This systematic narrative review has revealed a relatively sparse literature on the important concept of mindlines, first introduced in 2004 by Gabbay and le May. Whilst we included 340 publications in our final dataset, some authors appeared to have (with the best of intentions) used this term incorrectly and naively. But others had recognised and explored mindlines’ fundamental philosophical challenge to EBM. The relatively limited contributions from the guideline development community suggest that the concept has been largely ignored by the guideline industry—perhaps because, as American physicist and philosopher of science Thomas Kuhn wrote, ‘the proponents of competing paradigms practice their trades in different worlds’ .
‘Knowledge translation’, one-way transfer of knowledge from producer (research) to user (clinician) [49,50], remains a (contested) policy challenge. Multiple research traditions have contributed to a vast body of literature on how evidence from research is disseminated [51,52]. Underlying this literature is a fundamental hope for the possibility of optimising the ‘intermediation’ of knowledge—that is, the managed processes by which practitioners can be supported to interact with knowledge . But an alternative metaphor that goes beyond the concept of ‘spreading’ good ideas is needed .
In their book from 2011, Gabbay and le May explain in great detail the origins of mindlines, their implications and related theories . Mindlines fit a view that knowledge is not so much a set of external facts that are waiting to be ‘translated’ or ‘disseminated’ but a more fluid and multi-directional phenomenon in which knowledge is ‘re-created’ in different contexts by different people again and again as previously postulated by Mol and Nonaka [55,56]. From this perspective, improving knowledge intermediation is more like maximising the opportunity to create knowledge. How this might be achieved differs from the EBM paradigm in a number of ways uncovered by Gabbay and le May throughout their book. We explore these philosophical questions further below.
Reality—single or multiple?
Firstly, mindlines question our assumptions about reality. Without doing full justice to the vast literature and on-going debate on the philosophy of knowledge, we acknowledge in particular the philosophical difference between naïve realism (in which there is one reality, which exists independently of human thought and can ultimately be known by everyone) and more pluralist philosophies (such as subjective idealism and critical realism) which assume multiple realities, none of which are fully shared .
The possibility of multiple realities explains why different national guidelines for even simple conditions like urinary tract infections draw different conclusions from the latest research . Naïve rationalists (some protagonists of EBM, for example) think that we will overcome these differences and ultimately have a clear set of ‘facts’, recommendations, international clinical guidelines and policies that everyone except the misguided and ignorant will all agree upon. In contrast, as Gabbay and le May point out , the concept of mindlines presents us with the idea of a shared (but by no means homogeneous) reality consisting of multiple very individual and temporary realities of people: clinicians, researchers, guideline makers and patients. Mindlines offer a view that the number of guidelines on any topic will never cease to expand because we expect new individual realities and scientific paradigms to emerge continuously.
The nature of knowledge
Absent in the conventional EBM paradigm, but very present in the concept of mindlines, is the acknowledgement that, to use Polanyi’s phrase, ‘we know more than we can tell’—that is, not all knowledge is conscious or explicit . Gabbay and le May discuss how mindlines include ‘knowledge-in-practice-in-context’: practical knowledge formed not only by the setting but also by the need for that knowledge . Conventional EBM views knowledge narrowly as factual data, only a tiny fraction of which are relevant to the decision at hand (and that it will be largely self-evident which ‘facts’ are needed). The mindlines concept envisions a ‘negotiating space’  where clinical decision-making by clinicians and patients involves a process of reduction and prioritisation from a vast realm of potentially relevant knowledge of different kinds.
In this sense, we would argue that mindlines stress both the act of and the need for reducing possible options for action. EBM does not reject the idea of reducing knowledge—for example, the guideline development process encourages the dismissal of evidence and knowledge that is considered of low quality [60,61]. But in mindlines, the process of reduction applies not only to explicit knowledge. All kinds of knowledge are competing for attention [12,39]. In clinical consultations, not everything is or can be taken into account; there is limited time , our brains do not process everything , we are forgetful and the ‘whole’ story is not told to us. Reduction is an essential and ever-present process to create knowledge and experience reality. As Heisenberg wrote ‘what we observe is not nature in itself but nature exposed to our method of questioning’ . Arguably, what we observe as clinicians is not reality itself but the reality exposed to our method of reducing or filtering the various potentially relevant streams of knowledge of which we are consciously or unconsciously aware and from those, constructing a picture of current reality.
As we explore a clinical case by questioning, examining and testing, some things get more ‘certain’, but the overall picture will not necessarily become clearer. For example, in the case of abdominal pain, a stool sample makes us more certain about the organisms involved. However, it says little about the social context in which a particular type of pain occurs and recurs. If we ask a patient about her daily life, we may build a fuller, clearer and more holistic understanding of the abdominal pain—and the stool sample may become less of a priority. Indeed, an incidental growth of a pathogen in the stool sample may confuse and distract us if the ‘real’ cause of the abdominal pain is migraine, marital stress or the side effects of medication. Reducing in the spirit of mindlines is a creative process, not a reductionist one, like carving a particular ‘abdominal pain’ out of a piece of buzzing, blooming reality  with an infinite number of dimensions.
How the ‘truth’ is arrived at
Gabbay and le May offer many detailed examples of how clinicians co-construct knowledge and discuss several theories that help to understand these processes such as the knowledge creation cycle developed by Nonaka and Takeuchi explaining the growth of tacit knowledge in organisations . But unlike the papers we found that simply see mindlines as heuristics, Gabbay and le May explain that logical thinking is embedded in mindlines. What we derive from this is that the set of tools used to reduce or construct (depending on our perception of realities) the knowledge base for a clinical decision differs dramatically between conventional EBM and mindlines. In the former, the dominant tool for identifying knowledge is rational, conscious questioning, and the main requirement of that knowledge is validity—that is, if it fulfils the criteria of correspondence (to the real world—for example, through a robust sampling procedure), coherence (with what we define as a logical system of high-quality knowledge—for example, derived from a well-conducted randomised trial and meeting the standards of statistical prediction) and consensus (experts agree—for example, through peer review). The literature on mindlines, as reflected for instance in the circular evidence model suggested by Walach et al. , fundamentally challenges these philosophical criteria by which a finding will become classified as ‘true’ as they are inadequate to make predictions in the real world of clinical practice.
In particular, EBM intentionally focuses on so-called ‘frequentist’ reasoning, in its quest to discard mechanism-based reasoning and reliance on (potentially unreliable) clinical expertise [65,66]. This type of reasoning is an evasion of the fundamental inability to predict the future—or more precisely the problem of induction . If we look at an association between A (such as a taking a tablet) and B (such as a clinical outcome) in the rich context of everyday practice, we may discover a web of interacting influences linking A with B. Evidence-based reasoning is essentially a process of stripping the causal net (compare Pearl ) from the association between A and B (‘bias’), in order to find a single general yet virtual rule—and then applying this rule to predict the future in another situation where A and B, but also the bias, are present. This might be termed reality-to-rule-to-reality reasoning. In contrast, mindlines allow other evasions of the induction problem (such as Bayesian learning from a one-off experience ) in a chain of reasoning that might be termed reality-to-pattern-to-reality, which allows practitioners to keep the network of causality intact from one case to the next.
As articulated in the idea of systems 1 and 2 knowledge , mindlines draw more on tacit knowledge, the knowledge we subconsciously use when focusing on the things we want to do . Subconsciously knowing how to interpret gestures, smell, interaction, environment and time during a consultation reduces prevailing uncertainties and helps us to further shape our holistic understanding and make predictions of what is likely to happen in this case.
More importantly, mindlines encapsulate a more sophisticated and comprehensive concept of truth than traditional EBM. As Gabbay and le May eloquently explain , mindlines take a constructivist approach to knowledge, assuming that it is created in social processes, through discourse, influenced by cultural and historic forces. This chimes with the work of the Russian philosopher Mikhail Bakhtin, who proposed that a unified truth involves sharing personal knowledge with others who provide a separate perspective. ‘The idea lives not in one person’s isolated individual consciousness—if it remains there only, it degenerates and dies. The idea begins to live, that is, to take shape, to develop, to find and renew its verbal expression, to give birth to new ideas, only when it enters into genuine dialogic relationships with other ideas, with the ideas of others’ . Similarly, the Austrian philosopher Ludwig Wittgenstein argued by the same reasoning that all knowledge is collective .
Aligning with this focus on the intersubjective nature of knowledge creation [72,73], we contend that contrary to the fears expressed by Glasziou in the quote above, mindlines are not void of validation processes in spite of being mainly tacit. They convey strong and rich elements of shared sense-making (and hence consensus-making), both conscious and unconscious; they address correspondence with reality as it pushes back in the local context; and they address coherence using other types of evasions of the induction problem. In sum, mindlines can be accurate and useful in a local setting and provide useful predictions, despite not being construed according the set of reduction tools and beliefs underpinning the EBM paradigm.
Economics, politics and ethics
The political, economic and ethical dimensions of the processes of knowledge creation in the papers in our sample were almost never directly linked to mindlines. This is surprising, given that mindlines and traditional EBM differ considerably in this perspective. Several papers in our sample noted that politicians, research leaders, management consultant firms, lobbying groups, the pharmaceutical industry and many other powerful actors use their influence to define research priorities, what counts as medical evidence, how knowledge is distributed and how resources are allocated [31,39,74-76]. In EBM, population-derived statistical estimates fit the needs of policymakers as they provide truths that are—apparently—‘right’ for groups and those who interact with those groups, such as governments and the pharmaceutical industry. What is viewed as good care for a defined group as a whole is in some way regarded as good for individuals. Mindlines however lack this overarching ‘built-in’ criterion of what is right or wrong patient care. With other authors, Gabbay and le May worry that mindlines can spread ‘collective folly’ .
EBM strongly adheres to the ‘deficit model’ , which entails that clinicians and patients are regarded as deficient in certain knowledge: evidence-based knowledge. This is considered a moral problem of ‘leaving people incapable of understanding the world around them’ . Mindlines, on the other hand, correspond more with the idea that anyone, including patients, create valid knowledge too and can be ‘experts’ in consultations . With the current evolution towards person-based medicine and practices , the deficit model may be set to give way to a more pluralist and constructivist one. But at the same time, this may uncomfortably question our basic assumptions about who decides what is good or bad care.
Knowledge management, knowledge intermediation
Although Gabbay and le May aptly remark ‘how ironically inconsistent it would be if [they would] try to dictate how [their] work should be put in practice’ , many of the implications they discuss have been explored in the literature we found. We discovered articles that explored which sources of knowledge clinicians actually use, projects that aimed to bring research and practice closer together and the development of organisational structures such as communities of practice and virtual social networks to support the use of ‘knowledge-in-practice-in-context’. But what strikes us when considering these implications for practice is the question of whether controlling knowledge creation is actually feasible. EBM assumes that knowledge can be managed, and that, through intermediation, the knowledge deficit of both practitioners and patients can be rectified.
In contrast, mindlines remind us by their emphasis on tacit knowledge that knowledge creation is in large part unmanageable. A one-off event experienced by an individual is all too real for them, ‘anecdotal’ though it may be to others. Furthermore, we cannot control all interactions nor can we control all aspects of resources and contexts. Knowledge development is an organic rather than rational process, which can only be controlled to a limited extent. Currie et al. note that the implication of mindlines is that ‘any attempt at managing professionalised and tacit knowledge in health care through the mobilisation of explicit and codified knowledge faces significant challenge’ . As Contandriopoulos et al. conclude in their review on the dissemination of knowledge: ‘…the quest for context-independent evidence on the efficacy of knowledge exchange strategies is probably doomed’ . This picture of knowledge as fluid, multiple, uncontainable and defying rationality is a long way from the hopes and dreams of the EBM movement  or even science itself. Thomas Kuhn touches the core of the matter writing: ‘We are all deeply accustomed to seeing science as the one enterprise that draws constantly nearer to some goal set by nature in advance. But need there be any such goal?’ .
However, it would be wrong to conclude that because mindlines are constructed, tacit, emergent and shared, they are directionless. On the contrary, there is evidence that the knowledge of mindlines is ‘self-organising’, tending to achieve stability over time. Broekaert noticed ‘real human commitment consists of an open, methodical, meaningful search for the best solution for a certain problem’ . Consciously and subconsciously, we collectively create and continuously refine more or less enduring frameworks to look at the world, based on our previous experiences, opinions of colleagues and experts, practical knowledge, guidelines and articles, produced in discourse, agreement and consensus with others, limited by psychological abilities, contexts and the physical world. The more closely the statement, ‘This patient probably has a viral illness, tomorrow she will be OK, even if she doesn't take antibiotics today’ persists in such a framework, the more real, useful and valid it becomes.
Persistence is not without risk. In evidence-based guideline development and research, we are used to synthesising a single version of reality to settle differences  and provide consistency of care. But ‘consistency’ of care can harm as well as help individual patients since such an approach may limited our list of differential diagnoses to common or obvious options, removing the possibility of managing the unusual case differently . Similarly, we have to prevent our mindlines from becoming too rigid. Unanimity, or the absence of logical contradiction, prevents the development of new, competing theories and innovations, which needed from time to time to replace the current paradigms .
This raises the question of why persistence of knowledge in the form of mindlines is valuable if knowledge is inherently ephemeral and too much persistence risks making our decisions too rigid. Ernst Mach argued: ‘If our dreams were more regular, more connected, more stable, they would also have more practical importance for us’ . Haridimos Tsoukas contends that ‘[U]nderstanding presupposes an Archimedes’ point, a perspective (undoubtedly an irremediably open-ended and evolving perspective, but a perspective nonetheless) from which the world may be viewed, accounted for, and interpreted. Ironically, abundantly available information leads to formlessness and, thus, to a diminished capacity for understanding.’ . Persistence keeps a perspective open for exploration, prediction and guidance of human behaviour. We need persistence to help us to see and find new events, insights and practices so we limit discontinuity or instability that nobody agrees with. Most of infinite reality is not created, distributed, translated or mediated. We miss things because we have to in order to experience anything at all.
Hasok Chang  argues that scientific realism should commit to pursue many theories to find where reality ‘resists’, whilst investing to preserve theories that did not seem to work that well. In the future, those might turn out to give helpful alternative insights. Applied to mindlines, this may translate to a call to create a broad menu of mindlines to find where collective reality ‘resists’ using many methods of truth finding. If we want to intermediate the process of knowledge creation (to the limited extent that this is possible), further research needs to look into how to speed up the cycle of building and turning over many more persistent mindlines, whilst keeping alternative, less persistent ones afloat efficiently.
In sum, mindlines offers a philosophically and theoretically sophisticated perspective on knowledge and clinical method. Yet in 10 years since the concept was introduced, the study of mindlines has remained a minority sport within critical social science whilst research within the EBM movement on the generation, circulation and use of evidence has remained predominantly (though by no means exclusively) wedded to a naïve rationalist view of knowledge.
The strength of this review is its tight focus on the word ‘mindlines’ and Gabbay and le May’s original 2004 paper in the literature to address the question of how the word and their work have been used, and the meta-narrative approach which allowed us to consider different philosophical assumptions behind different uses of the word. The limitation is that the concept of mindlines may have been discussed more extensively in grey literature, institutional reports and other forums, which we would not have detected using our search.
We hope this review will encourage practitioners and policymakers, along with academics, to embrace fully the implications of the mindlines paradigm. In our sample, Malterud noticed that the EBM movement does not limit the best evidence to randomised controlled trials (RCTs) and meta-analysis, ‘Yet, the foundation for integrating the available sources of knowledge remains unclear. We still do not know whether convincing information leads to optimal decision making’ . Similarly, Jonas argues that ‘we need to broaden and deepen our understanding of what counts as “evidence” and which types of evidence are best used to inform differing aspects of clinical decision making’ . A new research agenda is needed, which should centre first and foremost on the processes and interactions by which mindlines are validated by both clinicians and patients. Research should also seek to break down the walls between EBM and mindlines, for example, by exploring how mindlines emerge and are negotiated in guideline development groups and research communities. Through such interdisciplinary work, it should be possible to identify ways to broaden the methods that such groups could use to create richer and more valid forms of ‘evidence-based’ knowledge.