VOL: 100, ISSUE: 22, PAGE NO: 42
Dawn Dowding, PhD, RN, is senior lecturer, Hull York Medical School, University of York
Carl Thompson, DPhil, RN, is senior research fellow, Department of Health Sciences, University of York
This is the third of four papers discussing judgement and decision-making in nursing. The first paper in this series (Thompson et al, 2004) highlighted the importance of judgement and decision-making to nursing practice. The second (Dowding and Thompson, 2004) discussed how complexity associated with decision problems could be made sense of by using an approach to structuring decisions known as decision analysis. The aim of this article is to discuss the issue of judgement in nursing. In particular, it examines the way nurses may use information to inform their judgements, and ways in which this process can be assisted to improve the accuracy of judgements.
Judgement in nursing
The process of judgement involves integrating different aspects of information (which may be about a person, object or situation) to arrive at an overall evaluation (Maule, 2001). In nursing this could be considered as the process of using different types of clinical information about the patient (such as appearance, vital signs, and behaviour) to make an assessment of her or his current health status (Dowding and Thompson, 2003).
Judgements feed into decision-making (Box 1) in that the evaluations or assessments an individual makes can be used as the basis of choice between alternatives. For example a nurse may assess a patient as being ‘at risk’ of developing a pressure ulcer (judgement) and then ‘choose’ a particular intervention to reduce that risk (decision) on the basis of the assessment.
Examining judgements in nursing is important, as they have an effect on decisions taken about patient care. Harvey (2001) suggests decisions may be poor because the judgements on which they depend are inaccurate or because individuals combine different judgements inappropriately. Therefore, a key issue for nurses and patients is ensuring judgements are as accurate as possible.
There are two main reasons for inaccuracy:
- The nurse may be using information that has no utility for the judgement in question (Cioffi 2002);
- The nurse may be placing too much importance on particular information (Dowding, 2002).
Therefore, the type of information individuals use to inform their judgements, some knowledge of the information they should be using to inform their judgements, and how that information is (or should be) combined is required to investigate and improve accuracy. The two main ways these issues have been investigated are descriptive research and social judgement analysis.
Descriptive research into judgement
Most of the research examining judgement in nursing is qualitative and descriptive in nature - an appropriate design for the research questions being addressed. The aim of many of the studies is to describe the nature of judgements through the analysis of how nurses manage clinical situations, including the information they use to inform their judgements and decisions.
The use of intuition and rules
Perhaps the most well known research in this area is that carried out by Benner et al (Benner, et al 1999; 1992; Benner, 1982). This research highlighted characteristics of ‘expert’ nursing practice and judgement and how that expertise develops. Benner (1982) suggests expert nurses mainly use intuition, which is defined as ‘knowing without necessarily having a specific rationale or making explicit all that goes into one’s sense of a situation’ (Benner, 1999).
This is in direct contrast to less experienced individuals who may use rules to combine common attributes such as a patient’s vital signs (Benner, 1982). This combination may eventually be combined into some form of ‘global’ pattern that guides action.
Although Benner’s work has provided insight into the nature of expert nursing practice, it fails to give details of how information is processed to inform accurate judgements. This is due in part to the research methods used - predominantly observation of practice and interviews.
However, observation cannot provide insight into all the information used in reaching a judgement, and self-reporting has been shown to be an unreliable method of investigating judgement and decision-making as individuals often have little insight into how they make judgements and decisions (Harries et al, 1996). Also, the critical incident method used by Benner et al (1999) may mean individuals only examine situations where their reasoning processes have been successful (Lamond and Thompson, 2000), meaning a full exploration of issues of judgement accuracy is not possible.
Another set of studies used the psychological theory of information processing (Newell and Simon, 1972) as the basis for exploring the reasoning processes nurses use when making judgements and decisions. This theory suggests humans have limited capacity for processing information, meaning a variety of strategies is employed to assist the process. Examples of this type of study have been carried out by Cioffi (1997), Tanner et al (1987), and Corcoran (1986). These studies have suggested that nurses use a process of hypothetico-deductive reasoning when making judgements, together with mental short cuts or ‘heuristics’.
Hypothetico-deductive reasoning involves using available information to formulate hypotheses, which are then tested and reformulated until a conclusion is reached (Thompson and Dowding, 2002). The types of information that appear to be used vary considerably. For instance in a very early study examining the information nurses use to make a judgement about patient pain, Hammond et al (1966) found they used 165 different information cues. Hypothetico-deductive reasoning appears to be used by individuals in situations where they have no experience of the task in question. In situations where people have more experience, they are more likely to use a process of ‘pattern matching’, which involves the recognition of similarities between the patient case being considered and ones that have been encountered in the past (Elstein et al, 1990). These ‘short cuts’ are the focus of the fourth paper in this series.
The main strategies used to examine reasoning and information use in information processing studies are variations of a ‘think aloud’ technique and retrospective interviewing (Tanner et al, 1987; Corcoran, 1986). Simulations are typically used to compare individuals across cases. The process of thinking aloud involves the subject of the study verbalising everything they think of while carrying out the judgement task. They may be interviewed after the task to discuss any other information they think they used and their rationale.
There are a number of problems with this type of study: the use of simulations may mean the judgements made by the subject do not reflect what they would do with a patient. Also, thinking aloud relies on the participant’s ability to make their judgement policies explicit (Harries and Harries, 2001), and retrospective interviewing suffers from the same problems as highlighted above.
Limitations of descriptive research
In summary, if we are interested in the accuracy of judgements, much of the descriptive research into nursing practice fails to provide the evidence that is needed to inform practice. These types of study are a useful representation of practice but it is difficult to observe a sufficient range of scenarios for a given judgement in order to determine how information is used to make that judgement (Harries and Harries, 2001).
Many of the studies look at a broad range of practice, which means detail about the information cues is often lacking. Also, a reliance on self-report methods (such as interviews and thinking aloud) means the research is dependent on a participant’s insight into her or his judgement processes and ability to verbalise these processes.
By definition expert judgement usually involves the use of automatic, unconscious thought processes (often referred to as intuition). Such experts often will not be able to verbalise their thoughts - a characteristic that limits the analysis of their judgements (Lamond and Thompson, 2000).
Social judgement analysis
The lens model
The theoretical basis of social judgement analysis is the ‘lens’ model of cognition proposed by Brunswik. This is a representation of the relationship between a person and her or his environment (Harries and Harries, 2001). Brunswik suggested that to investigate judgement, researchers should take into account the unpredictable nature of the environment in which they operate, and that a range of judgements, in a range of situations, needs to be investigated (Harries and Harries, 2001).
The lens model can be represented diagrammatically (Fig 1). In this diagram the left-hand side represents the environment (such as a patient’s state of health). A number of different information cues will be related probabilistically to this environment. The right-hand side represents the individual making the judgement. This person uses information cues to make her or his judgement on the environment (for example, do I need to call a doctor?) and in doing so will attach more weight to some cues than others. By comparing the way the information cues are related to the state in the environment and the weighting assigned to information cues by the judge, one can identify:
- If the person’s judgement is accurate (is there a correspondence between patient state and the judgement?);
- Whether the judge uses appropriate information, and if so, does she or he put appropriate importance or weighting on different pieces of information (Dowding, 2002).
Disagreement between judges
Using this model highlights where disagreements occur between judges: they could be using different information or be placing different importance on certain cues, which would lead to differences in judgements.
In order to model the environment statistical techniques are used to identify possible relationships between information and a patient state. For instance, in a study examining doctors’ diagnoses of heart failure, Skanér et al (2000) used information from patient cases to model how different cues were related to the diagnosis. This ‘optimal strategy’ suggested cardiac enlargement was the most important cue to determine if a patient had heart failure. To model the clinician’s judgement, a number of scenarios of patient cases are constructed containing information considered important for the judgement under investigation, and designed to represent the range of situations in the environment.
In social judgement analysis studies, the number of scenarios is often very large to make the judgements as real as possible (Harries and Harries, 2001). The judge(s) are then asked to make a judgement about each of the scenarios, and this is then also modelled using statistical techniques (usually of linear multiple regression). This provides a statistical analysis of the information the judge uses to make judgements, and the importance she or he attaches to each of the cues.
For instance, Skanér et al (2000) studied the diagnostic judgements of GPs, cardiologists, and medical students. Through their modelling of how individuals used information to make a diagnosis, they highlighted the variation in the use of information cues. One-third of participants used relative heart volume as the most important cue - as opposed to cardiac enlargement, which was identified in the optimal strategy.
As well as being able to identify possible sources of error in judgement - which may affect judgement accuracy - the results of social judgement analysis studies can be used to provide ‘cognitive feedback’ to participants as a way of improving their accuracy. Cognitive feedback is different to outcome feedback, which provides participants with the outcome of each case, in that it contains information about the ‘optimal’ strategy (how information is related to the patient state in the environment) and the individual’s own policy (how she or he uses the information). With this knowledge they can identify disparities and be aware of how to improve their use of information (Wigton, 1996). Various studies that used cognitive feedback have shown it can improve diagnostic accuracy and prognostic predictions (Wigton, 1996).
Social judgement analysis requires individuals to make judgements as they normally would, and then uses statistical techniques to describe the relationship between the information available to the judge and the judgement or decision made (Harries and Harries, 2001). The focus of these studies is not the process of judgement, rather an analysis of how information use is linked to judgement accuracy, so in this way studies are able to analyse in detail how and why judgements may differ among individuals, as well as offering a way of improving accuracy through the use of cognitive feedback.
Another strength of social judgement analysis is that it is not reliant on the ability of participants to self-report their judgement processes, and can identify policies that judges are unaware of (Harries and Harries, 2001). However, social judgement analysis studies are often reliant on the construction of scenarios, frequently with limited sets of information presented in a way not found in reality. So, as with all other types of study, they do have limitations.
As highlighted by Hammond et al (1966) nursing judgements are complex, often involving the need to process a large number of information cues. Key issues in the study of such judgements are the analysis of judgement accuracy and ways of improving accuracy.
More traditional approaches to the study of nursing judgement have provided valuable insights into the nature of expert nursing practice and the complexity of practice. However, they have limitations in terms of being able to provide the specific data needed to address judgement accuracy.
Social judgement analysis approaches may be a way of overcoming these limitations. However, as yet these approaches have been more common in medicine, examining the nature of medical diagnosis and prescribing (Skanér et al 2000; Harries et al 1996), than in nursing practice.
With nurses taking on roles requiring accurate judgement, it is time for clinicians and researchers to grapple with this thorny issue in ways that will reveal possible routes forward rather than offering just description.
- This article has been double-blind peer-reviewed.
SERIES ON CLINICAL DECISION-MAKING
This is the third in a four-part series on decision-making:
1. Strategies for avoiding the pitfalls in clinical decision-making
2. Using decision trees to structure clinical decisions
3. How to use information ‘cues’ accurately when making clinical decisions
4. Tools for handling information in clinical decision-making.