Applying data mining techniques to explore factors contributing to occupational injuries in Taiwan's construction industry

Construction accident research involves the systematic sorting, classification, and encoding of comprehensive databases of injuries and fatalities. The present study explores the causes and distribution of occupational accidents in the Taiwan construction industry by analyzing such a database using the data mining method known as classification and regression tree (CART). Utilizing a database of 1542 accident cases during the period 2000–2009, the study seeks to establish potential cause-and-effect relationships regarding serious occupational accidents in the industry. The results of this study show that the occurrence rules for falls and collapses in both public and private project construction industries serve as key factors to predict the occurrence of occupational injuries. The results of the study provide a framework for improving the safety practices and training programs that are essential to protecting construction workers from occasional or unexpected accidents.     

Longitudinal relationship between economic development and occupational accidents in China

The relativity between economic development and occupational accidents is a debated topic. Compared with the development courses of both economic development and occupational accidents in China during 1953–2008, this paper used statistic methods such as Granger causality test, cointegration test and impulse response function based on the vector autoregression model to investigate the relativity between economic development and occupational accidents in China from 1953 to 2008. Owing to fluctuation and growth scale characteristics of economic development, two dimensions including economic cycle and economic scale were divided. Results showed that there was no relationship between occupational accidents and economic scale during 1953–1978. Fatality rate per 10⁵ workers was a conductive variable to gross domestic product per capita during 1979–2008. And economic cycle was an indicator to occupational accidents during 1979–2008. Variation of economic speed had important influence on occupational accidents in short term. Thus it is necessary to adjust Chinese occupational safety policy according to tempo variation of economic growth.     

Trends in mortality due to motor vehicle traffic accident injuries between 1987 and 2011 in a Spanish region (Comunitat Valenciana)

Objective

To analyse the time evolution of the rates of mortality due to motor vehicle traffic accidents (MVTA) injuries that occurred among the general population of Comunitat Valenciana between 1987 and 2011, as well as to identify trend changes by sex and age group.

Methods

An observational study of annual mortality trends between 1987 and 2011. We studied all deaths due to MVTA injuries that occurred during this period of time among the non-institutionalised population residing in Comunitat Valenciana (a Spanish Mediterranean region that had a population of 5,117,190 inhabitants in 2011). The rates of mortality due to MVTA injuries were calculated for each sex and year studied. These rates were standardised by age for the total population and for specific age groups using the direct method (age-standardised rate – ASR). Joinpoint regression models were used in order to detect significant trend changes. Additionally, the annual percentage change (APC) of the ASRs was calculated for each trend segment, which is reflected in statistically significant joinpoints.

Results

For all ages, ASRs decrease greatly in both men and women (70% decrease between 1990 and 2011). In 1990 and 2011, men have rates of 36.5 and 5.2 per 100,000 men/year, respectively. In the same years, women have rates of 8.0 and 0.9 per 100,000 women/year, respectively. This decrease reaches up to 90% in the age group 15–34 years in both men and women. ASR ratios for men and women increased over time for all ages: this ratio was 3.9 in 1987; 4.6 in 1990; and 5.8 in 2011. For both men and women, there is a first significant segment (p < 0.05) with an increasing trend between 1987 and 1989–1990. After 1990, there are 3 segments with a significant decreasing APC (1990–1993, 1993–2005 and 2005–2011, in the case of men; and 1989–1996, 1999–2007 and 2007–2011, in the case of women).

Conclusion

The risk of death due to motor vehicle traffic accidents injuries has decreased significantly, especially in the case of women, for the last 25 years in Comunitat Valenciana, mainly as of 2006. This may be a consequence of the road-safety measures that have been implemented in Spain and in Comunitat Valenciana since 2004. The economic crisis that this country has undergone since 2008 may have also been a contributing factor to this decrease.Despite the decrease, ASR ratios for men and women increased over time and it is still a high-risk cause of death among young men. It is thus important that the measures that helped decrease the risk of death are maintained and improved over time.     

Individual and occupational factors related to fatal occupational injuries: a case-control study

This study has been designed in order to identify factors increasing the risk of a fatal outcome when occupational accidents occur. The aim is to provide further evidence for the design and implementation of preventive measures in occupational settings. The Spanish Ministry of Labour registry of occupational injuries causing absence from work includes information on individual and occupational characteristics of injured workers and events. Registered fatal occupational injuries in 2001 (n = 539) were compared to a sample of non-fatal injuries in the same year (n = 3493). Risks for a fatal result of occupational injuries, adjusted by individual and occupational factors significantly associated, were obtained through logistic regression models. Compared to non-fatal injuries, fatal occupational injuries were mostly produced by trapping or by natural causes, mostly related to elevation and transport devices and power generators, and injured parts of body more frequently affected were head, multiple parts or internal organs. Adjusted analyses showed increased risk of fatality after an occupational injury for males (adjusted odds ratio aOR = 10.92; 95%CI 4.80–24.84) and temporary workers (aOR = 5.18; 95%CI 2.63–10.18), and the risk increased with age and with advancing hour of the work shift (p for trends <0.01). Injuries taking place out of the usual occupational setting (aOR = 2.85, 95%CI 2.27–3.59), or carrying out atypical tasks (aOR = 2.08; 95%CI 1.27–3.39) showed increased risks of a fatal result too, as occupational accidents in agricultural or construction companies. These data can help to select and define priorities for programmes aimed to prevent fatal consequences of occupational injuries.     

Spatiotemporal analysis of vehicle collisions involving wild boar and roe deer in NW Spain

Ungulate–vehicle collisions pose a serious traffic safety hazard in the North of Spain. The understanding of underlying temporal and spatial structure of these non-random events is imperative to develop appropriate mitigation measures. This study analyses the temporal, spatial and spatiotemporal patterns of car crashes involving wild boar and roe deer in the province of Lugo (NW Spain) in the period 2006–2010 using geographic information systems (GIS) and spatial statistics. The temporal analysis – conducted at three scales: daily, weekly and seasonal – revealed that accidents are related to specific animal's life cycles and to interactions with human activities. The localization of collision points with GIS discovered the sections of the autonomic road network where accidents with the two studied species concentrate. Besides, the spatial arrangement of significant hotspots was mapped through kernel density estimation over two time scales (daily and seasonal), distinguishing among 41 sets, sequentially arranged to facilitate clustering comparison and determination of spatiotemporal risky areas. This work is of valuable help for road managers to design the appropriate mitigation measures that will improve traffic safety and animal welfare.     

Developing techniques for cause-responsibility analysis of occupational accidents

The aim of this study was to specify the causes of occupational accidents, determine social responsibility and the role of groups involved in work-related accidents. This study develops occupational accidents causes tree, occupational accidents responsibility tree, and occupational accidents component-responsibility analysis worksheet; based on these methods, it develops cause-responsibility analysis (CRA) techniques, and for testing them, analyzes 100 fatal/disabling occupational accidents in the construction setting that were randomly selected from all the work-related accidents in Tehran, Iran, over a 5-year period (2010–2014). The main result of this study involves two techniques for CRA: occupational accidents tree analysis (OATA) and occupational accidents components analysis (OACA), used in parallel for determination of responsible groups and responsibilities rate. From the results, we find that the management group of construction projects has 74.65% responsibility of work-related accidents. The developed techniques are purposeful for occupational accidents investigation/analysis, especially for the determination of detailed list of tasks, responsibilities, and their rates. Therefore, it is useful for preventing work-related accidents by focusing on the responsible group’s duties.     

Pressure to produce = pressure to reduce accident reporting?

Each year, more than 4 million U.S. workers are injured on the job – several thousand die (Bureau of Labor Statistics, 2008). Despite these staggering numbers, research suggests that they are gross underestimates of the true volume of workplace related illnesses and injuries due to accident under-reporting. Although accident under-reporting has been well-documented, less is known regarding why this occurs. The current study suggests that under-reporting may in part be due to high levels of perceived production pressure. Specifically, this study tested the hypotheses that production pressure would be related to more experienced accidents overall and more negative attitudes toward reporting accidents. Further, we expected that production pressure would exacerbate the under-reporting of accidents. Survey data were collected from a sample of 212 copper mining workers located in the southwestern United States. The survey measured employee perceptions regarding production pressure, attitudes toward reporting accidents, perceived consequences of reporting accidents, and actual reporting behaviors (e.g., types and numbers of accidents experienced vs. reported). As predicted, the average number of experienced accidents per employee was significantly higher (M = 2.84) than the number of reported accidents (M = .49). In addition, production pressure was related to more negative reporting attitudes. Individuals who had positive reporting attitudes were injured less frequently; however, when an incident occurred, they were more likely to report it. Finally, higher levels of production pressure were related to greater accident under-reporting. Additionally, employees who perceived high levels of production pressure not only experienced more accidents overall, they also reported fewer of them to the organization. Implications for occupational safety initiatives – particularly in the current economic climate – are discussed, as are methodological challenges of conducting research in this area.     

Occupational exposure of medical radiation workers in Lithuania, 1950-2003

This study presents the summary of historical exposures, measurementpractice and evolution of the recording of the individual dosesof medical radiation workers during 1950–2003 in Lithuania.The aim of this study is to present occupational exposure ofmedical radiation workers in Lithuania since the earliest appearanceperiod. Data from publications have been used for the earliesttwo periods prior to 1969; data from the archives of the largesthospitals, for the period 1970–1990 and data from LithuanianSubdivision of Individual Dosimetry of Radiation ProtectionCenter, for the period 1991–2003. The analysis of thedata obtained from personal records allows to conclude thatthe average annual effective dose of Lithuanian medical radiationworkers was greatly reduced in radiology, radiotherapy and nuclearmedicine in all occupational categories from 1950 to 2003. Duringthe last period 1991–2003 extremity doses clearly decreasedand after 1994 were no longer present in Lithuania.     

Association of work-related accidents with noise exposure in the workplace and noise-induced hearing loss based on the experience of some 240,000 person-years of observation

This retrospective study explores the association between occupational noise exposure at the time of hearing tests, permanent noise-induced hearing loss and work-related accident risk. Log-binomial analysis was used to first ascertain the association between study variables according to activity sector (North American Industry Classification System, NAICS) and accident context while controlling for age. Second part of the paper estimates the overall number of accidents attributable to occupational noise or the associated hearing loss (excess fraction). Study was carried on a sample of 52,982 male workers exposed to a minimum of 80dBA on a daily basis and whose hearing was examined at least once between 1983 and 1996 by public health authorities of Quebec. These participants evidenced bilateral average hearing threshold levels at 3, 4 and 6kHz ranging from normal (/=90dBA) and noise-induced hearing loss.     

Fatal train accidents on Europe's railways: 1980-2009

This paper presents an analysis of fatal train accident rates and trends on Europe's main line railways from 1980 to 2009. The paper uses a new set of data for the European Union together with Norway and Switzerland, assembled partly under the auspices of the European Railway Agency and partly on the author's own account. The estimated overall trend in the number of fatal train collisions and derailments per train-kilometre is −6.3% per year from 1990 to 2009, with a 95% confidence interval of −8.7% to −3.9%. The estimated accident rate in 2009 is 1.35 fatal collisions or derailments per billion train-kilometres, giving an estimated mean number of fatal accidents in 2009 of 6.0. The overall number of fatalities per fatal accident in 1990–2009 is 4.10, with no apparent long term change over time, giving an estimated mean of 24.6 fatalities per year in train collisions and derailments in 2009. There are statistically significant differences in the fatal train accident rates and trends between the different European countries, although the estimates of the rates and trends for many individual countries have wide confidence limits. The distribution of broad causes of accidents appears to have remained unchanged over the long term, so that safety improvements appear to have been across the board, and not focused on any specific cause. The most frequent cause of fatal train collisions and derailments is signals passed at danger. In contrast to fatal train collisions and derailments, the rate per train-kilometre of serious accidents at level crossings remained unchanged in 1990–2009. The immediate causes of most of the serious level crossing accidents are errors or violations by road users.     

Permanent impairments, disabilities and disability pensions related to accidents in Norway

The purpose of this study was to estimate the burden of accidents in terms of new disability pensioners related to accidental injuries in Norway and to use this information to supplement the picture of the epidemiology of accidents in Norway. The basis of the study was 7,241 new disability pensioners due to accidents in the Disability Register of the National Insurance Administration in Norway for 1992-1997. Included in the analyses are some data from health surveys in USA and Norway. The rate of disability pensioners (16-66 years old) due to accidents increased 4% annually during 1992-1997, and in the age group 16-44 years 8% annually. The increase is found in all types of accidents, except for home and the group other accidents, where the rate is constant. 45% had been injured in traffic accidents, 33% in occupational accidents, 16% in leisure accidents, 4% in home accidents and 2% in other accidents. The rates of disability pensionings due to occupational accidents are three to four times higher among 'blue collar' workers than among 'white collar' workers. While disability pensioning rate in the age group 16-66 years due to accidents has increased 4% annually, the accident mortality for the age group 15-64 years has decreased 3% annually. The relationship between disability pensioning and mortality in the age group 15-64 years is found to be approximately 2:1 for all accidents. For occupational accidents the relationship is 5:1, for traffic accidents 2.5:1, for leisure and other accidents 1:1 and for home accidents 1:2.7. The complete epidemiologic picture due to accidents in Norway per 100,000 of population seems to be: medically treated, 10,000; hospitalised, 1,200; permanently impaired, 800 (of whom probably 50% are treated as out-patients); disabled (restricted activity), 400; fatalities, 40; and disability pensioned in the age group 16-66 years, 50. Medically treated and hospitalised patients due to accidents seem to show a slightly decreasing rate, fatalities a higher decreasing rate. This supplement to the epidemiological picture of rates and trends of impairments, disability and disability pensionings (the more serious consequences of accidents), points to the need to focus on the current trends for accident prevention.     

Modeling fault among accident—Involved pedestrians and motorists in Hawaii

Using a comprehensive database of police-reported accidents in Hawaii, we describe the nature of pedestrian accidents over the period 2002–2005. Approximately 36% of the accidents occur in residential areas, while another 34% occur in business areas. Only 41.7% of the pedestrian accidents occur at intersections. More pedestrian crashes occur at non-intersection locations—including midblock locations, driveways, parking lots, and other off roadway locations. Approximately 38.2% of the crashes occur at crosswalk locations, while proportionately more (61.8%) of the pedestrian accidents occur at non-crosswalk locations. Using this database the human, temporal, roadway, and environmental factors associated with being “at-fault” for both pedestrians and drivers are also examined. Using techniques of logistic regression, several different explanatory models are constructed, to identify the factors associated with crashes producing fatalities and serious injuries. Finally, two pedestrian models (drunk males and young boys) and one driver model (male commuters) are developed to provide further understanding of pedestrian accident causation. Drunk male pedestrians who were jaywalking were in excess of 10× more likely than other groups to be at-fault in pedestrian accidents. Young boys in residential areas were also more likely to be at-fault. Male commuters in business areas in the morning were also found to have higher odds of being classified at-fault when involved in pedestrian accidents. The results of this study indicate that there should be a combination of enforcement and educational programs implemented for both the pedestrian and drivers to show those at-fault the consequences of their actions, and to reduce the overall number of accidents.     

Dose level of occupational exposure in China

This paper discusses the dose level of Chinese occupational exposures during 1986-2000. Data on occupational exposures from the main categories in nuclear fuel cycle (uranium enrichment and conversion, fuel fabrication, reactor operation, waste management and research activity, except for uranium mining and milling because of the lack of data), medical uses of radiation (diagnostic radiation, nuclear medicine and radiotherapy) and industrial uses of radiation (industrial radiography and radioisotope production) are presented and summarised in detail. These are the main components of occupational exposures in China. In general, the average annual effective doses show a steady decreasing trend over periods: from 2.16 to 1.16 mSv in medical uses of radiation during 1990-2000; from 1.92 to 1.18 mSv in industrial radiography during 1990-2000; from 8.79 to 2.05 mSv in radioisotope production during the period 1980-2000. Almost all the average annual effective doses in discussed occupations were lower than 5 mSv in recent years (except for well-logging: 6.86 mSv in 1999) and no monitored workers were found to have received the occupational exposure exceeding 50 mSv in a single year or 100 mSv in a five-year period. So the Chinese protection status of occupation exposure has been improved in recent years. However, the average annual effective doses in some occupations, such as diagnostic radiology and coal mining, were still much higher than that of the whole world. There are still needs for further improvement and careful monitoring of occupational exposure to protect every worker from excessive occupational exposure, especially for the workers who were neglected before.     

The link between fatigue and safety

The objective of this review was to examine the evidence for the link between fatigue and safety, especially in transport and occupational settings. For the purposes of this review fatigue was defined as ‘a biological drive for recuperative rest’. The review examined the relationship between three major causes of fatigue – sleep homeostasis factors, circadian influences and nature of task effects – and safety outcomes, first looking at accidents and injury and then at adverse effects on performance. The review demonstrated clear evidence for sleep homeostatic effects producing impaired performance and accidents. Nature of task effects, especially tasks requiring sustained attention and monotony, also produced significant performance decrements, but the effects on accidents and/or injury were unresolved because of a lack of studies. The evidence did not support a direct link between circadian-related fatigue influences and performance or safety outcomes and further research is needed to clarify the link. Undoubtedly, circadian variation plays some role in safety outcomes, but the evidence suggests that these effects reflect a combination of time of day and sleep-related factors. Similarly, although some measures of performance show a direct circadian component, others would appear to only do so in combination with sleep-related factors. The review highlighted gaps in the literature and opportunities for further research.     

Alcohol, psychoactive drugs and fatal road traffic accidents in Norway: a case-control study

A case–control study was conducted on 204 drivers fatally injured in road traffic accidents in south-eastern Norway during the period 2003–2008. Cases from single vehicle accidents (N = 68) were assessed separately. As controls, 10 540 drivers selected in a roadside survey in the same geographical area during 2005–2006 were used. Blood samples were collected from the cases and oral fluid (saliva) samples from the controls. Samples were analysed for alcohol, amphetamines, cannabis, cocaine, opioid analgesics, hypnotics, sedatives and a muscle relaxant; altogether 22 psychoactive substances. Equivalent cutoff concentrations for blood and oral fluid were used. The risk for fatal injury in a road traffic accident was estimated using logistic regression adjusting for gender, age, season of the year, and time of the week. The odds for involvement in fatal road traffic accidents for different substances or combination of substances were in increasing order: single drug < multiple drugs < alcohol only < alcohol + drugs. For single substance use: medicinal drug or THC < amphetamine/methamphetamine < alcohol. For most substances, higher ORs were found when studying drivers involved in single vehicle accidents than for those involved in multiple vehicle accidents, but confidence intervals were wider.     

Work-related versus non-work-related road accidents, developments in the last decade in France

The aim of this research was to analyze the changes that have affected work-related road accidents between 1997 and 2006, using police data. The study focused on drivers aged between 14 and 64 years. The characteristics considered were the age, gender, type of vehicle and occupation of the individuals involved and the location, time and severity of the accident. Two periods were compared, 1997-2000 and 2003-2006. Three types of journey were considered: while at work, commuting (going to and from work), and non-work-related.The percentage of all accidents which were work-related varied little over the decade (10% while at work and 18% while commuting). The accidents that occurred while at work still had the lowest fatality rates (1.4% among women and 3.4% among men in 2003-2006). Men accounted for the majority of the casualties: 90% of fatalities while at work and approximately 80% for the other types of journey. The greatest reduction took place in the number of motorists, and this has led to an increase in the proportion of motorized two-wheelers both while at work and while commuting. In the case of accidents while at work, the professional drivers still had the highest risk, craftsmen and shopkeepers were also at higher risk than manual workers of both genders and male employees. Extending analysis of this type to other European countries would be of major interest both for epidemiological monitoring and the prevention of occupational road accidents.     

Accidents in the construction industry in the Netherlands: An analysis of accident reports using Storybuilder

As part of an ongoing effort by the Ministry of Social Affairs and Employment of the Netherlands, a research project is being undertaken to construct a causal model for occupational risk. This model should provide quantitative insight into the causes and consequences of occupational accidents. One of the components of the model is a tool to systematically classify and analyse reports of past accidents. This tool ‘Storybuilder’ was described in earlier papers. In this paper, Storybuilder is used to analyse the causes of accidents reported in the database of the Dutch Labour Inspectorate involving people working in the construction industry. Conclusions are drawn on measures to reduce the accident probability. Some of these conclusions are contrary to common beliefs in the industry.     

Fatal falls among Hispanic construction workers

This study evaluated occupational deaths resulting from fall injuries among Hispanic construction workers using data from the Census of Fatal Occupational Injuries and the Current Population Survey. The demographics and characteristics of fatal falls among Hispanic workers were examined and compared with that of their white, non-Hispanic counterparts. The results show that fatal injuries among Hispanic construction workers were more likely to be caused by a fall than their white, non-Hispanic counterparts (OR = 1.48, 95% CI: 1.05–2.10) after controlling for possible confounders. The rate of fatal falls for foreign-born Hispanic construction workers was 5.5 per 100,000 FTE, which is significantly higher than 4.1 per 100,000 FTE for Hispanic workers who were born in the U.S. (OR = 1.36, 95% CI: 1.08–1.67). The disparities in fatal injuries from falls were found in age groups, job tenure, occupations, and types of construction projects. This study also found that about every two of three fatal falls in construction occurred in establishments with 10 or fewer employees. More prevention, intervention, and training measures must be applied to Hispanic workers, especially those who are new immigrants. OSHA enforcements should target small construction establishments in order to lower overall fatality rates, costs, and unnecessary losses of life.     

Reductions in transport mortality in Australia: Evidence of a public health success

Objective

To describe trends in transport mortality for a range of common transport types in Australia over a 30-year period (1975–1977 to 2005–2007).

Methods

Mortality data on all-cause and transport-related causes of death were supplied by the Australian Institute of Health and Welfare (AIHW). Mortality rates, expected number of deaths and probabilities of death were compared for three time periods: 1975–1977, 1990–1992 and 2005–2007.

Results

There were significant decreasing trends between 1975–1977 and 2005–2007 in all-cause and most other transport mortality types for both men and women. There were significant reductions in the contribution of transport-related mortality to all-cause mortality; however the difference in mortality between men and women (higher for men) changed little over the evaluated period.

Conclusions

Between 1975–1977 and 2005–2007 there were marked reductions in key causes of transport-related mortality amongst Australian adults, and the reductions in transport-related mortality exceeded reductions in all-cause mortality. The reductions could be attributed to better preventive measures and improved medical treatment for people involved in transport crashes. Although there is scope for further improvement, the reductions are evidence of a success in the prevention of crashes and the medical treatment of crash victims.     

Strategies for improving safety performance in construction firms

Over the years many prevention management practices have been implemented to prevent and mitigate accidents at the construction site. However, there is little evidence of the effectiveness of individual or combined practices used by companies to manage occupational health and safety issues. The authors selected a sample of 1180 construction firms and 221 individual practices applied in these companies to analyze their effectiveness reducing injury rates over a period of four years in Chile. Different methods were used to study this massive database including: visual analyses of graphical information, statistical analyses and classification techniques. Results showed that practices related to safety incentives and rewards are the most effective from the accident rate viewpoint, even though they are seldom used by companies; on the other hand, practices related to accidents and incidents investigation had a slight negative impact on the accident rate because they are frequently used as a reactive measure. In general, the higher the percentage of prevention practices implemented in a strategy, the lower the accident rate. However, the analysis of the combined effect of prevention practices indicated that the choice of the right combination of practices was more important than just the number of practices implemented.