点击显示 收起
美国内科医师学会2005年会循证医学热点
Highlights of Evidence Based Medicine From The American College of Physicians 2005 Annual Session
2005年4月14－16日
美国加利福尼亚州旧金山
April 14 - 16, 2005, San Francisco, California
Evidence-Based Approach to the Management of Irritable Bowel Syndrome and Chronic Constipation
Richard Glickman-Simon, MD
IBS – a potentially debilitating condition characterized by abdominal discomfort, bloating, and disturbed patterns of defecation – affects approximately 10% to 15% of the North American population. The prevalence is equally divided among 3 subtypes: IBS with constipation, IBS with diarrhea, and IBS with alternating constipation/diarrhea. A 2:1 prevalence ratio favors women.[1,2]
IBS is in a class of conditions often seen in CAM practices. It is a chronic, fluctuating illness that inconsistently responds to conventional treatment, which is often administered on a trial and error basis. It is a functional syndrome, meaning it has no known etiology, undocumented structural or biochemical abnormalities, and an incompletely understood pathophysiology involving nonspecific neuropsychiatric effects. Under the circumstances, patients often turn to CAM practitioners, from whom they find persuasive and appealing explanations for their symptoms, whether or not the therapies they receive work. This begs an important question: Can conventional medicine offer effective treatments for this functional condition?
In his session on the Evidence-Based Approach to the Management of Irritable Bowel Syndrome and Chronic Constipation, Dr. Philip Schoenfeld (University of Michigan School of Medicine) attempts to answer this question by drawing on a systematic review he and others recently published.[3] The review produced graded recommendations based on the strength of evidence, the scale for which appears in Table 1.
Modified from Cook D, Guyatt G, Laucacis A, Sackett D. Chest. 1992;102:305S.
Diagnosis. The pretest probability of inflammatory bowel disease, colorectal cancer, and infectious diarrhea in IBS patients without alarm findings is less than 1%, which is the same as in healthy controls. (Alarm findings include, but are not limited to, a family history of colon cancer, hematochezia, weight loss greater than 10 pounds, recurring fever, anemia, and chronic severe diarrhea.) Therefore, the routine use of endoscopic, imaging, or microscopic testing to rule out organic disease is not justified unless such findings are present. On the other hand, recent evidence suggests that celiac sprue may be at least 5 times more likely in IBS patients free of alarm findings, making an evaluation for sprue justifiable (C recommendation).
Treatment. There is no established threshold above which treatment for IBS should be considered. Patients should be informed of the natural history of their condition and reassured that it is non-life threatening. Benefits of treatment are likely to outweigh risks when patients feel their symptoms are adversely affecting their quality of life. In such cases, the most reliable measures of treatment effect are the relief of global IBS symptoms, which include the triad of abdominal discomfort, bloating, and altered bowel habits (an important research consideration).
Drug treatment options for IBS include antispasmodics, bulking agents, antidiarrheal agents, antidepressants, and 5HT (serotonin) receptor agonists and antagonists. There is currently insufficient data to recommend hyoscyamine and dicyclomine, the two antispasmodic medications available in the United States (B recommendation). Bulking agents (such as wheat bran, calcium polycarbophil, psyllium, and ispaghula husk) and antidiarrheal agents (loperamide) are no more effective than placebo at relieving global IBS symptoms (B recommendation). Tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) are more effective than placebo at treating global IBS symptoms (B recommendation). However, only diarrhea-predominant IBS patients tend to respond to TCA treatment (presumably due to their anticholinergic side effects). The 5HT4 receptor agonist tegaserod is more effective than placebo at treating global IBS symptoms in female patients with constipation (A recommendation). Only 20% of subjects in studies to date have been men, and tegaserod increases the risk of diarrhea 2-fold, making its use less attractive in diarrhea-predominant and alternating diarrhea/constipation IBS cases. The 5HT3 receptor antagonist alosetron is more effective than placebo at treating global IBS symptoms in female patients with constipation (A recommendation). However, because of concerns regarding the risk of ischemic colitis and serious diarrheal complications, the FDA has approved only restricted marketing of alosetron for severe diarrhea-predominant IBS that has not responded to other treatment.[3]
The majority of IBS patients who seek subspecialty services are found to have a comorbid psychiatric diagnosis, most commonly depression or anxiety. This has led to the speculation that IBS is largely the somatic manifestation of a mental disorder. However, since less than 20% of IBS suffers in population-based studies are found to have primary psychiatric disorders, this is probably not the case.[3] Nevertheless, the overlap between psychological disorders and IBS has promoted research into behavioral therapies, many of which are viewed as unconventional interventions. The theoretical mechanism for most of these therapies rests on the stress hypotheses of disease. In this case, a maladaptive response to chronic or recurrent stressful stimuli exacerbates IBS symptoms by stimulating the autonomic nervous system and its effects on gut motility, smooth muscle contraction, and pain perception. Table 2 lists behavioral therapies that have been studied for IBS.[3]
At least 16 randomized trials have evaluated the effectiveness of these therapies in the treatment of IBS. Although all had methodologic flaws (largely due to multiple biases unique to behavioral interventions), the authors concluded that behavioral therapy is more effective than placebo for individual IBS symptoms (B recommendation), but not necessarily for global symptoms.[3]
Dr. Schoenfeld went on to consider CFC, which is defined as unsatisfactory defecation characterized by infrequent stools, difficult stool passage, or both for at least 3 months. Prevalence is approximately 15% in North America and is more commonly reported in women (2-3:1 ratio), the elderly, nonwhites, and individuals of lower socioeconomic status.[4] His conclusions regarding the evidence for the diagnosis and treatment of chronic constipation are similar to that for IBS:
In the absence of alarm findings, there is inadequate data to support the routine use of endoscopy, imaging, thyroid function testing, serum calcium, or any other diagnostic evaluation. Alarm findings include family history of colon cancer or inflammatory bowel disease, hematochezia or positive fecal occult blood test, anemia, weight loss > 10 pounds, and the acute onset of constipation in elderly patients (C recommendation). Patients with other signs and symptoms of organic disease (eg, hypothyroidism) should be further evaluated, and all patients over the age of 50 should be screened for colorectal cancer.
The bulking agent psyllium (eg, Metamucil) increases stool frequency in patients with CFC (B recommendation), but there is insufficient evidence to make a recommendation for calcium polycarbophil, methylcellulose, and bran. There is also insufficient evidence to make a recommendation for stool softeners (eg, docusate sodium) or stimulant laxatives (eg, senna or bisacodyl).
The osmotic laxatives, polyethylene glycol and lactulose, are effective for improving stool frequency and consistency (A recommendation). There is insufficient evidence, however, to make a recommendation for milk of magnesia.
Tegaserod is more effective than placebo for decreasing straining, increasing stool frequency and number of spontaneous bowel movements, and improving stool consistency in men and women aged < 65 years (A recommendation).
There is insufficient evidence to make a recommendation for lubricants (eg, mineral oil) for CFC in adults, but several randomized trials indicate that mineral oil is more effective than stimulant laxatives, but less effective than osmotic laxatives, in improving stool frequency and consistency in children.[3,5]
Dr. Schoenfeld did not cover behavioral therapies for CFC except to point out that uncontrolled trials indicate that biofeedback can improve stool frequency compared with baseline (C recommendation).
Since this session was based almost exclusively on a thorough systematic review of the literature, it was truly an evidenced-based approach to IBS and CC as advertised. It is interesting to note, however, that Dr. Schoenfeld readily disclosed his business relationships with virtually all the major pharmaceutical companies (GlaxoSmithKline, Movartis, Solvay, Wyeth, Merck and AstraZeneca) before launching into his evidence-based discussion.
The primary purpose of any respectable scientific assembly is to bring updated and practical information to its participants so they may improve the care of their patients. Although most of the information exchanging hands at these conferences is certainly current and useful, the extent to which it is also evidence-based varies considerably from one session to the next. Advocates of evidence-based medicine are quick to point out that expert opinion is the least reliable form of evidence. The experts who are invited to present at these conferences have a choice. They can select individual studies that support their own preconceived clinical recommendations, or they can systematically review the literature (where it exists) and formulate evidence-based recommendations. Most presenters land somewhere in between.
While it is not possible to gauge the evidence-based nature of all 260 events at this year's ACP conference, the meeting did include a number of sessions specifically addressing evidence in clinical practice. Numerous workshops offered hands-on training in the retrieval of Web-based information and the use of PDAs. There were 4 sessions in particular that highlighted evidence-based medicine and its application at the point of care.
Complementary and Alternative Medicine: An Evidence-Based Approach
Evidence-Based Approach to the Management of Irritable Bowel Syndrome With Constipation
Prevention and Screening in Primary Practice and Adult Immunization
Information Mastery: Making Evidence-Based Medicine Practical at the Point of Care
A common, sweeping definition of CAM is that of a diverse collection of presumably healthy lifestyles, therapies and systems of healing that falls outside the mainstream of conventional medicine and has yet to be validated by scientific methods. One way to interpret this definition is that unlike conventional therapies, CAM therapies have yet to be proven safe, effective, or both. But once they are, they will promptly join the ranks of mainstream practice. This overly simplistic view fails to account for a number of principles of evidence-based medicine.
First, supportive evidence is not synonymous with proof of effectiveness. Although large randomized controlled trials (RCTs) provide valuable information about the relationships between interventions and outcomes, they do not prove cause and effect. The boundary between what works and what doesn't is not a bright, narrow line easily crossed with the publication of a single or even multiple studies. It is a wide and shifting zone, continuously fed by conventional and unconventional therapies moving in various directions as the evidence for their effectiveness accumulates or diminishes with time. CAM therapies, therefore, do not simply "cross over" when proven effective, and neither should conventional ones.
Second, evidence-based medicine is not simply the direct application of outcomes research to the care of patients. It involves the judicious use of scientific evidence to support, not supplant, clinical judgment. Even a preponderance of evidence supporting the effectiveness of an unconventional therapy would not persuade many clinicians to adopt it. Glucosamine and acupuncture, for example, both have evidence supporting their effectiveness for osteoarthritis of the knee.[1,2] If this evidence were to be equally consistent between the two, glucosamine would stand a better chance of acceptance among conventional physicians. Whereas the biological effects of a dietary supplement like glucosamine are plausible, if not well understood, the theoretical basis of acupuncture's effectiveness is akin to eastern mysticism, a conceptual stretch beyond the reach of most western-trained clinicians.
The standard argument goes: when it comes to CAM, we should be patient and wait for the research. In the meantime, it is best to err on the side of skepticism and assume that an intervention is ineffective until proven otherwise. This way of thinking certainly has merit, particularly from the perspective of safety and cost. We should not expose unwitting patients to potentially dangerous and expensive treatments until we can reasonably weigh their risks against their benefits.
However, not all clinical questions can, or should, be answered. The standard position presupposes that patients can wait for evidence that, in many cases, will never exist. Evidence is expensive, and no research budget will ever be large enough to answer even the majority of clinical questions. Certainly, if the risks and costs are high and the benefits unknown, the intervention should be rejected until it can be adequately studied. In the case of many "unproven" CAM therapies, however, the risks are small and their modest costs are largely borne by the patients themselves. Adopting an evidence-based approach means recognizing and accepting its limitations. When it comes to treating patients, balancing risks against benefits of any therapy – conventional or unconventional – will always be necessary in the absence of data.
Dr. Margaret Chesney, Deputy Director of the National Center for Complementary and Alternative Medicine (NCCAM), took on the issue of the scientific evidence and CAM in her session, Complementary and Alternative Medicine: An Evidence-Based Approach.
To set the stage for her discussion of CAM research, Dr. Chesney reviewed its popularity among American consumers and its penetration into the healthcare system. In a 2002 national survey of 31,044 randomly sampled adults,[3] 36.5% said they used CAM within the past 12 months. This number jumped to 62% if prayer was included as a CAM intervention. The most common therapies used were herbal therapies (19%), deep breathing (12%), meditation (8%), chiropractic (8%), yoga (5%), massage (5%) and diet-based therapies (4%). Hospitals offering CAM service increased from 8% in 1998 to 17% in 2002.[4] Specific services offered were massage (78%), pastoral counseling (62%), stress management (61%), and yoga (58%). Patient demand, not clinical effectiveness, was the key reason hospitals gave for offering these services. It is this consumer interest that has been largely driving the research agenda in CAM rather than the possibility of clinical effectiveness.
Yet despite this popular interest, CAM remains conspicuously understudied. Several factors have combined to dissuade world-class scientists and well-funded research institutions from taking it on. First, rigorous research requires strict adherence to consistent standards across interventions. A hallmark of many CAM practitioners, however, is their devotion to individualization and therapeutic heterogeneity. Classifying subjects into uniform groups and treating them all the same often contradicts their practice philosophy. Moreover, in the case of herbal medicine where diagnostic uniformity should be possible, it is still a challenge for researchers to guarantee consistent quality, purity, and dose across batches of plant extracts.
Second, the private sector is reluctant to invest in CAM research, foreseeing no return on their investment. This drawback is again well illustrated in the case of herbs. Since these products are rarely patentable, pharmaceutical companies see no profit in investing large sums of research dollars to determine their safety and effectiveness. They will either ignore the products in favor of patentable drugs, or market them without doing the research (which federal law permits).
And third, CAM research has been hampered by a lack of direction and cooperation. Poorly funded researchers operating in isolated silos succeed only in moving their agenda forward in fits and starts. Notwithstanding her personal stake in the matter, Dr. Chesney believes all this changed in 1999, when Congress required the National Institutes of Health to create NCCAM with an initial annual budget of $50 million (its FY 2005 budget is now $123.1 million). Unlike most NIH entities, the majority of NCCAM's research budget supports clinical studies rather than basic science research. NCCAM's 4 broad strategic goals are to:
Invest in CAM research
Train CAM practitioners in research and conventional practitioners in CAM
Disseminate evidence-based CAM information to healthcare professionals and the public
Facilitate the integration of CAM into the practice of allopathic medicine
NCCAM divides the vast field of CAM into 4 domains: biologically based practices, mind-body medicine, manipulative and body-based practices, and energy medicine. In turn, NCCAM identifies several whole "systems of healing" that may incorporate these 4 domains into a single entity, much like allopathic medicine incorporates pharmaceuticals, surgical procedures, rehabilitative services, etc. Examples of these healing systems include traditional Chinese medicine, ayurvedic medicine, naturopathy, and homeopathic medicine.
Biologically based practices. These practices use substances found in nature to treat illness and enhance health, including dietary supplements (such as herbs) and special diets. The widespread popularity of these products raises a number of concerns, which NCCAM plans to address through its research agenda. First, consumers are in the habit of equating "natural" with "safe." While most dietary supplements are safe, their safety cannot be guaranteed. The Dietary Supplement Health and Education Act (1994) classifies these products as foods rather than drugs. This means they are not subject to the same regulatory standards imposed by the FDA on pharmaceuticals, and can be marketed without any clinical evidence supporting their safety and effectiveness. It is up to consumers and their advocates to show that a product is unsafe before it can be removed from the market. Second, consumers must deal with the same lack of product standardization clinical researchers face. Herbs such as ginseng, for example, have been found to contain anywhere from 0% to 300% of the active ingredient advertised on the label. Third, without regulatory oversight, these products may contain numerous impurities such as pesticides and pharmaceuticals. Finally, many of these substances interact with numerous medications. St. John's wort, for example, may affect the metabolism of some HIV medications, lowering their serum concentrations.
Dr. Chesney provided 3 examples of dietary supplements that have been subject to numerous studies: glucosamine and chondroitin, saw palmetto, and ginkgo biloba. Seventeen RCTs have been published on the effectiveness of glucosamine, chondroitin, or both for osteoarthritis. Many of these studies suffer from weak design and small size, and most of them have been funded by manufacturers. Nevertheless, on balance, both glucosamine and chondroitin appear to offer some benefit in terms of pain and function. Daniel Clegg of the University of Utah is currently leading a study that may provide more definite answers on whether glucosamine and chondroitin are efficacious in the treatment of knee arthritis. This 6-month, placebo-controlled, double-blind trial includes 1588 osteoarthritic adults over the age of 40 divided into 5 treatment arms: glucosamine alone, chondroitin alone, glucosamine and chondroitin combined, a placebo, and a COS-2 inhibitor. Outcomes will include safety, function, pain, and preservation of joint space.
Seventeen placebo-controlled trials have also been conducted to evaluate saw palmetto, most of them small and short-term. On balance, they show improvement in symptoms and flow compared with placebo, and the effect is comparable to that of finasteride. In fact, there is evidence that saw palmetto, like finasteride, inhibits 5-alpha reductase. Other proposed mechanisms for saw palmetto include an anti-inflammatory effect and blockage of alpha-adrenergic receptors. A large, multisite phytotherapy trial comparing saw palmetto against African plum and placebo for the treatment of benign prostatic hyperplasia is in its early stages.
A number of small trials have suggested that ginkgo biloba may slightly improve cognitive function in patients with dementia, particular Alzheimer's. There has been some recent speculation that ginkgo extract may inhibit beta-amyloid aggregation. Steven DeKosky of the University of Pittsburgh has undertaken a large prevention trial to determine whether ginkgo biloba can decrease the incidence of dementia among 3073 adults over age 75. Outcomes will include not only changes in cognitive function but also the incidence of cardiovascular disease and overall mortality
Manipulative and body-based practices. These involve the manual movement of body parts to produce a variety of health outcomes. Chiropractic, osteopathic manipulation, and massage therapy are popular examples. Dr. Chesney gave relatively little attention to these services, focusing exclusively on the comparative safety and effectiveness of chiropractic. The evidence to date indicates that chiropractic and conventional regimens are comparable in relieving symptoms and restoring function for acute, subacute, and chronic low back pain.[5,6] In experienced hands, chiropractic is safe. Minor adverse effects, such as short-term discomfort in the treated area, are not uncommon. Although the risk of serious harm is low, stroke, for example, has been reported following neck manipulation.[7] Since the effectiveness of chiropractic for cervicogenic pain has yet to be established, it remains a matter of debate whether this unknown benefit is outweighed by this small risk.
Energy medicine. This practice involves the detection and manipulation of vital energy fields for therapeutic benefit. In so-called "biofield therapies," such as Gi Gong, Reiki, and therapeutic touch, the energy is undetectable by current scientific instruments. In bioelectromagnetic field therapies, detectable EM fields are used in unconventional ways to produce therapeutic effects. Examples include transcutaneous electrical nerve stimulation (TENS), magnets, and light therapy. Consumer interest in the use of magnets for the treatment of chronic pain prompted a recent placebo-controlled trial investigating the use of magnetized insoles for the treatment of plantar fasciitis, which failed to show any benefit.[8] Research investigating the efficacy of bioelectromagnetic field therapies is relatively straightforward given the measurable nature of these fields. Studying biofield therapies, however, is more challenging, and until these fields are detected, the conventional medical community will likely greet positive outcomes with a great deal of skepticism. NCCAM's position is that the same standards for experimental design in other disciplines should apply to biofield therapies. Fortunately, energy medicine is one of the safest of all CAM modalities.
Traditional Chinese medicine (TCM). TCM is an ancient system of healing encompassing an array of therapies, including meditation, exercise (Tai Chi), energy healing (Gi Gong), and herbal remedies. Acupuncture, though, is the most familiar TCM therapy to most Americans and has been extensively studied over the past several decades, with mixed results. Acupuncture is not easily suited to the RCT model of investigation, making it difficult to draw firm conclusions about its efficacy. A recent relatively well-designed trial investigated the usefulness of acupuncture as adjuvant treatment for knee osteoarthritis.[2] Five-hundred and seventy subjects were randomized to receive true acupuncture, sham acupuncture, or education for 26 weeks. Although 43% of the education group and 25% each of the true and sham acupuncture groups were lost to follow-up, true acupuncture was superior to the two controls in terms of improvement in knee pain and function. (Note: A well designed trial published after this ACP session investigated the efficacy of acupuncture for migraine headache and showed no benefit compared to a sham control.[9])
Mind-body medicine. This approach consists of numerous practices that draw on the complex bidirectional pathways between the mind and body. Meditation, yoga, hypnosis and Tai Chi all attempt to tap into the deep reservoir of the mind as a means to achieve therapeutic effects in the body. According to Dr. Chesney, this is a high-priority area for NCCAM. One area of particular interest is how mind-body medicine can favorably influence the practice of conventional medicine through the placebo effect, which remains poorly understood. NCCAM is also interested in the following:
Identifying the common and specific features of widely used meditative practices, including the relaxation response, transcendental medicine, and mindfulness meditation[10-14]
Discovering the neurologic, endocrine, and immunologic pathways underlying the observed therapeutic effects of mind-body practices
Evaluating the ability of mind-body interventions to reduce the prevalence and impact of stress-related chronic disease by enhancing positive affect and resilience
Given the enormous breadth of the topic, Dr. Chesney wisely did not attempt to provide a systematic review of the effectiveness of CAM as a discipline. That was not her charge or intent. She did attempt to summarize selected research findings and presented both positive and negative study results. However, the clinical relevance of citing individual studies out of context is questionable. The tone of her presentation was clearly one of optimism, rather than skepticism or even neutrality. Strictly speaking, therefore, her presentation was not evidence-based. Still, it was a thoughtful review of selected CAM research and NCCAM's agenda for the future.
Like complementary and alternative medicine (CAM), clinical prevention poses a number of challenges for proponents of evidence-based medicine. First, randomized trials measuring patient-oriented outcomes are few and far between. In primary prevention, the incidence of target conditions is too low and the time frame too long to make many of these studies economically feasible. While it is relatively straightforward to determine how well an antihypertensive agent lowers blood pressure, for example, it is more difficult to determine its effect on stroke risk. Second, lifestyle interventions -- a mainstay of preventive medicine -- are not easily studied in a controlled fashion. This is why most of the evidence for clinical prevention comes from (1) randomized trials measuring surrogate endpoints, (2) secondary prevention trials done on high-risk groups, and (3) large, population-based observational studies (eg, The Nurses Health Study). Occasionally, of course, sizable well-funded randomized trials producing guideline-altering evidence do get published (eg, Women's Health Initiative study on hormone replacement therapy).[1]
Yet, despite these inherent shortcomings, no other discipline requires higher-quality evidence than prevention. Unlike treatment, the benefits of preventive interventions occur in the distant future (if they occur at all), while their costs and possible harmful effects are incurred early on. Whereas most patients can expect to benefit from a proven treatment in the short term, only a small minority will benefit from a preventive intervention in the long term. Consider the use of selective estrogen receptor modulators in women at high risk for breast cancer. Raloxifene has been shown to reduce the risk of breast cancer by 50%, but only the fraction of women who were destined to get breast cancer could possibly benefit. Everyone who takes raloxifene, however, is at risk for its thrombotic complications and must pay the $70 per month for the privilege. While it may be ethically justifiable to expose a patient with a serious illness to the risk of treatment, it is ethically indefensible to expose a well person with only a statistical probability of future disease to the same risk.
This line of reasoning brings us to 2 conclusions. First, the decision to proceed with a preventive intervention always involves a careful balance between its benefits, harms, and costs. Since, as a general rule, this balance shifts toward a net benefit as the risk of the disease increases, a preventive intervention should not be undertaken without some idea of an individual's risk of the target condition. Second, the risks and costs of these interventions are immediate and real, and the benefits remote and speculative. Therefore, the principle of informed consent requires that patients be given the evidence regarding their safety and effectiveness before proceeding.
The general lack of high-quality evidence in preventive medicine has led to wide variations in preventive services. Many patients miss out on beneficial, low-risk interventions (eg, blood pressure screening), while others are harmed by more questionable practices (eg, prostate specific antigen testing). In response, the federal government created the US Preventive Services Task Force (USPSTF), originally convened by the US Public Health Service in 1984, and now sponsored by the Agency for Health Research and Quality.[2] The USPSTF carefully considers the evidence from systematic reviews in clinical prevention, estimates the magnitude of benefits and harms for each preventive service, and makes a recommendation. The preventive services covered are screening, chemoprevention, immunizations, and lifestyle counseling. The Task Force grades the strength of its recommendations and the quality of evidence according to the scales depicted in the Table.
In his session on Prevention and Screening in Primary Practice and Adult Immunizations, Harold C. Sox, the current Editor of the Annals of Internal Medicine and former Chair of the USPSTF (1990-1996), provided an overview of the USPSTF evidence-based guidelines for screening interventions, immunizations, and chemoprevention.[2] He chose not to cover lifestyle change, remarking that the evidence for its effectiveness is weak. He did say, however, that counseling is "worth it if we have the time" since it is so inexpensive and can make a difference for highly prevalent conditions (eg, smoking cessation counseling to reduce the incidence of tobacco-related morbidity).
Breast cancer. The USPSTF recommends screening mammography in average-risk women every 1 to 2 years beginning at age 40 (B recommendation). However, the balance of benefit vs harm is more favorable with advancing age. In older women (ages 50 to 69), the higher pretest probability of breast cancer coupled with a greater risk reduction for mortality makes mammography more cost-effective in this age group than in younger women (40 to 49). Although there are no randomized controlled trials supporting the effectiveness of mammography in women aged 70 and older, extrapolation of the data to relatively healthy women in this age group (with a life expectancy of at least 4 to 5 more years) argues for continued mammography. There is insufficient evidence to make a recommendation for clinical breast exam alone or self-breast exam as a means to reduce breast cancer mortality (I recommendation).[2]
Cervical cancer. Strong evidence supports the effectiveness of Pap smears at reducing cervical cancer mortality starting within 3 years of the onset of sexually activity or age 21, whichever comes first (A recommendation). Most mathematic models show that the cost-effectiveness of Pap smears peaks at a screening frequency of every 3 years. In women aged 65 and older with a history of normal Pap smears, the pretest probability of cervical dysplasia or cancer is too low to recommend continued screening (D recommendation). Women with total hysterectomies also should not be screened, irrespective of the indication for the hysterectomy (D recommendation). There is insufficient evidence for the USPSTF to make a recommendation for new screening technologies or HPV testing (I recommendation).[2]
Prostate cancer. Definitive evidence on the benefits of screening awaits the results of 2 large randomized trials begun in 1994 and due to be published in 5 to 6 years. In the meantime, recommendations must be based on observational trials and other sources of indirect evidence. Prostate cancer is an extremely prevalent but slow-growing tumor. The natural history of organ-confined prostate cancer remains poorly understood. Most men with prostate cancer die of something else, and there is no reliable method to predict who is likely to die from their cancer at the time of diagnosis. The sensitivity and specificity of digital rectal examination (DRE) is so low as to make the test essentially nonpredictive in average-risk men. The sensitivity of prostate specific antigen testing is a vast improvement over DRE, but its specificity remains low. Furthermore, the harms associated with prostate cancer treatment (radical prostatectomy and radiation therapy) are substantial, and can severely diminish quality of life. Since, based on current data, it is not possible to balance benefits against harms, the USPSTF concludes that at this time there is insufficient evidence to recommend for or against prostate cancer screening by any means (I recommendation).[2]
Colon cancer. The USPSTF found fair to good evidence that several screening methods are effective in reducing mortality from colorectal cancer (A recommendation). These include DRE, office fecal occult blood testing (FOBT), 3-sample home FOBT, sigmoidoscopy, double-contrast barium enema (DCBE), colonoscopy, and CT colography (virtual colonoscopy). The USPSTF recommends colorectal screening in all average-risk adults over age 50 (A recommendation). There is good evidence that home FOBT, and fair evidence that sigmoidoscopy alone or in combination with FOBT, reduces mortality. There is no direct evidence that colonoscopy effectively reduces mortality; however, its efficacy is supported by case-control evidence, extrapolation from sigmoidoscopy studies, and by its ability to study the proximal colon. DCBE is less sensitive than colonoscopy, and there is no direct evidence that it reduces mortality. Insufficient data exist to make a recommendation regarding CT colography. Dr. Sox emphasized the results of a recent carefully designed trial[3] indicating that office FOBT had a sensitivity of only 5% for advanced neoplasia. This is particularly worrisome since, according to a survey published at the time, about one third of doctors said office FOBT is the only colorectal cancer screen they use.[2]
Lung cancer. Although there is fair evidence that bronchial sampling cytology, chest-x-ray, and low-dose chest CT can effectively detect presymptomatic lung cancer, there is only fair to poor evidence that early detection with any of these modalities alters lung cancer mortality. Taking into account the invasive nature of diagnostic follow-up and the harms associated with false-positive results, the USPSTF states there is insufficient evidence to conclude that the benefits of lung cancer screening outweigh the harms (I recommendation).[2]
Osteoporosis. USPSTF recommends routinely screening all women over the age of 65 and women aged 60 to 64 at high risk for fracture (B recommendation). Other than age, the major risk factors for osteoporosis are a weight of less than 70 kg (best predictor) and no estrogen therapy. Other risk factors include smoking, weight loss, family history, low calcium or vitamin D intake, and sedentary lifestyle. As of 2002, when this recommendation was made, no randomized controlled trials on the effectiveness of screening had been published. However, bone density measured at the femoral neck by dual-energy x-ray absorptiometry is airly good at predicting fracture risk in older women with moderate sensitivity (but low specificity). Also, there is effective treatment: biphosphonates can reduce fracture risk by 40% to 50% in women with low bone density.[2]
Type 2 diabetes. The long preclinical phase of type 2 diabetes (DM) – 10-12 years or longer – provides ample opportunity for screening. While there is good evidence that interventions in patients with impaired glucose can delay the onset of DM for 3 to 6 years, it is too early to know whether these interventions prevent DM over the long term or if they have an impact on DM complications. Tight glycemic control in clinically detected (not screen detected) DM can reduce the incidence of microvascular complications, such as nephropathy and retinopathy. These complications, however, take at least 15 years to develop, making screening studies difficult to carry out. Even though existing studies have not shown the same benefits of tight glycemic control on macrovascular complications (myocardial infarction and stroke), knowing that a patient at high risk for cardiovascular disease has diabetes could favorably influence this risk. It would make more sense, then, to screen for diabetes in response to elevated cardiovascular risk and not the other way around. The USPSTF takes the view that there is insufficient evidence to recommend for or against routine screening for type 2 DM or impaired glucose tolerance (I recommendation). However, it does recommend screening patients with hypertension or dyslipidemia (B recommendation).[2]
Hypothyroidism. Although it is possible to reliably detect subclinical hypothyroidism, there is poor evidence that its treatment improves health outcomes. Furthermore, over-treatment is a potential harmful consequence of treating subclinical disease, and the long-term effects are unknown. Since the balance between benefits and harms cannot be established, the USPSTF cannot make a recommendation for or against routine thyroid screening (I recommendation).[2]
Aspirin to prevent cardiovascular disease. In men, there is good evidence that low-dose aspirin (75 mg/day) reduces the risk of coronary heart disease (CHD) by about 30%, with no significant effect on all-cause mortality or ischemic stroke. For 1000 patients with a 5% risk of a CHD event over 5 years, aspirin would prevent 14 myocardial infarctions (range, 6 to 20) but would cause 1 hemorrhagic stroke (range, 0 to 2) and 3 major gastrointestinal bleeds (range, 2 to 4). For patients with 1% risk of an event over 5 years, aspirin would prevent 3 myocardial infarctions (range, 1 to 4) but the incidence of intracranial and extracranial hemorrhage would not significantly change. The USPSTF takes the view that in both men and women at high risk of CHD (5-year risk greater than 3%), the benefits of low-dose aspirin outweigh its harms (A recommendation).[2]
In women, the story is a bit different. In a recently published trial from the WHI), almost 40,000 women over age 45 were randomized to receive 100 mg of aspirin every other day or a placebo for 10 years. Aspirin therapy did not significantly affect the incidence of fatal or nonfatal cardiovascular events, hemorrhagic stroke, or all-cause deaths. However, it did reduce the risk of ischemic stroke by 24% (absolute risk reduction 2.5 per 1000). The relative risk of any GI bleeding increased by 22% (absolute risk increase 8 per 1000), and the relative risk of major GI bleeding increased by 40% (absolute risk increase 1.8 per 1000). Unlike in men, the major benefit of low-dose aspirin appears to be a slight reduction of ischemic stroke, not myocardial infarction. Some women may feel, however, that the risk of GI bleeding offsets this benefit.[4]
Antioxidant vitamin supplementation. Despite encouraging results from numerous laboratory and observational studies, recent clinical trials have provided conflicting evidence for the effectiveness vitamins A, C, and E in the prevention of either cancer or cardiovascular disease (I recommendation). For beta-carotene, there is good evidence of no effect in nonsmokers and increased mortality in smokers. For this reason, the USPSTF gives beta-carotene a D recommendation.[20] A recent meta-analysis goes further to implicate vitamin E in causing increased cardiovascular harm at doses above 400 IU per day,[5] and a report from the WHI presented at the latest American College of Cardiology meeting indicated no cardiovascular benefit from 600 IU.
To the extent that Dr. Sox's presentation rehashed the latest USPSTF recommendations, it can be considered evidence-based. His comment regarding counseling to affect behavior change, however, should not be construed in the same way.
As mentioned above, lifestyle interventions are not easily studied in a controlled fashion. It is feasible to assign a small group of subjects to a specific diet or exercise program and strictly monitor their adherence over a short period of time. Conducting such a trial over the long term, however, is too expensive and vulnerable to follow-up losses. And, even if it were possible, extrapolating the results to free living, uncontrolled populations would be questionable.
While counseling, then, does have the weakest evidence supporting its efficacy, in 2003 the USPSTF strongly recommended that clinicians screen for tobacco use in all their adults patients and provide tobacco cessation interventions for all smokers (A recommendation). So, for tobacco and selected other behavioral risks, the evidence suggests that physicians should make the time for counseling.
Dr. Mark H. Ebell (Michigan State University) led the session, Information Mastery: Making Evidence-Based Medicine Practical at the Point of Care. The entire discussion of evidence-based medicine is meaningless unless the evidence reaches the clinical decision makers and affects their behavior at the point of care. The problem, of course, is that no clinician, no matter how well intentioned, could possibly translate even a fraction of the existing evidence into clinical action. Dr. Ebell mentioned that the average physician spends 30-60 minutes per week keeping up with the literature. The top 120 English language journals publish over 20,000 articles per year. Even if a physician felt that no more than 10% of this information was relevant and useful, he or she would still need to review 2000 articles per year. Spending as little as 15 minutes per study (barely enough time to get past the abstract) would take roughly 10 hours per week. Physicians clearly need a strategy to obtain only the information they need, when and where they need it.
In his session, Dr. Ebell pointed out the following equation as a way to determine the usefulness of information.
Information usefulness = Relevance x Validity/Work
Relevance. The relevance of a study depends on the prevalence of the clinical problem it addresses in a given population and the impact its results are likely to have on clinical practice. The nature of a study's outcomes is critical to its relevance. Patient-oriented outcomes are directly related to patients' experience of their illness: mortality, morbidity, and quality of life. Disease-oriented outcomes are surrogates for the patient-oriented ones: pathophysiologic markers, laboratory tests, and physical exam findings. Most clinically relevant evidence is built on predominantly patient-oriented studies.
Patient-oriented and disease-oriented outcomes may agree or disagree. For example, the evidence that diuretics effectively decrease blood pressure in hypertensive patients (a disease-oriented outcome) agrees with the evidence that they reduce mortality from stroke (a patient-oriented outcome). On the other hand, evidence that vitamin E prevents LDL oxidation and platelet aggregation (disease-oriented outcome) disagrees with the evidence that vitamin E has a neutral, or even harmful, effect on cardiovascular mortality (patient-oriented outcome).
Since studies with disease-oriented outcomes are less expensive and produce more repeatable results, they almost always predate studies on patient-oriented outcomes. There is evidence, for example, that PSA levels rise early in the development of prostate cancer; however, to date there is no evidence that PSA screening reduces mortality from prostate cancer. Time will tell whether disease-oriented and patient-oriented evidence agree or disagree with regard to prostate cancer screening.
Validity. This is a measure of a study's ability to get to the "truth." The design and execution of a highly valid study, for example, minimizes the chance for bias. Validity is communicated on 2 levels depending on the number of studies involved:
A levels-of-evidence scale is used to grade the validity of individual studies.
A strength of recommendations scale, which is based on a body of evidence, weighs the benefits against the risks of an intervention.
As we saw above, the USPSTF table and that presented by Dr. Schoenfeld use both scales to assess the validity of preventive interventions. Transparency is the key to evidence-based medicine. If an evidence source does not provide levels of evidence or strength of recommendations, its reliability should be questioned.
Work. Work refers to the time and effort it takes to find and use valid and relevant clinical information. Classic evidence-based medicine emphasizes exhaustive Medline searches and critical appraisal of the literature, a non-starter for most physicians. A better approach, of course, is to have other people (who are paid for their time) review the journals, search the literature, evaluate the studies and publish their summaries. This not only saves work, but allows people familiar with the literature's landscape to provide useful interpretation and comment on the clinical relevance they uncover. Many online resources provide this service:
The Cochrane Collaboration - http://www.cochrane.org
ACP Journal Club - http://www.acpjc.org
Bandolier - http://www.jr2.ox.ac.uk/bandolier
InfoPOEMs (Patient Oriented Evidence that Matters)
- http://www.infopoems.com
Clinical Evidence - www.clinicalevidence.org
How should physicians fulfill their obligation to apply the latest and best evidence to the care of their patients? Dr. Ebell's view is that physicians need to construct a framework for efficiently staying up to date and determining what new clinical intervention to adopt or reject. This requires a set of evidence-based medicine (EBM) tools that bring highly relevant and valid information directly to the practicing physician. Ideally, these tools would explicitly state the process used to search the literature, filter out useless information, rate the strength of evidence, provide useful summaries of important topics, and employ calculators and other decision support features to reduce work.
For information to be useful at the point of care it must be accessible electronically and fully interactive. The Internet provides great versatility and capacity but requires a network to be in place wherever patients are seen. Portable digital assistants (PDAs), which are handheld devices, have become more than adequate for most EBM applications and do not have this restriction. They can go wherever and whenever patients are seen and still have the capability of wireless communication with the Internet. The 2 most common types are currently Palm and Pocket PC. Although each has its advantages and disadvantages, their memory, wireless capability, expandability, screen quality software and costs are rapidly converging. The decision to purchase one over the other, therefore, is probably best made by reaching a consensus with other colleagues.
The Table lists selected EBM tools to consider. Of note, Dr. Ebell owns InfoPoems and so spent a fair amount of time discussing InfoRetriever.
To be useful, evidence must be brought to the physician and not the other way around. Unfortunately, this is not how most physicians are exposed to new evidence. Highly successful conferences like the ACP Annual Session draw physicians away from their practices – sometimes thousands of miles away – to teach them about the evidence with no patients in sight. In fact, this method is so popular and well respected that state government licensing boards recognize participation in these gatherings as proof of continuing education for license renewal. There is no evidence, however, that attendance at CME conferences has a lasting effect on the participants' behavior or their patients' health outcomes. Traditional CME can highlight clinical advances and make physicians aware of their deficiencies, but it cannot be expected to change practice. The most effective adult learning occurs through active problem solving at the point of care rather than the passive reception of information independent of care. It is ironic that the ACP sessions on evidence-based medicine are being delivered in a format for which there is no evidence of effectiveness.