- Email: [email protected]
Work injury trends during the last three decades in the construction industry
Safety Science 85 (2016) 60–66
Contents lists available at ScienceDirect
Safety Science journal homepage: www.elsevier.com/locate/ssci
Work injury trends during the last three decades in the construction industry Flemming Lander a,⇑, Kent Jacob Nielsen b, Jens Lauritsen c a
Dept. of Occupational Medicine, Odense University Hospital, Denmark Danish Ramazzini Centre, Department of Occupational Medicine, Herning Regional Hospital, Denmark c Accident Analysis Group, Dept. of Orthopedic Surgery, Odense University Hospital, Denmark and Institute of Clinical Medicine, University of Southern Denmark, Denmark b
a r t i c l e
i n f o
Article history: Received 1 April 2015 Received in revised form 25 August 2015 Accepted 27 October 2015
Keywords: Longitudinal study Age Type of injury Work activities Business cycle Emergency department Denmark
a b s t r a c t Objectives: The aim was to analyze injury trends according to age, severity, work activity and business cycle in the construction sector. Methods: From 1980 to 2010 a total of 23.464 work related injuries were treated at the emergency department at Odense University Hospital. The annual incidences were calculated. Employment levels in the construction sector were used as an indicator of fluctuations in the business cycle since 1980. Results: Through the last three decades the overall trend of work-related injuries was unchanged. For some subgroups of injuries, such as major injuries and injuries due to young workers use of small powered tools significant downward trends were seen, but trends within different age groups of workers were unchanged and young workers have at least twice the risk compared to older workers. The fluctuations in work injury trends among workers under 30 years of age were significantly related to the business cycle, where the risk of injuries was higher during economic booms than during recessions. Further, periods with economic booms are positively related to the rate of minor injuries and injuries due to all other work activities than the use of power tools. Conclusion: Overall the number of injuries in the construction sector have not changed significantly during the recent three decades, except for minor subgroups related to ‘major injuries’ and ‘injuries due to use of small power tools’. Re-evaluation of safety prevention programs is needed in order to break the high injury level among young workers compared to older workers, especially during economic booms. Ó 2016 Published by Elsevier Ltd.
1. Introduction A recent study revealed that the overall trend of work injuries in the Danish construction industry has been unchanged during the last three decades, which stands in contrast to the general downwards trend seen in work injuries in Denmark across all sectors (Nielsen et al., 2015). This long term trend as well as the pattern of annual fluctuations were similar for work injuries treated at a emergency department (ED) and for injuries reported to the Danish Working Environment Authority (Nielsen et al., 2015). This is in spite of a very small individual overlap between these two datasets. In addition, each data set represents to a large extent different injury events and different types of injuries (Lander et al., 2014). Thus, the strong concordance in the time trends and fluctuations indicates that injury frequency is influenced by one or more common underlying factor. ⇑ Corresponding author. Tel.: +45 26353208; fax: +45 65 41 49 88. E-mail address: [email protected] (F. Lander). http://dx.doi.org/10.1016/j.ssci.2015.10.013 0925-7535/Ó 2016 Published by Elsevier Ltd.
So far, no other European studies or surveillance reports covering work injury rates in the construction industry for this period exist. In a US setting, a significant decline in rates of reported work-related injuries and illnesses has been observed over the preceding 20-year period among union carpenters (McCoy et al., 2013). In particular, a substantial decline was observed for falls from height among these construction workers (Lipscomb et al., 2014). Also, the US Bureau of Labor Statistics observed a downward trend in reported work injuries in the construction industry as a whole (Welch et al., 2007). Inconsistencies in the information suggest that some of the apparent decrease may be due to changes in the ways injuries are treated, misclassification of employees, or underreporting (McCoy et al., 2013; Welch et al., 2007). Construction has always been regarded as one of the most hazardous industries (Swuste et al., 2012). It is considered as a special case, as the nature of the work entails that automation and many technological improvements that have led to improved safety in other industries has not had the same impact within construction (Swuste et al., 2012). Thus, in this context the European Union (EU)
61
F. Lander et al. / Safety Science 85 (2016) 60–66
2. Materials and methods
A total of 23.464 injuries within the construction industry between 1980 and 2010 were observed. Table 1 shows the dichotomized age distribution of the injuries. Almost half of all injuries involved construction workers below 30 years of age. Clinical information consisted of injury diagnosis based on ICD8 (1980– 1992) and ICD10 (1993–2010), which was transformed into main type of injuries and body parts. ‘Major injuries’ include amputations, bone fractures, joint sprains and strains, all types of soft tissue damage, eye corrosion and burns, and electrical shock. ‘Minor injuries’ mainly consisted of superficial lacerations and wounds, and being struck by a foreign body (mainly in eyes). The ratio between major and minor injuries was one to two, see Table 1. The ED register also contains individual information on work activity leading to the injury (Table 1). These different types of activities were transformed into three main activities: (1) ‘Movement activities’, which include slips, trips, falls, striking an object when walking, handling or carrying materials. (2) ‘Work with small power tools’, e.g. electric or air pressure powered hand tools such as angel grinders, nail guns, drilling machines, as well as machinery in a stationary frame such as smoothing plane or band saw. (3) ‘Miscellaneous work activities’ include use of non-powered hand tools, various assembly or disassembly work tasks and a heterogeneous number of activities e.g. activities involving big machines such as cranes and trucks, and exposure to chemicals. We assumed that the proportion of workers at risk on the construction sites who perform these three categories of work tasks remains approximately constant from year to year during the observation period. In the present study, the annual level of employment within the Danish construction sector was used to describe the business cycle from 1980 to 2010, see Fig. 1. Three major economic recessions can be observed with steep falls in employment in the first years of the 1980s, the first years of the 1990s, and the late years of the 2000s. A minor recession is also seen in the mid 2000s. The observed trend corresponds closely to trends in other measures of the business cycle, such as general unemployment, indicating that they measure the same underlying economic phenomenon (Asfaw et al., 2011). The correlation between annual level of employment within construction and the general unemployment level in Denmark was 0.83 (P < 0.01). Data was obtained from Bureau of Statistics Denmark.
2.1. The observation data
2.2. Statistics
The population base of the study is work sites situated in the catchment area of the Emergency Department (ED) at Odense University Hospital (OUH), which is situated in the city of Odense on the island of Funen, Denmark. The department is the only emergency room in that part of the island and the ED is a free 24-h emergency service. The catchment area for OUH is a well defined mixed rural and urban geographic area with a population of 362,000 inhabitants, which represents approximately 6.7% of the Danish population. The area is demographically and industrially comparable to Denmark as a whole (Statistic Denmark, 2015). Since 1980, the Accident Analysis Group (UAG) at the ED, OUH, has conducted systematic and quality-assured recording of all treated injuries. Each year 35–40.000 injury patients are treated at the ED. About 10% of these injuries occur during paid work and are classified as occupational. In 2010 the UAG ED-database contained information on approx. 160.000 occupational injuries. All injured individuals are indentified by a unique 10-digit personal civil registry number (CRN), and the CRN combined with year of injury was the key for linking to Statistics Denmark’s Integrated Database for Labour Market Research, IDA (Statistic Denmark, 2015). IDA contains individual information on industry ties on an annual basis from 1980 to 2010. The IDA 6 digit industry codes were transformed into an overall construction sector code.
Total annual employment and age stratified employment in construction in the ED catchment area formed part of the denominator in the equations of the injury incidence rate calculations. Incidence was defined as the number of injuries per 1000 workers employed (Table 2). The table shows the distribution of injured construction workers, and the total number of employed workers in the catchment area of the Emergency Department according to
since the early 1990s had enacted several directives, guidelines and standard directed toward safety and health at work e.g. Directive 92/57/EEC and the majorities of the member states including Denmark had fully implement specific legislation directed toward occupational safety in the construction sector (osha.europa.eu, 2015). In Denmark, as well as other western countries, much has been done during the last 20–30 years to regulate and improve work environment standards on building sites (Fabiano et al., 2001; Haslam et al., 2005; Danish Labor Inspection, 2015). Many initiatives have repeatedly been introduced, either through legislation or labor market agreements between employers and employees, both in specific industries and at enterprise level e.g. promotion of drug and alcohol-free workplaces, safety training programs, establishing safety groups on work-sites, but unfortunately few formal scientific evaluations have been undertaken measuring the effect on work injuries (Loomis et al., 2004; Fabiano et al., 2001; Wickizer et al., 2004; The Danish Health and Medicines Authority, 2008). Parallel to these initiatives, recurring economic recessions and booms within the construction sector, as well as other industries, continuously alter work conditions e.g. through changes in the number of hours worked, pacing of work, and changes in the level of recruitment of inexperienced labor (Asfaw et al., 2011). As a consequence, a declining number of reported work injuries are observed during recessions while an increase is seen during booms – a trend that the construction industry might be especially sensitive to (Nielsen et al., 2015; Asfaw et al., 2011). It is clear that in spite of formal work standards and legislations work environment practices on construction sites are not fixed, but are altering constantly over time. However, very little is known about the impact of these dynamic processes on injuries in different subgroups of workers or on labor force characteristics. The aim of the current study was to describe long term trends in injury incidence rates in different age-groups, in different type of injuries, and in injuries due to different work activities in the construction sector, and to analyze the impact of the business cycle on work injuries according to these labor force characteristics.
Table 1 Background indicators. Indicators
N
% (95% CI)
Age (years) <30 P30
11,028 12,436
47.0 (46.3–47.6) 53.0 (52.0–53.9)
Type of injuries Minor injuries Major injuries
15,972 7492
68.1 (67.0–69.1) 31.9 (31.2–32.7)
Injuries due to Movement activities Use of small powered tools Other work activities
7893 5024 10,547
33.6 (32.9–34.4) 21.4 (20.8–22.0) 45.0 (44.1–45.8)
Total number of injuries
23,464
F. Lander et al. / Safety Science 85 (2016) 60–66 200000
variability and trend using linear regression analysis. All analyses were performed in STATA 11.
160000
180000
3. Results
140000
Total employment in construction industry
62
1980
1990
2000
2010
year Fig. 1. The course of the business cycle indicator expressed by the annual level of employment within the construction sector in Denmark from 1980 to 2010.
Table 2 Distribution of injured construction workers, and the total number of employed workers in the catchment area of the Emergency Department according to age and year of injuries. Year
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Age < 30 years
Age P 30 years
Injuries n
Total employment N
Injuries N
Total employment N
501 368 360 426 452 441 520 492 459 406 353 247 297 186 261 237 278 386 340 388 390 367 279 262 313 339 384 451 301 288 256
4382 3777 3356 3182 3113 3370 3707 3885 3853 3525 3312 3160 3083 2840 2410 2651 2761 2821 2926 2961 2870 2977 2728 2521 2521 2705 2864 3091 3242 2925 2262
429 428 380 399 432 424 491 489 471 423 416 220 328 198 270 256 356 380 423 418 493 409 425 407 410 507 464 541 375 380 394
8285 6918 6518 6137 6374 6337 6940 7111 7055 6849 6682 6577 6787 6160 5505 6035 6213 6277 6374 6406 6723 7291 7376 6634 6733 7244 7827 8163 8318 7775 7146
age and year of injuries. The business cycle indicators were tested for stationarity using the augmented Dickey–Fuller (ADF) test. The ADF test revealed that the indicator ‘‘annual level of employment within the total construction sector” was non-stationary, whereas in turn the first difference of the variable (the change score from one year to next year) was stationary (t = 4.12, P = 0.001). Thus, this change score was used as an estimate of the impact of the business cycle on the incidence of work injuries. The longitudinal, annual incidence data and its correlation to age, type of injury, and work activity, as well as the business cycle indicator, was tested for
Fig. 2 shows the mean annual incidence of injuries according to 3 different indicators from 1980 to 2010. Graph A shows the courses and trends in two age groups. The estimated linear trends are time constant for both age groups – see also Table 2, which shows coefficients and p-values. Visually, the course shape of the incidence seems to be close to that of the business cycle indicator in Fig. 1 – with a general decrease in injury incidence during recession periods and visa versa during booms. This picture seems far more profound in the age groups below 30 years compared to the workers older than 30 years. Further, it can be seen that the injury incidences are approx. 2–3 times higher for workers below 30 years of age than for those more than 30 years of age. The graph B shows the course and linear trends stratified into major and minor injuries according to age groups. For both age groups a moderate but significant reduction at 0.25 and 0.13 for major injuries per 1.000 workers per year is observed, whereas for the minor injuries, the trend is constant for workers below 30 years of age and slight increasing for those over 30 years, see Table 2. Again, the shape of the incidence courses seems to be closely related to that of the business cycle indicator. This is especially noticeable for the minor injuries. The graph C shows the course and linear trends stratified into injuries due to the three different main work activities according to age groups. Overall, a general declining trend through all work activities among the youngest age group is seen, which is contrary to the trends observed among the oldest worker group, – the injuries related to work with power tools and other work activities revealed small, but significant decreases in incidence rates at 0.13 and 0.18 injuries per 1000 workers per year, whereas the declining trend is borderline significant for injuries due to movement activities (see Table 2). Again, the shape of the incidence courses during the thirty years of observation seems to be close to that of the business cycle indicator, especially for injuries due to movement activities and other work activities. Fig. 3 shows the correlations expressed by the linear regression trend line between annual injury incidences and annual change in total employment in the construction sector, from 1980 to 2010, according to the three different indicators. The matching linear coefficients and p-values are shown in Table 3. Overall, the linear trends do not indicate a strong relationship between the business cycle indicator and the three indicators, but for workers under 30 years, the relationship is significant and indicates that these workers have a slightly higher risk for work injuries during economic booms. Workers over the age of 30 seem also to be affected by changes in the business cycle, but the association is only borderline significant. Also, minor types of work injuries and injuries due to the category ‘‘other work activities” occur more frequently during economic booms than during recessions, whereas major types of injuries and injuries due to use of power tools are unaffected by changes in the business cycle. A declining number of injuries due to movement activities are borderline associated to periods of economic booms (see Table 4). 4. Discussion Longitudinal observations over a long time span are rare, but in turn might provide insight into possible accumulated effects of repeated preventive initiatives as well as changes in safety attitudes on work sites. This is rarely measurable in studies with short time spans or specific intervention trials (van der Molen et al., 2012; Sancini et al., 2012). Long time series also provide
63
F. Lander et al. / Safety Science 85 (2016) 60–66
>= 30 years
50
75
100
125
150
175
< 30 years
25
Injuries per 1000 workers (95% poisson CI)
Graph A
1980
1990
2000
Construction sector.
2010 1980
1990
2000
2010
Minor injuries & >= 30 years
Major injuries & < 30 years
Major injuries & >=30 years
10
20
30
40
0
10
20
30
40
Minor injuries & < 30 years
0
Injuries per 1000 workers (95% poisson CI)
Graph B
1980
1990
2000
2010 1980
1990
2000
2010
Construction sector.
Small powered tools & <30 years
Other work activities & < 30 years
Movement activities & >=30 years
Small powered tools & >=30 years
Other work activities & >=30 years
10
20
30
0
10
20
Movement activities & <30 years
0
Injuries per 1000 workers (95% poisson CI)
30
Graph C
1980
1990
2000
2010 1980
1990
2000
2010 1980
1990
2000
2010
Construction sector. Fig. 2. Work injury incidence from 1980 to 2010 by age, severity and work activity.
opportunities to assess transient progress and regressions in prevention of work injuries, including those that follow economic cycles, which seem to have a profound influence on levels of work injuries (Asfaw et al., 2011; Davies et al., 2009; Boone et al., 2006; Nielsen et al., 2015). Short term intervention studies are
particularly sensitive to these external time dependent fluctuations when assessing the results of a specific intervention on work injuries. In this context, our study revealed some characteristic patterns of work injuries in relation to different labor force characteristics, and some of the associations suggest new perspectives:
F. Lander et al. / Safety Science 85 (2016) 60–66 Minor injuries
Major injuries
60
100
Incidence pr 1000 workers
80
>= 30 years
0
20
50
Incidence pr 1000 workers
150
< 30 years
40
64
-20
-10
0
10
-20
-10
0
-20
10
0
10
-20
-10
0
10
10 20 30 40
Small powered tools
-20
-10
0
10
Other work activities
10 20 30 40
Incidence pr 1000 workers
Movement activities
-10
delta (employment in construction(N/1000))
delta (employment in construction(N/1000))
-20
-10
0
10
delta (employment in construction(N/1000)) Fig. 3. Linear regression of mean annual injury incidences from 1980 to 2010 and the national employment rate in construction industry by age, severity and work activity.
Table 3 The regression coefficient for injury trends between 1980 and 2010 among construction workers according to different age groups, different type of injuries and injuries due to different work activities. Indicators Age (years) <30 P30 Type of injuries Minor injuries <30 P30 Major injuries <30 P30 Injuries due to Movement activities <30 P30 Use of powered tools <30 P30 Other work activities <30 P30 *
Coeff. 0.03 0.12
SE 0.40 0.21
t-Value 0.07 0.57
0.14 0.24
0.09 0.10
1.54 2.43
0.25 0.13
0.04 0.04
5.69 2.97
Table 4 The association between mean annual work place injury incidences and total annual employment in the constructions sector according to different indicators. Coeff.
SE
t-Value
p-Value
Age <30 P30
231.81 276.95
69.63 147.04
3.33 1.88
0.002* 0.07
Type of injuries Minor injuries Major injuries
480.84 260.51
158.91 303.77
3.03 0.86
0.005* 0.40
Injuries due to Movement activities Use of powered tools Other work activities
670.11 469.74 702.54
337.18 454.10 233.17
1.99 1.03 3.01
0.06 0.309 0.005*
p-Value 0.95 0.58
0.14 0.02* 0.0001* 0.006* *
0.08 0.11
0.04 0.06
1.99 1.92
0.13 0.05
0.04 0.03
3.52 1.53
0.001* 0.14
0.18 0.05
0.06 0.06
2.87 0.80
0.008* 0.43
p-Values < 0.05.
0.06 0.06
p-Values < 0.05.
First, general, but modest long term declines were seen in time trends for major injuries combined regardless of workers age, for injuries among young workers due to work with powered tools and for the heterogenic category of other work activities. Conversely, work-related minor injuries have increased for workers that are more than 30 years old during the last three decades. A few longitudinal studies covering all industries during the preceding decades also have observed a decline in both the number of
serious and non-serious injuries, but so far no such observations have been observed for the construction sector as a whole. The most consistent observation in several studies has been a decline in fatal injuries across all industries, as well as in the construction sector (Sancini et al., 2012). Fatal injuries have been more than halved from the 1980s to the present in the Danish construction industry (www.at.dk, 2015). This indicates that some improvements in work safety have been achieved and according to our study, the same repeated initiatives might also have affected major injuries, but not minor injuries. A previous study have shown a decline of serious injuries (‘‘fall from height”) in a subgroup of construction workers (carpenters) during the preceding 2 decades (Lipscomb et al., 2014). In our study, ‘‘fall from height” is included in the category ‘‘movement work activities”, but represents only a small fraction of all types of injuries due to these activities, and thus, has no real impact on the overall trend. One remarkably
F. Lander et al. / Safety Science 85 (2016) 60–66
observation in our study is the decline in injuries due to work with powered tools, but only among the young workers, not the older workers. This is new information, which to our knowledge has not previously been described. In our study power tools represent many different hand held tools and small tools in a stationary frame. From another study of ED treated work injuries, saws, nail guns, welding tools and drills seem to be the most common cause of work injuries, representing 65% of all powered tool-related injuries (Lipscomb et al., 2010). As in the US study, our results indicate that many of the technological improvements and changes in safety standards that have led to improved safety elsewhere have also had similar impact in some subgroups of construction workers (Lipscomb et al., 2010). Even though the overall long term work injury trends are constant, the picture of young workers having at least twice the risk compared to older workers is maintained during the whole observation period. Hypothetically, the result could be biased, if health behavior has changed during the last 30–35 years, but we have no information on any age-specific changes in health behavior, including the use of hospital emergency departments. During the whole observation period, there has been free and open access to unrestricted treatment at all EDs in Denmark, with no economic consequences for either workers or companies and without any notification to authorities or employer. Thus, an injured worker can seek treatment at any ED free of charge, without company knowledge, and independent of possible future compensation claims (Nielsen et al., 2015; Lander et al., 2014). In many studies, age seems to act as a strong risk predictor of injuries. It is well documented, especially in the construction sector, that the frequency of injuries is higher among young workers (Salminen, 2004; Siu et al., 2003; Westaby and Lowe, 2005). Two US-studies, both based on ED sources between 1998 and 2005, revealed incidence rates between 51–81 and 28–54 per 1.000 construction workers under 35 years respectively, a range 2–3 times higher than found in older workers (Schoenfisch et al., 2010). Secondly, we found significant cyclical fluctuations in annual injury incidences through all three decades in both age groups. However, the fluctuation was especially pronounced among young workers and less for the oldest age group. The reason for that is not clear, but for the youngest workers, some of that fluctuation was related to changing periods of economic boom and recession. In our study at least 4 such periods were observed. During booms the risk for work injuries among young workers was significantly higher than for older workers, who were less affected by the business cycle. The relationship between business cycles and injury frequency is well known (Asfaw et al., 2011; Davies et al., 2009; Boone et al., 2006). However, no previous studies have looked at specific age-stratified trends. Multiple factors have been suggested to explain this general relationship between the business cycle and workplace safety. Some of these changes are probably a mix of task-specific factors (such as higher work pace during booms), and organisational factors (such as less training of new employees or longer working hours during booms) (Asfaw et al., 2011). These cyclic changes seem to affect the younger workers more than older workers, probably because young workers are less experienced and trained (Salminen, 2004; Siu et al., 2003; Westaby and Lowe, 2005). Further, in line with a Dutch and a British study, our results also show that economic booms are positively related to the rate of minor injuries, whereas major injuries are independent (Davies et al., 2009; Boone et al., 2006). Why the seriousness of work injuries is sensitive to the business cycle is not clear. In the context of reported injuries, the association might be explained as a consequence of changes in reporting behavior during economic recessions. Because of high unemployment, greater fear of dismissal could lead to a lower level of reporting of minor injuries compared to major injuries (Davies et al., 2009). But our results rest on ED
65
data, which in a Danish context are more or less independent of changes in reporting behavior, because injured workers do not have to fear dismissal when companies have no knowledge of the ED treatment, and there are no treatment costs or economic incentives keeping injured workers from seeking treatment (Nielsen et al., 2015). Thus, it is plausible to presume that higher work pace, poorer training of new employees and longer working hours during booms negatively influence worksite conditions, e.g. order and tidiness on building sites, which are well known risk factors associated to work injuries (Laitinen et al., 2010; Laitinen and Päivärinta, 2010). The study has several strengths. First, we were able to determine the exact age specific annual injury incidence based on reliable employment data during more than three decades. The key for that is the individual unique 10-digit personal civil registry number, which links every single person in Denmark to a work place since the introduction of the system in 1972. Calculation of annual incidences according to different work activities rests on the total employment in the ED catchment area, and the calculation of the trends rest on the assumption that the proportion of each main work activity is somewhat uniform from one year to the next. Thus, we assume that that this calculation is reliable and represents real life in the construction sector. Second, according to the ‘‘Statistic Denmark” our samples are representative of the construction sector in Denmark concerning age distribution, as well as the distribution of small, middle range and large construction companies (Statistic Denmark, 2015). Further, the data cover all persons who work in the construction sector, including the self-employed and those belonging to a family company, who are not obliged to insure themselves, and thus, are absent from official registers for compensation claims. Finally, the ED data has been validated continuously since 1980. The main limitation of our study is the lack of information on some common injuries that are rarely observed in the ED, but frequently reported to insurance companies/authorities e.g. low back pain (Lander et al., 2014; Welch and Hunting, 2003). 5. Conclusions In summary, the study shows that through the last three decades, the overall trends of work-related injuries within the construction sector have been remarkably constant. However, for minor subgroups such as ‘major injuries’ and ‘injuries due to young workers use of small power tools’ a significant decline was seen. Further, young workers have twice the injury risk compared to older workers during the entire observation period. We found significant cyclical fluctuations in the annual injury incidences for both age groups, but these were much more pronounced among young workers. The fluctuations among the youngest groups were significantly related to repeated economic periods where the risk for work injuries was higher during booms and lower during recessions. Further, periods with economic booms are positively related to the rate of minor injuries, whereas major injuries are unaffected. Why the seriousness of ED treated work injuries should be sensitive to business cycles is not clear, whereas changes in reporting behavior related to the business cycle may well play a role when it comes to reported injuries. Acknowledgements The project was financed by the Danish Working Environment Research Fund, project number 24-2008-09. Without meticulous registration and commitment performed by secretaries on patient arrival over three decades and managerial support for such registration this study would not have been possible. For further study
66
F. Lander et al. / Safety Science 85 (2016) 60–66
of specific hypotheses the collected data can be made available following agreement with author J. Lauritsen at [email protected]. References Asfaw, A., Pana-Cryan, R., Rosa, R., 2011. The business cycle and the incidence of workplace injuries: evidence from the U.S.A. J. Saf. Res. 42, 1–8. Boone, J., van Ours, J.C., Wuellrich, J.P., Zweimüller, J., 2006. Recessions are bad for workplace safety. J. Health Econ. 25, 1069–1093. Danish Health and Medicines Authority, 2015. (in Danish). Danish Labor Inspection, 2015. (in Danish). Davies, R., Jones, P., Nuñez, I., 2009. The impact of the business cycle on occupational injuries in the UK. Soc. Sci. Med. 69, 178–182. Fabiano, B., Currò, F., Pastorino, R., 2001. Occupational injuries in Italy: risk factors and long term trend (1951–98). Occup. Environ. Med. 58, 330–338. Haslam, R.A., Hide, S.A., Gibb, A.G., Gyi, D.E., Pavitt, T., Atkinson, S., Duff, A.R., 2005. Contributing factors in construction accidents. Appl. Ergon. 36, 401–415. Laitinen, H., Marjamäki, M., Päivärinta, K., 2010. The validity of the TR safety observation method on building construction. Accid. Anal. Prev. 31, 463–472. Laitinen, H., Päivärinta, K., 2010. A new-generation safety contest in the construction industry – a long-term evaluation of a real-life intervention. Saf. Sci. 48, 680–686. Lander, F., Nielsen, K.J., Rasmussen, K., Lauritsen, J.M., 2014. Patterns of work injuries: cases admitted to emergency room treatment compared to cases reported to the Danish working environment authority during 2003–2010. Occup. Environ. Med. 71, 97–103. Lipscomb, H.J., Schoenfisch, A.L., Cameron, W., Kucera, K.L., Adams, D., Silverstein, B. A., 2014. How well are we controlling falls from height in construction? Experiences of union carpenters in Washington State, 1989–2008. Am. J. Ind. Med. 57, 69–77. Lipscomb, H.J., Schoenfisch, A.L., Shishlov, K.S., Myers, D.J., 2010. Nonfatal tool- or equipment-related injuries treated in US emergency departments among workers in the construction industry, 1998–2005. Am. J. Ind. Med. 53, 581–587. Loomis, D., Richardson, D.B., Bena, J.F., Bailer, A.J., 2004. Deindustrialisation and the long term decline in fatal occupational injuries. Occup. Environ. Med. 61, 616–621.
McCoy, A.J., Kucera, K.L., Schoenfisch, A.L., Silverstein, B.A., Lipscomb, H.J., 2013. Twenty years of work-related injury and illness among union carpenters in Washington State. Am. J. Ind. Med. 56, 381–388. Nielsen, K.J., Lander, F., Lauritsen, J.M., 2015. The relationship between macroeconomic and industry-specific business cycle indicators and workrelated injuries among Danish construction workers. Occup. Environ. Med. 72, 271–276. OSHA.EUROPA.EU, 2015.. Salminen, S., 2004. Have young workers more injuries than older ones? An international literature review. J. Saf. Res. 35, 513–521. Sancini, A., Fioravanti, M., Andreozzi, G., Giorgio, V.D., Tomei, G., Tomei, F., et al., 2012. Meta-analysis of studies examining long-term construction injury rates. Occup. Med. 62, 356–361. Schoenfisch, A.L., Lipscomb, H.J., Shishlov, K., Myers, D.J., 2010. Nonfatal construction industry-related injuries treated in hospital emergency departments in the United States, 1998–2005. Am. J. Ind. Med. 53, 570–580. Siu, O.L., Phillips, D.R., Leung, T.W., 2003. Age differences in safety attitudes and safety performance in Hong Kong construction workers. J. Saf. Res. 34, 199–205. Statistic Denmark, 2015. . Swuste, P., Frijters, A., Guldenmund, F.W., 2012. Is it possible to influence safety in the building sector? A literature review extending from 1980 until the present. Saf. Sci. 50, 1333–1343. van der Molen, H.F., Lehtola, M.M., Lappalainen, J., Hoonakker, P.L., Hsiao, H., Haslam, R., et al., 2012. Interventions to prevent injuries in construction workers. Cochrane Database Syst. Rev. 12, CD006251. Welch, L.S., Dong, X., Carre, F., Ringen, K., 2007. Is the apparent decrease in injury and illness rates in construction the result of changes in reporting? Int. J. Occup. Environ. Health 13, 39–45. Welch, L.S., Hunting, K., 2003. Injury surveillance in construction: what is an ‘‘injury”, anyway? Am. J. Ind. Med. 44, 191–196. Westaby, J.D., Lowe, J.K., 2005. Risk-taking orientation and injury among youth workers: examining the social influence of supervisors, coworkers, and parents. J. Appl. Psychol. 90, 1027–1035. Wickizer, T.M., Kopjar, B., Franklin, G., Joesch, J., 2004. Do drug-free workplace programs prevent occupational injuries? Evidence from Washington State. Health Serv. Res. 39, 91–110.
Contents lists available at ScienceDirect
Safety Science journal homepage: www.elsevier.com/locate/ssci
Work injury trends during the last three decades in the construction industry Flemming Lander a,⇑, Kent Jacob Nielsen b, Jens Lauritsen c a
Dept. of Occupational Medicine, Odense University Hospital, Denmark Danish Ramazzini Centre, Department of Occupational Medicine, Herning Regional Hospital, Denmark c Accident Analysis Group, Dept. of Orthopedic Surgery, Odense University Hospital, Denmark and Institute of Clinical Medicine, University of Southern Denmark, Denmark b
a r t i c l e
i n f o
Article history: Received 1 April 2015 Received in revised form 25 August 2015 Accepted 27 October 2015
Keywords: Longitudinal study Age Type of injury Work activities Business cycle Emergency department Denmark
a b s t r a c t Objectives: The aim was to analyze injury trends according to age, severity, work activity and business cycle in the construction sector. Methods: From 1980 to 2010 a total of 23.464 work related injuries were treated at the emergency department at Odense University Hospital. The annual incidences were calculated. Employment levels in the construction sector were used as an indicator of fluctuations in the business cycle since 1980. Results: Through the last three decades the overall trend of work-related injuries was unchanged. For some subgroups of injuries, such as major injuries and injuries due to young workers use of small powered tools significant downward trends were seen, but trends within different age groups of workers were unchanged and young workers have at least twice the risk compared to older workers. The fluctuations in work injury trends among workers under 30 years of age were significantly related to the business cycle, where the risk of injuries was higher during economic booms than during recessions. Further, periods with economic booms are positively related to the rate of minor injuries and injuries due to all other work activities than the use of power tools. Conclusion: Overall the number of injuries in the construction sector have not changed significantly during the recent three decades, except for minor subgroups related to ‘major injuries’ and ‘injuries due to use of small power tools’. Re-evaluation of safety prevention programs is needed in order to break the high injury level among young workers compared to older workers, especially during economic booms. Ó 2016 Published by Elsevier Ltd.
1. Introduction A recent study revealed that the overall trend of work injuries in the Danish construction industry has been unchanged during the last three decades, which stands in contrast to the general downwards trend seen in work injuries in Denmark across all sectors (Nielsen et al., 2015). This long term trend as well as the pattern of annual fluctuations were similar for work injuries treated at a emergency department (ED) and for injuries reported to the Danish Working Environment Authority (Nielsen et al., 2015). This is in spite of a very small individual overlap between these two datasets. In addition, each data set represents to a large extent different injury events and different types of injuries (Lander et al., 2014). Thus, the strong concordance in the time trends and fluctuations indicates that injury frequency is influenced by one or more common underlying factor. ⇑ Corresponding author. Tel.: +45 26353208; fax: +45 65 41 49 88. E-mail address: [email protected] (F. Lander). http://dx.doi.org/10.1016/j.ssci.2015.10.013 0925-7535/Ó 2016 Published by Elsevier Ltd.
So far, no other European studies or surveillance reports covering work injury rates in the construction industry for this period exist. In a US setting, a significant decline in rates of reported work-related injuries and illnesses has been observed over the preceding 20-year period among union carpenters (McCoy et al., 2013). In particular, a substantial decline was observed for falls from height among these construction workers (Lipscomb et al., 2014). Also, the US Bureau of Labor Statistics observed a downward trend in reported work injuries in the construction industry as a whole (Welch et al., 2007). Inconsistencies in the information suggest that some of the apparent decrease may be due to changes in the ways injuries are treated, misclassification of employees, or underreporting (McCoy et al., 2013; Welch et al., 2007). Construction has always been regarded as one of the most hazardous industries (Swuste et al., 2012). It is considered as a special case, as the nature of the work entails that automation and many technological improvements that have led to improved safety in other industries has not had the same impact within construction (Swuste et al., 2012). Thus, in this context the European Union (EU)
61
F. Lander et al. / Safety Science 85 (2016) 60–66
2. Materials and methods
A total of 23.464 injuries within the construction industry between 1980 and 2010 were observed. Table 1 shows the dichotomized age distribution of the injuries. Almost half of all injuries involved construction workers below 30 years of age. Clinical information consisted of injury diagnosis based on ICD8 (1980– 1992) and ICD10 (1993–2010), which was transformed into main type of injuries and body parts. ‘Major injuries’ include amputations, bone fractures, joint sprains and strains, all types of soft tissue damage, eye corrosion and burns, and electrical shock. ‘Minor injuries’ mainly consisted of superficial lacerations and wounds, and being struck by a foreign body (mainly in eyes). The ratio between major and minor injuries was one to two, see Table 1. The ED register also contains individual information on work activity leading to the injury (Table 1). These different types of activities were transformed into three main activities: (1) ‘Movement activities’, which include slips, trips, falls, striking an object when walking, handling or carrying materials. (2) ‘Work with small power tools’, e.g. electric or air pressure powered hand tools such as angel grinders, nail guns, drilling machines, as well as machinery in a stationary frame such as smoothing plane or band saw. (3) ‘Miscellaneous work activities’ include use of non-powered hand tools, various assembly or disassembly work tasks and a heterogeneous number of activities e.g. activities involving big machines such as cranes and trucks, and exposure to chemicals. We assumed that the proportion of workers at risk on the construction sites who perform these three categories of work tasks remains approximately constant from year to year during the observation period. In the present study, the annual level of employment within the Danish construction sector was used to describe the business cycle from 1980 to 2010, see Fig. 1. Three major economic recessions can be observed with steep falls in employment in the first years of the 1980s, the first years of the 1990s, and the late years of the 2000s. A minor recession is also seen in the mid 2000s. The observed trend corresponds closely to trends in other measures of the business cycle, such as general unemployment, indicating that they measure the same underlying economic phenomenon (Asfaw et al., 2011). The correlation between annual level of employment within construction and the general unemployment level in Denmark was 0.83 (P < 0.01). Data was obtained from Bureau of Statistics Denmark.
2.1. The observation data
2.2. Statistics
The population base of the study is work sites situated in the catchment area of the Emergency Department (ED) at Odense University Hospital (OUH), which is situated in the city of Odense on the island of Funen, Denmark. The department is the only emergency room in that part of the island and the ED is a free 24-h emergency service. The catchment area for OUH is a well defined mixed rural and urban geographic area with a population of 362,000 inhabitants, which represents approximately 6.7% of the Danish population. The area is demographically and industrially comparable to Denmark as a whole (Statistic Denmark, 2015). Since 1980, the Accident Analysis Group (UAG) at the ED, OUH, has conducted systematic and quality-assured recording of all treated injuries. Each year 35–40.000 injury patients are treated at the ED. About 10% of these injuries occur during paid work and are classified as occupational. In 2010 the UAG ED-database contained information on approx. 160.000 occupational injuries. All injured individuals are indentified by a unique 10-digit personal civil registry number (CRN), and the CRN combined with year of injury was the key for linking to Statistics Denmark’s Integrated Database for Labour Market Research, IDA (Statistic Denmark, 2015). IDA contains individual information on industry ties on an annual basis from 1980 to 2010. The IDA 6 digit industry codes were transformed into an overall construction sector code.
Total annual employment and age stratified employment in construction in the ED catchment area formed part of the denominator in the equations of the injury incidence rate calculations. Incidence was defined as the number of injuries per 1000 workers employed (Table 2). The table shows the distribution of injured construction workers, and the total number of employed workers in the catchment area of the Emergency Department according to
since the early 1990s had enacted several directives, guidelines and standard directed toward safety and health at work e.g. Directive 92/57/EEC and the majorities of the member states including Denmark had fully implement specific legislation directed toward occupational safety in the construction sector (osha.europa.eu, 2015). In Denmark, as well as other western countries, much has been done during the last 20–30 years to regulate and improve work environment standards on building sites (Fabiano et al., 2001; Haslam et al., 2005; Danish Labor Inspection, 2015). Many initiatives have repeatedly been introduced, either through legislation or labor market agreements between employers and employees, both in specific industries and at enterprise level e.g. promotion of drug and alcohol-free workplaces, safety training programs, establishing safety groups on work-sites, but unfortunately few formal scientific evaluations have been undertaken measuring the effect on work injuries (Loomis et al., 2004; Fabiano et al., 2001; Wickizer et al., 2004; The Danish Health and Medicines Authority, 2008). Parallel to these initiatives, recurring economic recessions and booms within the construction sector, as well as other industries, continuously alter work conditions e.g. through changes in the number of hours worked, pacing of work, and changes in the level of recruitment of inexperienced labor (Asfaw et al., 2011). As a consequence, a declining number of reported work injuries are observed during recessions while an increase is seen during booms – a trend that the construction industry might be especially sensitive to (Nielsen et al., 2015; Asfaw et al., 2011). It is clear that in spite of formal work standards and legislations work environment practices on construction sites are not fixed, but are altering constantly over time. However, very little is known about the impact of these dynamic processes on injuries in different subgroups of workers or on labor force characteristics. The aim of the current study was to describe long term trends in injury incidence rates in different age-groups, in different type of injuries, and in injuries due to different work activities in the construction sector, and to analyze the impact of the business cycle on work injuries according to these labor force characteristics.
Table 1 Background indicators. Indicators
N
% (95% CI)
Age (years) <30 P30
11,028 12,436
47.0 (46.3–47.6) 53.0 (52.0–53.9)
Type of injuries Minor injuries Major injuries
15,972 7492
68.1 (67.0–69.1) 31.9 (31.2–32.7)
Injuries due to Movement activities Use of small powered tools Other work activities
7893 5024 10,547
33.6 (32.9–34.4) 21.4 (20.8–22.0) 45.0 (44.1–45.8)
Total number of injuries
23,464
F. Lander et al. / Safety Science 85 (2016) 60–66 200000
variability and trend using linear regression analysis. All analyses were performed in STATA 11.
160000
180000
3. Results
140000
Total employment in construction industry
62
1980
1990
2000
2010
year Fig. 1. The course of the business cycle indicator expressed by the annual level of employment within the construction sector in Denmark from 1980 to 2010.
Table 2 Distribution of injured construction workers, and the total number of employed workers in the catchment area of the Emergency Department according to age and year of injuries. Year
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Age < 30 years
Age P 30 years
Injuries n
Total employment N
Injuries N
Total employment N
501 368 360 426 452 441 520 492 459 406 353 247 297 186 261 237 278 386 340 388 390 367 279 262 313 339 384 451 301 288 256
4382 3777 3356 3182 3113 3370 3707 3885 3853 3525 3312 3160 3083 2840 2410 2651 2761 2821 2926 2961 2870 2977 2728 2521 2521 2705 2864 3091 3242 2925 2262
429 428 380 399 432 424 491 489 471 423 416 220 328 198 270 256 356 380 423 418 493 409 425 407 410 507 464 541 375 380 394
8285 6918 6518 6137 6374 6337 6940 7111 7055 6849 6682 6577 6787 6160 5505 6035 6213 6277 6374 6406 6723 7291 7376 6634 6733 7244 7827 8163 8318 7775 7146
age and year of injuries. The business cycle indicators were tested for stationarity using the augmented Dickey–Fuller (ADF) test. The ADF test revealed that the indicator ‘‘annual level of employment within the total construction sector” was non-stationary, whereas in turn the first difference of the variable (the change score from one year to next year) was stationary (t = 4.12, P = 0.001). Thus, this change score was used as an estimate of the impact of the business cycle on the incidence of work injuries. The longitudinal, annual incidence data and its correlation to age, type of injury, and work activity, as well as the business cycle indicator, was tested for
Fig. 2 shows the mean annual incidence of injuries according to 3 different indicators from 1980 to 2010. Graph A shows the courses and trends in two age groups. The estimated linear trends are time constant for both age groups – see also Table 2, which shows coefficients and p-values. Visually, the course shape of the incidence seems to be close to that of the business cycle indicator in Fig. 1 – with a general decrease in injury incidence during recession periods and visa versa during booms. This picture seems far more profound in the age groups below 30 years compared to the workers older than 30 years. Further, it can be seen that the injury incidences are approx. 2–3 times higher for workers below 30 years of age than for those more than 30 years of age. The graph B shows the course and linear trends stratified into major and minor injuries according to age groups. For both age groups a moderate but significant reduction at 0.25 and 0.13 for major injuries per 1.000 workers per year is observed, whereas for the minor injuries, the trend is constant for workers below 30 years of age and slight increasing for those over 30 years, see Table 2. Again, the shape of the incidence courses seems to be closely related to that of the business cycle indicator. This is especially noticeable for the minor injuries. The graph C shows the course and linear trends stratified into injuries due to the three different main work activities according to age groups. Overall, a general declining trend through all work activities among the youngest age group is seen, which is contrary to the trends observed among the oldest worker group, – the injuries related to work with power tools and other work activities revealed small, but significant decreases in incidence rates at 0.13 and 0.18 injuries per 1000 workers per year, whereas the declining trend is borderline significant for injuries due to movement activities (see Table 2). Again, the shape of the incidence courses during the thirty years of observation seems to be close to that of the business cycle indicator, especially for injuries due to movement activities and other work activities. Fig. 3 shows the correlations expressed by the linear regression trend line between annual injury incidences and annual change in total employment in the construction sector, from 1980 to 2010, according to the three different indicators. The matching linear coefficients and p-values are shown in Table 3. Overall, the linear trends do not indicate a strong relationship between the business cycle indicator and the three indicators, but for workers under 30 years, the relationship is significant and indicates that these workers have a slightly higher risk for work injuries during economic booms. Workers over the age of 30 seem also to be affected by changes in the business cycle, but the association is only borderline significant. Also, minor types of work injuries and injuries due to the category ‘‘other work activities” occur more frequently during economic booms than during recessions, whereas major types of injuries and injuries due to use of power tools are unaffected by changes in the business cycle. A declining number of injuries due to movement activities are borderline associated to periods of economic booms (see Table 4). 4. Discussion Longitudinal observations over a long time span are rare, but in turn might provide insight into possible accumulated effects of repeated preventive initiatives as well as changes in safety attitudes on work sites. This is rarely measurable in studies with short time spans or specific intervention trials (van der Molen et al., 2012; Sancini et al., 2012). Long time series also provide
63
F. Lander et al. / Safety Science 85 (2016) 60–66
>= 30 years
50
75
100
125
150
175
< 30 years
25
Injuries per 1000 workers (95% poisson CI)
Graph A
1980
1990
2000
Construction sector.
2010 1980
1990
2000
2010
Minor injuries & >= 30 years
Major injuries & < 30 years
Major injuries & >=30 years
10
20
30
40
0
10
20
30
40
Minor injuries & < 30 years
0
Injuries per 1000 workers (95% poisson CI)
Graph B
1980
1990
2000
2010 1980
1990
2000
2010
Construction sector.
Small powered tools & <30 years
Other work activities & < 30 years
Movement activities & >=30 years
Small powered tools & >=30 years
Other work activities & >=30 years
10
20
30
0
10
20
Movement activities & <30 years
0
Injuries per 1000 workers (95% poisson CI)
30
Graph C
1980
1990
2000
2010 1980
1990
2000
2010 1980
1990
2000
2010
Construction sector. Fig. 2. Work injury incidence from 1980 to 2010 by age, severity and work activity.
opportunities to assess transient progress and regressions in prevention of work injuries, including those that follow economic cycles, which seem to have a profound influence on levels of work injuries (Asfaw et al., 2011; Davies et al., 2009; Boone et al., 2006; Nielsen et al., 2015). Short term intervention studies are
particularly sensitive to these external time dependent fluctuations when assessing the results of a specific intervention on work injuries. In this context, our study revealed some characteristic patterns of work injuries in relation to different labor force characteristics, and some of the associations suggest new perspectives:
F. Lander et al. / Safety Science 85 (2016) 60–66 Minor injuries
Major injuries
60
100
Incidence pr 1000 workers
80
>= 30 years
0
20
50
Incidence pr 1000 workers
150
< 30 years
40
64
-20
-10
0
10
-20
-10
0
-20
10
0
10
-20
-10
0
10
10 20 30 40
Small powered tools
-20
-10
0
10
Other work activities
10 20 30 40
Incidence pr 1000 workers
Movement activities
-10
delta (employment in construction(N/1000))
delta (employment in construction(N/1000))
-20
-10
0
10
delta (employment in construction(N/1000)) Fig. 3. Linear regression of mean annual injury incidences from 1980 to 2010 and the national employment rate in construction industry by age, severity and work activity.
Table 3 The regression coefficient for injury trends between 1980 and 2010 among construction workers according to different age groups, different type of injuries and injuries due to different work activities. Indicators Age (years) <30 P30 Type of injuries Minor injuries <30 P30 Major injuries <30 P30 Injuries due to Movement activities <30 P30 Use of powered tools <30 P30 Other work activities <30 P30 *
Coeff. 0.03 0.12
SE 0.40 0.21
t-Value 0.07 0.57
0.14 0.24
0.09 0.10
1.54 2.43
0.25 0.13
0.04 0.04
5.69 2.97
Table 4 The association between mean annual work place injury incidences and total annual employment in the constructions sector according to different indicators. Coeff.
SE
t-Value
p-Value
Age <30 P30
231.81 276.95
69.63 147.04
3.33 1.88
0.002* 0.07
Type of injuries Minor injuries Major injuries
480.84 260.51
158.91 303.77
3.03 0.86
0.005* 0.40
Injuries due to Movement activities Use of powered tools Other work activities
670.11 469.74 702.54
337.18 454.10 233.17
1.99 1.03 3.01
0.06 0.309 0.005*
p-Value 0.95 0.58
0.14 0.02* 0.0001* 0.006* *
0.08 0.11
0.04 0.06
1.99 1.92
0.13 0.05
0.04 0.03
3.52 1.53
0.001* 0.14
0.18 0.05
0.06 0.06
2.87 0.80
0.008* 0.43
p-Values < 0.05.
0.06 0.06
p-Values < 0.05.
First, general, but modest long term declines were seen in time trends for major injuries combined regardless of workers age, for injuries among young workers due to work with powered tools and for the heterogenic category of other work activities. Conversely, work-related minor injuries have increased for workers that are more than 30 years old during the last three decades. A few longitudinal studies covering all industries during the preceding decades also have observed a decline in both the number of
serious and non-serious injuries, but so far no such observations have been observed for the construction sector as a whole. The most consistent observation in several studies has been a decline in fatal injuries across all industries, as well as in the construction sector (Sancini et al., 2012). Fatal injuries have been more than halved from the 1980s to the present in the Danish construction industry (www.at.dk, 2015). This indicates that some improvements in work safety have been achieved and according to our study, the same repeated initiatives might also have affected major injuries, but not minor injuries. A previous study have shown a decline of serious injuries (‘‘fall from height”) in a subgroup of construction workers (carpenters) during the preceding 2 decades (Lipscomb et al., 2014). In our study, ‘‘fall from height” is included in the category ‘‘movement work activities”, but represents only a small fraction of all types of injuries due to these activities, and thus, has no real impact on the overall trend. One remarkably
F. Lander et al. / Safety Science 85 (2016) 60–66
observation in our study is the decline in injuries due to work with powered tools, but only among the young workers, not the older workers. This is new information, which to our knowledge has not previously been described. In our study power tools represent many different hand held tools and small tools in a stationary frame. From another study of ED treated work injuries, saws, nail guns, welding tools and drills seem to be the most common cause of work injuries, representing 65% of all powered tool-related injuries (Lipscomb et al., 2010). As in the US study, our results indicate that many of the technological improvements and changes in safety standards that have led to improved safety elsewhere have also had similar impact in some subgroups of construction workers (Lipscomb et al., 2010). Even though the overall long term work injury trends are constant, the picture of young workers having at least twice the risk compared to older workers is maintained during the whole observation period. Hypothetically, the result could be biased, if health behavior has changed during the last 30–35 years, but we have no information on any age-specific changes in health behavior, including the use of hospital emergency departments. During the whole observation period, there has been free and open access to unrestricted treatment at all EDs in Denmark, with no economic consequences for either workers or companies and without any notification to authorities or employer. Thus, an injured worker can seek treatment at any ED free of charge, without company knowledge, and independent of possible future compensation claims (Nielsen et al., 2015; Lander et al., 2014). In many studies, age seems to act as a strong risk predictor of injuries. It is well documented, especially in the construction sector, that the frequency of injuries is higher among young workers (Salminen, 2004; Siu et al., 2003; Westaby and Lowe, 2005). Two US-studies, both based on ED sources between 1998 and 2005, revealed incidence rates between 51–81 and 28–54 per 1.000 construction workers under 35 years respectively, a range 2–3 times higher than found in older workers (Schoenfisch et al., 2010). Secondly, we found significant cyclical fluctuations in annual injury incidences through all three decades in both age groups. However, the fluctuation was especially pronounced among young workers and less for the oldest age group. The reason for that is not clear, but for the youngest workers, some of that fluctuation was related to changing periods of economic boom and recession. In our study at least 4 such periods were observed. During booms the risk for work injuries among young workers was significantly higher than for older workers, who were less affected by the business cycle. The relationship between business cycles and injury frequency is well known (Asfaw et al., 2011; Davies et al., 2009; Boone et al., 2006). However, no previous studies have looked at specific age-stratified trends. Multiple factors have been suggested to explain this general relationship between the business cycle and workplace safety. Some of these changes are probably a mix of task-specific factors (such as higher work pace during booms), and organisational factors (such as less training of new employees or longer working hours during booms) (Asfaw et al., 2011). These cyclic changes seem to affect the younger workers more than older workers, probably because young workers are less experienced and trained (Salminen, 2004; Siu et al., 2003; Westaby and Lowe, 2005). Further, in line with a Dutch and a British study, our results also show that economic booms are positively related to the rate of minor injuries, whereas major injuries are independent (Davies et al., 2009; Boone et al., 2006). Why the seriousness of work injuries is sensitive to the business cycle is not clear. In the context of reported injuries, the association might be explained as a consequence of changes in reporting behavior during economic recessions. Because of high unemployment, greater fear of dismissal could lead to a lower level of reporting of minor injuries compared to major injuries (Davies et al., 2009). But our results rest on ED
65
data, which in a Danish context are more or less independent of changes in reporting behavior, because injured workers do not have to fear dismissal when companies have no knowledge of the ED treatment, and there are no treatment costs or economic incentives keeping injured workers from seeking treatment (Nielsen et al., 2015). Thus, it is plausible to presume that higher work pace, poorer training of new employees and longer working hours during booms negatively influence worksite conditions, e.g. order and tidiness on building sites, which are well known risk factors associated to work injuries (Laitinen et al., 2010; Laitinen and Päivärinta, 2010). The study has several strengths. First, we were able to determine the exact age specific annual injury incidence based on reliable employment data during more than three decades. The key for that is the individual unique 10-digit personal civil registry number, which links every single person in Denmark to a work place since the introduction of the system in 1972. Calculation of annual incidences according to different work activities rests on the total employment in the ED catchment area, and the calculation of the trends rest on the assumption that the proportion of each main work activity is somewhat uniform from one year to the next. Thus, we assume that that this calculation is reliable and represents real life in the construction sector. Second, according to the ‘‘Statistic Denmark” our samples are representative of the construction sector in Denmark concerning age distribution, as well as the distribution of small, middle range and large construction companies (Statistic Denmark, 2015). Further, the data cover all persons who work in the construction sector, including the self-employed and those belonging to a family company, who are not obliged to insure themselves, and thus, are absent from official registers for compensation claims. Finally, the ED data has been validated continuously since 1980. The main limitation of our study is the lack of information on some common injuries that are rarely observed in the ED, but frequently reported to insurance companies/authorities e.g. low back pain (Lander et al., 2014; Welch and Hunting, 2003). 5. Conclusions In summary, the study shows that through the last three decades, the overall trends of work-related injuries within the construction sector have been remarkably constant. However, for minor subgroups such as ‘major injuries’ and ‘injuries due to young workers use of small power tools’ a significant decline was seen. Further, young workers have twice the injury risk compared to older workers during the entire observation period. We found significant cyclical fluctuations in the annual injury incidences for both age groups, but these were much more pronounced among young workers. The fluctuations among the youngest groups were significantly related to repeated economic periods where the risk for work injuries was higher during booms and lower during recessions. Further, periods with economic booms are positively related to the rate of minor injuries, whereas major injuries are unaffected. Why the seriousness of ED treated work injuries should be sensitive to business cycles is not clear, whereas changes in reporting behavior related to the business cycle may well play a role when it comes to reported injuries. Acknowledgements The project was financed by the Danish Working Environment Research Fund, project number 24-2008-09. Without meticulous registration and commitment performed by secretaries on patient arrival over three decades and managerial support for such registration this study would not have been possible. For further study
66
F. Lander et al. / Safety Science 85 (2016) 60–66
of specific hypotheses the collected data can be made available following agreement with author J. Lauritsen at [email protected]. References Asfaw, A., Pana-Cryan, R., Rosa, R., 2011. The business cycle and the incidence of workplace injuries: evidence from the U.S.A. J. Saf. Res. 42, 1–8. Boone, J., van Ours, J.C., Wuellrich, J.P., Zweimüller, J., 2006. Recessions are bad for workplace safety. J. Health Econ. 25, 1069–1093. Danish Health and Medicines Authority, 2015.
McCoy, A.J., Kucera, K.L., Schoenfisch, A.L., Silverstein, B.A., Lipscomb, H.J., 2013. Twenty years of work-related injury and illness among union carpenters in Washington State. Am. J. Ind. Med. 56, 381–388. Nielsen, K.J., Lander, F., Lauritsen, J.M., 2015. The relationship between macroeconomic and industry-specific business cycle indicators and workrelated injuries among Danish construction workers. Occup. Environ. Med. 72, 271–276. OSHA.EUROPA.EU, 2015.