Investigation of Causative Factors Associated with Summertime Workplace Fatalities
This study has reviewed the workplace fatality data recorded by the Department of Labour from 2000 to 2005 inclusive in order to investigate seasonal trends and possible contributory factors. The findings of this study are discussed in the following sections, along with associated strategies for the reduction of workplace fatality risks.
This study has confirmed an annual trend of workplace fatalities with a peak between 2001 and 2002 (73-74 cases per year). Since then there was a gradual decline in the workplace fatal incidents until 2004. However, the occurrence of fatal incidents at work increased by some 38% from 2004/05 to 2005/06 (Figure 1). Such a trend has been confirmed by normalised data (Figure 2) and by statistical test.
Trends of workplace fatalities and serious harm injuries
By comparing the trend of fatality data with the serious harm data over the same 6-year period (Figure 3), it is interesting to see that the pattern of changes of the two types of incidents are different year on year. The pattern of serious harm incidents is not indicative of the pattern of workplace fatalities. For example, the workplace fatality rate first decreased from 0.97 per 100,000 workers in 2002 down to 0.79 in 2003, and to 0.58 in 2004; then increased to 0.78 in 2005. During the same period, however, the rate of serious harms first increased from 0.75 per 1000 workers in 2002 to 0.85 in 2004, and then dropped to 0.5 by 2005. The reasons for such a controversial trend in these two types of incident data are unknown, but one explanation is that it may be due to the change in the incident reporting/data recording system over the past two to three years. It was pointed out by some experts that there had been an issue of over-reporting of serious harms in the past, but this situation may have been improved through policy changes for the assessment and recording of work-related serious harms in the database.
Preliminary analysis of seasonal trends in the serious hard data showed further disparities with departmental fatality data; see the Discussion section on 'Seasonal Trends in the serious harm data'.
While the focus of the present study is on the issue of workplace fatalities, it is suggested that further research should be carried out to study the data of work-related serious harms for their seasonal trends and for their causal factors.
Trends by industry and by regions
The annual trends by industry (Figures 4 & 5) indicate that industrial sectors such as 'Agriculture, Forestry & Fishing', 'Transport & storage', and 'Construction' have shown a tendency of increased workplace fatalities since 2003. However, the current data set is not large enough to provide a conclusive result on these trends. Similar issue exists with the annual trend by regions (shown in Figure 6). Therefore, given the current data set, it is considered to be more appropriate to look at the annual trend of workplace fatalities at a national level, rather than by industries or by regions (Figure 6).
The seasonal trend of workplace fatalities is the primary focus area of the present study.
The monthly trend of the 6-year data (2000-05) appears to show a January peak (Figure 7), although such a trend cannot be confirmed with statistical testing because of the small data set.
However, when analysing the data by seasons (Figure 8), the data did show a significant seasonal variation (p≤0.05) and a summer peak, followed by autumn, in the occurrence of workplace fatalities. This trend has been supported by the normalised results (Figure 9) as well as by a sufficient amount of independent evidence (summarised in Table 5) confirming that more work-related incidents (including fatal and non-fatal) tend to occur in the summer and autumn months.
Seasonal Trends in the serious harm data
A preliminary analysis of the serious harms data over the same period (2000-05) indicated that the workplace incidents resulting in serious harm injuries occurred more often during autumn and winter than in spring and summer. In fact, summer season had the lowest level of serious harm injuries. This trend is highly significant (p<0.001). The reason for the difference in the seasonal trends between workplace fatalities and serious harms is not well understood at this stage. There is an assumption that lower serious harm numbers may be an artefact of lower workplace participation over the summer holiday period. Further investigation is required to understand and analyse the serious harms data in greater detail, using normalisation to account for the effects of seasonal labourers etc.
Seasonal trend by industries
When analysing the seasonal trend of workplace fatalities by industry, classified according to the ANZSIC system, only the agriculture industry showed a significant seasonal variation in the fatality rate (p≤0.01, Figure 10). This also suggests that the overall national trend is mainly influenced by the trend of this sector. This result is not surprising because the work activities in the agriculture industry are highly seasonal, and much independent evidence (Table 5) has indicated that, as far as work-related incidents are concerned, the agriculture industry is more vulnerable to the summer season than other industries.
For the agriculture sector, January is seen to have the highest workplace fatality rate, followed by two more peaks, one in April, and the other in October (Figure 11). These trends are found to be statistically significant (p≤0.05). This implies that some intervention measures could be developed to target the work activities of workers in this industry during these peak periods.
Further data analysis within the agriculture sector suggested that 'Forestry and logging' is the sub-sector with the highest summertime fatality rate, followed by 'Horticulture & fruit growing' (Figure 12). However, while these trends may exist, the current data set cannot technically confirm that these trends are not by chance. This issue will need to be considered with other evidence. Discussions with the experts who are familiar with these sub-sectors suggest that the work activities in these two sectors tend to be more intensive during the summer months than other seasons of the year. There is also an issue of casual labour in these sectors, especially in 'Horticulture & fruit growing', along with the increased work demand in summer months. This will be further discussed under 'Workplace factors'.
Seasonal trend by regions
The mid-north region and the southern region have been found to be most affected by the summer season with regard to the occurrence of work-related fatal incidents (Figure 13). These results are not surprising because one region (mid north) is relatively more populated with seasonal related industries and work activities, and the other region (southern) is sensitive to seasonal changes with more agricultural activities. Nevertheless, the findings have suggested that some focused campaigns may be beneficial to the improvement of workplace safety in these regions during the summer.
Seasonal trend by employment status
Initial analysis of Department of Labour data suggests that it is the employed who are most at risk of having a workplace fatality. This result is in agreement with a previous study which also showed that in New Zealand the majority of work-related deaths were of employees (New Zealand Environmental and Occupational Health Research Centre, 1999). However, this result cannot be made general until denominator data can be obtained. Given that the numbers of self employed are likely to be relatively low, normalised data may show a very different picture.
An American study has reported that, although wage and salary workers suffered more than three times as many fatal work-related injuries as did self-employed workers, when the data were normalised and fatality rates were compared, it became evident that self-employed workers were 2.7 times more likely to be victims of fatal work injuries than their wage and salary counterparts (Pegula, 2004). Combining this independent evidence with the findings of higher seasonal related workplace fatalities in the 'Agriculture, Forestry & Fishing' sector (shown in Figure 15), it appears to be reasonable to suggest that more preventative efforts can be made to the self-employed workers in the agricultural industry to improve their safety at work.
For different regions, there are some apparent seasonal changes in workplace fatalities with the employee status (Figure 16). For example, in the northern region, there were more work-related fatalities involving employees in the summer than in other seasons; and in the southern region there were more fatal incidents involving employees in the autumn. In general, there appears to be a trend that in the northern part of the country (northern, mid north), more work-related fatal incidents involving the employees tend to occur in the summer; and in the southern part of the country (central, southern) these tend to occur in the autumn (Figure 17).
Seasonal trend by age groups
An important finding of this study is that, throughout New Zealand, workers of 55-64 years of age have the highest incidence of summertime workplace fatalities (Figure 18). This result has been confirmed by data normalisation (Figure 19) and by statistical test. An independent study has also confirmed that "regardless of the type of workers, the fatality rate of workers aged 55 and older is greater than the fatality rate of workers aged 16 to 54 (Pegula, 2004). The same author (Pegula, 2004) also pointed out that the self-employed workers are also generally older than those wage and salary workers.
The present study also found that the workers between 35 and 44 years of age are more likely to have a work-related fatal incident in the autumn. The trend has been confirmed by data normalisation and statistical tests. This is an interesting finding since little published information has been found which directly links the workers of the 35-44 age group with the work-related fatalities in autumn.
Further data analysis by industry confirmed that both seasonal peaks of work-related fatalities (i.e., 55-64 age group for summer, 35-44 for autumn) were significant (p≤0.05) for the agriculture sector (Figure 20). Therefore, It is evident to conclude that the workers aged 55-64 are most vulnerable during the summer, and those aged 35-44 are most vulnerable during the autumn, and most of these workers are working in the agriculture industry.
Several ISEs interviewed through the course of this work speculated that the development of ATVs and improved ride-on farm machinery had served to extend the physical working lives of workers in the agriculture industry. Farmers that would, 10 years ago, have had to stop working in the field now continue working as the ATV driver. The very high percentage of vehicle related fatalities (over 50% see Figure 29) and the very high number of deaths arising because of vehicle (primarily ATV) rollovers supports further speculation that older men, extending their working lives through the use of farm vehicles, are the ones involved in the fatal vehicle related incidents.
The normalised results also indicate that the workers aged 65 and above had a greater work-related fatality rate than any other age groups throughout the year (Figure 19), but due to the relatively small number of fatality cases involving the people aged 65 and above in the current data set, this result was statistically insignificant. The normalised results by age group and by industry show that the workers aged 65+ tend to have a higher fatality rate in the agriculture industry during the harvest season in autumn (Figures 20 & 21). A previous New Zealand study also identified a dramatic increase in the rate of work-related fatal injuries for those over the age of 65 years (New Zealand Environmental and Occupational Health Research Centre, 1999).
Further analysis of the workplace fatality data by age group and by geographical region suggests that the peak in summertime fatalities for older workers, is more pronounced in the 55-64 age group, northern and mid north regions, and that further south (central and southern regions) the fatal incidents with this age group, along with the 35-44 age group, become more likely in the autumn (Figure 22).
In general, it is evident that, as far as worker age is concerned, the current workplace fatality data, and the relevant denominators available, support a reliable analysis of seasonal trends at the national level, and in some cases, at industry level (top ANZSIC level). While attempts have been made to analyse the data at a regional level, the results are limited by the size of the dataset and by the reliability of seasonal worker population data in these regions.
Time of day; day of the week
The overall national data from 2000 to 2005 showed two peak times when work-related fatal incidents occurred, one is between 10:30 and 12:30, and the other is around 15:00. This result is confirmed to be highly significant both for the national trend (Figure 23) and for the agriculture industry (Figure 24) (p<0.001).
While it may be that the peaks of just before lunch and during the middle of the afternoon can be explained across industry by factors such as loss of concentration before lunch hunger, or the 'post-lunch dip', closer examination of the data within the industry sub- sectors (preferably using denominator data) might reveal specific sectors where time of day is a particular concern (e.g. because of shift hand overs or end of day activities).
Similar 'time of day' effects have been reported in some previous independent studies often in the area of road transport, indicating that most incidents tend to occur between 10:00 and 15:00 during a working day (New Zealand Environmental and Occupational Health Research Centre, 2003; Nofal and Saeed, 1997; Olowokure et al., 2004).
A closer examination of the time of day data by season suggests that work related fatal incidents in summer tend to occur more in the morning time (from sunrise to noon) and work related fatal incidents in autumn tend to occur more in the afternoon (from noon to sunset). See Figure 25
By looking at the workplace fatality data by day of the week, it is within expectation that most fatal incidents occurred during weekdays than during weekends (Figure 26). This result is also largely in agreement with what is reported by the New Zealand Environmental and Occupational Health Research Centre (2003), apart from that on Saturday, the rate of work-related fatal incidents is not as high as that of fatal traffic injuries.
Primary causes of work-related fatal incidents
According to the workplace fatality investigation reports, the primary cause (or 'agent') of workplace fatalities is vehicle rollover, accounting for over 23% of the total work-related deaths that have been investigated over the past 6 years (2000-05). This was followed by fall from height as a second cause, accounting for 9.7%, and non-vehicle crush injuries as a third cause at 8.7%. The fatal crush injuries were up to 17% when considering both vehicle and non-vehicle crushes, (as shown in Figure 27).
It is noticeable that the three types of primary causes (i.e. vehicle rollover, machinery/vehicle crushes, and fall from height) accounted for almost 50% of the total workplace fatalities from 2000 to 2005. This suggested that if some effective measures can be taken to tackle the factors/reasons behind these major causes, the number of work-related fatal incidents could be significantly reduced.
Further details of the major causes for work-related death or fatal injuries, and the way that these are classified are explained below:
- Vehicle rollover (23.08%): All types of vehicles used for work, such as cars, vans, tractors and ATVs.
- Fall from height (9.70%): Where victim fell from a height more than 1 metre.
- Crush injuries (non vehicle) (8.70%): Incidents where the victim has received fatal crush injuries not directly involving a vehicle, e.g., crushed by machinery such as a recycling / rubbish compactor of truck.
- Vehicle crushes (8.36%): When victim has received fatal crush injuries involving a vehicle, e.g., crushed between vehicle and an obstacle. Excluding rollovers
- Felling incident (5.69%): Fatality incurred when victim involved in the process of felling trees.
- Vehicle incident (5.02%): When victim is involved in a vehicle incident that is other than crushes or being struck by a vehicle.
- Struck by object (5.02%): When victim was struck by object such as rocks and logs (when not actually felling), excluding struck by vehicle.
- Electrocution (5.02%): Where the victim was electrocuted.
- Drowning (4.68%): Drowning incidents, including diving fatalities where victims suffered unknown problems whilst on the dive.
- Machinery incident (3.68%): Where machinery/equipment was involved in the incident, e.g., moving parts of machinery striking/entangling victim.
- Animal incident (3.34%): If an animal contributed to the fatality.
- Slip / trip / fall (3.01%): When victim slipped, tripped or fell whilst walking or running (includes falling down stairs but excludes fall from height incidents).
- Natural disaster / weather conditions (2.34%): Eruptions, avalanches etc.
- Extreme heat / burning / smoke / fumes (2.01%): Where victims were fatally injured by heat, burning, smoke and fumes inhalation.
- Asphyxiation (1.67%): Incidents of asphyxiation and strangulation, excluding drowning.
- Struck by vehicle (1.67%): When victim was struck or run over by a vehicle (e.g., bystander hit by work vehicle)
- Machinery incident - Crane (1.34%): Where machinery / equipment was involved in the incident but specifically involving cranes.
- Vehicle incident - ATV (non rollover) (1.34%): When victim was involved in an ATV incident where the vehicle did not rollover (e.g. struck fence without rolling over and crushing victim).
- Vehicle incident - other farm vehicle (1.34%): Other farm vehicle incident (e.g., Tractors, this excludes ATVs).
- Assault (1.00%): Incidents where the victim was assaulted (e.g., shot, bashed, etc.).
- Rail incident (1.00%): Incident involving train.
- Health problems (1.00%): For example heart attack at work.
It should be noted that these 'primary causes' were recorded as being directly related to the death of the victims at the time when the incident occurred. These are the events at the moment when the incident actually happened. For the purposes of incident prevention, it is more important to understand what the main reasons (or contributing factors) were that lead to the fatalities recorded. The present study has attempted to identify these contributing factors surrounding the workplace, the environment, the society and the people.
As shown in Figure 28, a list of high level 'contributing factors' for the work-related fatal incidents has been identified from the workplace fatality investigation reports. The top six factors (each of which contributed in some way to more than 8% of the fatalities investigated) were:
- Human error (>43.1%)[*]
- Procedural violation (>27.7%)*
- Poor/inadequate equipment/workplace design (22.1%)
- Poor safety culture (11.7%)
- Unsafe supervision 10.0%)
- Lack of personal protective equipment (PPE) (8.7%)
The research indicates that 'human error' and 'procedural violations' have contributed to nearly 70% of work-related fatalities over the past six years. It is critical that the term 'human error' is properly understood. Identifying 'human error' allows us to ask why a person's decisions and actions made sense to them at the time. It should be seen as a symptom of other things that are wrong deeper in the work system. 'Human error' is not simply identifying the mistakes people make.
In the report, 'human error' has been used to classify a variety of incidents where memory lapses, slips in behaviour or mistakes (e.g. misjudgements, misinterpretations, distraction errors, silly decisions, inadequate knowledge) have led to a fatality. These slips, lapses and mistakes are only one part of the cause of injuries. A memory lapse may have occurred because a person was asked to do a task they had not done for some time. A slip in behaviour could have occurred as a result of fatigue. A mistake, such as a misjudgement, may have been made because the person making a decision was given incorrect information to base a decision on. A procedural violation could have occurred because of pressure to complete a task, resulting in a risky shortcut being taken.
The vast majority of human errors and violations can be avoided, or at least minimised. Consequently this finding demonstrates an area where significant improvements can be made. The research points to areas such as workplace systems design, training and workplace culture, where improvements could make a huge difference to incident, injury and fatality rates.
A wide range of additional workplace factors have been investigated in the present study for their association with the occurrence of work-related fatalities. Independent evidence was collected in relation to these factors. For some factors, there appears to be strong independent evidence to support the findings, and for other factors, the evidence appears to be relatively weak. The following factors have been identified as having contributory effect on the occurrence of work-related fatal incidents with a strong or moderate independent evidence support:
Other workplace factors with strong independent evidence support:
- Hours worked per day
- Lack of recovery from fatigue
- Tight timescales/deadlines
- Time of day (10:30-12:30; 14:30-15:30); during summer: AM; autumn: PM)
- Day of the week (weekdays)
- Type of industry/workplace (especially 'Agriculture, Forestry & Fishing')
- Work involving vehicles
Other workplace factors with moderate independent evidence support:
- Level of staffing
- Level of casual labour
Strategies for preventing workplace fatalities should concentrate on these factors with strong or at least moderate support from independent evidence.
The present study has identified several environmental factors that are significantly associated with the occurrence of work-related fatal incidents. The literature research has also provided some supporting evidence for these factors, although the published studies are often in the area of transport related incidents, and the reports relating a wide range of industrial incidents to specific environmental factors as used in this study are still rare.
On the basis of the detailed factor analysis and the review of independent evidence, the following environmental factors have been identified as playing contributing roles in the occurrence of work-related fatal incidents:
- Rainfall and wet days
- High temperature
Rainy or wet days following a shower make the ground surface wet and slippery. This may create problems for ATVs or quadbikes especially when the ground underneath is still hard, due to reduced traction between the tyre and the ground (Moore, 2006). From a human factors point of view, the user expectation that those farm vehicles can work well in soft ground conditions may encourage people to ride the vehicles on wet but relatively hard slopes which do not provide enough traction for the vehicle to stay in balance. This situation becomes worse when the vehicle is loaded.
Other harsh environmental conditions such as strong wind may have contributed to the occurrences of some particular incidents in some industries in the past (e.g., forestry & logging; construction), but due to the limitation of the data available, this condition was not confirmed to be a general contributing factor associated with the workplace fatalities. However, this does not mean that such factors are not important for the industries concerned. On the contrary, when developing preventative strategies for a particular industry, known environmental hazards (e.g. wind, fog) must still be accounted for despite them not being formally recognised as contributory factors to workplace fatalities as a result of this desk top review.
In addition, it should be understood that a certain contributing factor to the occurrence of a workplace incident takes effect in combination with other factors. It is often the case that an incident happens at the 'wrong time' (environmental factor), in the 'wrong place' (workplace factor), and with the 'wrong person' (individual factor). Therefore, when developing preventative measures to tackle work-related fatalities, the environmental factors will need to be considered in conjunction with other workplace or sociological factors. For example, the risks of a vehicle rollover on a steep slope increase on a rainy/wet day if it is just before or just after lunch.
The likelihood of workplace incident increases in the summer when people work longer hours and therefore have limited time to recover from fatigue. This situation becomes worse under poor/harsh environmental conditions.
A number of sociological factors have been identified in the present study, with support from some independent evidence sources.
Sociological factors with relatively strong independent evidence support:
- Summer months/season
- Alcohol and drugs
Sociological factors with moderate independent evidence support:
- Tourists/temporary labour
- School holiday and public holiday
- Daylight saving
- Tobacco smoking
The inclusion of 'alcohol' as a contributing factor to the occurrence of workplace fatalities does not necessarily mean that the victims involved in the workplace fatalities investigated had higher blood alcohol levels. Rather this factor has been identified largely from the statistical point of view, on the basis of independent evidence, and under the consideration of the long-term effect of alcohol on human abilities. This factor will also need to be considered in conjunction with other factors (workplace/environmental/individual) when developing preventative strategies. For example, as alcohol can still be detected in the body and have an effect some 13 hours after drinking 6 pints of premium strength beer (12 units) (Morgan and Ritson, 2003), it is advisable not to drink more than a certain amount of alcohol a certain number of hours before carrying out work at height, or handling vehicles/machineries under harsh environmental conditions, even though the alcohol level in the body is within the legal limit for work. Several ISEs consulted through the course of this work expressed the opinion that marijuana usage could be a contributor factor in workplace fatalities. No evidence was available during the course of this study to support these opinions.
The following worker individual factors have been identified as major contributing factors to the occurrence of work-related fatal incidents particularly in summer, with strong independent evidence support:
- Age (55-64 years old for summer; 35-44 for autumn, and older workers 65+)
- Gender (male workers)
- Employment (employees, and self-employed)
- Return to work after a long holiday
One of the issues that have not been covered in the incident database is 'return to work after a long holiday'. This has been identified as a confounding factor with a combination of various factors such as fatigue (e.g., due to sporting activities during the holiday), jetlag and sleep loss, excessive alcohol consumption, and skill erosion. These issues have been documented in some published reports (some are give in Table 6), although systematic studies on the relation between holiday taking (or absence from work) and work-related incidents have been very rare. A relatively recent study found that train drivers are almost twice as likely to pass a red signal after taking a period of long holidays (7 days or more) (Li and Lock, 2003). While more scientific evidence may still be needed in support of this topic, it is advisable to consider this issue when developing preventative strategies to tackle work-related fatalities.
Fatal Injuries in relation to Serious Harms
Prevention strategies for workplace fatalities should not be guided entirely by this analysis of the Department of Labour fatality data set. As far as work-related incidents are concerned, fatal incidents are only the tip of the iceberg. Workplace fatalities occur only when a number of contributing factors co-exist simultaneously. Depending on the type of industry, research suggests that some 500-2000 smaller injuries take place for each fatality. This is illustrated in the incident pyramid below. A more detailed analysis of the work-related serious harm injury statistics along with the fatality data is recommended in order to identify some specific relationships between the two data sets for different industries. It is anticipated that such a study will result in a set of more practical solutions for the mitigation of work-related injuries.
[*] In an additional 16% of cases, an unsafe behaviour was identified that could have been either a human error or a procedural violation