- Email: [email protected]
Why evidence-based medicine failed in patient care and medicine-based evidence will succeed
Journal of Clinical Epidemiology 84 (2017) 14e17
Why evidence-based medicine failed in patient care and medicine-based evidence will succeed Ralph I. Horwitza,*, Burton H. Singerb a
Department of Medicine, Temple Transformative Medicine Institute, Temple University, 3500 N. Broad Street, Philadelphia, PA 19140, USA b Emerging Pathogens Institute, University of Florida, P.O. Box 100009, Gainesville, FL 32610, USA
Abstract Evidence-based medicine (EBM) has succeeded in strengthening the evidence base for population medicine. Where EBM has failed is in answering the practicing doctor’s question of what a likely outcome would be when a given treatment is administered to a particular patient with her own distinctive biological and biographical (life experience) profile. We propose Medicine-based evidence (MBE), based on the profiles of individual patients, as the evidence base for individualized or personalized medicine. MBE will build an archive of patient profiles using data from all study types and data sources, and will include both clinical and socio-behavioral information. The clinician seeking guidance for the management of an individual patient will start with the patient’s longitudinal profile and find approximate matches in the archive that describes how similar patients responded to a contemplated treatment and alternative treatments. Ó 2017 Elsevier Inc. All rights reserved. Keywords: Evidence-based medicine; Personalized medicine; Medicine-based evidence
Evidence-based medicine (EBM) has succeeded at the narrowly defined task for which it was best designed, to evaluate treatment efficacy by estimating the average results in randomized controlled trials (RCTs). Where EBM failed is in providing evidence to guide decisions in clinical care for individual patients. In the sections that follow, we describe the success of EBM, indicate where and why it failed, and suggest what is needed in the new era of personalized medical care. 1. The success of EBM The RCT and meta-analyses of multiple trials of the same topic (disease and treatment) are the major tools of EBM. It is worth remembering how recently the RCT was introduced into medical research. Although R.A. Fisher used the method of randomization in agricultural settings in the 1920s, the first report of an RCT in medicine was not published until 1948 (‘‘Streptomycin treatment of pulmonary tuberculosis’’) [1]. It was quickly recognized that the RCT would be useful in providing valid estimates of the average benefits of treatment in groups of patients. For the regulator interested in licensing pharmaceuticals
* Corresponding author. Tel.: (215)707-5671; fax: (215)707-8431. E-mail address: [email protected] (R.I. Horwitz). http://dx.doi.org/10.1016/j.jclinepi.2017.02.003 0895-4356/Ó 2017 Elsevier Inc. All rights reserved.
and for the company interested in developing them, these average estimates proved acceptable [2]. Indeed, RCTs used for this purpose have been superb in separating useful from useless drugs and have enabled the development and approval of numerous medicines that have transformed patient care. It is important to celebrate this accomplishment. During the past several decades, new therapies have led to effective control of risk factors for myocardial infarction and stroke; transformed HIV infection from a rapidly fatal disease to a chronic illness; cured Hepatitis C and achieved substantial improvements in the outcomes of some cancers. Many more advances in clinical therapeutics also owe their success to the findings that emerged from well-designed trials assessing the efficacy of new medicines. Even these accomplishments of RCTs in drug approval have been tarnished by disappointments. Since the 1990s, over 20 medicines approved by regulators on the basis of valid, well-designed and appropriately implemented RCTs have later been withdrawn from use in clinical practice, and many others have been associated with adverse complications that were unrecognized in the preapproval studies [3]. Among the drugs withdrawn were well-known medicines like Vioxx (Rofecoxib) that was associated with an increase in cardiovascular risk and Rezulin (Troglitazone) that caused life-threatening liver failure. These errors in
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
What is new? Despite the success of evidence-based medicine in strengthening the scientific basis of population medicine, EBM has been less successful in answering the physician’s question: What is the best treatment for my particular patient? Medicine Based Evidence will build an archive individual patient profiles to provide the evidential base to guide clinical decision making in the personalized care of patients.
drug approval are not accidental. They are embedded in the design of trials that (1) test the efficacy of drugs in younger patients only to observe adverse effects in older patients; (2) test the effectiveness of drugs in trials that employ short-term follow-up when the medicine is intended to be used for much longer durations; or (3) report commonly occurring adverse effects but are unable to detect uncommon side effects even in trials with large sample sizes. As serious as these shortcomings are in the use of RCTs as the foundation for EBM in drug development, there are even more profound shortcomings of EBM for the practice of Medicine. Most notably, EBM did not have a focus on evidence needed to guide decision-making about management of an individual patient. 2. EBMs greatest failures The practice of medicine is complex. Clinical care is not directed toward groups of patients but to an individual patient whose biology and biography (social, behavioral, and environmental experiences) are distinctive, and whose goals of therapy are often at odds with the goals of EBM. Because EBM relies so heavily on data from RCTs, the data do not include many types of treatments or patients seen in clinical practice; the results show the efficacy of treatment for an ‘‘average’’ randomized patient, not those in subgroups formed by cogent clinical features such as severity of illness, comorbidity or other clinical nuances; or psychosocial features such as stress, allostasis, or neighborhood deprivation that are known to affect both risk for disease and response to treatment. Randomized trial data are also seldom available for issues that are prominent in clinical practice such as etiology, diagnosis, and prognosis of disease. Physicians who seek to provide counseling to patients or take their personal preferences into account or learn strategies for giving comfort or reassurance to patients will find little help from the literature of EBM. In the sections that follow, we describe some of the most prominent concerns about EBM and suggest how to develop a new approach to evidence for personalized care in clinical medicine.
15
3. EBM disregards lack of applicability of RCT results in clinical practice RCTs are carried out in highly selected patients who are required to meet inclusion and exclusion criteria that often omit many of the patients who would later be candidates for treatment. Many trials impose restrictions on eligible patients who are judged to be too old or too young, to have illness that is too severe or not severe enough, or who have comorbidities that are common in patients with the main disease being treated. Recent reviews of trials that evaluated the benefits of statins and nonsteroidal antiinflammatory drugs reported that women, older adults, and minorities were underrepresented [4]. Studies of the medicine montelukast for treatment of asthma illustrate the limits of EBM as currently practiced. Numerous RCTs of the asthma drug montelukast had shown that it was inferior to inhaled corticosteroids as a first-line treatment for asthma control and also inferior to long-acting beta agonists (LABAs) as a second-line ‘‘addon’’ therapy. Montelukast is an oral medicine with higher adherence than that of inhaled agents when used in realworld settings. However, as often occurs in short-term (|6 months), placebo-controlled trials, patients maintain high levels of adherence within the trial that are not achievable outside the trial. A study that illustrates the apparent advantage of inhaled steroids and LABAs comes from the results of an RCT in which montelukast was compared with beclomethasone and placebo. In this study, both montelukast and beclomethasone were superior to placebo [in improvements in forced expiratory volume in 1 second (FEV1)], and beclomethasone was superior to montelukast [5]. Despite these findings, montelukast remains popular with patients and physicians. Evidence to support the wisdom of this popularity was observed when montelukast was tested in realworld effectiveness studies. In these real-world studies, montelukast was just as effective as both inhaled corticosteroids and LABAs. What was different? [6]. The real-world studies enrolled patients with asthma who were more like those seen in actual clinical practice; had lower adherence rates that reflect levels seen commonly in customary care, not those artificially achieved in placebo-controlled trials; and measured clinical outcomes such as symptoms, function, and well-being. When discussing the results of their real-world trial, the authors noted that they deliberately included patients typically excluded in most EBM RCTs including those who smoke and those with coexisting conditions, poor adherence, or poor inhaler technique. If there are serious concerns about the applicability of RCTs to real-world effectiveness, there is even greater concern that EBM fails to generate applicable evidence on the harms of treatment. RCTs are designed with sample sizes large enough to detect clinical benefits but not large enough to identify less commonly occurring potential
16
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
harms. Brain hemorrhage associated with phenylpropanolamine use [7] and liver failure after use of troglitazone [8] are serious life-threatening complications that were missed in registration trials because they occurred at low rates. Other medicines have risks that occur at higher rates but require longer lengths of follow-up for detection than are typical in RCTs. 4. EBM and the fallacy of design hierarchies EBM fostered the ascent of design hierarchies that placed RCTs at the pinnacle of the hierarchy and observational studies, including cohort and caseecontrol designs, in subordinate roles [9]. The adoption of this view, now memorialized in over 40 different grading systems, was extended to the development of practice guidelines with the further effect that the restricted populations, comparative treatments, and hard end points of RCTs became the basis for individualized decision-making. Data from observational studies were accorded a low level of importance along with the more nuanced but ‘‘softer’’ measures of patient preference, and physical and social functioning [10]. The original proponents of the RCT feared this doctrinaire approach to clinical research and warned against it. Bradford Hill, one of the principle architects of the RCT, wrote in his Heberden Oration, ‘‘.any belief that the controlled trial is the only way would mean not that the pendulum had swung too far but that it had come right off the hook’’ [11]. Ironically, empirical studies fail to confirm the primacy of RCTs in assessing the benefits of therapy. When topics and studies from defined sampling frames have been assessed, several research groups find that well-designed observational and randomized designs produce equivalent results [9,12]. If specious criticisms of observational designs based on comparing RCTs with historical controlled trials are dismissed, the evidence confirms that wellconducted observational cohort and caseecontrol studies can provide a similar level of internal validity as RCTs. EBM fails clinical practice when it adopts hierarchies to replace judgment with overly simplistic methods that create the appearance of quantitative precision that does not exist. Clinicians are decision makers and like all decision makers need to incorporate judgment, often based on idiosyncratic cases from their own and others clinical practices as a crucial element in the appraisal of evidence. It does not advance the practice of medicine to claim that RCTs are trustworthy and other forms of research are not. Advances in clinical care occur even in the absence of RCT evidence. In 2011, the FDA approved belimumab, a B-lymphocyte stimulator inhibitor, for the treatment of systemic lupus erythematosus (SLE) [13]. Before the approval of belimumab, the FDA last approved a drug to treat SLE in 1955 (hydroxychloroquine). What happened in the outcomes of SLE treatment during the
50 years when no EBM validated new treatments were approved? [13]. Urowitz followed a group of 1,241 patients seen at the University of Toronto between 1970 and 2005 [14]. Over the 36-year period of the study, 211 deaths occurred. The overall standardized mortality ratio was 12.6 [95% confidence interval (CI): 9.1, 17.4] in the first decade and 3.46 (95% CI: 2.7, 4.4) in the last decade when stratified by calendar year of enrollment. Some of the improved mortality was attributable to recognition of mild disease, but a large part was due to improved management achieved through clinical experimentation, careful observation, and the assessment of the effects of outcomes in practice. 5. Biology, social environment, and individualized care: medicine-based evidence EBM should not be abandoned. RCTs will continue to be valuable for drug companies and regulatory authorities who need to decide whether the average treatment results of new medicines will create more good than harm when used in the general population. For the new era of individualized patient care that is now upon us, we require new ways of thinking; new types of data describing clinical, social, behavioral, and neighborhood features of individuals; and new methods of analysis. The starting point for any clinically applicable analysis of personalized care is a detailed characterization of the individual patient. This apparently innocent recommendation is sure to meet with great resistance. Clinical medicine is all about individual care, but clinical research has little experience with studies that are oriented around individual level design and analysis. Individual level data will need to emphasize detailed and repeated measure over time of each subject across multiple domains including deep biology, clinical physiology, social and behavioral features, and work and neighborhood description. The social environment in which the patient is embedded matters a great deal. All of these data need to be amalgamated into a unified ‘‘profile’’ and, for diverse types of patients, assembled into an archive to enable identification of approximate matches to a given patient. The clinician seeking guidance for management of an individual patient starts with the patient profile and then finds approximate matches in the archive that describe how like patients responded to a contemplated treatment and alternative treatments. Medicine-based evidence (MBE) as a concept was elegantly described over 20 years ago [15]. As envisioned now, MBE will build an archive of profiles that emerge from all types of studies and data sources: RCTs, cohorts, caseecontrol studies, disease registries, and more. Investigators need to embrace the challenge of developing new methods and improving existing ones. Investigators and clinicians need to reject dogmatic assertions about the hierarchy of evidence. Finally, all who are involved in the care of patients need to recognize that the best decisions for the
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
care of patients require the role of judgment in the research that generates evidence, judgment in the interpretation of evidence, and judgment in the application of evidence to the care of individual patients.
[7]
[8]
References [1] Streptomycin treatment of pulmonary tuberculosis: a Medical Research Council Investigation. Br Med J 1948;2:769. [2] Horwitz RI, Singer BH, Makuch RW, Viscoli CM. Can treatment that is helpful on average be harmful to some patients? A study of the conflicting information needs of clinical inquiry and drug regulation. J Clin Epidemiol 1996;49:395e400. [3] Rawlins MD. Harveian Oration. De Testimonia on the evidence for decisions about the use of therapeutic interventions. Lancet 2008; 372:2152e61. [4] Bartlett C, Doyal L, Ebrahim S, Davey P, Bachmann M, Egger M, et al. The causes and effects of socio-demographic exclusions from clinical trials. Health Technol Assess 2005;9:38. [5] Malmstrom K, Rodriguez-Gomez G, Guerra J, Villaran C, Pi~neiro A, Wei LX, et al. Oral montelukast, inhaled beclomethasone, and placebo for chronic asthma. A randomized, controlled trial. Ann Intern Med 1999;130:487e95. [6] Price D, Musgrave MD, Shepstone L, Hillyer EV, Simms E, Gilbert RT, et al. Leukotriene antagonists as first-line or add-
[9]
[10]
[11] [12] [13] [14]
[15]
17
on asthma-controller therapy. N Engl J Med 2011;364: 1695e707. Kernan WN, Viscoli CM, Brass LM, Broderick JP, Brott T, Feldmann E, et al. Phenylpropanolamine and the risk of hemorrhagic stroke. N Engl J Med 2000;343:1826e32. Graham DJ, Green L, Senior JR, Nourjah P. Troglitazone-induced liver failure: a case study. Am J Med 2003;114:299e306. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342:1887e92. Owens DK, Lohr KN, Atkins D, Treadwell JR, Reston JT, et al. Grading the strength of a body of evidence when comparing medical interventions. In: Agency for Healthcare Research and Quality Methods Grade for Comparative Effectiveness Reviews. [posted July 2009]. Available at http://effectivehealthcare.ahrq.gov/healthInfo. cfm?infotype5rr&ProcessID560. Hill AB. Reflections on controlled trial. Ann Rheum Dis 1966;25: 107. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med 2000;342:1878e86. Stohl W, Hilbert DM. The discovery and development of belimumab: the anti-BLyS-lupus connection. Nat Biotechnol 2012;30:69e77. Urowitz MB, Gladman DD, Tom BD, Iba~nez D, Farewell VT. Changing patterns in mortality and disease outcomes for patients with systemic lupus erythematosus. J Rheumatol 2008;35:2152e8. Knottnerus JA, Dinant GJ. Medicine based evidence, a prerequisite for evidence based medicine. BMJ 1997;315:1109e10.
Why evidence-based medicine failed in patient care and medicine-based evidence will succeed Ralph I. Horwitza,*, Burton H. Singerb a
Department of Medicine, Temple Transformative Medicine Institute, Temple University, 3500 N. Broad Street, Philadelphia, PA 19140, USA b Emerging Pathogens Institute, University of Florida, P.O. Box 100009, Gainesville, FL 32610, USA
Abstract Evidence-based medicine (EBM) has succeeded in strengthening the evidence base for population medicine. Where EBM has failed is in answering the practicing doctor’s question of what a likely outcome would be when a given treatment is administered to a particular patient with her own distinctive biological and biographical (life experience) profile. We propose Medicine-based evidence (MBE), based on the profiles of individual patients, as the evidence base for individualized or personalized medicine. MBE will build an archive of patient profiles using data from all study types and data sources, and will include both clinical and socio-behavioral information. The clinician seeking guidance for the management of an individual patient will start with the patient’s longitudinal profile and find approximate matches in the archive that describes how similar patients responded to a contemplated treatment and alternative treatments. Ó 2017 Elsevier Inc. All rights reserved. Keywords: Evidence-based medicine; Personalized medicine; Medicine-based evidence
Evidence-based medicine (EBM) has succeeded at the narrowly defined task for which it was best designed, to evaluate treatment efficacy by estimating the average results in randomized controlled trials (RCTs). Where EBM failed is in providing evidence to guide decisions in clinical care for individual patients. In the sections that follow, we describe the success of EBM, indicate where and why it failed, and suggest what is needed in the new era of personalized medical care. 1. The success of EBM The RCT and meta-analyses of multiple trials of the same topic (disease and treatment) are the major tools of EBM. It is worth remembering how recently the RCT was introduced into medical research. Although R.A. Fisher used the method of randomization in agricultural settings in the 1920s, the first report of an RCT in medicine was not published until 1948 (‘‘Streptomycin treatment of pulmonary tuberculosis’’) [1]. It was quickly recognized that the RCT would be useful in providing valid estimates of the average benefits of treatment in groups of patients. For the regulator interested in licensing pharmaceuticals
* Corresponding author. Tel.: (215)707-5671; fax: (215)707-8431. E-mail address: [email protected] (R.I. Horwitz). http://dx.doi.org/10.1016/j.jclinepi.2017.02.003 0895-4356/Ó 2017 Elsevier Inc. All rights reserved.
and for the company interested in developing them, these average estimates proved acceptable [2]. Indeed, RCTs used for this purpose have been superb in separating useful from useless drugs and have enabled the development and approval of numerous medicines that have transformed patient care. It is important to celebrate this accomplishment. During the past several decades, new therapies have led to effective control of risk factors for myocardial infarction and stroke; transformed HIV infection from a rapidly fatal disease to a chronic illness; cured Hepatitis C and achieved substantial improvements in the outcomes of some cancers. Many more advances in clinical therapeutics also owe their success to the findings that emerged from well-designed trials assessing the efficacy of new medicines. Even these accomplishments of RCTs in drug approval have been tarnished by disappointments. Since the 1990s, over 20 medicines approved by regulators on the basis of valid, well-designed and appropriately implemented RCTs have later been withdrawn from use in clinical practice, and many others have been associated with adverse complications that were unrecognized in the preapproval studies [3]. Among the drugs withdrawn were well-known medicines like Vioxx (Rofecoxib) that was associated with an increase in cardiovascular risk and Rezulin (Troglitazone) that caused life-threatening liver failure. These errors in
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
What is new? Despite the success of evidence-based medicine in strengthening the scientific basis of population medicine, EBM has been less successful in answering the physician’s question: What is the best treatment for my particular patient? Medicine Based Evidence will build an archive individual patient profiles to provide the evidential base to guide clinical decision making in the personalized care of patients.
drug approval are not accidental. They are embedded in the design of trials that (1) test the efficacy of drugs in younger patients only to observe adverse effects in older patients; (2) test the effectiveness of drugs in trials that employ short-term follow-up when the medicine is intended to be used for much longer durations; or (3) report commonly occurring adverse effects but are unable to detect uncommon side effects even in trials with large sample sizes. As serious as these shortcomings are in the use of RCTs as the foundation for EBM in drug development, there are even more profound shortcomings of EBM for the practice of Medicine. Most notably, EBM did not have a focus on evidence needed to guide decision-making about management of an individual patient. 2. EBMs greatest failures The practice of medicine is complex. Clinical care is not directed toward groups of patients but to an individual patient whose biology and biography (social, behavioral, and environmental experiences) are distinctive, and whose goals of therapy are often at odds with the goals of EBM. Because EBM relies so heavily on data from RCTs, the data do not include many types of treatments or patients seen in clinical practice; the results show the efficacy of treatment for an ‘‘average’’ randomized patient, not those in subgroups formed by cogent clinical features such as severity of illness, comorbidity or other clinical nuances; or psychosocial features such as stress, allostasis, or neighborhood deprivation that are known to affect both risk for disease and response to treatment. Randomized trial data are also seldom available for issues that are prominent in clinical practice such as etiology, diagnosis, and prognosis of disease. Physicians who seek to provide counseling to patients or take their personal preferences into account or learn strategies for giving comfort or reassurance to patients will find little help from the literature of EBM. In the sections that follow, we describe some of the most prominent concerns about EBM and suggest how to develop a new approach to evidence for personalized care in clinical medicine.
15
3. EBM disregards lack of applicability of RCT results in clinical practice RCTs are carried out in highly selected patients who are required to meet inclusion and exclusion criteria that often omit many of the patients who would later be candidates for treatment. Many trials impose restrictions on eligible patients who are judged to be too old or too young, to have illness that is too severe or not severe enough, or who have comorbidities that are common in patients with the main disease being treated. Recent reviews of trials that evaluated the benefits of statins and nonsteroidal antiinflammatory drugs reported that women, older adults, and minorities were underrepresented [4]. Studies of the medicine montelukast for treatment of asthma illustrate the limits of EBM as currently practiced. Numerous RCTs of the asthma drug montelukast had shown that it was inferior to inhaled corticosteroids as a first-line treatment for asthma control and also inferior to long-acting beta agonists (LABAs) as a second-line ‘‘addon’’ therapy. Montelukast is an oral medicine with higher adherence than that of inhaled agents when used in realworld settings. However, as often occurs in short-term (|6 months), placebo-controlled trials, patients maintain high levels of adherence within the trial that are not achievable outside the trial. A study that illustrates the apparent advantage of inhaled steroids and LABAs comes from the results of an RCT in which montelukast was compared with beclomethasone and placebo. In this study, both montelukast and beclomethasone were superior to placebo [in improvements in forced expiratory volume in 1 second (FEV1)], and beclomethasone was superior to montelukast [5]. Despite these findings, montelukast remains popular with patients and physicians. Evidence to support the wisdom of this popularity was observed when montelukast was tested in realworld effectiveness studies. In these real-world studies, montelukast was just as effective as both inhaled corticosteroids and LABAs. What was different? [6]. The real-world studies enrolled patients with asthma who were more like those seen in actual clinical practice; had lower adherence rates that reflect levels seen commonly in customary care, not those artificially achieved in placebo-controlled trials; and measured clinical outcomes such as symptoms, function, and well-being. When discussing the results of their real-world trial, the authors noted that they deliberately included patients typically excluded in most EBM RCTs including those who smoke and those with coexisting conditions, poor adherence, or poor inhaler technique. If there are serious concerns about the applicability of RCTs to real-world effectiveness, there is even greater concern that EBM fails to generate applicable evidence on the harms of treatment. RCTs are designed with sample sizes large enough to detect clinical benefits but not large enough to identify less commonly occurring potential
16
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
harms. Brain hemorrhage associated with phenylpropanolamine use [7] and liver failure after use of troglitazone [8] are serious life-threatening complications that were missed in registration trials because they occurred at low rates. Other medicines have risks that occur at higher rates but require longer lengths of follow-up for detection than are typical in RCTs. 4. EBM and the fallacy of design hierarchies EBM fostered the ascent of design hierarchies that placed RCTs at the pinnacle of the hierarchy and observational studies, including cohort and caseecontrol designs, in subordinate roles [9]. The adoption of this view, now memorialized in over 40 different grading systems, was extended to the development of practice guidelines with the further effect that the restricted populations, comparative treatments, and hard end points of RCTs became the basis for individualized decision-making. Data from observational studies were accorded a low level of importance along with the more nuanced but ‘‘softer’’ measures of patient preference, and physical and social functioning [10]. The original proponents of the RCT feared this doctrinaire approach to clinical research and warned against it. Bradford Hill, one of the principle architects of the RCT, wrote in his Heberden Oration, ‘‘.any belief that the controlled trial is the only way would mean not that the pendulum had swung too far but that it had come right off the hook’’ [11]. Ironically, empirical studies fail to confirm the primacy of RCTs in assessing the benefits of therapy. When topics and studies from defined sampling frames have been assessed, several research groups find that well-designed observational and randomized designs produce equivalent results [9,12]. If specious criticisms of observational designs based on comparing RCTs with historical controlled trials are dismissed, the evidence confirms that wellconducted observational cohort and caseecontrol studies can provide a similar level of internal validity as RCTs. EBM fails clinical practice when it adopts hierarchies to replace judgment with overly simplistic methods that create the appearance of quantitative precision that does not exist. Clinicians are decision makers and like all decision makers need to incorporate judgment, often based on idiosyncratic cases from their own and others clinical practices as a crucial element in the appraisal of evidence. It does not advance the practice of medicine to claim that RCTs are trustworthy and other forms of research are not. Advances in clinical care occur even in the absence of RCT evidence. In 2011, the FDA approved belimumab, a B-lymphocyte stimulator inhibitor, for the treatment of systemic lupus erythematosus (SLE) [13]. Before the approval of belimumab, the FDA last approved a drug to treat SLE in 1955 (hydroxychloroquine). What happened in the outcomes of SLE treatment during the
50 years when no EBM validated new treatments were approved? [13]. Urowitz followed a group of 1,241 patients seen at the University of Toronto between 1970 and 2005 [14]. Over the 36-year period of the study, 211 deaths occurred. The overall standardized mortality ratio was 12.6 [95% confidence interval (CI): 9.1, 17.4] in the first decade and 3.46 (95% CI: 2.7, 4.4) in the last decade when stratified by calendar year of enrollment. Some of the improved mortality was attributable to recognition of mild disease, but a large part was due to improved management achieved through clinical experimentation, careful observation, and the assessment of the effects of outcomes in practice. 5. Biology, social environment, and individualized care: medicine-based evidence EBM should not be abandoned. RCTs will continue to be valuable for drug companies and regulatory authorities who need to decide whether the average treatment results of new medicines will create more good than harm when used in the general population. For the new era of individualized patient care that is now upon us, we require new ways of thinking; new types of data describing clinical, social, behavioral, and neighborhood features of individuals; and new methods of analysis. The starting point for any clinically applicable analysis of personalized care is a detailed characterization of the individual patient. This apparently innocent recommendation is sure to meet with great resistance. Clinical medicine is all about individual care, but clinical research has little experience with studies that are oriented around individual level design and analysis. Individual level data will need to emphasize detailed and repeated measure over time of each subject across multiple domains including deep biology, clinical physiology, social and behavioral features, and work and neighborhood description. The social environment in which the patient is embedded matters a great deal. All of these data need to be amalgamated into a unified ‘‘profile’’ and, for diverse types of patients, assembled into an archive to enable identification of approximate matches to a given patient. The clinician seeking guidance for management of an individual patient starts with the patient profile and then finds approximate matches in the archive that describe how like patients responded to a contemplated treatment and alternative treatments. Medicine-based evidence (MBE) as a concept was elegantly described over 20 years ago [15]. As envisioned now, MBE will build an archive of profiles that emerge from all types of studies and data sources: RCTs, cohorts, caseecontrol studies, disease registries, and more. Investigators need to embrace the challenge of developing new methods and improving existing ones. Investigators and clinicians need to reject dogmatic assertions about the hierarchy of evidence. Finally, all who are involved in the care of patients need to recognize that the best decisions for the
R.I. Horwitz, B.H. Singer / Journal of Clinical Epidemiology 84 (2017) 14e17
care of patients require the role of judgment in the research that generates evidence, judgment in the interpretation of evidence, and judgment in the application of evidence to the care of individual patients.
[7]
[8]
References [1] Streptomycin treatment of pulmonary tuberculosis: a Medical Research Council Investigation. Br Med J 1948;2:769. [2] Horwitz RI, Singer BH, Makuch RW, Viscoli CM. Can treatment that is helpful on average be harmful to some patients? A study of the conflicting information needs of clinical inquiry and drug regulation. J Clin Epidemiol 1996;49:395e400. [3] Rawlins MD. Harveian Oration. De Testimonia on the evidence for decisions about the use of therapeutic interventions. Lancet 2008; 372:2152e61. [4] Bartlett C, Doyal L, Ebrahim S, Davey P, Bachmann M, Egger M, et al. The causes and effects of socio-demographic exclusions from clinical trials. Health Technol Assess 2005;9:38. [5] Malmstrom K, Rodriguez-Gomez G, Guerra J, Villaran C, Pi~neiro A, Wei LX, et al. Oral montelukast, inhaled beclomethasone, and placebo for chronic asthma. A randomized, controlled trial. Ann Intern Med 1999;130:487e95. [6] Price D, Musgrave MD, Shepstone L, Hillyer EV, Simms E, Gilbert RT, et al. Leukotriene antagonists as first-line or add-
[9]
[10]
[11] [12] [13] [14]
[15]
17
on asthma-controller therapy. N Engl J Med 2011;364: 1695e707. Kernan WN, Viscoli CM, Brass LM, Broderick JP, Brott T, Feldmann E, et al. Phenylpropanolamine and the risk of hemorrhagic stroke. N Engl J Med 2000;343:1826e32. Graham DJ, Green L, Senior JR, Nourjah P. Troglitazone-induced liver failure: a case study. Am J Med 2003;114:299e306. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342:1887e92. Owens DK, Lohr KN, Atkins D, Treadwell JR, Reston JT, et al. Grading the strength of a body of evidence when comparing medical interventions. In: Agency for Healthcare Research and Quality Methods Grade for Comparative Effectiveness Reviews. [posted July 2009]. Available at http://effectivehealthcare.ahrq.gov/healthInfo. cfm?infotype5rr&ProcessID560. Hill AB. Reflections on controlled trial. Ann Rheum Dis 1966;25: 107. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med 2000;342:1878e86. Stohl W, Hilbert DM. The discovery and development of belimumab: the anti-BLyS-lupus connection. Nat Biotechnol 2012;30:69e77. Urowitz MB, Gladman DD, Tom BD, Iba~nez D, Farewell VT. Changing patterns in mortality and disease outcomes for patients with systemic lupus erythematosus. J Rheumatol 2008;35:2152e8. Knottnerus JA, Dinant GJ. Medicine based evidence, a prerequisite for evidence based medicine. BMJ 1997;315:1109e10.