and the characterization of deaths as “truly work-related”. It further facilitates evaluation of differences between these datasets in terms of the types of cases found in each.
North Carolina has a centralized medical examiner system in which consistent effort is made to identify and record information for deaths on-the-job. In addition, North Carolina has its own Occupational Safety and Health Act (OSHA) program which collects information on fatal events within the purview of OSHA guidelines as well as a Worker's Compensation Insurance Program and a death certificate file wherein the OTJ variable is collected on every death certificate. North Carolina also participates in the US Department of Labor's Census of Fatal Occupational Injury Program (CFOI) in which each case is verified as work-related from at least two different sources, including all of those noted above.
For 1994 and 1995, the ME, death certificate, worker's compensation, and OSHA data sets for North Carolina are examined to assess the concordance and discordance between them and, from that, to provide a more complete estimate of the “true” annual incidence (including confidence intervals) of fatal occupational injury using multiple source methodology (capture-recapture methods). The resultant aggregate estimate as well as its component parts (datasets) are compared to the CFOI dataset (the gold standard since inclusion in this set as a fatal occupational injury is confirmed by more than one source). Sensitivity and specificity analyses for each dataset are performed to identify areas of variation by age, race, gender, and other variables such as the decedent's occupation/industry. Univariate and multivariate logistic regression are used to estimate the log odds (odds ratio) that a given type of case will be found in the component datasets and aggregate data when compared to CFOI. Policy and practice modifications that will have direct effects on the completeness of the data sources individually and in aggregate are discussed along with implications for the application of these methods to the development, implementation, and evaluation of prevention efforts.
A Decision Model for Use in the Allocation of Safety and Health Resources—Warrack B, Redekop T
There has been an increasing emphasis in jurisdictions on directing safety and health prevention activities to where they are most needed. To effectively accomplish this, better data and improved data analysis methods are needed. Data on occupational injuries and illnesses is useful to assist in deciding where to direct safety and health inspection activities. The model being proposed uses a number of decision variables to assist in deciding how to priorize and focus efforts. Key to any model such as this is being able to use data to define, characterize the risks and quantify the risks to workers so that risks can be appropriately ranked and risk comparisons made so as to focus efforts. Manitoba jurisdiction data will be used to demonstrate how this model can be used in practice. Also discussed will be some shortcomings and problems associated with using this type of data in decision models.
Re-conceptualization of Measuring Musculoskeletal Disorders: A Pilot Test of Telecommunications Workers—Griffin J
While reports of repetitive trauma injuries and disorders among workers have continued to increase over the past decade, little research has been done to either modify existing measures or to develop new measures that are valid and reliable. Having valid and reliable measures is essential to understanding the relationship between early signs of musculoskeletal strain and chronic musculoskeletal disorders; moreover, they may be important to understanding how strain increases the vulnerability to other work-related injuries.
For this project, the current literature on work-related musculoskeletal disorders was reviewed. Experts were consulted to determine the strengths and weaknesses of the frequently used measures of musculoskeletal disorder symptoms. Combining the information from experts, findings from previous research, and other hypothesized relationships, a new measure of musculoskeletal symptoms was then developed and tested for validity and reliability. Eight anatomical areas (wrists, hands, fingers and forearms; neck; shoulders and upper arms; elbows; upper back; lower back; hips and thighs; feet and ankles) are included in the new measure. For each anatomical area, general questions about symptoms, such as pain, weakness, and limited motion are asked. Those reporting any symptoms are then asked more specific questions about limited activity, previous injury, medical treatment, and severity.
To test this new measure, surveys were sent to 179 telecommunication workers from a local labor union. All the workers surveyed have jobs that require them to use repetitive motions on computer display terminals for a majority of the workday. This work group is primarily women (80%), who are married (60%), with an average age of 37 years. The average length of time working for the current employer is 8.5 years.
Workers report general symptoms of musculoskeletal disorders most commonly in the neck; lower back; and hands, fingers, wrists or forearms. For each of these groups, about 70% of the sample report at least one general symptom (e.g., pain, limited movement) specific to that group. Approximately 45% reported at least one general symptom in their shoulders and upper arms.
Responses to questions about general symptoms of musculoskeletal disorders were also compared to a self-report of being medically diagnosed with a musculoskeletal disorder. Workers were asked to report if they had ever been diagnosed with certain disorders, such as carpal tunnel syndrome or a back disorder to the muscles, nerves or discs. Of those who report having been diagnosed with carpal tunnel syndrome by a doctor, 88% answered “yes” to having all of the general symptom questions associated with hands, fingers, wrists or forearms. Of those who report ever having been diagnosed with a back disorder to the muscles, nerves or discs, 79% answered “yes” to having all of the general symptoms associated with the lower back.
This presentation will include a short analysis of the current measures, a discussion of the development of the new measure, a critique of the validity and reliability properties of the measurement, and possible applications of this measurement to evaluate risk for acute and chronic injuries at the workplace.
New Biostatistical Approaches to Measure and Monitor Health Trends—Brant LJ, Bos AJG
Recent interest in monitoring health trends in individuals through repeated health examinations has led to a realization that one of the challenges facing these programs is to distinguish aging and disease from other health-related events. Aging and disease progression are both processes of change in individuals occurring over the course of time, and can be affected by the occurrence of numerous events in the individual's living and working environment. Both age changes and disease have a direct impact on the function of the individual and thus needs to be accounted for in any monitoring process of the individual. The process of change occurs in everyone, but it is a highly individual phenomenon that occurs differently from person to person. Some individuals may change very little with regard to a particular health-related outcome, while others may show increasing or decreasing trends of the outcome. In order to monitor and better understand health trends in individuals, biostatistical approaches must be available that allow investigators and clinicians to examine the different patterns of change over time among persons as well as account for the natural heterogeneity among individuals. Such an approach must be able to predict each person's response even if that response differs substantially from the average response for the population. Information for describing patterns of individual response requires the development of a medical history that is obtained by monitoring a particular outcome for each individual over time. This paper discusses a new biostatistical approach that is appropriate for monitoring individual health trends and describing the natural heterogeneity in the study population. Results from the approach can be used in identifying individuals with deviate or abnormal patterns of response. Examples of applying the approach include the monitoring of prostate specific antigen as a diagnostic tool for prostate cancer and the monitoring of blood pressure as a diagnostic tool of heart disease. These biostatistical tools have significant potential for research and practice in health evaluation, health prediction, and the development and promotion of injury and disease prevention strategies.
Counting Concepts: Estimating the Population-at-risk for Regulatory Action—Botkin A
Before an OSHA regulation can be enacted, a considerable amount of study and analysis is devoted to the supporting materials needed for the rule, and its regulatory impact analysis, or RIA. The need for regulatory impact assessments is created by Executive Orders, as well as Congressional and judicial actions which require Federal agencies to evaluate the impact of proposed regulations.
A significant issue in the development of a regulatory impact analysis is estimation of the population-at-risk. This value characterizes the baseline, drives the estimate of potential benefits as well as the cost of compliance in achieving reductions in the hazard following implementation of regulations. But for most regulatory actions there is no statistical data service that can provide a count of the specific entities or occupations affected by the proposed rule.
Like many other important aspects of safety and health regulation, a confined space is a concept. It is instantly recognizeable to informed safety professionals, but it is of little importance in any other business or economic context and thus not enumerated or examined in statistical data.This presentation will illustrate the research and development of this important metric for the OSHA general industry confined spaces standard, and the development of its analogue in the construction industry standard on confined spaces.
Slipping, Tripping, and Falling Incidents: Steel Erection and Metal Roofing Workers—Murphy PL, Cotnam J, Sorock G
Falls in construction are a serious problem. In a recent study (Leamon and Murphy 1995), 38% of workers' compensation claim costs in the construction industry were associated with falls. The incidence of falls was 4.9 falls per 100 full time workers. A per capita cost was calculated at $560 per employee for direct medical and indemnity costs. Steel erection workers and roofers have been identified as high risks for falls in the construction sector, but limited international research on the risk factors or causes have been identified in the literature (Suruda et al., 1995, Flett 1992, Hardesty et al., 1993).
This study is a retrospective analysis of workers compensation claims over a seven year period, with over 90,000 claims specific to worker's compensation job classifications for steel erection workers, bridge workers, and metal roofing. The number of slipping, tripping and falling incidents and the relative cost of falls are examined by injury type, body part and occupation. Aggregate frequency and severity of events, and relative individual claims costs are presented. Details on type of fall, the location of the slip, trip or fall, the walking surface and condition or contaminant, employee activities, and although very limited, footwear worn, when the incident occurred will be extracted from individual narrative accident descriptions. The available narrative description is limited to 120 characters. Algorithms based on word frequency and usage were developed for this analysis.
Specific goals of the study are to identify the scope of occupational injuries, and antecedent events for workers exposed to metal roofing, decking, and beams, determine the significance of slipping incidents for falls on the same level and falls from elevation and compare with tripping, loss of balance, falls through openings, from perimeters, from scaffolds and other locations. Additionally, this study determines the significance of slipping incidents that occur on steel beams, decking and roofing and compares them with slipping incidents on other surfaces like ladders, concrete and plywood.
References
Leamon, T.B. and P.L. Murphy. Occupational slips and falls: more than a trivial problem, Ergonomics, 1995, Vol. 38, No. 3, pp. 487- 498.
Suruda, A., D. Fosbroke, and R. Braddee. Fatal Work-Related Falls from Roofs, Journal of Safety Research, 1995, Vol. 26, No.1, pp. 1-8
Flett, D.S. Reducing Risks in the Erection of Structural Steel, Safety Science, 1992, Vol. 15, Nos. 4-6, pp. 215-224.
Hardesty, C., C. Culver, and F. Anderson. Ironworker Fatalities in Construction, Occupational Hazards, June 1993, Vol. 55, No. 6, pp. 47-49.
Nonfatal Falls in Construction Workers: Predictors of Injury Severity—Gillen M, Faucett J, Beaumont JJ, McLoughlin E
The study evaluated injury severity in a group of construction workers who sustained nonfatal falls at work. The convenience sample consisted of 255 adults, predominantly males (97%) with a mean age of 34.6 years (SD = 9.3), who were identified from Doctor's First Reports submitted to the California Department of Industrial Relations. A full range of construction trades was represented in the sample population. More than one quarter of the sample were union members, and more than half worked in residential construction. For those that fell from heights (76%), the mean height of the fall was 9.2 feet (SD = 7.1).
Two measures of injury severity were used--the Injury Severity Score (ISS) and the disability section of the Stanford Health Assessment Questionnaire (HAQ). There were 518 injuries including 61 extremity fractures. Thirty-two individuals were hospitalized and 41 required surgery. Seventeen participants (7%; 95% CI, 4 to 10%) were deemed permanently disabled. A simultaneous multiple regression model, using five independent variables, explained 21% of the variance in HAQ scores. Nonunion status and poorer safety climate scores indicating increased risk were positively correlated with higher HAQ scores, as were greater heights and impact on concrete surface. The mean number of lost days was 44.3 days (SD = 58.6), and the median was 10 days. Both measures of injury severity were moderately associated with a greater number of days lost from work ( ISS: r = .43; p = <.001; HAQ: r = .52; p = <.001).
Almost half of the sample (46%) were involved in direct installation when they fell, and 91% were performing their usual duties. The most frequently reported specific activity being performed at the time of the fall was walking (22%), followed by descending (11%), and climbing (10%). Workers who fell from a height, fell most often from ladders, scaffolds, or planks (41%), followed by roofs and wood skeletons (24%). Forty six percent of the sample fell from heights of six feet or more.
The Safety Climate Measure score was significantly associated with union status (r = .225; p = <.001), and is most likely explained by five items in the instrument where union and nonunion members differed in their perceptions regarding safety conditions at their job site. Union members were more likely than nonunion members to perceive supervisors as caring about their safety, be made aware of dangerous work practices and conditions, have received safety instructions when hired, have regular job safety meetings, and not perceive that taking risks was a part of their job.
These findings confirm that falls in construction are far too common, suggest that injury severity and disability associated with falls is notable, and identify key target areas for intervention and prevention such as management commitment to safety, ongoing worker training, and hazard identification and control.
Risk Factors of Task Performance at Elevated and Inclined Surfaces— Bagchee A, Bhattacharya A, Succop P, Medvedovic M, Mitchell T
A majority of fatal and non-fatal accidents at the workplace occur due to fall-related incidents, particularly in the construction industry with twice the rate of nonfatal falls as the average industry. Most falls result from momentary loss of postural stability of the worker. Working on elevated and/or inclined surface can produce excessive demand on the postural control system of the worker. For example, ironworkers and roofers traditionally work at raised and/or inclined surfaces and account for 70% of fall-related fatalities. A laboratory-based study was performed to investigate the effect of surface elevation, surface inclination, environmental lighting and noise on postural stability of workers performing simulated tasks. Twenty young (age =3D 25.0 =F1 2.2 years) and twenty old (age =3D 53.7 =F1 3.2 years) industrial workers, with equal number of males and females in each group, participated in the study. The subjects performed three kinds of tasks of Stationary, Bending, and Reach. The subjects performed these tasks on specially constructed surfaces with combinations of three levels of elevation (0, 30.5, and 61 cm high), and three levels of inclination (0, 14, and 26 degrees). The environmental lighting was varied between poor ( < 0.2 footcandles) and good ( > 40 footcandles). The subjects were fitted with lightweight wireless headphones that presented audible distraction in the form of pre-recorded construction sound at a setting sufficiently loud to interfere with levels of normal human conversation. The forces and moments exerted by the subjects were recorded using a piezo-resistive force platform and the data was analyzed to yield the movement of the center of pressure (CP) as a quantitative measure of postural sway. Stepwise regression analysis was used for identifying the significant interactions. Dependent variables included the measures of sway area and sway length, with the age, height/weight ratio, and gender included as covariates. Mixed model analysis showed highly significant fixed factor effects for elevation (p < 0.001) and environmental lighting (p < 0.001) for all three tasks for both sway area and length, with increasing sway area and length for increasing elevation and inclination The sway length increased significantly with increasing inclination for the Bending and Stationary tasks, but was non-significant for the Reach task. Sway area increased significantly (p < 0.05) with increasing inclination for all three tasks. The results are indicative of postural balance performance that deteriorates with increase in elevation of the working surface as well as the inclination of the surface. This introduces additional burden and may be a major contributor to momentary loss of balance resulting in falls. Environmental lighting was also found to be a significant risk factor and introduced further burden on the postural stability of the worker. The experiment was performed under laboratory conditions, with no workload and on firm support surface with sufficient coefficient of friction. These risk factors may further jeopardize the postural stability of the worker at an elevated and/or inclined surface. Results from this study would be helpful in identification of specific risk factors and their relative contribution in postural imbalancement during dynamic task performance on inclined and/or elevated surfaces.
Fall Prevention in Construction by Organizational Intervention— Becker P
This paper will present the results of a research project to evaluate an innovative organizational intervention to prevent falls in con struction. This intervention research is funded by the Center to Protect Workers' Rights and NIOSH.
Falls are the leading cause of injury in construction. The construction safety literature recognizes available engineering controls, work practices, and personal protection which are effective in preventing construction falls. However the equipment and practices are not widely used in the industry.
This project implements a construction contractor certification program called Fall-Safe to improve management use of existing fall prevention methods and to develop an accountability system on job sites to maintain appropriate controls in the dynamic construction setting. WVU Safety and Health Extension serves as the certifying organization for contractors in West Virginia, and is assisting contractors in developing office and site fall prevention programs, training of supervision and workers, and quarterly audit of both company and site fall prevention efforts.
Fall-Safe has been successfully marketed to contractors as a way to increase the contractor's marketability to large purchasers of construction services. Initial development of the program enlisted the Appalachian Construction Users Council, an organization of industrial plants who regularly use construction services. With the support of the construction users, contractors eager to work on these sites have enlisted as Fall-Safe Participants.
The Fall-Safe program incorporates a rigorous evaluation component using comparisons of intervention and control groups of workers and contractors. Baseline measurements and twice yearly measurements are compared prospectively between contractors in northern and southern WV working for American Electric Power.
Measures include contractor, worker, supervisor attitude and knowledge surveys, and an observational scoring system carried out on sites determining use of appropriate fall prevention techniques. In the longer run, the project will include sufficient power to compare Workers' Compensation claims for falls between intervention and control groups.
Severe Repetitive Strain Injury-Focal Hand Dystonia: Central Neural Consequences Versus Local Tissue Injury—Byl NN
Repetitive strain injuries are reaching epidemic levels in the workplace despite increased ergonomic attention to the work site. It has been hypothesized that repetitive strain injuries result in tissue microtrauma following the biomechanical stress of repetition. Recently central neural consequences have been reported. The question is whether severe RSI-Focal hand dystonia results from severe sensory degradation of the hand representation on the somatosensory cortex or local tissue microtrauma including adhesions or inflammation. Owl monkeys were trained in a behavioral paradigm of rapid active opening/ closing of the hand under conditions of high cognitive drive. This led to the development of motor problems which appeared to simulate focal hand dystonia. Severe degradation of the hand on the somatosensory cortex was measured. Post mapping, the monkeys were taken to the anatomy lab for careful dissection of the flexor tendons and histological study of the tendons and the median nerves. The anatomical dissections of the flexor tendons and the nerve were normal for all monkeys except the monkey that developed the focal hand dystonia, most specifically involving the fourth digit, in an unusually short period of time (5 weeks). The cortical penetrations in area 3b showed large receptive fields and a high proportion of multiple receptive fields involving D4 and adjacent digits, as well as multiple receptive fields which overlapped the glabrous and dorsal surfaces. On the anatomical dissection, the profundus tendon of D4 was adherent across the middle phalanx. Interestingly, the same adhesion was seen on D3 on the untrained side. There were no signs of inflammation of tendons or the median nerve on any of the digits on either hand. These studies suggests that severe RSI-focal hand dystonia may result from a degradation of the somatosensory representation of the hand without local pathology. This suggests that treatment needs to address the neural consequences of the disorder and not just the biomechanical effects.
Reliability, Validity, and Clinical Utility of a Battery of Physical Performance Tests for Patients with Low Back Pain—Olson SL
Low back pain is a wide spread and costly work-related problem that leads to physical dysfunction. Clinical evaluations are based upon measures of impairment such as range of motion, strength, and pain. Physical performance measures have potential utility for evaluation, treatment planning, and determination of treatment outcome, yet few measures have proven reliability and validity. The purpose of this study is to determine the reliability, validity, and clinical utility of eight physical performance measures. Subjects: A control group of 48 healthy subjects and a group of 44 subjects with current low back pain (LBP) participated in this study.
Procedure: The following physical performance tests were used: 1) Ten repeated trunk flexions; 2) Five repeated sit-to-stands; 3) Timed up and go; 4) Loaded forward reach (with load of 10% body weight); 5) Unloaded forward reach; 6) Five-minute walk (distance measured); 7) Fifty-foot walk ( preferred and fast speeds); and 8) Sorensen fatigue test. The reach tests were measured in cm. and the timed tests were measured in seconds. Lumbar flexion was also measured. The subjects repeated this battery of tests twice a session at two separate sessions. The order of tests was randomized except for the Sorensen test, which was performed last. Two testers were randomly selected from a group of six to assess each subject.
Data Analysis: ICC Model One analyses were used to determine intertester, test-retest, and day-to-day reliability values. A two-way multivariate analysis of variance (group by gender) and follow-up univariate analyses were conducted on all variables to test validity. Pearson correlation coefficients were calculated among the physical performance measures to determine convergent and discriminant validity. Results: Intertester reliability was above .95 for all variables across both groups. Test-retest reliability was above .83 for all measures across both groups except repeated flexion, which was .7 for the control group and .45 for the LBP group. Day-to-day reliability ranged from .46 to .76 for the control group and .59 to .88 for the LBP group. Day-to-day reliability values were higher when comparing the averages of 2 trials in the control and LBP groups, .62 - .89 and .76 - .91, respectively. The results of the MANOVA showed a significant effect of group only (F=3.25; df=11,62; p=.002). Univariate analysis revealed significant group differences on all variables except unloaded forward reach and 50- foot walk at preferred speed. In the LBP group, strong correlations were found among the 5-minute walk, 50-foot walks, and timed up and go (r=.78-.96). Lumbar flexion angle correlated poorly with repeated flexions (r= -.16). For the control group, the highest correlation was between the timed up and go and the 50-foot walk (r=.67). Conclusions: Overall these measures had adequate reliability and validity. Averaging 2 trials for the repeated flexion and sit-to-stand tests is recommended to increase reliability. The measures show good clinical utility as they are easy to perform, acceptable to all subjects, and require little equipment.
Ergonomics and Work Injury Management: 15 Years of Application in a Clinical Setting—Khalil TM, Steele-Rosomoff R, Abdel-Moty E, Rosomoff HL
The goal of rehabilitation of work injuries encompasses preventing disability through functional restoration and immediate return to a productive lifestyle. Due to the complex nature of the problem, it has been well recognized that chronic pain management requires a multidisciplinary approach since no one physician or therapist has the expertise or resources to manage this condition. One discipline that can contribute, significantly, to pain management is Ergonomics. Ergonomics studies deal with safety, human performance analysis, work environment, and other studies of value to the rehabilitation process. Also, due to its nature as an interdisciplinary science, it can offer solutions to many problems related to injury and its prevention. The Ergonomics Division of the University of Miami Comprehensive Pain and Rehabilitation Center (Miami Beach, Florida, USA) is an example where ergonomists work daily with members of the multidisciplinary rehabilitation team to solve problems in pain management and return to work issues. Over the past 15 years, the Ergonomics Division has integrated its activities and resources into the organization chart of the Center to address many complex problem from both the engineering and the medical perspectives. For the first time, this has provided the scientific basis for the rationalization of many treatment approaches. Through applied research activities, dissemination of valuable information and data was possible and new untraditional treatment approaches and techniques were developed. In the area of patients care, ergonomics contributes to the determination of the functional status of the low back pain patient through the establishment of profiles of functional abilities. This is done through a battery of quantitative measurements that help establish a human performance profile for each patient upon admission and throughout rehabilitation. The goal of treatment is, then, to condition the injured individual and to restore functional levels to the “normal” capacities of healthy uninjured individuals. The performance profiles are then compared to the physical demands dictated by the job. The objective here is to determine intervention strategies for matching the physical capabilities of the individual to specific job task. Ergonomics job analysis, job simulation, and job-site visits are used to prepare the injured worker to reenter the productive job market and lead a normal life style. Ergonomics knowledge pertaining to workplace design and expert systems is used to help patients adjust their workplace in order to minimize potential stresses due to poorly designed and/or improperly adjusted workplaces. Ergonomists also assist in the selection of jobs that match the measured functional capabilities of the rehabilitated persons. This paper will outline the history, rationale, methods, and various interventions which have been developed and tested in our facility.
Forearm Muscle Oxygenation Decreases During Low Levels of Brief, Isometric Contraction—Murthy G, Hargens AR, Kahan NJ, Bach JM, Rempel DM
Introduction. Regional muscle pain syndromes can be caused by repeated and sustained exertion of a specific muscle. Such exertion may elevate local tissue fluid pressure, reduce blood flow and tissue oxygenation (TO 2 ), and cause fatigue, pain and functional deficits of the involved muscle. Low levels (less than 20% maximum voluntary contraction (MVC)) of prolonged static contraction of the upper extremity are common in many occupational settings and may cause fatigue (1). The purpose of our investigation was to determine whether TO 2 decreases significantly at low levels of static contraction of the extensor carpi radialis brevis (ECRB).
Methods. Healthy male and female subjects (n=9) participated in the study after providing written informed consent. The protocol was approved by the University Human Research Institutional Review Board. Each subject was seated, right arm was abducted to 45º, elbow was flexed to 85º, right forearm was pronated 45º, and wrist and forearm were supported on an arm rest throughout the protocol. Altered TO 2 was measured noninvasively using near infrared (NIR) spectroscopy (2). This technique has been validated previously (3). The NIR probe was placed over the ECRB muscle and gently secured with an ace wrap. MVC was determined initially and the subject rested for an hour prior to subsequent tests. After one minute of relaxed, baseline measurements, four different loads (randomly ordered) were placed just proximal to the metacarpalphalangeal joint such that subjects isometrically contracted the ECRB at 5, 10, 15, and 50% of MVC for 1 minute each. A 3 minute recovery period followed each contraction level. At the end of the protocol, with the NIR probe still in place, an ischemic TO 2 was obtained to establish a zero level for each subject. NIR data were normalized to a relative scale between the physiologic minimum (0%) established during ischemia and the spectrophotometer output at baseline (100%).
Results. After 35 and 40 seconds of contraction, TO 2 plateaued at below baseline levels and remained at that level throughout the contraction period. Mean TO 2 decreased from resting baseline (100% TO 2 ) to 89 ± 4% (SE), 81 ± 8%, 78 ± 8%, and 47 ± 8% at 5, 10, 15, and 50% MVC, respectively. TO 2 levels at 10, 15, and 50% MVC were significantly lower (p < 0.05; RANOVA and Tukey's follow-up) than baseline values. TO 2 recovered to baseline values within 3 minutes following contraction.
Discussion. This study demonstrates a significant reduction in TO 2 even at sustained contraction levels as low as 10% MVC. Tissue deoxygenation during prolonged isometric muscle contraction may play an important role in the development of work-related muscle fatigue and pain. Static or dynamic contraction with inadequate recovery time may sustain elevated intramuscular pressures, and reduce blood flow and TO 2 , and cause muscle fatigue and pain. Although the duration of static contraction in our study was only 1 minute, the observation that recovery to baseline TO 2 took between 30 seconds to 3 minutes indicates that a low contraction level even for a brief period is sufficient to reduce TO 2 significantly. Therefore, sustained tissue hypoxemia associated with low levels of sustained contraction may provide a mechanism to explain work-related muscle dysfunction.
References
1. Sjögaard et al. Eur J Appl Physiol 57:327-335, 1988
2. Chance et al. Analyt Biochem 174:698-707, 1988
3. Belardinelli et al. Eur J Appl Physiol 70:487-492, 1995.
Preventing Drownings in Alaska's Commercial Fishing Industry— Conway GA, Lincoln JM
Introduction. The Arctic and subarctic waters of Alaska provide a very hazardous work setting, with great distances, seasonal darkness, cold waters, high winds, brief fishing seasons, and icing. Deaths have been inordinately common in Alaska's commercial fishing industry. Over 90% of these deaths have been due to drowning or drowning plus hypothermia, following vessel capsizings and sinkings. During 1991 through 1994, the U.S. Commercial Fishing Vessel Safety Act of 1988 (USCFVSA) required the implementation of comprehensive prevention measures for all fishing vessels in offshore cold waters, including immersion suits, survival craft (life rafts), EPIRBs and crew training in emergency response and first aid.
Purpose. To examine the effectiveness of the measures instituted under the USCFVSA in reducing the high occupational fatality rate (200/100,000/year in 1991-1992) among Alaska's commercial fishermen
Method. Comprehensive surveillance for commercial fishing occupational fatalities was established by our office during 1991 and 1992 in Alaska. Demographic, risk factor, and incident data for 1991 through 1996 were compiled and analyzed for trend. Findings. During 1991-1996, there was a significant (p=.002) decrease in Alaskan commercial fishing-related deaths, from 36 in 1991 to 35 in 1992, 22 in 1993, 11 in 1994 (artificially reduced number due to closure of crab fisheries that year), 18 in 1995, and 24 in 1996. While man-overboard drownings and vessel-related events in crabbing (often conducted far offshore and in winter) have continued to occur, marked progress (significant downward trend, p<0.0002) has been made in saving lives of those involved in vessel-related events:
| Year | Vessels Lost | Persons on Board | Persons Killed | Case Fatality Rate* |
| 1991 | 39 | 93 | 25 | 27% |
| 1992 | 44 | 113 | 26 | 23% |
| 1993 | 24 | 83 | 14 | 17% |
| 1994 | 36 | 131 | 4 | 3% |
| 1995 | 26 | 106 | 11 | 10% |
| 1996 | 38 | 114 | 13 | 11% |
*Case Fatality Rate = (number killed/number at risk) x 100 percent
Conclusions. Specific measures tailored to prevent drowning in vessel capsizings and sinkings in Alaska's commercial fishing industry have been very successful so far. However, these events continue to occur, placing fishermen and rescue personnel at substantial risk. Additional efforts must be made to reduce the frequency of vessel events, enable similar progress in crabbing fisheries, and to prevent man-overboard events and drownings associated with them.
Methodological Criteria for Evaluating the Effectiveness of Accident Prevention Programs—Shannon HS, Robson LS, Guastello SJ
Despite the large volume of occupational safety literature, which is often descriptive or examines physical risk factors, accident prevention programs are often not evaluated. Moreover, the quality of evaluations that do take place has been criticized. Commonly occurring methodological weaknesses include: not taking into account length of exposure when calculating accident rates; the use of proxy measures for outcome measures, without validating the use of the proxy; not monitoring large workplace/societal changes which could influence accident rates, especially in the case of before-after designs. There are, no doubt, many reasons for this situation--among them is a lack of training in evaluation in those responsible for occupational safety within organizations. Yet rigorous assessments of our efforts in promoting safety are essential if we are to avoid using limited human and financial resources on ineffective (or even harmful) measures.
In the following, we describe a set of methodological criteria, derived in order to assess the quality of reported evaluations of safety interventions. These criteria can also serve as a reference for those planning such evaluations. Identification of the criteria was based on an examination of the safety, health promotion, program evaluation and research design literatures. Eight areas were deemed relevant:
1. Program objectives—e.g., Did the program objectives provide a measure against which outcomes could be compared? 2. Program design—e.g., Was experimental, quasi-experimental or non-experimental design used? 3. Program participants—e.g., Was selection bias considered? 4. Description of the safety intervention(s)—e.g., Was exposure to additional societal/workplace factors considered? 5. Measurement of program implementation— e.g., Was description/measures of compliance of program recipients provided? 6. Measurement of program outcome—e.g., Were true outcome measures provided? 7. Analysis of results— e.g., Was sample size/statistical power considered? 8. Conclusions— e.g., Were conclusions supported by the analysis?
Some of the criteria are applicable to field experimentation in general, but emphasis has been given to issues in the safety field.
Baseline Safety Measures in the First Year of the New England Safety Project—Halperin K
In the fall of 1996 the United Brotherhood of Carpenters Health and Safety Fund (UBCHSF), working with researchers from the Johns Hopkins School of Public Health, was awarded a three year grant from NIOSH in injury prevention among union carpenter contractors in New England. The aim of this project is to demonstrate that injuries to carpenters can be measurably reduced by the implementation of written health and safety programs by small construction contractors. The methodology is a controlled prospective trial among small firms employing unionized carpenters. Twenty-two small-to-medium sized carpenter contractors (with average annual employment approximating 10 to 50 carpenters) in the Boston, MA, Hartford, CT and Providence, RI areas were recruited into the treatment group. The control group consists of 50 similar (in size and types of work) contractors in upstate New York. The geographical separation should minimize contamination. The treatment group will implement, with UBCHSF help, a written health and safety program. Using OSHA-required injury logs and workers'compensation data, this project seeks to demonstrate how the implementation of written health and safety programs by such contractors can lead to a measurable reduction in the rate of occupational injuries. Implementing this intervention over a two-year period is expected to result in a reduction in OSHA-recordable injuries, lost workday injuries, and days lost from work due to injuries, and in reduced workers' compensation experience modification rates.This project will also seek to demonstrate that the implementation of health and safety programs will result in measurable changes in workplace safety and health practices by participating contractors. Four "sentinel" safety and health practices are being measured through direct observation in actual workplaces: use of eye protection, use of ground fault circuit interrupters, use of hearing protection, and fall protection. Contractor and worker interviews are being used to gather information about the implementation of the intervention, and particularly about the perceptions of these two groups regarding the degree of acceptance of the intervention.The control group will be given an initial workplace visit with occupational safety and health advice (comparable to the baseline visit for the treatment group) and the promise of help in installing health and safety programs in their companies at the end of the two year period. Periodic measures of the sentinel practices will occur for both the control and treatment groups throughout the period, and the OSHA logs and experience modification rates collected before and after the period.Baseline information about workers' compensation experience modification rates, OSHA recordable injury rates, and occupational safety and health practices as represented by observation of the four "sentinel" practices will be presented for both the treatment and control groups. The next steps of the project will be outlined.
A Meta-Analysis of Long-Term Results for a Behavior-Based Method to Reduce Workplace Injuries—Krause TR, Sloat KCM, Seymour KJ
Research and applications of behavioral principles have established behavior-based safety initiatives as effective, proactive, and long-term solutions to occupational health and safety challenges. This study adds to the existing literature a longitudinal evaluation of an injury prevention process implemented in real-world industrial settings where the behavioral causes of injury varied from one site to the next. These highly individual initiatives shared 4 components: specification of critical behavior, observation/data collection, feed-back, and problem-solving. Up to four years' injury data from sixty companies representing the chemical, petroleum, paper, lumber, electronics, transportation, food, and other industries who implemented this behavior-based safety process were examined. The average reduction from baseline amounted to 29% after 1 year of observation and feedback, 46% from baseline after 2 years, 50% after 3 years, and 59% after 4 years. Results did not depend on union status, industry, or baseline recordable rates and perceptions of success within the organizations concurred with these findings.
Preventing Accident Repetition—Developing Effective Multi-Partner Teams for Reducing Workplace Injuries—Gallie KA, Jessup BA
In Part one we provided background information on workers having experienced 20 or more workplace injuries reported to the British Columbia, Canada Workers' Compensation System (n = 15,042; X = 25.4 injuries/worker per working lifetime, range = 20 - 91 injuries; SD = 10.9). Our analyses showed that this group had an increased chance of permanent disability and risk of fatality. These workers reflected the “Tip of the Iceberg” and if effective proactive measures were not quickly taken there would soon be a larger number of workers facing similar workplace suffering. As example, our records showed that an alarming number of workers experienced at least 1 injury/year.
Given the need to intervene with workers most likely to benefit from our interventions we widened our selection criteria to include those workers with at least 5 injuries/5 years. This latter group represents 80% of our intervention group.
Communication with our occupational health and safety stakeholders including employer, worker, union/shop stewards, health and safety committees as well as in-house personnel (i.e., prevention officers, adjudicators) shaped development of the three injury reduction interventions we developed. These interventions were guided by two main goals: To show stakeholders the win-wins of adopting Health and Safety as their common goal and the role each could take (i.e., employer and disability/case management etc.) and focus their efforts on preventing accidents before they happened.
Three main interventions were developed. The first was a personalized letter from our President/CEO requesting all BC employers and workers with 20+ or 5 injuries/5 years to join in a partnership with WCB in bringing down the number of workplace injuries and provided guidelines on how this could be done. The second intervention consisted of a Preventing Accident Repetition Workshop offered to workers currently away from their job due to a workplace injury. This one day workshop included information on the causes of workplace accidents and how to prevent them, proper body mechanics and work simplification, stress and coping strategies, as well as a personal injury profile analysis and commitment towards safety . There was an overall psychosocial focus with coverage on the legal responsibilities towards Occupational Health and Safety. The third intervention comprised meetings with employers and their health and safety/union representatives where WCB representatives presented a Health and Safety analysis on that workplace including their injury profile (i.e., type and mode of injury, body region, occupation group, cost and lost person hours) as well as suggestions/guidance on establishing/maintaining an effective and comprehensive occupational safety program. Results of each intervention and conclusions will be discussed.
Work-related Injury in a Rural Emergency Department (ED) Population—Williams JM, Furbee P, Dirk S, Higgins D, Prescott J
Methods. An emergency department-based injury surveillance system (EDBISS) was used to collect injury data on all ED patients seen over a one year period. A patient was classified as injured if their record contained an ED log injury code, an ICD-9 N-code between 800 and 995 in any diagnostic field, an E-code, or an entry in the trauma registry. An injury was considered work-related if the patient reported that the injury had occurred while at work. Descriptive analysis of the data was performed.
Results. Work-related injuries accounted for 1539/12,321 (12.5%) of all injuries. The mean age of patients injured on the job was 33.8 years (range, 16-77) compared to a mean age of 27.7 for all injured patients. Males accounted for 1026/1537 (67%) of the work-related injury visits compared to 57% of all injury visits. The most common mechanisms of work-related injuries were: overexertion ( 313 or 20%); cut or pierced by sharp implements (248 or 16%); falls (250 or 16%); struck by object (202 or 13%) and transportation- related injuries (71 or 5%). Sprains and strains were the most common type of injury sustained (415 or 27%), followed by wounds to upper limbs (283 or 18%), contusions (182 or 12%), and fractures (151 or 10%). Of the 1,539 patients presenting with occupational injuries, 178 (12%) presented to the ED via EMS. 1,401 (91 %) were treated and released from the ED and 136 (9%) were hospitalized. The mechanisms of injury that most commonly resulted in hospitalization included struck by an object (28 or 21%), transportation (26 or 19%), falls (27 or 20%), crushing mechanism (13 or 10%), and machinery (20 or 15%). Of those requiring hospitalization, 132/136 (97)% were male and the average length of stay was 4.4 days. Four of the hospitalized persons died of their work-related injuries. Known medical charges incurred by patients injured at work were as high as $62,622. The average charge for those treated and released was $273 dollars; the average charge for those who required hospitalization was $10,910.
Conclusions. Occupational injuries contribute significantly to the overall incidence of injuries seen in our emergency department and are responsible for tremendous medical charges each year.
Work-Related Visits to Emergency Departments in the United States, 1995—Stussman B
Introduction. It is well known that a significant portion of health care utilization occurs for occupational injuries. Care is sought in a variety of settings including physician offices, clinics, employer health units, and emergency departments (EDs). An injury resulting in a visit to the emergency department may be more severe than one ending up in another setting. This paper examines various aspects of emergency department visits resulting from work-related injuries. An estimated 96.5 million visits were made to hospital emergency departments in 1995. Of these, 4.8 million visits were work-related. This paper examines the extent of ED utilization for injuries that occur while working. Data are presented on the most frequent causes of injury, the types of injury, diagnostic services and procedures, and demographic characteristics of the patient.
Purpose. To describe work-related visits to emergency departments in the U.S.
Method. The data presented were collected from the 1995 National Hospital Ambulatory Medical Care Survey (NHAMCS). The NHAMCS is a national probability survey of visits to emergency and outpatient departments of non-Federal, short-stay and general hospitals in the U.S. Sample data were weighted to produce national estimates. In 1995, there were 391 emergency departments that provided data for the survey. Hospital staff were asked to complete Patient Record forms for a systematic random sample of patient visits occurring during a randomly assigned 4-week reporting period. The number of Patient Record forms completed for EDs was 21,911, of which 1,058 were work-related. Work-related visits were defined as visits for injuries that occurred while the patient was engaged in work-related activity on or off the employer's premises. Data were collected on up to three external causes of injury, up to three physician's diagnoses, procedures provided, and diagnostic and screening services ordered or provided.
Findings. Of the 37.2 million ED visits that were for injury, nearly 13 percent were related to work (4.8 million). One-fifth of injury-related ED visits for persons 18-64 years were related to work. Open wound of finger was the leading diagnosis for work-related ED visits. About seventy percent of visits were made by males (73.8 percent). The leading external cause of injury for work-related ED visits was overexertion and strenuous movements. Wound care was provided at 33.7 percent of work-related visits, and 8.0 percent had orthopedic care provided . Some form of imaging was performed or ordered for 64 percent of visits, with extremity X-ray being the most frequent (29.0 percent). Three-quarters of work-related ED visits had worker's compensation as an expected source of insurance. The work relationship was unknown for a quarter of injury-related ED visits.
Conclusions. The average total cost for an emergency department visit has been found to be $209, totaling a billion dollars annually for work-related visits. Work-related visits comprise a small but significant portion of the health care provided in emergency departments. It has been estimated that 30% of injuries to the hands and feet could possibly be prevented by the wearing of appropriate personal protective gear in the workplace. Since open wound to finger and contusion of lower limb were the top two diagnoses for work-related emergency department visits in 1995, determining appropriate preventive action may result in financial savings for industry and health care systems.
Surveillance of Work Injuries Using Hospital Discharge Data— Trent RB
In a growing number of states, records of hospital stays, hospital discharge files, are coded for cause of injury with one or more E-codes from the International Classification of Diseases, Ninth Edition. Discharge records also include information on the patient, treatment, outcome, and charges. When the payer is coded as workers compesantion the condition can be considered work-related as determined by the clinicians assessment of the patient's history. Using 1995 discharges from California s 587 acute care hospitals, we identified nonfatally injured patients age 16-64 where the expected payer was workers compensation (N=7,622, or 6%) and other payers (N=111,081, or 94%). We compared work versus nonwork patients and found that work injuries are distinct from nonwork injuries. Work injuries were 6 in 10 of all machine injuries, 5 in 10 of electrical injuries, and about 2 of 10 injuries due to explosions, struck bys, and fire/flames. We confirmed this pattern when we looked at fatal work injuries (using modified NTOF methods) and found that work injuries comprised the majority of injuries due to machines, explosions, electricity, and struck bys.
Within broad injury types, work injuries may be common for some specific etiologies and not others. For example, falls are the most common hospitalized injury, comprising 9% of work injuries and 6% of nonwork injuries. However, some specific types of falls are much more likely to be work related. Examples are falls from scaffolds (56% work related), ladders (23%), buildings (18%), and to a lower level (17%). In contrast, there are very few work related falls from wheelchairs (3%) or playground equipment (2%).
Hospital discharge data have several strengths for work injury surveillance: (1) one can document serious nonfatal work injuries in a number of states using a passive, uniform system, (2) for severe, nonfatal injuries, hospital discharge data are likely to be highly specific and sensitive, (3) differences between work and nonwork injuries can be examined, and (4) some specific types of injuries are so often work related (e.g., scaffold falls) as to be useful sentinels for work injury problems.
Weaknesses of hospital discharge data for work injury surveillance: (1) injury event narratives are not computerized or even centralized, (2) unlike the ANSI Z16 system, E-codes are not designed specifically to document work injuries, and (3) industry and occupation is not available, precluding the development of rates specific for risk.
Conclusion: hospital discharge data offer a useful, easy way to document severe work injuries, contrast them with nonwork injuries, and locate sentinel injuries. Hospital discharge data lack detail on work injury circumstances, occupation, and industry and therefore cannot replace analyses based on existing systems such as physicians reports, absences from work, or workers compensation claims.
Occupational Injury Surveillance Using the Alaska Trauma Registry— Husberg B, Conway G
Introduction. From 1980 to 1989 Alaska had the highest rate of any state for occupational fatalities, 34.8 deaths per 100,000 workers per year, five times higher than the U.S. average of 7.0 deaths per 100,000 workers per year. The majority of the occupational fatalities occurred in the fishing, logging, and aviation industries. The Alaska Trauma Registry (ATR) provides a population-based tool for occupational injury surveillance for moderate to severe injuries. Each hospital in Alaska participates in contributing data to the ATR.
Methods. To be included in the ATR, patients either have to be admitted to a hospital, transferred from an Emergency Department (ED) to a facility with a higher level of care, or declared dead in the hospital's ED. Data for the ATR is collected retrospectively from medical record charts and the information is sent to the Alaska Department of Health and Social Services, Division of Public Health, Section of Community Health and Emergency Medical Services to be compiled into the ATR. Data fields related to occupational injury surveillance go through additional cleaning and coding by personnel at the National Institute for Occupational Safety and Health, Division of Safety Research, Alaska Field Station. Cause of the injury can be examined via the ICD-9-CM "E-Code" and the injury description narrative.
Results. Currently the ATR has information for 20,842 cases from January 1991 through December 1995; 2,421 (12%) of these injuries were classified as occupational injuries. There are 40 fatalities among the occupational injuries (1.7%). Commercial fishing, construction, and logging led the industry categories for number of occupational injuries. The most common causes of injuries in the fishing industry were caused by machinery onboard vessel (74), fall between levels (38), and cuts (13). There are a wide variety of machines used on fishing vessels: a hydraulic lifting platform known as a crab pot launcher was the most common machine mentioned in causing injury. In the construction industry, different types of falls lead all causes with falls from or out of building or other structure (64), fall on or from ladder (43), and fall on or from scaffolding (36). The top three causes in the logging industry were being struck by an object (94), falls (33), and machinery (28).
Conclusions. The main causes of occupational fatalities in Alaska have been drowning (primarily in commercial fishermen) and trauma related to aircraft crashes. With the emphasis of the ATR primarily on non-fatal injuries, we have identified further areas of study for crab fishing injuries, falls in construction, and logging-related injuries in Alaska. As it is population-based, ATR data can be used to calculate injury incidence rates. The use of rates in injury surveillance will make the ATR data useful for industries to prioritize areas for injury prevention.
Using Injury Surveillance and Workers' Compensation Data to Facilitate Injury Prevention Activities—Pastula S, Reeve G
During the past five years, the authors have directed an effort to design a near real-time data analysis system for occupational injuries in the North American locations of a major automobile manufacturer. The major obstacle for the project was obtaining conceptual support from certain areas within the company that did not initially accept the premise that such a data system was really needed in order to reduce injuries. This initial lack of support was based on a fundamental misconception about the utility of injury incidence data in the real-time environment of large manufacturing plants. The rationale of this misconception is best illustrated by the following statement: “You don't need a sophisticated data system to reduce injuries. You go out on the shop floor, look at the jobs, make a list of the “bad” jobs, and then fix them.” However, in a plant with 3,000 to 5,000 workers, the number of “bad” jobs always exceeds the available capacity to fix them. Therefore, timely plant-based injury surveillance is critical to correctly prioritize the redesign of jobs that can or do cause injuries. Injury surveillance must include not only data about rates and types of injuries, but also cost information for all occupational injuries regardless of OSHA recordability. In addition, surveillance must continue after the “bad” job has been redesigned in order to determine: whether the changes have decreased the injuries of initial concern; and, did not result in a subsequent increase in injuries of a different type.
Several examples which illustrate the need for timely injury surveillance data in the manufacturing environment will be presented. These examples include: the occurrence of a new set of injuries following a process change that was made to solve a specific injury problem; an effective use of injury cost data to facilitate a product design change; and, a rapid evaluation of a work-hardening program which prevented its company-wide implementation.
Fatal and Non-Fatal Incidents Associated with Forklifts and Other Powered Industrial Vehicle Incidents—Collins JW, Baker SP, Smith GS, Kisner SM, Landen DD, Warner M, Johnston, JJ
This research examines the circumstances of work-related injuries and fatalities involving powered industrial vehicles (PIVs), which include forklifts or other mobile power-driven vehicles used to carry, push, pull, lift, or stack material. Descriptive analyses were conducted on 946 PIV-related fatalities in the National Traumatic Occupational Fatality (NTOF) surveillance system from 1980 through 1993 and 916 incidents in 54 U.S. automobile manufacturing plants from July 1989 to June 1992. The NTOF surveillance system provides data from death certificates from the 50 states, the District of Columbia, and New York City. Death Certificates are collected for persons 16 years of age and older who died of external causes and for whom the certifier indicated that the fatality was associated with an injury while on the job. The automotive surveillance system is run jointly by the medical and safety departments in the plant and includes information on employee characteristics, characteristics of the workplace and injury-producing event, and description of the injury.
The three most common types of fatal incidents in the NTOF database involved PIV overturns (22%), pedestrian struck by PIV (20%), and decedent crushed by forklift (17%). The highest frequency of fatalities by industry division occurred in manufacturing (33%), transportation, communication and public utilities (16%), and construction (14%). The highest fatality rate by industry occurred in wholesale trade, mining, and agriculture/forestry/fishery. The highest forklift-related fatality rates by occupation occurred to laborers and transport operators.
The 916 PIV-related incidents in the automotive surveillance system resulted in 913 injuries and three fatalities. Of the 913 injury incidents, 41% (372 of 913) of the injuries resulted in an employee missing work. The 372 lost workday incidents resulted in a total of 22,730 lost workdays, an average of 61 days away from work per lost workday incident. The three most common types of injury incidents in the automotive manufacturing surveillance system involved pedestrians being struck by PIVs (n=35%), PIV collisions with fixed objects/other PIVs (n=16%), and mounting/dismounting PIVs (15%). Recommendations are presented with regard to the factory environment, vehicle safety features, and driver and pedestrian training for reducing the risk of powered industrial vehicle incidents.
The Incidence of Injuries Involving Robots in a Large Manufacturing Company—Pastula S, Howe J, Smitt R, Reeve G
Robots are in widespread use in the automotive manufacturing environment. Robots work side by side with people in the plants, and the robots are programmed and repaired by these same workers. In recent years, there has been a growing discussion of the possibility to change safety standards for robots. This discussion has been driven in large part by a presumable absence of reports of severe injuries associated with robots. The purpose of this study was to identify robot-related injury cases among workers in a heavy manufacturing environment. In addition, the actual or potential severity was evaluated for each case.
The Ford Motor Company maintains an Occupational Health and Safety (OHS) System, which captures all work-related injuries seen at the 55 plant medical departments in the U.S. There is also a reporting tool that allows near real-time surveillance of these injuries at the plant. At the time the study was conducted, the database contained 390, 518 First-Time Occupational Visits (FTOVs) for the period June 1993 through August 1996. Of these, 200, 985 (51%) were OSHA-recordable. Text searches of the patient statement of the incident were determined to be the most accurate way to capture injuries involving robots. The first search of the database for any visit with the word "Robot" embedded in the text identified 695 injuries, 44% of which were OSHA recordable, and 53 cases had days away from work. A manual review of these cases to determine the validity of these injuries as involving robots found that a majority of the employees injured were doing manual work normally performed by the robot; working on the robot or equipment related to the robot; or, working adjacent to the robot.
Nineteen injuries actually involved being struck by the robot or pinched by robot tooling. These workers could have easily been crushed or hit hard enough to cause severe or fatal injuries. There were a significantly higher number of cases involving injuries to the eye. These were primarily due to foreign bodies in the eye caused by welding robots throwing sparks. Many of the employees reported wearing safety glasses when their eye was injured.
While the incidence of robot-related injuries is low, the potential injury severity of each of these incidents is troubling. In several of these cases had the robot rotated or extended another couple of degrees, the worker could have been killed. Under slightly different circumstances the incidents with foreign bodies in the eyes (sparks) could have caused permanent blindness. The findings of this study do not support a relaxation of current robot safety standards.
Relationship between Hand-Arm Vibration Exposure and Hand-Arm Vibration Disorders among Workers Using Impact Wrench in Automobile Assembly Lines—Jeung J-Y, Lee J-Y
This study is conducted to clarify the relationship between hand-arm vibration exposure and hand-arm vibration disorders among workers using impact wrench in automobile assembly lines.
Hand-arm vibration exposure was assessed using ISO/DIS 5349. Musculoskeletal disorders of occupational origins were assessed by the pain of hand, wrist, elbow, shoulder and low back, neck. Hand-arm disorders of vascular and sensorineural component were assessed using Stockholm Workshop hand-arm vibration syndrome classification system. The change of sensorineural component was assessed using vibration perception threshold of 8, 16, 31.5, 63, 125, 250, 500Hz.
Dominant acceleration and impulsiveness were 4.42m/sec 2 and 71.12m/sec2 respectively and dominant frequency was 369.19Hz. Daily exposure time and year of working were 0.79 hours and 2.86 years respectively.
Prevalences and odds ratios of pain by the site of musculoskeletal system were 71.1% and 10.31 in the hand, 59.7% and 5.41 in the wrist, 53.0% and 3.65 in the elbow, 50.3% and 2.47 in the shoulder, 39.6% and 2.66 in the low back, 33.6% and 2.21 in the neck. Association between hand-arm vibration exposure and musculoskeletal disorders of occupational origin was observed. Symptom prevalences and odds ratios of vascular and sensorineural component were 51.0% and 5.24, 77.9% and 10.83 respectively. Association between hand-arm vibration exposure and vascular component, sensorineural component was observed.
Vibration perception threshold in right and left hand of the impact wrench workers by the frequency were higher than the controls. Increase of vibration perception threshold in 125, 250Hz was higher than any other frequency and this phenomenon was obvious for right hand highly exposed to hand-arm vibration. In the right hand, the trend of increase in vibration perception threshold by the frequency was observed by the increasing of sensorineural stage but this phenomenon was not observed in vascular stage. Correlation coefficient between lifetime vibration dose and vibration perception threshold by the frequency was high in all frequency and according to correlation coefficient, 63Hz was 0.68, 250Hz was 0.64, 125 was 0.62. Four hour equal energy acceleration, dominant acceleration and total tool operating time were highly correlated with vibration perception threshold of 63Hz, 125Hz, 250Hz. According to increase of lifetime vibration dose, prevalences of pain in hand, wrist, elbow, shoulder were increasing and symptom prevalences in stage of vascular and sensorineural component were also increasing.
Lifetime vibration dose will be useful index for the musculoskeletal disorders of occupational origin of hand, wrist, elbow, shoulder and the vascular and sensorineural component. Vibration perception threshold by the frequency was increasing by the increasing of lifetime vibration dose. Lifetime vibration dose will be useful to expect vibration perception threshold by the frequency and to establish preventive measures to vascular and sensorineural disorder deterioration. 63Hz, 125Hz, 250Hz will be effective frequency in measuring and following vibration perception threshold among workers using impact wrench in automobile assembly lines.
| NOIRS 1997 Agenda Directory Page |
| NOIRS 1997 Menu Page |
| NOIRS Home |
| NIOSH home page | CDC home page | NORA |
|
|
|