How is therapeutic knowledge generated and justified in medicine? Modern medical knowledge in terms of both diagnostic and therapeutic procedures is certainly more dependent on technical innovation than fifty years ago, as detailed in Chapter 5. But that dependency is more than simply a need for technical devices in order to practice medicine; rather, it has a definite epistemological component. "The episteme of technology," according to Ian McWhinney, "has become the episteme of medicine" (1978, p. 299). In other words, modern biomedical knowledge is driven by technical and research innovation, whether in terms of mechanical or cognitive devices. In this chapter, the epistemological issues concerning the cognitive or research devices are examined in the first section, followed by technical devices in the next section. The chapter ends with an examination of the epistemological issues associated with narrative therapy, often championed by humanistic or humane practitioners to address quality-of-care issues.
10.1 Biomedical Research
Biomedical research is patterned after research in the natural sciences. "The biomedical model," according to Engel, "was devised by medical scientists for the study of disease. As such, it was a scientific model; that is," he argues, "it involved a shared set of assumptions and rules of conduct based on the scientific method and constituted a blue print of research" (1977, p. 130). Biomedical research, especially in terms of clinical trials, is the single most important factor in the rise and for the success of the biomedical model in the twentieth century (Le Fanu, 2002).
Clinical trials are the systematic investigation or search for "facts" or "generalizable knowledge" concerning the efficacy of therapeutic treatments, such as pharmaceutical drugs and surgical procedures (Pellegrin and Nesbitt, 2004, p. 2). Such research often begins not at the bedside with patients but rather in the laboratory as preclinical research, e.g. in discovering new drugs and in testing them on laboratory animals. The goal of such research is to evaluate a therapeutic treatment's efficacy and its risk with respect to injurious side effects. Once a treatment is shown to be effective and safe within the laboratory on experimental animals, human trials are then under taken in the clinic.
There are three phases to clinical trials that must be completed successfully before a new treatment or procedure is approved for therapeutic purposes. A fourth phase may also be undertaken, after a treatment is marketed publicly, to insure its continued efficacy and safety. The "gold standard" of biomedical research is the randomized controlled trial (RCT): "Inherent in this scientific approach [RCT] is the epistemic status of the knowledge linked to therapeutic claims" (Christensen and Hansen, 2004, p. 68). The RCT exhibits two chief features: a concurrently controlled group to which the treatment group is compared and randomized allocation of patients to the two groups to remove bias (Matthews, 2000). However, not all medical knowledge or practice depends upon the RCT and other designs of clinical trials suffice for the generation of medical knowledge.
10.1.1 Clinical Trials
"Clinical trials," according to J.N.S. Matthews, "are experiments performed on human subjects, usually patients, in order to assess the efficacy of a treatment that is under investigation" (2000, p. xiii). Besides efficacy, clinical trials are also used to test a treatment's safety (Spodick, 1982). A treatment may be either a pharmaceutical drug or a surgical procedure. Clinical trails are also performed to analyze diagnostic protocols or screening programs. The clinical trial represents the introduction of the experimental method of the natural sciences into clinical medicine, which gave rise to twentieth-century clinical science. Today, it is the formalization of daily medical practice: "The practice of medicine is in effect the conduct of clinical research, in which questions are asked and new facts are obtained, synthesized, analyzed, and acted upon" (Chalmers, 1981, p. 325).
Empirical evidence is the standard by which epistemological issues concerning the efficacy and safety of a treatment are resolved in modern medicine. No longer is anecdotal evidence or authoritarian opinion adequate to justify a therapeutic treatment. "Simply believing that one treatment is superior to another," notes Matthews, "is not justification for acting on that belief: such justification requires you to collect evidence to prove or refute your beliefs and the RCT is the currently accepted tool for doing this" (2000, p. 3)
There are three types of clinical trials (Lilienfeld, 1982). The first is the therapeutic trial, in which a pharmaceutical drug, such as insulin, or surgical procedure, such as by-pass surgery, is used to treat a disease or its condition. The second type of clinical trial is intervention, in which the clinical scientist intervenes prior to the development of a disease in patients who exhibit the disease's symptoms partially or who are at risk for the disease. An example of such intervention is a double mastectomy on a middle-aged woman with predominant genetic markers for breast cancer. The final type of clinical trial is preventive or prophylactic, in which a drug or procedure is used to prevent the appearance of the disease, when a person is asymptomatic or normal. The classic example is a vaccination trial.
Each of the three types of trials can be further divided into explanatory and management trials (Sackett, 1983). The former trials are concerned with providing the mechanism by which the treatment works, generally under specified conditions and with a well-defined subject population. The latter trials are concerned with "all real-world consequences, good or bad, of treating an illness in a certain way and to determine whether therapy works, usually under as close toroutine clinical circumstances as possible" (Sackett, 1983, p. 66).
The structure of the RCT contains five important elements (Matthews, 2000). The first is the identification of a uniform population of eligible subjects, especially patients. The second is the selection of test subjects from the eligible population. Appropriate criteria for selection are critical to ensure that comparison between the experimental or treated and controlled groups is permissible and valid. To ensure that the results are free of bias, the assignment of subjects to either experimental or controlled groups is accomplished through randomization. The final element is a robust analysis of the results, especially in terms of statistical analysis. There are three key components to a RCT: randomization of the subjects, blinding of patients and clinicians to randomization, and controlled groups either for current treatment protocols or for placebo effect. It is these components that make the RCT a "gold standard"
10.1.1.1 Prerequisites for Clinical Trial Success
Before presenting the four phases of a clinical trial, several prerequisites for the success of such a trial must be discussed first (Sackett, 1983; Tobias et al., 2000). Success does not mean obtaining the results anticipated for a trial, but rather it means that the results bring consensus to the medical community (Tobias et al., 2000). The very first prerequisite is certainly the need for a drug or procedure to treat a widespread or debilitating disease. In other words, there is no current drug or procedure that is sufficiently effective or safe for treating the disease, or if there is it is not optimal and improvement would be beneficial.
The next prerequisite involves both the appropriateness and unambiguousness of the question asked of the effectiveness and/or safety of the drug or procedure. Moreover, the question must be appealing to patients in order to gain their participation in the trial. Sackett identifies two main questions asked in clinical trials. The first question seeks to determine if the drug is effective, e.g. "Can drug A reduce tumor size?", while the second asks whether it is worthwhile or safe, e.g. "Does prescribing drug A to patients with tumors do more good than harm?" (Sackett, 1983, p. 66). An inability to distinguish between these types of questions often leads to controversy over the interpretations of a trial's evidence.
Another prerequisite is the possibility or feasibility of the trial. According to Sackett, there are three elements to this prerequisite: "the protocol must be attractive to the potential clinical collaborators, appropriate types and numbers of study patients must be available, and minimal performance criteria for containing or abandoning the trial must be set" (1983, p. 74). Central to the first element is a clinical design that is aesthetically appealing to clinicians, especially in terms of providing not only valid conclusions but also relevant generalizations. The second element involves appropriate criteria for including or excluding test subjects in the trial, in order to attain appropriate numbers for statistical significance. Finally, reasonable and clearly defined criteria for abandoning a trial are required. Often the feasibility of a trial is initially determined through a pilot study, as long as it does not compromise these elements.
A critical prerequisite is a trial's practicality, particularly in terms of a protocol that makes patient participation straightforward and unproblematic: "The study must be perceived as attractive and appealing to the patient, to ensure enthusiastic participation" (Tobias et al., 2000, p. 1372). Many studies are unsuccessful or fail because the trial does not clearly communicate to the patient the importance or significance or the demands or sacrifices on the patient's part.' Thus, the success of a clinical trial, just as the success of a therapeutic protocol, depends on a vital and dynamic relationship between the patient as test subject and the physician as clinical scientist: "A healthy research community is ... dependent on the integrity and creativeness of the individual doctor-patient partnership, committed to a joint responsibility both for the study design and also to its implementation" (Tobias et al., 2000, p. 1372).
Another vital prerequisite for the success of a clinical trial is an administrative structure that is effective (Sackett, 1983). That structure must be able to manage the daily activities and problems, especially for trials that involve more than a single center. Joint ownership of a trial, especially for a multi-center trial, is vital for maintaining high-level quality of output among the different participants. Moreover, the trial's directors must hire top-quality personnel, especially laboratory and clinical participants and statisticians, to conduct the study and to use the most current and best technology available. Finally, besides being effective a trial's administration must also be efficient, especially in terms of a trial's cost and utilization of resources. Sufficient financial support of a trial is critical not only for conducting a top-quality study but also for the necessary follow-up studies.
The final prerequisite involves a valid "trial architecture" (Sackett, 1983, p. 69). A major problem with most clinical trials is that they are often invalid because a confounding variable is unaccounted for and produces bias-"the arrival at a conclusion that differs systematically from the truth" (Sackett, 1983, p. 69). Sackett identifies several ways to avoid confounding variables, from restricting inclusion of particular patients within a study to random allocation of patients within the various control and experimental groups. Randomization is the best way to avoid bias, although ethical concerns may surface over randomizing a trial for a highly morbid disease in which the drug or procedure is possibly efficacious. Besides randomization, controlled and blinding groups are also important for eliminating confounding variables that lead to bias.
10.1.1.2 Four Phases of Clinical Trials
The first phase of a clinical trial involves a modest group of around forty to eighty healthy volunteers and lasts for about one month. The goal of this phase is to determine the safety of a drug and the maximum tolerance of volunteers to it. Generally these goals are accomplished by an ascending dose-response test. The protocol involves administration of the drug, after which blood samples are drawn at different times to determine the drug's pharmokinetics, including its absorption, distribution, metabolism, and excretion. The type of information obtained from this phase involves a drug's physiological effects, especially any side effects or toxicities, and the maximal dose tolerated by volunteers.
The second phase involves a large group of well-screened patient volunteers of around two hundred, suffering from an appropriate disease. The goal of this phase is to determine a drug's efficacy, as well as its safety. This phase can last for several months. Again, the pharmacokinetics, as well as the efficacy and safety, of the drug are determined at various doses. Generally, the studies are double-blind, randomized, contain several control groups, and are often controlled for placebo effect. This phase is really a pilot study with patients selected by strict and well-defined criteria to determine whether phase III studies are warranted.
The third phase mimics the anticipated treatment regime in terms of duration and design. The size of the patient population is in the hundreds to thousands and the criteria for selection of test subjects are not as strict as in a phase II trial. The goal of this phase is to determine both the efficacy and safety of the drug at a particular dosage determined from the second phase. The design of a phase III clinical trial is the randomized, double-blind, placebo-controlled trial. If this phase of the trial is successful, then FDA approval of the drug for therapeutic use generally follows.
Although not required by the FDA, a fourth phase may be conducted to monitor a larger and more diverse patient population for both the efficacy and safety of the treatment. This phase can last for years and may provide data on the general use of a drug. Moreover, the criteria are even less strict for selecting patients than in phase III trials. For example, patients with co-morbid diseases, such as diabetes, may participate in a phase IV trial. Another important goal of this phase is pharmaco- economic data, especially if another treatment modality is available.
10.1.2 Randomized, Double-Blind, Concurrently Controlled Clinical Trials
The development of the standard clinical trial enjoys a history that stretches back to the eighteenth century, beginning with the French royal investigation into mesmerism (Green, 2002).2 In that trial, the royal investigation, headed by Benjamin Franklin, introduced two important elements of the current clinical trial: "sham intervention and subject ignorance about the bogus nature of that intervention" (Green, 2002, p. 311). An RCT prototype eventually emerged in mid twentieth century, with the streptomycin clinical trial that cemented the biomedical model for medical knowledge and practice (Doll, 1984).' "The streptomycin trial demonstrated," according to Christensen and Hansen, "that therapies can be evaluated in an empirical and experimental manner and require validation regardless of subject" (2004, p. 68).
The evolution of the RCT to its current state has included randomization, blinding, and control groups to remove possible bias and confounding variables that might comprise the integrity or validity of a trial's results, in terms of determining a treatment's or procedure's actual efficacy and safety (Green, 2002; Lilienfeld, 1982). RCTs achieve their objectives in terms of "ceteris paribus, i.e., `all other factors being equal"' (Lilienfeld, 1982, p. 3).
10.1.2.1 Bias and Placebo Effect
The main purpose of RCTs is to eliminate possible biases, such as selection bias, allocation bias, assessment (or observer or information) bias, or stopping bias, which can compromise the comparison of the results between the experimental and control groups (Matthews, 2000). Bias is a "distortion of judgment, or action, based on personal preference or a wished-for result" (Spodick, 1982, p. 21). Selection bias refers to entering a patient into a controlled or treated group preferentially: "Selection bias can occur when the decision to enter a patient into an RCT is influenced by knowledge of which treatment the patient will receive when entered" (Matthews, 2000, p. 14).
Allocation bias refers to a preferential distribution of subjects with a certain prognostic indicator, such as a subject's immunological competence, to either the test or control group. The skewing of allocation is often the result of the stochastic nature of simple randomization, while more robust means of randomization often reduce this bias. Assessment bias is the result of subjective evaluation or assessment of the trial's outcomes: "If the observer knows the treatment being given to the patient and if the measurement of an outcome variable contains an element of subjectivity, then it is possible that the value of an observation might be influenced by knowledge of the treatment" (Matthews, 2000, p. 19). Finally, stopping bias can be introduced when a trial is conducted until a significance difference is obtained between the treated and controlled groups.
Besides bias, another important factor that could influence or invalidate a clinical trial's results is the placebo effect (Macedo et al., 2003; Papakostas axnd Daras, 2001).4 Although the placebo effect was recognized for centuries, it took on greater importance in the last several decades vis-d-vis evidence-based medicine. Although there are a number of definitions proposed for the placebo effect, there is no current consensus definition (de Craen et al., 1999; Macedo et at., 2003). According to David Cockburn, "the placebo effect can be measured but not adequately explained" (2002, p. 1). At best, the effect is defined in operational terms as "the difference in outcome between a placebo treated group and an untreated control group in an unbiased experiment" (Gt tzsche, 1994, p. 925).
The placebo effect is an important factor in many therapeutic encounters between patients and physicians. "Even without a consensual definition, and assuming that the placebo effect does not seem to be fully dependent on a placebo administration," claim Macedo and colleagues, "one issue seems unquestionable: the placebo effect is present in clinical practice and in clinical trials, no matter which name we choose to call it" (Macedo et al., 2003, p. 337).
The placebo effect is a general result of therapeutic protocols to treat diseases. It certainly contributes to some extent in almost all protocols and most likely cannot be eliminated entirely from them, although randomization and blinding can minimize its role or effect. Although it cannot be explained in mechanistic terms, the placebo effect appears to be a function of the mind itself since it cannot be elicited in an unconscious patient. The main theories to account for the placebo effect are classical conditioning, response expectancy, and psychoneuroimmunological response (de Craen et al., 1999; Papakostas and Daras, 2001).
Randomization is by far the least problematic means by which to avoid bias.' It ensures that "chance alone assigns a patient to a particular treatment" (Spodick, 1982, p. 21). In the 1920s and 1930s, R.A. Fisher championed its significance and necessity and randomization became accepted and mandatory in clinical medicine after 1940 (Green, 2002; Lilienfeld, 1982). RCTs must be conducted such that the test subjects are randomly assigned to either experimental or controlled groups. Briefly, randomization is achieved by assigning treatment and subjects according to standard protocols, as simple as flipping a coin, by using random number tables, or by computer generated random numbers.
Randomization can remove bias, such as selection or assessment bias, which is associated with age, sex, social status, among other confounding variables. Such bias could easily nullify the significance of a clinical trail's results. However, "randomization limits the expression of the various forms of bias that might otherwise shift more of those subjects who will have better outcomes into one or another of the treatment groups" (Heaney, 1991, p. 105). Finally, randomization permits the results from the treated and controlled groups to be compared in order to determine causation: "Randomization is the means by which the ability to state that the difference in treatment groups is caused by the difference in treatments is achieved" (Matthews, 2000, p. 10).
Randomization cannot avoid all possible sources of bias and it cannot guarantee the validity of a clinical trial's results. Sackett identifies several issues that need to be kept in mind in order to avoid biasing a clinical trial, even tough the trial is randomized (1983, pp. 71-72). The first is that ancillary techniques not directly associated with a drug or procedure being tested must also be performed on the control group. The next issue is avoiding any exposure of the control subject to the test drug or protocol. Another issue is to avoid any contamination of the control with test subjects who have received the drug or procedure. Finally, as noted above, simple randomization cannot eliminate allocation bias, "because knowledge about which treatment is about to be assigned can influence whether or not a patient is deemed suitable to enter a trial, as well as how hard the physician tries to persuade a reluctant patient to volunteer" (Chalmers, 1981, p. 330). It is imperative, then, that the randomizing itself is blinded by an appropriate technique.
Another important factor in clinical trials is blinding, which is employed to remove bias, especially assessment bias, associated with a clinical trial.' Blinding is a "[p]lanned concealment from the physician, the patient, or both, of the nature of the actual substance being tested" (Spodick, 1982, p. 21). The term "blind test" was introduced into the literature in the 1930s by Harry Gold and colleagues, who adapted it from the work of British psychologist H.H.R. Rivers (Green, 2002; Strong, 1999)."
A single-blind trial refers to the situation in which the clinician knows which patient is receiving the drug or experimental treatment, while the patient does not. This type of blinding is often used when the treatment has serious side effects that require constant monitoring. The obvious problem with this type of blinding involves possible subjective or subconscious communication or behavior on a clinician's part, which may lead to a placebo effect. To remove this bias, a doubleblind trial is often conducted in which both investigator or clinician and patient do not know who is receiving the drug. It is the randomized, double-blind controlled clinical trial that is the preferred method for testing the efficacy of a drug or procedure.
In a triple-blind study, another level of blinding is introduced in terms of blinding the person who assigns which group receives treatment or it may refer to the statistician or data analyst and/or the person interpreting or assessing or collecting the results who is kept ignorant of which group represents what. If the statisticians or assessors are two separate persons and if both are kept blind of the assignments, then the study is quadruple blind.'
10.1.2.4 Concurrently Controlled Groups
Besides randomization and blinding, bias is also eliminated by adding concurrent controlled groups to the experimental trials (Matthews, 2000). There are at least three ways by which to control a trial. The first is the traditional control-group which does not receive the same experimental protocol or treatment as the test or experimental group, e.g. those receiving the drug. The comparison of the treated group and the non-treated, controlled group allows a clinical scientist to conclude whether the treatment or drug is effective with respect to the disease.
The second sense of control involves neutralizing any placebo effect, generally by undergoing the same experimental protocol but without receiving the active treatment. If it is a drug that is being tested, then one gives a pill or placebo that does not contain the active ingredient-traditionally such pills contain sugar.L° There are also active placebos, which mimic the effects or side-effects of the treatment being tested. This control-group permits the clinical scientist to determine whether any improvement in the disease is the result of a psychological effect due to general manipulation or is the result of the treatment.
The final control is an active control-group, which receives a different treatment protocol that is effective but not the same as the experimental treatment (Pellegrin and Nesbitt, 2004). The purpose of this control is comparison of the efficacy of the experimental or new treatment to a known or an older treatment and to demonstrate that the newer treatment is better than or superior to a current treatment.
10.1.3 Other Clinical Trails
Although RCTs are considered the "gold standard" in medical epistemology, not all medical knowledge is justified by such trials (Hennekens and Buring, 1987; Thagard, 1999). There is often no single or possible way by which to run a clinical trial to obtain the necessary information concerning treatment efficacy for every drug or procedure. For example, the current recommendation for the daily dose of fluoride is half of the earlier recommendation of 75mg/day. "How is it," queries Robert Heaney, "that we know this? There has been no randomized, controlled trial dealing with this issue. The answer is partly the shared experience of the community of clinical investigators working with fluoride for the past 20 years" (1991, p. 105).
Moreover, "in everyday practice a multitude of management decisions must still be taken without good evidence" (van Gijn, 2005, p. 69). The reasons for a lack of such evidence includes that a trial may be impractical or not feasible or that a trial's results may be equivocal. Finally, good clinical practice often means doing the best with what is available, even if it is not evidence based but antidotal: "Evidence based medicine is not restricted to randomized trials and meta-analysis. It involves tracking down the best external evidence with which to answer our clinical questions" (Sackett et al., 1996, p. 72).
Thagard (1999), utilizing Hennekens and Baring (1987), identifies several different types of medical studies that are divided into two major groups: descriptive and analytic studies. The first descriptive study consists of correlational studies, in which the disease frequency from different populations is compared for a specific time period. The next type of descriptive study consists of the single case study, which involves a detailed description of a single patient and the course of the disease. The final descriptive study is the cross-sectional survey, in which large amounts of data are collected on a particular risk factor, such as cigarette smoking, and the incidence of disease, in a particular time period.
The analytic studies involve "an explicit comparison of the risk factor of disease between those exposed to a factor and those not exposed" (Thagard, 1999, p. 76). Besides RCT, there are two analytic studies. The first is the case-control study, in which a population of patients with a disease is compared to a control population not expressing the disease. The next analytic study is the cohort study, in which a population exposed to a risk factor and one not exposed to it are followed over time to determine the disease's development and expression.
10.2 Biomedical Technology
A major epistemological problem with technical innovations is their assessment. Whereas a drug's efficacy is determined by RCT, no standard method of assessment is available or accepted for technical devises, especially new surgical protocols. "For evaluating surgical operations and multimodality therapeutic regimes such as intensive care, RCTs are much less consistently employed, leading to editorial scolding," according to Jennett, "on both sides of the Atlantic" (1986, p. 233). Surgical research presents special problems for the application of RCT not encountered in drug research, such as postoperative procedures and the placebo effect.
Although surgical procedures do lend themselves easily to RCT analysis, an analysis of the surgical literature revealed that around 40% of surgical protocols could be evaluated using RCT (Solomon and McLeod, 1995). Even though limitations exist, the use of RCT is still encouraged in the evaluation of surgical protocols: "the RCT is the undoubted gold standard for the evaluation of medical therapies. This holds true for surgical operations too, and we definitely encourage every surgeon to conduct such studies" (Sauerland et al., 1999, p. 426). However, others claim that the limitations of RCT hinder it from assessing adequately surgical protocols. For example, Nick Black (1999) lists several limitations, including lack of generalization of RCT to individual patient and oversimplification of the application of RCT to surgery, and asserts that RCT is nothing more than a "passing fad." "
Jonathan Meakins (2002) proposes strategies by which to evaluate surgical protocols, which depend on rules of evidence that are tailored for surgical research. He locates room within the current rules of evidence initially proposed by Sackett (1989) for surgeons to utilize outcomes from observational studies to determine the efficacy and safety of surgical procedures that are not equipoise. "The only way to reduce the `do it my way' approach that has plagued operative surgical research," according to Meakins, "is to define the best data and where it is seen to be absent and to do the studies required to get the answer" (2002, pp. 401-403).
To accomplish a systematic analysis of surgical research, especially in terms of the application of RCT, Meakins (2002) advocates an initial step in which the problem under consideration is reviewed systematically and exhaustively, to determine the adequacy of prior solutions and whether they make sense given current conditions. If an RCT is not possible, then observational studies that are prospective and nonrandomized must be conducted, in which outcomes are defined ahead of time. The evaluation of outcomes must be conducted by a third party to reduce bias.
Besides these more general problems medical technology also produces epistemic problems, particularly with respect to knowledge about the patient. Although Cassell agrees that today our knowledge, especially scientific knowledge, is technical, he bemoans the fact that "For medicine, the scientific knowledge and subsequent technology developed in response to the challenge posed by sickness and suffering has assumed an actuality more convincing than the reality of sick persons themselves" (1997, p. 75). The result is that technology drives a wedge in between the patient and the physician, with the physician focusing not on the patient's suffering as a person but on the patient's pain caused by a diseased body part.
Contemporary medical knowledge and practice is often limited to the diseased part and does not include the patient's suffering. The solution, according to Cassell (1991), is not to jettison modern medical technology but to teach physicians, whom he considers to be "the primary instruments of diagnosis and treatment," to resist the development of technology for its own sake, to tolerate a certain amount of ambiguity and uncertainty in medical practice, and to share power with patients.
10.3 Narrative Therapeutics
Narrative looms large in the discussion of the humanistic or humane models for medical knowledge and practice, especially in terms of the epistemological analysis with respect to reasoning, judging, and explaining. Moreover, while narrative is important for diagnostic procedures it is also critical with respect to therapeutics. The physician plays a crucial role in not only obtaining a full understanding of the patient's illness experience in order to make a correct diagnosis but also in providing an adequate or effective therapeutic protocol or procedure.
Besides a patient's narrative of the illness experience, a physician is also called upon to provide a medical rendition of it. There is a fundamental difference between the two narratives, a difference that is important for effective therapeutics (Hunter, 1991). If that difference is not respected, especially by the physician, then the patient may not be healed fully or even adequately. Traditionally, a physician interprets a patient's narrative by transposing or translating it into medical terms and concepts. "The medical interpretation of the patient's story," according to Hunter, "bears great power for healing" (1991, p. 124). As for the justification of narrative diagnostic knowledge, however, the empirical warrant for such healing power is not possible to justify epistemologically.
Part of the problem for the biomedical model is that a physician's official, medical rendition of a patient's narrative of the illness experience is assumed to capture completely a patient's story for effective therapeutics. Howard Brody protests: "Physician's should not simply assume that the medical story is the patient's story or that no negotiation between the two stories is needed for the patient to receive the full benefit of the medical work" (2003, p. 10). Rather for Brody, as well as for Hunter and other humanistic practitioners, the physician has to pay particular attention to the meaning embedded in a patient's narrative and to provide hope for the patient, especially in terms of a faithful and an accurate prognosis. Studies on the placebo effect, for example, demonstrate that a physician's ability to provide an accurate account and projection of a patient's illness significantly affects a patient's recovery (Brody, 2003). According to Brody, "the ability to prognosticate accurately, to tell the story of future illness, maintains a sense of control and thus may symbolically, even if not pharmacologically, lead to an enhanced healing" (2003, p. 15).
Fundamentally, narrative therapeutics involves the restoration of a patient's broken life-narrative. Hunter refers to Sigmund Freud's analysis of his own work in psychiatry as repairing a patient's narrative. "The patient presents a malady in both body and story," notes Hunter, "hoping for a rewriting of the narrative of illness in and through the medical narrative, an interpretation that will lead to an understanding of the symptoms and thereby to their relief and cure" (1991, p. 130).
Brody also acknowledges the importance of a physician's ability to retell a patient's story of illness, in order to provide effective therapy. "Patients come to physicians with broken stories," claims Brody, "as much as with broken bones and broken bodies" (2003, p. 16). Importantly, the retelling of a patient's story of sickness cannot be reduced to a formula, as is often the case in the biomedical model where the physician asks standard questions concerning a patient's disease story and expects only pertinent facts. "In fact, if [narrative therapy] is seen as a formula or used as a recipe," according to Gerald Monk, "clients will have the experience of having things done to them and feel left out of the conversation" (1997, p. 24). The result is that therapy is less efficacious.
Although there is no standard protocol for narrative therapy to ensure it epistemologically, especially in terms of a set of narrative questions, there is a "form" to ensure its effectiveness, especially when compared to traditional biomedical therapy (White and Epston, 1990). This form or structure of narrative therapy is composed of several different components. The first is that narrative therapy takes seriously or privileges a patient's story of illness or lived experience, rather than the universal, medicalized story of a biomedical practitioner. For in a patient's story are embedded the meanings associated with the illness that are important for therapeutic success. The next component consists of the temporal sequence of a patient's narrative, in which meaning can be reshaped.
The third component pertains to the language utilized in a narrative. Instead of the indicative mood of biomedical discourse a narrative therapist utilizes "the subjunctive mood to create a world of implicit rather than explicit meanings, to broaden the field of possibilities through the `triggering of presupposition,' to install `multiple perspective,' and to engage `readers' in unique performances of meaning" (White and Epston, 1990, pp. 81-82). Associated with this dimension is an invitation for multiple readings of a patient's narrative, rather than the standard, univocal text of the biomedical physician.
The fourth component of narrative therapy involves personal, active agency. Rather than subjugating the patient as a passive agent, as in the biomedical model, narrative therapy engages the patient as an active agent in the healing process, especially in terms of revising a broken narrative. A patient's narrative reflects "a world of interpretative acts, a world in which every retelling of a story is a new telling, a world in which persons participate with others in the `re-authoring,' and thus in reshaping, of their lives and relationships" (White and Epston, 1990, p. 82). The final component involves the relative positions of physician and patient. Instead of a physician being "above" a patient, as in the biomedical model, the patient is a vital and important co-producer of the healing narrative. For narrative therapy, the patient is not objectified as in the biomedical model but rather is personalized.
Brody (2003) also argues that joint construction of the physician's and patient's stories of sickness is imperative for successful and effective therapy. To that end, he identifies four criteria required for constructing the joint narrative. The first involves the process itself for jointly constructing the therapeutic narrative. A physiciandominated story of the patient's illness is generally ineffective. "Ideally," according to Brody, "the physician's role in `coauthorship' consists of hints, nudges, and offers of bits of narrative raw material. The patient," he continues, "is the best person to put the pieces together, in a way that allows her finally to own the resulting story, a story about what is happening in her life" (2003, pp. 16-17). Hunter (1991) also comes to a similar conclusion. She claims that physicians must restore the narrative to the patient to make it his or her own again. The next criterion is that the narrative must be in line with the best of biomedical knowledge. The narrative is not to make the patient simply feel better but to heal him or her.
The third criterion of successful narrative therapy is a patient's responsibility to own the joint healing narrative. According to Brody, "the ideal healing narrative is not merely `I know what has caused my problem, I feel that others care that I get better, and something can be done to control what ails me.' Rather," he contends, "the ideal narrative continues, `...and I see myself actually taking the concrete steps I know to be necessary to carry out the program of treatment that I have (ideally) agreed to"' (2003, p. 17). For example, if a patient agrees that part of the healing narrative is to reduce his or her cholesterol level, then that patient must comply with that part of the healing narrative and follow the protocol(s) needed to reduce cholesterol.
The final criterion depends on whether the illness is acute or chronic. For acute illness, the joint therapeutic narrative assists in helping a patient return to a normal life which he or she lived prior to the illness. For chronic illness, the therapeutic narrative is more complex and demanding. "The patient's task," writes Brody, "is both to grieve the loss of the old life story, which now can never be completed the way the patient had intended, and also to construct a modified life story that carries on within the realities and constraints forced by the sickness" (2003, p. 17).
The biomedical approach to the generation of therapeutic knowledge and its justification is through RCT and biomedical technology. These epistemic instruments assure biomedical practitioners that their therapeutic protocols and techniques are both effective and safe. Although only a small percentage of therapeutic intervention to date is justified by these instruments, the goal is to justify all medical practice through them. The therapeutic care of biomedical practitioners then depends upon a highly technical story that is often incommensurable with a patient's existential or emotional needs story.
"Expectations of care that ignore the difference between the physician's and patient's stories," cautions Hunter, "contribute to the widespread dissatisfaction with contemporary medicine" (1991, p. 123). In response to that dissatisfaction, humane physicians practice a narrative therapy that incorporates a patient's existential or emotional needs into healing stories. Although such stories cannot be justified through the highly technical instruments used to justify biomedical therapy, narrative therapy proponents claim that its effectiveness mimics the benefits associated with the placebo effects (Brody, 2003). Consequently, narrative therapy goes a long way to relieving the quality-of-care crisis provoked by the biomedical model.