White, Wellman, and Nazer 22 make the case that objective maps of intellectual structure produced using author co-citation analysis (ACA) have a deep affinity with insiders' perceptions of the structure of their own fields. We held such an insider perception as we began, and that perception is reflected in the following brief overview of the knowledge utilization field and its most obvious subsets (domains). These domains (knowledge utilization, diffusion of innovation, technology transfer, evidence-based medicine or EBM) are, we argue, substantively similar on the basis that they all address the idea of solving social problems with knowledge. They differ along such dimensions as core problems of concern, knowledge used, audiences of relevance, and sometimes modes of transfer.

Rich has argued that the roots of the knowledge utilization field date back to the time of the ancient Greeks 23, although most scholars date it no further back than the earliest studies in innovation diffusion credited to the French sociologist Gabriel Tardé over a century ago 24. Numerous literatures and traditions (some overlapping) are subsumed within the broad knowledge utilization domain. Some authors have conceptualized knowledge utilization as a broad domain over-arching all others 2526. We believe that there has been a strong thread that constitutes knowledge utilization proper whose scholars concern themselves with the relationship of knowledge (often in the form of scientific research) to policy 1723272829303132333435. The most often cited source from this broad overarching knowledge utilization field is Glaser, Abelson, and Garrison's encyclopedic review of the literature on the topic 36. Backer described the evolution of the knowledge utilization field specifically 37; Valente and Rogers 38 and Rogers 10 described evolution of the closely related field of innovation diffusion. Beal, Havelock and Rogers offered additional insights into the origins of the field of knowledge utilization, termed by them "knowledge generation, exchange, and utilization" (KGEU) 39. Havelock argued that the parent discipline of KGEU was sociology, and acknowledged social and organizational psychology as important contributors. Rogers in this same volume clarified the importance of the agricultural extension model and its influence on the thinking of scholars in the field.

One of the most identifiable domains within a knowledge utilization framework, and until recently the most dominant, is diffusion of innovation. The history of the development of innovation diffusion as a research tradition is well-documented 103840. Rogers credited the Ryan and Gross classical agricultural study on hybrid corn as creating the template for classical diffusion theory for 40 years 41. Rogers 1042 identified nine diffusion research traditions: anthropology, early sociology, rural sociology (dominant until the 1960's), education, public health/medical sociology, communication, marketing, geography, general sociology, and a miscellaneous "other". Valente and Rogers used a Kuhnian framework for their analysis of the rise and fall of the diffusion paradigm among rural sociologists – arguing that the diffusion paradigm faded as a result of a paradigm shift. Although innovation diffusion theory is often described as Rogers' "Theory of Innovation Diffusion", it is more accurate to talk about Rogers' representation of innovation diffusion theory. Crane 40 and Valente and Rogers 38 show that the Ryan and Gross publication formulated the diffusion model. By the mid-1950s, a group of rural sociologists had filled in the major elements. Lionberger's 1960 "Adoption of New Ideas and Practices" 43 contains most of the elements of the diffusion model.

Technology transfer has a 60-year history of scholarship 44, with interest beginning primarily post World War II, and with periods of heightened interest in the Western world in response to events such as the Cold War, the development of the Space Age, and the emergence of economic competition in the 1970s 45. In Canada, for example, the role of technology transfer has been spearheaded by the Federal Partners in Technology Transfer, while in the United States a legislative approach has been adopted; these different approaches to technology transfer have subsequently affected each country's progress. For instance, post World War II Canada was slower than its American and British counterparts to establish technology transfer policies 45.

In 1992 a new group and a new style of knowledge utilization emerged, heralded by the publication of the influential paper "Evidence-Based Medicine: A New Approach to Teaching the Practice of Medicine" 46. This group of physicians declared a new way of doing medicine – one based on the explicit incorporation of empirical research findings into clinical decision-making processes. Their approach coincided, particularly in the United States, with increasing pressures to manage health care, in large part by reducing variation across both individual and group physician practices. They drew their lineage from the work of epidemiologist Archie Cochrane, who stressed the importance of evaluating medical interventions. Cochrane's work 47 had an important influence on the field of medicine and ultimately resulted in the establishment of the Cochrane Collaboration in 1993. Since the publication of the 1992 EBM manifesto, western society has witnessed a rapid emergence of numerous evidence-based centers, journals and resources.

Bibliometric analysis (bibliometrics) uses citation data and quantitative analysis to trace published literature and to study the patterns of publication within a field. In analyzing scholarly fields, investigators map structures over time using techniques such as co-citation, co-word, and author co-citation analyses [48, Chap 1]. In our work, we used ACA in the manner of White and McCain 49.

White and McCain argued that co-citation maps/citation analyses were powerful tools for mapping the intellectual structure of a field over time 5051. More recently, they reported longitudinal analyses of the structure and evolution of fields 4952. Small proposed that the cited documents are concept symbols 53. Normative sociologists, among them Zuckerman 54 and Merton 55, viewed citations as markers of intellectual influence and as reward and payment of intellectual debts, respectively. Constructivists Latour 56 and Callon 57 viewed citation as a way of "enrolling allies" to strengthen one's own knowledge claims.

Merton argued that citations denote scholarly influence 58, that they can be used as a measure of scholarly value; they serve the instrumental function of transmitting knowledge, and the symbolic function of rewarding scientists by recognizing their intellectual property rights 59. In short they are symbolic payment of intellectual debts 60. Alternatively, constructivists such as Latour 56 have argued that authors use citations to legitimate knowledge claims. By citing another's work, an author strengthens his or her own knowledge claim by tying it to those cited. The social process of making knowledge consists of the successful alignment of initially diverse claims, and if the network is strong enough, the author's knowledge claim becomes an obligatory passage point 57. Future authors wishing to make claims on the topic must go through this passage point (i.e., the author's work) by citing it. Consistent with Small 60, we argue that both normative and constructivist interpretations of citation patterns are valid.

In ACA, cited and co-cited authors are the unit of analysis 51. As White and Griffith point out, "Co-citation of authors results when someone cites any work by any author along with any work by any other author in a new document of his own" [61, p. 163]. Spatial maps are produced using one of a number of statistical techniques (e.g., cluster analysis, multi-dimensional scaling, factor analysis). Heavily co-cited authors appear grouped in space, with authors having many links occupying central locations on the maps and authors with weaker links (fewer co-citations) appearing on the periphery of maps 51. White and McCain argued that ACA simplifies literatures to "writings by use" providing "a more rigorous grouping principle than typical subject indexing, because it depends on repeated statements of connectedness by citers with subject expertise" [49, p. 329]. Several reports of ACA are available in the literature. White and Griffith covered seven years of the information science literature, finding identifiable author groups, which they call schools 61. They identified border authors who connect areas of research. White and colleagues recently argued that co-citations reflect intellectual structure more strongly than they reflect social structure 22.

One of the uses to which co-citation analysis is put is the identification of invisible colleges 6263 – groups of elite, interacting scientists who are geographically dispersed, but who exchange information to monitor progress in their field 406465. Invisible colleges are generally agreed to represent social networks or significant thought (i.e., cognitive) collectives within a field. The former are commonly studied with sociometric methods, the latter with bibliometric methods. The emergence or strengthening of an invisible college on one hand or the weakening or loss of one altogether on the other, signal important changes scientifically and intellectually – potentially serving to herald significant changes in the ongoing negotiations between science and society of their (sometimes uneasy) social contract. Author co-citation as a method maps intellectual structure, and does not provide direct evidence of social networks in a field.

In the study reported in this paper, we undertook a domain analysis 4952 using bibliometric methods, specifically ACA to trace historical development of the field of knowledge utilization between 1945 and 2004. Our specific objectives were to map the development over time of knowledge utilization as a scientific field, and to identify the intellectual structure of this scientific community.

We searched the Web of Science online database covering 1945 to October 2004 with combinations of keywords derived from concepts within the scope of the study (see Additional File 1 for the complete search strategy). Bibliographic information from 14,968 papers was downloaded. The goal of the search was for a balance between recall (exhaustivity) and precision (specificity). Recall is the number of relevant documents retrieved compared to the total relevant documents 66. Our recall was 88.7%, based upon how many of the possible 200 most cited documents were retrieved in our initial search. Precision is the number of relevant documents retrieved compared to the total documents retrieved 66. We addressed precision by reviewing all titles and screening for inclusion/exclusion based on pre-determined decision rules. All reviewer pairs had an inter-rater agreement of more than 80%, the first author reviewed the final exclusion decisions; 7,183 titles were excluded. More detailed methods are described in Additional File 2 and further additional information is available in the technical report on request.

We removed 336 duplicates and 3,099 titles that were not "articles" (from the document type field), as articles most often represent new scientific production in a field of study 6768. From the initial 14,968 titles, 5,278 articles were retained. Data files were cleaned prior to analysis by correcting for variance in author name, cited author name, cited documents, journal name, and country, and the data were categorized by decade.

Analyses were conducted for each decade starting with 1945. The data were analyzed using Bibexcel freeware, Excel, and Systat 4.0. Descriptive analyses – including most prolific countries, journals, cited authors, and cited documents – were completed by aggregating the data. For co-citation analysis, selection of authors was by frequency of citation. Selection of authors for co-citation analysis can be by a variety of means, such as personal knowledge, review articles, or directories 5163.

We produced maps for each decade using the twenty-five most cited authors. Twenty-five was chosen as a reasonable number of key authors to produce maps that were interpretable and not visually overwhelming. In one instance (1965 to 1974), 13 authors were chosen, as greater or less than 13 authors produced a map that was not readily interpretable. To create the author co-citation maps, co-citation matrices were first developed from raw citation co-occurrences using Bibexcel. The matrix of co-citation frequencies was entered into Systat 4.0, which uses a multidimensional scaling (MDS) algorithm to find the best-fitting two-dimensional representation of the matrix co-citation entries in the form of a visual map. We assessed the goodness-of-fit of each of the co-citation maps produced using Kruskal's Stress measure 51. Values for Kruskal's Stress 1 49 measure were 0.06, 0.16, 0.12, and 0.13 for each of the decades respectively; a stress value less than 0.2 is considered acceptable 51. We elected to present raw frequency maps because they were more interesting, with variation in the size of the nodes indicating frequency of citation. We reproduced our maps using Salton's cosine normalization 6970 and found no significant differences or changes to interpretation of the maps. The circles or nodes on the co-citation maps represent frequency of author citations; the lines joining the circles represent author co-citation 5171. Thicker lines and closer nodes indicate that the pair are co-cited more frequently, and therefore their work is considered to be conceptually similar 71. We demonstrated structural change over time by producing a separate map for each decade 72. The first map is for the decade of 1965 to 1974; prior to that there were insufficient authors to create meaningful maps.

White and McCain 49 define a canonical author as someone who appeared on the citation maps in three or more decades. We identify seven canonical authors whose work has enduring importance to the field and who were on at least the last three maps (1975 to 2004). We argue that the most cited works of these authors constitute the canonical literature of the science of knowledge utilization as it entered the twenty-first century. These canonical authors are: Everett Rogers and Gerald Zaltman (innovation diffusion), Carol Weiss and Ronald Havelock (knowledge utilization), and Edwin Mansfield, Thomas Allen, and Vijay Mahajan (technology transfer). Rogers, management scientist Allen and economist Mansfield are the only authors who were top-cited in all decades, excluding 1945 to 1954.

The intellectual structure of the field in all decades is dominated by the work of Everett Rogers. His theory has been the dominant and most consistently used theory since inception of diffusion research 125126, and is emblematic of the diffusion paradigm until the emergence of EBM in the last decade. We argue that Rogers is a canonical author 49, that his book "Diffusion of Innovations" is a canonical text for all the domains, and that his approach to innovation diffusion represents the dominant paradigm for conducting diffusion research. "Diffusion of Innovations", first published in 1962, went through five editions before his death in 2004 (1971, 1983, 1995, 2003) 107677127128. Each edition of his book was based on an analysis of all retrievable diffusion studies. The presence of such a dominant author stands in contrast to, for example, the information sciences field, characterized by the absence of a strong central author 49. It also stands in contrast to some of our own canonical authors such as Mansfield, whose early publications are the only work top-cited in all decades.

In addition to Everett Rogers' canonical status in this domain, Gerald Zaltman's most cited works were "Innovations and Organizations" 129 and "Strategies for Planned Change" 130. While Zaltman's work was strongly pro-innovation, it also reflected a belief that innovation should not be accepted unquestionably. He showed the importance of individual users of innovations to the diffusion process, and he proposed that to understand how innovation diffusion truly occurs we need to study demand characteristics, such as users' ability and willingness to seek and process innovations.

Carol Weiss was most cited for her works "Knowledge Creep and Decision Accretion" 35 and "Using Social Research in Public Policy Making" 131. She continues to publish 132133, and has expanded initial views of research use as an instrumental application of research to inform a decision, to include conceptual or enlightened use – when findings from research influence decision makers' attitudes to and perceptions of a social problem 134. She described several models of utilization: knowledge-driven, problem-solving, interactive, political, tactical, and research as part of the intellectual enterprise of society 34. An advocate of using research findings to inform public policy, she was among the first to examine the utilization of evaluation findings in improving program processes and program outcomes 135. Her articulation and extension of the concept "research utilization" was an important contribution to the field of knowledge utilization 34.

Ronald G. Havelock is recognized for his extensive work on knowledge use, change planning, and technology transfer. Author of "Planning for Innovation through the Dissemination and Utilization of Knowledge" 101, his work spanned many fields including education and medicine. Building on Rogers' work 76, Havelock developed a framework which aided in the understanding and improvement of the dissemination and utilization of knowledge in the social sciences. Guided by an extensive analysis of this literature from education and beyond, he proposed an often cited "linkage model" that connects researchers with end users in a two-way exchange of information that mutually enhances problems solving.

Edwin Mansfield's early work from the 1960s formed "concept symbols" for the domain 53 and exerted a continuing impact on the field 110111112. An influential economic analyst of technology, he explored the length of time required for firms to uptake decisions and products used by rival firms, and how this information is spread from one firm to the next. He also studied the proportion of new products and processes that are based on academic research, and the amount of time needed for such research findings to be incorporated into the commercial environment.

Among other contributions, Thomas J. Allen studied the influence of distance on information transfer and developed the Allen curve, which depicts the inverse relationship that exists between distance and the frequency of communication 86113136. Allen also studied the ways in which formal and informal associations within organizations contribute to the diffusion of knowledge. He identified "gatekeepers" as important individuals within organizations who bring new knowledge into their organization both by reading literature and by engaging with others outside of the organization. He found that new ideas are spread most commonly through informal mechanisms such as personal contact.

Vijay Mahajan wrote on developing knowledge in the areas of marketing strategies, product diffusion, and research methodology. He is responsible for adding the temporal element into a model originally designed by Blackman 137 to explain technological substitution. Mahajan has made important contributions in the area of product diffusion and has studied the diffusion process in developing countries, which he argues is an important but understudied area 138139. A more detailed version of the longitudinal findings is available in Additional File 3.

Our first major finding is that new domains within the field now generally referred to in the literature as knowledge utilization have emerged over time; in earlier generations, the term most widely used was innovation diffusion. We argue that although all of the domains we identify are concerned with the use of knowledge in some way, they change and take on distinct specializations and perspectives over time and continue to be strongly linked to innovation diffusion. This finding is in contrast to White and McCain's longitudinal ACA of the information science literature, where they found tremendous inertia, or lack of change over time 49. They argue that their maps could have looked different at the separate time points if there had been major changes in the field. Our maps do look different in each decade, reflecting continuing change and growth in the intellectual structure of the field. The 1965 map represents authors from a wide variety of academic disciplines whose common object of inquiry is conditions surrounding the use or application of scientific knowledge. In the 1975 to 1984 decade knowledge utilization and technology transfer emerge as distinct areas of study, and in the 1985 to 1994 decade EBM emerges. Over the decades new areas emerged, centered on the work of canonical authors who were already working in the field, before it divided into subfields.

The origins of this broad knowledge utilization field lie in the study of the diffusion of agricultural innovations in rural sociology, credited by Rogers as dating back to Ryan and Gross' hybrid corn study. In a modified co-citation analysis of the diffusion of innovation and technology transfer literature between 1966 and 1972, Cottrill, Rogers, and Mills found that the majority of the members of their diffusion of innovations cluster were from sociology 21. In Table 4, Wilkening's work on the diffusion of agricultural innovations is the most-cited work in the 1955 to 1964 decade. By the late 1960s, research on the diffusion of innovations in rural sociology had virtually died out, possibly because it solved the particular problem of producing and disseminating means by which high yield crops are produced.

A second important finding is that, over time, the initial core set of authors in diffusion research branched off to become the intellectual center of their new fields. In other words, new fields branch off from the original field of innovation diffusion, and at the core of each of these new fields (with the exception of EBM), we see one of the authors in the first or second decades. Rogers and Mansfield emerge from the first decade as central to innovation diffusion and technology transfer in later decades, respectively. Weiss, Caplan, and Havelock appear in the third decade and remain in each subsequent decade under the knowledge utilization domain. EBM first appears in the 1985 to 1994 decade with three authors, Eddy, Haynes, and Lomas. Lomas is a border author providing the primary connection between EBM and innovation diffusion (and to a lesser extent knowledge utilization) 61.

The social sciences have been characterized, most famously by Kuhn, as being "pre-paradigmatic," a state where no accepted set of principles or theories guide research in the area 140. Although sometimes disputed, Kuhn suggested that the social sciences were characterized by disagreement and lack of consensus, and argued that the natural sciences are characterized by long periods of normal science, where practitioners are guided by a single theoretical model, which aids them in solving puzzles that fit within the paradigm. In 1979 Small and Crane found evidence that the fields of economics, psychology, and sociology were developing in a manner more characteristic of the natural sciences 75. Kuhn further argued that natural science does not progress in a cumulative fashion, but instead is punctuated by revolutions that radically alter the theoretical rules that inform practice. Here, in the overarching field of knowledge utilization, we find no evidence of the fragmentation and allegiance to multiple-paradigms predicted by Kuhn 140.

Valente and Rogers claim that the pre-paradigm period in innovation diffusion was in the 1930s 38, and that the paradigm was set by Ryan and Gross 141. Although in 1983 Rogers claims the Ryan and Gross article as the top cited one in diffusion literature 127, our analysis found that that various editions of Rogers' books are by far the highest cited documents in the innovation diffusion literature. Our findings also provide strong evidence that this social science field is more like Kuhn's portrayal of the natural sciences during periods of normal science. There is a paradigm from the beginning – when Rogers publishes the first edition of his book. We believe that Rogers' work had such a significant impact because it was the first, and continued to be the only work, by virtue of its ongoing syntheses, that described (at least his representation of) a general theory of innovation diffusion. Further, Rogers' synthesis of most known empirical studies of diffusion remained useful to scholars.

If Kuhn's claim of the social sciences as pre-paradigmatic is correct, then in knowledge utilization we have an atypical social science field, growing in an atypical way. The maps show growth and specialization over time, not fragmentation as predicted by Kuhn. The knowledge produced about diffusion is taken up and used – it does not languish. People on this map share many assumptions, and their intellectual debt to the work of Everett Rogers shows in the persistent size of his node, and in the strong links to each newly emerged field, including EBM. Zuckerman argued that citations indicate intellectual influence 54. If so, no single person has been more influential in this field than Rogers; many if not most on this map share an intellectual debt to him. Rogers put forth a representation of a "generalized model of diffusion" [142, p. 16] in his first book 76 and "set forth common findings to date, arguing for a general diffusion model, and for more standardized ways of adopter categorization" [142, p. 16].

Although the study of diffusion in rural sociology became exhausted in the 1960s 38, Rogers argued [142, p. 19] that diffusion research was not dead or dying: "The number of diffusion publications completed per year continues to hold steady. Unlike most models of human behavior that begin to fade after some years of use, the diffusion model continues to attract strong interest from scholars." We also did not find a 'fading' of the diffusion paradigm. Our citation maps show that there is not a shift away from the diffusion paradigm; rather, there is a spread of the paradigm to other fields and areas of specialization. We do see, however, from the titles of the articles of the most-cited authors, a content shift away from the concerns of agriculture.

Everett Rogers, with the publication of his 1962 book, became an obligatory passage point 57. His continued publication of updated editions of that book 107677127128 ensured that he remained an obligatory passage point making it, we propose, nearly de rigueur to cite Rogers when writing in the knowledge utilization field. This may change if EBM continues its explosive growth; we observed that its proponents were less consistent in their citation of Rogers than those in other subfields.

Rogers is generally viewed as one, if not the most influential, social scientists of the last one hundred years. His book is the second most cited in the social sciences 143. Some reasons for this influence are obvious. He worked in several universities (among them Iowa State, Michigan State, University of Southern California, University of Michigan, University of New Mexico) and had many academic associates, among them some of the original diffusion scholars – Beal, Coleman, Gross, Ryan, and Wilkening. He worked on projects in many countries; he was invited to speak widely and often. He had many graduate students and colleagues and was known for his generosity and gift for bringing people and institutions together. He shared authorship and ideas widely.

Sociologist of science Knorr-Cetina argued that each of the sciences produces knowledge in a different way 144. For example, in molecular biology, she argued that scientist's biographies affect their epistemology – that through their careers their bodies become finely tuned measurement instruments, learning to perform delicate operations that cannot be taught. In much the same way, there is evidence that Rogers' personality and his way of doing research – his epistemology – was an embodiment of the same networking principles which he studied. His personal biography embodied his theorizing. After Rogers' death, four of his former students recalled their personal associations with him 145. Thomas Backer recalled that "Ev had a remarkable gift for bringing together people and institutions that otherwise didn't talk to each other much – he was the best example in the world of the kind of natural networker he studied in his research" [145, p. 291], and well known for his enthusiastic ability to "help connect people he thought should know each other, or who could work with him on a project" [143, p. 285]. James Dearing, argued that "Perhaps intuitively, Ev understood the social capital advantages of heterophilous relationships" [145, p. 294]. Thomas Valente wrote that after the first conference on Entertainment Education in 1989, Rogers invited everyone to his house for dinner. It became a "raucous celebration," and he goes on to note that " [e]veryone who was there marks it as the time when entertainment education became a cohesive field of scholarship" [145, p 297].

Rogers lived a long and productive life. He completed his doctoral work in 1957 at the age of 26, and remained active as a scholar for the next 47 years. He wrote 36 books and more than 350 refereed journal articles 145 including a new edition of his well known "Diffusion of Innovations" about every ten years (1962, 1975, 1983, 1995, 2003). His last published paper appeared posthumously in 2005 146. In it, Rogers reflected on his own unique place in the emergence of the diffusion model and on the model's origins in the literature review chapter in his dissertation. Of central importance is that he created a common language with which scholars could talk about diffusion – by emphasizing the term diffusion "rather than the plethora of terms that had been used for this concept" [142, p. 16]. He concluded that "... it seems there is indeed a general diffusion model" [142, p. 19].

From a science studies perspective, by studying every known diffusion study regardless of discipline, Rogers unknowingly enrolled thousands of allies to his cause 56. We do not claim that this was his intent, but by taking something from a wide range of fields (the empirical studies of diffusion), and by giving back something to every field (a general theory of diffusion and a common language), Rogers built a formidable knowledge claim. It is also important to note, however, that citation studies find that authors are cited for their usefulness and for their merit. The story could have unfolded differently if the scholars in the field had not found Rogers' representation of innovation diffusion theory to be of use in solving problems in their own areas. In other words, Rogers is not important just because he studied all known diffusion studies, but because he did this and produced a useful representation of diffusion theory; he got it right, so to speak. His representation of innovation diffusion theory has been shown to be stable and of use to a wide array of scholars in diverse fields. However, getting it right is no guarantee of success. The history of science is replete with people who got it right but who were not credited as such, or people that only time proved were right. Perhaps most famously, in the nineteenth century Pasteur's germ theory won out over Pouchet's theory of spontaneous generation – not because the evidence supported Pasteur, but because in part at least, the members of the French Académie des Sciences were biased in favor of Pasteur. At the time, an "evidence-based" decision would have supported Pouchet's theory 147. In addition to getting it right, success of the kind that Rogers and the theory of innovation diffusion have earned over the decades is a complex blend of crafting and winning credibility among peers 148, and having one's knowledge claim noticed and taken up by some relevant community 5674.

In this paper, we have presented a largely descriptive picture of the growth of the field known broadly as innovation diffusion. We have shown that until recently, research in this field has been informed largely by one theoretical paradigm, laid out in the work of innovation diffusion scholar Everett Rogers. We have shown how the field began with a core set of scholars from many disciplines with a common interest in innovation diffusion. We have shown that, over time, this core set of scholars formed the core of new, but related fields. We demonstrated that in the mid-1980s another field emerges (EBM). This field is linked intellectually particularly to Everett Rogers, primarily through the work of Jonathan Lomas, who is strongly linked to Haynes in EBM. The work of Eddy pulls more widely from the field of knowledge utilization, as well as from scholars in the parent domain of innovation diffusion. In the 1990s, we see the field of EBM growing and drawing from the fields of technology transfer, knowledge utilization, and innovation diffusion. Terminology, even in this subfield and within the health disciplines, is complex. For many groups working in the field that we have labeled with the cover term EBM, a broader cover term will be necessary. One sees for example terms such as evidence-based practice, evidence-based nursing, evidence-based decision-making, and evidence-informed decision-making used widely.

We argue that the rapid emergence of this new domain is possible because its adherents practice a form of knowledge production and scientific output that differs significantly from those in related fields. This production of outputs is characterized in particular by an emphasis on systematic reviews of the research literature. Its adherents tend to publish in journals with unusually high impact factors and wide dispersion. Their emphasis is arguably more vigorously focused on instrumental and normative ends (the use of clinical research to improve outcomes), and their emergence coincided with emerging foci on accountability and cost containment, and more recently foci on value for money and accountability for performance in health services. Although not yet apparent in this decade, it is likely that subsequent decades if mapped with a wide enough net will reveal an explosion of related fields within EBM, comprised at least of quality improvement and safety subsets. EBM adherents may represent a new epistemic culture of knowledge production 144.

We propose that it is the EBM Group's emphasis on systematic reviews and their active dissemination and transformation, coupled with a highly normative and prescriptive orientation, that creates the conditions for this new form of knowledge production. Rogers, working without the aid of computer databases (for much of his early career at least) also synthesized, pouring laboriously through all known diffusion studies. In this regard there are similarities. Rogers' goal, however, was a synthesis and representation of a general theory of innovation diffusion. EBM's goal is prescriptive while being rigorously empirical – to guide and inform medical practice, working from a model in which the 'gold standard' in medical knowledge has been defined. Rogers sought to understand the process of how new innovations were diffused. EBM may be creating by example a new way of diffusing innovations. Its members have used the innovation diffusion literature and are linked in many ways to this literature. But this epidemiology based group has not as yet evidenced an intent (nor we argue should they necessarily) to build theory about knowledge utilization, technology transfer, or innovation diffusion. Rather, it is highly prescriptive and characterized by a strong underlying assumption that practitioners of EBM have (or can get) the best knowledge and the best knowledge production model. They are linked to the original problematic of innovation diffusion through the age-old problem of the know-do gap 149.