Developing the conceptual model

It is argued in Chapter 2 that healthcare operates within a dynamic complex socio-technical system (Plsek and Greenhalgh, 2001, Plsek and Wilson, 2001, Braithwaite et al., 2009, Sweeney and Griffiths, 2002). As noted, the extant literature is weak in providing a conceptual model to consider how the characteristics and dynamics of a socio-technical system influence patient safety. It is therefore, necessary to synthesise the literature to develop a suitable conceptual model that takes account of the dynamic and systemic characteristics that influence patient safety.

5.1.1 Synthesising the literature

There are a number of important concepts relating to safety arising from the

characteristics of ‘systems’ that are found in the accident theory literature. However, the conceptual models used in healthcare situations, such as the Swiss Cheese model

(Reason, 1997), are limited as they are do not fully take account of the dynamic non-linear feedback that occurs in complex socio-technical systems (Roelen et al., 2010, Dekker, 2011).

Rasmussen (1997) argues that for us to consider safety at a higher conceptual level, it is necessary to take account of the complex sociotechnical nature of the problem space.

His SWE model is used by ‘resilience engineering’ theorists and applied to a limited extent to healthcare (Cook and Rasmussen, 2005, Miller and Xiao, 2007, Amalberti et al., 2006). For the purpose of this thesis, the SWE model provides a means to articulate two underlying themes. First, the model takes into account the complex and competing dynamics that occur in the performance of a system. Second, a resilient system can be conceptualised as one where the operating point is able to remain within the SWE at times of perturbation or continuous stress and is therefore more likely to keep patients

safe. The model therefore provides a ‘system resilience’ perspective to consider safe performance.

The reader is reminded that the boundaries of the envelope depict the constraints within which the system is expected to operate. The performance of the system in relation to the boundaries is depicted by the movement and location of the OP. The dynamics under which the system operates is depicted by the gradients exerting competing pressures on the OP. The SWE model can also incorporate a number of the concepts drawn from patient safety and accident theory.

At this stage there are four observations to make about the model. The first is that Rasmussen (1997), and subsequently Cook and Rasmussen (2005), identify three gradients that impact on the OP. These are the gradients of ‘management pressure towards efficiency’ ‘least effort’ and ‘campaigns for safety’ (see Figure 5.1). It is argued that these three gradients represent a range of variables and are therefore

thematic rather than specific. For example, the gradient ‘towards efficiency’ can include a number of specific issues such as costs, income, the use of equipment and speed of working. This means that what might be found in a context specific situation is a multitude of influences which may not always fit neatly into the three thematic gradients.

The second observation is that the envelope needs to sit within a wider context. Such a context has a bearing on the attention paid by the decision making agents to the

boundaries and movement of the OP (Williams and Smart, 2009). The context also contributes to the type and strength of the gradients. The context of healthcare is a complex mix of social, political, financial and regulatory factors which result in

conflicting and fuzzy policy (Ham, 2009, Klein, 1995). There is the potential for policy decisions to create latent conditions which have unforeseen non-linear interactions, delayed feedback with potential safety consequences. It is argued in Section 5.2 that the context influences the performance and decision making.

The third observation is that the gradients can be a mix of downward and upward

influences (‘scale interactions’), considered in Chapter 3, which are part of the complex dynamic (Woods, 2006). Where there is downward influence, the system context

influences the OP. Where there is upward influence it originates from the actions of the

staff where, in the face of the downward influences, they make trade-off decisions, find work around solutions or seek local optimisation. Scale interaction between decision making agents and the wider context can be incorporated by taking into account the link between SD and social theory. As noted in Chapter 4, the assumption made in this research is that the underlying social theory of SD is the contemporary dialectic relationship between the social structure and agency (Lane, 2001b).

The fourth and related observation is that the movement of the OP is due to multiple and dynamic factors, not all of which can be understood or predicted (Stacey and Griffin, 2005). Currently, the model helps us to conceive of compensating actions to maintain the OP within the envelope in the face of competing influences. However, the model does not fully include the dynamics of the stocks, flows and feedback loops within the envelope. This means that the Rasmussen (1997) model does not address how the flow of work through the system creates a series of interactions with decision makers who are seeking to balance a number of competing priorities in a complex environment.

To overcome some of the weaknesses observed, the SD approach is used to extend the SWE model to take account of the dynamics of stocks, flows, feedback, delays and the interaction with decision making agents. SD modelling can be used to illustrate the interrelationships between the parts of a system and the potential consequences of a change in those relationships. SD is used to incorporate the concepts of ‘coupling’ and

‘feedback’ between the parts of the system, which in turn influence the behaviour of the whole system. When a system is disrupted or under continuous stress the dynamics generated by the relationships may change with potential safety implications (Cook and Rasmussen, 2005, Perrow, 1984). By depicting the potential change in the relationship of the parts through SD diagrams, further insights can be developed into the movement of the OP in relationship to the boundaries.

Figure 5.1: SWE (v2) model developed from synthesis of the literature