• No results found

CHAPTER 1. INTRODUCTION

1.4 Complexity of hospital medication systems and processes associated with medication

1.4.2 Understanding the problems of complexity

Much of the understanding of system complexity and the susceptibility of health care systems to patient incidents in national reports (Kohn et al., 1999; Department of Health, 2000a) have been based on human error theory by Reason (1990), which in turn was partly built on the work by Perrow (1984), from his analysis of major accidents in high risk industries. This section provides an overview of the work by Perrow (1984) on systems contributions to accidents and highlights the theoretical applications for understanding systems effects on MAEs.

Perrow (1984) suggested that the susceptibility of a system to ‘accidents’ is attributable to two concepts: interactiveness, and coupling. Interactiveness relates to the notions of linear and complex interactions within the system. Linear interactions are the predominant interactions within a system and are characterised by predictable, visible, and one-to-one relationships between components, for example, A is always follwed by B which is followed by C, therefore if B is not working, then C will not function, and exploration into A or other upstream components is expected to reveal the cause. By contrast, complex interactions are characterised by the opposite; the interactions involve sequences that are unfamiliar, unplanned, or unexpected, and are either not visible or are not immediately comprehensible. As highlighted earlier, hospital medication systems and the medication administration process are dependent on a large number of components. It can be seen from figure 1.3 that many of the components either are or have the potential to be involved in a complex interaction.

For example, checking the drug chart may involve first searching for the drug chart;

availability of the drug chart is not always predictable and the subsequent actions taken to locate the drug chart is not always consistent. This may be further complicated by

interruptions experienced during the task and subsequent actions are likely to be situation dependent and therefore unpredictable.

Coupling is a mechanical term with origins in engineering and was used by Perrow (1984) to describe the flexibility of systems in response to unpredictable changes and failures. In tightly coupled systems, there are more time-dependent processes; A is immediately followed by B and there is little waiting time in-between. The sequence is invariable (B must follow A) and the overall process is designed to reach the goal in one way, with little slack or buffer in how resources are used. Consequently, buffers and redundancies need to be designed into tightly coupled systems to support the recovery process if/when a component fails (regardless of whether the interactions are complex or linear). By contrast, loosely coupled systems can accommodate time delays, the sequence can be reordered, there are multiple methods to achieve the same goal and resources may be ‘wasted’ without impacting greatly on the goal. The arrangement of components in a loosely coupled system may also facilitate recovery from failure. In general, health care is considered to be a loosely coupled system (Pinelle & Gutwin, 2006), and this probably includes hospital medication systems due to the dynamic nature of the processes involved. For example, in the drug distribution system, medications can be ordered within a range of times from the pharmacy, often via one of a selection of methods. Furthermore, medications may be supplied to the ward at a range of times, and then put away in the relevant ward-based storage facilities at a time that is convenient for the nurse, rather than interrupt them. If a new medication is prescribed and is not available on the ward, the system is generally sufficiently flexible to enable nurses to obtain the drug via an alternative method to avoid a dose omission.

Conversely, due to the ‘looseness’ of the hospital medication system, a newly prescribed

dose may not be identified and/or supplied within the required period of time, thus potentially contributing to a dose omission error.

As a potential problem, the notion of complex interactions suggest that the solution would be to simplify and make interactions more linear. However, Perrow (1984) highlights that this is not the case; in practice, complex systems can be more efficient than linear systems (for example, due to multi-functional components). Furthermore, it is not always possible to reduce complexity and produce the same ‘output’. As with complex interactions, tight coupling in systems is sometimes necessary and not always seen as a problem. However, in general, systems that have tightly coupled, complex interactions are more prone to accidents than loosely coupled linear interactions as there are less opportunities and time to recover from component failure. Overall, the type of interactions and coupling within the hospital medication system and associated medication administration process are situation dependent. This creates a challenge for developing systems-based interventions because potential latent error-producing conditions may not be recognised at the time of investigation (Kohn et al., 1999).

Furthermore, as most health care processes were not designed but have evolved (Vincent, 2011), it is suspected that system and process variations exist across the NHS.

This presents a barrier for identifying, prioritising, and developing system-based interventions to reduce error. The concepts of variation are next discussed.

1.5 Variations in hospital medication systems and