Jenicek M (1987). Translation

Jenicek M (1987). Méta-analyse en médecine. Évaluation et synthèse de l’information clinique et épidémiologique. St.Hyacinthe and Paris: EDISEM and Maloine Éditeurs.

Meta-analysis in medicine. Evaluation and synthesis of clinical and epidemiological information.

- personal experience and apprenticeship
- planned research on groups of patients
- synthesis of individual research studies: this is the sphere of meta-analysis in medicine

How can one know whether a medical article is good or bad?
How should one judge whether a presentation made at scientific meeting should be trusted?
How should one choose the clinical, therapeutic and preventive information necessary for making medical decisions, and how should one use it in practice?
How should one evaluate data, with a view to applying them at the level of health policies and administration?
How should one proceed when different studies on the same topic yield conflicting results?

Chapter 2. Evaluating an individual original research study, a medical article, or an essay

“All my life when I did not know how, I have looked for a book that told me how. …
A book is not enough, of course. It is what you bring to the book; your ability to
absorb the information it contains and make those facts, those insights, yours. If you
know how to absorb information, you can do just about everything – start your own
business, build a happy marriage, even shingle a roof.”

We have learned the theory and practice of medicine. We have mastered the rules of valid research and we know how to do scientific studies designed to address concrete, clearly defined problems.

Our teachers, our predecessors and our colleagues have inculcated in us some idea of the art and the science of medicine, without always specifying, however, the limitations of each (Table 0-0).

It remains to be known whether we have succeeded in confronting the great challenges of medicine today. Although there have been many mistakes and many blind alleys, the success of modern medicine is undeniable: reasonable control of diabetes, emergency medicine, organ transplantation, immunisations against viral illnesses, etc.

The results of such efforts and accomplishments is tangible. Over 200 biomedical journals are currently being published. The number of original articles, syntheses and other papers is increasing by 6 to 7% a year. Thus the volume of new medical information doubles every 10 to 15 years and increases tenfold in 23 to 50 years.¹

All medical information is not of the same quality. The reader must make an assessment and choose which information is most likely to be useful to him.

- Which summary information is the most valid and objective, among a multitude of individual observations (studies) on the same topic? For example, if several studies aim to evaluate the preventive role of physical activity on the aetiology of myocardial infarction, what is its quantitative importance across the hundreds of studies on the topic? Are there contradictory results?

- If one constructs a summary table of multiple pieces of medical information on the same topic, do other data in this body of evidence enable one to pose new questions, to discover unrecognised and important problems, to formulate new hypotheses, to direct further research?

These questions illustrate clearly the importance of obtaining good primary medical information, whether in an article or a study. An analysis of medical information, as structured and as complete as possible, ought to achieve this.

Aside from the analysis of primary information, one needs a kind of “research on research”, an analysis, critique and synthesis of multiple pieces of information. A kind of meta-analysis in medicine emerges. The quantitative and qualitative aspects of meta-analysis of research have been developed and applied in a variety of biological and human sciences. Medicine needs this more than most because the flood of medical information shows no signs of abating.

Originally, meta-analysis was defined as a method for the systematic and statistical integration of numerous results of research, for analysis and synthesis.

In medicine, we can think of meta-analysis as qualitative and quantitative evaluation, systematic synthesis and integration of available medical information, based on results of several sources of information or independent studies on the same topic”.

During his professional training and subsequent medical practice, the doctor soon becomes aware that he has learned some things, but not others. Studies of normal and abnormal morphology and function were followed by clinical training. During the latter he learned to recognise health problems and what to do with them. Finally, modern general epidemiology (basic or “classical”) and “clinical” epidemiology encouraged logical organisation and structured medical thinking and decision making. Gradually a new basic science of clinical medicine (in D.L. Sackett’s terms ²) is being born. The concept is not entirely new. A.R. Feinstein ³ devoted one of his books to “clinical judgement”. This trend has been reinforced by the efforts of E.A. Murphy ⁴ in formulating his “logic of medicine”. Most recently, O.S. Miettinen has proposed the term “theory of medicine”. As a result of his initiative and in the spirit of the Viennese “think tanks” of the beginning of the century, the Cercle de Montréal saw the light of day in 1985. The Cercle took on the task of maintaining, pursuing, stimulating and developing the domain of organised medical thinking.

Contemporary “classical” and “clinical” epidemiologists have developed an effective approach to health problems in populations and groups. They have done the same for the individual approach, in a logical, structured, organised, practical and effective conceptual framework. Among these activities, one in particular plays an absolutely crucial role: the acquisition of good quality medical information, equally indispensable for identifying important health problems as for working out how to deal with them, how to evaluate the results of efforts to tackle them, and how to anticipate the future course of health phenomena that will interest us in the long term. This book belongs to the new domain of the “logic of medicine” or the organisation of medical thinking.

Three steps are needed to arrive at a more complete understanding: after having mastered the ways of everyday practice of medicine, one needs to know how to do original, specific research studies, with a view to obtaining answers to specific questions. Third, one needs to learn how to integrate medical information. Systematic integration of scientific information is becoming, as suggested by Feldman ⁵, the subject of a veritable “mini-explosion” of new initiatives, of methodological development and of applications in a variety of spheres in research and practice. This applied domain, adapted and extended to medicine in this introductory book, corresponds to several terms in English ⁵: meta-analysis, integrative research review, research integration, research consolidation, data synthesis, quantitative assessment of research domains, combining studies, combining results, empirical cumulation, empirical evidence. The concern is always the same: to obtain the best information possible, which an individual research study can almost never achieve.

This book addresses this last domain: assessing medical information and retaining those elements that are most useful and which reflect reality most closely.

How many doctors have acquired the skills necessary to synthesise and review systematically multiple pieces of information? Such research synthesis though is essential in planning research, in interpreting results, in editing review articles and position papers, not to mention for taking practical, everyday medical decisions.

With this perspective, this book’s objective is to offer the reader the main rules of the game which apply to the analysis of primary information and the meta-analysis of the totality of knowledge on a given topic. The book covers general principles about obtaining data and medical information, primary and secondary analysis of an article, meta-analysis of the totality of medical information, and, finally, the practical implications of such more complete methodology dealing with important medical topics.

The book is aimed more at the “consumers” of professional and scientific information than at the “producers”. All the same, the methodological elements of analysis and meta-analysis of medical information have been highlighted and can be useful to researchers, whether in basic, clinical of public health sciences. They will understand what is expected from their results, what has been done, and what the consumers of their message look for in their reports.

Meta-analysis in medicine today is a considerably larger domain than that originally defined by Glass and Rosenthal. Besides its original quantitative dimension and structure, the qualitative dimension of research analysis and synthesis appears as its increasingly important counterpart and complement. Standardisation of criteria for assessing the quality of medical scientific studies is developing rapidly in several areas of clinical and epidemiological research. Thus, meta-analysis in medicine comprises both quantitative and qualitative methods. It is in this latter aspect that epidemiologists and clinical pharmacologists have introduced elements lacking in the original domains of psychology and education sciences, at least in terms of what is desirable for the medical domain. So it is that a kind of “epidemiology of medical information” is beginning bit by bit to see the light of day.

- First, we will try to structure the way one should evaluate the quality of original studies in medicine, as they are presented in specialist journals.

- Second, we will cover, technical aspects included, the means to obtain good quality information and/or new information about methods for analysis and synthesising information from several original studies on the same topic. The latter aspect represents a new asset, albeit not without its limitations, for acquiring new knowledge in medicine, which cannot be provided by individual studies.

This subject will be presented primarily from the point of view of the reader, who will often be a novice information consumer. The authors and designers of new studies will doubtless want more technical information on the topics covered. They will find this in the bibliography at the end of each chapter. As the majority of specialised and technical publications are written for information producers rather than information consumers, we have tried to restore a balance.

Furthermore, it is not our intention to present a manual on meta-analysis in medicine. Any such aim would discourage people reading about the topic for the first time. A colleague and friend recently remarked: “… I like short, succinct documents, which I can read in two or three evenings…” We suggest this book should be regarded as an hors-d’oeuvre; the main course will provided elsewhere, hopefully in the not-too-distant future. The relevance of the topic certainly justifies it.

If we succeed in alerting readers to the importance of the topics covered in this short text, our expectations will have been more than accomplished. There is no contradiction between humane medicine and rational, pragmatic and effective medicine. It’s rather the latter that constitutes good medicine, as was made clear by Hippocrates and many others.

Several consumers of medical information will find that the use of information in this book will be defined by their respective individual needs.

The clinician-practitioner will use it to improve his prophylactic and therapeutic decisions.

The researcher and epidemiologist will use it to improve the objectives of research.

The applicants for research grants and those who assess research proposals will find, besides general rules of how to present a research proposal, a definition of the quality required for proposed studies and a description of the most basic and frequent errors.

Health service managers and politicians will use these facts for establishing the best health policies.

Lawyers and legal experts will look for reliable information about health problems that are the subject of litigation. The reliability of a proof of a cause and effect relation (a substance with a purported harmful effects on health, the beneficial and unwanted effects of a medical intervention) have to be demonstrated and inform judgements.

Students in health sciences and health professionals need a rapid and reliable guide to make the best use of their time. They are not always comfortable when faced with information, particularly if it is in numerical form.⁶

To all these potential readers we propose an overview of analysis and meta-analysis in medicine, without wishing this overview to include the more advanced methodology which needed for analysis and meta-analysis in medicine. We assume that the reader already has basic biostatistical and epidemiological knowledge, without which reading this book will certainly provoke a healthy unease in the mind of the reader and will stimulate him to seek out the epidemiological and statistical principles he needs. Those who want to fill gaps will find the bibliography at the end of each of the three chapters useful, which, in addition to the ad hoc references, include basic methodological references intended to provide further guidance for the novice.

We hope that the reader, after reading this book, will be better able to evaluate the uncertainties, ignorance and solid knowledge of which medical science consists. Hopefully he will finish reading a selected article with the pleasant certainty that the information he has just acquired is sound, or that he will complete his overview of scores of articles on the same topic with a better idea of how they should be evaluated.

To give a concrete example, the reader must be able to assess whether, to minimise the chances of getting a coronary, he really does have to start running and how fast.

1. de Solla Price D.: The development and structure of the biomedical literature (pp. 3-16) & Warren K.S.: Selective aspects of the medical literature (pp. 17-30) in: Coping with the Biomedical Literature. A Primer for the Scientist and the Clinician. Edited by K.S. Warren. New York: Præger Special Studies/Præger Scientific, 1981.

2. Sackett D.L., R.B. Haynes & P.X. Tugwell: Clinical Epidemiology. A Basic Science for Clinical Practice. Boston: Little, Brown, 1985.

3. Feinstein A.R.: Clinical Judgement. Huntington, N.Y.: Robert E. Krieger Publ. Comp., 1967.

4. Murphy E.A.: The Logic of Medicine. Baltimore: The Johns Hopkins Press, 1976.

5. Feldman K.A.: Analyse du livre “Issues in Data Synthesis” (W.H. Yeaton and P.M. Wortman eds. San Francisco: Jossey-Bass, 1984). J Am Stat Ass, 81:253-254, 1986.

6. Berwick D.M., H.V. Fineberg & M.C. Weinstein: When doctors meet numbers. Am J Med, 71:991-998, 1981.

Records