Jeremy Howick: The Philosophy of Evidence-Based Medicine
- First Online:
- Cite this article as:
- Marcum, J.A. Medicine Studies (2011) 3: 125. doi:10.1007/s12376-011-0069-1
- 86 Views
Since its introduction several decades ago, evidence-based medicine (EBM) has often served to polarize the medical community. On the one hand are clinicians who maintain that clinical expertise and pathophysiology are adequate for practicing effective clinical medicine, while on the other hand are champions of EBM who claim that medical practice should be based solely on the best evidence currently available, especially evidence obtained from randomized clinical trials (RCTs). Indeed, proponents of EBM claim that EBM represents a Kuhnian revolution with the replacement of the older pathophysiology paradigm with the newer, incommensurate evidence-based paradigm. Until now, there has been no sustained philosophical analysis of EBM. In this book, Jeremy Howick, a philosopher of medicine at the Centre for Evidence Based Medicine, University of Oxford, provides a considerably insightful analysis of the philosophical issues surrounding EBM and the debate it has aroused among medical pundits and commentators. Indeed, EBM’s popularity and the debate surrounding it are ripe for philosophical investigation and analysis. What sets Howick’s book off from other philosophical analyses of EBM is his defense of an expanded notion of what constitutes quality clinical evidence. Howick divides the book into three parts, with a concluding chapter on EBM’s future.
In the first part, consisting of three chapters, Howick introduces the notion of EBM and explores its relationship to traditional medicine. In chapter 1, he briefly presents the pyramidal hierarchy of evidence informing medical practice obtained from three distinct methods: clinical expertise/mechanical reasoning representing the pyramid’s base, RCTs its apex, and observational clinical studies its middle. The main question animating Howick’s philosophical investigation of EBM is whether the claim of EBM proponents—that the EBM method provides superior evidence compared to the other two—is justified. In chapter 2, he takes a historical approach to explicating the nature of EBM, beginning with its initial definition and following its development in response to critics until EBM integrated best research evidence with clinical expertise and patient values and circumstances. However, he maintains that EBM proponents claim best evidence still refers to clinical research and not to mechanistic reasoning or clinical expertise or observational studies. Indeed, he pursues what constitutes quality evidence for clinical decision making in the third chapter. According to Howick, quality evidence allows a clinician to make “clinically effective” decisions in which treatment benefits outweigh harm and in which treatment is patient appropriate and represents the best available option. With this preliminary introduction, he then analyzes the assertion that only RCTs provide the best evidence for making clinical decisions, compared to observational studies.
In the second part, Howick proposes a method, in the fourth chapter, by which to assess the quality of evidence obtained from clinical trials. The method involves the ability of evidence to eliminate rival hypotheses through reducing the number of confounding factors and biases. He also addresses the fallibility of background knowledge, at the end of the chapter, and the problems it raises for assessing the relative strength of evidence. In the next four chapters, he examines the impact of randomization, double blinding, and placebo controls in assessing the quality of evidence vis-à-vis rival hypotheses elimination. In chapter 5, he introduces the paradox of effectiveness in which dramatic therapies, such as the Heimlich maneuver, have not been or would not be subjected to random clinical trials. Although randomization does rule out self-selection and allocation biases and thus provides stronger evidence than observational studies, still the latter studies, Howick argues, can provide adequate evidence to rule out competing hypotheses if the overall outcome or effect exceeds the combined effect of possible confounding factors and biases. In chapter 6, he introduces the Philip’s paradox in which double-blinded studies are not necessary to support dramatic therapies. For justifying such therapies, blinding is simply inconsequential or benign. Unfortunately, masking is seldom successful, because of what he calls “malicious unmasking,” in which researchers can often determine which arm of the trial is experimental or control.
In the final two chapters of the second part, Howick examines placebo controls in terms of the quality of clinical evidence. In chapter 7, he discusses the nature of placebos, especially in terms of defining them. He proposes what he calls a “legitimate placebo control,” which consists of two components. The first is that the placebo exhibits the relevant non-characteristic features of the experimental agent. If the placebo mimics the non-characteristic features of the experimental agent, then the agent’s characteristic effect is accurate. However, if it does not, then the agent’s effect could be underestimated or overestimated. The second component involves no other features that exceed the agent’s non-characteristic features. If such features infect placebos, then the placebos do not represent adequate controls for determining the agent’s effect. He then explores the problems associated with designing placebos for examining the therapeutic effect involved in exercise and acupuncture, which expose the practical limitations of designing legitimate placebos. He concludes that placebos should be considered treatments in their own right and that researchers should provide detailed information concerning placebos. In the next chapter on placebos, chapter 8, Howick contrasts placebos with active controls. He concludes that active controls are not inferior to placebos and can provide, under specific conditions, quality evidence for assessing therapeutic effect.
In the final part of the book, Howick transitions, in chapter 9, to examining whether mechanical reasoning and clinical expertise provide adequate evidence for making quality clinical decisions. In chapter 10, he takes up mechanical reasoning. Howick argues that although mechanical reasoning does not present evidence on par with clinical trials it can offer evidence through inference from how things work. In other words, if a particular molecule functions critically within a specific pathway, then drugs modulating its activity would warrant exploration for developing therapies for a disease in which that pathway is essential. However, such reasoning is fraught with two problems, according to Howick. The first is the existence of unknown factors, which might have an impact on the pathway that then alters results, while the other is the sheer complexity of mechanical processes in terms of their stochastic nature. To address these problems, he advocates high-quality mechanical reasoning that is sensitive to these problems. In chapter 11, Howick entertains whether clinical expertise provides evidence for robust decision making. Although he rejects the idea that it does provide evidence, citing Paul Meehl’s pioneering work, he does recognize several roles such expertise can play in clinical practice. For example, clinical expertise can help to integrate patient values and preferences with available therapeutic options or it can help enhance the placebo effect.
In a concluding chapter, chapter 12, Howick rehearses several claims concerning EBM, particularly the need to modify EBM to include mechanical reasoning and clinical expertise in clinical practice. He also explores two frontiers for EBM. The first is conflict of interest. He is concerned that funding and publication biases could compromise the quality of evidence obtained from RCTs. This is certainly a legitimate concern, given that both biases are prevalent in the biomedical community and can have a devastating impact on medical decisions using such compromised evidence. Unfortunately, with more and more clinical trials sponsored by industrial funds, no immediate solution to this problem appears on the horizon. The second frontier for EBM, claims Howick, is how best to situate EBM within the larger healthcare framework, especially in terms of preventive health care. EBM certainly functions well with respect to down-stream care but can EBM methodology be applied to evaluate studies for strategies to prevent healthcare problems. The answer to this question is not obvious, and Howick believes that it will take further development of EBM’s methodology to answer it.
Howick makes an interesting and important assertion about EBM: epistemology and ethics go hand in hand. What he means is that how we obtain medical knowledge carries with it ethical implications. For example, he claims that RCTs are ethically unsustainable if sufficient evidence is available from either mechanical reasoning or observational studies to support a drug’s efficacy. However, he fails to appreciate the role uncertainty plays in the discovery process and its ethical implications for clinical practice. For example, researchers use the principle of equipoise to justify not treating some patients with a new drug in a RCT. The principle states that researchers cannot guarantee that a new drug is better than an older one, at least not until the trial is conducted. In other words, uncertainty of the drug’s action, even though laboratory evidence suggests otherwise, is the underlying factor that justifies enlisting some patients into control groups. Moreover, analyses of series of clinical studies support the principle, which yields a curious paradox. On the one hand, equipoise drives discovery in that a trial’s success cannot be predicted prior to conducting it; while, on the other hand, it limits discovery in that roughly only half of conducted trials succeed (Djulbegovic 2009). Finally, uncertainty also influences the intersection of epistemology and ethics vis-à-vis EBM in terms of differences in which clinicians utilize best evidence to treat individual patients (Ghosh 2004).
Although I am sympathetic to the paradox of effectiveness, I believe it fails if a distinction is posited between complex and simple medical conditions or chronic and acute diseases. For example, the Heimlich maneuver—one of Howick’s favorite examples to illustrate the paradox—is a straightforward approach to a simple or acute medical condition (this is not to say that it is not a serious medical condition). Compared to a complex and chronic disease like cancer, choking is a condition that requires immediate attention and resolution, which involves dislodging the obstruction from the airway passage. What causes the problem and how to solve it are rather well known and unproblematic (this is not to say that the maneuver is not ingenious). In such cases, designing randomized and double-blinded clinical studies would not simply be unnecessary but immoral. However, for a disease like cancer, what to do is fraught with problems beginning with disease etiology to designing effective therapies. Here, the design of clinical trials is imperative, and approaches such as randomization and double blinding are essential to evaluate a therapy’s effectiveness. I do not think Howick disagrees with this; but, I believe this distinction should be made in order not to conflate between complex and simple medical conditions and diseases and then to support mechanical reasoning as comparable to RCTs.
Finally, Howick claims at the book’s beginning that the reader should be able to answer by its end the question, “What is the evidence for the EBM philosophy of evidence?” (p. 9). At the end of the book (p. 188), he reminds the reader of this question; however, he does not provide an answer to it, in terms of an alternative model to counter the pyramidal, hierarchical model of evidence that he claims proponents of EBM espouse. Although he provides various renditions of a parachute model for illustrating particular components of clinical practice, I believe that he has argued successfully for a relational model for evidence-based clinical practice, as depicted below in an alternative pyramidal model.
In this model, evidence from each of the sources contributes to or envelopes the practice of evidence-based clinical practice. Thus, clinical trials—whether randomized, observational, or comparative—contribute in their own way, within specific limitations, to what constitutes evidence for best clinical practice. Likewise, both clinical or mechanical reasoning and clinical expertise—taking into mind Meehl’s work—and patient values and circumstances provide valuable qualified evidence for clinical practice. Implicit in this model also are interacting relationships among the various sources, so that patient values, for example, can influence directly the design of clinical trials. Clearer depiction and explication of this model would have benefited the reader in terms of the author’s important question.
In conclusion, Howick provides an able philosophical analysis of EBM and an insightful discussion of the debate over the issue of evidence for practicing clinical medicine. The book is a very useful introduction to EBM, and we look forward to an updated edition as both EBM and the author’s understanding of it evolve.