Methodological Standards and Strategies for Establishing the Evidence Base of Animal-Assisted Therapies

Psychotherapy is defined broadly to encompass interventions that use psychosocial techniques (e.g. interpersonal interaction, learning experiences, role playing, practice, coping skills) to reduce distress, maladaptive behavior, and psychological and psychiatric problems and to enhance adaptive functioning and positive experiences in everyday life. The therapist provides conditions (e.g. support, encouragement, acceptance) through which these techniques are applied. Key concepts that are addressed or emerge in therapy include the patient/therapist relationship and bond, support, attachment, and friendship (e.g. Norcross, 2002; Schofield, 1986; Wallin, 2007). Also, the benefits of therapy are considered to include developing awareness, empathy, and increased interpersonal sensitivity, making individuals feel better, reducing stress, and improving the quality of life. Each of the concepts and benefits central to psychotherapy figures prominently in writings of human/animal relationships (e.g. Anderson, 2008; Olmert, 2009; Salotto, 2001). Thus, the systematic use of animals in the context of therapy is reasonable, intuitive, and consistent with core concepts of traditional psychotherapy.

Outside of the context of therapy, the benefits of close contact with animals are widely recognized. This can be attested to in part by pet ownership, which encompasses 63% of all households in the USA, based on 2005–06 data ( The number of pets (≈350 million) in the USA actually exceeds the number of people (≈303 million). The benefits of animal contact are evident from personal experience with pets, observation of the experience of others, and reliance on pets among many cultures currently and throughout history, as well as from scientific research on health and well-being. The challenge is to harness these benefits so they can be systematically applied in the context of therapy.

The appeal, widespread belief, and everyday experience of the benefits of human/animal contact are at once a strength and liability for developing the science base of animal-assisted therapy (AAT). The strength draws on the keen interest in these benefits and extending these benefits to many whose lives might be improved with animal contact. The liability stems from the almost universal acceptance of the benefits of animal/human contact. One might ask: Do we need research when all signs point to the huge impact of animals on human experience? Is it not obvious that animals and people help each other? That animals are subjectively valued and improve the quality of life are easily evaluated by just asking people. It is quite another matter to raise the empirical question of whether AATs can ameliorate social, emotional, behavioral adjustment problems and diagnosable psychiatric disorders (e.g. autistic spectrum disorders, anxiety, depression, conduct disorders) and mitigate the impairment with which these are associated. Claims have been made that AATs can effect change in these latter domains. Given the public health implications of advocating and delivering interventions, scientific scrutiny is essential.

The goal of this chapter is to foster further scientific evaluation of AATs in contexts in which the goal is to improve the social, emotional, behavioral adjustment or adaptive functioning or to ameliorate some psychological or psychiatric condition among children and adolescents. This chapter highlights the current status of psychotherapy (non-animal assisted) for children as a backdrop to convey current methodological standards for intervention research. The discussion moves from conceptual underpinnings of an investigation, to the specific question that is asked about treatment, and to control and comparison conditions pivotal to AAT research. Common methodological practices in AAT research that interfere with drawing conclusions about the effects of treatment are highlighted. The chapter concludes by proposing next steps for research; these are intended to advance both the conceptual understanding of AATs and the conclusions about their impact on child functioning.

25.2 Context: current advances in psychotherapy outcome research

AATs are often designed to improve adjustment and functioning of individuals and to decrease various sources of social, emotional, cognitive, and behavioral problems. Psychotherapy shares these goals but of course without the use of animals. The status and accomplishments of psychotherapy research are relevant because they convey a body of literature to which AAT research will be compared and integrated. Although most psychotherapy research focuses on adults, let me highlight progress by referring to psychotherapies for children and adolescents.1

There has been enormous progress in child therapy research in the past few decades (see Kazdin, 2000b; Weisz and Kazdin, 2010). First, the quantity of controlled therapy outcome studies is remarkable. The last formal estimate of such studies in 1999 placed the number at 1,500 (Kazdin, 2000a). The number continues to increase in light of a constant stream of journals in which therapy studies are routinely published. It is difficult to evaluate that number without some comparison. It is useful to mention in passing that a recent review of AATs identified six studies of children and adolescents in which there were control or comparison groups (Nimer and Lundahl, 2007). There are hundreds of child and adolescent clinical problems and sources of impairment (e.g. anxiety, depression, oppositional defiant disorder, autistic spectrum disorders) (e.g. American Psychiatric Association, 1994). A large number of studies are needed to test variations of treatment across a range of dysfunctions.

Second, the quality of child psychotherapy research has improved over the years. Several methodological practices currently guide psychotherapy research and they are listed in Table 25.1. Randomized controlled clinical trials (RCTs) are the rule rather than the exception and many other practices listed in the table raise the methodological bar even higher. Not all studies include all of these practices, but it is difficult to obtain funding for treatment research or to publish the results of a psychotherapy outcome study in the scientific literature without including most of these features.

Table 25.1 Methodological practices that guide psychotherapy outcome research

  1. Random assignment of participants to conditions

  2. Careful specification of the client sample and the inclusion and exclusion criteria required for participation

  3. Use of strong control or comparison groups (e.g. treatment as usual or another viable treatment rather than or in addition to no treatment or wait-list control groups)

  4. Use of treatment manuals to codify procedures and practices so as to permit training of therapists and replication of treatment by other investigators

  5. Assessment of treatment integrity, i.e. the extent to which the intervention was carried out as intended

  6. Use of multiple outcome measures with multiple assessment methods (e.g. self-report, parent report, direct observation) and measures of multiple domains of functioning (e.g. symptoms, prosocial functioning)

  7. Evaluation of the clinical significance of change, i.e. whether the changes at the end of treatment make a difference in returning individuals to adaptive functioning

  8. Evaluation of follow-up weeks, months, or years after post-treatment assessment of child functioning.

Third, quantitative (meta-analytic) reviews of the research consistently conclude that many forms of psychotherapy for children are effective (Weisz, 2004). These reviews place diverse outcome measures from the individual investigations on a common metric (effect size) so that studies can be combined and conclusions can be reached about different treatments, clinical problems, patient samples, and other characteristics spanning the studies. Improvements among children in treatment groups surpass the changes made by children in control-group conditions. Moreover, the magnitude of the effects of treatment is reasonably large (effect size ≈.70).2

Fourth and related, there are now several evidence-based psychotherapies. These refer to psychotherapies that have controlled studies to support them, where the effects of treatment have been replicated, and where several of the methodological practices noted in Table 25.1 have been included (Christophersen and Mortweet, 2001; Nathan and Gorman, 2007; Weisz and Kazdin, 2010). As an illustration, consider the treatment of aggressive and antisocial behavior among children (current psychiatric diagnosis is conduct disorder). Aggressive and antisocial behavior among children is the most commonly referred clinical problem to child inpatient and outpatient services (33–50% of all referrals in the USA). There are now at least seven evidence-based treatments. The treatments encompass young children through adolescents and mildly oppositional and aggressive behavior to severe and repetitive antisocial behavior that has led to adjudication.3 This is a remarkable accomplishment in relation to a serious clinical problem. This is one example. Evidence-based psychotherapies are available for many other clinical problems, as reviewed in the previous citations.

My comments highlight advances in psychotherapy research. In passing, it is important to acknowledge the current status of child therapy as carried in clinical practice. There has been a huge lag in disseminating treatments that have a strong empirical base to clinical practice or to training programs in the mental health professions (e.g. clinical psychology, psychiatry, social work, and nursing). In clinical practice with children, more than 550 techniques are in use (Kazdin, 2000b). Most treatments in use have never been investigated empirically; many treatments known to be effective are not in widespread use; some treatments in widespread use are known not to be effective; and well-intentioned efforts to help children (e.g. horticulture therapy, smudge art therapy, wilderness camps) continue to emerge with no empirical evidence to support them.

For the purposes of this chapter, it is relevant to refer to a large and actively used set of treatments in clinical work that have little evidence to support them. Historically, unevaluated psychotherapies have had little scrutiny or accountability. Currently, the absence of evidence for a given intervention provides a sharp contrast to many other interventions now available. Scrutiny by state and federal agencies and third-party payers increasingly will influence what interventions are administered and reimbursed. Consequently, unevaluated treatments or poorly researched treatments will have renewed pressure to justify their use.

The progress I have highlighted conveys there is now a large outcome literature on psychological treatment for children. While the substantive gains have advanced, the methodological standards have evolved as well. Those standards reflect a broader movement in intervention research and health care. The standards for clinical trials across many disciplines have become higher and more explicit. As a prominent illustration, the Consolidated Standards of Reporting Trials (or CONSORT) have emerged to guide how clinical trials are conducted and reported. A checklist for investigators is available to address critical facets of the trial, such as how the sample was identified, how they were allocated to conditions, how many started in the trial and completed treatment, and whether they received the intended treatment (please see The standards have been adopted by hundreds of professional journals from many disciplines and countries (see—journals/) in an effort not only to improve reporting of trials but also to increase their quality.

The implications of progress in psychotherapy research and explicit standards (e.g. CONSORT) for conducting intervention trials are clear. Methodological standards are in place and evolving; these standards convey what is required of any intervention for it to be added to the body of knowledge. Any newly proposed treatment or treatment that has been available for some time but has not undergone careful empirical evaluation has a methodological template to follow in order to gain the attention of the scientific community.

25.3 Conceptualizing the study and its focus

The preceding comments provide a backdrop for AAT by conveying advances in developing a list of evidence-based psychotherapies and in setting the methodological standards for adding less well-studied or evaluated interventions to that list. The context is useful but, by itself, is not very helpful in beginning or designing a study to evaluate an AAT. Development of a specific study begins with conceptualization of the study, treatment, and therapeutic change. Conceptualization of a study is discussed here at three levels proceeding from the more general to the specific issues that guide the design of a study and how the treatment is evaluated.

25.3.1 Small theory: the investigator’s view of animal-assisted treatments

The design of a study begins with making explicit the theory that one has about the treatment. This has been referred to as a “small theory” or “treatment theory” to convey that it is confined to the particular study and need not be something grandiose to explain all therapy or all clinical problems (Lipsey, 1996). This theory specifies the clinical problem, what the treatment is, how or why that treatment can be expected to have impact on the problem, what the critical components of treatment are, and what outcomes best reflect therapeutic change. This small theory makes explicit the rationale for the treatment, its application, and evaluation. As I discuss later, common methodological problems in AAT research often emerge from not clarifying the rationale for the intervention or how the intervention might be expected to help participants.

The small theory proposes how the therapy works, i.e. through what processes. For example, what facet of the animal’s presence in the session is responsible for or contributes to therapeutic change? Is it the presence of an animal, the interactions of the child with the animal, or the bond or relationship of the child to the animal that leads to or is likely to lead to change? Of course, it could be all of these and many more alternatives than those I am using for illustrative purposes. The processes are important to specify. Once specified, the investigator knows what to measure during treatment to ensure that in fact the process was altered. This might include evaluation of the extent of direct interaction or the bond the child reports.

Apart from assessment, specifying the process through which change occurs can guide the activities of the therapy sessions. Thus, if the child/animal relationship is important, the treatment procedures ought to maximize that. If direct contact and interaction are important, the sessions should be structured to ensure these occur at a high rate. One would expect the critical process(s) to relate to therapeutic change but the first task is to ensure that the process itself was invoked or occurred as intended.

Our small theory also directs attention to the outcome of treatment, i.e. assessments administered at the end of treatment (and perhaps follow-up) to evaluate efficacy. We might propose that the process of therapy reduces arousal in the session, perhaps because of the calming effect of direct physical contact with an animal. For outcome assessment, perhaps a measure of arousal also would be good to include in light of our small theory about what therapy is likely to accomplish. Arousal may be the primary measure and, in light of our view, and more likely than some other measure of anxiety (e.g. teacher ratings) to reflect change. Multiple measures might be included to evaluate treatment but the small theory conveys clearly what is likely or expected to happen and, at the very least, guides us in selecting some of the outcome measures to ensure our focus connects the problem, processes in treatment, and therapeutic change.

In short, a small theory guides the investigator in what to emphasize during the sessions and what to measure both during treatment (process measures) and after treatment (outcome measures). The absence of a small theory can foster a weak test of treatment and weak outcome effects. The small theory of why treatment works and what to emphasize might well be wrong, but remains a good place to begin in deciding what treatment to provide and how to deliver the treatment in ways likely to maximize its impact.

25.3.2 Questions to guide animal-assisted treatments

Another way to guide the design is to make explicit the specific question that may serve as the impetus for the study. Studies of psychotherapy (or other psychological, educational, and health care interventions) usually focus on one or more of several questions. These questions codify the substantive focus but in the process also influence many methodological decisions such as what the control or comparison groups will be. Table 25.2 lists several treatment strategies that encompass specific questions and control or comparison conditions required for their evaluation. These are frequently asked questions that guide intervention research and apply to AATs.

Table 25.2 Questions that guide psychotherapy research

Treatment strategy Question asked of treatment Control and comparison conditions required
Treatment package What is the impact of treatment relative to no treatment? Treatment vs. no treatment or wait-list control
Dismantling strategy What components contribute to change? Two or more treatment groups that vary in the components (or ingredients) of treatment they receive—some components are “subtracted” from the main treatment to see if they are needed
Constructive strategy What treatments can be added (combined treatments) to optimize change? Two or more treatment groups that vary in components. New components are added to the basic treatment to see if they enhance outcome
Parametric strategy What parameters can be varied to influence (improve) outcome? Two or more treatment groups that differ in one or facets of the treatment. A component central to treatment is varied (e.g. amount of contact with the animal; type of animal)
Comparative treatment strategy How effective is this treatment relative to other treatments for this problem? Two or more different treatments for a given clinical problem (could be two different models of AATs or AAT vs. some entirely different treatment)
Treatment moderator strategya What patient, therapist, treatment, and contextual factors influence (moderate) outcome, i.e. affect the magnitude or direction of impact Treatment as applied separately to different types of cases or as administered by different types of therapists to see what variables influence the effectiveness of an AAT
Treatment mediator/mechanism strategyb What processes within or during treatment influence, cause, and are responsible for outcome (therapeutic change)? Treatment groups in which processes during the course of treatment are evaluated to identify critical constructs on which therapeutic change depends

a A moderator refers to any variable that changes the magnitude or the direction of a relationship. For example, if an AAT is more or less effective with boys rather than girls (or younger rather than older children), then sex (or age) is considered be a moderator.

b A mediator refers to a statistical relation between an intervening variable and the relationship between an independent variable (e.g. AAT) and dependent variable (treatment outcome). A mechanism is the underlying basis for the effect and identifies the processes or events that are responsible for the change and how these changes come about (see Kazdin, 2007a for further discussion).

The first and most fundamental question is whether treatment is effective and surpasses changes over time that might otherwise occur without treatment. This is not an easy study to carry out in many situations because of ethical issues alone raised by withholding treatment for a no treatment or wait-list treatment control condition. From a methodological standpoint, a comparison of treatment versus no treatment is the most basic of the studies to show that treatment is effective. Also, this is the study most likely to show an intervention effect. Effect sizes (the magnitude of change) usually are stronger for treatment versus no treatment studies than treatment versus some other treatment condition. When two treatments are compared (e.g. AAT vs. treatment as usual), the differences between means of the groups and the effect sizes are smaller and a much larger sample size is needed for statistical power (Kazdin and Bass, 1989).4 In principle, the comparison of treatment versus no treatment is a place to begin to establish the efficacy of treatment. In practice, withholding treatment makes this difficult to do with clinical samples. Also, now with many viable (evidence-based psychotherapies) the treatment/no treatment comparison is of diminished interest in many contexts (e.g. treatment of anxiety, depression, disruptive behaviors). There are already effective treatments and, in such cases, there is interest in showing some benefit or advantage in relation to one of these (e.g. lower cost, broader impact, less attrition from treatment, stronger maintenance of changes). An alternative is comparing a treatment (AAT) with a treatment that is routinely used (referred to as “treatment as usual”).

The second question in the table, what components contribute to change, may be of special relevance to AATs. In evaluating and establishing the effectiveness of AATs, we take almost as a given that use of an animal in therapy contributes to therapeutic change. A study in which an AAT is compared to no treatment does not provide evidence that the animal made a difference to treatment outcome. There is strong evidence that meeting with a client and establishing a relationship contributes to therapeutic change when no animal is present. These influences are called the non-specific treatment factors or common factors of therapy and they alone seem to effect change. The factors are referred to as common because they characterize many even if not all therapies (Wampold, 2001). Among the elements are attending sessions, meeting with a therapist, expecting improvement, and having others (e.g. parents, other relatives) who may also expect change. If an AAT were more effective than no treatment, the most parsimonious interpretation is that the effect was due to the impact of common factors. Another parsimonious interpretation would be that treatment with a human therapist led to change, again without suggesting the animal made a special contribution. These are plausible and parsimonious interpretations because they explain larger sets of studies in a uniform fashion in which animals (other than humans) were not present.

To test whether the animal contributes to therapeutic change requires that one group receives an AAT and another group receives a very similar treatment but without the animal present. From my perspective, of all the questions listed, question two is the highest priority for AAT research. There is a firm belief that presence of the animal makes a difference. I am not challenging that. Yet, I do not believe there is a strong body of empirical research in the context of child treatment to make the case persuasively, i.e. to the standards mentioned previously (see Table 25.1).

I mentioned that treatment versus no treatment control group studies are difficult to do with clinical samples and individuals in need of services. In contrast to that, question two is more feasible and user (clinician, administrator, and client) friendly. Among the groups needed for the study are two variations of treatment, one with and one without the presence of an animal. Because the vast majority of psychological treatments do not involve non-human animals, a comparison intervention without an animal is not likely to raise ethical concerns. The question is whether the animal makes an addition to treatment outcome and that can be readily addressed by a study comparing two genuine treatments.

The questions in Table 25.2 provide a broad portfolio of research. A well-designed study addressing any of the questions would be an excellent contribution. I have favored question two; other questions might be equally compelling. For example, the question of moderators addresses various conditions (e.g. type of children, type of animal, and child/animal combinations) that might influence therapeutic change. The investigator may have a view that some children (e.g. older versus younger; boys versus girls; individuals with anxiety versus other types of problems) will respond better or that one type of animal (or breed of a given animal) is better for a specific problem or age group. One could begin with a test of one of these hypotheses and not only evaluate the efficacy of treatment but also whether efficacy varies as a function of some other variable (moderator). There is no a priori reason to begin with one question rather than another. Yet, as one moves down the list of questions in Table 25.2, the experimental designs, assessment procedures, and data analyses can be increasingly complex and demanding.

25.3.3 Control and comparison conditions

Arguably, the two most fundamental questions for AATs are: Does the presence of an animal improve the effectiveness of therapy? and if it does, What is it about the animal contact that makes a difference? These questions nicely illustrate the control and comparisons needed in research and move to a more helpful level of specificity than enumerating the broad portfolio of treatment questions.

The first question about the contribution of the presence of an animal is relatively easily addressed. It is helpful to begin by recognizing that there is a large therapy literature of children interacting with animals where that animal is a human. We know that child/human contact can lead to therapeutic change. AATs propose that therapeutic change is enhanced when an animal is added and new experiences are built around that addition. It would be important to include a control condition in which an AAT is compared with the same or as similar as possible treatment with just the therapist. Support for the first question is easily tested by showing the AAT leads to greater therapeutic change or is better in some other way (e.g. subjective evaluation of the session by the child, better participation of the child in the treatment process) when compared to that treatment or very similar treatment without the addition of an animal. In brief, to show that the non-human animal makes a difference requires controlling for the influence of the human animal (therapist) in the session. That study is not difficult to do.

Let us assume that in fact the above comparison, treatment by a therapist with an animal leads to greater therapeutic change than that same or similar treatment without the animal. We move to the second question: What is it about the animal/child contact or about introducing an animal in the session that makes a difference? Is it the use of a live animal (e.g. dog) in the session or would a non-living substitute (e.g. stuffed animal) do just as well? We know that children use stuffed animals to cope with fear and anxiety (e.g. Muris et al., 2001) and hence these “animals” too might be reasonable therapeutic aides. Tinkering with stuffed animals so they can be heated (Weiner, 2001) and perhaps warm and cozy might make their utility in therapy even more plausible to some. Presumably, most of us believe strongly that there is something very special about the live animal/child bond that is therapeutic. From a methodological standpoint, all we need to do is to specify that feature or set of features and ensure that any comparison or control group does not receive that part. Again, this is precisely where the small theory of the investigator not only dictates the therapy group but also the appropriate control group.

Based on that small theory, a credible control condition might be based on robotic pets, especially since robotic pets are becoming increasingly sophisticated in terms of their response to children, their ability to acquire information (learn) and patterns of interaction, and their ability to “see” and react (through small cameras). For example, the robotic dog, named AIBO™ (by Sony), has been subjected to careful research pertinent to the discussion of AAT.5 After a period of interacting, children accord AIBO mental states (e.g. AIBO can feel happy, tries to obtain a nearby toy) and biological characteristics (e.g. can grow, breathe, feel pain). Also, children ascribe to AIBO the ability to establish social rapport (e.g. AIBO likes the child, can be a friend, wants to spend time with the child) (Kahn et al., 2006). Children at different ages view the capacities and social attributes of a robotic pet, stuffed animal, and live animal differently, however, the similarities are striking (see Melson et al., 2009). Differences in child interactions with AIBO and a live dog favored AIBO on some measures (e.g. social interaction, touching) and a live dog on others (e.g. stroking, participation in the sessions) (Ribi et al., 2008). AIBO has been suggested as an alternative to use in AAT. Indeed, AAT research using AIBO has increased activity and social behavior among the elderly and surpassed the impact of a control condition using a stuffed animal (Tamura et al., 2004).

The purpose here is not to advance the use of robotic pets in the context of treatment. Yet, the notion of robotic animals and their increased sophistication can help sharpen our view and research on AATs. Precisely what is it about the child/animal contact or connection in AAT that is an aid to therapeutic change? This question refers to the process of treatment and what an AAT may uniquely provide. Can robotic animals in the session, stuffed animals, or other means (e.g. virtual animals and pets on a computer screen) be used to provide the benefits of animals in therapy? Many, if not most of us, believe that the live animal (and indeed live pet) are without peer in what they provide for human interaction. In everyday life all we need to support that belief are subjective reports of individuals from surveys. However, the AAT professional literature has a more sophisticated agenda and challenge, namely, empirical demonstration that live human/non-human interaction has effects that are unique. That agenda requires demonstration against strong control conditions that omit the unique components of live animal/human interaction.

I have made the assumption that most professionals interested in AATs would view live animal/human interaction as critical. It is possible that many would view animals in all of their living and non-living forms (e.g. robotic pets) as potential therapeutic aids. That view would greatly change the nature of a treatment study. If a live animal is conceived as critical, then contact with a stuffed or robotic animal is a possible control condition. If animal contact in any form is critical, then stuffed and robotic animals become two viable and possibly equally effective treatments. Again, the methodology, in this case what might be a control condition, stems from the small theory about a given AAT and the facet(s) of child/animal contact critical to therapeutic change.

Only gold members can continue reading. Log In or Register to continue

Feb 16, 2017 | Posted by in GENERAL | Comments Off on Methodological Standards and Strategies for Establishing the Evidence Base of Animal-Assisted Therapies
Premium Wordpress Themes by UFO Themes