A little neglect may breed mischief: for want of a nail the shoe was lost; for want of a shoe the horse was lost; and for want of a horse the rider was lost.— Benjamin Franklin, Maxims prefixed to Poor Richard's Almanac, 1757
Technology alone is rarely a solution to disorder in work processes. And in medicine, disorder is pervasive. Disorder in medical decision making is the inevitable outcome of reliance on the unaided human mind for applying medical knowledge to patient data. That information processing function is carried out under severe constraints of time and complexity. Operating under those constraints, the mind of the physician is necessarily fallible, inefficient, expensive, and scarce.
This problem goes far beyond “medical error” as usually conceived. The prevailing medical culture—and in particular the educational and credentialing institutions that confer legal authority on the physician—remain in denial about the scope of the problem and the wrenching changes needed to solve it. The consequence is that reliance on the physician's mind stifles use of a superior alternative. For the want of that alternative, cognitive inputs to medical decision making are uncontrolled. For want of controlled inputs to medical decision making, the quality of care, the cost of care, the education and credentialing of caregivers, and the development of medical knowledge itself, are unmanageable. With the threat of malpractice forever hanging overhead, physicians are forced to function in the face of avoidable ignorance of crucial details from medical science and from the patients themselves. We are all trapped in an archaic infrastructure where the transmission of knowledge from its source in libraries and laboratories to patients is faulty and where the coordination among providers of care has broken down. Physicians and patients alike are left with a sense of futility and despair.
Medicine needs a new infrastructure in the form of information tools that serve two purposes: [1] bridging the gap between the physician's mind and the unbearable burden of coupling patient data with medical knowledge, and [2] organizing the multiple processes involved in patient care (1). The first purpose is achieved with software tools specifically designed for use by caregivers and patients in real-world conditions. The second purpose is achieved with structured medical records that compel transparency and organization as part of caregiving itself. These powerful tools embedded in a well-defined system of care can lead to better medical science and a better science of medical practice.
Of these two types of information tools, the most immediately accessible is the first—software for applying medical knowledge to patient data. Tools of this kind can be a great liberating force for caregivers who have felt impossible pressures to live up to the image of the all-knowing, all-caring physician—the storybook image that many students imagine for themselves as they enter medical school. As their careers develop, they specialize by body system and technique to acquire competence. But their competence as specialists does not make them competent to meet all of a patient's real needs. The complexity and continuous flood of new knowledge even within one specialty exceeds their capacity to master. Moreover, a single problem may implicate multiple body systems, crossing specialty boundaries, and patients frequently have multiple problems that interact in complex ways. Unless equipped with tools for coping with these realities, neither highly trained specialists nor primary care practitioners alike can consistently function competently or safely.
A prerequisite to understanding the new tools and the opportunities they offer is understanding the multiple work processes in which caregivers are engaged and why those processes are in a state of disorder. The author has been engaged in such an effort since the mid-1950s. That effort revealed first of all that disorder exists because the functioning of caregivers is not organized around the four basic steps that orderly problem solving requires in every field: gathering information, formulating problems, developing plans for each problem, and acting on each plan and subsequent feedback. The question then arose—how is this kind of organization to be achieved? The answer turned out to be an information tool that is already familiar to practitioners: the medical record.
Traditionally, the medical record had little utility as an organizing tool, because its contents were arranged in a source-oriented manner (doctors' notes in one place, nurses' notes in another, laboratory data in another). A source-oriented arrangement bears no relation to the four basic steps of problem solving just described. To address this failing, a problem-oriented structure was developed for medical records 2, 3. Problem-orientation has two interrelated meanings: [1] data in the record are arranged by the patient problem to which the data relate, and [2] problems are defined based on supporting data and patient needs, not provider hypotheses and purposes. The problem-oriented medical record (POMR) turned out to be a powerful tool. And its design is relevant to any electronic form of medical record (4). As one user wrote in 1974: “the POMR compels relationships and interdependencies as conditions of physician conduct. And now, through computer technology, it draws us into a thorough-going consistency, which is unfamiliar to most natures and threatening to our tolerated caprices” (5).
The POMR has four components, corresponding to the four problem-solving steps described above: a database, problem list, plans, and progress notes for each problem.
Problem-oriented progress notes
Required: Progress notes and data flowsheets labeled by the problem to which they relate, thereby placing information in its logical context, permitting multiple providers to coordinate their actions without omission or duplication, and providing feedback on the actions taken.
Revealed: Frequently, disciplined implementation of plans is lacking and plans are not appropriately modified as relevant new data accumulate. Moreover, providers often do not communicate effectively, and risky or expensive procedures may be duplicated. Clinicians try to deal with many variables over time without flowsheets and trends are missed and necessary changes are not made (6). These points emerged from experience with POMR in paper form from the mid-1950s into the 1970s. It was obvious that adapting computers to the POMR offered great potential to enhance its utility. Most importantly, the computer could make the POMR a tool that would itself contain parameters of guidance and currency of knowledge unavailable from the physician's mind. Accordingly, from the late 1960s to the late 1970s, work progressed on developing a minicomputer-based, electronic version of the POMR. This system included not only problem-oriented patient data but also thousands of up-to-date displays of general medical knowledge. Bypassing the limitations of the human mind as a storage and retrieval device in this way, however, highlighted the mind's limitations as an information-processing device. A sonnet by Edna St. Vincent Millay perfectly captures the dilemma, one not limited to medicine:
Upon this gifted age, in its dark hour,
Rains from the sky a meteoric shower
Of facts…they lie unquestioned, uncombined.
Wisdom enough to leech us of our ill
Is daily spun, but there exists no loom
To weave it into fabric… (7)
In response to this need for a device to weave medical knowledge into the fabric of care, a new form of specialized database software was developed, beginning in the early 1980s. Known as problem-knowledge coupling software, it is designed as a tool for coupling patient data with medical knowledge directly relevant to diagnostic and treatment problems faced in everyday patient care. The software performs this coupling function by identifying detailed data needed to investigate a given diagnostic or treatment problem and then, once the data on a patient is entered, analyzing the data's significance in light of comprehensive medical knowledge. If the discipline of collecting the needed data is accepted by the provider and patient, then the software presents to them, based on the patient's unique combination of findings, a set of directly relevant diagnostic and management options and the pros and cons of each option for that particular patient. The patient's values determine the options that are chosen, and the provider's skills are called upon for those options that the patient cannot implement.
Problem-Knowledge Couplers® have been developed by PKC Corporation for over 100 diagnostic and management problems, as well as for screening and health maintenance. Users find that the existing body of diagnostic and management Couplers, although not complete, is useful for at least 80%–90% of the patients who appear in a typical primary care practice and that the screening and health maintenance Couplers are useful for all patients. The universal use of these systems allows the unique patterns of each individual to become apparent and prevents the errors that occur when treatment is provided based on the generalities of population studies.
Knowledge-coupling software enables non-physician caregivers and patients themselves to achieve a level of reliability, completeness and efficiency that physicians aspire to but cannot consistently achieve. Because physician time is expensive and scarce, their initial workups can be meager. In addition, most physicians act according to their own preconceived notions about what history, physical, and laboratory findings are worth checking. Equally idiosyncratic are the conclusions they draw from whatever data they select. Both selection and analysis of data are influenced heavily by their medical education, prior clinical experience, specialty orientation, contradictory clinical guidelines, financial concerns, cultural background, personal biases, and day-to-day time constraints, all of which vary enormously among individual practitioners. This variation has always been apparent to laymen when they shop for doctors and has not been lost upon the press (8).
The first components of the POMR (a defined database and complete problem list) implemented with knowledge coupling software together represent a concrete, practical approach for preventing random variation in initial data collection and analysis 9, 10, 11.
Knowledge coupling tools make possible a fundamental change in the way we move knowledge from its source in laboratories and libraries to the people who need the rigorous application of that knowledge. The physician's mind is no longer required to be the vehicle for bringing medical knowledge to the point of need in patient care. In turn, medical education and credentialing of providers will have to change from a knowledge-based to a skills-based approach. We must shed the illusions instilled by graduate medical education. Physicians are “educated” to believe that, in Herman Blumgart's words, “The application of knowledge at the bedside is largely the function of the sagacity inherent in or personally developed by the individual physician” (12). Sherwin Nuland has further described the traditional ideal:
The quest of every doctor in approaching serious disease is to make the diagnosis and to design and carry out the specific cure. This quest I call The Riddle, and I capitalize it so there will be no mistaking its dominance over every other consideration. The satisfaction of solving The Riddle is its own reward, and the fuel that drives the clinical engines of medicine's most highly trained specialists. It is every doctor's measure of his own abilities; it is the most important ingredient in his professional self-image. Our most rewarding moments of healing derive not from the works of our hearts but from those of our intellects—it is there that the passion is most intense (13).
Yet, personal sagacity is usually a poor substitute for effective information retrieval and processing. The decision-making process requires recognizing simple data linkages that can be ascertained from the medical literature and captured in software tools. Given an abnormality to diagnose, for example, the medical literature reveals both the associated possible diagnoses and the associated data needed to confirm or rule out each possible diagnosis cost-effectively. Rather than relying on the human mind to recall or figure out these associations, the far more effective approach is to use computer tools to guide caregivers in determining what data to collect for a given problem and what medical knowledge has to say about the results.
Shifting this data-processing burden from the human mind to electronic tools exposes an enormous gap between the data collection and analysis that clinicians normally engage in and what they can do when freed from the constraints of their own minds. It becomes possible to employ systematic combinatorial analysis of detailed data, rather than hit-or-miss hypothesis formation from incomplete data. Combinatorial analysis extracts maximum guidance from simple, inexpensively acquired bits of information arranged in meaningful patterns, minimizing trial and error with risky or expensive procedures.
To illustrate, diagnosing abnormal vaginal bleeding requires taking into account more than 63 possible causes. Efficiently identifying which of these causes should be considered for an individual patient requires considering over 160 simple clinical findings at the outset of care, and then analyzing each patient's unique combination of those findings in light of current medical knowledge. Similarly, for a common problem such as chest pain there are over 100 causes and over 320 findings. Safe and cost-effective management of contraception requires taking into account more than 20 different management options and approximately 218 distinct initial observations. For the management of diabetes, 120 management options and 380 findings on the unique patient should be taken into account. Often the effect of this comprehensive approach is, paradoxically, to simplify decision making. As Chris Weed wrote in 1982 about the effect of systematically combining detailed data with comprehensive knowledge:
[this combinatorial approach] uses the power of simple observations in the aggregate to restrict the range of possible solutions to our problems. When we know many things, even many simple things, about a patient, we are in a better position to see the full range of possible causes for the patient's problems and to find combinations of those things that discriminate clearly between the possibilities. The more information we have about a patient, and the more effective our tools for combining and matching within that information are, the less the need there is to split hairs over the value of each piece of data (14).
In contrast, most physicians are able to take into account only a portion of the diagnostic and management options potentially relevant to their patients and only a fraction of the evidence needed for choosing among those options. As a result, physicians fall back on their personal knowledge and sagacity. They hope that whatever data they choose to obtain is enough and then hope they can recognize the implications of those data. The result is a significant voltage drop between what is known to medical science and each physician's limited and variable application of knowledge at the bedside. Physicians do little better with the usual aids to medical decision making, such as practice guidelines and use of Internet resources. Those aids provide general knowledge, but do not ensure that the physician will recall all the data or successfully link it with a particular patient's specific problem.
The way physicians are taught to function flies in the face of decades of research in cognitive psychology, decades of research in health care quality, decades of experience in other industries, and common sense. Common sense tells us to rely on maps and a compass or GPS device, not on our sense of direction, when navigating in unfamiliar territory. An airline pilot uses radar; he does not claim to be able to see through clouds. In other areas we have extended our muscles with machines, and our eyes with microscopes and telescopes. Similarly, we should extend our cognitive capacities to recall and process the many variables in solving clinical problems. Relying on recall is unsafe, unreliable and unnecessary. We must use technology and system organization to create a rational division of labor, where people and machines are assigned functions to which they are suited. The present infrastructure of the medical system with its flawed beliefs, inadequate information tools, and poorly defined linkages among its parts does not support such a rational division of labor. Until a new infrastructure is put into place, acceptable quality and productivity will remain out of reach.
According to a working committee of the National Institutes of Health, “the dominant trend in biomedical science and medical practice, as in every realm of science, is the increasing value and usage of computers” (15). Yet medical practice lags far behind medical science in usage of computers. In contrast to clinical imaging tools and other devices for medical practice, the device primarily used for medical decision making—the physician's brain—functions largely without the aid of computer software. Patients can no longer be expected to depend on those graduates who “rely on their own creative intuition, experience, and random and informal contacts with other concerned people. Without the routine use of powerful knowledge coupling tools to generate specific linkages of the knowledge base to practical decision-making for unique individuals, scientific medicine affects practice primarily through new procedures and associated technologies, while the application of such procedures and technologies is left to a sort of cottage industry or folk art based on something approaching oral tradition” (16).
Physicians and legislators overlook indispensable software tools because medical education and specialty board and licensing examinations instill illusory ideals of personal knowledge and sagacity. In the words of Robyn Dawes, a cognitive psychologist at Carnegie Mellon University:
States license psychologists, physicians, and psychiatrists to make lucrative global judgments in the form of ‘It is my opinion that …’ People have a misplaced confidence in their global judgments, a confidence that is strong enough to dismiss an impressive body of research findings and to find its way into the legal system. The greatest obstacles to using external aids may be the difficulty of convincing ourselves that we should take precautions against ourselves. The idea that self- imposed, external constraints on action can actually enhance our freedom by releasing us from predictable and undesirable internal constraints is not a popular one. (17)
Providers of medical care must shed all the illusions that were bought at such a high price from an antiquated system of medical education and accreditation. Not only has the effort to move knowledge through heads, instead of through the disciplined use of knowledge tools, failed for many patients, but the massive amount of time and money that are going into the existing systems of medical education and licensure are preventing the development of an infrastructure in medicine in which a complete, up-to-date set of knowledge tools are used routinely by all patients and caregivers traveling across the medical landscape. The global judgments of medical “experts” would become irrelevant. High quality and efficient care would emerge case-by case, in a progression of many small logical steps tailored to each patient's unique combination of needs. Each provider would be responsible for helping patients at one or more steps in the system where technical skill is necessary. Much of the knowledge and the medical logic is available in the new information tools themselves, allowing anyone with high proficiency in a specific medical procedure and willingness to perform that procedure to do so but only when the knowledge tools presented it as an option for a specific patient and only when the patient or the family chose that option. Providers' training and accreditation for a given skill would be inexpensive and far more focused and rigorous than at present, which in itself should do much to reduce the high level of errors that practitioners now commit.
No single provider would need to maintain the illusion of being a “total physician,” nor would the educational system need to support that illusion by requiring students to sweep superficially through and memorize an enormous amount of material at the expense of developing a sense of responsibility and excellence in the activities that they actually perform for patients. Instead, a sense of the whole would emerge from the providers and patients functioning within a system that achieves the whole, total care. Patients would have their own medical records, they would use knowledge coupling tools in conjunction with providers to support medical decision making, and they would be taught from childhood how to interact with the system, just as they now learn how to interact with travel systems. As patients assume decision-making power through the use of new information tools, their demands will reshape the healthcare marketplace in these directions. Once new premises, powerful new tools, linkages among all the parts, controlled inputs, automatic outcome studies and corrective feedback loops are built into the infrastructure, the system itself becomes the arbitrator of “who gets what.” As our country needed a constitution as a foundation for a government of laws and not of men, so the field of medicine needs a system with rules to control the expectations of patients and rules to organize the efforts and competing interests of the providers and the payers.
In shedding their illusions, the medical establishment and the public must embrace a new division of intellectual labor between man and machine. As Francis Bacon saw 400 years ago: “The unassisted hand and the understanding left to itself possess little power. Effects are produced by means of instruments and helps, which the understanding requires no less than the hand, so that those that are applied to the mind prompt and protect the understanding . The sole cause and root of almost every defect in the sciences is this, that while we falsely admire and extol the powers of the human mind, we do not search for its real helps” (18). Providers who pursue that search will encounter presently available “knowledge coupling” tools that enable a new way of looking at the practice of medicine 6, 10, 14, 19, 20, 21. And those interested in the science of medicine, independent of the problems of everyday practice, will find that as the telescope changed our understanding of the universe, so will the new tools open up a picture of medical science never visible to us before. What will emerge with increasing clarity is the gap between the fallible generalizations of medical “knowledge” and the realities of unique, individual patients. Those realities, in turn, intelligibly recorded in structured medical records, will be a fertile source of new medical knowledge.
FULL TEXT
