Chapter 2: The Role of Theory in Clinical Research

Clinical research is a systematic method for evaluating the effectiveness of treatment and for establishing a basis for inductive generalizations about intervention. The ultimate goal is to further intellectual progress by contributing to the scientific base of practice through the development of theory. Theories are created out of a need to organize and give meaning to a complex collection of individual facts and observations.

Methods are the means by which we conduct investigations in a reliable and valid way so that we can understand clinical phenomena. But it is theory that lets us speculate on the questions of why and how treatment works, accounting for what we observe. Theories provide the explanations for findings within the context of what is already known from the successes and failures of previous investigations. As we continue to examine observations, we try to create theoretical generalizations to form a basis for predicting future outcomes. Without such explanations we risk having to reinvent the wheel each time we are faced with a clinical problem.

A theory is a set of interrelated concepts, definitions, or propositions that specifies relationships among variables and represents a systematic view of specific phenomena.¹ Theories have always been a part of human cultures, although not all theories have been scientific. Philosophy and religion historically have played a significant part in the acceptance of theory. The medieval view that the world was flat was born out of the theory that angels held up the four corners of the earth. Naturally, the men of the day were justified in believing that if one sailed toward the horizon, eventually one would fall off the edge of the earth. Such theories went untested because of a lack of instrumentation and because it was not considered necessary to test that which was already known to be true.

In contrast, scientific theory deals with the empirical world of observation and experience, and requires constant verification. We use theory to generalize beyond a specific situation and to make predictions about what should happen in other similar situations. The validity of these predictions can be tested through research. The purpose of this chapter is to define the elements of theory and to describe mechanisms for developing and testing clinical theories.

Theories can serve several purposes in science and clinical practice, depending on how we choose to use them. Theories summarize existing knowledge, giving meaning to isolated empirical findings. They provide a framework for interpretation of observations. For example, theories of motor learning bring together the results of many separate studies that have examined schedules of practice, types of skills, psychomotor components of performance, and other elements of the learning process. Theories are also used to explain observable events by showing how variables are related. For instance, a theory of motor learning would explain the relationship between feedback and feedforward mechanisms in the learning, performance, and refinement of a motor skill.

Theories allow us to predict what should occur, given a set of specific circumstances. For example, one theory of motor learning states that greater changes take place during stages of initial learning than during later stages, as illustrated by a decelerating learning curve. On the basis of this theory, we could anticipate that a patient using an exercise device for the first time will experience a spurt of improvement in force output during early trials as a result of practice that will not necessarily be related to strength increases.²

Theories can also provide a basis for predicting phenomena that cannot be empirically verified. For instance, through deductions from mathematical theories, Newton was able to predict the motion of planets around the sun long before technology was available to confirm their orbits. The element of prediction also affords us a measure of control. This is illustrated by analysis of the germ theory of disease, which explains how organisms in the environment cause disease states. The theory allows us to predict how changes in the environment will affect the incidence of disease. This, in turn, suggests mechanisms to control disease, such as the use of drugs, vaccines, or attention to hygiene.

Theories also help to stimulate the development of new knowledge by providing motivation and guidance for asking significant clinical questions. On the basis of a theoretical premise, a clinician can use the process of deduction to formulate a hypothesis which can then be tested, providing evidence to support, reject, or modify the theory. For instance, based on the theory that reinforcement will facilitate learning, a clinician might deduce that verbal encouragement will decrease the time required for a patient to learn a program of home exercises. This hypothesis can be tested by comparing patients who do and do not receive reinforcement, and, if supported, the hypothesis will lend credence to the original theory. A wide variety of hypotheses can be deduced from this same theory. For instance, a clinician may hypothesize that reinforcement will improve learning for spinal cord injured patients working to master the use of a hand splint. The results of testing each hypothesis will provide additional affirmation of the theory or demonstrate specific situations where the theory is not substantiated.

Theory provides the basis for asking a question in applied research. Sometimes there will be sufficient background in the literature to build this framework; other times the researcher must build an argument based on what is known from basic science. In descriptive or exploratory research, the study's findings may contribute to the development of theory. The researcher uses a theoretical premise to project how the variables being studied should be related and what outcomes are expected. The theoretical framework is usually discussed within the introduction or discussion section of a paper. Without a theoretical framework a researcher will be unable to understand the implications of his findings, and observations will not have a context.

Concepts and Constructs

The role that theory plays in clinical practice and research is best described by examining the structure of a theory. Figure 2.1 shows the basic organization of scientific thought, building from observation of facts to laws of nature.

The essential building blocks of a theory are concepts. Concepts are abstractions that allow us to classify natural phenomena and empirical observations. From birth we begin to structure empirical impressions of the world around us in the form of concepts, such as "mother," "father," "play," or "food," each of which implies a complex set of recognitions and expectations. We develop these concepts within the context of experience and feelings, so that they meet with our perception of reality. We supply labels to sets of behaviors, objects, or processes that allow us to identify them and discuss them.

We use concepts in professional communication in the same way. Even something as basic as a "wheelchair" is a concept from which we distinguish chairs of different types, styles, and functions. Almost every term we incorporate into our understanding of human and environmental characteristics and behaviors is a conceptual entity. When concepts can be assigned values, they can be manipulated as variables, so that their relationships can be examined. In this context, variables become the concepts used for building theories and planning research. Variables must be operationally defined, that is, the methods for measuring or evaluating them must be clearly delineated.

Some concepts are observable and easily distinguishable from others. For instance, a wheelchair will not be confused with an office chair. But other concepts are less tangible, and can be defined only by inference. Concepts that represent nonobservable behaviors or events are called constructs. Constructs are invented names for abstract variables that cannot be seen directly, but are inferred by measuring relevant or correlated behaviors that are observable. The construct of intelligence, for example, is one that we cannot see, and yet we give it very clear meaning. We evaluate a person's intelligence by observing his behavior, the things he says, what he "knows." We can also measure a person's intelligence using standardized tests and use a number to signify intelligence. An IQ score of 125 tells us something about that individual, but the number by itself has no empirical value. We cannot observe 125 intelligence "points" like we can 125 degrees of shoulder joint motion. Constructs are often manipulated as variables in psychosocial and behavioral research.

Propositions

Once the concepts that relate to a theory are delineated, they are formed into a generalization, or proposition. Propositions state the relationship between variables, which can be described in several ways. For example, a hierarchical proposition shows a vertical relationship, establishing ordered levels of concepts. Maslow's theory of the relationship of human needs to motivation demonstrates this principle.³ He described five levels, beginning at the bottom with basic physiological needs, moving up to safety, social needs, esteem, and finally ending at the top with self-actualization, or the fulfillment of one's self (see Figure 2.2).

A quantitative proposition is based on the frequency or duration of a specific behavior. For example, theories of fatigue are based partly on the concept of repetitions of exercise and how that relates to muscular endurance.⁴ A temporal proposition orders concepts in time and states a sequence of events. For instance, the transtheoretical model explains behavior change as a process along a continuum of motivational readiness, with five stages: precontemplation, contemplation, preparation, action, and maintenance.⁵ Rather than seeing change as a unidimensional act, such as simply quitting smoking or starting to exercise, this model suggests that individuals progress through a series of stages in recognizing the need to change and finally engaging in a new behavior (see Figure 2.3). The model also clarifies the importance of identifying the stage an individual is in before a successful intervention can be implemented.

Models

Many of the concepts we deal with in professional practice are so infinitely complex that we cannot truly comprehend their real nature. In an effort to understand them we try to simplify them within the context of a model that serves as an analogy for the real phenomenon. To understand the concept of an "atom," for example, it was helpful for scientists to delineate a conceptual model that is likened to a solar system. The intricacies of genetic processes were clarified by the development of a helical model of DNA. Function of the neuromuscular system is often taught using a model of the muscle spindle. These models are considered simplified approximations of reality. The model leaves out much of the detail, but describes the conceptual structure closely enough to give us a better understanding of the phenomenon. Models are symbolic representations of the elements within a system. Where a theory is an explanation of phenomena, a model is a structural representation of the concepts that comprise the theory.

Some physical models are used to demonstrate how the real behavior might occur. For example, engineers study models of bridges to examine the stresses on cables and the effects of different loading conditions. The benefit of such models is that they obey the same laws as the original, but can be controlled and manipulated to examine the effects of various conditions in ways that would not otherwise be possible. Rehabilitation engineers develop prototypes of prostheses or motor-driven wheelchairs, to evaluate their performance and to perfect their design. Scientists also use animal models to mimic specific anatomical or physiological deficits in the human to examine the effects of pathology, trauma and intervention.

Sometimes a model is a schematic representation, such as an architect's plans or a map. Therapists might use this type of model when evaluating a client's home for architectural barriers, by drawing a diagram of rooms and doorways and plotting out the spatial requirements for use of a wheelchair. Such a model provides opportunities for considering the implications of different approaches without physically carrying them out, and facilitates making appropriate changes when necessary. Computer simulations are the most recent contributions to the development of physical models. Scientists can experiment with an infinite number of variations in design and can analyze the implications of each without risk or major expense.

A model can also represent a process rather than a real object. For example, decision-making models can be used to suggest the most effective progression of intervention with specific disorders.⁶ The transtheoretical model provides a framework for understanding the process of behavior change (see Figure 2.3).⁵ The International Classification of Functioning and Disability (ICF) model (shown in Figure 1.1 in Chapter 1) creates a structure to understand the theoretical relationship between impairments and functional activities, and can be used to offer explanations as to why specific impairments might lead to certain types of disability.⁷

Quantitative models are used to describe the relationship among variables by using symbols to represent them. Models in physical science allow for accurate prediction of quantities, such as the summary of the relationship between force, mass, and acceleration (F = m × a). In the behavioral sciences, however, quantitative models are less precise, usually containing some degree of error resulting from the variability of human behavior and physical characteristics. For instance, a clinician might want to determine the level of strength a patient could be expected to achieve following a period of training. A model that demonstrates the influence of a person's height, weight, and age on muscle strength would be useful in making this determination.^8,9,10 This type of quantitative model can serve as a guide for setting long-term goals and for predicting functional outcomes. Research studies provide the basis for testing these models and estimating their degree of accuracy for making such predictions.

As the previous examples illustrate, theories are not discovered, they are created. A set of observable facts may exist, but they do not become a theory unless someone has the insight to understand the relevance of the observed information and pulls the facts together to make sense of them. Certainly, many people observed apples falling from trees before Newton was stimulated to consider the force of gravity. Theories can be developed using inductive or deductive processes.

Inductive Theories

Inductive theories are data based and evolve through a process of inductive reasoning, beginning with empirically verifiable observations. Through multiple investigations and observations, researchers determine those variables that are related to a specific phenomenon and those that are not. The patterns that emerge from these studies are developed into a systematic conceptual framework, which forms the basis for generalizations. This process involves a degree of abstraction and imagination, as ideas are manipulated and concepts reorganized, until some structural pattern is evident in their relationship.

For instance, this process was used by Skinner in the formulation of his theories of learning and behavior, based on previous work and his own observations of human behavior.¹¹ Through the examination and clarification of the interrelationships between stimuli and responses, he formulated a systematic explanation for the observed behaviors. Glaser and Strauss¹² used the term "grounded theory" to describe the development of theory by reflecting on individual experiences in qualitative research. As an example, Resnik and Jensen¹³ used this method to describe characteristics of therapists who were classified as "expert" or "average" based on the outcomes of their patients. Building on their observations, they theorized that the meaning of "expert" was not based on years of experience, but on academic and work experience, utilization of colleagues, use of reflection, a patient-centered approach to care, and collaborative clinical reasoning.

Deductive Theories

The alternative approach to theory building is the intuitive approach, whereby a theory is developed on the basis of great insight and intuitive understanding of an event and the variables most likely to impact on that event. This type of theory, called a hypothetical-deductive theory, is developed with few or no prior observations, and often requires the generation of new concepts to provide adequate explanation. Freud's theory of personality fits this definition.¹⁴ It required that he create concepts such as "id," "ego" and "superego" to explain psychological interactions and motivations. Because they are not developed from existing facts, hypothetical-deductive theories must be continually tested in the "real world" to develop a database that will support them. Einstein's theory of relativity is an excellent example of this type of theory; it was first advanced in 1905 and is still being tested and refined through research today.

Most theories are formulated using a combination of both inductive and hypothetical-deductive processes. Observations initiate the theoretical premise, and then hypotheses derived from the theory are tested. As researchers go back and forth in the process of building and testing the theory, concepts are redefined and restructured. This process occurs along a circular continuum between fact and theory, whereby a theory can be built on facts, but must also be tested by them (see Figure 2.1).

As we explore the many uses of theories in clinical research, we should also consider criteria that can be used to evaluate the utility of a theory. First and foremost, a theory should provide a thorough and rational explanation of observed facts. It should provide a basis for classifying relevant variables and predicting their relationships. A theory should also provide a means for its own verification; that is, it should be sufficiently developed and clear enough to permit deductions that form testable hypotheses.

A Good Theory Is Economical. It should be the most efficient explanation of the phenomenon, using only those concepts that are truly relevant and necessary to the explanation offered by the theory. Complex theories are difficult to interpret and less likely to provide meaningful direction to practice or research. Theories are also most useful when they apply to a broad range of situations, not one specific segment of a discipline.

A Theory Should Be Important. It should reflect that which is judged significant by those who will use it. In this sense, theories become the mirror of a profession's values and identity. When we examine the theories that are adopted by clinicians in the course of their practice, their intellectual investments become clear. For example, many therapists rely on neurophysiological theory as a basis for choosing therapeutic exercise techniques that use diagonal and rotational patterns of motion, as opposed to the traditional use of anatomical theory as a basis for exercises in straight planes. This suggests that research is needed to test hypotheses that predict the superiority of multiple-plane movement over single-plane exercise for given purposes.

Acceptance of Theory Can Change. Theories must be consistent with observed facts and the already established body of knowledge. Therefore, our acceptance of a particular theory will reflect the present state of knowledge and must adapt to changes in that knowledge as technology and scientific evidence improve. Therefore, a theory is only a tentative explanation of phenomena. It should be reasonable according to what has been observed, but may not be the only explanation. Many theories that are accepted today will be discarded tomorrow (see Box 2.1). Some will be "disproved" by new evidence, and others may be superseded by new theories that integrate the older ones. For example, Gardner's theory of multiple intelligences challenged long-held assumptions about general intelligence and the ability to measure it with a single score, such as an IQ test.¹⁵ He proposes eight distinct intelligences and suggests that different cultures will perceive these differently.

Theory recognition also evolves with social change. For instance, the disengagement theory of aging was originally proposed to account for observations of age-related decreases in social interaction.¹⁶ The explanation this theory offered was that older individuals withdrew from social involvements in anticipation of death. As sociological theory progressed, however, new perspectives emerged, such as exchange theory, which suggested that interactions in old age become limited because the old have fewer resources to offer, therefore bringing less to a relationship.¹⁷ In a further generation of exchange theory, socioemotional selectivity theory tried to explain reduced social exchange of older persons as a function of increasing selectivity in interactions.¹⁸ This theory suggests that older persons decide to reduce emotional closeness with some people while they increase closeness with others; that is, interactions reflect the rewards of specific emotional support with a selective group of individuals. The most recent progression of this theory is gerotranscendence, which looks at human development as a process that extends into old age. The theory proposes that aging, from childhood through old age, is a process that can be obstructed or accelerated by life crises, culture and support systems. Old age is yet another phase in development. When optimized, the process ends in a new and qualitatively different perspective on life.¹⁹

This evolution of theory illustrates how the explanations of an observed psychosocial phenomenon have continued to change as our understanding and perceptions of social interaction have grown. It also demonstrates how caregivers and health professionals may change their perspective on how to support and interact with aging individuals, depending on how they view the psychological focus of the aging process.

Every theory serves, in part, as a research directive. The empirical outcomes of research can be organized and ordered to build theories using inductive reasoning. Conversely, theories must be tested by subjecting deductive hypotheses to scientific scrutiny. The processes of theory development and theory testing are represented in the model shown in Figure 2.1. It integrates the concepts of inductive and deductive reasoning as they relate to the elements of theory design.

Theory Testing

When we speak of testing a theory, we should realize that a theory itself is not testable. The validity of a theory is derived through the empirical testing of hypotheses that are deduced from it and from observation of the phenomenon the theory describes. The hypotheses predict the relationships of variables included in the theory. The results of research will demonstrate certain facts, which will either support or not support the hypothesis. If the hypothesis is supported, then the theory from which it was deduced is also supported.

When we compare the outcomes of individual research studies with predicted outcomes, we are always aware of the potential for disconfirmation of the underlying theory. In essence, the more that research does not disconfirm a theory, the more the theory is supported. This may sound backwards, but in actuality we can never "prove" or "confirm" a theory. We can only demonstrate that a theoretical premise does not hold true in a specific situation. When a research hypothesis is tested and it is not rejected, that is, the study turns out the way we expected, we cannot state that the underlying theory is definitely true. To make such a statement, we would have to verify every possible application of the theory and demonstrate that the outcomes were absolutely consistent. As this is not feasible, we can only interpret individual hypotheses and conclude that a theory has not been disproved.

Utilization of Theory in Research and Practice

Clinicians are actually engaged in theory testing on a regular basis in practice. Theories guide us in making clinical decisions. Specific therapeutic modalities are chosen for treatment because of expected outcomes that are based on theoretical assumptions. Treatments are modified according to the presence of risk factors, based on theoretical relationships. Therefore, the theory is tested each time the clinician evaluates treatment outcomes. When a theory is used as the basis for a treatment, the clinician is, in effect, hypothesizing that the treatment will be successful. If results are as expected, the theory has been supported. When evidence is obtained that does not support a theory, or that cannot be explained by the theory, alternative explanations must be considered. There may be reason to question how measurements were taken and how concepts were defined, to determine if these were truly consistent with the theory's intent. The validity of the theory may be questioned, or the application of the theory to the specific problem being studied may need to be re-evaluated. It may also be necessary to re-examine the theory and modify it, so that it does explain the observed outcome. If this is not practical, a new theory may need to be considered that will encompass this and all previous observations.

As an example of the application of theory to clinical decision making, Mueller and Maluf²⁰ have described physical stress theory, which states that changes in levels of physical stress cause a predictable adaptive response in all biological tissue. According to this premise, stresses less than normal will result in decreased tolerance of tissues to subsequent stresses, and stresses greater than normal will result in increased tolerance. If we accept this premise, we can assume that when muscle is not sufficiently stressed (not exercised), we would predict decreased tension and power (weakness), which would limit the muscle's future tolerance to outside forces. When muscle is challenged at high stresses (as through exercise), we will see increases in contractile strength. Stresses at either extreme will cause the tissue to fail. When stresses are absent, the muscle will atrophy; when stresses are excessive, the muscle will be strained.

We can use this theory to help with decision making when an individual's muscle performance does not fall within normal limits. As shown in Figure 2.4, for a condition of weakness, due to prolonged low stress levels, the threshold for adaptation will decrease. Therefore, a patient who has a weakened muscle is likely to suffer an injury at a lower force threshold than someone who is stronger. Similarly, a weakened muscle will increase in strength with a lower level of exercise than a stronger muscle would require. This theory goes beyond this specific example, to demonstrate how intrinsic and extrinsic factors can modify the adaptive responses of various tissues. The authors clearly illustrate how continued testing of the theory and its relationships is needed to contribute to the foundations of practice.²⁰

Theory as a Foundation for Understanding Research

Understanding clinical phenomena cannot be achieved in a single study. It is a process within a community of researchers involving discussion, criticism and intellectual exchange to analyze the connection between new and previous findings and explanations. This type of exchange allows the inconsistencies to surface, to identify findings that cannot be explained by current theories. This process can serve as a catalyst for a paradigm shift, or a change in the basic framework that governs the way knowledge is pursued.²¹ For instance, the focus on outcomes in research represents a paradigm shift in rehabilitation science, as constructs such as disability and quality of life become a major focus, rather than the traditional emphasis on changes in physical impairments.²² The ICF is another example, allowing interactions with theories of self-efficacy and health behaviors, and incorporating environmental factors as influences on performance.²³ By looking at outcomes in this way, researchers recognize that questions and their underlying theories will take a different direction than in the past.

The importance of theory for understanding research findings is often misunderstood. Whenever a research question is formulated, there is an implicit theory base that suggests how the variables of interest are related. Unfortunately, many authors do not make this foundation explicit. Empirical results are often described with only a general explanation, or an admission that the author can find no explanation. It is the author's responsibility, however, to consider what is known, to examine potential relationships, and help the reader understand the context within which the results can be understood. It is incumbent upon all researchers to project their expectations into the realm of theory, to offer an interpretation of findings, and thereby contribute to the growth of knowledge.²⁴

When a theory does reach the level of absolute consistency in outcomes, it is called a law. Laws generally have a wider scope than theories, and allow precise predictions. For example, Newton made many mathematical observations that were used to describe the motion of the planets around the sun. These motions can be described with great precision, allowing great accuracy of prediction. What started as a theoretical observation eventually came to be accepted as a universal law. Generally, laws are not established in the applied sciences as they are in the physical sciences. The nature of human beings and their interactions with the environment does not allow our theories to become so precise in their prediction. We are left, therefore, with the necessity of continuing the quest for affirmation of our theories.