Measurement – the scale of harm
There are several ways to measure or estimate the extent of errors or adverse events in healthcare, including analysis of administrative or reporting data, case record review, observations of patient care, and active clinical surveillance. Each of these methods has advantages and disadvantages in terms of availability of data, cost, reliability, observer bias, and clinical relevance. Clearly, the measurement of safety overlaps with the measurement of quality of care, described in other chapters. An adverse event is defined as an unintended injury caused by medical management rather than the disease process, which is sufficiently serious to lead to prolongation of hospitalization or temporary or permanent disability or death.4 This definition is important because it has been traditionally used in studies of the nature and scale of harm, described below. A critical incident or ‘near miss’ may be considered the next step down from this incident, which may have caused harm but did not actually do so.
Over the last 40 years, a number of international adverse event studies have been published in which retrospective case record review was used to identify adverse events and assess the nature and scale of harm in acute hospitals. A selection of these studies are shown in Table 11.1. Rates of adverse events in most recent studies lie between 8 and 12%, a range now accepted as typical of advanced healthcare systems. The rate per patient is always slightly higher, as some patients suffer more than one event, and about half of adverse events are generally judged to be preventable. Rates in the United States are much lower, and those in Australia appear to be much higher. The lower US rates might reflect better‐quality care but probably reflect the narrower focus on negligent injury rather than the broader quality improvement focus of most other studies.
Studies of errors
A number of methods of studying errors and adverse events have evolved over time and been adapted to different contexts.19 Each method has particular advantages and disadvantages. Some methods are oriented toward detecting the incidence of errors and adverse events (Table 11.2), whereas others address the causes and contributory factors of those errors and adverse events (Table 11.3). There is no perfect way of estimating the incidence of adverse events or errors. For various reasons, all the approaches give a partial picture. Any retrospective review is vulnerable to hindsight or outcome bias, where knowledge that the outcome was bad leads to unjust simplification and criticism of preceding events. Record review is comprehensive and systematic but, by definition, is restricted to matters noted in the medical record. Reporting systems are strongly dependent on the willingness of staff to report and are a very imperfect reflection of the underlying rate of errors or adverse events. Michel, et al. compared three methods of estimating adverse events rates: cross‐sectional (data gathered in one day), prospective (data gathered during a hospital stay), and retrospective (review of medical records). The prospective and retrospective methods found similar numbers of cases; however, the former identified more preventable cases with high reliability and validity.20 The cross‐sectional method showed a large number of false positives and did not pick up serious events.
Why do adverse events occur?
To obtain a true understanding of patient safety, it is not enough to assess the scale of healthcare‐associated harm: we have to look deeper to understand the processes that underlie it. As described later, attempts to understand the causes of adverse events are routinely made within individual organizations in several ways, such as through local morbidity and mortality meetings or the use of more structured approaches such as root cause analysis.
Learning from other industries
On a more general level, patient safety researchers have used many methods to enhance our knowledge of the causes of adverse events. It is important to appreciate that when an adverse event occurs, we may be quick to judge or to blame the actions or omissions of individuals, but careful inquiry usually shows that deficiencies in our systems are also at fault. We have learnt much from other industries in this respect. Investigation of major disasters such as the Chernobyl nuclear explosion, the Space Shuttle Challenger crash, and the Paddington rail accident identified ‘violations of procedure’ or problems resulting from actions or omissions by people at the scene. However, further analysis of these events revealed ‘latent conditions’21 further upstream in the process, which allowed these violations to occur and have such a devastating effect. ‘Latent conditions’ are often a result of gradual and unintentional erosion of safety‐enhancing processes because of other pressures: for example, cutting training budgets to reduce costs. Further in the background are often deeply ingrained cultural and organizational issues, some of which may be elusive and difficult to resolve. Of course, it is very well to learn about the underlying causes of these non‐healthcare‐related disasters, but the question that most clinicians will ask at this stage is how they are relevant to us. Although healthcare is similar to these industries in some respects, such as the high level of inherent risks and the presence of well‐meaning and dedicated staff, it is very different in others, such as diversity, often non‐centralized administration, uncertainty, and unpredictability.
Human error
Human error is not easy to define, as boundaries are often blurred between the actions or inactions of individuals and the deficiencies of the systems in which they work. However, it is important to define and classify different sorts of errors in medicine, largely because this may help us learn from incidents. We can think about errors in medicine in relation to the clinical processes involved – for example, prescribing errors or diagnostic errors – but perhaps it is also useful to look at the underlying psychological themes. In his analysis of different types of error, Reason22 divided them into two broad types of error: slips and lapses. These are errors of action and mistakes that are, broadly speaking, faults of knowledge or planning. He also discusses violations that, as distinct from errors, are intentional acts that, for one reason or another, deviate from the usual or expected course of action.
Table 11.1 International adverse events studies, showing data for older patients.
Sources: Mills6; Brennan, et al.7; Wilson, et al.8; Thomas, et al.9; Vincent, Neale, and Woloshynowych10; Davis, et al.11; Baker, et al.12; Forster, et al.13; Michel, et al.14; Sari, et al.15; Sousa, et al.16; Rafter, et al.17; Nilsson, et al.18
| Study | Year | No. of subjects | No. (proportion, %) of elderly subjects | Definition of elderly (years) | Overall adverse event rate (%) | Incidence in elderly (%) | Incidence in young (%) | Difference |
|---|---|---|---|---|---|---|---|---|
| California (Mills) | 1977 | 20,864 | 3826 (18.34%) | ≥65 | 4.65 | 7.22±0.82 | 4.07±0.30 | p < 0.05 |
| Harvard (Brennan) | 1991 | 30,121 |
|
