Abstracts of
Poster Sessions

ABSTRACT 1
“Baseline” SF-36 Scores of Drivers Involved in Injury Crashes Compared with the General Driving Population

Shanthi Ameratunga, M.B.Ch.B., M.P.H. 
Jennie Connor, M.B.Ch.B., M.P.H. 
Injury Prevention Research Centre, University of Auckland, New Zealand 

Robyn Norton, M.P.H., Ph.D. 
Institute for International Health, University of Sydney, Australia 

Rod Jackson, M.B.Ch.B., Ph.D. 
Department of Community Health, University of Auckland, New Zealand

Investigations of longer-term health outcomes following injuries often rely on serial measurements of health status in a defined cohort of injured individuals. While some studies include retrospectively assessed estimates of the pre-injury health status of participants, the distribution of the latter in relation to the health status of a “control” population not experiencing the index injury has received little attention.

We aimed to address this issue as part of a large population-based prospective study investigating the longer-term health outcomes following injury-involved car crashes in Auckland, New Zealand. The principal health status measure used in the study is the Short Form-36 (SF-36). The pre-crash health status of 270 “case” drivers involved in injury crashes was determined at recruitment to the study soon after the crash. Concurrently, the health status of approximately 270 “control” drivers representing a random sample of the driving public in the Auckland region was also assessed.

The comparative distributions of the baseline SF-36 scores in the “case” and “control” driver populations will be presented. The potential biases in estimating pre-injury health status following an injury and the methodological implications for measuring the burden of injury will be considered.

ABSTRACT 2
An Analysis of Hospital Related Charges Associated 
with Injuries Due to Automobile Crashes

Daryl R. Baker, M.P.H.
S. Ross Clarke, Ph.D.

This study analyzes factors contributing to hospital charges from injuries incurred in automobile crashes. Studies have clearly documented the relationship between seatbelt usage, injuries and hospital charges. The purpose of this study is identify the components related to variation in hospital charges of drivers involved in crashes.

Methodology

1995 data files from the Oklahoma Department of Public Safety (ODPS) and the Oklahoma State Department of Health (OSDH) were probabilistically linked to obtain a data set matching individual hospital information to individual crash information. The asymmetric nature of the data necessitated use of the natural logarithm of total hospital charges as the outcome variable in a regression analysis. The predictor variables included: 1) age, 2) gender, 3) seatbelt use, 4) alcohol use, 5) vehicle type, 6) vehicle damage (in dollars), 7) time of day (night vs. day), 8) location (urban v. rural), 9) day of the week, 10) weather conditions, 11) road conditions and 12) speed before the crash.

Results

Significant predictor variables at p-value < .0001 were: age, gender, speed before the crash, failure to use a seatbelt, alcohol usage. Location was also a significant predictor at p-value < .05.

Conclusions

A demographic profile of drivers involved in crashes shows that increasing hospital charges are associated with males, increasing age, urban settings, and lack of seatbelt use. Individual behaviors associated with increasing hospital charges are increased speed prior to the crash, increased vehicle damage and alcohol use.

ABSTRACT 3
A European Working Group on Post-Injury Levels of Functioning

E.F. Van Beeck 
M.L. Essink-Bot 
Department of Public Health, Erasmus University Rotterdam, the Netherlands 

C. Falck Larsen 
University Hospital, Copenhagen, Denmark 

S. Mulder 
Consumer Safety Institute, Amsterdam, the Netherlands

Background

Disabilities as a consequence of injury or disease are increasingly recognized as an important public health indicator. Information on disabilities is, for example, necessary to calculate composite public health measures like DALYs (disability-adjusted life years) and QALYs (quality-adjusted life years). As these measures are increasingly used in health care, they should be applied in the injury field as well. However, within the injury field, empirical knowledge on the incidence and determinants of disabilities is scarce and severity weights by level of disability are hardly available. Therefore, a European working group on post-injury levels of functioning has been established, which aims to improve the scientific basis for future calculations on the burden of injury at the population level.

Activities Of The Working Group

The working group brings together European researchers and clinicians with experience in measuring post-injury disability. Disciplines involved are injury epidemiology, public health research, health status measurement, traumatology and rehabilitation medicine. Each participant analyses and reports own data on the incidence and determinants of post-injury disabilities. Moreover, the working group prepares a state-of-the-art report on the strengths and weaknesses of available data and methods needed to quantify the impact of post-injury disabilities on public health. Topics included in this state-of-the-art report are: available systems of injury coding (eg ICD 10, AIS) in relation to functional outcome of injury patients, available methods to describe functional outcome of injury patients (eg SF-36, EuroQol, FIM, FCI), available data on post-injury disabilities (at different moments in time) and their probability of occurrence, and available methods to assess the severity of both the initial injury and their functional outcome.

At the conference, a summary will be given of the theoretical and empirical needs so far identified by the working group. This will be illustrated with data on the incidence and determinants of post-injury disabilities so far obtained.

ABSTRACT 4
Non-Institutionalized Medical Resource Use and Costs After Initial Emergency Visit: Results From a 5-Months Follow-Up Study

E. Birnie 
W.J. Meerding 
E.F. Van Beeck 
Department of Public Health, Erasmus University, Rotterdam, The Netherlands 

H. Toet 
P.C. Den Hertog 
S. Mulder 
Consumer Safety Institute, Amsterdam, The Netherlands

Introduction

Injuries have considerable effects on health status, medical resource use and costs. In The Netherlands, 1.1 million patients with injuries visit emergency departments each year, of whom 10% are hospitalized. We have developed a surveillance-based model on the costs of injury to monitor the epidemiology and costs of injury continuously, to support priority setting in injury control and to evaluate specific preventive measures. Data on the short-term and long-term resource use and costs outside hospitals, nursing homes and rehabilitation centers are hardly available. To fill this gap, we conducted a patient survey to collect data on non-institutionalized care after the initial emergency visit.

Methods

Included in the follow-up study were 5,766 patients with injuries who had visited emergency departments between July 1997 and October 1998. All had been sampled randomly from the Dutch Injury Surveillance System (LIS), a continuous registry of injury-related emergency visits. Excluded were patients who had died and patients suspected of suicide or self-mutilation. At 2 and 5 months after the initial emergency visit, patients were asked to report (questionnaires or interviews) on the following non-institutionalized medical resource use: GP care, outpatient care, physical therapy, home care, and pharmaceuticals. To obtain representative estimates of resource use, ‘crude’ data were adjusted for response bias by inverse probability weighting. Costs were calculated as resource use multiplied by costs per unit estimates.

Results

The resource use of 2,387 patients (at 2 months) and 1,588 patients (at 5 months), respectively, was analyzed. The 5 months cumulative costs were Dfl 661 mln and consisted of care provided by general practitioners (7%; 1.2 mln contacts), outpatient care (49%; 1. 1 mln visits), physical therapy (15%; 2.6 mln visits), home care (27%; 3.4 mln hours), and pharmaceuticals (2%; at least 10.8 mln ‘medication days’). Resource use and costs differed considerably by location and type of injury, external cause of injury, age and sex, e.g., women were responsible for 77% of the costs of home care whereas 59% of the costs of outpatient care were attributable to men.

Discussion

Non-institutionalized medical care (Dfl 661 mln or 30%) is responsible for a considerable share of the injury costs in The Netherlands (Dfl 2,179 mln (1997)). Resource use and costs vary considerably between patient groups and over time. The well-developed level of primary health care in the Dutch health care system may limit the generalizability of our findings. Resource use and costs at 9 months after the initial emergency visit are currently being analyzed.

ABSTRACT 5
The Value of Statistical Life in Transport Safety: A Meta-Analysis

Arianne de Blaeij 
Raymond J.G.M. Florax 
Piet Rietveld 
Erik Verhoef
Free University, Department of Spatial Economics, Amsterdam, The Netherlands

Accident costs are an important part of external costs of traffic. One of the components of accident costs is fatalities. In order to evaluate these costs, an estimate of the economic value of a statistical life has to be available. The aim of this poster presentation is to present an overview and analysis of the literature on the economic valuation of statistical life in road safety. The studies used in this paper are transport safety studies only.

This study will have an empirical follow-up, with the goal of finding the value of a statistical life (VOSL) for the Netherlands, making use of the preference approach. Until now, in the Netherlands the VOSL has been estimated by means of the human capital method. The general opinion among economists is that the most appropriate way to measure VOSL is with a stated or revealed preference measure.

We used the meta-analysis of Elvik (1995) who collected a substantial number of VOSL studies. Furthermore, we looked in the economic database Econlit and on the Internet. In this study, we give an overview of the studies available to us and made a database of these studies containing most information available in all the studies. Since it is never sure that all the relevant literature is found, we invite experts attending this conference to share knowledge on the available literature.

On the basis of the database, we did a descriptive analysis and a meta-analysis. The descriptive analysis gives already some intriguing results. For example, there are studies wherein the VOSL is more then 1,000 times the GDP per capita of that country in the same year.

We will make use of meta-analytic methods to determine which variables are most appropriate in terms of explaining the variance of the VOSL. A meta-regression was carried out with VOSL as the dependent variable. The independent variables are categorized in five groups. Firstly, there are the variables reflecting specific underlying causes. For this purpose we choose GDP per capita, and risk level. Secondly, there are the cause-effect variables. These are the dummy variables behavior, car and road, which are related to the way in which the valuation question is asked. Furthermore, there is a dummy reflecting if the safety good is a private versus a public good. Thirdly, there are the research design variables. Is the study a stated or a revealed preference study and is the purpose of the study political or scientific? Fourthly, there is the time variable, i.e. year of the data. The last group variable is the location variable (type of country).

ABSTRACT 7
Measures of Outcome Following Childhood Traumatic Brain Injury

Rob Forsyth, Ph.D., M.R.C.P.C.H. 
Marion Crouchman* , F.R.C.P.C.H.
Tom Kelly, Ph.D.
University of Newcastle, UK and *Kings College Hospital, London UK

Introduction

Childhood traumatic brain injury (TBI) remains the most important cause of childhood mortality and morbidity beyond the first year. A fundamental obstacle to research into primary prevention strategies remains the lack of simple measures of outcome. Adult approaches are not applicable to children because of dynamic relationships (i) between the injury and ongoing development (thus measures must relate to changing age-appropriate expectations of independence) and (ii) between the component domains (physical, cognitive, behavioural etc) whose relative contributions to the total morbidity vary with age at injury, interval since injury and injury severity. We compared two candidate global outcome scores for use in the context of childhood TBI.

Methods & Key Findings

The Child Health Questionnaire

The CHQ is a generic childhood HRQOL measure. CHQ profiles six weeks and six months after injury were regressed against quasi-pre morbid CHQ injury severity (ISS score and GCS) and injury-independent parameters of family functioning (Family Assessment Device and the Family Crisis Orientated Personal Evaluation Scales). Six-week scores correlated weakly with ISS but there was no correlation between the six month CHQ and either ISS or family-related parameters. Correlation with GCS was weak. The factors determining CHQ scores remain largely unidentified.

The King’s Outcome Score in Child Head Injury (KOSCHI)

The KOSCHI is a paediatric adaptation of the adult Glasgow Outcome Score (GOS) for TBI. It extends the five-point scale of the GOS to provide greater sensitivity at the mild disability end of the outcome spectrum. A total of eight categories are defined. Extended inter-reliability exercises revealed systematic inter-rater errors arising from ambiguities of category definition that highlight future development issues.

In an attempt to define the factors leading to assignment of category membership, a functional scoring system rating potential impairments (mobility, behavioural problems, communication, self care, etc) was regressed against category membership. The optimal model correlated highly with independent global assessments of category membership (giving confidence in construct validity) however there was a high residual SD of 0.7 implying that factors independent of these impairments were influencing global outcome assessments.

Conclusions

These findings highlight the challenge in developing global outcome scores for child TBI. In evaluating injury-prevention strategies, measures of handicap (extended to include effects of injury on the family, so-called “third-party handicap”) would seem most relevant. However injury-independent factors emerge as strong determinants of outcome at these levels after childhood TBI.

ABSTRACT 8
The Impact of Traumatic Brain Injury on Mortality

Barbara Gabella, M.S.P.H 
Chun-Lo K. Meng, Ph.D. 
Holly B. Hedegaard, M.D., M.S.P.H.
Colorado Department of Public Health and Environment

From 1991 through 1997, approximately 100 per 100,000 Colorado residents sustained a hospitalized or fatal traumatic brain injury (TBI) each year. In 1997, 737 (17%) Colorado residents died prior to admission, 3305 (77%) were discharged alive from a hospital, and 244 (6%), died in the hospital¹. To assess the impact of TBI on mortality, the state health department is comparing the nature and pattern of fatal outcomes for Colorado residents hospitalized with TBI to those outcomes in the non-TBI trauma population.

Methods

The state health department oversees the Colorado Trauma Registry, which collects data on all trauma-related deaths and acute care hospitalizations in Colorado. Data are obtained from three sources: the statewide hospital discharge data set, death certificates, and case abstracts from hospitals designated as a Level I, Il or III trauma center. Trauma discharges that occurred January through June of 1998 were linked with all death certificates where the death was due to any cause and the death occurred in 1998. A TBI was defined as a skull fracture or intracranial injury and was identified by ICD-9-CM codes of 800, 801, 803, 804, 850-854, comparable to the TBI case definition recommended by the Centers for Disease Control and Prevention².

Results

Of the 1589 persons hospitalized with a TBI during January through June 1998, 116 (7.3%) died as an inpatient. The age-adjusted death rate was 8.4 per 100 hospitalized TBI patients (95% confidence interval: 6.8, 10.0). Of the 11,503 persons hospitalized with a trauma but without a ICD code for TBI in any of the 15 discharge diagnoses, 181 persons (1.6%) died as an inpatient. The age-adjusted death rate was 1.5 per 100 hospitalized non-TBI trauma patients (95% confidence interval: 1.3, 1.7). These rates were not adjusted for severity, but when looking at the death rates for specific ISS groups, only the age-adjusted death rates for ISS 25 to 75 were significantly higher for the TBI group compared to the non-TBI trauma group. The rates for deaths occurring within six months of discharge were not significantly different for the TBI group compared to the non-TBI trauma group. Of those patients who died after discharge, half of the TBI group died within 3 weeks of their discharge. In comparison, half of the non-TBI trauma group died within 7 weeks.

These results suggest that TBI may impact mortality, especially among persons with multiple trauma. The next steps are to determine if the nature and manner of death varied between both groups and to better describe the duration of survival using a longer time period.

1. Injury Epidemiology Program; Traumatic brain injury in Colorado, 1998. Denver, CO: Colorado Dept. of Public Health, 2000. 
2. Thurman DJ, et al. Guidelines for Surveillance of Central Nervous System Injury. Atlanta: Center for Disease Control and Prevention, 1995.

ABSTRACT 9
Burden of Injuries in Pakistan

Abdul Ghaffar M.D., M.P.H. (Ph.D. candidate)
Department of International Health, Johns Hopkins University, Baltimore

Introduction

Injuries are gaining recognition as one of the major public health concerns in the world. WHO estimates that by the year 2020, the years of healthy life lost worldwide will be equal for infectious diseases and for injuries (WHO 1999). Despite this data, injuries are not appreciated as a public health problem by the health sector in Pakistan.

Limitation of knowledge, absence of reliable estimates of the current level of injuries and traditional views on health and disease are limiting factors to this understanding. To provide better estimates for burden of injuries in Pakistan, a national household survey {National Injury Survey of Pakistan 1997-99} was carried out to:

1. Establish the descriptive epidemiology of injuries. 
2. To estimate the Burden of injuries using summary measures of population health [DALYs and HeaLYs], and 
3. To suggest policy interventions based on the survey results and available literature on injury control.

Methods

The universe was all of Pakistan. A two stage, stratified sample, in which the primary sampling unit was a village for rural population and an enumeration block for urban population was carried out. The sample size was almost 30,000. A questionnaire was administered asking if anyone was injured in the last three months, and needed a consultation with a health care provider (formal/ informal) or could not do their routine work for half a day or more.

Results

The preliminary results reveal that the annual incidence of injuries is 41.5 per 1000 population. Motor vehicle injuries ranked 1st followed by falls and work place injuries. 76% injured were male and mean age was 27 years with the poor and uneducated at a higher risk. The average workdays lost due to an injury were 17.4 days, which resulted in almost 4 million person-days loss to the nation excluding years lost from deaths.

Conclusion

We conclude that legislation for gun control laws, and road safety in combination with health education programs are required in Pakistan.

ABSTRACT 10
Inconsistencies Between Hospital Discharge Data and Death Certificates for Inpatient Trauma Deaths

Holly B. Hedegaard, M.D., M.S.P.H. 
Chun-Lo K. Meng, Ph.D.
Colorado Department of Public Health and Environment

Background

The Colorado Trauma Registry collects data on all trauma-related deaths and acute care hospitalizations in Colorado. Data are obtained from three primary sources: a statewide hospital discharge data set, death certificates and case abstracts from designated trauma centers.

Methods

The hospital discharge data set includes up to fifteen ICD-9-CM diagnosis codes. Trauma records are selected based on having at least one of a specific list of injury-related ICD-9-CM codes within the first six diagnoses. Trauma records identified from the hospital discharge data set were matched to 1998 Colorado death certificates (regardless of manner of death) using patient name, date of birth, and date of death.

Results

In reviewing impatient deaths, differences were found between the data provided from death certificates and data from the hospital discharge data set. In 1998, 26,607 individuals were admitted for trauma to an acute care hospital in Colorado. Of these, 596 individuals died as an inpatient (overall mortality, 2.2%). For 355 deaths (59.6%), the underlying cause of death listed on the death certificate indicated trauma; for 241 deaths (40.4%), the underlying cause of death was a medical condition.

For those patients whose underlying cause of death was a medical condition, trauma was not mentioned on the death certificate as either an underlying or contributory cause for 172 deaths (71.4%). This occurred more frequently for patients whose underlying cause of death was cerebrovascular disease (30 of 35 deaths, 85.7%), chronic obstructive pulmonary disease (COPD) (22 of 29 deaths, 75.9%), and heart disease (46 of 80 deaths, 57.5%). Seventy-eight percent of these patients were age 65 and older. In reviewing the hospital discharge records for these patients, 43.6% had a primary admission diagnosis related to injury, 48.3% had an Injury Severity Score (ISS) of 9 or greater (indicating moderate to severe injury), and the length of hospital stay prior to death was 3 days or more for 71.5%. Only 21 of these patients (12.2%) had an autopsy.

The medical condition listed as the underlying cause of death on the death certificate was not mentioned in the hospital discharge data set for 33 deaths (13.7%). This occurred more frequently for patients whose underlying cause of death was stroke (15 of 35 deaths, 42.8%), COPD (I of 29 deaths, 13.8%) and heart disease (5 of 80 deaths, 6.3%).

Conclusion

These results suggest that one should use caution in relying on data from a single source. If data from death certificates were the only source of information, 172 of 596 deaths during hospital admission for trauma (28.9%) would have been missed, and the percent mortality for hospitalized trauma patients would have been calculated as 1.6%. If one relied solely on data from the hospital discharge data set, significant medical conditions that were ultimately listed as the underlying cause of death would have been missed in 33 deaths (13.7%). This study highlights the need for review of multiple data sources to identify trauma deaths and contributing medical conditions.

ABSTRACT 11
Using Healthy Life Years (HeaLYs) to Measure the Injury Burden in Pakistan

Adnan A. Hyder M.D., M.P.H., Ph.D. 
Department of International Health, Johns Hopkins University, USA 

Abdul Ghaffar M.D., M.P.H., M.H.A. 
Health Services Academy, Pakistan 

Richard H. Morrow M.D., M.P.H., F.A.C.P. 
Department of International Health, Johns Hopkins, University, USA

The Global Burden of Disease Study for 1990 has renewed interest in the use of composite indicators for the measurement of disease burden. Composite indicators combine the mortality and morbidity effects of a disease into a single number, which is usually measured in terms of time loss. The disability adjusted life year (DALY) and the healthy life year (HeaLY) are two such indicators. This study explores the application of the HeaLY concept to a developing world case study for measuring the burden of disease from injuries.

Pakistan is a developing country in South Asia with a population of 135 million and a GNP per capita of US$ 400. A National Injury Prevalence Survey has recently been completed in the country and preliminary data from that source is being used to evaluate the impact of injuries in the country in terms of healthy life lost. Premature mortality and disability consequences from all injuries over all age groups and both genders are being included. HeaLY losses and rates are being estimated.

The paper will present preliminary results at the meeting on the following issues:

• The HeaLYs lost from premature mortality in Pakistan due to injuries 
• The HeaLYs lost from disability as a result of injuries in Pakistan 
• The overall HeaLYs lost from injuries in the country 
• Comparison and relative rank of injuries compared to other communicable and non-communicable diseases 
• Comparison of the results with regional and global estimates of healthy life lost from injuries

This is the first time that such a nationally representative survey has been carried out on injuries in a developing country. This presentation is the first presentation of the HeaLY losses in Pakistan from injuries. A discussion of the utility of the method and the results will be presented.

ABSTRACT 12
Consequences of Traffic Injury in the Netherlands

Boudewijn van Kampen
Paul Wesemann
SWOV Institute for Road Safety Research, the Netherlands

In the Netherlands, consequences of traffic accidents are registered and measured in terms of the number of injury producing accidents and the number of traffic casualties. Additionally, a limited severity scale is used to divide these numbers of casualties in categories: fatal, hospitalized and other (lighter) severity.

The Dutch traffic safety policy (including safety improvement goals set for the near future) is based on these numbers, mainly regarding only serious outcome (fatal and hospital). These numbers do not reflect any kind of short or long-term consequences of injury, such as handicaps, impairments, other functional or social consequences or costs of injury. In view of the current trend of (the Dutch) society to care about the quality aspects of life, it is appropriate to focus on the consequences of injury and to develop an additional source to express their importance (number, type and severity), along with the more direct consequences of traffic accidents.

The final goal of the proposed study is to gain access to data describing these other types of consequences of injury sustained in traffic accidents, including their severity. Since it is expected that such data are not or only partly available, the study practically aims at producing an outline for a registration system of such data.

The study should start with a study of available literature, describing existing systems, scales, etc. concerning the coding of type and severity of consequences of injury.

It is to be expected that such studies have already been carried out, and this poster presentation invites experts to share knowledge on this topic.

The second stage of the study is to find sources that already contain the data or may be used as base to acquire the data that the study aims at (i.e. type and severity of injury consequences). Though these type of data would have to be primarily Dutch data, here again research experience gained from comparable activities in other countries will be most welcome.

ABSTRACT 13
Burden of Motor Vehicle Injury in Latin America

Kelly Ketchen, B.S.
Timothy Baker, M.D., M.P.H.
Johns Hopkins University, School of Public Health

Introduction

The “reducible burden” of disease or injury expands a concept introduced by the Pan American Health Organization in 1997. It does not simply measure the absolute burden of a health condition, but compares the burden in a given country to a “gold standard” attained in a reference country with similar resources. This paper measures the reducible burden of motor vehicle deaths in countries of Latin America having reasonable quality of death registration.

Sources of Data

Injury death data by age and type of injury for the countries of Latin America come from the latest Health Statistics from the Americas (PAHO). Morbidity estimates are based on World Bank “Daly” estimates. Numbers of motor vehicles were obtained from World Road Federation reports.

Methods of Analysis

Years of productive life (15-65) lost per 1,000 population by each age group from 1-65 are calculated and discounted at 3%. (See CDC (ypll) and World Bank (Daly’s). The years lost/1,000 are subtracted from the years of life lost/1,000 in each age group for the “gold standard” country (Argentina). The results are presented as reducible in Potentially Productive Years of Life Lost (RPPYLL) from motor vehicle injuries for selected countries of Latin America. To emphasize the magnitude of the problem, RPPYLL’s are compared to the number of labor force entrants. Data from Canada and the United States are presented for comparison.

Results

Results show a large and increasing reducible burden of motor vehicle injuries in the poorer countries. Rates of PPYLL/1,000 for all but one Latin American country and the United States are greater than those of Canada which has a high ratio of vehicles/1,000 population. The size of reducible gaps in burden of motor vehicle injuries should help countries to set appropriate priorities for motor vehicle injury control.

ABSTRACT 14
Underestimating the Burden of Construction-Related Injury

Pamela Kidd 
Susan Wojcik 
Mark Parshall 
Tim Struttmann

Background

In an attempt to assess the cost effectiveness of a safety training program in small construction companies (companies with ten or less employees), use of worker compensation claims did not adequately quantify the economic impact of injury. The study used a pre-test/post-test design to test the effectiveness of a training program consisting of six simulations (three for fall and three for back injury prevention) in decreasing injuries, claims, and costs. Worker compensation claims reports and self report data were reviewed for injury data.

Procedures

The simulations were developed using data obtained from a series of nine focus groups recruited through collaboration with the a state fund workers’ compensation carrier. The focus groups were audio-recorded. Data were transcribed from the audiotapes, entered into a qualitative data analysis software, and analyzed for thematic content.

Findings

Many elements that relate to the examination of cost effectiveness were identified through the focus group discussions of the impact of the costs of injury. These elements include safety practices identified to prevent construction-related injuries such as: pacing work and work load, paying attention to the task at hand, planning and coordinating work activities, sizing up job and task requirements, maintaining a clutter-free worksite, coaching co-workers or employees to do tasks safely, and communicating safety through: actions and words. Groups suggested several disincentives to submitting claims for minor injuries (e.g. beliefs that premiums will necessarily increase, or preferring to avoid medical attention or time off work unless absolutely necessary). Thus, it is likely that claims data underrepresent the full spectrum of work-related injury in small construction companies. Several other negative outcomes from injury (indirect costs) were identified that are important to explore from the perspective of the company. These include: (1) negative impact on the reputation of the company from being unsafe; (2) time lost in finding and training replacements; (3) compromised production schedules and quality if replacement workers are inexperienced or not motivated; and (4) possible loss of current or future contracts if work is not completed on time. Worker compensation claim reports did not address these indirect costs of injury. There were no follow-up data available from any data source from which to examine how the reputation of the company may have suffered as a result of the injury as well as how many contracts were lost due to failure to complete the job in a timely manner.

Conclusions

If failures to perform safety practices identified in this study were highlighted in injury claim investigations, along with a more detailed assessment of indirect injury costs, the economic ramifications of performing more safely could be communicated to company owners in a more realistic and meaningful way.

ABSTRACT 15
Guidelines for Assessment of the 
Indirect Costs of Non-Fatal Injuries in Denmark

Marie Kruse
National Institute of Public Health, Denmark

Summary

The paper assesses the data available for analysing the indirect costs of injuries in Denmark. The Danish ‘prevention register’ provides unique opportunities for analysing the consequences of injuries, consequences for the patient’s monetary situation and association with the labour market.

A priori the friction cost approach is applied. The paper will discuss the relevance of this in light of the data opportunities.

The paper will provide guidelines for further analysis using the available data.

Data

The unique Danish personal registration system allows for multiple analyses involving the ample data from the economic, socio-demographic and health related records of each person with a permanent address in the country. These data can be used for precise estimates of the indirect costs of traffic-, work- and home and leisure time accidents.

Results

The main results are

1. A presentation of the data and the opportunities provided by the data 
2. A discussion of the friction cost approach in this light 
3. Guidelines for further analysis

ABSTRACT 16
Comparative Analysis of Costs 
Among Motor Vehicle Crashes in Maryland

Joseph A. Kufera 
Patricia C. Dischinger 
Timothy J. Kerns 
Shiu M. Ho
National Study Center for Trauma and EMS, University of Maryland, Baltimore, MD

Linkage of existing data sources may provide valuable information regarding the epidemiology and economic impact of motor vehicle crash injuries in Maryland. In this study of 9886 hospitalized drivers, motor vehicle crash data from the Maryland Automated Accident Reporting System (1991-1995) were combined with same period costs from hospital discharge data (Health Services Cost Review Commission), using probabilistic linkage methods. Average costs were expressed as medians, and univariate comparisons between groups were analyzed by the Wilcoxon rank-sum test. Comparisons of costs were based on groups defined by demographic characteristics (age, gender), safety equipment (seatbelt, airbag), direction of impact (frontal vs. lateral), and police perception of alcohol or drug impairment. Costs were then transformed by natural logarithm and modeled by analysis of variance (ANOVA). The median hospital cost for all drivers was $3348 (range $0 - $301,961). Rankings indicated that unbelted drivers incurred the highest median cost ($3901), followed by airbag deployment ($3817), impaired drivers ($3702), and older drivers (age > 35 years, $3521). Belted drivers incurred the smallest median cost ($3095). Univariate comparisons between groups indicated that drivers who were unbelted, impaired, and older (each p<0.001) incurred significantly higher costs. Airbag deployment also resulted in higher costs (p = 0.01). Costs did not differ by gender or direction of impact. Multifactor ANOVA revealed significant main effects (p<0.10) on log costs for drivers who were unbelted, impaired, older, injured in a lateral crash and had a deployed airbag. Interaction of seatbelt use by point of impact was significant (p=0.08). After adjusting for other variables, unbelted drivers in lateral crashes ranked highest in mean log costs. These data validate the influence of seatbelt use as a major determinant of hospital costs. High costs were also incurred in airbag deployed crashes, as drivers were more likely to survive with serious injuries and require hospital services. For this presentation, data will be updated to include years 1994 through 1998.

ABSTRACT 17
Comparison of Methods of Coding Severity of Injury Using New Zealand Hospital Data

John Langley
Shaun Stephenson

Aim

To replicate part of a recent study by Sacco et al to compare five measures of injury severity using New Zealand (NZ) hospital inpatient data.

Method

The severity measures, considered were the Modified Anatomic Profile (MAP), Anatomic Profile Score (APS), Injury Severity Score (ISS), New Injury Severity Score (NISS) and ICD based Injury Severity Score (ICISS).

The first four measures were calculated using the ICD-9-CM to AIS conversion via ICDMAP-90.

Source data were all injury discharges, dead or alive, from NZ public hospitals from 1993 to 1996.

Two analyses were conducted. The first involved calculating the ICISS survival risk ratios and testing them with all the data. The second involved dividing the data in half and using the first half to calculate the ICISS ratios and the second half to test how accurate they were.

The measures were compared by fitting logistic regression models where the outcome was survival/death. The concordance of the fitted models was calculated and compared and calibration curves were plotted.

Results

ICISS gives the highest levels of concordance of any of the models for both analyses. MAP was the best performing AIS-based measure. Calibration results are mixed.

Conclusion

ICISS performs as well as, or better than, any of the alternative measures for assessing injury severity derived from ICD-9-CM.

Given that: 1) many countries do not use the CM version of ICD-9, 2) some are using or plan to use ICD- 10, or the CM derivative, and there is no ICDMAP program for ICD- 10; the utility of the ICD based Injury Severity Score (ICISS) based on ICD-10 needs to be assessed.

ABSTRACT 18
Health Related Quality of Life Following Injury 
is a Function of Time Since and Type of Injury

Ronan A Lyons 
Martin Heaven 
Sarah Jones 
Jo Sibert 
Stephen R Palmer
University of Wales College of Medicine, Cardiff, United Kingdom

Background

The 1995 Welsh Health Survey was a random sample of 28,095 community dwelling adults aged 18-98 years. Questions were asked about current health status using the SF-36 questionnaire and current and past use of health services, including injuries in the previous 3 or 12 months resulting in hospital attendance. Data were analysed to determine whether health status was related to the time since injury or type of injury.

Methods

Regression analyses were performed with the SF-36 variables as dependant, binary injury categories (attended emergency department for any reason in the last 3 months, last 12 months, or last 3 months, with a burn, fracture, laceration, head injury, poisoning or other injury) as explanatory variables and adjustment made for age and sex.

Results

Among the respondents 2149 people attended hospital with an injury in the previous 3 months and 4584 in the previous year. There were 484 fractures, 107 poisonings, 143 head injuries, 349 lacerations, and 56 burns among the 3 months attendees. All eight variables of the SF-36 were significantly lower in patients in the 3 month or 12 month attender categories versus never, with lower scores in 0 – 3 versus 3 – 12 month attenders. Patients with head injuries or fractures had lower scores than patients with lacerations or burns. In patients with head injuries and fractures the greatest problems were bodily pain (15 and 14.1 points lower), social function (14.3 and 9.9 points lower) and role limitations due to physical problems (16.1 and 15.9 points lower).

Conclusions

Health related quality of life in injury patients is related to both severity of injury and time since injury. Studies measuring the burden of injuries need to adjust for severity of injury and the natural history of recovery and disability.

ABSTRACT 19
Missed Educational Opportunity 
Contributes to the Burden of Injury

Ronan A Lyons 
Alison Maddocks 
Peter Barnes 
Mike McCabe 
Pam Nash 
Arun Midha
University of Wales College of Medicine, Cardiff, United Kingdom

Background

All injuries have multiple consequences, to the individual and society. Measurement of the burden of injury has tended to focus on the economic and physical consequences of injury related morbidity and mortality. However, other consequences, such as missed educational opportunities, are rarely if ever evaluated. About 1 in 3 children in Wales attend emergency departments each year, predominantly with minor injuries. The purpose of this study was to determine how much educational time was lost to children as a result of minor injuries.

Methods

We carried out a case control study in children attending three emergency departments with minor injuries on Sundays for three months in 1999. School attendance for each half day in the following week was recorded in cases and matched controls taken from the class register. The difference in attendance between cases and controls was compared with the one sample Kolmogorov-Smirnov test.

Results

During the period 440 cases meeting inclusion criteria were identified; 422 were matched with cases in 130 schools. Bruises (27.3%), sprains (26. 1 %), lacerations (17.3%), and minor fractures (13.5%) accounted for the majority of injuries. Cases missed an average of 1.45 days from school compared with 0.3 days in controls (p <0.001). Patients with fractures lost an average of 2.28 days compared to 1.31 for other injuries. Differences between cases and controls were unaffected by adjustment for age, sex, and deprivation score.

Conclusions

The combination of excess school loss following minor injury and the high incident of these injuries indicates that missed educational opportunity should be considered as another and important aspect of the burden of injury.

ABSTRACT 20
Development of the Pediatric Functional Capacity Index: Descriptive Results of the Scaling Exercise

L.G. Mangus
M.J. Hanfling 
C. Contant

Introduction

Development of the Pediatric Functional Capacity Index (FCI) is based on the methodology used by E. M. MacKenzie in developing the Adult FCI. Funded under a cooperative agreement with the National Highway Traffic Safety Administration, the FCI is intended to measure the societal impact of non-fatal motor vehicle injuries.

Methodology

Ten dimensions of functioning were defined: eating, excretory function, ambulation, hand and arm function, bending and, lifting, visual function, auditory function, sexual function, speech, and cognitive function. All of these dimensions but sexual function, which was to be a predictor of adult function, was defined for four different age groups; based on developmental differences in children: 1 to <2 years, 2 to <5 years, 5 to <10 years, and 10 to 17 years of age. Within each age-specific dimension, three to seven severity levels were developed to discriminate among important, measurable differences in function.

The next step involved category-scaling, an exercise in which outcomes are evaluated according to their relative importance, often employed by health services researchers to measure societal preferences for health outcomes. These data describe the responses of three groups of raters: teachers of handicapped children, teachers of non-handicapped children, and clinical pediatric experts front across the country who also assisted in defining functional capacity within the 10 dimensions of functioning. On a ruler where 0 is no limitation in function and 100 is the most severe level of function, Part I of the exercise involved assigning values to the levels within each dimension. In Part II, raters placed at 100 the dimension in which they felt its most severe level had the greatest impact on a child’s everyday living. They then placed the other dimensions on the ruler above 100 relative to the effect that the most severe level of function for each dimension would have on a child of that age. We compared the responses for the four different age groups, as well as the responses of the 3 rater groups.

Results

We found, as did MacKenzie, that the clinical experts’ mean responses for both levels and dimensions were generally lower than the teachers of non-handicapped children. There were fewer differences between the two groups of school teachers and between the teachers of handicapped children and the clinical experts. Differences were also found in ratings for the different age groups, with the responses for the functional limitation on the youngest age group most often rated as having less of an impact than it would for older children.

Conclusions

The average level and dimension ratings obtained in this scaling exercise will be used as level values and dimension weights for predicting functional limitations at 12 months post injury for each of the four pediatric age groups. In order to obtain a more global, societal response to the impact various functional limitations might have on a child, efforts are under way to administer the valuation assessment to four other groups of raters: parents with and without children with functional limitations and children with and without limitations. The FCI will provide a means of quantifying individual and societal consequences of nonfatal injuries, which have typically been measured by their economic costs.

ABSTRACT 21
The Role of Conjoint Analysis in the Measurement of Exposure to Injury Risk for Children and Parents

Paul McNamee

A number of weaknesses can be identified in currently used methods to measure the effects of injury prevention programs. For example, difficulties in establishing the true frequency and severity of non-fatal injury have meant that studies need to be very large in order to be able to detect differences in outcome. In addition, many evaluations confine their objectives to changes in risk prevalence (eg the wearing of cycle helmets) and fail to establish other exposures (eg cycling prevalence).

One area which has not been explicitly addressed in the literature thus far relates to injury prevention programs producing effects over and above those related to the avoidance of the final outcome, ie injury. For example, improved urban road safety measures may produce benefits for parents (in terms of peace of mind), children (independence/improved access to the environment) and others (less road congestion).

Valuation of these ‘process’ measures by children, and/or parents using traditional methods (Time Trade Off and Standard Gamble for Cost Utility Analysis, Willingness To Pay for Cost Benefit Analysis) poses difficulties. An alternative approach is to use the method of Conjoint Analysis (CA), which offers a way to estimate preferences and can be used to establish the relative importance of different attributes, in the provision of a particular service.

CA has a number of appealing features. First, it is particularly suited for the valuation of non-health effects. These are likely to be important in childhood prevention programs, where factors such as freedom of movement and peace of mind affect all users of any proposed program. Second, the technique can serve a number of different purposes; in particular, the values produced can be used as benefit measures in Cost Utility Analysis (Maas and Stalpers) and Cost Benefit Analysis (Roe et al).

In this paper an outline of the main stages required for CA will be presented: ie identification of attributes; specification of attribute levels; presentation of scenarios; preference elicitation; analysis of responses, along with consideration of the theoretical perspectives underpinning the method. The applied CA literature to date will also be reviewed, in order to draw implications over the applicability of the methods for future research of injury prevention programs.

ABSTRACT 22
Assessment of the Economic Burden of Injury:
Outcomes as Many as Methods

W.J. Meerding
E.F. van Beeck
Department of Public Health, Erasmus University Rotterdam, Netherlands 

S. Mulder 
Consumer Safety, Institute, Amsterdam, Netherlands

Introduction

A considerable number of cost of injury studies from different countries have been published in the last decade. These studies can help policy makers to set priorities and provide baseline information for evaluative studies in injury prevention. However, results from different studies need to be comparable in order to provide a solid basis for these purposes.

Methods

We searched Medline and the documentation centre of the Consumer Safety Institute for English language articles and reports on the economic burden of injury. An ultimate selection of twelve published from 1980 includes at least the main studies in this area. We analyzed these studies and investigated their methodology in-depth. The original results were adjusted for inflation and converted to 1997 US dollars. We computed per capita costs for making comparisons among countries and through time.

Results

Per capita medical costs due to injury ranged from $42-292. Within countries, this range was $69-303 (US) and $72-88 (Netherlands). These differences could be attributed to numerous causes, including methodological differences. This is illustrated by the following examples. Four studies included only hospital costs, while others included total healthcare. In the latter hospital costs were 60-99% of total medical costs. Two studies only included hospitalized patients, while others were based on A&E incidence or total incidence. Four studies included indirect costs of lost productivity. These costs accounted for 71-78% of total costs when the human capital method was used, and 42% when the friction cost method was used.

Conclusions

Huge differences in methodology lead to wide variations in findings on the economic burden of injury. Standardization of conducting and reporting cost of injury studies is needed before true differences in the economic burden among injury groups, among countries, and through time can be analyzed.

ABSTRACT 23
Towards an International Classification 
of External Causes of Injuries (ICECI)

Saakje Mulder 
Consumer Safety Institute, Amsterdam, The Netherlands 

James Harrison
Research Centre for Injury Studies, Flinders University of South Australia

Etienne Krug 
WHO, Geneva

Introduction

The External Causes section of the International Classification of Diseases (ICD) of the World Health Organisation has long been the main classification system for recording the circumstances and causal factors of injury. The ICD has considerable limitations for application for injuries, prominent among which are (1) poor discrimination for morbidity data (the ICD was developed for mortality statistics) and (2) internal inconsistencies (many of these result from attempting to compress several conceptual dimensions into a uni-dimensional classification). It also lacks much of the information and the user guidelines that are needed to facilitate practical application of a classification for injury surveillance and research. There is a need for a classification system that meets the requirements of injury control practitioners and is compatible with ICD. The task of developing the International Classification of External Causes of Injuries (ICECI) has taken up by the ‘WHO Working Group on Injury Surveillance and Methodology Development’.

Purpose

To develop an international classification system, including guidelines, principally for use to record morbidity data on the external causes of injuries that is useful for injury prevention and is as much as possible compatible with the 10th revision of the ICD.

Method

The classification and guidelines are developed in close collaboration with the world’s leading institutes in injury surveillance. In 1999 the classification and its guidelines will be tested in various settings and various parts of the world by reviewing, coding case scenarios, and by actual field testing. It is intended to launch the first version in November 2000. An important aspect of development is to establish mechanisms to maintain and update the system after the first version has been launched.

Results

The data elements included are ‘intent,’ ‘injury mechanism,’ ‘place of the injury event,’ ‘activity at time of injury,’ ‘objects involved,’ and ‘alcohol and drug use.’ More specific information on violence and transport can be collected by means of a module that will be part of the classification.

Conclusions

The results of the testing phase of ICECI will be presented. The final classification will be one of the most important tools for injury surveillance on morbidity.

ABSTRACT 24
Psychosocial Outcomes and Predictors of Recovery from Vehicular Injury
of External Causes of Injuries (ICECI)

Kathleen Read, M.S.W. 
Patricia Dischinger, Ph.D. 
Andrew Burgess, M.D. 
Timothy Kerns, M.S. 
Joseph Kufera, M.A.

With the increasing availability of modern restraint systems, more vehicle occupants are able to survive serious crashes. Aside from their physical injuries, psychosocial factors such as substance use, demographics, mental and emotional health status, occupation, and financial resources, among others, play a key role with regard to the extent of recovery.

As part of the Crash Injury Research and Engineering Network (CIREN) a multi-center study investigating all aspects of automotive crash safety, psychosocial data was collected on patients admitted to a Level I trauma center following injuries received in a car crash. This component of CIREN was begun in April 1998 and is on-going. All patients were initially interviewed at the R Adams Cowley Shock Trauma Center at the University of Maryland shortly after their admission to obtain baseline data and pre-injury history. Patients were again interviewed at six and twelve months post trauma to determine their physical and psychosocial outcome. To date, seventy-five patients have begun the interview process. Of these patients, 67% were drivers and 40% were males. Standardized measurements included the SF36, CAGE screens for alcohol, and drugs, Behavioral Risk Factor Survey, and questions regarding depression, anxiety, and post-traumatic stress disorder. In addition, culpability for the crash was also determined and related to pre-injury factors. An initial analysis has shown that the degree of impairment and return to functional status and outcome are closely related to the patient’s pre-morbid history. The presentation will address the psychosocial predictors of recovery in this population.

ABSTRACT 25
Cost Effectiveness Analysis of Alternative Smoke Alarm Systems

Maria Segui-Gomez 
Jeffrey L. Fellows 
Mark Stevenson

Background

In the US, a residential fire occurs every 70 seconds and a resident dies almost every 2 hours due to those fires. Deaths from residential fires remain a major public health challenge despite the fact that existing smoke alarms are effective in reducing the toll. The prevalence of smoke alarms in the US is estimated to be 94% (with a range from 78% to 99%), although it is estimated that approximately 20% of these alarms are inoperable because the battery has been removed or the life of the battery has been exceeded. In response to this problem, longer-life battery and hard-wired smoke alarms have been developed.

Methods

We assessed the average and incremental cost-effectiveness of three types of smoke alarms: standard alarms that use a one-year battery; alarms that use a 10-year lithium battery; and AC-powered (hard-wired) alarms. For our analysis, we developed a model that simulates the residential fire-related deaths and associated costs for a fictitious cohort of 10 million households (23 million persons) over a 20-year period. We quantified the health and economic impacts of each smoke alarm type using incidence and effectiveness data available in the literature. Costs were obtained from the literature or retailers. Methods used followed (to the maximum possible extent) the recommendations by the US Panel on Cost-Effectiveness. A sensitivity analysis and a threshold analysis were conducted for several of the input values.

Results

Health Outcomes, Net Costs, and Cost-Effectiveness of Alternative Smoke Alarm Systems

Conclusions

All three types of smoke alarms result in cost-savings benefits per each life saved. When the remaining years of life saved are quality-adjusted, the average cost-effectiveness ratios ranged from less than $3,000 to $14,000, well within the range of other public health-related interventions implemented in the US. The incremental cost-effectiveness shows that shifting from one-year batteries to ten-year batteries may actually result in more cost-savings due to their much higher effectiveness and only slightly higher costs.

ABSTRACT 26
Elicitation Methods in Defining 
Quality-of-Life Weights: Practical Implications

Maria Segui-Gomez, M.D., Sc.D.

Background

Direct scaling, time trade-off, and standard gamble are among the most commonly used elicitation techniques used by researchers. To this day, no consensus exists as to which method is better. Indeed, a recent review of quality-of-life preference weights used in health-related economic evaluations emphasized the lack of homogeneity among researchers regarding this issue as a major problem in this area. In 1996, the US Panel on Cost-Effectiveness in Medicine recommended standard-gamble based preference weights.

Direct scaling, the most commonly used elicitation method, is thought to provide very low estimates (i.e., weights too close to 0 – the worst health state). In the time trade-off, the respondent is asked to “give away” years of his or her own life in order to improve his or her health state. The standard gamble method is the only one that complies with the tenets of decision theory and cost-effectiveness analysis, although this is an argument criticized by some authors. In theory, the latter two provide more conservative estimates since the interviewee is asked to trade sickness for health by giving away “hypothetical” years of life. But these two techniques are also much more difficult to apply to general populations, given their conceptual complexities and the use of probabilities and trading-off of time in their formulations.

The survey available here was developed to compare the performance of these three elicitation methods, when they are used in relation to the Functional Capacity Index (for which development the direct scaling method was used).

Methods

The survey was developed using U-Titer 11, a web-based preference elicitation instrument developed at the University of San Diego. The health states selected for this survey include the ten most frequent injury-related conditions reported in a State Trauma Registry. The survey takes about 20 minutes to respond.

ABSTRACT 27
The Burden of Injury in Russia 
and the Newly Independent States

M. Sector, P.T., M.P.H. 
Timothy Baker, M.D. 
Kelly Ketchum
The Johns Hopkins School of Public Health

Introduction

At the end of the 1980’s, the USSR fragmented into the Russian Federation and the Newly Independent States (NIS). The full extent of the burden of injury in these countries has not been well recognized. Although it is known that overall death rates are high, the terrible burden of preventable injury deaths has not been fully documented. This study examines the burden of Potentially Productive Years of Life Lost (PPYLL) in these countries. The countries for which data is available were divided as follows: Group I low-income countries with GNP per capita $330-$1,340: Azerbadjan, Kazakkhstan, Kyrgystan, Moldova, and Tadjikistan. Group 2 lower middle-income countries, GNP $2,230-$3,330: Latvia, Lithuania, Estonia, and the Russian Federation. We also compared their PPYLL to other countries and compared the Russian Federation to the USSR in 1987.

Sources Of Data

Injury mortality rates for 1994-5 were derived from the World Health Statistics Annuals; 1995, 1996 World Health Organization (1996 is the most recent Annual). Injury deaths are classified using external cause codes from the International Classification of Disease-9th or 10th revision. Per capita GNP data was obtained from the World Development Report 1999 World Bank.

Methods Of Analysis

The human capital approach was used to estimate the value of lost productive capacity due to injuries. We calculated the PPYLL (15-65 years of age) per 1,000 population discounted at 3% (the value used in the World Bank Development Report on Health, 1993). We compared the PPYLL in the low-income group with a country of a similar economic level, China, to determine the reducible PPYLL. A similar analysis was made for the lower-middle income group using Thailand (similar economic level) as the comparison country. Lastly, we compared the USSR in 1987 to the Russian Federation 1995 in PPYLL.

Results

The low-income countries had the lowest productive years of life lost per 100,000. The Baltics and the Russian Federation had the highest rate of productive years lost. In comparison to China, the southern NIS had excessive rates. The Baltics and the Russian Federation had much higher rates than Thailand. The Russian Federation had greater PPYLL rates compared to the USSR 1997 rates.

Conclusions

There is a great disparity in injury mortality rates in the Russian Federation and the NIS. The NIS had the lowest PPYLL per 100,000, and the Russian Federation and the Baltics had alarmingly high rates. The Russian Federation and the Baltics would gain most from implementing public health intervention strategies to reduce the terrible economic burden of injuries. The other NIS should implement injury control efforts now to prevent increases in the burden of injuries as these countries develop economically.

ABSTRACT 28
A Swedish Model for Estimation of Injury Costs

Bengt Springfeldt 
Bjarne Jansson 
Karolinska Institutet, Stockholm 

Börge Bengtsson 
National Board of Occupational Safety and Health, Solna 

Bo Landgren

Objective

In a research project a program called EVA was tested to discern whether economic estimations of road construction works and traffic accidents can be applied to accidents of different types, primarily occupational injuries.

Methods

From the basic health cost calculations of the EVA program a new calculation program EVIS was constructed. In the model studies accidents caused by vehicles and accidents caused by falls were cost calculated. In another special study, injuries while using portable ladders were studied. The calculations were based on information on the accident cases from official statistics of occupational injuries as well as labor market insurance statistics. Information from the two sources were matched together in a new data base. The injury cases were divided in severity groups. The average material costs for health care expenses and loss of production were calculated by use of the economic production method. To these costs a human value was added, using the willingness-to-pay method, the same values that were utilized in the EVA project.

Results

In the method studies the material costs of occupational vehicle and fall injuries during the period 1991-93 have been estimated at an average of 38,000 and 33,000 US dollars per case respectively. The total costs inclusive of the human values were 220,000 and 150,000 dollars per case. Accidents with not more than 30 sick days cost about 33,000 dollars materially, and in total 46,000 dollars. By application of the EVIS method on all registered occupational accidents in 1995 the mean costs and sum costs have been calculated on different types of accidents. By special cross calculations the average costs have been estimated on accidents in different branches of industry, professions, main events and principal external agencies. Average material costs per case in the whole material are estimated to about 6,000 dollars and to 30,000 dollars. The material mean costs in agriculture and forestry were 10,000 and 8,000 dollars respectively, 60,000 and 54,000 dollars in total.

Conclusions

The method can be used for estimation of the effects for society when people, for different reasons are obliged to leave their work and be hospitalized for injuries: occupational accidents, school, sports and home accidents.

ABSTRACT 29
The Scope and Cost of Work-Related Injury 
in Victoria, Australia

Wendy Watson 
Joan Ozanne-Smith 
Monash University Accident Research Centre Melbourne, Victoria, Australia

Introduction

A study on the cost of injury in Victoria (1993/94) suggested that the proportion of work-related injuries compensated by the state worker’s compensation scheme is just under half of all injuries that occur in the workplace in Victoria. The aim of this study was to examine the relationship between compensated and non-compensated injury in Victoria in greater depth by comparing the epidemiology, the patterns of severity and estimated cost of these injuries.

Method

The total annual incidence of work-related injury was estimated through analysis of State health sector databases held at the Monash University Accident Research Centre. The incidence of work-related fatalities was derived from a recent reconciliation of work-related injury cases recorded in the Victorian Coroners Facilitation System, WorkCover’s Payments File and fatalities investigated by WorkCover’s Health & Safety Division.

Known treatment costs from WorkCover and private insurers plus an estimate of the cost of remaining cases based on previous work were used to establish an estimate of the total direct cost of workplace injury in Victoria. Indirect costs were established using a mortality cost model developed by the Australian Bureau of Transport Economics (BTE). Estimates of impairment from injury were derived using impairment fractions and probabilities of permanent and partial disability. The morbidity costs, associated with the estimated disability from injury cases, were calculated using the lost earnings and lost household production values contained in the BTE’s lost production model. An attempt was also made to quantify lost quality of life using values derived from court awards.

Results

It is estimated that, overall, there were about 61,000 medically-treated, work-related injuries in adults (15 years and over) in Victoria in 1996. Of these, some 56 resulted in death and over 5,900 were hospitalised. WorkCover made some form of compensation payment to over 27,000 injured workers in 1996, about 5,452 of whom required hospital admission. The total lifetime cost of work-related injury in Victoria in 1996 was estimated at $1.134 billion AUS.

A comparison with WorkCover data provided details of those areas in which the State’s compensation scheme had high or low coverage of the injured workforce. Non-hospitalised injuries, particularly those attending hospital Emergency Departments, rather than medical practitioners, are less likely to be compensated than hospitalised cases.

Conclusion

The estimate of the incidence of workplace injury in Victoria is comparable to that identified in an earlier study. Although less severe injuries are more likely to be uncompensated, the health care sector, injured workers and their families bear a considerable portion of the cost of workplace injury. Uncompensated work injuries should not be ignored when calculating the total burden of workplace injury to the state economy.

ABSTRACT 30
Indicators For Evaluation of Traffic Safety Measures

Paul Wesemann
SWOV Institute for Road Safety Research, the Netherlands

Decisionmakers on road safety measures increasingly strive for economic evaluations ex ante, whereby the social costs and effects are being confronted.

Sometimes they want the social profitability of alternative expenditure options to be assessed. This is especially necessary when the total budget for road safety policy has to be decided on, weighing its utility with other policy options. Social cost-benefit analysis is the appropriate method to this end. It requires all costs and effects, of the prevented casualties and injuries, to be monetized.

At other occasions decisionmakers select concrete measures, searching for the package with the maximum effect on safety and other stated social objectives, within the budget available. Social cost-effectiveness analysis is the appropriate method for this purpose. It does not require that the prevented casualties and injuries be monetized. They can be expressed in another appropriate unit as well, provided that for all evaluated measures the same units are being used.

SWOV is preparing a cost-benefit analysis of a comprehensive road safety programme in the Netherlands. It is presupposed that all effects can be quantified sufficiently reliable. However, it is uncertain whether the monetization can be realized timely and sufficiently supported by the decisionmakers. This requires an accepted method (at least by the decisionmakers themselves) with the data to match. Research of SWOV (together with the Free University of Amsterdam) into the valuation of human life is still on going.

If monetization is not yet feasible (for the time being) other safety indicators, in stead of or in addition to the monetary one, have to be considered. It seems advantageous to choose the same indicator(s) which are being used in other policy-sectors that strive for the prevention of accidents or medical care (e.g. public security, health care, prevention of accidents at home, in sports or at work). On the one hand this increases the chance that the indicator will be accepted; on the other hand it becomes possible to compare the cost-effectiveness of measures in several policy sectors. This is to some extent what cost-benefit analysis also is intended to accomplish.

A review of the (inter)national literature has begun in order to make an inventory of the safety-indicators that are being used in several policy sectors, to assess their degree of acceptance. The poster will present the preliminary results aiming at an exchange of knowledge and experiences with participants at the session.

• Continue to Acronyms

• Return to Table of Contents