- Email: [email protected]
OBSOLETE: Clinical Epidemiology
Clinical Epidemiology☆ DL Sackett and R Brian Haynes, Kilgore S. Trout Research & Education Centre a Irish Lake, Markdale, ON, Canada ã 2014 Elsevier Inc. All rights reserved.
Definitions of Clinical Epidemiology A Brief History of Clinical Epidemiology Typical Clinical Epidemiologic Investigations Education Programs for Training Clinical Epidemiologists Clinical Epidemiology’s Detractors The Current State of Clinical Epidemiology Clinical Epidemiology and the Clinical Journals Clinical Epidemiologic Contributions to Five Recent Evolutions Evidence Generation The Critical Appraisal, Storage, and Retrieval of Evidence Evidence-based Medicine Evidence-synthesis and the Cochrane Collaboration Challenges and Frontiers References
1 1 2 2 3 3 3 3 4 4 4 4 4 5
Definitions of Clinical Epidemiology Clinical epidemiology is the application of epidemiological and related methods from a clinical (rather than public health) perspective. Whereas classical epidemiology studies the distribution and determinants of disease at a population level, clinical epidemiology employs these methods to study the etiology, diagnosis, prognosis, and treatment of health problems of individual and groups of patients. In its inception in the mid-19th century, clinical epidemiology employed methodological strategies and tactics mostly confined to epidemiology and biostatistics. Over the past 6 decades it has evolved in two ways. First, it has widened the research methods it employs, including health economics and the social and behavioural sciences. Second, it has expanded the uses to which these methods are put: the explosion in the generation and accessibility of relevant evidence, the critical appraisal of that evidence for its validity and clinical applicability, the efficient storage and retrieval of that evidence, the development of evidence-based medicine and health care, the systematic review and synthesis of evidence, and the understanding and enhancement of knowledge translation and implementation. While the scope of application continues to grow rapidly in the “information age”, the core scientific approach of clinical epidemiology remains consistent: development and deployment of methods and tools, from many scientific disciplines, that minimize bias in seeking answers about the nature and management of threats to, and disorders of, health.
A Brief History of Clinical Epidemiology The term “clinical epidemiology” was introduced by John Paul (born 1893 – died in 1971), an infectious disease internist who was appointed head of the Section of Preventive Medicine in Yale’s Department of Medicine in 1940. In his president’s address to the American Society for Clinical Investigation in 1938 (when it was still an organization with broad interests that included intact humans), he proposed clinical epidemiology as a “new basic science for preventive medicine” in which the exploration of relevant aspects of human ecology and public health began with the study of individual patients” (Paul, 1938). Dr. Paul wrote the first book and offered the first course in clinical epidemiology for undergraduate medical students. However, his concept of clinical epidemiology had a population rather than individual patient orientation in which he described the role of the clinical epidemiologist as being “like that of a detective visiting the scene of the crime” who then “branches out into the setting in which that individual became ill.” Thus the procedure in his course for 3rd and 4th year Yale medical students was to “start the student at the bedside and lead him gradually away from it.” This was in sharp contrast to the orientation of other pioneers who, although they didn’t refer to themselves as clinical epidemiologists, exemplified the application of epidemiology in bedside neonatology (where William Silverman showed that the traditional prophylactic antimicrobial therapy of babies with kernicterus did more harm than good) and gastroenterology (where Thomas Chalmers showed that the traditional regimen of prolonged bed rest for Type A hepatitis was unnecessary). The shift in the focus of clinical epidemiology from community ecology to individual patients and groups of patients took place in the 1960s in the form of the first Clinical Epidemiology Research Unit in the Department of Medicine at the State University of ☆
Change History: August 2014. DL Sackett and R Brian Haynes made changes in authors and updated abstract, keywords, text, and references.
Reference Module in Biomedical Research
http://dx.doi.org/10.1016/B978-0-12-801238-3.02865-8
1
2
Clinical Epidemiology
New York at Buffalo, USA in 1966, followed shortly by the Department of Clinical Epidemiology and Biostatistics at McMaster Medical School in Canada in 1967. In the prospectus for each of them clinical epidemiology was defined as “the application, by a physician who provides direct patient care, of epidemiologic and biostatistical methods to the study of diagnostic and therapeutic processes in order to effect an improvement in health” (Sackett, 1969). Thus, at McMaster the external, public health orientation was set aside and replaced with a focus on individual patients and groups of patients in clinical, not community, settings. In 1968 Alvan Feinstein published a series of papers on clinical epidemiology in the Annals of Internal Medicine (Feinstein, 1968). He defined the “territory” of clinical epidemiology as: “the clinicostatistical study of diseased populations. The intellectual activities of this territory include the following: the occurrence rates and geographic distribution of disease; the patterns of natural and post-therapeutic events that constitute varying clinical courses in the diverse spectrum of disease; and the clinical appraisal of therapy. The contemplation and investigation of these or allied topics constitute a medical domain that can be called clinical epidemiology.” Thus, he cast a wider net, and included elements of classical epidemiology and public health. His inclusion of public health in his definition of clinical epidemiology was repeated 18 years later in his book of that name: “clinical epidemiology represents the way in which classical epidemiology, traditionally oriented toward general strategies in the public health of community groups, has been enlarged to include clinical decisions in personal-encounter care for individual patients.“ Early clinical epidemiologists received important encouragement from Archie Cochrane, who provided a powerful rationale for the rigorous scientific evaluation of diagnosis, treatment, and preventive medicine (Cochrane, 1972). The first modern textbook in clinical epidemiology was written by Robert Fletcher, Suzanne Fletcher and Edward Wagner at the University of North Carolina, published in 1982. Now in its fourth edition, it continues to be a favoured introductory text. It was followed by ones from McMaster (now in its 3rd edition) and Yale in 1985, from the University of Washington in 1986 (now in its 3rd edition), from McGill in 1988, and from the University of California at San Francisco (now in its 4th Edition). Each has its own flavour and niche, and they are now available in several languages.
Typical Clinical Epidemiologic Investigations Typical clinical epidemiological investigations include: (1) Comparing patients’ symptoms and signs with the results of “reference standard” diagnostic tests. High quality studies of this sort sample patients in whom it is clinically sensible to suspect a specific diagnosis (e.g., patients suspected of chronic airflow limitation), carry out specific bits of the clinical examination (e.g., the position of the thyroid cartilage relative to the suprasternal notch), and then carry out an independent, “reference standard” test (e.g., spirometry) while “blind” to the results of the clinical examination. The results of these studies, commonly expressed as likelihood ratios, improve both the accuracy and efficiency of the clinical examination. In doing so, they have confirmed the usefulness of some traditional signs and symptoms (e.g., the presence of an S3 gallop on cardiac auscultation of a patient with suspected heart failure), added new signs and symptoms (e.g., clinical prediction rules for deep vein thrombosis and the “Ottawa ankle rule” for ruling out the need for ankle radiographs) and, equally important, shown that other signs and symptoms are useless (e.g., the tourniquet test for carpal tunnel syndrome). (2) Relating patients’ later outcomes (“prognoses”) to the results of earlier (“baseline”) clinical and laboratory findings. High quality studies of this sort sample patients at the very start of their illness (an “inception cohort”), perform baseline clinical and laboratory examinations on them, and then follow them to the conclusion of their disease. The results of these studies, often referred to now as “clinical prediction guides”, provide more accurate predictions and advice to patients at the start of an illness. (3) Randomized clinical trials in which consenting patients are assigned, by a system analogous to tossing a coin, to receive or not receive a new (“experimental”) treatment and then closely followed for the occurrence or prevention of unfavourable outcomes. High quality studies of this sort have been applied to medications (e.g., caffeine for premature babies), operations (e.g., carotid endarterectomy for threatened stroke), behavioural and educational interventions (e.g., behavioural manoeuvres for improving compliance with medications and exercise), health professionals (e.g., the nurse practitioner), and the like. The results of these studies determine whether new treatments or other health care interventions do more good than harm. For example, the preventive and therapeutic interventions for coronary heart disease that have been validated in randomized trials are credited with halving both the incidence and case-fatality of myocardial infarction in high-income countries. (4) Cluster randomized trials in which groups of patients, clinicians, clinics, hospitals or communities are randomly allocated to receive or not receive an intervention to improve the application of validated health interventions or services. This approach has become the mainstay of knowledge translation research and implementation science and is used, for example, to assess the value of computerized clinical decision support, continuing education, and quality improvement interventions.
Education Programs for Training Clinical Epidemiologists Opportunities for clinicians to obtain education and training in clinical epidemiology gradually spread from Yale and McMaster to other North American health sciences centres and to centres in Europe and the Far East. Combined training in clinical medicine and
Clinical Epidemiology
3
clinical epidemiology greatly expanded in the US in 1974 with the creation of the Robert Woods Johnson Clinical Scholars Program. The internationalization of clinical epidemiology began in 1980 when Kerr White and the Rockefeller Foundation initiated the International Clinical Epidemiology Network (INCLEN). In this programme, young clinicians from low-income countries came for training in clinical epidemiology to “training Centres” at McMaster in Canada, Newcastle in Australia and the University of Pennsylvania. A key element of their career development was linkage to a mentor who spent part of each year working with them back at their home institutions. The organization now includes 89 medical institutions and 31 training centres in 34 countries. In the eyes of many, its most important accomplishments have been the repeated redefinition of clinical epidemiology to suit local needs and the taking over of the training of clinical epidemiologists by regional centres in Africa, China, India, Latin America, and South East Asia.
Clinical Epidemiology’s Detractors Clinical epidemiology has not been without its detractors, especially among more traditional epidemiology departments who perceived (often correctly) their loss of resources and bright young minds to this new discipline. For example, in 1983 Walter Holland urged the abandonment of the term clinical epidemiology altogether (Holland, 1983). While acknowledging its usefulness over the previous 15 years, he found it a divisive term that conferred “respectability” only on those epidemiologists who practiced medicine, created the impression that one form of teaching (using epidemiology for solving clinical problems) was more appropriate than another (mastering classical epidemiological methods), and fashioned students’ perceptions of the priorities and needs of societies. In response to his criticism, it was suggested that the distinction between clinical and non-clinical epidemiologists was on a nominal, not ordinal, scale, but that his other criticisms were not only true, but to be applauded: clinical epidemiology was a better way to teach medical students, and clinical epidemiology was reshaping the perceptions of not only medical students (who began to see it as a relevant basic science) but entire faculties (departments of clinical epidemiology were growing in number and size; clinical departments were carrying out more and better “clinical-practice” research), and learned societies were acknowledging the relevance of clinical epidemiology to “clinical research” in ways that classical epidemiology had been unable to achieve (Sackett, 1984).
The Current State of Clinical Epidemiology Having established itself, gained formal recognition at universities, granting agencies, and learned societies, and populated academic departments and research groups around the world, the field of clinical epidemiology became increasingly able to emphasize its similarities to, rather than its differences from, classical public health epidemiology and the related sciences of economics, political science, psychology, and sociology. This is perhaps best seen in low- and middle-income countries, where clinical epidemiologists have begun to perform highly pragmatic cluster trials of public health interventions (such as insecticidebearing mosquito nets) that have incorporated economic analyses and sociological inquiries. As predicted by Walter Spitzer, all of these disciplines carry out and collaborate in studies of “diagnostic and therapeutic processes in order to effect an improvement in health.” (Spitzer, 1986)
Clinical Epidemiology and the Clinical Journals Although a wide spectrum of clinical journals have published the concepts, methods and results of clinical epidemiological research, and The Journal of Clinical Epidemiology has been a natural home for the discipline, some general medical journals also fostered the field and its recent evolutions. In the 1970’s the Journal of Clinical Pharmacology and Therapeutics turned Donald Mainland’s “Notes from a Laboratory of Medical Statistics” over to Alvan Feinstein for his landmark series in “Clinical Biostatistics.” In the 1980’s the Canadian Medical Association Journal hosted series on “How to Read Clinical Journals” and “How to Interpret Diagnostic Data” from the clinical epidemiology group at McMaster. In the 1990’s, the Journal of the American Medical Association initiated the “Rational Clinical Examination” series that hosted reviews of the accuracy and precision of the clinical history and examination. JAMA went on to host a series of “Users’ Guides to the Medical Literature” that were collated into a major text. Throughout this era, the Annals of Internal Medicine published several fundamental papers and series, culminating in the ACP Journal Club and a series of journals of secondary publication. The British Medical Journal published several “evidence-based” journals and became a major international publisher of clinical epidemiological research.
Clinical Epidemiologic Contributions to Five Recent Evolutions Clinical epidemiology has played a central or major role in five recent evolutions (some say revolutions) in health care: in evidence generation, its rapid critical appraisal, its efficient storage and retrieval, evidence-based medicine, and evidence synthesis.
4
Clinical Epidemiology
Evidence Generation The evolution in evidence generation since 1970, although most easily documented in the growth in reports of and about the randomized trial (with more of them published in the single year 2000 than in the entire decade 1965–75), is paralleled by similar, although less spectacular, increases in the numbers and sophistication of reports about diagnosis, prognosis, and the appropriateness and quality of clinical care. Clinical epidemiologists are providing leadership in both the generation and continuing methodological development of this burgeoning body of clinically-relevant evidence.
The Critical Appraisal, Storage, and Retrieval of Evidence The price to be paid for this vast increase in relevant evidence was an increasing difficulty in finding it, retrieving it, and keeping up-to-date with it. By 1972 there were about 4M articles published in the biomedical literature per year (in all languages). Restricting one’s reading to just the journals that provided the content that is sound and relevant for internal medicine required clinicians to read 33 articles every day of the year. When the dramatic decline in the currency of general medical knowledge after certification was documented by a group of clinical epidemiologists, it became impossible to ignore this growing problem. A second problem became evident when this growing body of evidence was subjected to the critical appraisal of its validity: the majority of it was found wanting. These two situations combined to place clinicians at increasing risks of “drowning in doubtful data.” The parallel evolutions in the rapid critical appraisal of evidence (for its validity and potential clinical usefulness) and in the efficient storage and rapid retrieval of evidence combined to rescue clinicians who were striving to track down the evidence than might help their patients. Several clinical epidemiologists, as well as library scientists, statisticians, and qualitative researchers made contributions to these parallel evolutions, most notably Brian Haynes at McMaster University, who brought these evolutionary streams together in powerful and clinically-relevant ways (Haynes et al., 1994). The example he set by reducing the internal medicine literature to just the 2% that was both valid and clinically relevant in the ACP Journal Club introduced the revolution that today provides front line clinicians in a number of clinical fields with manageable chunks of up-to-date, reliable evidence, right at the bedside (Wu and Straus, 2006). These days, clinical epidemiology knowledge refineries provide online alerts about current best evidence for health care, customized for a broad range of clinical disciplines, and ongoing support for updating online clinical texts, systematic reviews, and clinical practice guidelines.
Evidence-based Medicine As more and more clinicians, armed with the strategies and tactics of clinical epidemiology, cared for more and more patients, they began to evolve the final, vital link between evidence and direct patient care. Building on the prior evolutions, and manifest in clinically useful measures such as Andreas Laupacis’s NNT (the Number of patients a clinician would Need to Treat in order to prevent one more bad outcome) (Laupacis et al., 1988), and often incorporating the patient’s own values and expectations as in Sharon Straus’s LHH (the Likelihood that a treatment would Help vs. Harm the patient’s achievement of their health objectives) (Straus, 2002), the revolution of Evidence-Based Medicine was introduced by Gordon Guyatt (Evidence-Based Medicine Working Group, 1992). Since its first mention in 1992, its ideas about the use (rather than just critical appraisal) of evidence in patient care and in health professional education have spread worldwide and have been adopted not only by a broad array of clinical disciplines but also by health care planners and evaluators. In 2006, the British Medical Journal designated evidence-based medicine one of the 15 greatest medical breakthroughs since 1840.
Evidence-synthesis and the Cochrane Collaboration Simultaneous with these other evolutions and revolutions, and both supporting and building upon them, has been the evidencesynthesis evolution of strategies and tactics for assembling and systematically reviewing the totality of evidence about the effects of health care. This evolution is epitomized in the Cochrane Collaboration, a worldwide collaboration of patients, clinicians and methodologists, who prepare, maintain and promote the accessibility of systematic reviews of the effects of healthcare interventions (Cochrane Collaboration, 2007). Conceived and initiated by Iain Chalmers in Oxford, this undertaking has been characterized as equal in importance to the human genome project. Many consider the contributions of clinical epidemiologists to evidence-synthesis their greatest accomplishment since the term was introduced 65 years ago.
Challenges and Frontiers (1) Despite the advances of knowledge about the cause, course, diagnosis, prevention and treatment of health disorders, the pace of improvement in health care and patient outcomes remains distressingly slow. The barriers to evidence application are not well understood and effective, acceptable, affordable interventions to overcome them are limited. Clinical epidemiologists are leading the assault on these barriers under the 21st century banners of knowledge translation research, implementation science, comparative effectiveness research and patient-oriented outcomes research. Greater collaboration than exists today
Clinical Epidemiology
5
will be needed with health policy researchers and managers, practitioners, patients and the public if the benefits of health research are to be realized. (2) The era of “Big Data” is upon us, in which limitless amounts of health data in digital format, collected for purposes other than research, are available for automated statistical analysis. When these data have been collected with reasonable fidelity, and are analysed with careful attention to methodology for managing and minimizing bias, they may greatly facilitate addressing certain types of questions. For example, tapping population-based health registries can greatly reduce the costs of follow-up in clinical trials for some outcomes such as deaths, hospitalizations, procedures, and medications prescribed within insurance plans. But analysis of such data without regard for the meaning, accuracy, timeliness and completeness of the data leads to false positive and false negative results that will predictably swamp the true signals from sound research. Clinical epidemiologists are needed here to set standards for when and how to analyze non-research data, and how to interpret and weigh the findings.
References Cochrane AL (1972) Effectiveness and efficiency: Random reflections on Health Services. London: Nuffield Provincial Hospitals Trust. Cochrane Collaboration (2007). http://www.cochranelibrary.com/Collaboration/. Evidence-Based Medicine Working Group (1992) Evidence-based medicine. A new approach to teaching the practice of medicine. Journal of the American Medical Association 268: 2420–2425. Feinstein AR (1968) Clinical epidemiology. The populational experiments of nature and of man in human illness. Annals of Internal Medicine 69: 807–820. Haynes RB, Wilczynski NL, McKibbon KA, Walker CJ, and Sinclair JC (1994) Developing optimal search strategies for detecting clinically sound studies in MEDLINE. Journal of the American Medical Informatics Association 1994(1): 447–458. Holland W (1983) Inappropriate terminology. International Journal of Epidemiology 12: 5–7. Laupacis A, Sackett DL, and Roberts RS (1988) An assessment of clinically useful measures of the consequences of treatment. New England Journal of Medicine 318: 1728–1733. Paul JR (1938) Clinical epidemiology. Journal of Clinical Investigation 17: 539–541. Sackett DL (1969) Clinical epidemiology. American Journal of Epidemiology 89: 125–128. Sackett DL (1984) Three cheers for clinical epidemiology. International Journal of Epidemiology 13: 117–119. Spitzer WO (1986) Clinical epidemiology. Journal of Chronic Diseases 39: 411–415. Straus SE (2002) Individualizing treatment decisions: The likelihood of being helped versus harmed. Evaluation and the Health Professions 25: 210–224. Wu R and Straus SE (2006) Use of PDAs in health care: Systematic review of the literature. Evidence for handheld electronic medication records in improving care: A systematic review. BioMedCentral Medical Informatics and Decision Making 6: 26.
Further Reading Cochrane AL (1999) Effectiveness and efficiency: Random reflections on health services, 3rd edn. London: Royal Society of Medicine Press. Feinstein AR (1985) Clinical epidemiology; The architecture of clinical research. Philadelphia: WB Saunders. Fletcher RW and Fletcher SW (2005) Clinical epidemiology: The essentials, 4th edn. Philadelphia, PA: Lippincott Williams & Wilkins. Guyatt GH and Rennie D (eds.) (2002) User’s guides to the medical literature. A manual for evidence-based clinical practice. Chicago: American Medical Association Press. Haynes RB, Sackett DL, Guyatt GH, and Tugwell P (2006) Clinical epidemiology: How to do clinical practice research. Philadelphia, PA: Lippincott Williams & Wilkins. Hulley SB, Cummings SR, Browner WS, Grady DG, and Newman DB (2013) Designing clinical research, 4th edn. Philadelphia, PA: Lippincott, Williams & Wilkins. Kramer MS (1991) Clinical epidemiology and biostatistics. Berlin: Springer-Verlag. Paul JR (1966) Clinical epidemiology. Chicago: University of Chicago Press, Revised Edition. Straus SE, Richardson WS, Glasziou P, and Haynes RB (2005) Evidence-based medicine. How to practice and teach EBM, 3rd edn. Edinburgh: Elsevier Churchill Livingstone. Weiss NS (2006) Clinical epidemiology: The study of the outcome of illness, 3rd edn. Oxford: Oxford University Press.
Relevant Websites http://www.cochranelibrary.com/Collaboration/ – Cochrane Collaboration (2007). http://www.inclen.org/ – International Clinical Epidemiology Network.
Definitions of Clinical Epidemiology A Brief History of Clinical Epidemiology Typical Clinical Epidemiologic Investigations Education Programs for Training Clinical Epidemiologists Clinical Epidemiology’s Detractors The Current State of Clinical Epidemiology Clinical Epidemiology and the Clinical Journals Clinical Epidemiologic Contributions to Five Recent Evolutions Evidence Generation The Critical Appraisal, Storage, and Retrieval of Evidence Evidence-based Medicine Evidence-synthesis and the Cochrane Collaboration Challenges and Frontiers References
1 1 2 2 3 3 3 3 4 4 4 4 4 5
Definitions of Clinical Epidemiology Clinical epidemiology is the application of epidemiological and related methods from a clinical (rather than public health) perspective. Whereas classical epidemiology studies the distribution and determinants of disease at a population level, clinical epidemiology employs these methods to study the etiology, diagnosis, prognosis, and treatment of health problems of individual and groups of patients. In its inception in the mid-19th century, clinical epidemiology employed methodological strategies and tactics mostly confined to epidemiology and biostatistics. Over the past 6 decades it has evolved in two ways. First, it has widened the research methods it employs, including health economics and the social and behavioural sciences. Second, it has expanded the uses to which these methods are put: the explosion in the generation and accessibility of relevant evidence, the critical appraisal of that evidence for its validity and clinical applicability, the efficient storage and retrieval of that evidence, the development of evidence-based medicine and health care, the systematic review and synthesis of evidence, and the understanding and enhancement of knowledge translation and implementation. While the scope of application continues to grow rapidly in the “information age”, the core scientific approach of clinical epidemiology remains consistent: development and deployment of methods and tools, from many scientific disciplines, that minimize bias in seeking answers about the nature and management of threats to, and disorders of, health.
A Brief History of Clinical Epidemiology The term “clinical epidemiology” was introduced by John Paul (born 1893 – died in 1971), an infectious disease internist who was appointed head of the Section of Preventive Medicine in Yale’s Department of Medicine in 1940. In his president’s address to the American Society for Clinical Investigation in 1938 (when it was still an organization with broad interests that included intact humans), he proposed clinical epidemiology as a “new basic science for preventive medicine” in which the exploration of relevant aspects of human ecology and public health began with the study of individual patients” (Paul, 1938). Dr. Paul wrote the first book and offered the first course in clinical epidemiology for undergraduate medical students. However, his concept of clinical epidemiology had a population rather than individual patient orientation in which he described the role of the clinical epidemiologist as being “like that of a detective visiting the scene of the crime” who then “branches out into the setting in which that individual became ill.” Thus the procedure in his course for 3rd and 4th year Yale medical students was to “start the student at the bedside and lead him gradually away from it.” This was in sharp contrast to the orientation of other pioneers who, although they didn’t refer to themselves as clinical epidemiologists, exemplified the application of epidemiology in bedside neonatology (where William Silverman showed that the traditional prophylactic antimicrobial therapy of babies with kernicterus did more harm than good) and gastroenterology (where Thomas Chalmers showed that the traditional regimen of prolonged bed rest for Type A hepatitis was unnecessary). The shift in the focus of clinical epidemiology from community ecology to individual patients and groups of patients took place in the 1960s in the form of the first Clinical Epidemiology Research Unit in the Department of Medicine at the State University of ☆
Change History: August 2014. DL Sackett and R Brian Haynes made changes in authors and updated abstract, keywords, text, and references.
Reference Module in Biomedical Research
http://dx.doi.org/10.1016/B978-0-12-801238-3.02865-8
1
2
Clinical Epidemiology
New York at Buffalo, USA in 1966, followed shortly by the Department of Clinical Epidemiology and Biostatistics at McMaster Medical School in Canada in 1967. In the prospectus for each of them clinical epidemiology was defined as “the application, by a physician who provides direct patient care, of epidemiologic and biostatistical methods to the study of diagnostic and therapeutic processes in order to effect an improvement in health” (Sackett, 1969). Thus, at McMaster the external, public health orientation was set aside and replaced with a focus on individual patients and groups of patients in clinical, not community, settings. In 1968 Alvan Feinstein published a series of papers on clinical epidemiology in the Annals of Internal Medicine (Feinstein, 1968). He defined the “territory” of clinical epidemiology as: “the clinicostatistical study of diseased populations. The intellectual activities of this territory include the following: the occurrence rates and geographic distribution of disease; the patterns of natural and post-therapeutic events that constitute varying clinical courses in the diverse spectrum of disease; and the clinical appraisal of therapy. The contemplation and investigation of these or allied topics constitute a medical domain that can be called clinical epidemiology.” Thus, he cast a wider net, and included elements of classical epidemiology and public health. His inclusion of public health in his definition of clinical epidemiology was repeated 18 years later in his book of that name: “clinical epidemiology represents the way in which classical epidemiology, traditionally oriented toward general strategies in the public health of community groups, has been enlarged to include clinical decisions in personal-encounter care for individual patients.“ Early clinical epidemiologists received important encouragement from Archie Cochrane, who provided a powerful rationale for the rigorous scientific evaluation of diagnosis, treatment, and preventive medicine (Cochrane, 1972). The first modern textbook in clinical epidemiology was written by Robert Fletcher, Suzanne Fletcher and Edward Wagner at the University of North Carolina, published in 1982. Now in its fourth edition, it continues to be a favoured introductory text. It was followed by ones from McMaster (now in its 3rd edition) and Yale in 1985, from the University of Washington in 1986 (now in its 3rd edition), from McGill in 1988, and from the University of California at San Francisco (now in its 4th Edition). Each has its own flavour and niche, and they are now available in several languages.
Typical Clinical Epidemiologic Investigations Typical clinical epidemiological investigations include: (1) Comparing patients’ symptoms and signs with the results of “reference standard” diagnostic tests. High quality studies of this sort sample patients in whom it is clinically sensible to suspect a specific diagnosis (e.g., patients suspected of chronic airflow limitation), carry out specific bits of the clinical examination (e.g., the position of the thyroid cartilage relative to the suprasternal notch), and then carry out an independent, “reference standard” test (e.g., spirometry) while “blind” to the results of the clinical examination. The results of these studies, commonly expressed as likelihood ratios, improve both the accuracy and efficiency of the clinical examination. In doing so, they have confirmed the usefulness of some traditional signs and symptoms (e.g., the presence of an S3 gallop on cardiac auscultation of a patient with suspected heart failure), added new signs and symptoms (e.g., clinical prediction rules for deep vein thrombosis and the “Ottawa ankle rule” for ruling out the need for ankle radiographs) and, equally important, shown that other signs and symptoms are useless (e.g., the tourniquet test for carpal tunnel syndrome). (2) Relating patients’ later outcomes (“prognoses”) to the results of earlier (“baseline”) clinical and laboratory findings. High quality studies of this sort sample patients at the very start of their illness (an “inception cohort”), perform baseline clinical and laboratory examinations on them, and then follow them to the conclusion of their disease. The results of these studies, often referred to now as “clinical prediction guides”, provide more accurate predictions and advice to patients at the start of an illness. (3) Randomized clinical trials in which consenting patients are assigned, by a system analogous to tossing a coin, to receive or not receive a new (“experimental”) treatment and then closely followed for the occurrence or prevention of unfavourable outcomes. High quality studies of this sort have been applied to medications (e.g., caffeine for premature babies), operations (e.g., carotid endarterectomy for threatened stroke), behavioural and educational interventions (e.g., behavioural manoeuvres for improving compliance with medications and exercise), health professionals (e.g., the nurse practitioner), and the like. The results of these studies determine whether new treatments or other health care interventions do more good than harm. For example, the preventive and therapeutic interventions for coronary heart disease that have been validated in randomized trials are credited with halving both the incidence and case-fatality of myocardial infarction in high-income countries. (4) Cluster randomized trials in which groups of patients, clinicians, clinics, hospitals or communities are randomly allocated to receive or not receive an intervention to improve the application of validated health interventions or services. This approach has become the mainstay of knowledge translation research and implementation science and is used, for example, to assess the value of computerized clinical decision support, continuing education, and quality improvement interventions.
Education Programs for Training Clinical Epidemiologists Opportunities for clinicians to obtain education and training in clinical epidemiology gradually spread from Yale and McMaster to other North American health sciences centres and to centres in Europe and the Far East. Combined training in clinical medicine and
Clinical Epidemiology
3
clinical epidemiology greatly expanded in the US in 1974 with the creation of the Robert Woods Johnson Clinical Scholars Program. The internationalization of clinical epidemiology began in 1980 when Kerr White and the Rockefeller Foundation initiated the International Clinical Epidemiology Network (INCLEN). In this programme, young clinicians from low-income countries came for training in clinical epidemiology to “training Centres” at McMaster in Canada, Newcastle in Australia and the University of Pennsylvania. A key element of their career development was linkage to a mentor who spent part of each year working with them back at their home institutions. The organization now includes 89 medical institutions and 31 training centres in 34 countries. In the eyes of many, its most important accomplishments have been the repeated redefinition of clinical epidemiology to suit local needs and the taking over of the training of clinical epidemiologists by regional centres in Africa, China, India, Latin America, and South East Asia.
Clinical Epidemiology’s Detractors Clinical epidemiology has not been without its detractors, especially among more traditional epidemiology departments who perceived (often correctly) their loss of resources and bright young minds to this new discipline. For example, in 1983 Walter Holland urged the abandonment of the term clinical epidemiology altogether (Holland, 1983). While acknowledging its usefulness over the previous 15 years, he found it a divisive term that conferred “respectability” only on those epidemiologists who practiced medicine, created the impression that one form of teaching (using epidemiology for solving clinical problems) was more appropriate than another (mastering classical epidemiological methods), and fashioned students’ perceptions of the priorities and needs of societies. In response to his criticism, it was suggested that the distinction between clinical and non-clinical epidemiologists was on a nominal, not ordinal, scale, but that his other criticisms were not only true, but to be applauded: clinical epidemiology was a better way to teach medical students, and clinical epidemiology was reshaping the perceptions of not only medical students (who began to see it as a relevant basic science) but entire faculties (departments of clinical epidemiology were growing in number and size; clinical departments were carrying out more and better “clinical-practice” research), and learned societies were acknowledging the relevance of clinical epidemiology to “clinical research” in ways that classical epidemiology had been unable to achieve (Sackett, 1984).
The Current State of Clinical Epidemiology Having established itself, gained formal recognition at universities, granting agencies, and learned societies, and populated academic departments and research groups around the world, the field of clinical epidemiology became increasingly able to emphasize its similarities to, rather than its differences from, classical public health epidemiology and the related sciences of economics, political science, psychology, and sociology. This is perhaps best seen in low- and middle-income countries, where clinical epidemiologists have begun to perform highly pragmatic cluster trials of public health interventions (such as insecticidebearing mosquito nets) that have incorporated economic analyses and sociological inquiries. As predicted by Walter Spitzer, all of these disciplines carry out and collaborate in studies of “diagnostic and therapeutic processes in order to effect an improvement in health.” (Spitzer, 1986)
Clinical Epidemiology and the Clinical Journals Although a wide spectrum of clinical journals have published the concepts, methods and results of clinical epidemiological research, and The Journal of Clinical Epidemiology has been a natural home for the discipline, some general medical journals also fostered the field and its recent evolutions. In the 1970’s the Journal of Clinical Pharmacology and Therapeutics turned Donald Mainland’s “Notes from a Laboratory of Medical Statistics” over to Alvan Feinstein for his landmark series in “Clinical Biostatistics.” In the 1980’s the Canadian Medical Association Journal hosted series on “How to Read Clinical Journals” and “How to Interpret Diagnostic Data” from the clinical epidemiology group at McMaster. In the 1990’s, the Journal of the American Medical Association initiated the “Rational Clinical Examination” series that hosted reviews of the accuracy and precision of the clinical history and examination. JAMA went on to host a series of “Users’ Guides to the Medical Literature” that were collated into a major text. Throughout this era, the Annals of Internal Medicine published several fundamental papers and series, culminating in the ACP Journal Club and a series of journals of secondary publication. The British Medical Journal published several “evidence-based” journals and became a major international publisher of clinical epidemiological research.
Clinical Epidemiologic Contributions to Five Recent Evolutions Clinical epidemiology has played a central or major role in five recent evolutions (some say revolutions) in health care: in evidence generation, its rapid critical appraisal, its efficient storage and retrieval, evidence-based medicine, and evidence synthesis.
4
Clinical Epidemiology
Evidence Generation The evolution in evidence generation since 1970, although most easily documented in the growth in reports of and about the randomized trial (with more of them published in the single year 2000 than in the entire decade 1965–75), is paralleled by similar, although less spectacular, increases in the numbers and sophistication of reports about diagnosis, prognosis, and the appropriateness and quality of clinical care. Clinical epidemiologists are providing leadership in both the generation and continuing methodological development of this burgeoning body of clinically-relevant evidence.
The Critical Appraisal, Storage, and Retrieval of Evidence The price to be paid for this vast increase in relevant evidence was an increasing difficulty in finding it, retrieving it, and keeping up-to-date with it. By 1972 there were about 4M articles published in the biomedical literature per year (in all languages). Restricting one’s reading to just the journals that provided the content that is sound and relevant for internal medicine required clinicians to read 33 articles every day of the year. When the dramatic decline in the currency of general medical knowledge after certification was documented by a group of clinical epidemiologists, it became impossible to ignore this growing problem. A second problem became evident when this growing body of evidence was subjected to the critical appraisal of its validity: the majority of it was found wanting. These two situations combined to place clinicians at increasing risks of “drowning in doubtful data.” The parallel evolutions in the rapid critical appraisal of evidence (for its validity and potential clinical usefulness) and in the efficient storage and rapid retrieval of evidence combined to rescue clinicians who were striving to track down the evidence than might help their patients. Several clinical epidemiologists, as well as library scientists, statisticians, and qualitative researchers made contributions to these parallel evolutions, most notably Brian Haynes at McMaster University, who brought these evolutionary streams together in powerful and clinically-relevant ways (Haynes et al., 1994). The example he set by reducing the internal medicine literature to just the 2% that was both valid and clinically relevant in the ACP Journal Club introduced the revolution that today provides front line clinicians in a number of clinical fields with manageable chunks of up-to-date, reliable evidence, right at the bedside (Wu and Straus, 2006). These days, clinical epidemiology knowledge refineries provide online alerts about current best evidence for health care, customized for a broad range of clinical disciplines, and ongoing support for updating online clinical texts, systematic reviews, and clinical practice guidelines.
Evidence-based Medicine As more and more clinicians, armed with the strategies and tactics of clinical epidemiology, cared for more and more patients, they began to evolve the final, vital link between evidence and direct patient care. Building on the prior evolutions, and manifest in clinically useful measures such as Andreas Laupacis’s NNT (the Number of patients a clinician would Need to Treat in order to prevent one more bad outcome) (Laupacis et al., 1988), and often incorporating the patient’s own values and expectations as in Sharon Straus’s LHH (the Likelihood that a treatment would Help vs. Harm the patient’s achievement of their health objectives) (Straus, 2002), the revolution of Evidence-Based Medicine was introduced by Gordon Guyatt (Evidence-Based Medicine Working Group, 1992). Since its first mention in 1992, its ideas about the use (rather than just critical appraisal) of evidence in patient care and in health professional education have spread worldwide and have been adopted not only by a broad array of clinical disciplines but also by health care planners and evaluators. In 2006, the British Medical Journal designated evidence-based medicine one of the 15 greatest medical breakthroughs since 1840.
Evidence-synthesis and the Cochrane Collaboration Simultaneous with these other evolutions and revolutions, and both supporting and building upon them, has been the evidencesynthesis evolution of strategies and tactics for assembling and systematically reviewing the totality of evidence about the effects of health care. This evolution is epitomized in the Cochrane Collaboration, a worldwide collaboration of patients, clinicians and methodologists, who prepare, maintain and promote the accessibility of systematic reviews of the effects of healthcare interventions (Cochrane Collaboration, 2007). Conceived and initiated by Iain Chalmers in Oxford, this undertaking has been characterized as equal in importance to the human genome project. Many consider the contributions of clinical epidemiologists to evidence-synthesis their greatest accomplishment since the term was introduced 65 years ago.
Challenges and Frontiers (1) Despite the advances of knowledge about the cause, course, diagnosis, prevention and treatment of health disorders, the pace of improvement in health care and patient outcomes remains distressingly slow. The barriers to evidence application are not well understood and effective, acceptable, affordable interventions to overcome them are limited. Clinical epidemiologists are leading the assault on these barriers under the 21st century banners of knowledge translation research, implementation science, comparative effectiveness research and patient-oriented outcomes research. Greater collaboration than exists today
Clinical Epidemiology
5
will be needed with health policy researchers and managers, practitioners, patients and the public if the benefits of health research are to be realized. (2) The era of “Big Data” is upon us, in which limitless amounts of health data in digital format, collected for purposes other than research, are available for automated statistical analysis. When these data have been collected with reasonable fidelity, and are analysed with careful attention to methodology for managing and minimizing bias, they may greatly facilitate addressing certain types of questions. For example, tapping population-based health registries can greatly reduce the costs of follow-up in clinical trials for some outcomes such as deaths, hospitalizations, procedures, and medications prescribed within insurance plans. But analysis of such data without regard for the meaning, accuracy, timeliness and completeness of the data leads to false positive and false negative results that will predictably swamp the true signals from sound research. Clinical epidemiologists are needed here to set standards for when and how to analyze non-research data, and how to interpret and weigh the findings.
References Cochrane AL (1972) Effectiveness and efficiency: Random reflections on Health Services. London: Nuffield Provincial Hospitals Trust. Cochrane Collaboration (2007). http://www.cochranelibrary.com/Collaboration/. Evidence-Based Medicine Working Group (1992) Evidence-based medicine. A new approach to teaching the practice of medicine. Journal of the American Medical Association 268: 2420–2425. Feinstein AR (1968) Clinical epidemiology. The populational experiments of nature and of man in human illness. Annals of Internal Medicine 69: 807–820. Haynes RB, Wilczynski NL, McKibbon KA, Walker CJ, and Sinclair JC (1994) Developing optimal search strategies for detecting clinically sound studies in MEDLINE. Journal of the American Medical Informatics Association 1994(1): 447–458. Holland W (1983) Inappropriate terminology. International Journal of Epidemiology 12: 5–7. Laupacis A, Sackett DL, and Roberts RS (1988) An assessment of clinically useful measures of the consequences of treatment. New England Journal of Medicine 318: 1728–1733. Paul JR (1938) Clinical epidemiology. Journal of Clinical Investigation 17: 539–541. Sackett DL (1969) Clinical epidemiology. American Journal of Epidemiology 89: 125–128. Sackett DL (1984) Three cheers for clinical epidemiology. International Journal of Epidemiology 13: 117–119. Spitzer WO (1986) Clinical epidemiology. Journal of Chronic Diseases 39: 411–415. Straus SE (2002) Individualizing treatment decisions: The likelihood of being helped versus harmed. Evaluation and the Health Professions 25: 210–224. Wu R and Straus SE (2006) Use of PDAs in health care: Systematic review of the literature. Evidence for handheld electronic medication records in improving care: A systematic review. BioMedCentral Medical Informatics and Decision Making 6: 26.
Further Reading Cochrane AL (1999) Effectiveness and efficiency: Random reflections on health services, 3rd edn. London: Royal Society of Medicine Press. Feinstein AR (1985) Clinical epidemiology; The architecture of clinical research. Philadelphia: WB Saunders. Fletcher RW and Fletcher SW (2005) Clinical epidemiology: The essentials, 4th edn. Philadelphia, PA: Lippincott Williams & Wilkins. Guyatt GH and Rennie D (eds.) (2002) User’s guides to the medical literature. A manual for evidence-based clinical practice. Chicago: American Medical Association Press. Haynes RB, Sackett DL, Guyatt GH, and Tugwell P (2006) Clinical epidemiology: How to do clinical practice research. Philadelphia, PA: Lippincott Williams & Wilkins. Hulley SB, Cummings SR, Browner WS, Grady DG, and Newman DB (2013) Designing clinical research, 4th edn. Philadelphia, PA: Lippincott, Williams & Wilkins. Kramer MS (1991) Clinical epidemiology and biostatistics. Berlin: Springer-Verlag. Paul JR (1966) Clinical epidemiology. Chicago: University of Chicago Press, Revised Edition. Straus SE, Richardson WS, Glasziou P, and Haynes RB (2005) Evidence-based medicine. How to practice and teach EBM, 3rd edn. Edinburgh: Elsevier Churchill Livingstone. Weiss NS (2006) Clinical epidemiology: The study of the outcome of illness, 3rd edn. Oxford: Oxford University Press.
Relevant Websites http://www.cochranelibrary.com/Collaboration/ – Cochrane Collaboration (2007). http://www.inclen.org/ – International Clinical Epidemiology Network.