Published online by Cambridge University Press: 29 March 2005
Should patients request antibiotics from their physicians for common upper respiratory infections?
Usually this question has been construed as a scientific and not as an ethical question. The scientific answer today is that most upper respiratory infections, including colds, bronchitis, and most cases of sinusitis, are thought not to require antibiotics.1
For examples of evidence-based guidelines, see Gonzales R, Bartlett JG, Besser RE, Cooper RJ, Hickner JM, Hoffman JR, Sande MA. Principles of appropriate antibiotic use for treatment of acute respiratory tract infections in adults: Background, specific aims, and methods. Annals of Internal Medicine 2001;134:479–86; Hickner JM, Bartlett JG, Besser RE, Gonzales R, Hoffman JR, Sande MA. Principles of appropriate antibiotic use for acute rhinosinusitis in adults: Background. Annals of Internal Medicine 2001;134:498–505; Gonzales R, Bartlett JG, Besser RE, Cooper RJ, Hickner JM, Hoffman JR, Sande MA. Principles of appropriate antibiotic use for treatment of uncomplicated acute bronchitis: Background. Annals of Internal Medicine 2001;134:521–9; and Gonzales R, Bartlett JG, Besser RE, Hickner JM, Hoffman JR, Sande MA. Principles of appropriate antibiotic use for treatment of nonspecific upper respiratory tract infections in adults: Background. Annals of Internal Medicine 2001;134:490–4.
For a guideline suggesting the use of penicillin, see Cooper RJ, Hoffman JR, Bartlett JG, Besser RE, Gonzales R, Hickner JM, Sande MA. Principles of appropriate antibiotic use for acute pharyngitis in adults: Background. Annals of Internal Medicine 2001;134:509–17. For analyses of evidence suggesting a lack of benefit with antibiotics in strep throat, see Del Mar CB, Glasziou PP. Antibiotics for sore throats? Journal of Paediatrics and Child Health 1998;34:498–9; and Graham A, Fahey T. Evidence-based case report. Sore throat: diagnostic and therapeutic dilemmas. British Medical Journal 1999;319:173–4.
If the question were to be construed as an ethical question, the dominant approach in bioethics, based on respect for patient autonomy, would argue that patients certainly might request antibiotics, and moreover, physicians might well be obligated to provide them, after sharing all the information about their risks and benefits. By contrast, those who hold to the doctrine of futility would argue that using antibiotics for viral infections is the very height of futility, and physicians of conscience ought never to prescribe futile interventions, no matter what the patient requests.
Here I will develop the answer to the question, “Is it ethical for patients to request antibiotics for common upper respiratory infections?” as, “Mostly not.” I will try to justify this answer in terms of a patient-centered ethic derived from the practice of evidence-based medicine (EBM). Both the “not” and the “mostly” will turn out to be indicative of important features of EBM, properly understood.
EBM can and has been defined in various ways and has numerous champions who do not always see eye to eye.3
Evidence-Based Medicine Working Group. Evidence-based medicine: A new approach to teaching the practice of medicine. JAMA 1992;268:2420–5; Sackett DL, Richardson WS, Rosenberg W, Haynes RB. Evidence-Based Medicine: How to Practice and Teach EBM. New York: Churchill-Livingstone, 1997; Tonelli MR. The philosophical limits of evidence-based medicine. Academic Medicine 1998;73:1234–40.
Many promoters of EBM equate the practice with doing what is consistent with the results of randomized controlled trials or meta-analyses of randomized trials. This, I believe, is a perversion of EBM. In many instances, a randomized trial will be the best evidence and so should be given great weight. But in a number of cases the best available evidence will not consist of randomized trials. For many aspects of medicine, especially those with significant ethical import, qualitative studies are better suited than randomized trials to provide the best possible evidence, because qualitative studies capture better what patients think and why. In many areas of medicine, randomized trials have not been conducted or are of poor quality or cannot be properly generalized to the patient in question. In such instances traditional medical judgment, based on understanding disease mechanisms, may be the “best” evidence.
Many well-designed randomized trials have now demonstrated that in many diverse patient populations, antibiotics provide no benefit compared to placebo in the treatment of acute bronchitis. EBM would teach us that in deciding what to do, these randomized trials would take precedence over all of the following:
The bulleted list reflects the sorts of reasons why physicians have typically practiced the way that they have. EBM challenges physicians to stop appealing to these less adequate types of evidence when better evidence is at hand.
As commonly practiced, EBM has a large ethical blind spot. It is usually put forth as a totally physician-centered activity. Physicians are supposed to review the medical literature and decide what to do. That implies that physicians get to decide what sorts of clinical outcomes matter most to patients. If, for example, “evidence-based” guidelines state that physicians should not prescribe combined estrogen–progestin products for postmenopausal women who have not undergone hysterectomy, the guideline simply assumes that women want to avoid any risk of heart disease and breast and colon cancer, and care much less about relieving menopausal symptoms or avoiding osteoporosis. But this ethical blind spot can be filled in so long as EBM advocates adopt a commitment to a shared decisionmaking strategy, in which patients are invited to determine the clinical outcomes that matter most.4
Holmes-Rovner M, Llewellyn-Thomas H, Entwhistle V, Coulter A, O'Connor A, Rovner DR. Patient choice modules for summaries of clinical effectiveness: A proposal. British Medical Journal 2001;322:664–7.
Bodenheimer T, Lorig K, Holman H, Grumbach K. Patient self-management of chronic disease in primary care. JAMA 2002;288:2469–75.
Why would EBM place an ethical obligation on patients? Two reasons can be given, grounded in professional integrity and in just resource allocation.
Patients have an interest in their physicians practicing medicine with the greatest degree of professional integrity—assuming that when the sick person approaches the physician for help, she expects to receive not merely certain services as freely negotiated between consenting adults, but rather good medical care. Technical competence is one aspect of professional integrity.6
Miller FG, Brody H. The internal morality of medicine: An evolutionary perspective. Journal of Medicine and Philosophy 2001;26:581–99.
The second reason why patients ethically should be familiar with and support EBM-based practice arises from the scarcity of resources to provide and pay for health care. Not all healthcare interventions that patients might wish for can be provided for all patients. In the face of scarcity, we need fair principles of allocation. We would seem to treat the most patients as fairly as possible if we eliminate costly treatments that have been shown in well-designed scientific studies not to produce the outcomes that patients themselves desire.
Patients who have a basic understanding of EBM principles and who accept well-grounded recommendations from their physicians not to use certain interventions are therefore good healthcare citizens. They use their fair share, but no more than their fair share, of the socially available pool of healthcare resources. When the scientifically demonstrated likelihood of benefit falls below a reasonable threshold, they stop demanding resources and leave those resources in the social pool for the benefit of their fellow citizens. All citizens benefit because fewer resources are wasted on useless interventions and more resources are preserved for addressing health problems that are truly amenable to improvement through available technology.
Appeals to professional integrity and the just allocation of limited resources therefore explain the “not” part of my “mostly not” answer to the question as to whether patients should request antibiotics for viral respiratory infections. Why, then, the “mostly”?
EBM requires physicians to use the best available evidence to make treatment recommendations to patients. In many (but not all) cases, the best available evidence will consist of randomized controlled trials or meta-analyses. It follows that the recommendations that flow from EBM are limited by whatever limits randomized trials might have as a way of gaining medical knowledge.
EBM and similar applications of clinical epidemiology to medical practice have been criticized for replacing a focus on the needs of the individual patient with data drawn from large populations.7
Kassirer JP. Managing care—Should we adopt a new ethic? New England Journal of Medicine 1998;339:397–8.
The physician who believes that he is an advocate for the individual patient might still object, “When I recommend drug A for my patient's hypertension, I do so because of his various individual characteristics—for instance, that he's Hispanic, overweight, and somewhat anxious; it's my clinical judgment that drug A tends to work better for patients with those characteristics.” But this does not negate the point. All this physician is saying is that if one had a large population of Hispanic, overweight, hypertensive patients with anxiety states, and then randomized them to either A or B, the A group would do better on average. The fact that no one has or perhaps ever will actually carry out such a randomized trial does not negate the conceptual point that the physician relies on the hypothetical results of such a trial to make the recommendation. On this analysis, the only difference between basing one's therapeutic recommendations on EBM or on “individualized” clinical judgment is that the basis for the recommendation is explicitly made clear in EBM.
The implication of these considerations is that the individual patient always has potential grounds for rejecting the physician's evidence-based treatment recommendation—the randomized trials that were carried out (or that the physician hypothesized, for that matter) were performed on a population of patients unlike the present patient in some important way. A related objection is that all a randomized trial can usually show is the average impact of a treatment upon a group of patients; individual patients for whatever reason may have an especially good or an especially bad response to treatment, but on average the results of the trial may show no significant difference compared to placebo.
What does all this have to do with antibiotic prescribing? It is not illogical for an individual patient to receive the physician's EBM recommendation not to use antibiotics for a particular infection and yet object that he has other evidence that tends to show that he would benefit from antibiotics. Usually, in my own practice, this argument has come in the form of a recital of numerous previous episodes of similar illness that did not get better with other forms of treatment, but that got better promptly as soon as an antibiotic was given. In one study of patient interactions with physicians, such a rationale was employed in 13% of visits in which an antibiotic for an apparently viral infection was discussed.8
Scott JG, Cohen D, DiCicco-Bloom B, Orzano AJ, Jaen CR, Crabtree BF. Antibiotic use in acute upper respiratory infections and the ways patients pressure physicians for a prescription. Journal of Family Practice 2001;50:853–8.
Should the physician grant the patient “exceptional” status based on such a recital of past events? Assuming that the patient wishes to be the “good healthcare citizen” previously described, the patient is ethically as well as scientifically obligated to consider alternative explanations for the previous illness episodes, besides the effectiveness of antibiotics. The alternative explanation takes into account the post hoc fallacy and regression to the mean. Imagine that the patient is suffering from a viral illness that will, in fact, resolve without antibiotic treatment. The typical patient does not seek medical care right away, but waits several days or a week or more. As a rule, the decision to seek care comes after a worsening trend. That means that purely by statistical chance, the vast majority of patients see the physician around the time that their self-resolving illnesses are at their worst. If nothing at all was done, the natural history of the infection would lead to slow improvement over the next several days following the visit. If at the visit an antibiotic is prescribed and promptly taken, the patient will naturally conclude (erroneously) that the antibiotic caused the improvement.
If physician and patient discussed this possibility candidly and thoughtfully, many supposedly “exceptional” patients would be forced to admit that this could in fact explain why they had become convinced that they “need” antibiotics to recover from an infection like this one. As good healthcare citizens, they should then be willing to try to ride out this infection without the immediate use of antibiotics. A strategy some physicians are starting to favor is the so-called deferred prescription. In this plan, a patient is given the prescription for an appropriate antibiotic, but advised not to fill the prescription for several more days, and then only if symptoms continue to worsen or if specific new, worrisome symptoms (like fever) develop. Although antibiotic use remains high among patients given deferred prescriptions, the approach nevertheless reduces antibiotic usage significantly compared to current practice.
There still remains the occasional patient who can give detailed descriptions of numerous previous episodes of similar infections in which antibiotic use was finally followed by resolution, and where the time course and other factors make the standard explanation of regression to the mean unlikely. The physician in such a case has only two options—say to the patient in effect, “I am so wedded to EBM that I don't believe what you are telling me,” or admit that this patient might reflect an outlier population who does, in fact, derive benefit from antibiotic use for certain upper respiratory infections. If the physician takes the first position, she is overreading what information can actually be derived from EBM and randomized trials. How certain can we be, when all is said and done, that no upper respiratory infection is ever caused by a bacterium that could be effectively treated with antibiotics? So, in the spirit of shared decisionmaking (which demands before all else that the physician be willing to listen sympathetically to what the patient reports), the second position seems the only legitimate one.
The inherent limits of randomized trials leads to the “mostly” in “mostly not” as a rule for whether patients are ethically justified in requesting antibiotics for upper respiratory infections. I would hope that we could respect the (apparently) legitimate requests of this small cadre of patients and still reduce unnecessary antibiotic use and the attendant selection of resistant microorganisms by as much as 80%. From the public health standpoint, that would be almost as good as a complete ban on the use of these antibiotics.
Will practitioners follow the advice suggested by this analysis? The biggest reason for them not to do so is that both the EBM model of deriving therapeutic recommendations from the best available evidence and the shared decisionmaking model that allows EBM ethically to be brought to bear on the individual patient's case require time and energy. In today's world of primary care in the United States, time and energy are both in exceedingly short supply. The implications of the primary physician reaching for the prescription pad because she lacks time to have a thorough discussion with the patient about the evidence-based alternatives has serious implications for much more than bacterial resistance. The ability to practice both scientifically competent and ethical medicine hangs in the balance as well.