Nursing Inquiry 2012; 19(1): 6–17 Feature A critical realist approach to knowledge: implications for evidencebased practice in and beyond nursing Stuart Nairn University of Nottingham, Derby, UK Accepted for publication 27 May 2011 DOI: 10.1111/j.1440-1800.2011.00566.x NAIRN S. Nursing Inquiry 2012; 19: 6–17 A critical realist approach to knowledge: implications for evidence-based practice in and beyond nursing This paper will identify some of the key conceptual tools of a critical realist approach to knowledge. I will then apply these principles to some of the competing epistemologies that are prevalent within nursing. There are broadly two approaches which are sometimes distinct from each other and sometimes inter-related. On one side, there is the view that all healthcare interventions should be judged on the principles of randomised controlled trials and the other is a preoccupation with language in which healthcare interventions are subjected to a discursive interrogation. These debates are configured through the idea of a hierarchy of knowledge that is accorded uncritical acceptance by some and virulent distaste by others. I will argue that the notion of hierarchy is problematic and is largely argued for in unproductive epistemological terms. What is required is a shift towards a theory that emphasises the contextual nature of the ways that knowledge is produced and disseminated. In other words, there is no single hierarchy of knowledge, but there are multiple hierarchies of knowledge. Key words: critical theory, evidence-based practice, mixed method, nursing theory, poststructuralism. Historically, there has been a tension between science and philosophy which began to emerge more explicitly during the enlightenment when science was referred to as natural philosophy (Thorne 2009). The gradual development of a conflict between science and philosophy is based on science’s propensity for prediction and experimental control. The openness of philosophical discourse is closed by a preoccupation with methodological robustness and a commitment to the objective world and its erasure of the subjective and contextual aspects of knowledge. By the early nineteenth century therefore the common view of science was that: You can call a body of knowledge science only once you’ve decided it shouldn’t be called philosophy, and scientists, similarly, are otherwise learned people who maintain a principled silence on matters of ontology and morality and politics and God (Thorne 2009, 119). Correspondence: Stuart Nairn, Derby Education Centre, School of Nursing, Midwifery and Physiotherapy, University of Nottingham, London Road Community Hospital, Derby DE1 2QY, UK. E-mail: <stuart.nairn@nottingham.ac.uk> This anxiety that science, and in the context of healthcare the medical model, closes of knowledge, is deterministic and reductionist in character, is a central concern for nurses who have an ongoing commitment to holistic care and to multiple sources of knowledge rather than to a restrictive scientific one. But this anxiety is also associated with a general critique of knowledge as a source of power and influence, the recurrent suspicion that the accumulation of knowledge simply sets up artificial parameters of what it is possible to include as legitimate (Foucault 1972). Thus, nurses lack influence, are a subordinate profession, but if we turn our knowledge base into a scientific discourse then we will gain influence and status, which it is then assumed will improve patient care (Davies 1995). It is in the contentious area of evidence based practice (EBP) that this debate has become particularly intense, producing a mixture of debate and in some cases angry exchanges (Porter and O’Halloran 2009). It is the purpose of this paper to analyse these arguments from a broadly critical realist perspective. This approach offers the possibility of  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge finding a way through this debate, which incorporates a realist approach to knowledge. It is unlikely and indeed not my intention to please everyone and there are some arguments in the literature that I believe are unhelpful and should be rejected, but the central tenets of critical realism and its prioritisation of ontology over epistemology offers a route into a relational approach to knowledge that incorporates a pluralist approach to finding out about the world alongside a realist commitment to the structural properties of both the natural and the social world. This is not to suggest that there are no other realist ways of addressing these problems (see Niiniluoto 2002, chapter one), but the purpose of this paper is to outline the possibilities of a critical realist approach. This paper is structured in the following way. I will start with a general overview of ontology from a critical realist perspective. I will then examine how the competition between different epistemologies is addressed in the literature. First, I will look at how causation, when reduced to a set of empirical regularities, can produce a reductionist account of knowledge. Second, I will address the strengths and limitations of focusing on language and the discursive. Finally, I will situate these ideas within the idea of a hierarchy of knowledge and how a critical realist account might tackle the notion of a hierarchy. I will cover a number of diverse theories and will doubtless oversimplify the nuances and tensions within these theories. For example, I have not addressed the extensive literature on mixed methods approaches to research which attempt to acknowledge many of the nuances of nursing knowledge. However, the main purpose of this paper is to grapple with the central tenets and the theoretical approaches of these theories and the tensions between them. Finally there is an extensive literature in this area and I will be addressing only a fraction of those writers who have examined this issue, but my main purpose is to develop an approach that makes EBP and nursing knowledge more generally, usable for nursing practice. kar the purpose is to re-orientate theory towards ontology. The tendency for both theories is to reduce ontological questions to epistemological ones or to confuse our descriptions of the world with the world itself, the so called epistemic fallacy. So, positivism is preoccupied with developing robust methods, axiomatic rules of enquiry that are usually statistical, as a means of accessing precise knowledge about the world, while hermeneutics is often focused on a self-referential preoccupation with the endless play of language as a substitute for their erasure of the real. Both are ways of knowing, strategies for making sense of the world. While the former is committed to accessing the world, it nonetheless accords the methods employed as determinants of how the world is, it therefore has a tendency to erase ontology or the epistemological trumps the ontological. Interpretivists adopt a variety of approaches to the problem ranging from a questioning scepticism to a complete rupture between concept and object that can sound perilously close to a denial that a world even exists outside our perceptions of it. The central problem for critical theory, whether in its realist or non-realist forms, is the relationship between how we describe the world and the world itself, between concept and object. For critical realism maintaining the category distinction is essential and is summed up in the terms such as the intransitive and the transitive. The intransitive is equated with ontology and a real world of objects with their own causal powers and structures. These are not inert objects, the real changes, social structures change and mountains erode, but they nonetheless occupy a single world and are therefore not part of multiple worlds as poststructuralists are prone to argue. The transitive, however, is multiple and relativist and is associated with epistemology, it is related to how we make sense of the world and crucially is situated within certain socio-historical contexts. CRITICAL REALISM AND ONTOLOGY Underpinning this distinction is the notion of a stratified ontology, which has two elements: first, the empirical, the actual and the real and second the way that these concepts operate within different domains of the real. The empirical, the actual and the real are central to Bahskar’s assertion of a non-positivist approach to scientific knowledge. The empirical is simply what a person perceives from their senses, what they see and feel, what they experience. The actual exists on the level of events, what actually happens or occurs and the real is the identification underlying mechanism that may or may not occur. So for example I may, on an empirical level, experience a heart attack or may observe an individual having a heart attack, or I may identify how many heart attacks The relationship between ontology and epistemology The relationship between epistemology and ontology is important to critical realism. Ontology is the world as it is, the constituents of the world, how it is made up, while epistemology is concerned with the ways we can adopt to access knowledge about the world. Bhaskar’s (1998, 2008a) central arguments about ontology are to critique both positivist and hermeneutic approached to knowledge. They both adopt epistemological approaches to knowledge, whereas for Bhas 2011 Blackwell Publishing Ltd The empirical, the actual and the real 7 S Nairn occur over a time period or between different groups such as class and gender and then may produce calculations about whether there are any significant differences or changes over time. On the level of the actual, some heart attacks may escape our attention, a silent MI for example or a person may collapse, die and be buried without the heart attack being identified. However, none of this information explains why the heart attacks occur or why different social classes experience more or less incidents than another. To do this requires us to identify underlying mechanisms such as diet, stress, material circumstances, behaviour, biology, and so on. A mechanism is therefore something, which is capable of producing an observable event and can be a natural biological mechanism that then produces a thrombus or a social mechanism, such as inequalities in health that can generate a stress response and contribute to thrombus formation. In short, there are multiple mechanisms at work that need to be identified before we can understand the ‘real’ underpinning heart disease. Positivism tends to stay within the realm of the empirical, whereas critical realism operates within the real defined as a potential rather than an actual event. For example, my body has the potential to produce a blood clot, but the mechanism that activates a blood clot will only occur if there is some damage to my body that requires a blood clot to repair it. The mechanism is dormant more often than it is active. The creation of a clot formation during a heart attack is a specific event, the breaking off of atheroma within a coronary artery, but its activation could have multiple causative mechanisms that occur both immediately and over a prolonged period of time. Therefore, a mechanism is where the real resides and is the object of scientific enquiry. Therefore, critical realism is primarily a philosophy of causation based on ontological realism. It asks the question, what must the world be like to make science possible? (Bhaskar 2008a). Because the world has underlying mechanism that may or may not be activated, scientific questions are about when and how those mechanisms are enacted and when countervailing mechanisms prevent their activation. So, the mechanism of clot formation is dependent on the causal powers intrinsic to the properties of blood. Drugs such as anticoagulants are a countervailing mechanism that may under the right circumstances prevent clot formation and alleviate cardiac problems but can be disastrous for the person if they are involved in a road traffic accident and experience hypovolaemic shock. So, critical realism: gives priority to potentiality over actuality and to actuality over experience. Foregrounding the causal powers of entities and processes, it stresses that tendencies of generative mechanisms may be real, yet unexercised, exercised, yet unactua- 8 lised, and actualised independently of human perception or detection. A causal power can be exercised over and again or not be exercised at all. Although exercised, its power may be neutralised by other countervailing powers …, so that actually nothing results and factually nothing happens. And when something happens, it may well happen without anyone noticing it … the experimental situation in which the real (the potential), the actual (the factual) and the empirical (the observable) coincide is an exceptional one and should not be taken as the rule (Vandenberghe 2009, 218). Therefore, the experimental method creates an artificial situation that attempts to identify the causal powers of an object outside of their natural setting. So if the researcher is able to identify, in laboratory conditions that an antibiotic has the capacity to destroy a bacteria, there is no guarantee that in the real world it will work in the same way. Its efficacy may be undermined by other mechanisms (Pawson and Tilley 1997) such as nutritional status or poor housing conditions and have unintended consequences such as the production of bacterial resistant strands. Closed experimental conditions are particularly difficult to attain in either medical or nursing research and it is for this reason that randomised controlled trials are difficult to construct and apply to practice (Pawson and Tilley 1997). These epistemological problems of accessing knowledge about the world are therefore particularly acute in health research. While the most solid evidence can be attained in the laboratory, real world research is more focused on probabilities rather than certainty. A stratified reality The second aspect of ontology is the relational and in depth character of the real. The properties of the real are not the same from one object to another and therefore there is little possibility of adopting the same methodological technique to access knowledge about it (Bhaskar 1998, 2008a). So there are different mechanisms at work within microbes than there are in animal behaviour. A geneticist may identify a certain mechanism at work within a gene that may, when interacting with other genes, produce blue eyes or brown hair but it does not provide an model for explaining human behaviour such as criminality despite the best efforts of socio-biologists’ to make direct links in this way. As human beings, we live within a number of different ontological levels which interact with each other but need not be conflated with each other. So we have a biological level, an individual psychology and a social situation. None of these different levels are by themselves sufficient to explain the other ontological levels (Elder-Vass 2010). So, one cannot explain global finance by genetics or vice versa. The social is not possi-  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge ble without our biology but it is emergent from and not reducible to the other (Rose and Rose 2000). For example, the medical model is not a complete explanation for all aspects of the illness experience, but symptoms are often emergent from the underlying biological mechanism. I will explore these ideas further as the paper develops and use them as tools to critique existing approaches to nursing knowledge. COMPETING EPISTEMOLOGIES 1: IT’S ALL ABOUT NUMBERS of a research paper is whether the research has identified an accidental association or a genuine causal effect (Hitchcock 2008). In other words, have they neutralised all the confounding variables that could have undermined the internal validity of the research? Indeed, one of the central distinctions one can make between positivism and critical realism is that the former is largely against or at the very least highly sceptical about causative claims (Persson 2010), while one of the defining features of critical realism is its commitment to a theory of causation and attendant terminology such as emergence, retroduction and causal powers (Sayer 1992). The legacy of positivism Empirical realism The CONSORT group, whose stated aim is to improve the quality of clinical trials, starts its update on its guidelines for reporting randomised trials by arguing, in language that they seem to assume as being non-controversial that: ‘Randomised controlled trials (RCTs), when appropriately designed, conducted, and reported, represent the gold standard in evaluating healthcare interventions’ (Schulz et al. 2010, 698). The purpose of the guidelines is to promote the adoption of clear criteria for journals to assess the quality of trials for inclusion in academic journals and they state on their website that 50% of core medical journals endorse these principles (Consort Group, 2010). RCTs are therefore presented as the gold standard of knowledge and what is required for any health intervention is a methodologically and statistically robust approach to empirical data. This fits into a view of science as an anti-philosophy, the scientific rigour of the trial is what counts, so why would any philosophical enquiry be necessary? The main complaint about this type of thinking is that it represents a medically reductionist approach to knowledge and presents the researcher as some independent, decontextualised observer. The result of this criticism is usually a call for qualitative research to be included as evidence with its own legitimacy outside of traditional hierarchies of evidence. Further criticisms might include labelling the proponents of RCT’s as positivist as a means of incorporating them into a widely discredited philosophical system (Traynor 2002; White and Willis 2002). The term positivism, however, could be considered a misnomer in this context. Two of the key components of positivism are its emphasis on verification and its anti-causal approach to knowledge (Hacking 1983). While the former is a usually a feature of clinical trials through the adoption of statistical methods to verify that a specific healthcare intervention works, the second anti-causal logic of positivism is not. Causal claims are a ubiquitous feature of clinical trials and the key to understanding the methodological principles  2011 Blackwell Publishing Ltd A more useful way of characterising the predominant view about EBP and the prioritisation of the statistical is empirical realism. Bhaskar (2008a) uses this term to critique a method of producing science that relies upon statistical regularities. In this approach, a constant conjunction of events is sufficient to verify and establish law like regularities, which are then translated in practice based on the probability that these regularities provide sufficient and robust evidence for clinical work. This approach can provide useful information and I would certainly prefer knowledge from an RCT before taking some medication than anecdotal evidence from someone who tells me it felt good for them. RCTs can be used in very sophisticated and complex ways (Richards and Hamers 2009), but some caution is required in extrapolating this approach to knowledge as a gold standard for all healthcare interventions. Empirical realism’s theory of causation is based on the principle that when X is followed by Y and then regularly follows from X to Y then an association has been established. This successionist approach is confirmed by comparing an experimental group with a control group where the experimental group is exposed to an intervention and the control group is not. The effect of the intervention is then measured by comparing the two groups post test. A causal effect is then inferred by the association. The strength of this approach is that it can identify certain causal tendencies within the real world at a specific moment of time, which can then be utilised to go further in trying to identify what mechanisms are at work and what mechanism are obstructing their realisation. But to do this, entails a more robust ontological approach to the empirical tendencies identified. Context, mechanism and outcome A critical realist approach argues that empirical realism is operating at the ontological level of the empirical rather 9 S Nairn than the real. Inferring a causal relationship between two events requires an understanding of the mechanism (M) that connects the two events and the context (C) within which they occur (Pawson and Tilley 1997). An outcome (O) can only be measured when both context and mechanism have been understood. Therefore, C+M = O. For example, the contexts of cardiac arrests vary considerably, occurring at home, in the street, on airplanes or in hospitals. The mechanisms that operate in an arrest also vary, they may be a consequence of a traumatic, a cardiac or a respiratory event and within those categories numerous individual and contextual variables will be at work. Interventions activate mechanisms, such as chest compressions or defibrillation, social interventions such as lay training, access to technology, and so on. The research on the efficacy of CPR has major problems in addressing the multiple contexts in which CPR occurs and the contingency and relative rarity of the event make organising a coherent sample a further problem (Cummins 1997). The adoption of empirical realism, which attempts to neutralise context so that a pure CPR intervention can be adopted is therefore problematic and the often very different results in mortality between different research, suggests that the artificiality of the experimental method and the desire to produce a closed system that this method implies is difficult. This suggests that research in this area must be more local and specific, reflecting the multiple mechanisms at work within the varied contexts within which cardiac arrests take place. In other words, context should not merely be a variable to be neutralised but an essential component of the research process (Nairn and Timmons 2010). The traditional scientific approach of identifying population, intervention, comparison and outcome (PICO) works to an extent, but is explicitly based on neutralising the context of the object of knowledge and the context of the research process. In CPR, for example, the use of drug therapy is based on animal experiments and delivered in the early circulatory phase on an arrest, whereas most clinical trials are conducted at a later stage when metabolic acidosis is already taking place (Reynolds, Rittenberger, and Menegazzi 2007). The contexts of real cardiac arrests are erased by a preoccupation with empirical realism rather than evaluating the contexts and mechanisms within which (in this case) weak evidence, based on animal experiments, is translated into clinical practice with no convincing evidence of efficacy. PICO is a key approach in developing knowledge within the EBP movement and is explicit in the Cochrane review handbook (Higgins and Green 2011) While the Cochrane collaboration is cautiously examining other approaches, such as a chapter on qualitative reviews, the fundamental approach 10 remains within a quantitative ⁄ PICO approach. Furthermore, pharmaceutical companies have little interest in examining the context within which their medications operate. As Musschenga, van der Steen, and Ho (2010) argue, the knowledge that non-pharmaceutical interventions can be more effective than pharmaceutical ones is unwelcome, but as RCTs are expensive and therefore dominated by corporations, the type of research carried out is biased towards the needs of shareholders rather than human need. The context and dissemination of evidence is as important as the methodologies employed by researchers. The result is a tension between academics pursuing knowledge, but who then need to access research funding with all the attendant problems of the commercialisation of knowledge (Chan and Fisher 2008). This strategy no doubt pleases pharmaceutical companies, but the results of such research, while often useful in identifying certain potential tendencies and the causal powers of a drug, cannot necessarily be extrapolated into clinical usefulness with a high level of certainty. This is not to suggest that the EBP movement has ignored these criticisms; the limitations of RCTs are now widely discussed and the argument to incorporate the CMO approach is on the agenda in established medical journals. For example, Berwick (2008) makes a strong case for RCTs when appropriate, but argues that they should not be the default position. He draws upon the tension over the evidence for the use of rapid response teams, which has been questioned by RCTs, but argues that such evidence is problematic on the basis that this type of complex, multi-component intervention is about processes of social change which cannot be evaluated adequately by RCT’s. This is not a rejection of this method. RCT’s are an essential method for understanding what works and what does not work, but it needs to be emphasised that they work best when the intervention is relatively straightforward, such as a single drug against a placebo. For many nursing interventions, such as identifying the efficacy of a stroke nurse, the multiple causative agents at work make RCTs inherently problematic and require a multiple and extensive research programme utilising diverse methodologies (Pawson and Tilley 1997). COMPETING EPISTEMOLOGIES 2: IT’S ALL ABOUT WORDS The turn to language The so called linguistic turn in philosophy has had an impact on many of the criticisms of EBP. Writers inspired by Derrida, for example, Freshwater and Rolfe (2004), expend much  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge effort critiquing certain texts and deconstructing them. By identifying the tensions and contradictions within texts and the rhetorical devices associated with those promoting EBP, these authors attempt to undermine these texts through a process of deconstruction. While this strategy provides some useful insights into the ways that language can be used to obscure these problems and then present EBP as a more coherent philosophy than it really deserves, it nonetheless presents clinical subjects such as nursing with unnecessary problems. The preoccupation with language as a discourse, largely disconnected from the real, or to see the real simply as an expression of perception, places this approach firmly in the world of the empirical. As language is both a sophisticated means of expression, but also incompetent when accessing knowledge about the world around us, all we are left with is language games, identifying how the meanings of words are constantly deferred and put under erasure the moment they are written. All texts become a blurring if not an outright erasure of the object of knowledge (Derrida 1998). While the rupture between concept and object is never complete in Derrida’s work (cf. Norris 1997) the referent is little more than a trace, and while this may be useful in subjects such as cultural studies and literary theory whose main focus is the study of texts (Norris 2002), it is not clear what use this has to a clinical subject like nursing. If nursing texts are so removed from the real world of nursing practice and if this is regarded as an intrinsic feature of language then it is unclear what use any nursing texts are in relation to clinical practice. What tends to happen with this type of work is that it takes a problem, the relation between the transitive and the intransitive, concept and object, and turns it into an irresolvable problem. All one is left with is a constant deferral of meaning where even one’s own comments and critiques are constantly undermined at the very moment that they are written. The result is a kind of postmodern empiricism where one is unable to say anything about an event, an object and occurrence of any kind for fear of imposing a definition on it that simply reflects a power ⁄ knowledge discourse. So, what starts off as a sophisticated theoretical engagement with the ways that knowledge can be misused, becomes a general critique of all claims to knowledge, a profound scepticism that is disabling rather than enabling. To avoid this, it is important to assert the real underpinning such discourses and their relationship to pre-existing cultural and structural features of the social world (Archer 1995, 1996). From a critical realist perspective, the real is anything that has causative properties. This is not simply material objects but also language. Language, words and texts have an effect, which is why a managerial discourse, a  2011 Blackwell Publishing Ltd nursing discourse and a medical discourse are important to analyse, precisely because of their causal effects. If causation is put on hold, or if the researcher is unwilling to identify causative mechanisms, then qualitative research stays within the domain of the empirical which is epistemologically distinct from, but ontologically congruent with an empirical realist approach. So what does the real mean from a critical realist perspective? Something is real if it has an effect or makes a difference. Since entities such as mountains and discourses clearly make a difference, in the sense that they cause human beings to act in ways they would not in the absence of these entities, then mountains and discourses are real. Entities such as fairies are not real, although entities such as the discourse of fairies are real: if people think fairies are real, they may undertake actions such as trying to photograph them (Fleetwood 2004, 29). In relation to researching qualitative phenomena, critical realists are as much concerned with the real as when they are researching in the natural sciences, but the object of knowledge is different and requires different approaches, and what unites them is an emphasis on causation, of identifying generative mechanisms. So if research identifies a group of people who hold racist opinions then the purpose of the research is not simply to report it but to attempt to identify what social, contextual circumstances generate these false beliefs. Similarly from a quantitative perspective, it is not simply the epidemiology of health inequalities that should interest us, but what generative mechanisms are at work in producing those inequalities. For critical realists, it is not the primacy of text or the primacy of numbers that matters, but the primacy of context. This shifts the argument away from which epistemology is best to what is our ontological theory about a particular object of knowledge and then to examine the strengths and weaknesses of that description through whichever epistemological approach is likely to give the most convincing answer. In short, language, despite its problems, is not the main object of knowledge for critical realists, but rather a route into trying to make sense of a world that is more than mere language. No matter how clumsy or accurate our use of language may be we can still identify the difference between a useful concept and one that is not. Although concepts occupy the domain of the transitive and are therefore inherently fallible I can still make claims that are ‘practically adequate’ if never certain (Sayer 2000). So if I see a patient bleeding from their carotid artery and identify that their pulse has increased and blood pressure has dropped, it is reasonable to conclude that hypovolaemic shock is occurring. I can then respond by putting pressure on the bleed 11 S Nairn and then initiating intravenous fluids. Few healthcare interventions are this straightforward which is why care should be taken about applying our concepts too readily to the real but this does not mean that we can only limit ourselves to discussions about those concepts. For example, the placebo effect works at a number of different levels, but is primarily about how we interpret the world. The placebo effect and the context of meaning We attach meanings to objects and are able to make some sense of those relationships and how they work. So, the placebo effect operates at the level of meaning. Objects, such as a drug, are not simply a cluster of biochemical processes that have causal efficacy on our biological bodies, but also a meaning centred mechanism that can also causally affect the body from the cultural meanings we attach to the intervention. It is suggested that the placebo response may constitute 30% of the efficacy of a drug, but this is not consistent from one drug to another but varies with the healthcare intervention and above all it only works where the meaning system employed is culturally appropriate (Brody 2010). The ritual aspects of health delivery are different in different cultures and so a Shamanic ritual may work in some parts of the world but not in others, while the ritual of the drug round may be more effective in some cultures than others. From an orthodox biomedical reductionist approach, the placebo is a ‘suspect category’, to be neutralised or erased (Brody 2010), but it is also problematic in that causation seems to go the wrong way, from context to biology rather than underlying pathophysiology to symptom to psychological status with the hope that the sociocultural context can be completely ignored. The placebo effect is primarily about the context of meaning or as Miller and Kaptchuk (2008) put it ‘contextual healing’. It is about the symbolic effects of language, interactions and objects and how people interpret these meanings. These have direct biological consequences and have been identified at a neurobiological level (Benedetti et al. 2005), producing increasingly sophisticated understandings of the complex and diverse ways in which the mechanism of the placebo, or contextual healing, operates. The critical realist point to make here is that the discursive, the semiotic and the biological all have effects but they operate in different domains of the real: the cultural, the biological, the psychological and so on. The activation of the placebo effect cannot be reduced to empirical realism nor to discourse, but through an understanding of the potentiality of the placebo response and making sense of what contexts and in what way the effect can be utilised by healthcare professionals. The 12 placebo is not simply a case of the cultural being an epiphenomena of the biological, nor is it a case of the biological being reducible to the cultural. We could not have the ability to attach meanings to objects, such as a pill, without our neurons, but our interpretation of those objects is influenced by the sociocultural, which then feeds back into our biology. It could therefore be argued that homoeopathy may effectively utilise contextual healing; it fails on a biochemical level, while in too much orthodox medicine the opposite is the case. This presents us with a relational ontology in which the object of knowledge influences our ways of knowing, rather than starting with epistemological assumptions and then conflating those assumptions with a predetermined view about how the world works. Foucault and the real In short, words and discourses do not exhaust our knowledge about the world, although they are key to communicating our understanding of it. We experience the world in non-language forms, but discourses have powerful effects. This is why a writer such as Foucault, despite being philosophically associated with poststructuralist ideas, is an important theorist and has been defended by some academics sympathetic to the critical realist approach (Al-Amoudi 2007; Frauley 2007). Although these writers have different views about Foucault, I would argue that, particularly in his early archaeological work (Foucault 1972), Foucault is open to a realist reading or perhaps more cautiously one could say that realists need not dismiss Foucault out of hand. Foucault makes clear links between how discourses operate and their relationship with institutions and therefore can only exist within certain structural relationships. Therefore, discourses are not simply free floating but rooted in the social and economic circumstances of the time, be they hospitals, prisons or universities. Foucault also distinguishes between connaissance and savoir. A connaissance is the formal rules that govern a particular topic such as biology or economics, while savoir refers to the totality of knowledge’s and the conditions that are necessary for a type of object to emerge. So these develop into a scientific domain which is bound by certain laws of analysis and an archaeological territory which may include science but is also associated with reflexivity, narrative and the socio-historical context that allows specific knowledge to emerge. This distinction provides an implicit acknowledgement of a stratified ontology. The strength of Foucault’s analysis is related to his understanding of how cultural discourses are created and the means by which certain ideas are included or excluded in a prevailing discourse. The problem is when these discourses become deterministic descriptions of scientific knowledge, as  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge Foucault himself and some of his followers are prone to do (Armstrong 1987), which have a tendency to reduce biology to a problem of perception. With this caveat in mind, his work can be a fruitful epistemological tool for accessing knowledge about topics such as surveillance mechanism and the problematic relationship between power and knowledge in the social and cultural domain. In short, the power of language need not be dismissed on the basis of its indeterminacy and its resistance to an empirical realist approach, but requires its own mode of analysis that can be relationally understood in the contexts in which these discourses emerge (Fairclough, Jessop, and Sayer 2004). COMPETING EPISTEMOLOGIES 3: HIERARCHIES OF KNOWLEDGE Carper’s patterns of knowing It is perhaps the notion of hierarchy that worries many of us when looking at topics such as EBP. Nursing history is characterised by a subordinate position to medicine, both in practice and as an academic subject. The medical model reigns supreme and began to dominate our understanding of health and illness long before it had any efficacy in alleviating human suffering (Foucault 1973; Porter 1997). Like any subordinate profession, strategies are employed to tackle this in a number of ways but confront a number of structural and cultural obstacles (Davies 1995). From a knowledge perspective, attempts to see nursing as a distinct body of knowledge that exists independently of medicine has been a key strand in asserting the status of nursing. One of the strategies used is to level out hierarchical approaches to knowledge. Carper’s (1978) seminal article can be seen as one way of doing this. Her four patterns of knowing the empirical, the aesthetic, personal knowing and ethics are all different epistemological strategies for knowing about nursing. There is an explicit concern about the over emphasis on the empirical, which is conflated by Carper with the term scientific at the expense of the other ways of knowing, but there is no attempt to deny the importance of this pattern of knowing; is as fundamental as the others. The four types of knowledge are all said to be separate but interrelated and each way of knowing should be restricted to the circumstances in which they are deemed to be valid. This is not so much an attempt to make out that each pattern is equally valid, although it is one possible interpretation of her model, but that nursing needs to incorporate each pattern for a truly holistic model of care to be implemented. This approach has been criticised for its willingness to separate these patterns out without convincingly bringing them back together again or  2011 Blackwell Publishing Ltd to explain what is meant by the interrelated nature of these patterns (Risjord 2010, 51–2). However, the emphasis on ‘knowing’ or epistemology, reflects a problem with this type of approach. Again the problem becomes which epistemology is most useful rather than starting with ontological presuppositions about an object of knowledge and then identifying what strategy might be the best for accessing knowledge about that topic? The hierarchy is undermined by Carper’s model but then re-emerges as a competition between different epistemologies as in Freshwater’s (2004) argument that the aesthetic component of Carper’s model has been marginalised. Moreover, there is no attempt to negate any of the four patterns of knowing but there is an argument about one pattern being sidelined by the others and in particular the dominance of the empirical. Porter’s (2010) intervention is particularly interesting in this sense as the emphasis of his critique of Carper is that EBP is an empirical approach to knowledge that should underpin the other patterns of knowing. Porter argues that it is the ‘ineffability’ of the aesthetic, the personal and the ethical that explains why empirical knowledge is more influential within EBP. The empirical puts its evidence in plain sight, its methodological approach, its data analysis and outcome measurements, and can therefore be challenged by others including academics, practitioners and patients. The other three patterns seem to contribute to an esoteric and unchallengeable basis for professional decision-making. The empirical for Porter can be either qualitative or quantitative and should be inserted into the other patterns of knowing without, he argues, reducing those components to the empirical. So in relation to ethics he argues that: While the danger of doing violence to the complexities of ethical issues by reducing them to empirical questions is one that should not be underestimated, this does not relieve nurses of the responsibility to make their ethical approaches transparent. Nor is it beyond our ken to develop sensitive and occupational tools to enable us to do this (Porter 2010, 11). In other words, Porter is expecting nurses not only to adopt an ethical position but also that the implementation of those decisions in practice should be empirically identified. So, are nurses treating patients with respect and so on? This would require the researcher to operationalise the concept of respect, identify an outcome measurement and then evaluate clinical practice on the basis of these constructs. As Porter admits, the risk is that complex social and ethical interactions can be reduced to a set of instrumentalist tools in which nurse’s ethical interactions are judged on pre-determined criteria, a tick box approach that says more 13 S Nairn about managerial concerns about presentation than a genuine engagement with nurses ethical decision-making. The reductionist approach to patient satisfaction research is a good example of how complexity can be reduced to a managerial discourse of efficacy (Williams 1994). I would also argue that qualitative research is no guarantee that these complexities can uncovered, but used sensitively and taking due account of structural pressures (Nairn 2009) they can be an effective way of exploring these ideas. While I am sympathetic to Porter’s argument and defence of EBP, it rests upon the unresolved problem of what constitutes evidence and for Porter this tends to be somewhat broader than what might be called the orthodox discourse of EBP. Defining a hierarchy The ‘orthodox’ view of EBP maintains a clear hierarchy of knowledge with RCTs at the top and the anecdotal at the bottom with cohort studies, case studies and so on somewhere in between. However, even within this discourse there is considerable variation in what constitutes the hierarchy and one author cites research that has identified some 40 different ranking scales (Worrall 2010). There are for example disagreements within these rankings about which is better, cohort studies or case control studies and some include meta-analysis and others do not (Worrall 2010). Of course qualitative research barely gets a mention, but while the notion of hierarchies is problematic, emphasising and indeed fetishising particular approaches at the expense of others, this does not mean that hierarchies of knowledge are irrelevant, that all types of knowledge are all equally valid, a relativistic discourse that rather attractively deconstructs the biomedical preoccupation with statistics, but then leaves us with no criteria for evaluating the quality of the evidence to be used in clinical practice. The problem is that the central concern is primarily epistemological, whether it is Carper’s patterns of knowing, Freshwater’s (2004) concern about aesthetics, Porter’s emphasis on the empirical or the EBP emphasis on some deterministic hierarchy appropriate for all health interventions with RCT’s at the pinnacle of efficacy. From a critical realist perspective, there is no need to start with epistemological suppositions, but rather from an ontological theory about the nature of the object of knowledge, followed by an epistemological discussion about how to access knowledge and test one’s ontological theory. One might say that as a consequence that there is no single hierarchy of knowledge, nor should we give up on hierarchies, but that we should acknowledge that there are hierarchies of knowledge for 14 each object of knowledge, or there is one world but multiple hierarchies of knowledge. For example, a cardiac arrest occupies a number of different levels of the real. One can study it at a biological level, so what biological mechanisms are at work. An object of knowledge may be a defibrillator and assessing the efficacy may require a cautious adoption of RCT’s as a high level of knowledge based upon the known properties of cardiac arrhythmias and the causal powers of a defibrillator. But if one wanted to know whether a relative would benefit from being present during CPR, the debate over witnessed resuscitation, it may be that RCT’s are at the bottom of the hierarchy of knowledge as the personal experience of the relative trumps a deterministic set of statistics about the relative’s responses. The aesthetics of a resuscitation room emits a set of signs that are technological in nature and how we respond to that environment while a relative is being resuscitated is a matter of meaning and related to a semiotics of high technology environments that is unstable rather than regular and predictive. The context of knowledge There is therefore a cultural context about medicine, nursing, technology, self-identity, psychology that require diverse research approaches to understanding and making sense of what is happening within the event of a cardiac arrest. Postmodernists may over-emphasise uncertainty, but they are correct in being sceptical about deterministic claims to knowledge. This is not simply a problem for qualitative research. Ioannidis (2005) has argued that most quantitative research findings are false at the level of methodology and one can further argue that the context in which research takes place, the funding sources, the managerial needs of healthcare systems, the institutional and career pressures on academics to publish also compromises the quality of evidence (Bourdieu 2004). The point here is that EBP and more broadly, nursing knowledge, is part of a political discourse and not simply the patient accumulation of data for the benefit of patients. While science likes to present itself as being outside of culture, ‘the culture of no culture’ in Haraway’s (1997) wry terminology, knowledge is inevitably a part of the social context of its own production. For critical realist theory, with its emphasis on ontology, the real is not simply at the level of the empirical, but fundamentally about the different causal mechanisms at work and then to examine the different ways that they interact with each other. There is one world, but multiple ways of examining that world; the transitive is always potentially fallible knowledge, but its purpose is to access as much of the truth as possible by  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge understanding how biological approaches might be influenced by social factors and vice versa. The more mechanisms that are at work, the more difficult it is to identify the real and the more problematic it is to produce research that can access that reality. This is why physics is often held up as the closest to a purely scientific subject, while biology is less so and sociology even less so. But from a critical realist perspective, it is not that these subjects are more or less scientific. It is our methodological tools that have to reflect the object of knowledge which then affects our ability to accurately grasp that reality. So the tools required to understand the anatomy and physiology of the human body are going to be different to the methods required to understand how society works or the economy. As physics can set up closed experimental conditions to identify physical mechanisms and the social sciences cannot, or not with the rigour of the natural sciences, then the level of certainty produced by these methods will be different. An ontologically more rigorous approach is therefore central to critical realist approaches and this is not a straightforward matter and dependent on one’s object of knowledge (ElderVass 2010). Furthermore not everything can be reduced to the scientific method. Bhaskar does not argue that science should be perceived as a supreme or overriding value and so not all knowledge can be assessed by science, nor do the objects of science exhaust reality: On the contrary, they afford only a particular angle or slant on reality, picked out precisely for its explanatory scope and power. Moreover, alongside ethical naturalism I am committed to moral realism and I would like also to envisage an adjacent position in aesthetics, indeed viewing it as a branch of practical philosophy, the art of living well … Starting with knowledge as a systematic phenomenon I reject that cognitive triumphalism, the roots of which lie in the epistemic fallacy, which identifies what is (and what is not) with what lies within the bounds of human cognitive competence. Reality is a potentially infinite totality, of which we know something but not how much (Bhaskar 2008b, 15). So science, statistics, words, experiments, experience, do not exhaust reality; they are instead the always problematic means we adopt to try and access that part of the world we happen to be focused on. So some objects of knowledge are susceptible to the scientific method because of their intrinsic properties but others, such as ethics cannot, or perhaps should not, be subsumed under the rubric of science. CONCLUSION Within nursing there is a growing interest in the use of critical realism to tackle the epistemological problems of access-  2011 Blackwell Publishing Ltd ing knowledge about nursing practice (Clark, Lissel and Davis 2008). I believe that this is an encouraging strand that has the potential to reorient nursing to a scientific realist approach to knowledge that can more explicitly identify its object of knowledge and then retroductively adopt strategies of enquiry appropriate to that object. This is premised on a stratified ontology, and given the multi-faceted range of objects, subject areas and conceptual problems that characterise nursing practice, I would argue that critical realism offers a useful way of addressing these problems in a nuanced and practical way. Knowledge about health and illness does not need to be confined to any one particular methodological principle because illness operates in a number of different ways and reflects the different stratified contexts in which illness occurs. The different contexts may or may not require different approaches to understanding, from the science of biology to subjective meaning. As a philosophical approach, critical realism adopts epistemologically diverse approaches to the ways in which knowledge of the world can be accessed but within realist ontology. Therefore, knowledge is a practical product of how human beings interact with the world rather than an abstract product of human thought; as nursing is very much about how we intervene in a world of practical and real problems, critical realism can provide a useful approach to engaging with those problems. ACKNOWLEDGEMENTS I would like to thank Martin Lipscomb for his helpful observations on an earlier draft of this paper. I am also very grateful to the two anonymous reviewers for their productive comments. REFERENCES Al-Amoudi I. 2007. Redrawing Foucault’s social ontology. Organization 14: 543–63. Archer M. 1995. Realist social theory: The morphogenetic approach. Cambridge: Cambridge University Press. Archer M. 1996. Culture and agency: The place of culture in social theory. Cambridge: Cambridge University Press. Armstrong D. 1987. Bodies of knowledge: Foucault and the problem of human anatomy. In Sociological theory and medical sociology, ed. G Scambler, 59–76. London: Tavistock Publications. Benedetti F, HS Mayberg, TD Wager, CS Stohler and JK Zubieta. 2005. Neurobiological mechanisms of the placebo effect. The Journal of Neuroscience 25: 10390–402. 15 S Nairn Berwick DM. 2008. The science of improvement. Journal of the American Medical Association 299: 1182–4. Bhaskar R. 1998. The possibility of naturalism. 3rd edn. London: Routledge. Bhaskar R. 2008a. A realist theory of science. London: Verso. Bhaskar R. 2008b. Dialectic: The pulse of freedom. London: Routledge. Bourdieu P. 2004. Science of science and reflexivity. Cambridge: Polity Press. Brody H. 2010. Ritual, medicine, and the placebo response. In The problem of ritual efficacy, eds WS Sax, J Quack and J Weinhold, 151–68. Oxford: Oxford University Press. Carper BA. 1978. Fundamental patterns of knowing in nursing. Advances in Nursing Science 1: 13–23. Chan AS and D Fisher. 2008. The exchange university: Corporatization of academic culture. Vancouver: UBC Press. Clark AM, MA Lissel and C Davis. 2008. Complex critical realism: Tenets and application in nursing research. Advances in Nursing Science 31: E67–79. Consort Group. 2010. The Consort Statement. http:// www.consort-statement.org/consort-statement (accessed 18 May 2011). Cummins RO. 1997. Human research on cardiopulmonary resuscitation: Current constraints on implementation. New Horizons 5: 120–7. Davies C. 1995. Gender and the professional predicament in nursing. Buckingham: Open University Press. Derrida J. 1998. Of grammatology. Corrected edn. London: The John Hopkins University Press. Elder-Vass D. 2010. The causal powers of social structures: Emergence, structure and agency. Cambridge: Cambridge University Press. Fairclough N, B Jessop and A Sayer. 2004. Critical realism and semiosis. In Realism discourse and deconstruction, eds J Joseph and JM Roberts, 23–42. London: Routledge. Fleetwood S. 2004. An ontology for organisation and management studies. In Critical realist applications in organisation and management studies, eds S Fleetwood and S Ackroyd, 27–53. London: Routledge. Foucault M. 1972. The archaeology of knowledge. London: Routledge. Foucault M. 1973. The birth of the clinic: An archaeology of medical perception. London: Routledge. Frauley J. 2007. Towards an archaeological-realist Foucauldian analytics of government. British Journal of Criminology 47: 617–33. Freshwater D. 2004. Aesthetics and evidence-based practice in nursing: An oxymoron? International Journal for Human Caring 8: 8–12. 16 Freshwater D and G Rolfe. 2004. Deconstructing evidence-based practice. Oxon: Routledge. Hacking I. 1983. Representing and intervening: Introductory topics in the philosophy of natural science. Cambridge: Cambridge University Press. Haraway DJ. 1997. Modest_Witness@Second_Millenium. FemaleMan_Meets_OncoMouse: Feminism and Technoscience. London: Routledge. Higgins JPT and S Green., eds. 2011. Cochrane handbook for systematic reviews of interventions version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. http://www.mrc-bsu.cam.ac.uk/cochrane/handbook/ (accessed 5 September 2011). Hitchcock C. 2008. Causation. In The Routledge companion to philosophy of science, eds S Psillos and M Curd, 317–26. London: Routledge Ioannidis JPA. 2005. Why most published research findings are false. PLoS Med 2: e124. DOI: 10.1371/journal.pmed.0020124. Miller FG and TJ Kaptchuk. 2008. The power of context: Reconceptualising the placebo response. Journal of the Royal Society of Medicine 101: 222–5. Musschenga AW, WJ van der Steen and VKY Ho. 2010. The business of drug research: A mixed blessing. In The commodification of academic research: Science and the modern university, ed. H Radder, 110–31. Pittsburgh: University of Pittsburgh Press. Nairn S. 2009. Social structure and nursing research. Nursing Philosophy 10: 191–202. Nairn S and S Timmons. 2010. Scientific uncertainty and the creation of resuscitation guidelines. Social Theory and Health. 8: 289–308. Niiniluoto I. 2002. Critical scientific realism. Oxford: Oxford University Press. Norris C. 1997. Against relativism: Philosophy of science, deconstruction and critical theory. Oxford: Blackwell. Norris C 2002. Deconstruction. Revised edn. London: Routledge. Pawson R and N Tilley. 1997. Realistic evaluation. London: Sage. Persson J. 2010. Misconceptions of positivism and five unnecessary science theoretic mistakes they bring in their train. International Journal of Nursing Studies 47: 651–61. Porter R. 1997. The greatest benefit to mankind: A medical history of humanity from antiquity to the present. London: Harper Collins. Porter S. 2010. Fundamental patterns of knowing in nursing: The challenge of evidence-based practice. Advances in Nursing Science 33: 3–14.  2011 Blackwell Publishing Ltd Critical realism and nursing knowledge Porter S and P O’Halloran. 2009. The postmodern war on evidence-based practice. International Journal of Nursing Studies 46: 740–8. Reynolds JC, JC Rittenberger and JJ Menegazzi. 2007. Drug administration in animal studies of cardiac arrest does not reflect human clinical experience. Resuscitation 74: 13–26. Richards DA and JPH Hamers. 2009. RCTs in complex nursing interventions and laboratory experimental studies. International Journal of Nursing Studies 46: 588–92. Risjord M. 2010. Nursing knowledge: Science, practice and philosophy. Oxford: Wiley-Blackwell. Rose H and S Rose. 2000. Alas, poor Darwin: Arguments against evolutionary psychology. London: Jonathan Cape. Sayer A. 1992. Method in social science: A realist approach, 2nd edn, London: Routledge. Sayer A. 2000. Realism and social science. London: Sage.  2011 Blackwell Publishing Ltd Schulz KF, DG Altman, D Moher and for the CONSORT Group. 2010. CONSORT 2010 statement: Updated guidelines for reporting parallel group randomised trials. PLoS Med 7: e1000251. DOI: 10.1371/journal.pmed.1000251 Thorne C. 2009. The dialectic of the counter-enlightenment. London: Harvard University Press. Traynor M. 2002. The oil crisis, risk and evidence-based practice. Nursing Inquiry 9: 162–9. Vandenberghe F. 2009. Realism in one country? Journal of Critical Realism 8: 203–32. White K and E Willis. 2002. Positivism resurgent: The epistemological foundations of evidence-based medicine. Health Sociology Review 11: 5–15. Williams B. 1994. Patient satisfaction: A valid concept? Social Science and Medicine 38: 509–16. Worrall J. 2010. Evidence: Philosophy of science meets medicine. Journal of Evaluation in Clinical Practice 16: 356–62. 17