Patient Safety as the Management of Risk Over Time
We have now arrived at a rather different view of patient safety which includes, but does not conflict with, definitions focused on the reduction of error and harm. The revised aim of patient safety is to maximise the overall balance of benefit and harm to the patient, rather than specifically to reduce errors and incidents. Patient safety becomes the management of risk over time as the patient and family move through the healthcare system. The benefit may be expressed as recovery whenever possible, reduction of suffering or extended survival. This is of course the aim of clinicians everywhere when treating individual patients but we are concerned with how this might be achieved across a system.
The reduction of harm remains important, as does the reduction in errors and incidents, but it is not the dominant perspective. Incidents associated with care will always occur during episodes of care since no human activity can be error free, especially across a system with open access 24 h a day and 7 days a week. Harm may occur because of single safety incidents but more commonly from an accumulation of poor care that impedes recovery, worsens the prognosis or prolongs disability unnecessarily. Patient safety is both the art of minimizing these incidents and managing risk over longer time periods which will require additional skills and methods. We accept in this vision that errors will inevitably occur but that, in a safe system, very few will have any consequences for the patient. This is in essence a clinical vision but at the level of the system as well as the individual patient. Note that this view gives considerable emphasis to the achievements of patients, families and staff in monitoring, negotiating, adapting and recovering from the inevitable hazards and failures along the patient journey.
Adopting a Range of Safety Models
Safety needs to be approached very differently in different environments. We have initially distinguished three classes of safety models that fit different field demands: the adaptive model embracing risks, the high reliability model managing risks, and the ultra-safe model in which risk is controlled or avoided wherever possible. These different responses to risk give rise to different models of safety, each with their own advantages and limitations. The differences between these models lie in the trade-off between the benefits of adaptability and the benefits of standardisation and control.
Healthcare has many different types of activity and clinical settings and so we cannot use one primary model (Box 11.3). We can see parallels and applications of the three models relatively easily in the hospital environment. Radiotherapy, blood products, imaging systems and the management of drugs in pharmacy are all highly regulated, very reliable and operate to industrial standards of precision. Many of these systems rely on a high degree of automation and decision support and the professionals working in these areas are accustomed to working in a highly ordered manner. In other settings, such as obstetrics and elective surgery, risk has to be accepted and managed with coordinated teamwork. High risk surgery, trauma medicine and the treatment of rare and dangerous infections require a more adaptive approach though all benefit from a foundation of standard procedures. We should also bear in mind that much adaptation and resilience in healthcare is unnecessary in that it is employed not from clinical necessity but to compensate for wider system deficiencies (Wears and Vincent 2013).
Box 11.3. Safety Models for Healthcare
• There is no one universal model of safety in healthcare that can apply across every setting. Each model has its own advantages, limitations and challenges for improvement.
• The choice of a safety model will derive from professional consensus, from real world experience, an understanding of safety and judgements as to what is politically feasible in the context in question.
• Imposition of a given safety model that is inappropriate to the context in question may not be effective and may sometimes even degrade safety.
• Each model has similar potential to improve safety in healthcare by a factor of 10, although the maximum attainable safety figures are context dependent and can vary considerably from one model to another.
In healthcare we may find we need a wider array of models than the three we have outlined. It would be a mistake to assume that these three broad approaches are all we need; they are a helpful simplification of a more complex problem. For instance care in the community is unusual in being highly distributed amongst different people and organisations and also only partially reliant on strict standards. Many industries would manage a very distributed system by careful standardisation of core procedures but this may not be possible when, for instance, managing the care of people with severe mental health problems in the community. We are also aware that the industries we have chosen to illustrate the differing approaches to safety are high hazard, high technology and, while those who work in them support each other, they are not simultaneously concerned with delivering compassionate care to vulnerable people. We will probably need a more thoughtful approach to the systemic management of risk in the care of people with learning disabilities for instance, which will retain the broader strategic understanding but achieve the objective of managing risk through personal relationships as much as through formal strategy.
We will also need to consider how we can move between models. When, for instance, does a previously adaptive approach become sufficiently embedded and understood to begin the transition to a high reliability approach? In part this comes about from innovation, familiarisation and the building of expertise within a community. Innovative surgery for instance always begins in a context of risk and challenge. As experience grows in, for instance the management of aortic aneurysm, the surgery still carries risks but these are known, understood and managed rather than endured.
A patient's journey crosses many medical settings and services, in different contexts, and therefore is necessarily exposed to the whole range of safety models. Controlling risk not only requires managing each setting and the transitions between settings, but also being alert to the fact that safety interventions that are effective in one setting may adversely affect safety in other contexts. For instance a cautious and restrictive control of laboratory services aimed at reducing error that is effective in raising standards locally, might adversely affect safety more widely through the reduction in the availability of timely laboratory results.
The external environment is also a critical determinant of which approach to safety can be adopted. An ultra-safe system relies not only on internal procedures, standardisation and automation but also on being able to control the external environment and working conditions. This is achieved by limiting exposure to risk, as when an airline grounds flights in bad weather, and also by controlling working conditions so that there are, for example, strict controls on how many hours civil aviation pilots can fly and how long they must rest before flying again. With enough resource this would be achievable in some areas of healthcare, and indeed some areas are already very safe. However if we cannot control the demand and working conditions, we necessarily have to rely on more adaptive approaches to safety; a different model may be intrinsically safer but simply not feasible in a particular context. While civil aviation is indeed a source of inspiration and learning such a model is only currently applicable in a relatively limited set of circumstances in healthcare. The approach taken to safety in any healthcare setting may ultimately depend in part on what is politically feasible which will vary by discipline, organization and jurisdiction.