Chapter 4 Research in Pediatric Critical Care

“A fool is a man who never tried an experiment in his life.”

• Research is the right thing to do for critically ill children, their families, and our profession. Whether or not it is supported at your institution does not alter this fact, nor does it relieve you of your scholarly responsibilities as a physician.

• Make inquiry, comparison, and analysis a permanent part of your critical care practice. Keep a follow-up card as a lifelong habit in the practice of medicine. Go to postmortems, and keep looking through the microscope.

• Before you accept dogmatic teaching from anyone, make a habit of analyzing its basis. Keep a “reading” section on your follow-up card.

Among the factors that define a medical specialty is the recognition of a clearly defined body of knowledge that is intrinsic and unique to that specialty. This body of knowledge is determined by the disease processes that the specialists treat, comprehensive understanding of those disease processes, and, most importantly, the academic and intellectual constructs that allow the advancement of medical knowledge, not only narrowly in the specialty but also in general. Endeavors directed at increasing the specialty’s knowledge base are the research interests of that specialty. Thus a recognized medical specialty has a clearly defined patient population, clearly defined disease processes, and clearly defined research interests. Medical specialties frequently have associated societies, professional organizations, and national institutes, all of which facilitate research funding, sharing of research information, and advancing knowledge in the specialty.

Because organ system classification of specialties is fairly obvious, it is a common basis for specialization. The specialists who treat diseases of the lungs are familiar with the wide spectrum of pulmonary diseases, their pathogenesis, and their treatment. There exists a large body of research-derived knowledge and a great deal of ongoing related research activity. This activity is presented at national meetings, such as the annual American Thoracic Society meeting. There is a network of extensive funding sources available to this specialty, such as the American Lung Association; the American Heart Association; and the National Heart, Lung, and Blood Institute (NHLBI). Taken together, these form the clearly defined specialty of pulmonary medicine. Not all specialties are organ specific; for example, infectious disease and immunology are specialties that are concerned with diseases that affect the entire patient. These specialties have managed to clearly define a body of knowledge that is both intrinsic and unique, as well as crucial, to the specialty.

How does critical care medicine measure up to this standard? Have we succeeded in defining disease processes, patient populations, and research efforts that are unique to critical care medicine? What are these areas? Is it necessary for intensivists to participate in some area other than critical care medicine for their academic and research involvement? Commonly, physicians whose clinical practice is critical care are involved in research integrated into multiple other specialty areas. To define critical care medicine as a medical specialty able to stand on its own, not only must the clinical practice and the pathophysiology of the unique disease entities be defined, but also critical care researchers must make a unique contribution to understanding and treating the pathophysiologic conditions that affect critically ill children. Without a body of research that is unique to critical care medicine, the certainty of the specialty’s future will be in question.

Research Areas

There are clearly pathophysiologic processes that appear to be unique to critical care medicine. Probably the most clear-cut of these processes is acute respiratory distress syndrome (ARDS),^1,² although this is only one manifestation of a systemic process. The syndrome appears to develop in critically ill patients whose initial injury arises from a wide variety of organ-specific insults. Although the lungs appear to be the primary target organ, all organs are affected by the same complex underlying pathophysiologic process. This process causes widespread endothelial injury involving many organ systems.³ This results in tissue edema, decreased organ perfusion, ischemia, and multiple organ failure. This pansystemic disease is familiar to the intensivist and is one of the most extensive areas of unique research interest in critical care today. Understanding this systemic inflammatory response is central to understanding critical care.

Another major disease process that clearly interests critical care physicians is shock. Shock, whether it be hypovolemic, septic, or of some other etiology, by its very nature is a multisystem, non–organ-specific, acute, life-threatening process.^4,⁵ Another natural area for critical care research would be specifically aimed at preventing and/or ameliorating the multisystem insults that occur because of the body-wide activation of potentially lethal mediators triggered by episodes of shock. The spectrum of research opportunities in this area ranges from molecular biologic to large-scale clinical trials and has grown exponentially in the past few years.⁶

Organ system interaction also provides an area of primary interest in critical care. In particular, cardiorespiratory interaction has immediate, everyday application in ventilatory management, cardiopulmonary resuscitation, and cardiac support. The physiology of how changes in the pleural pressure affect cardiac function during either spontaneous or positive pressure ventilation is an area of constant relevance to the intensivist.⁷ In a broader sense, cardiorespiratory interaction extends beyond this arena. How changes in cardiac function alter ventilation, airway resistance, and lung compliance is an integral part of critical care. The pulmonary endothelial synthesis, release, and degradation of multiple mediators, ranging from myocardial depressant factors to systemic vasoactive substances, that alter cardiovascular function can be considered cardiorespiratory interaction and are of unique interest to the intensivist.⁷ This area also encompasses a broad spectrum of possible approaches from molecular biology to cell physiology to integrated physiology. An area of research that is particularly important to critical care medicine is cell biology. All organ system failure can be described in terms of cellular failure. For example, the general effects of superoxide radicals produced by leukocytes on other cell functions are legion and clearly of interest to intensivists.^4,⁸ Extending the argument of cellular specialization is possible for other cell types. An argument could be made to consider the endothelium the specialty organ of intensivists.⁹ All organs that fail in critically ill children contain large areas of biochemically active endothelium. This endothelium is important because it elaborates hormones and autacoids that have systemic and local effects. These effects include alteration of coagulation and blood viscosity, superoxide generation and tissue damage, smooth muscle regulation, metabolism of circulating vasoactive substances, and interaction with the immunologic system.^10,¹¹ All generalized stressors (e.g., sepsis, hypoxia, hypovolemia) alter endothelial cell function. One way of looking at multiple organ system failure is to view it as an “endotheliopathy.” This endotheliopathy gives rise to widespread organ-specific damage, such as renal failure, ARDS, myocardial depression, and alterations in the blood-brain barrier, with subsequent cerebral edema. Endothelial cell function changes over time as the child develops and may be specifically affected by disease processes particularly prevalent in children, such as the infectious vasculitides.¹² This sort of theoretical construct could serve as an organizational basis for research efforts in pediatric critical care and offer new insights into our understanding of critical illness in children.

Because of the astonishing success of the human genome project, the development of gene chips, and the increasingly realized potential of proteomics, understanding the genetic nature of critical illness no longer seems beyond our reach.¹³ Understanding the genotype of individuals who physiotypically display fatal responses to meningococcal disease while their classmates remain asymptomatic, albeit colonized by the same serotype organism, holds the promise of prospectively tailoring therapy to each patient individually. The many genomic and proteomic projects developing within pediatric critical care will help our understanding of the genetic bases of critical illness and holds out hope for understanding systemic inflammatory response syndrome (SIRS) and sepsis.¹³^–¹⁵

An area unique to pediatric critical care is that of caring for children and their families who must cope with acute critical and sometimes fatal illness. No other physicians deal with death and dying more frequently than intensivists. This role is especially important in the care of children. Supporting the family facing the acute, unexpected, critical illness and death of a child requires masterly physician interpersonal skills that may attenuate family problems long after the child’s death. The psychosocial impact of critical illness and palliative care has been explored very little, and we need to know more. Such issues are intrinsic to critical care medicine, and it is imperative that intensivists become responsible for the research in this area. Potential avenues for the intensivist-investigator include epidemiologic study, such as family and sibling bereavement patterns, and randomized therapeutic interventions, such as the effect of frequent postmortem follow-up on the high incidence of divorce among couples who have lost a child.

Finally, and perhaps most importantly, the burgeoning area of medical informatics holds the promise of enabling us to understand complex, critical, but rare disease processes previously impossible to study.¹⁶^–¹⁹ Understanding knowledge discovery in databases and building national and international collaborative partnerships to understand the scope of pediatric critical care have come within our reach.²⁰ Informatics research with national funding may provide real-time guidance for management of the rarest critical illnesses. In addition participation in national databases to support quality improvement and multisite research studies provide a new and fertile area for critical care research.^21,²²

These examples indicate part of the wide spectrum of research opportunities in pediatric critical care medicine. There are many more, including the now well-established multicenter National Collaborative Pediatric Critical Care Research Network, now in its second 5-year epoch of funding by the National Institute of Child Health and Human Development (NICHD), one of the National Institutes of Health. Networked research is necessary in critical care and the Pediatric Lung Injury and Sepsis Investigators (PALISI; http://pedsccm.org/PALISI_network.php) is an important example of a national research network focused on pediatric critical care. Constant sensitivity to identifying the questions (incorrectly answered, unanswered, and unasked) is the character trait required in the academic intensivist if our specialty is to continue to grow. Identifying areas unique to critical care medicine provides the knowledge base for the specialty necessary for its growth and for our patients’ well-being. Given the many unique areas of interest in critical care, how do we uncover them and encourage research in the subspecialty?

Wellsprings of Research

Collective Needs

Why must critical care collectively, as a specialty, support research in pediatric critical care? The arguments mentioned previously suggest that without the academic, intellectual, and scientific pursuit of areas that are specifically unique and relevant to critical care, we have no specialty. In this broad sense, research establishes a collegial respect for physicians who practice critical care and engenders academic support for the specialty. If all we do is provide clinical care for desperately ill or dying children, and clinical care that other physicians do not understand by virtue of their not being dedicated to it, the challenge of “what do intensivists do?” should only in part be answered by “spend long hours with children who are critically ill and their families.” This question addresses a deeper issue: Are intensivists contributing to the further understanding of critical illness? Are intensivists contributing to the intellectual advancement of medicine and the rich intellectual and academic milieu in the universities in which they find themselves? Are intensivists obtaining extramural funding for university-wide interactive and collaborative research efforts? Are intensivists training future generations of physician-scientists? Unless we can affirmatively answer these questions from a uniquely critical care point of view, it should not be surprising that physicians in other subspecialties do not understand or reliably value our labors. This justification—to ensure the viability and respectability of a specialty whose primary concern is treating the critically ill—is a major reason we must encourage and support research. Failure to do so is a failure to critically ill children.

A further reason for the commitment to research by the critical care medicine community is to allow young investigators ample introduction to research. They must be able to discover what research is, be provided with the tools to answer the outstanding questions in the field, and eventually make contributions both scientifically and educationally. These contributions will occur only if sufficient foresight is exercised to ensure that the facilities and resources are available. One of the chief rewards of developing this integrated structure will be the advancement of a specialty that truly is able to improve patient care. There are other rewards from this organized research endeavor and the education it provides. If the young investigator returns to clinical medicine, never to enter a basic science laboratory again or to organize even one clinical trial, this physician will at least be sensitive to the critical questions in patient care and be able to read and apply the literature with a better understanding. Many otherwise perfectly adequate young physicians are unable to critically evaluate the medical literature as a tool to improve management of their patients until they have been involved in contributing to it. Didactic methods are inadequate for teaching the rigorous effort required to write an article published in a first-class journal. It requires practical and challenging mentoring and arduous doing; however, once done, the emerging clinician is better able to understand and critically review the contributions of her/his colleagues and their relevance to critically ill children.

There is a further benefit from encouraging our fellows to research. The research effort provides excellent opportunities to observe how modulation of biochemistry, biology, and physiology alters the status of living beings. The practical knowledge gained in learning how to accurately measure the aortic/systemic and pulmonary artery pressures in animal models, determining the growth requirements of endothelial cells, maintaining sterile tissue cultures, and measuring pharmacokinetics provides insights into daily clinical practice unobtainable in any other way and practical for the care of every critically ill child. For this reason, it is nearly impossible to become a well-rounded clinician without having learned the basics involved in, and completed the exercise of, addressing a research question. The spinoffs of how to do cutdowns, insert catheters, start arterial lines, measure tidal volumes and pressures, manipulate ventilators, care for cultures, bioassay eicosanoids, perform chromatographic separation of bioactive lipids, and sequence the messenger RNA for endothelin immediately improve the bedside care of children, both intellectually and practically.

Individual Motivation

Beyond the learning relevant to clinical practice, individual motivation to answer questions and change the field is essential. Research is difficult, expensive, and time consuming. It removes the clinician from patient care, it is frequently thankless, often difficult to plan and organize, and hard to execute. Even when excellently done, it may be challenging to present and possibly not well received. Why then should any physician dedicated to the critical care of children be even slightly interested in becoming involved in research? Surely the desire to be promoted in the academic setting, see your name in prestigious journals, and impress your family, friends, and colleagues is insufficient motivation to contribute vast amounts of time, exhaust your intellectual and physical energy, and reduce your availability for patient care and family life. Although these may be some of the benefits of a research career, they are merely some of the lesser fruits of research. They are inadequate to provide the primary motivation for being involved in research. Personal motivations for research are many, and there are numerous rewards.

One of the most obvious is that being an attending physician in a pediatric intensive care unit 12 months of the year is not something that either can or should be done, no matter how much physical and emotional stamina one may have (or think he or she has). Diversion from clinical and administrative responsibilities and refreshment and renewal are obvious rewards of research. This benefit, prevention of burnout, is not achieved merely by avoiding clinical work. Rather, the invigoration that comes from involvement in, and a commitment to, improving patient care and advancing the specialty are the source of the benefits. Active involvement in research provides reciprocal inspiration from the daily questions in clinical work and the value of the research endeavor. It helps the clinician intensivist see the long hours of clinical care in a broader perspective. Of course, research can (and should) be fun. If it is not fun, if the investigator does not look forward to being involved in the understanding, development, design, execution, analysis, preparation, and presentation of the research, then it is not worthwhile for that individual investigator to remain involved. You must like what you are doing, or the dedication and commitment required for Edison’s “99% perspiration for each 1% inspiration” will be lacking.

Personal motivation is the necessary starting point for research invovlement. No amount of external pressure can produce a successful researcher. Where does this personal motivation spring from? The noble goal of adding to the knowledge base of the field is excellent; however, it is unlikely that many clinicians wake up in the morning and say, “Aha! I will add to the knowledge base of critical care today.” In the practice setting clinicians are continually challenged by questions that arise from providing critical care for children, by questions about patient management, and by uncertainties and confusion in providing patient care. It must be very rare indeed for a young clinician not to wonder about, be curious about, or be interested in answering these questions. It is the role of all preceptors in critical care to make certain that the trainee is aware of these questions and that they are asked. To teach critical care as if it were dogma is destructive to these goals. To constantly point out and demonstrate where there are failures and conflicts in our understanding and where matters of style, rather than matters of substance, determine our clinical practice is to uncover fertile areas for research. Unless the young intensivist senses these exciting challenges, the personal enthusiasm toward research will not be discovered.

Another motivating factor is the simple desire to know what research is. The great shibboleth of research has been held up before medical students and pediatric residents for years. Nevertheless, most young physicians have no idea what research is. They may have seen a fragment of a clinical trial, but they likely did not participate in a basic research experience. All too frequently they have little understanding of the real application of the scientific method or statistical analysis. Curiosity for what research is should be recognized, fanned, and fed. The inquisitive clinician must not be lost because of a poor understanding of research or a feeling that it is an elitist club. For this reason, our fellowship programs must provide valid scientific research experiences guided by seasoned investigators. Not until junior physicians realize that they can acquire the skills to answer the questions that arise clinically, in a rigorous and scientific fashion, can we expect them to do so.

The previously mentioned personally motivating factors, which include personal aggrandizement such as fame, fortune, job security, avoidance of burnout, fun, ability, education, and training, still are not, however, sufficient. All of these possible motivations do not provide the major essential driving force. Individual curiosity, a tireless need to question, and the restless search for answers must be the source of the entire endeavor. Curiosity? Is this the crucial concept? Sir Peter Medawar calls it a “nursery word”—a motive too inadequate.²³ Everyone possesses curiosity, and yet not everyone makes a commitment to seek solutions, occasionally at great personal cost. So it must be more than mere curiosity. Medawar calls this driving compulsion the “exploratory impulsion”; Kant called it “restless endeavor.”²³ It is not merely curiosity but a surrender to the urge, often sacrificially, to seek the answer that motivates the investigator. This urge must be strong because it will require a great deal of time and energy before the question that originally piqued the clinician’s curiosity can be addressed. This innate, compelling motivation of the individual is the main driving force of medical investigation.

Where the collective needs of the specialty and the individual’s needs come together is that both have a genuine, deep-rooted desire to understand better how to help critically ill patients. This symbiosis of specialty needs and individual motivation forms the essential chemistry of discovery. In a fascinating address to the American Society for Clinical Investigation, J.L. Goldstein ²⁴ presented the formula:

The individual, when clinically stimulated, can make a fundamental contribution only with appropriate training. The specialty can meet the needs mentioned previously by providing that training. The collective combination of financial and intellectual resources and the individual’s blood, sweat, and tears is critical. Resources will be provided only if the physicians involved in critical care are committed to providing, for individual intensivists who have the curiosity and desire, the means to find answers to their individual questions. No matter how well organized, the specialty organizations can encourage individuals to labor toward solutions only for the problems that interest them. Selection of these trainees is critical. Erasmus Darwin said in 1792, “A fool is a man who never tried an experiment in his life.”

Let us not train too many fools. Constantly striving to recruit the seeker, doubter, questioner—and when they are recruited, to support them—is the responsibility of all physicians involved in critical care. Without them, critical care research will be nonexistent. Significant advances in our specialty can grow only from accepting this responsibility. Without this commitment to encourage and train, critical care medicine will lose promising young clinician-scientists to other specialty areas, because it will be only there that they will be able to seek answers to their questions. This crucial issue and its centrality to the growth of our specialty cannot be overemphasized. The training also must be thorough. This takes time, but without the commitment to train young investigators to think like basic scientists, they will be unable to apply the tools of basic science and will end up paralyzed and lost to the specialty of pediatric critical care.²⁴

Doing Research

“Gentlemen, do not think! Try and be patient. Have you performed the experiment?”

John Hunter

Variations in study populations, techniques of data gathering, study designs, questions asked, answers required, types of analyses, and whether the question should be addressed clinically or by a basic science approach (and, if basic science, whether by molecular, cellular, physiologic, biochemical, or biophysical experiments) can be very confusing. Then there are statistics. To understand how to address a given question, familiarity with the basic process of research is necessary. For example, the simple question, “Should I give my septic, acidotic patients sodium bicarbonate?” could be addressed in many ways (and indeed has been!). The options range from experiments to discern the subcellular effect of changes in pH on mitochondrial function to prospective, randomized, double-blind, multicenter clinical trials to determine whether bicarbonate therapy improves survival in septic shock. All of these factors have a part in answering what may first appear to be a simple question. Many factors influence how the researcher goes about answering any question, not the least of which are the researcher’s background and training. The availability of resources in the researcher’s institution and previous research relevant to the question being asked are also important.

So, what is research? Research is scientific investigation. If the motivation to do research can be matched by the commitment of the specialty to support research, is that sufficient? How is the bedside problem answered? If a keen investigator with a good question has a willing pediatric intensive care unit director with money, or at least one who is willing to help find resources, what next? How is research done? What is the scientific process?

Over the past 200 years, the scientific method has been developed by learning how to test our guesses about the universe.^25,²⁶ The key factors involved are the following:

• Observation

• Intuition

• Formulation of hypotheses

• Experimentation

• Development of scientific laws, theories, or axioms

• Testing these new theories

Sir Francis Bacon provided one of the first common-sense answers to the question, “How is research done?” The answer was “by observation and experimentation.” Which observations and data should be collected may seem fairly obvious at first, but deciding what should be observed and recorded are the crucial research questions. Approaches that may be useful in determining whether bicarbonate helps a critically ill child include observing and recording every physiologic parameter and every biochemical response, as well as measuring every enzyme’s activity and looking at urinary metabolic products. It is evident that these are not necessarily the best, most direct ways to answer the underlying question. Merely compiling a mountain of data without scientific reasons behind each observation (fishing) is risky, time-consuming, inefficient, and often futile. One of the main tasks of the investigator is to decide which observations in the whole set of possible observations are crucial. Lack of critical thinking may result in missing important observations and fruitless experimentation. Similarly, determination of what type of experiment should be performed is crucial.

What is an experiment? The original meaning of the term experiment was “a test made to demonstrate a known truth.” It served as a means of proof for an already “known” truth. This sort of experiment was not designed to generate new knowledge. This Aristotelian concept of experimentation, involving classical deductive logic, has limited application in modern medicine, other than perhaps that of pedantry and teaching high school chemistry. The essential concept of experimentation that has a more contemporary meaning entails an uncertain or unknown outcome. To the present-day researcher, the purpose of performing an experiment is to discriminate between possibilities. How experiments are designed to discriminate between possibilities depends on the underlying assumptions. Understanding the logic that underlies how an experiment is performed is useful in avoiding multiple traps, not only in reviewing the data but also in applying experimental results to real patients. Bacon ²⁷ noted: “If a man begins with certainties, he shall end in doubts, but if he will be content to begin with doubts, he shall end in certainties.” Many experiments are still designed—contrived may be a better word—to demonstrate the validity of preconceived ideas (sometimes called an ‘hypothesis’). This reasoning from preconceived ideas or premises to the specific situation is known as the process of deductive logic. Deductive logic is reasoning from the general to the specific. A clinical example of such deductive logic is the following syllogism:

Major premise: Penicillin is an effective treatment for pneumococcal pneumonia.

Minor premise: My patient has pneumococcal pneumonia.

Inference: I will treat my patient with penicillin.

The experiment performed in this case, treating a patient with penicillin, will have a certain outcome only inasmuch as the deductive logic is correct and the underlying primary assumption (major premise) and diagnostic result (minor premise) are true. In a general way, all specific conclusions that rest on authoritative statements of truth are deductive in origin. By their very nature, although they guide our clinical activity, they do not expand our medical knowledge. The hallmark of deductive logic is complete reliance on the certainty of known or revealed facts. Some 300 years after the modern scientific revolution and the birth of true scientific method, modern experimental medicine remains beset with this type of logic. Aristotelian experimentation is the process of clinical practice. We reason from general principles and accepted facts to specific interventions and treatments. The entire evidence-based practice movement is based on deducing therapy from sound premises.

The difficulty comes when the dogmatic assumptions that underlie our clinical practice, and deductively lead to our therapies, are incorrect. Aristotelian experimentation provides no way to approach outcomes in medicine that are exceptional, yet these are the very occurrences that may be enlightening. Charles Darwin has exhorted us never to allow these exceptions to go unnoticed.²⁸ Deductive logic is unable to assist in the discovery of new knowledge and therefore general principles. This was clearly noted by Bacon,²⁹ who stated in 1620:

“The syllogism consists of propositions, propositions consist of words, words are symbols of notions. Therefore if the notions themselves (which is the root of the matter) are confused and over-hastily abstracted from the facts, there can be no firmness in the superstructure. Our only hope therefore lies in a true induction.”

The great revolution in scientific and philosophic writing in the 1600s was typified by Bacon’s absolute refutation of the concept that any new truths could be discovered merely by a deductive act of the mind ⁴:

“The discoveries which have hitherto been made in the sciences are such as lie close to vulgar notions, scarcely beneath the surface. In order to penetrate into the inner and further recesses of nature, it is necessary that both notions and axioms be derived from things by a more sure and guarded way; and that a method of intellectual operation be introduced altogether better and more certain.”

The Baconian revolution in scientific thought was dependent on observation and experimentation. The underlying premise was that the general could be determined, inferred, and understood from observing the specific. This led to the realization that by the use of thoughtful inductive logic, linked to observation and understanding, specific discovery of new generalized truths was possible.

A clinical example of this sort of contribution to modern medicine is the well-known example of vaccination. Jenner’s recurrent observation of the specific immunity to smallpox of patients who had been infected by cowpox led to experimentation with observation and a series of inductive steps that ultimately led not only to the eradication of smallpox in the world but also to generalization of the concept of vaccination to other infectious processes and vast discoveries in the area of immunology. It is impossible to conceive how Aristotelian deductive logic could have led to these discoveries. Bacon’s contribution was to realize that observations of the specifics in nature would lead, through application of the intellect, through inductive logic, to the discovery of new truths. Bacon did realize that we were unable to rely on “the casual felicity of particular events”²³ to provide us with all the specific information required to discover scientific truths, even if we spent an entire lifetime observing nature. He thus realized the necessity to devise experiences and contrive occurrences to collect factual information by which we would understand the natural world. This is Baconian experimentation. However, this still was not experimentation as practiced in medicine today.

Current medical experimentation is more accurately described as Galilean experimentation.^23,²⁸ The Galilean experiment discriminates between possibilities and, in so doing, confirms a preconceived notion by supplying facts that support the inductive process and lead to a sound conclusion. The essence of an experiment as proposed by Galileo was a true test, a trial, or an ordeal of an hypothesis. This constructive experimentation more accurately reflects what we think of today when we want to further our understanding of critical illness. As Stephen Hawking²⁶ put it in A Brief History of Time:

“Our present ideas about the motion of bodies date back to Galileo and Newton. Before them people believed Aristotle, who said that the natural state of a body was to be at rest and that it moved only if driven by a force or impulse. It followed that a heavy body should fall faster than a light one, because it would have a greater pull toward earth.”

The Aristotelian tradition also held that you could work out all the laws governing the universe by pure thought: it was not necessary to check by observation. As incredible as it may seem, no one, until Galileo, bothered to see whether bodies of different weight did, in fact, fall at different speeds. Mythology reports that Galileo demonstrated that Aristotle’s belief was false by dropping weights from the leaning tower of Pisa. The story almost certainly is untrue, but Galileo did do something equivalent: he rolled balls of different weights down a smooth slope. The situation is similar to that of heavy bodies falling vertically, but it is easier to observe because the speeds are smaller. Galileo’s measurements indicated that each body increased its speed at the same rate, no matter what its weight. A scientific revolution ensued.

In Galileo’s experiment, the hypothesis tested was that objects of different mass fall at different velocities. The “control” group could have been a group of spheres with mass = X. The experimental group (or groups) would have been X × 1 kg (or X × 1 kg, X × 2 kg, and so forth). Because the null hypothesis: that all objects (regardless of mass) fall at the same velocity could be falsified or disproven, the hypothesis could be tested. His data failed to support the hypothesis (supported the null hypothesis, failed to reject the null hypothesis), and the hypothesis was irrevocably disproved and destroyed, instantly. The old system was dead. The deductively logical syllogism of Aristotle’s day was as follows:

Major: The velocity of a falling object is determined by the object’s mass.

Minor: These two objects are of different mass.

Inference: They will fall at different rates.

This syllogism was disproved by a single observation. Therefore the conclusion was incorrect and therefore if the minor premise was true, the major premise had to be false. A new intellectual universe became possible. Every preconceived notion was testable by experiment. All that is required is a testable hypothesis. The routine function of medical experimentation, the purpose of all medical science and the major modus vivendi of all the national institutes, is the testing of hypotheses. Although gathering facts and cataloging their relationships as in Aristotelian and Baconian experimentation remains of some value, testing hypotheses is our strongest research tool. Galilean experimentation provided the capability of constantly revising our hypotheses and avoiding unnecessary persistence in theoretical structures based on hypothetical errors that lead to no more than a house of cards.The generation of scientific hypotheses that can be critically tested is the basis of scientific discovery. Discovery has its beginnings in imaginative preconception, which is the creative act of mind that gives rise to a hypothesis.²⁸ Asking the question or conceiving the question is only the beginning of the scientific process. Casting the question in the form of testable, verifiable, scientific hypotheses is where the creative work of research really begins. The brilliant guess, the eureka moment—these scientific insights are the sources of these hypotheses. The hypothesis is a mark to be attained, a suggestion of the probable, a provisional proposal of an underlying truth, or some specific facet of it. The hypothesis has but one purpose, to be tested. But this approach means that an hypothesis, no matter how interesting, can never be proven. Absolute proof of a hypothesis is not, by the very nature of inductive logic and Galilean experimentation, ever possible because all possibilities cannot possibly ever be tested. Again, according to Stephen Hawking²⁶:

“Any physical theory is always provisional, in the sense that it is only a hypothesis: you can never prove it. No matter how many times the results of experiments agree with some theory, you can never be sure that the next time the result will not contradict the theory. On the other hand, you can disprove a theory by finding even a single observation that disagrees with the predictions of the theory. As philosopher of science Karl Popper has emphasized, a good theory is characterized by the fact that it makes a number of predictions that could in principle be disproved or falsified by observation. Each time new experiments are observed to agree with the predictions the theory survives, and our confidence in it is increased; but if ever a new observation is found to disagree, we have to abandon or modify the theory.”

Whereas a theory is an organized system of knowledge used to analyze or explain nature or behavior, a hypothesis has no such value. Theories may be built up from facts learned by testing hypotheses and may even contain partially substantiated hypotheses that are useful in predicting events, but hypotheses are useful only insofar as their testing acts as a focus for the discovery of truths. Without a doubt, hypotheses are the most important instruments in research. Developing a hypothesis is the initial phase of research and scientific investigation. It generates the plan for the research. Nevertheless, it is “a means, not an end,” as Thomas Huxley cautioned.²⁸ The ultimate goal of research is not to hunt blindly for unrelated facts but to test related hypotheses.

If we accept the fact that the purpose of experimentation is to test hypotheses, then it is clear that a necessity for research is to formulate appropriate hypotheses. The hypothesis must be focused, with a limited number of possible outcomes and limited number of implications that lead logically to further investigational steps. This minimizes futile activity. A hypothesis that accommodates all possible phenomena or outcomes is totally uninformative. The more restrictive it is, the more focused it is, the more instructive it is. One final warning about hypotheses: although they are the driving force of research, they must be kept in their place. Accepting unproved hypotheses can clearly lead you down a rabbit hole, often a time-consuming, expensive, and disastrous one. Failure to give up unsubstantiated or disproven hypotheses can lead (and has led) to a futile cycle of experimentation. Although scientists require hypotheses, find them attractive, and may not be able to live without them, they must not fall in love with them.^23,²⁸ The basic fact of science, that hypotheses are never proven and that they are only as good as the results they generate, must never be forgotten. Likewise, the physician-scientist as observer of nature must be encouraged to use quantitative and qualitative descriptive tools to develop the platform for meaningful hypothesis generation and testing.

The Null Hypothesis

Galileo’s revolutionary experiment proved nothing! Rather, it disproved the accepted dogma by a single observation. When deductive logic is correctly performed and the major and minor premises are correct, the inference is absolutely, positively true. This is not true in the other direction. Reasoning from the inferences is unreliable. The arrival of two objects at the ground at different times does not assure us that the objects are of different mass or that objects of different mass fall at different rates. In the penicillin case, the fact that our patient improved with penicillin proves neither that he had pneumococcal infection nor that penicillin is effective against pneumococcus. He could have just had erysipelas. Then again, if the objects of different mass arrive simultaneously—that is, the inference is wrong—then something also is very wrong with one or both of the premises. If penicillin does not reliably, reproducibly treat pneumococcal pneumonia, then something is wrong with the diagnosis or with the efficacy of penicillin against pneumococcus. Yes, an astute clinician sees all sorts of problems in this statement, but the problems only emphasize the importance of rigid control of nuisance variables (discussed later). Nevertheless, the fact that inference is asymmetrical demonstrates that falsification—the disproving of a hypothesis—is logically a stronger, surer process than the so-called (and impossible) proving of a hypothesis. As Hawking explained, absolute proof is not possible. Instead, to support scientific hypothesis we generally attempt to disprove the opposite hypothesis, that is, we try to “refute the null hypothesis.” For example, Galileo said: “All objects, irrespective of mass, fall at the same velocity.” The null hypothesis would be that objects of different mass fall at different velocities or, as previously asserted, mass determines velocity. Galileo absolutely refuted this null hypothesis; thus his data were consistent with his own hypothesis. Even so, they did not prove it; he merely disproved the null hypothesis.

As a clinical example, if the hypothesis is “steroids improve morbidity in shock,” then the null hypothesis is that they do not. To refute this null hypothesis, the investigator has to demonstrate a difference between steroid-treated and nontreated patients in an adequately randomized and powered trial. If so, the null hypothesis is rejected and the hypothesis survives this test—this time. Statistics are applied to determine the certainty of the rejection of the null hypothesis and actually are performed to demonstrate that the null hypothesis has been rejected with a degree of certainty. For example, if P = .05, then it is 95% certain that the null hypothesis is incorrect and that the results are consistent with the scientific hypothesis.

It is this asymmetry of inference that allows us to disprove major premises by demonstrating the untenability of the inference. This is how we support scientific hypotheses. This refutation of the null hypothesis is generally taken to affirm that the very opposite is true. This is done because falsification of the inference and/or minor premises proving the falseness of the major premise is such a potent tool. Proving the major premise true is, in fact, impossible. Thus the basic tool used to demonstrate that a hypothesis is true is that of proving that the null hypothesis is false. The scientist’s experimental goal is to reject the null hypothesis rather than to prove the actual scientific hypothesis. Because refuting the null hypothesis is such a potent tool, good scientific hypotheses must be of such a nature that their null hypothesis (or, indeed, many of their null hypotheses, because several may stem from one hypothesis) can be tested. This test is virtually always a statistical one.

Medical Research

“It is incident to physicians, I am afraid, to mistake subsequence for consequence.”

Samuel Johnson

The first great divide in medical research is between clinical and laboratory research. Many consider such a division arbitrary, and the bench to bedside to bench translational models that have enabled modern physician-scientists to bring breakthrough understanding to the care of critical illness mandate that effective pediatric critical care researchers have “feet” in both domains. Still, in this classical distinction, clinical research is carried out in patients. It is an extension of previous experience in patients or of results obtained from laboratory research. Clinical research can occur in any medical arena. Laboratory research clearly does not involve patients; rather, it relies on the results in animals or tissue-derived “subjects.”

Clinical research can be either retrospective or prospective. Retrospectively, epidemiologic studies, demographic studies, and studies of disease processes and outcomes of management regimens can provide useful information in directing future therapy. Certainly the great wealth of data now available in patient records can continue to provide worthwhile insights to aid our patients. Unfortunately, retrospective trials cannot convincingly answer therapeutic questions; rather, their utility is in hypothesis generation. Their solution requires true Galilean experimentation. Baconian studies such as these prospective trials require as much planning as possible before the patient actually is observed for the results of a therapeutic intervention, but this has started to change. Learning from reliable observations is the basis of physical science, and applying these research principles to data obtained from human subjects is useful in reliably suggesting a general theory if large enough numbers of observations are collected and analyzed. In medical informatics, this is the basis for knowledge discovery in databases. Thus with a sufficient number of observations (controlled, defined, verified data) we may learn how to manage our patients by applying analytical techniques to retrospective events. The information revolution may be driving knowledge discovery once again toward some reliance on deductive logic.¹⁸

The principles that guide all medical research, including clinical trials, are in place to minimize the possibility of an incorrect conclusion. A major cause of error is bias, either by the observer or the subject. A further cause of incorrect conclusions results from inadequate study design that may prevent accurate statistical analysis of the information obtained.

The other overwhelming principle that guides clinical research is to preserve the rights, autonomy, and safety of the individual subject.³⁰^–³² This is of particular importance in clinical research involving children. The issues of risk, informed consent, and the potential to benefit the patient are particularly finely focused in pediatrics. The spectrum of opinion runs from believing that research in children is not allowable to believing that child subjects should be treated exactly the same as adult experimental subjects. Any researcher who proposes doing clinical research in critically ill children must be familiar with all aspects of these arguments and realize the sensitivity of the issues involved in this area.^33,³⁵