Research and the cumulation of knowledge in physical therapy.

Research is increasingly emphasised within physical therapy. Calls are made in physical therapy journals for more research into physical therapy and for that research to be published.[1-3] More research is being published. Physical therapy journals first published prior to the 1950s have generally increased in size and, in some cases, frequency of publication. New physical therapy journals have been introduced.

Research is expected to assist in the development of physical therapy. Specifically, research is expected to contribute to the underlying and unique knowledge base of the profession and to demonstrating the effectiveness and scientific merit of the knowledge base.[4] Through these contributions, research is expected to assist in developing a more distinct identity for physical therapy. (Physical therapy is described in this article as a profession[5] and as a discipline, as it may be practiced full-time as a livelihood, requires the application of specialized knowledge to practical problems, requires at least a university degree or equivalent, and can support a doctoral program.)

Questions about physical therapy persist in the physical therapy literature. This study identified four categories of questions and concerns about the discipline in the physical therapy literature of the past two decades. First, questions continue to be asked in the discipline about what physical therapy is, what physical therapists do, and what core knowledge physical therapists should all have.[6-8] The extent of concern in physical therapy with the nature of the discipline is encapsulated in the choice of topic for an essay competition run in the journal Physiotherapy in 1988: "Are physiotherapists necessary?" Second, concern exists in physical therapy with the lack of agreed-on terms and concepts and with the apparent absence of a clinical literature in the discipline.[9-16] For example, the current editor of the journal Physical therapy has claimed that physical therapy lacks a body of clinical literature[13,14] or of scientific literature relevant to practice.[16] Third, concerns about the lack of development of relevant theory within physical therapy[17] are such that importing knowledge from other disciplines has been explicitly advocated.[18,19] Fourth, concerns exist about the extent of the reliance physical therapy has on medicine and the behavioral sciences for its knowledge base.[20] To summarize, these different categories of concerns together question whether physical therapy has a unique and identifiable knowledge base.

A unique knowledge base is a distinguishing feature of a discipline.[17,20] A knowledge base includes the core of knowledge relevant to the generally agreed-on subject matter of the discipline and to the accepted methods of investigating shared issues of concern.[21] The knowledge base of a discipline is articulated in formal communications, mostly the literature of a discipline. The literature embodies not only the conceptual knowledge base of the discipline but also the concepts and knowledge required for the performance of professional roles.[5]

The formally communicated or explicit knowledge embodied in a literature contrasts with the implicit and practice- or craft-based knowledge of a discipline. Formally communicated knowledge is usually refereed and is available for further general scrutiny and comment, and perhaps eventual integration into the discipline's body of knowledge. By contrast, informal communications or implicit knowledge are not generally monitored or filtered; may be vague, colloquial, and nonintegrated[22]; and often include incomplete work. Informal communications include oral conference presentations and personal contacts.[22-24] The extent of the reliance in physical therapy on informal communications remains unknown. A study of the content of the Australian Journal of Occupational Therapy found evidence of a high level of reliance on informal communications within the discipline of occupational therapy.[25] Anecdotally, similar situation prevails in physical therapy.

The cumulation of reliable knowledge in a discipline or research area is closely related to growth in the relevant literature. Individual published articles become part of the archives of an area, If cited and incorporated into reviews, individual articles then contribute to the scientific knowledge and understanding of an area.[22] Scientific knowledge is created through the development of a rational consensus on ideas and information.[26] That rational consensus is developed by publicly presenting research findings in one or more agreed-on forms for scrutiny and criticism. Publication alone, however, is insufficient to develop and cumulate reliable knowledge.[26] The cumulation of knowledge also depends on inbuilt methods of self-correction. Replications of existing studies can provide self-corrections and enable the gradual production of a relatively errorless scientific literature and body of knowledge.[27]

The cumulation of knowledge also requires a cohesive research program. Many separate but thematically related studies are required to successfully and systematically develop a cohesive research program. If cohesive, the result should be an interconnected network of thematically related research articles in the relevant literature. A literature with a higher thematic base can more effectively produce research directions for its users, as integrating research findings places them in a context in the relevant literature.[27] For example, a cohesive clinical research program involving a number of different researchers might be expected to result in the cumulation of knowledge about particular treatment interventions and relevant measures of change.[28]

A cohesive research style contrasts with an "independent style."[27] An "independent style" is not literature based and results in a series of experiments conducted outside the context of existing research. The net effect may be a thematically related body of literature but one in which contributing articles are not linked through either citations or concepts. A dominating independent style, with its limited generality and lack of research direction and cohesion, effectively reduces the extent of possible development of the relevant literature, Irrespective of whether the literature is predominantly clinical or nonclinical, a lack of cohesion is consistent with a lack of direction in the relevant literature.[27] By contrast, a cumulating literature is consistent with a cohesive research style and can benefit its users through its greater generality, cohesion, and research direction.

Some questions are raised by the coexistence of growth in the size of the physical therapy literature, the broad-based and ongoing concerns with the knowledge base of the profession, and the apparent absence of a clinical literature in a clinical discipline. First, to what extent is there evidence of cumulation within the physical therapy literature? Second, to what extent are thematic samples of the physical therapy literature cohesive? Third, to what extent is the physical therapy literature dependent on non-physical therapy sources?

Questions about the cumulation of knowledge in a discipline can be investigated by a quantitative analysis of the relevant literature. in particular, citation analysis provides a method for investigating the existence and extensiveness of conceptual networks and changes in the conceptual linkages among articles, the impact of individual articles on a literature, and the divergences and convergences among different fields of knowledge.[29,30] A study of citations in scientific articles showed an average of 15 citations per article; 12 of these citations referred to journal articles, and the remaining 3 citations referred to books or other Sources.[31] The same study also revealed that 49% of the articles were cited once, 16% were cited more than once, and 35% were not cited. Professional literatures are possibly different. Some evidence suggests that professional literatures cite more older materials and that they are more insular as they draw on more materials from their own fields.[31]

Citation networks and linkages develop as new articles cite existing articles. More heavily cited articles have more linkages to other articles than do uncited or self-cited articles. Self-cited articles contribute to the body of knowledge when used by researchers other than the authors of the articles. Conceptual links gradually extend as topics or thematic areas are systematically researched and some cited studies are gradually integrated into the relevant body of knowledge. Not all linkages will represent an affirming or even an accurate use of a cited article.[32,33] Linkages will also not indicate the relative merits of any contributions. An analysis of the relative contributions of articles to the conceptual development of an area requires content analysis. By contrast, citation analysis provides a method of evaluating the degree of development of a literature or the impact of different events on a body of literature.

Few analyses of the citations used in the physical therapy literature exist. Existing citation analyses identified the core journals in the discipline,[34-36] the journals most frequently used by contributors to Physical Therapy,[37] and the relations between citation frequency and professional reputation in parts of the physical therapy literature.[38] No existing studies have investigated the extent of cumulation or of cohesion within topics or thematic areas discussed within the physical therapy literature.

The principal aim of my study was to redress the absence of studies of the extent of cumulation and of cohesion within thematic samples of the physical therapy literature. There were three specific aims. The first aim was to investigate the extent to which thematic samples of the physical therapy literature show signs of cumulation. The second aim was to investigate the extent to which there is evidence of cohesion within the physical therapy literature. The third aim was to investigate the extent to which the physical therapy literature utilizes non-physical therapy sources.

Method

Sample

The sample comprised all 244 articles on the knee (n=102), the back (n=72), and electrical stimulation (n=70) published and indexed in Physical therapy between 1951 and 1990. The knee and the back were the two most frequently represented practice-oriented topics in that period in this journal, and electrical stimulation was the most frequently represented treatment modality. Physical Therapy was selected as the source journal as recent studies of the physical therapy literature showed it was the most frequently cited of four major physical therapy journals,[35] the most frequently cited journal by contributors to Physical Therapy,[37] and the most frequently cited physical therapy journal in the Physiotherapy Current Awareness Topic Search (CATS)[34] and in the Physiotherapy Index.[36] Further, Physical Therapy is indexed outside the professional literature, in Index Medicus, as well as within.

A "clinical treatment" subset of 78 of the 244 articles was identified. Clinical (treatment) articles reported the outcomes of the implementation of specified treatments applied to individual patients or groups of patients with the intention of maintaining or improving the function of those receiving the treatment. Of the 78 clinical articles, 25 were on the knee, 23 were on the back, and 30 were on electrical stimulation. The remaining 166 articles (77 on the knee, 50 on the back, and 39 on electrical stimulation) were designated "nonclinical" articles,

Procedure

Identity numbers were allocated to each of the 244 articles and recorded together with bibliographic details and the topic category. Details of all citations in each of the 78 clinical articles, irrespective of whether to a book, a journal, or some other type of item, were added, and the identity of the citing article was recorded. When cited articles were multiauthored, the name of only the first author was recorded. Otherwise, all details were recorded as provided in the citing clinical article.

The accuracy of the recorded details of all citations in each list was checked by the author and corrected where necessary. Errors among the published citations included incorrect page numbers and incorrect spelling of the names of some first authors. These errors were resolved by checking source material where possible or other citations of the same sources.

The relations between citations and articles in each sample were investigated in three stages. Stage 1 identified the following in each thematic set of sample articles and citations: the number of citations per topic, the numbers of unique items and different first authors, the sources of citations, and the journals cited most frequently.

Stage 2 excluded all items listed only once and all self-citations. Self-citations were defined as citations by the first author of an earlier article he or she first-authored. Only multiple-listed articles or books remained in each topic category, all films and operating manuals having been listed only once. The remaining multiple-listed items represented clinical articles and their citations, nonclinical articles and their citations, or nonsample articles cited in at least two clinical articles. The numbers of cited clinical, nonclinical, and nonsample articles, and the number of citations each received, were counted.

Stage 3 analyzed the relations between and among the multiple-cited articles, the clinical and nonclinical sample articles, and the nonsample articles. Specifically, any interconnections or linkages evident between cited and citing articles were sought and the period between publication and citation was established. Evidence of linkages included any co-citations by clinical articles of the same articles or any chains of citations in which a clinical article cited an existing clinical article and was itself subsequently cited by a clinical article.

Results

Table I shows the numbers and basic characteristics of items cited in clinical articles in each topic (knee, back, and electrical stimulation). The average number of citations in the 78 clinical articles was 11.3 per article. The table shows that the majority of items in each topic area were listed only once (unique articles) and most first authors in each topic area contributed only one item.

[TABULAR DATA OMITTED]

Table 1 also shows the number of different journals and other sources cited in the clinical articles. These sources comprised 50 different journals and 36 other sources for the knee, 60 different journals and 47 other sources for the back, and 122 different journals and 74 other sources for electrical stimulation. The other sources included a small number of films and operating manuals, but most were books. In all three areas, most books were cited only once. The last column in Table 1 shows the number of self-citations, or citations of articles previously published by the same author.

Table 2 lists the 10 most frequently cited journals in each topic area. The same 3 journals - Physical Therapy, Archives of Physical Medicine and Rehabilitation, and the Journal of Bone and Joint Surgery (American and British volumes combined) - were among the 4 most commonly cited journals in each thematic area. Of the 3 physical therapy journals listed, only Physical Therapy was included in more than one topic and cited more than seven times in any topic. The other journals listed were predominantly medical and concerned with orthopedics, rehabilitation, surgery, sports medicine, or pain, or were journals of sports science or physiology.

[TABULAR DATA OMITTED]

Table 3 shows the number of clinical, nonclinical, and nonsample articles cited in clinical articles in each thematic area and the frequency of citations received. The table shows that 9 (11.5%) of the 78 clinical articles received a total of 11 (1.25%) of the 878 citations in the clinical articles. One clinical article on the knee and 4 clinical articles on the back were each cited once in other clinical articles indexed on the same topic. Four clinical articles on electrical stimulation were cited in 6 other clinical articles on electrical stimulation, 2 twice each and 2 once each. Clinical articles were cited by other clinical articles from 1 to 32 years after publication (average=11 years, SD = 12 years). The two most recent clinical articles cited were published in 1988, and the earliest clinical article, published in 1951, was cited in a 1983 article.

[TABULAR DATA OMITTED]

Table 3 shows higher citation rates of nonclinical and nonsample articles than of clinical articles in all three topics. Of the 20 nonclinical articles cited, 15 articles were each cited once. Of the 5 multiple-cited nonclinical articles, 1 article on electrical stimulation received 4 citations, and 2 articles on the knee and 2 articles on electrical stimulation received 3 citations each. Nonclinical articles were cited in clinical articles from 1 to 13 years after publication (average=4.9 years, SD=3.2 years). The nonclinical articles each received an average of 1.55 citations, higher than the average 1.22 citations of clinical articles, but lower than the average of 2.27 times for nonsample articles.

Figure 1 illustrates the extent of the linkages found between cited clinical articles on electrical stimulation and the citing clinical articles. The citing clinical articles are shown on the left side of the figure, and the cited clinical articles are shown on the right side. (Clinical articles on the knee and the back and their relations with cited items were not illustrated because fewer clinical articles were cited for those topics and, unlike the articles on electrical stimulation, neither set exhibited any second-stage linkages.) identification numbers were used for each citing article (eg, S55, S31, S22), and the publication date for each article is shown under its identification number (eg, 1988 under S55). The articles identified as S55, S31, S14, and S16 cited a combined total of 61 articles. Two of those 61 articles were clinical articles and are shown in Figure 1 ("Baker" and "Burke"). Citing S22 "links" articles S55 and S31, similarly, through citing S13, articles S14 and S16 are "linked." Two other clinical articles ("Thurman" and "Weinstein") were among the 42 articles cited in clinical articles S54 and S26. None of the 45 articles cited by "Baker," "Burke," "Thurman," or "Weinstein," however, included any from among the clinical and nonclinical articles on electrical stimulation indexed in Physical Therapy; thus, no additional linkages to other articles were found.

Figure 2 illustrates the 34 linkages found between the nonsample articles cited three or more times and the citing articles on electrical stimulation. Linkages for the articles on electrical stimulation, rather than for the articles on the knee or the back, are shown in Figure 2 because of the greater prevalence of multiple-cited articles on that topic. Of the 72 sets of multiple-cited nonsample articles identified, all 4 articles cited five times and both articles cited four times dealt with electrical stimulation, as did 2 of the 4 nonsample articles cited three times and 29 of the 62 articles cited two times.

The first authors of the eight nonsample articles or books cited three or more times in the clinical articles on electrical stimulation are listed on the right side of Figure 2. The eight cited items comprised one book; one conference proceedings; and six orthopedic, rehabilitation, and physiology journals. The citing clinical articles on electrical stimulation, indicated by an identification number (eg, S34, S22), are shown on the left side of the figure. Figure 2 shows, for example, that S55 (published in 1988) cited "Gould" (1982), "Benton" (1981), "Bowman" (1979), "Munsatt" (1976), "Merletti" (1975), and "Waters" (1975). "Gould" (1982), a nonsample article or book, was also cited by S49, a 1987 clinical article on electrical stimulation.

Figure 2 provides some examples of co-citations. Most derive from the one first author who. contributed three clinical articles on electrical stimulation (S22, S26, and S38). Together, these three clinical articles contributed 10 of the 34 linkages illustrated in Figure 2 by citing four of the same articles twice and two other articles once each. Another example of co-citation is the citation by articles S34 (1984) and S22 (1979) of both "Munsatt" (1976) and "Waters" (1975).

Figures 1 and 2 together provide the two examples of a two-stage linkage of citations found in this study. Figure 1 shows that S55 and S31 both cited the clinical article "Baker" (S22). In turn, Figure 2 shows that "Baker" (S22) cited five other articles that were also cited by other clinical articles. Similarly, "Burke" (S13) cited two nonsample articles ("Liberson" and "Fischer") and was in turn cited by S13 (1963) and S14 (1964).

Discussion

This study revealed few linkages among clinical articles in the same thematic areas. The linkages identified were simple and had a maximum of two stages. This study also showed most cited items were unique and were from the medical, sports science, and physiology literatures.

The findings of this study suggest meager cumulation has occurred in the three thematic-based samples of physical therapy literature examined. Few articles in all three thematic areas cited papers from the same area, or even articles from the physical therapy literature. That is, few articles apparently used or extended existing findings published within the relevant areas of the literature of the discipline.

A possible criticism of the finding of a lack of cumulation is that articles published in 1989 and 1990 had few opportunities to be cited by others in the sample. Consistent with this conclusion, the most recently cited clinical articles were 2 published in 1988. The lack of opportunities available for the 1989 and 1990 articles to be cited, however, does not explain the overall lack of evidence of cumulation in the thematic samples examined. The later articles were able to cite and thus build on the content of existing clinical articles. Articles published successively earlier had successively more opportunities to be included among the 878 citations provided in the 78 clinical articles examined. Given the number of these opportunities, the findings that only 2 of the 78 clinical articles were cited twice and that only 7 articles were cited once suggest a problem related more to the literature examined than to limitations in the method of analysis. Perhaps each thematic area contains such a diversity of content and of opinions and approaches to any aspects of that content, and therefore little evidence of cumulation could reasonably have been expected. This possibility was not examined in this study but is consistent with the lack of evidence of cumulation found in the three thematic samples of clinical literature examined.

Physical therapy is a clinical discipline. The thematic areas examined comprised the two most frequently represented practice-oriented topics and the most frequently represented treatment modality topic published and indexed in Physical Therapy between 1951 and 1990. The citations examined were from the clinical subset of these articles in which the implementation and outcomes of treatments for patients were expressly investigated. The findings of this study suggest that concerns with the lack of a body of clinical literature expressed in editorials in Physical Therapy[13-15] were justified. Little evidence of cumulation was found in the three significant practice areas examined. That is, even though the number of articles had increased, little evidence of progressive building or cumulation of a body of knowledge was apparent within the thematic areas investigated. If these findings can subsequently be shown to apply to more recent articles in Physical Therapy, to other journals in the discipline, and to other thematic samples, then the physical therapy profession has reasons to be concerned about the identity and development of its knowledge base.

Reliable knowledge can cumulate only when made explicit. Perhaps the lack of cumulation found in this study was because little physical therapy knowledge is made explicit. Calls for more research are expected to provide evidence of the effectiveness of physical therapy practice and of its scientific merit. Calls for more research might also indicate an awareness in physical therapy that relatively few details of current practice are fully explicated and available for general scrutiny. This study did not attempt to investigate the extent to which knowledge in the discipline is explicit. Consequently, this report can only suggest that possibly much practice knowledge is yet to be made explicit and reiterate that explication and examination of otherwise implicit knowledge are required for cumulation and could have the effect of strengthening the identity of physical therapy.

This study demonstrated little evidence of cohesion within the thematic areas examined. Most citations within each area were to unique items, and few linkages were identified among the articles. This pattern typifies an "independent" research style, characterized by a lack of research direction in the literature and a lack of generality. Because the findings of this study pertain to the clinical areas of most interest in the discipline, they should concern those practitioners interested in research and in the role it must play in the development of the identity and knowledge base of physical therapy. More cohesive research programs on topics identified as important in physical therapy are required. More of these topics, as previously argued by Rothstein,[16,39] must address issues concerned with clinical practice. Perhaps more topic-oriented issues of journals or even more specialist journals are required. Whatever the course of future developments, the findings of this study suggest that more extensive reflection is needed within physical therapy on what the core knowledge of the discipline is and on how it can best be developed.

The degree of reliance of physical therapy researchers on non-physical therapy sources raises questions about the likely future cumulation of knowledge in the discipline. Clearly, there are very many shared areas of interest between physical therapy, medicine, sports science, and physiology. Physical therapists, however, see their profession as having or wishing to develop its own profession-identified knowledge base and have expressed concern in the physical therapy literature about the lack of development of a discipline-specific clinical literature. The extent of reliance on the medical, sports science, and physiology literatures identified in this study suggests that concern about whether physical therapy has a unique knowledge base of its own were well founded. Little evidence of cumulation of the physical therapy clinical literature in the thematic areas examined coexisted with a reliance on the literatures of medicine, sports science, and physiology. The question of why the physical therapy literature examined is so heavily reliant on non-physical therapy sources remains, as does a full exploration of the significance of these findings.

Calls in physical therapy for more knowledge to be imported from other professions appear misplaced. Contrary to findings of insularity within professional literatures generally, the extent of use made of other literatures in the physical therapy literature examined appears high. Perhaps health science literatures differ from other professional literatures in significant respects, or perhaps these findings indicate physical therapy is at an early stage of development. Alternatively, perhaps the findings of this study pertain only to the thematic areas of the physical therapy literature examined. Even so, these findings are significant because they concern the three clinical areas of prime importance in physical therapy and in the major journal of the discipline.

Further studies of the physical therapy literature are needed to validate and to extend the findings of this study. This study focused on the citations in the clinical subset of three thematic areas published in Physical Therapy between 1951 and 1990. A subsequent study could undertake a content analysis of the same thematic areas to establish whether the apparent lack of evidence of cumulation is due to the range of content covered in any one thematic area. Alternatively, this study could be extended by, for example, a citation analysis of the nonclinical articles from the same thematic areas. Further citation or content analyses of other thematic areas in the same journals and of the same and additional thematic areas from different physical therapy journals would add considerably to our knowledge about physical therapy and the development of its knowledge base. In turn, such knowledge would enable more accurate reflections on the stage of development already reached in physical therapy and a more rational basis for developing future plans.

A possible criticism of this study concerns the selection of Physical Therapy as the sole source of material relating to physical therapy. Evidence to date, however, suggests that Physical Therapy is the most significant journal within the discipline, and it remains one of only two physical therapy journals indexed in the discipline's indexes and in Index Medicus. The findings of this study also confirmed the prominence of Physical Therapy, in the discipline. Of the 32 most frequently cited journals identified in this study, 3 (including Physical Therapy) were physical therapy journals. Only Physical Therapy, however, was represented in more than one thematic area. In part, this is possibly a form of self-interest, as authors intending to contribute to a journal may cite articles from that journal to impress an editor and reviewers. In practice, though, existing studies have repeatedly shown Physical Therapy to be the leading journal in the discipline.

Another possible criticism of this study concerns the identification of articles by first authors only. Some teams of researchers might alternate which author is listed first. If so, the findings of this study may have underestimated the extent of cohesion in the thematic areas examined. Because very few clinical articles cited any other clinical or even nonclinical articles, however, any additional evidence of cohesion can only be minimal.

Another possible criticism might suggest that citation analysis of the physical therapy literature is inappropriate because citations are used differently within the thematic samples examined or within that literature. Earlier studies of citations in the physical therapy literature, however, indicated a distribution of citations to journals (77% and 71.5%[37]) similar to that (78.7%) found in this study. Further, the average number of citations per article in this study (11.3) is less, as might be expected of a professional literature, than the 15 citations per article found in an early study of scientific literature.[31] The significance of the time found between publication and citation cannot be evaluated due to the absence of information on both the physical therapy literature and related professional literature In the absence of more extensive information about health profession literature there is no reason to believe physical therapy literature has different citation patterns than do others and thus no reason to suggest citation analysis of the physical therapy literature is inappropriate.

Future Directions

Further studies of other topics and parts of the physical therapy literature will demonstrate the extent of generalizability of the findings of this study. The findings, however, are generally consistent with concerns raised in physical therapy about the lack of a clinical literature, of an agreed-on set of terms and concepts, of a body of physical therapy theory, and of an identifiable core body of knowledge in the profession. The findings are also generally consistent with concerns raised about the extent of the reliance on medicine and the behavioral sciences for the profession's knowledge base. Given these apparent consistencies, the following section will assume the generalizability of the findings and consider possible future directions for physical therapy.

What might the profession do to facilitate the cumulation of knowledge, particularly as it relates to practice? One possible way of answering this question already exists. The current stage of development in physical therapy is similar to that reported in developing scientific disciplines.[40] The experiences of a related profession that has passed through this stage suggest an option for physical therapy.

Psychology, like physical therapy, is concerned with practice and with researching that practice. Serious concerns in psychology during the 1940s and 1950s about the apparent lack of development of a scientific rationale for much practice, similar to those reported in the physical therapy literature, led to the development of a "scientist-practitioner" model. Initial expectations were not unlike those held, at least implicitly, in physical therapy. Teaching clinical psychologists research skills and methods during their graduate training was expected to lead to them to becoming scientists and practitioners.[41] Scientist-practitioners were expected to be expert clinicians who also conducted the research required to validate their practice. Subsequent experience shows that this did not occur.[42] One contributor to the psychology literature suggests that the failure was in part due to the move away from, and neglect of, the craft base of the discipline.[43] Consistent with this suggestion, the same contributor implies a "practitioner-scientist" model would be more constructive and more consistent with the history of other experimental sciences.

The adoption of a practitioner-scientist model might provide physical therapy with the focal point required for future development. Such an adoption would require the recognition of the centrality of practice in physical therapy and of the extent of the well-developed, but largely implicit, practice- or craft-based knowledge. Craft-based knowledge is developed by experienced practitioners, often over years, of what works and under what conditions. Some of this knowledge is undoubtedly passed on to students during their training, and to other therapists in the same departments and at workshops and seminars. As this method is primarily informal, however, the process is a fairly haphazard way for a profession of cumulating and transmitting knowledge. Further, it does not lead to effective scrutiny by others of that knowledge, which in turn cannot be integrated into the general body of knowledge that underpins and defines the profession.

Explicating craft-based knowledge is clearly not easy. The increasing focus on scientific research in physical therapy has seemingly produced few results in the clinical domain. For example, little evidence of cumulation of knowledge was found in the practice-oriented topics examined in this study. Changing the focus in physical therapy from research to clinical practice, however, provides a possible solution. This is definitely not an argument against doing research. It is, however, an argument against continuing on a track in which the outcomes so far appear quite limited. It is also an argument for approaching research in a manner expressly calculated to lead to the outcomes required and expected in a clinical profession, Furthermore, it is an argument for a full analysis, and some reconsideration, in physical therapy of how we expend our research time and effort. One aspect of this issue, the "costs" of entry-level student projects to the profession, has been addressed by the editor of Physical Therapy.[39] Other aspects have yet to be seriously addressed within the profession.

A focus on practice implies a greater emphasis on developing reliable methods for measuring, observing, and recording the changes of interest within the clinical context. This is happening to some extent, as is some questioning of which methods can best answer which questions of what works and when and how. Not all research methods currently favored or discussed within the profession are optimal for answering practice-oriented questions. Some methods require large numbers of homogeneous patients, average results across subjects, and can introduce ethical problems by requiring treatment to be withheld. Other methods are not generally accepted as scientific and yet may provide a basis for cumulating clinical knowledge. Some methods ignore the need to establish the reliability of findings and the extent to which the findings can be generalized. A change in focus for physical therapy from the implicitly held scientist-practitioner model to that of practitioner-scientist would facilitate an active reconsideration of research methods and of measurement, observation, and recording issues.

Current physical therapy editors generally recognize the need for more "research articles that provide relevant information for clinicians."[13](p796) Most practicing physical therapists no doubt endorse such calls. Perhaps the success rate of calls for more clinically relevant articles might increase if journals dedicated a set percentage of space per issue to articles explicitly discussing or surveying methods used to treat patients with commonly seen problems. Initially, these articles might have to be invited. Topics could include treatments of simple fractures, of acute whiplash, or of balance disorders, following strokes. The focus would be on practice and on material already published in the physical therapy literature. Clearly, the absence of materials with more than anecdotal Support on a number of central topics would raise questions for many physical therapists. Publishing these questions would form part of the process of making explicit what we do and do not know. Answering the questions would be the province of practitioners, at least in the first instance. Developing cohesive research programs with a range of contributors would greatly facilitate this process. Cumulating increasingly reliable and clinically relevant knowledge would, in time, be an inevitable outcome.

Cumulating reliable practice-oriented knowledge is central to the future development and success of physical therapy. This report has used the findings from an analysis of a small sample of articles, the content of discussions in the physical therapy literature, and the experiences of psychology to suggest a way of focusing future developments in physical therapy. The adoption of a practitioner-scientist model would affect all physical therapists, whether practicing therapists, academicians, administrators, journal editors, researchers, observers of the profession, or students. At the simplest level, changes in emphases within the profession would be required and more physical therapists would need to recognize the importance of researching practice and of the many ways available. Unless we make clinical knowledge explicit and seriously investigate which clinical practices are effective and under what conditions, the physical therapy knowledge base will remain, as at present, largely derivative and with often little direct relevance to clinical practice.

Conclusion

The findings of this study support concerns within physical therapy about the knowledge base of the profession. The findings show little evidence of cumulation of knowledge or of cohesion within the clinical literature examined. The findings also show a high degree of reliance on non-physical therapy literature sources. Within physical therapy, there is currently a strong emphasis on research and the publication of research. The findings of this study suggest a greater emphasis should also be placed on ensuring the publication of the implicit or craft knowledge of the profession and on developing cohesive research programs.

References

[1] Cooper BA, Saarinen-Rahikka HK. Interrelationships of theory, clinical models and research. Physiotherapy Canada, 1986;38:97-100. [2] Physiotherapy research: a new development. Physiotherapy. 1990;76:17. [3] Riddoch J, Lennon S. Evaluation of practice: the single case study approach. Physiotherapy Theory & Practice. 199 1;7:3-11. [4] Atkinson HW. Heads in the clouds, feet on the ground. Physiotherapy. 1988;74:542-547. [5] Pierce SJ. Characteristics of professional knowledge structures: some theoretical implications of citation studies. Library Information Science and Research. 1987;9:143-171. [6] Gemmell J. Are physiotherapists necessary? Physiotherapy. 1988;74:497-498. [7] Rose SJ. Editor's note: Our body of knowledge revisited. Phys Ther. 1989;69:297-298. [8] Rothstein JM. Pathokinesiology: a name for our times? Phys Ther. 1986;66:364-365. [9] Jensen GM. The work of accreditation onsite evaluators. Phys Ther. 1988;68:1517-1525. [10] Lister MJ. Editor's note: Farewell. Phys Ther. 1988;68:183-184. [11] Parry A, Stone S. Capturing the basics: the development of an expert computer system for physiotherapists. Physsiotherapy 1991;77:222-226. [12] Rose SJ. Description and classification: the cornerstones of pathokinesiological research. Phys Ther. 1986;66:379-381. [13] Rothstein JM. Editor's note: The Journal-past, present, and future. Phys Ther. 1989;69:795-796. [14] Rothstein JM. Editor's note: Clinical literature. Phys Ther. 1989;69:895-896. [15] Rothstein JM. "Science" isn't a dirty word. Clinical Managentent. 1989;9(4):12-13. [16] Rothstein JM. Editor's note: Physical therapy research - academic inadequacies. Phys Ther. 1992;72:92-94. [17] Tammivaara J, Shepard KF. Theory: the guide to clinical practice and research. Phys Ther. 1990;70:578-582. [18] Krebs DE, Harris SR. Elements of theory presentations in Physical Therapy. Phys Ther. 1988;68:690-693 [19] Winstein CJ, Knecht HG. Movement science and its relevance to physical therapy. Phys Ther. 1990;70:759-762. [20] Morrissey MC, Kanda LT, Brewster CE. Development of a clinical physical therapy research Phys Ther. 1987;67:1110-1114. [21] Arsenault AB, Cleather J. A perspective on Physiotherapy Canada: results of a 1981 readership survey. Phsiotherapy Canada 1981;34:28-36. [22] Garvey WD, Griffith BC. Scientific communication: its role in the conduct of research and creation of knowledge. Am Psychol 1971;26:349-362. [23] Meadows AJ. Communication in Science. London, England: Butterworth & Co (Publishers) Ltd; 1974. [24] Ziman JM. Public Knowledge. Cambridge., England: Univesity Press; 1968. [25] Madill H, Brintnell S, Stewin L. Professional literature: one view of a national perspective. Australian Journal of Occupational Therapy. 1989;36:110-119. [26] Ziman JM. Information, communication, knowledge. Nature. 1969;224:318-324. [27] Johnston JM, Pennypacker HS. Strategies and Tactics of Human Behavioral Research. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates Inc; 1993:43. [28] Barlow DH. On the relation of clinical research to clinical practice: current issues, new directions. J Consult Clin Psychol 1981;2: 147-155. [29] Garfield E, Kimberley R, Pendlebury DA. Mapping the social sciences: the contribution of technology to information retrieval. Current Contents. 1988;20(46):4-9. [30] Ziman JM. An Introduction to Science Studies. Cambridge, England: University Press; 1984. [31] Price DJD. Networks of scientific papers. Science. 1965;149:510-515. [32] Smith LC. Citation analysis. Library Trends. 1981;30:83-106. [33] Moravcsik MJ. Citation context classification of a citation classic concerning citation context classification. Social Studies of science. 1988;18:515-521. [34] Bohannon RW. Journals relevant to practice. Phys Ther. 1987;67:435-436. Letter to the Editor. [35] Bohannon RW. Core journals of physiotherapy. Physiotherapy Practice. 1987;3:126-128. [36] Bohannon RW. Physiotherapy Practice. 1989;5:99-100. Letters page. [37] Bohannon RW, Gibson DF. Citation analysis of Physical Therapy Phys Ther. 1986;66: 540-541. [38] Dean E, Davies J. Frequency of citation and reputational assessment of contributors in physical therapy. Phys Ther. 1986;66:961-966. [39] Rothstein JM. Editor's note: Living without student research projects. Phys Ther. 1992;72: 332-334. [40] Lemaine G, MacLeod R, Mulkay M, Weingart P. Problems in the emergence of new disciplines. In: Lemaine G, MacLeod R, Mulkay M, Weingart P, eds. Perspectives the Emergence of scientific Disciplines. The Hague, die Netherlands: Mouton Publishers; 1979:1-23. [41] Frank G. The Boulder Model: history, rationale, and critique. Professional Psychology: Research and Practice. 1984; 15:417-435. [42] Striker G. The relationship of research to clinical practice. Am Psyhol. 1992;47:543-549. [43] Lee VL. Organisms, things done, and the fragmentation of psychology. Behavior & Philosopby, 2091994;22:1-42.