علیت آسیب درهوای آزاد بزرگ: تجزیه و تحلیل سیستم های منجر شده حوادث آسیب فعالیت بیرون از منزل

Despite calls for a systems approach to assessing and preventing injurious incidents within the led outdoor activity domain, applications of systems analysis frameworks to the analysis of incident data have been sparse. This article presents an analysis of 1014 led outdoor activity injury and near miss incidents whereby a systems-based risk management framework was used to classify the contributing factors involved across six levels of the led outdoor activity ‘system’. The analysis identified causal factors across all levels of the led outdoor activity system, demonstrating the framework's utility for accident analysis efforts in the led outdoor activity injury domain. In addition, issues associated with the current data collection framework that potentially limited the identification of contributing factors outside of the individuals, equipment, and environment involved were identified. In closing, the requirement for new and improved data systems to be underpinned by the systems philosophy and new models of led outdoor activity accident causation is discussed.

There is an acknowledged risk of both severe and frequent injury associated with active pursuits, especially those participated in for sport, active recreation or leisure (e.g. Finch et al., 2007, Flores et al., 2008 and Gabbe et al., 2005). One important educational form of active recreation is led outdoor activities, which are defined as facilitated or instructed activities within outdoor education and recreation settings that have a learning goal associated with them (Salmon et al., 2010). Examples include activities such as school and scout camping, hiking, harness sports, marine aquatic sports and wheel sports. Although the organised, structured, and led nature of such activities provides a degree of risk management, these very same features engender a level of complexity that inevitably leads to injury causing incidents in which multiple contributory factors play a role. This is exemplified by the findings from an exhaustive investigation into the recent Mangatepopo gorge walking incident which identified multiple contributory factors related to various different actors, equipment, processes, and organisations (Brookes et al., 2009). The incidence of such ‘systems’ accidents and incidents during led outdoor activities necessitates the adoption of a systems approach when attempting to understand and prevent led outdoor activity accidents and injury-causing incidents. The systems approach to accident causation and analysis has a long history of applications in safety science efforts (Davis et al., 2013 and Waterson, 2009). Although the notion that multiple causal factors from across work systems play a role in accidents is not a new one (e.g. Heinrich, 1931), the systems approach has yet to become the dominant approach to understanding and preventing accidents in some areas (Davis et al., 2013). This is the case in the led outdoor activity domain, where the approach typically has not been underpinned by contemporary systems theory-based models of accident causation the like of which are widely used in other safety critical domains such as aviation, process control, and mining (Salmon et al., 2010). Moreover, although recent incident investigations such as Brookes et al. (2009) suggest that catastrophic led outdoor activity incidents are the result of multiple failures across the overall led outdoor activity system, there is little data or analyses available to demonstrate that this is the case for the everyday injury incidents that occur (e.g. falls, sprains and strains). This has some significant implications. First, it inhibits communication and uptake of the systems approach philosophy since information on the role of systems factors in incident causation is not forthcoming. Second, it hinders the development and implementation of data collection systems and analysis methods underpinned by systems thinking, since there is no evidence clarifying the role of systems factors in injury incidents. Third and finally, it limits the utility of interventions designed to prevent future incidents since they are based on a limited understanding of incident causation. As part of a wider program of research aiming to develop and implement a systems-based accident and incident reporting and learning system in the Australian led outdoor activity sector, the aim of the study described in this article was to utilise an in-depth led outdoor activity incident database in order to test a systems approach-based framework for its ability to classify the system wide contributory factors involved in led outdoor activity incidents. This incorporated the aim of investigating the nature of the contributory factors involved in led outdoor activity injury-causing incidents. The study involved the use of a Rasmussen's risk management framework and associated accident analysis method, Accimap (Rasmussen, 1997), to classify the contributory factors reported for 1014 led outdoor activity incidents in the New Zealand Outdoor Education/Recreation National Incident Database. This paper makes a contribution to the accident literature, since it presents a systems analysis of the causal factors involved in led outdoor activity injury-causing incidents. Such analyses have not yet been produced for multiple accident cases in the led outdoor activity context. Moreover, a practical contribution is made through the examination of existing data systems for their ability to support systems analyses of led outdoor activity injury incidents and in the specification of potential interventions designed to prevent future incidents. 1.1. The systems approach The systems approach to accident causation and analysis is a long and established philosophy that first emerged in part in the early 1900s (e.g. Heinrich, 1931) and has since evolved through a number of accident causation models and analysis methods (e.g. Leveson, 2004, Perrow, 1984, Rasmussen, 1997 and Reason, 1990). The systems approach centres on the notion that safety, and indeed accidents, are emergent properties arising from non-linear interactions between multiple components across complex sociotechnical systems (e.g. Leveson, 2004). Based on a review and comparison of models, Salmon et al. (2010) concluded that Rasmussen's (1997) risk management framework and accompanying Accimap accident analysis framework are suited for application in the led outdoor activity context. Rasmussen's risk management framework (see Fig. 1) argues that work systems comprise various levels (e.g. government, regulators, company, company management, staff, and work), each of which are co-responsible for production and safety. With regard to accident causation, the framework argues that decisions and actions at all levels of the system interact with one another to shape system performance: safety and accidents are thus shaped by the decisions of all actors, not just the front line workers in isolation, and accidents are caused by multiple contributing factors, not just one bad decision or action.To support use of the framework in accident analysis studies, Rasmussen (1997) outlined the Accimap framework for analysing accidents. Accimap is used to graphically depict the decisions, actions, and conditions that interacted with each another to produce the accident in question. Accimap typically structures these contributing factors across six organisational levels: government policy and budgeting; regulatory bodies and associations; local area government planning & budgeting; technical and operational management; physical processes and actor activities; and equipment and surroundings. Factors at each of the levels are identified and linked between and across levels based on cause–effect relations. Based on a comparison of three popular accident analysis methods, Salmon et al. (2012) concluded that the Accimap framework was the most suitable for analysing multiple led outdoor activity injury and near miss incidents. In testing Rasmussen's risk management framework in the led outdoor activity context, the authors wished to examine whether the framework could be used to classify the data in accordance with the key tenets of the systems approach. This involved testing whether the framework could be used to classify the contributory factors involved and identify where in the led outdoor activity system they reside (i.e. place the contributory factors across the different levels of the led outdoor activity system). 1.2. The systems approach in the great outdoors To enable the Accimap framework to be used in the analysis of led outdoor activity incidents, the six systems levels typically used in Accimap analyses were adapted to reflect the led outdoor activity domain. This led to the definition of the following six led outdoor activity system levels: 1. Equipment and surroundings: factors associated with the equipment used in support of the activity, the physical environment in which the activity was undertaken, and the ambient and meteorological conditions prior to or during the incident; 2. Physical processes and instructor/participant: activities undertaken ‘at the sharp end’ prior to, and during, the incident. It therefore describes the flow of events leading up to and during the incident in question. This includes decisions and actions made by instructors, participants, etc., but may also include decisions and actions made by other actors, such as supervisors, emergency responders, members of the public, etc.; 3. Technical and operational management: activities, decisions, actions, etc. made by personnel at the supervisory and managerial levels of the organisation providing the activity involved in the incident. These factors typically occur prior to the incident itself but can also include decisions and actions made during, or in response to, the incident. 4. Local area government, activity centre management planning and budgeting: activities, decisions, actions, etc. made by personnel working in local government and at the senior managerial levels of the activity centre involved (e.g. executive board level). These factors are related to higher level management, planning and budgeting activities and typically occur before the incident itself (this can even be years preceding the incident); 5. Regulatory bodies and associations, schools and parents: activities, decisions, actions, etc. made by personnel working for led outdoor activity regulatory bodies or associations and the schools involved. Decisions and actions made by the parents of the participants involved in the activity are also included at this level. 6. Government policy and budgeting: decisions, actions, etc. relating to the provision of led outdoor activities, such as those relating to funding and policy development. 1.3. New Zealand National Incident Database study The study involved the analysis of a series of led outdoor activity injury and near miss incidents using Rasmussen's risk management framework. Accordingly, factors identified from the incident data were classified, using the modified Accimap levels described above, in order to identify contributory factors from across the led outdoor activity system. The data were taken from the New Zealand Outdoor education/Recreation National Incident Database (OER NID). A description of the OER NID can be found at http://www.incidentreport.org.nz/. The database is used by commercial, educational, not for profit, or informal groups and individuals involved in outdoor activities. Full approval for the study was given by the New Zealand Mountain Safety Council who operate the NID on behalf of the Outdoor Sector. Although the authors of this article reside in Australia, there are very few Australian databases that collect exhaustive data on led outdoor activity injury incidents. The OER NID was the only readily available database that contained sufficient data to support the present analysis and thus provided the most appropriate data set for the study.

This study has classified the factors reported to be involved in a series of led outdoor activity incidents using Rasmussen's Accimap framework. The aim was to test the framework in the led outdoor activity context, and identify the factors involved in led outdoor activity incidents. First and foremost, the analysis led to the identification of contributing factors across all levels of the led outdoor activity system and there were no factors that could not be classified according to the framework. This confirms the requirement for a systems approach to accident analysis and prevention in this domain and demonstrates that Rasmussen's framework is suitable for accident analysis efforts. Although the identification of factors across the led outdoor activity system is not surprising given long standing accident causation theory, the analysis represents the first systems analysis of a led outdoor activity incident data set containing multiple injury causing incidents. Although systems analyses have been undertaken in response to catastrophic incidents involving multiple fatalities, systems analyses of multiple incident case databases have not been undertaken. The analysis confirmed that the framework was able to classify contributory factors across different actors (e.g. instructors, supervisors, participants) and the different levels of the led outdoor activity system (e.g. activity centre, local area government, government). Further the analysis demonstrated that causal factors from the six Accimap levels were reported to be present in the incidents analysed. Rather than being primarily the result of participant or instructor actions, the incident data suggests that the chain of events preceding led outdoor activity injury incidents can involve the decisions and actions of many actors from across the overall system. Importantly, these actors can reside outside of the activity providing organisations involved, for example government workers, parents and teachers. In addition, Rasmussen's argument that accidents are typically caused by multiple contributing factors across the different levels of the system, not just a single catastrophic decision or action, was confirmed. In the present study, the mean number of contributing factors across the incidents was 4.1 (SD = 2.33), suggesting that led outdoor activity incidents are caused by multiple factors rather than one decision or action in isolation. It is concluded that Rasmussen's risk management framework is an appropriate one for implementing the systems approach to accident analysis and prevention in the led outdoor activity domain. Although factors across all levels of the framework were identified, it is acknowledged that there were only a small number of factors identified at the higher levels (Government policy and budgeting, Regulatory bodies and associations, Local area government, activity centre management planning and budgeting, schools and parent, Technical and operational management) when compared to the lower levels (physical processes and instructor/participant activities and equipment and surroundings). Two explanations may be offered for this result. First, it may be that higher level factors do not commonly contribute to incident causation during led outdoor activities; however, the contemporary accident causation literature would suggest that this is unlikely. Second, the lack of higher level factors identified may be due to the structure of the database and incident reporting system, whereby causal factors outside of the instructor, participant, environment, and equipment can only be recorded in the incident description and causal narrative fields. That is, incident reporters are provided with multiple choice options for causal factors related to the instructor, participant, environment, and equipment, but not for causal factors outside of these four fields. Without training in the systems approach to accident causation, it is unlikely that reporters would seek or identify causal factors outside of the instructor, participant, environment, and equipment without some guidance. Moreover, it is notable that the systems approach to accident causation is not yet well entrenched within the led outdoor activity domain (Salmon et al., 2010). Compounding the issue, reporters sometimes did not fill in the causal narrative section, or would report that the incident was simply ‘an accident’ or ‘bad luck’. Notably, reporters would often select causal factors related to the participants and instructors from the multiple choice options, without providing an explanation for their selections. This indicates that there may be bias in reporting the causes of incidents in the NZ led outdoor activity sector which could lead to incorrect conclusions. A challenge moving forward is therefore the development of data collection systems that are capable of gathering data on contributory factors at the higher levels of the led outdoor activity system. Such systems need to be sensitive in that no additional workload should be imposed for simplistic incidents in which there are few contributory factors (e.g. a bee sting), however, they need also to support incident reporters and/or safety managers in collecting data on contributory factors across the system as and when the incident demands it. Whilst the design of the physical reporting system itself in terms of reporting forms and databases is important, other elements are also important such as training materials (emphasising the systems approach to accident causation) and the level of resources devoted to incident data collection and analysis (e.g. time support for reporters). It useful to compare the present analysis with existing perspectives on led outdoor activity incidents. Various led outdoor accident causation models exist (e.g. Brackenreg, 1999 and Davidson, 2007). For example, Davidson (2007) identifies three categories of contributory factors: unsafe conditions (environment), unsafe acts (on behalf of participants) and instructor judgement errors. These three categories were the most prominent contributory factors in the present study findings; however factors across the overall led outdoor activity system were identified. Whilst the present analysis confirms the presence of all three factors across the incidents analysed, it goes beyond this tripartite perspective to identify contributory factors outside of the instructor, participants, and environment. The requirement for new models of led outdoor activity injury incident causation, incorporating other contributory factors from across the led outdoor activity system, is thus supported. Although environmental, instructor and participant factors are prominent and highly visible contributory factors in led outdoor activity incidents, it is likely that other conditions across the system shape these factors or their impact (e.g. risk management systems, training programs, financial constraints). The challenge for future studies in this area is to clarify how system wide factors influence participant, instructor and environmental factors in a manner that leads to injury causing incidents. Although a test case study, the analysis does have implications for preventing injury causing incidents in led outdoor activities. Identification of systems failures is especially useful, since it supports treatment of failures across the system that influence the way in which the led outdoor activity system operates. Many researchers have argued that treatment of wider systems failures, identified through systems-based analyses, is more appropriate than the treatment of local factors at the sharp end of system operation, since the factors creating the front line behaviours are removed following analysis efforts (e.g. Reason, 1990, Dekker, 2002, Leveson, 2004 and Rasmussen, 1997). For example, identifying bad decisions or lack of skills on behalf of instructors is useful; however, without a systems analysis, the higher level factors that shaped how the instructor was trained or how they made their decisions is not well understood. Retraining, using the same training system, is not likely to be effective in improving instructors’ decision making or skills since behaviour shaping conditions across the wider system are still present. The analysis identified a number of higher level systems failures that played a contributory role in the injury incidents. These included a failure of Governmental organisations to undertake key safety related tasks (e.g. national park trail repair), various instances of parents and schools failing to provide activity organisers with critical information (e.g. pre-existing injuries, behavioural issues, medication requirements), inadequate risk management and training evaluation systems, inadequate activity policy, poor planning of activities and high participant to staff ratios. Injury incident countermeasures should thus focus on these factors rather than specifically on improving instructor and participant behaviour. For example, for the problem of parents and schools not providing critical information to activity centres, it may be pertinent to improve participant consent and information forms, focus on enforcement of parents and schools supply of such information, better communicate to parents and schools the importance of providing critical information to activity centres, and also communicate the risks associated with injured participants or participants with behavioural problems engaging in led outdoor activities. Moreover, in the event that they identify problem participants (e.g. injured or ill-behaved), led outdoor activity instructors should be given the power to abort activities or prevent participants from taking part. Another example would be to tailor risk management systems to focus on dealing with participant unsafe acts, instructor judgement errors, and environmental risks. These represent the most prominent causal factors in injury incidents therefore risk management systems should be tailored to deal with them. One example of this would be to develop more appropriate risk management strategies surrounding the presence of unsafe terrain or adverse weather during activities. Finally, the problem of high participant to staff ratios could be solved by identifying the factors that create this issue (e.g. financial constraints, lack of staff, no restrictions on number of activities provided for differing staff levels) and developing strategies to deal with them. The lower level factors identified are also useful to inform systems reform efforts. For example, the present analysis indicates that hazardous terrain was a prominent contributing factor across the incidents analysed. When combined with the identification of poor planning and inadequate risk management systems factors at the higher levels, and also the fact that 67.2% of cases had both a physical processes level and an environment and surroundings level factor, this suggests that the effects of environmental hazards such as terrain should be considered more explicitly in planning and risk management systems and should also form a key part of instructor training programs. Moreover, participants should be given explicit instructions on strategies for coping with hazardous terrain and activity programs should be designed with hazardous terrain management in mind. One notable limitation of this study surrounds the accuracy and validity of the incident reports analysed. It is worth pointing out that the incident reports were compiled by a range of different reporters and that they represent what causal factors each reporter believed were involved in the incident. Prior to the present study the incident reports had not been subject to any further analysis or validation and so may be vulnerable to under/over reporting or erroneous reporting of causal factors. Caution from the readers is therefore urged when considering the analysis findings. As it is not possible to verify the accuracy of the incident data reported, it is acknowledged that the analysis presented provides a description of what causal factors were reported rather than a verified accurate picture of incident causation in the activities analysed. That said, this is a limitation that most aggregate accident analysis studies suffer from. In conclusion, this analysis provides evidence that, in line with accident causation theory, led outdoor activity injury incidents do have causal factors from across the overall led outdoor activity system. Moreover, it is concluded that Rasmussen's framework and Accimap approach are appropriate for actualising the systems approach in future led outdoor activity accident and incident analysis and prevention endeavours. It is argued, therefore, that future efforts undertaken to understand, analyse, and prevent led outdoor activity injury incidents should be underpinned by the systems philosophy on accident causation. There are a number of pressing requirements to fulfil in order to actualise this paradigm shift. First, injury incident and accident data collection systems underpinned by the systems philosophy are required. Although the database used in the present study was useful, it was notable that causal factors were only coded in relation to participants, instructors, equipment, and the environment. Data on other causal factors was collected through the incident narrative, however, the database does not support coding of factors outside of participants, instructors, equipment, and the environment. Without data systems (and associated coding taxonomies) underpinned by the systems philosophy on accident causation, data on the system-wide causes of injury incidents will not be collected. A key challenge associated with developing new data systems is ensuring that data on causal factors at the higher system levels (e.g. government, local government, company management) can be accurately obtained. This will require training for instructors and activity centre staff on the systems approach, to ensure that they understand how to report and investigate incidents. Second, injury incident and accident analysis methods underpinned by the systems philosophy should be developed specifically for the led outdoor activity domain. Without such methods, practitioners will have no means to identify the system-wide causal factors involved in led outdoor activity injury incidents and accidents.