Hours of work and rest in the rail industry: to prescribe or not to prescribe, that is the question

In a recent publication in this Journal, Anderson and colleagues1 have suggested that the National Transport Commission (NTC) should introduce limits to working time arrangements as a specified element of the regulatory framework for managing fatigue-related risk. In their view, a performance-based approach (whereby the regulator requires a rail operator to develop a comprehensive fatigue risk management plan but does not provide prescriptive guidance on what is or is not an acceptable working time arrangement) runs the risk of permitting operators to introduce unsafe work practices. To support this argument, they refer to research linking shift length, cumulative working hours and night work with increased risk of accident and injury. In essence, they are concerned that a lack of clearly defined prescriptive limits will lead to extended working hours, increased likelihood of fatigue, fatigue-related error and reduced safety outcomes.

This debate between performance-based approaches to safety management and explicit regulatory prescription has a long and controversial history especially in the area of fatigue management. The current regulatory framework adopted by the NTC on behalf of the Rail Safety Regulator2 does not specify limits to working time arrangements within relevant legislation or regulation. The reasons for this are quite straightforward. It has proved difficult, if not impossible, to develop a prescriptive rule set that efficiently balances safety and operational effectiveness at the organisational level and covers the diversity of operational settings found in the Australian rail industry.

Simple prescriptive models (e.g. Schedule 2 of the Rail Safety Act, NSW)3 that have been used in a particular state or sector of the industry do not, themselves, have a definitive evidence base to support their effectiveness and often (not unreasonably) introduce limits based on other factors (e.g. industrial and/or psychosocial criteria) rather than safety outcomes alone. Moreover, they are often not as prescriptive as would initially appear to be the case. For example, recent publications4 by the NSW rail safety regulator (Independent Transport Safety and Reliability Regulator) permit operators to present a safety case to justify operating outside the prescriptive limits outlined in Schedule 2. Equally, traditional prescriptive limits on working times have often proved unwieldy or inefficient when applied in operational settings not envisaged in their development (e.g. automatic train and signal control systems, single-driver operations, relay running,1 or recent developments in highly automated trains and/or remote train operations2).

The NTC panel tasked with developing the new regulatory framework looked at these issues very carefully between 2010 and 2012, and, with the exception of the Rail Train and Bus Union3 representative, concluded that it would be extremely difficult to develop a simple cost-effective set of prescriptive limits that covered the diversity of rail operations in Australia. As is the case with the UK and US national rail safety regulators,5, 6 the decision was made to respect the situational complexity of managing fatigue-related risk and the diversity of the operational circumstances across the industry. Under the new approach, (i) individual rail operators would be required to develop an organisation-specific fatigue management plan, and (ii) the industry and regulator would be tasked with developing guidance materials to inform the development and likely approval of that plan. In order for a fatigue management plan to be approved by the national rail safety regulator, rail operators, in conjunction with relevant unions, would be required to risk-assess all working time arrangements at the organisational level and demonstrate that they had effectively identified and mitigated unacceptable risk. In the panel's view, this approach was more flexible, better targeted and would enable each organisation to develop a tailored fatigue management plan. Individualised plans at the organisational or work group level could address their specific operational requirements and take into account the likely impact of very specialised work practices and risk mitigation technologies that may or may not be in place (e.g. single-driver operations, automatic train control, vigilance control devices, etc).

In the paper by Anderson and colleagues,1 there are clear and cogent arguments as to why working time arrangements require risk assessment and mitigation. The Panel's recommendation to not prescribe working time limits at the legislative or regulatory level did not reflect any debate over the risk posed by fatigue or the well-established link between working time arrangements, the likelihood of fatigue or the consequence of fatigue-related errors. Neither does it reflect any difference in opinion as to the benefits of sleep disorders screening – in fact this is required by current and proposed regulations. The key issue is whether it is appropriate for the rail safety regulator to define a single national set of prescriptive working time limits (e.g. shift duration or successive shifts) or whether limits should be something more appropriately developed and approved at the organisational level.

The recommendation was based on several arguments. First, at a theoretical level, it is very difficult to provide clear definitive evidence as to what should be the upper limit for any given dimension of the working time relationship. As Anderson and colleagues point out, there is conflicting evidence on whether 12-h shifts are safer than 8-h shifts or whether we should limit shifts to 2, 4 or 6 in a row before a break period is provided. In a recent review, Ferguson and Dawson7 suggested that it is difficult to interpret data on a single dimension because they are rarely orthogonal and do not typically interact additively. In addition, fatigue-related risk (as distinct from fatigue per se) is not well predicted by shift duration or number of successive shifts not because they do not play a role but rather because they interact in complex non-linear ways. For example, a long sequence of 12-h day shifts will typically produce less sleep loss than a similar sequence of 12-h night shifts. Similarly, the consequence of a fatigue-related error can vary from the trivial to the catastrophic depending on the nature of the task (e.g. a single-driver operating a train without automatic train protection vs operating a photocopy machine). It would be reasonable to suggest that while we do understand the broad relationships between working time, fatigue and risk, there are rarely scientific data of sufficient quality or relevance to derive clear limits in the way that has been suggested. In response to this uncertainty, the committee suggested that the most appropriate way forward was to indicate the typical range of limits across different industries and jurisdictions. However, if an organisation wants to exceed a specific threshold limit, the burden of proof would be reversed, and the rail operator would be required to present an evidence-based safety case assessing the likely level of increased risk and demonstrating how the risk will be mitigated and monitored over time.

At a regulatory level, this approach is entirely consistent with that developing in other rail jurisdictions and in other industries including healthcare8 and aviation.9 The UK and US rail safety regulators have introduced a non-prescriptive requirement for a Fatigue Management Plan for all operators.5, 6 In the UK, there are no regulated or legislated limits on the working time arrangement. In the US, the current status is that individual operators are required to develop an individual Fatigue Management Plan that can include prescriptive limits. In the US, that some elements of the current hours of service (HoS) requirements will remain in the short term – although it is important to note that HoS limits in the US rail industry are considerably less restrictive than the Australian equivalents or anything that has been approved under non-prescriptive approaches.

Similarly, the Australian National Heavy Vehicle Regulator has recently proposed a fatigue-risk management regulatory system10 that provides guidance materials that suggest indicative limits to working time arrangements but permits operators to propose an organisational, driver or route-specific safety case using a risk classification system (Advance Fatigue Management) that permits operators to exceed designated limits (for Standard Hours and Basic Fatigue Management) provided the operator can demonstrate (to the regulator) that the additional risk has been appropriately assessed and mitigated.

At a practical level, there are two key issues in this debate. First, if prescriptive limits are embedded at the legislative or regulatory level, this makes it very difficult to respond appropriately to new information on safety outcomes, new developments in risk mitigation technologies or innovative work practices. Regulatory or legislative reform is always extremely slow. By embedding indicative limits within guidance materials, the regulatory landscape can be modified quickly in response to new or emerging information, technology or work practices. Second, there is a logical dilemma. Until we try something new, it is difficult to provide evidence to support the efficacy of a novel approach.

In my view, many of the current limits on working time arrangements are far from evidence based. Furthermore, we do not have sufficient scientific evidence to provide definitive evidence on the risk associated with duration or number of successive shifts given the diversity of operational circumstances. In the future, this complexity will only increase. Anderson and colleagues are right to be wary of the risks associated with a change in regulatory approach. A performance-based approach that fails to provide a rigorous approval process or requires comprehensive post-implementation surveillance datadoes carry the potential for de facto deregulation. On the other hand, a rigorous performance-based approach carries significant potential to simultaneously enhance operational effectiveness and safety outcomes. It is not unreasonable to suggest that in view of the uncertainty in assessing and controlling fatigue-related risk using work-rest limits alone, some degree of innovation, experimentation and evaluation could well be justified and, if well managed, could lead to significant improvements in the way we manage this hazard in the future. Is this not the historical basis of all progress?