Problem-knowledge coupling: a tool for physical therapy clinical practice.

Problem-Knowledge Coupling: A Tool for Physical Therapy Clinical Practice Recently, there has been a renewed emphasis in physical therapy on improving the scientific basis of clinical practice. This effort is considered important for the good of the patient [1,2] and to establish accountability [2,3,4] and scientific credibility within the profession. [5,6] To accomplish this goal, physical therapy researchers and practitioners have suggested a reappraisal of how clinical patient information is collected, measured, and evaluated, [7] and ideas have surfaced in the physical therapy literature on how to improve the clinical decision-making process itself. [3,8,9] In addition, use of the computer as an aid both in collecting patient data and in the process of making clinical decisions has been advocated. [4,9]

The purposes of this article are 1) to review the development and philosophy of a computerized information system--the Problem-Knowledge Coupler [R] (PKC [R])*--that was developed as an aid to physicians in making medical practice decisions and 2) to suggest the potential application this system has to current and future needs in the clinical practice of physical therapy.

Development of the

Problem-Knowledge Coupling

Concept

The basis and historical foundation for problem-knowledge coupling is the problem-oriented system (POS) of medical care developed by Lawrence L Weed, MD, and others in the 1950s. [10] The clinical patient care core of the POS is the medical record, which contains a defined database from which a problem list, initial plans, and progress notes are developed. Proponents of the problem-oriented medical record system have suggested that use of the problem-oriented medical record provides for greater consistency, thoroughness, and organization in data collection in addition to an opportunity for the clinician to demonstrate the logic of all clinical management decisions. [10] This system has been widely applied and currently is commonly used in the clinical practice of nursing, medicine, and physical therapy. Its specific use and benefits have been described in publications within each of these fields. [11-13]

In the 1970s, Weed added the use of computer technology to the POMR system concept as a means of expanding the POS concept and decreasing the heavy reliance on memory that exists in the practice of medicine. [14] With funding from a federal grant, he established the Problem-Oriented Medical Information System (PROMIS) laboratory at the Medical Center Hospital of Vermont in Burlington, Vt. This project, operating from 1970 to 1981, established a computerized problem-oriented medical record (CPOMR) system that operated on a large mainframe computer and that was used first in a gynecology unit and later in a medical unit at the hospital. With the patient as the focus of the system, the CPOMR was used to improve the coordination among a patient's health care providers, decrease the dependence on memory for patient and medical information, and provide a system that recorded the logic of an individual's actions and established feedback loops for subsequent audit. [14] At the conclusion of the PROMIS project, Weed began to develop the PKC [R] system.

Problem-Knowledge Coupler [R]

System

The PKC [R] system is a computer software system that can be run on microcomputers, is applicable to a wide variety of patient problems, and modifies the concepts of the PROMIS project in several important ways. Weed found from his experience with the PROMIS project that clinicians required help in synthesizing the massive amounts of information the computerized collection of information provided. [14] He designed the PKC [R] system to aid both in the collection and the synthesis of patient information so as to guide the practitioner in making diagnostic or patient management decisions. As its name implies, the ultimate goal of the system is to couple existing patient problems with the available knowledge in the field that is clinically relevant to a specific patient's unique set of problems (Fig. 1).

There are three major components in the system. Couplers are problem-oriented medical guidance modules. Each coupler is developed around a specific type of patient problem. For example, couplers that Weed and associates have already developed and marketed include such diverse topical problems as knee problems, hypertension, and dizziness (Appendix). The coupler development system is the component of the PKC [R] system that allows the user to either build an original coupler or to modify an existing one. The third component of the system is the knowledge network, which stores the referenced information from the clinical literature that is the basis for the couplers.

Basic Function of Couplers

There are two major types of couplers: 1) diagnostic couplers and 2) management couplers. Both types of couplers aid the clinical decision-making process by either providing the user with relevant possible diagnoses (diagnostic coupler) or presenting the clinician with follow-up management options (management coupler) that are based on the individual patient's pattern of signs and symptoms.

To determine possible diagnostic options or management priorities, patient information is first collected via a series of questions or statements to which the clinician or patient respond. Fig. 2 is an example of one screen of tests from a diagnostic shoulder problem coupler. Statements that are true for a patient are flagged with arrows, and following completion of all screens in the coupler, the software compiles and displays a list of all observed findings. In addition, depending on whether the coupler is a diagnostic or management type, the system also couples individual findings with specific diagnostic possibilities or management options. The linkage of patient findings to a specific diagnostic possibility or management option is made during the development of the coupler and derived from the available clinical literature by the developer of the coupler.

To guide the clinician in making diagnostic decisions, couplers in the PKC [R] system provide the user with a list of potential causes of the patient's signs and symptoms along with a comparison of the number of observed patient findings to the total number of findings stored in the system for each of the diagnoses that was generated (Fig. 3). For example, the finding of "resisted external rotation of the arm is painful," noted to be true for the patient in the question 13 screen (Fig. 2), was linked during the development of the shoulder coupler to a diagnosis of "rotator cuff tendinitis--impingement syndrome." When the clinician, therefore, found this sign to be true for the patient and flagged it appropriately, the system automatically generated the diagnosis to which it had been linked previously and "rotator cuff tendinitis -- impingement syndrome" appeared on the "possible causes" screen. "Resisted external rotation of the arm is painful" is only 1 of 11 findings derived from the clinical literature by the developer of the coupler and stored in the shoulder coupler when it was built. The patient whose possible causes of shoulder pain are demonstrated in Fig. 3 had 5 other findings (located on other question screens within the coupler) that were also suggestive of a diagnosis of "rotator cuff tendinitis -- impingement syndrome." The patient, therefore, had a total of 6 findings present out of the total of 11 that were stored in the system for that diagnosis. In addition, the patient demonstrated other positive findings during the examination that were suggest of diagnoses other than a rotator cuff problem, and these findings are also listed in Fig. 3 along with their respective comparisons of observed findings in the patient to total findings stored in the system for each of them. To further clarify any one diagnosis, the PKC [R] system can also produce a list of findings present and findings not present for each of the generated diagnoses. Clarifying comments and referenced resource backup for the information contained in the coupler can be accessed in the form of "facts" located in the knowledge network component of the system. An example of a fact from the knowledge network that supports the linkage of the finding "resisted external rotation of the arm is painful" to the diagnosis of "rotator cuff tendinitis--shoulder impingement syndrome" is demonstrated in Fig. 4. Fact screens typically include the findings, clarifying comments, and the reference for the stated information.

Management couplers operate in a format that is similar to diagnostic couplers. They solicit relevant information from the patient and clinician and compile and display lists of observed findings. In management couplers, however, patient findings are linked to specific management versus diagnostic options. As with diagnostic couplers, the documentation that supports the linkage of patient response to suggested management option is derived from the clinical literature and stored in the knowledge network along with the referenced sources of that information.

Philosophical Basis of

Problem-Knowledge Coupling

Weed has explained the philosophical basis and practical implementation of problem-knowledge coupling in detail in clinical practice journals, [15] in presentations to scientific societies, [16] and in his own publications. [17] To better understand the potential application of this system to the clinical practice of physical therapy, however, some of the major philosophical premises for the development of the PKC [R] system are summarized here.

Weed developed the PKC [R] as a workable clinical tool to bridge the gap he perceives to exist between medicine's ideal of patient management and the reality of how it is actually practiced. [17] He notes that patients entering the medical system have certain expectations. They expect that the physician will perform the appropriate evaluation and, based on their signs and symptoms, will know and choose the appropriate management strategy that will solve their problem. Weed, however, has questioned the ability of the clinician to meet these expectations. He notes the severe limitations of the unaided human mind to remember all the correct questions to be asked and relevant tests to be performed for the many different patients seen daily in a busy clinical practice. In addition, he challenges the ability of even the most, expert clinician to remember and process the large number of different potential causes or management options that each patient's findings suggest as well as the types of information needed to distinguish between competing diagnostic hypotheses or management strategies in each patient situation. Furthermore, Weed feels the constraints of having to make these decisions in the limited time of the patient-clinician encounter puts impossible burdens on the practicing clinician and likely results in less than optimal patient care.

Historically, Weed explains, medicine has handled these clinical practice problems by applying laws of probability to medical practice decisions. [17] He objects to the traditional medical decision-making model that relies solely on probability estimates of problems in the general population to guide management of individuals. He believes that the analysis and subsequent management of a patient's problems is best accomplished by first understanding the unique pattern of findings that exist in each individual patient situation. It is each patient's unique pattern of findings (not theoretically derived probability estimates based on large population statistics) that should be used initially to make clinical decisions related to diganosis or follow-up medical management. Once the uniqueness of the individual patient's situation has been exhausted, then, if ambiguity remains, probabilities can be used to choose among or further narrow diagnostic or management choices.

Weed acknowledges that an appreciation of the power of multiple variables to define a problem and access to tools that can accomplish the mechanics of collecting, synthesizing, and organizing such information are essential aspects of what he calls a combinatorial versus a probabilistic approach to patient management. [17] He suggests that if clinicians had a tool that helped them to collect, store, process, and organize a large number of easily obtained patient findings, then the process of making diagnostic or management decisions would be clarified and a patient's unique circumstances would not be obscured or ignored during the process. He believes the computerized PKC [R] system provides such a tool. Together, the computer and the PKC [R' software can be the technological link that guides the clinician in the evaluation and management process, helps to sort and process the collected patient data, and provides the clinician with links to the clinical practice literature that should serve as the rationale upon which patient care decisions are made.

In Weed's PKC [R] system, the numerical comparisons that accompany diagnostic couplers are designed to be indicators of the propensity of a patient for a certain diagnosis. [16] They cannot be considered probability estimates because the numbers used in the Weed ratio formulation are not derived from any analysis of a frequency of occurrence in a large population. Rather, the numbers represent only those findings for an individual patient (numerator) as compared with the total number of findings that the builder of the couplers stored in the system for that particular diagnosis (denominator). The purposes of the numerical comparisons are to give the clinician an overview of the range of diagnostic options that are suggested by the patient's own unique pattern of signs and symptoms and to provide a methodical and systematic way for the clinician to consider all possible options that are relevant to that patient inclusive of those that are uncommon or rare.

In contrast, Weed notes the traditinal probability-based model often relies on early hypothesis formation, which rejects low probability diagnoses at the outset of care and instead follows the path of the "most likely" diagnosis (ie, the one having the greatest chance of occurrence in the general population). [17] He cautions against this policy in making decisions about individual patients because serious pathological conditions often have an evolving nature that will change over time. These unlikely options, therefore, should not be discarded for an individual based on their low probability of occurrence in the general population. In addition, Weed notes that forming an early hypothesis implies rejection of others. This notion assumes single causes for sets of findings, an assumption that defies the reality of patient problems. Furthermore, Weed believes that the follow-up procedures that are necessary to test and rule out various diagnostic options considered "likely" based on general population occurrence may be considered costly, risky, or unnecessarily time consuming if they are being carried out on patients to whom in reality they do not apply.

Clinical medicine has just begun to acknowledge the value and need for external aids in making clinical decisions. [18] Computerized "expert" systems other than the Weed system have been developed and are evolving for the purpose of helping to make patient care decisions. Weed finds it ironic that many of these computerized expert systems seek to reproduce the frailties inherent in human experts and probability-based systems. [17] In addition, he notes that many of these expert systems are largely unavailable to the practitioner because of extensive computer hardware needs, cost constraints, or training requirements. Weed's PKC [R] system is designed to offer both a philosophical and practical alternative to these expert decision-support systems.

Potential for Application of

Problem-Knowledge Coupling in

Physical Therapy

Although problem-knowledge coupling was developed as an aid in resolving issues in the field of medicine, we believe the clinical practice of physical therapy could also benefit from its use. We believe that the Weed PKC [R] system has the same potential to improve the quality of care in physical therapy clinical practice as it does in medicine. Most importantly, the PKC [R] system, with its origin in the POS, maintains the patient with his or her unique set of findings as the focus of treatment interventions. Furthermore, the methods used in problem-knowledge coupling have the potential to provide support to the physical therapy clinician in critical professional practice issues such as professional autonomy, professional accountability, and scientific credibility of clinical practice.

Although the actual operation of PKCs [R] is simple and based on single keystroke activation at the computer's keyboard, the development of an original coupler is an intensive and time-consuming task. It requires an extensive search of the professional literature, which in the field of physical therapy may be severely hampered by the poor availability of scientific clinical data related to evaluation and management procedures. In addition, coupler development requires a familiarity with the mechanics of the PKC [R] system and a commitment to the philosophy of clinical practice on which the system is based.

The real advantages and disadvantages of problem-knowledge coupling in physical therapy clinical practice remain to be demonstrated as couplers that are useful in physical therapy are developed and implemented in clinical practice settings. We speculate, however, that problem-knowledge coupling has the potential to be of value to physical therapists in the following ways:

1. Use of diagnostic or management couplers would improve patient evaluations by improving the consistency and reliability with which evaluative information is sought on each patient. The software organization of the PKC [R] provides for a systematic and methodical way of questioning and testing the patient. The data collection phase of patient evaluation would not depend on human recall with its attendant deficiencies, and intertherapist examination of patients and evaluation of similar patient problems would be more consistent and equitable.

2. Development of diagnostic or management couplers would help physical therapists define and explore in a clinically relevant manner what constitutes the body of knowledge related to specific clinical problems. Using the coupler development component of the PKC [R] system, it is possible for physical therapists to build a coupler of their own choosing. Couplers are developed from either a search of the clinical literature or at least a complete examination of commonly accepted clinical practice. This development requires a thorough review and documentation of all possible diagnostic reasons or management options for the stated problem content area of the coupler as well as associated symptomatology or indicators for management strategies. This process could serve as an excellent educational tool for either the physical therapy student or the established clinician in reviewing all possible causes or management options that have been reported in the literature for a certain clinical problem.

3. Use of the system would provide clinicians access to the clinical literature as the basis for clinical practice. In conjunction with developing couplers, the knowledge network component of the system is established. The knowledge network acts as a repository of documented information about what is currently known in clinical diagnostics and management about the problem area under evaluation. It is the basis for the content of the couplers. Linkages of patient findings to either diagnostic or management options as they appear in the PKC [R] system, therefore, are not speculative or biased by the examiner's own prejudices in clinical care. Rather, they reflect current knowledge in the field, that is, the scientific basis of clinical practice as it is currently known. At any point in the diagnostic or management process, these documented references can be reviewed by accessing the knowledge network component of the system where they are stored.

4. Use of the PKC [R] system would help to make the logic of patient management decisions more explicit. Sound analytic sense is one of the major behavioral characteristics that is evaluated in audit procedures that review a clinician's decision-making capability. Because clinical decisions are often assumed to be intuitive, however, this characteristic is often difficult to assess. With problem-knowledge coupling, options that are based on the available literature are presented to the clinician. The clinician can demonstrate to the auditor the path of decision making that was taken based on the individual characteristics of the patient under consideration. Accountability to external auditors of physical therapy management would be engendered.

5. Use of the PKC [R] system would help in making management decisions for patients that respect, not ignore, their uniqueness. The importance of recognizing the uniqueness of patients' problems and of incorporating that uniqueness into patient management has often been noted. [3] The reality of being able to do so, however, has been elusive. The Weed PKC [R] system may give physical therapists one way to preserve the patient's uniqueness and respond to it in management. In addition, it may help physical therapists begin to understand the reasons why certain patients respond to some physical therapy interventions and others do not. Patients who seek physical therapy care often have problems that distinguish them from other patients and that therefore may be the basis for differences in the outcome of similar physical therapy management procedures.

6. Professional autonomy includes the responsibility of identifying when physical therapy is not appropriate management for a patient. At a time when recent physical therapy practice act and state law changes allow patients direct access to physical therapy without physician referral, the need to determine individual suitability for physical therapy procedures is especially acute. An awareness at the outset of care of potential contraindications for physical therapy procedures or of existing health problems that might interfere with rehabilitation is a professional and legal responsibility.

Appendix. Problem-Knowledge Coupler [R] (PKC [R]) System--Information, Specifications, and Costs

PKC [R] Software Packages and Costs PKC [R] system package cost: $1,495 PKC [R] system package includes: PKC [R] coupler system and 27 medical guidance modules (couplers) PKC [R] development system PKC [R] knowledge network system

Topics of currently available medical guidance modules are: screening history, screening physical, guidance toward better health and wellness, acute abdomen, unexplained vomiting, jaundice, hypertension, guidance in medical management of hypertension, chest pain, ECG interpretation, abnormal heart sounds, headache, memory loss and confusion, vertigo and dizziness, knee problems, low back pain, shoulder problems, abnormal vaginal bleeding, hematuria, dysuria, upper and lower respiratory complaints, hypercalcemia, diagnostic workup of suspected lung cancer in patients with abnormal chest films, anemia, assessment and management of acute asthmatic attack, obesity, hyperlipidemia management.

Individual components of the package and sets of couplers may be purchased separately.

Hardware Configuration

IBM personal computer, (a) IBM PC-XT [TM], (a) IBM PC-AT [TM], (a) IBM personal computer system/2 models 30, 50, 60, and 80, (a) or an IBM-compatible computer with at least 256K of random access memory.

Minimum of two floppy disk drives (5-1/4 or 3-1/2 in (b)) or one or more floppy disk drives and a hard disk with at least 3M of space free. Hard disk recommended for full-system package.

Operating System 

IBM PC DOS 2.00 (a) or newer version required.

Distribution Media

Double-sided, double-density 5-1/4- or 3-1/2-in soft-sectored diskettes, formatted for MS-DOS 2.xx/3.xx.

For more information, contact PKC Corporation, 10 Mary St, South Burlington, VT 05401.

It has been difficult, however, to easily assess and record a patient's overall health status in a comprehensive and efficient way. Couplers that survey general health and offer medical management options have been developed by Weed (Appendix), and the PKC [R] system includes the potential to develop similar couplers that could specifically address the needs of health screening in preparation for physical therapy intervention.

7. Use of problem-knowledge coupling may serve as a stimulus to involving clinicians in clinical research. The wide survey of clinical practice methods and their rationale that is needed to build a coupler and its associated knowledge network will inevitably demonstrate not only the agreed upon principles of practice but also the ambiguities and controversies that exist within the field. Weed has encouraged this self-examination of practice as an educational tool. [10] It alerts the clinician to what is going on outside narrow personal views or a certain focus of practice, and it may stimulate the clinician to more closely examine the evidence for the rationale and effectiveness of common practice procedures. Questions that cannot be answered from the literature or those that remain unresolved are the basis for good clinical research questions. Because these questions arise out the clinician's own contextual reference of trying to improve patient care, perhaps the clinician will abandon the view that most research is not relevent to clinical practice [19] and instead become an active participant in producing relevant research.

Summary

Physical therapists, in their efforts to improve the scientific basis of clinical practice, have begun to explore various ways to improve the process of making clinical practice decisions. We have described one computerized medical information system--the PKC [R]--that originated in the POS of medical care. Problem-knowledge coupling was designed as a tool to aid medical practitioners in making clinical decisions in medicine. We have speculated on the potential of problem-knowledge coupling to address some of the important patient care and professional practice issues in the field of physical therapy.

Acknowledgment 

We wish to express our appreciation to Lawrence L Weed, MD, for the many provocative and creative discussions he has evoked during the writing of this article.

(*1) PKC Corp, 10 Mary St, South Burlington, VT 05401.

(a) International Busines Machines Corp, PO Box 1328-S, Boca Raton, FL 33432.

(b) 1 in = 2.54 cm.

References

[1] Feinstein AR: Clinical Judgment. Baltimore, MD, Williams & Wilkins, 1967

[2] Ottenbacher K: Evaluating Clinical Change: Strategies for Occupational and Physical Therapists. Baltimore, MD, Williams & Wilkins, 1986, p 8

[3] Watts NT: Decision analysis: A tool for improving physical therapy practice and education. In Wolf SL (ed): Clinical Decision-Making in Physical Therapy. Philadelphia, PA, F A Davis Co, 1985

[4] Hislop HJ: Clinical decision-making: Educational, data, and risk factors. In Wolf SL (ed): Clinical Decision-Making in Physical Therapy. Philadelphia, PA, F A Davis Co, 1985, p 41

[5] Basmajian JV: Professional survival: The research role in physical therapy. Phys Ther 57: 283-285, 1977

[6] Rothstein JM: Measurement and clinical practice: Theory and application. In Rothstein JM (ed): Measurement in Physical Therapy: Clinics in Physical Therapy. New York, NY, Churchill Livingstone Inc, 1985, vol 7, p 1

[7] Campbell SK: On the importance of being earnest about measurement, or, how can we be sure that what we know is true? Phys Ther 67: 1831-1833, 1987

[8] Rothstein JM, Echternach JL: Hypothesis-oriented algorithm for clinicians: A method for evaluation and treatment planning. Phys Ther 66:1388-1394, 1986

[9] Shewchuk RM, Francis KT: Principles of clinical decision making--An introduction to decision analysis: A special communication. Phys Ther 68:357-359, 1988

[10] Weed LL: Medical records that guide and teach. N Engl J Med 278:593, 652, 1968

[11] Woody M, Mallison M: The problem-oriented system for patient-centered care. Am J Nurs 73: 1168-1175, 1973

[12] Walker HK, Hurst JW, Woody MF (eds): Applying the Problem-Oriented System. New York, NY, Medcom Press, 1973

[13] Feitelberg SB: The Problem-Oriented Record System in Physical Therapy. Burlington, VT, University of Vermont, 1975

[14] Wakefield JS (ed): Managing Medicine. Kirkland, WA, Medical Communications and Services Association, 1983

[15] Weed LL, Hertzberg R: Knowledge coupling for microcomputers in office practice. Update: Computers in Medicine 1(2):49-61, 1983

[16] Weed LLd Problem-Knowledge Coupling. Read at the New York Academy of Medicine Symposium on Biomedical Engineering, New York, NY, March 9, 1982

[17] Weed LL: Problem-Knowledge Couplers: Philosophy, Use, and Interpretation. South Burlington, VT, PKC Corporation, 1982

[18] Ziporyn T: Medical decision-making Analyzing options in the face of uncertainty. JAMA 249:2133-2142, 1983

[19] Bohannon RW, LeVeau BF: Clinicians' use of research findings: A review of literature with implications for physical therapists. Phys Ther 66:45-50, 1986

N Zimny, MS, PT, is Assistant Professor of Physical Therapy, Department of Physical Therapy, School of Allied Health Sciences, University of Vermont and State Agricultural College, Burlington, VT 05405 (USA).

C Tandy, BS, PT, is Advanced Clinician, Department of Physical Therapy, Medical Center Hospital of Vermont, Burlington, VT 05401.

This paper was supported in part by a grant from the Research Fund of the Associates in Physical and Occupational Therapy Inc, Burlington, VT.

This article was submitted May 3, 1988; was with the authors for revision for three weeks; and was accepted September 15, 1988.